.' _1EIf l i.. i.5i!...iillw M N N B. 1.. h~~~~~~~~~~~~~~~~~~~~~~~~ , l. = r =s s s ~~~~~~~~~~~~~~~~~~~~fo ss X Ie. Public Disclosure Authorized

Save this PDF as:

Size: px
Start display at page:

Download ".' _1EIf l i.. i.5i!...iillw M N N B. 1.. h~~~~~~~~~~~~~~~~~~~~~~~~ , l. = r =s s s ~~~~~~~~~~~~~~~~~~~~fo ss X Ie. Public Disclosure Authorized"


1 Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized al~~~~~~~~~~~~~~~~~~~~~~sl ; '-ujuluzmmw---- ~~~ It lif't5.> fl Elf-iWEtfWIi5I S -~ ~_, ~ S,, _ 3111 ' '! -!'_= a.' _1EIf l i.. i.5i!.....iillw M N N B.,,= 1.. h~~~~~~~~~~~~~~~~~~~~~~~~ = r =s s s ~~~~~~~~~~~~~~~~~~~~fo ss X Ie, l l l=4 1lill'%WYldii.Ul~~~~~~~~~~~~~~~~~~ ita=iii 1 w l~ 6t*t

2 Estimating Woody Biomass in Sub-Saharan Africa

3 Estimating Woody Biomass in Sub-Saharan Africa Andrew C. Miflington Richard W. Critdhley Terry D. Douglas Paul Ryan With contributions by Roger Bevan John Kirkby Phil O'Keefe Ian Ryle The World Bank Washington, D.C.

4 @ 1994 The International Bank for Reconstruction and Development/The World Bank 1818 H Street, N.W., Washington, D.C , US.A. All rights reserved Manufactured in the United States of America First printing March 1994 The findings, interpretations, and conclusiorns expressed in this publication are those of the authors and do not necessarily represent the views and policies of the World Bank or its Board of Executive Directors or the countries they represent Some sources cited in this paper may be informal documents that are not readily available. The manlerial in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommnercial purposes, without asking a fee. Permission to copy portions for classroom use is granted through the CopyrightClearance Center, Inc-, Suite 910,222 Rosewood Drive, Danvers, Massachusetts 01923, US.A. The complete bacdist of publications from the World Bank is shown in the annual Index of Publications, which contains an alphabetical title list and indexes of subjects. authors, and countries and regions. The latest edition is available free of charge from Distribution Unit, Office of the Publisher, The World Bank H Street, N.W., Washington, D.C 20433, USA., or from Publications, The World Bank, 66 avenue d'idna, Paris, France. The boundaries, colors, denonminations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any ternitory or the endorsement or acceptance of such boundaries. The cover shows a representation of the vegetation index for Africa,June 1986, derived from the NoAAsatellite's AvHRsensor. Library of Congress Cataloging-in-Publication Data Estimating woody biomass in Sub-Saharan Africa J Andrew C MilJington -.. [etal]. p. cm. Includes bibliographical references and index. ISBN Fuelwood crops-africa, Sub-Saharan-Geographical distribution. 2 Woody plants-africa, Sub-Saharan-Geographical distribution. 3. Forest biomass-afirica, Sub-Saharan-MeasurenenL 4. Vegetation classification-africa, Sub-Saharan. SD536.6.A357E '3911'0967-dc2O CIP

5 Contents- Preface and Acdnowledgments Authors and Contributors Acronyms and Abbreviations xiii xi. xiv Part I Application of Remote Sensing for Woody Biomass Assessment and Mapping 1 Introduction and Background I 2 Using Meteorological Satellite Data for Vegetation and Land Use Mapping 5 Remote Sensing 5 Sensors and Data Products 5 Advanced Very High Resolution Radiometer (AvHR) Data 6 Normalized Difference Vegetation Index (NDwvi) 7 Vegetation and Land Cover Mapping from NDvi Data 7 Production of NDvi Data by NASA,GSFC 8 3 ll pping of Land Cover Class 10 Phase I-Data Inspection and Preprocessing 10 Phase II-Initial Image InteIpretation 11 Phase I1-Automatic (Supervised) Classification and Mapping of Land Cover Classes 15 Land Cover Mapping 15 InterpretingAvHRRNDvI Data for Woody Biomass, Stock, and Sustainable Yield 16 4 Woody Biomass Assessment 19 Data Base for Biomass Estimation 19 Area Calculations of Land Cover Class 23 Interfacing Area Grouwng Stock and Data on Sustainable Yield 23 5 Description of Biomass Classes 24 6 Regional Summaries by Cass of Biomass 29 The West African Sahel 29 The West African Coast 31 The Hom of Africa 32 Central Africa 34 East Africa 35 Southern Africa 38 7 Condlusions and Future Directions 41 References.46 v

6 vi Contents Part II Regional Distribution of Land Cover Classes 8 The West African Sahel Richard W. Critchley 53 Class 0-Desert 53 Class 12-Hydromorphic Grassland 54 Class 21-Semidesert Wooded Grassland 54 Class 44-Sahel-Sudanian Acacia Wooded Bushland 55 Class 62-Dry Sudanian Woodland 56 Class 64-Sudanian Woodland 57 Class 65-Moist Sudanian Woodland 58 Land Cover Class Tables 59 References 59 9 The West African Coast Andrew C. Millington 64 Class 12-Hydromorphic Grassland 64 Class 44-Sahel-Sudanian Acacia Wooded Bushland 64 Class 62-Dry Sudanian Woodland 65 Class 64-Sudanian Woodland 66 Class 65-Moist Sudanian Woodland 67 Class 74-Guinean Woodland 68 Overview: West African High Woody Biomass Mosaics 71 Class 75-High-Productivity West African Cultivation andforestmosaic 71 Class 76-Medium-Productivity West African Cultivation and Forest Mosaic Class 81-Mangrove Land Cover Class Tables 77 References The Hom of Africa John lrkby 85 Class O-Desert 85 Class 12-Hydromorphic Grassland 86 Class 13-Ethiopian Montane Steppe 86 Class 21-Semidesert Wooded Grassland Class 22-Acacia Wooded Grassland Class 41-Dry Acacia-Commiphora Bushland and Tiket 89 Class 43-Moist Acacia-Commiphora Bushand and Thicket Class 44-Sahel-SudanianAcia Wooded Bushland Class 45-Escarpment Wooded Thicket 92 Class 52-East African Low Woody Biomass Mosaic 93 Class 62-Dry Sudanian Woodland 94 Class 63-Sudan-Ethiopian Woodland and Thicket 95 Class 64-Sudanian Woodland 95 Class 65-Moist Sudanian Woodland 96 Class 73-Cultivation and Forest Regrowth Mosaic 97 Class 74-Guinean Woodland 97 Class 77-Highland Cultivation Mosaic 98 Class 84-Montane Forest 99 land Cover Class Tables 100 References Central Africa Ternj D. Douglas 105 Class 11-Veld Grassland 105 Class 25-Edaphic Wooded Grassland 105 Class 62-Dry Sudanian Woodland 106 Class 64-Sudanian Woodland 107 Class 65-Moist Sudanian Woodland Class 66-SeasonalMiombo Woodland 108 Class 67-WetMiombo Woodland 108

7 Contents vii Class 72-Cultivation and Forest/Woodland Mosaic 109 Class 73-Cultivation and Forest Regrowth Mosaic 109 Class 74-Guinean Woodland 110 Class 82-Evergreen Forest 111 Class 85-Mesophilous Humid Tropical Forest III Class 86-Hunmid Tropical Swamp Forest 112 Class 87-OmbrophlUous Humid Tropical Forest 112 Land Cover Class Tables 123 References East Africa Phil O'Kfe,IanRyle,andJohnlCirkby 119 Class 0-Desert 119 Class 21-Semidesert Wooded Grassland 119 Class 24-Transitional Wooded Grassland 120 Class 33-Bushy Shnibland 121 Class 41-Dry Acacia-Commniphora Buslland and Thicket 121 Class 43-Moist Acacia-Commiphom Bushland and Thicket 122 Class 51-Acacia Woodland Mosaic 123 Class 52-East African Low Woody Biomass Mosaic 124 Class 61-Open Woodland 124 Class 65-Moist Sudanian Woodland 124 Class 66-Seasonal Aiombo Woodland 125 CLass 67-WetMiomb Woodland 126 Class 71-Evergreen Woodland Moaic 127 Class 72-Cultivation and Forest/Woodland Mosaic 127 Class 73-Cultivation and Forest Regrowth Mosaic 128 Class 74-Guinean Woodland 128 Class 82-Evergreen Forest 128 Class 8-Coastal and Gallery Forest- 29 Class 85-Mesophilous Humid Tropical Forest 130 Class 84-Montane Forest; Class 86-Humid Tropical Swamp Forest; Class 87-Ombrophilous Humid Tropical Forest 130 Land Cover Class Tables 131 References Southern Africa Roger Bevan 136 Class 0-Desert. 136 Class 11-Veld Grasslad 137 Class 14-Montane Grassland and Heatbland 138 Class 24-Transtonal Wooded Grassland 139 Class 31-Veld Shrubland and Cultivation Class 32-Hill Shrubland Class 33-Bushy Shrubland Class 35-Wooded Shrubland 145 Class 42-Fynbas Thicket 146 Class 34-Kalahari Shrubland Class 43-Moist Acacia-Comnziphomn Bushland and Thicket 147 Cls 51-Acadza Woodland Mosaic 149 Class 61-Open Woodland 151 Class 66-SeanMwmbo Woodland (including Tropical Coastl Woodland) 154 Class 67-WetMiombo Woodland (icluding Warm Temperate Woodland) 158.Class 71-Evergreen Woodland Mosaic 160 Class fl-cultivation and Forest/Woodland Mosaic 162 Class 82-Evergreen Forest 163 Land Cover Cla Tables 165 References 165

8 viii Contents Glossary 175 Index of Botanical Names 279 Index of Place Names 285 Figures 1-1 Project Methodology Characteristic Land Cover Reflectance Curves and Relation to AVHRR Sensor Bandwidths Small-Scale and Large-Scale Perspectives of Radiation Absorption and Reflection Relation Between NDVI and Above-Ground Biomass and Vegetation Cover Distribution of Cloud Cover in 1986 Imagery rcvi Profiles, Summary Land Cover Classes 0, 1,, NDvi Profiles, Summary Land Cover Classes 3,4, Nrn Profiles, Summary Land Cover Classes 6,7, Regional Summary Land Cover Classes Growing Stock for Sub-Saharan Africa, Sustainable Yield for Sub-Saharan Africa NDVI Profile, Hydromorphic Grassland (Class 12) NDVI Profile, Semidesert Wooded Grassland (Class 21) NDVI Profiles, Sahel-Sudanian Acacia Wooded Bushland (Class 44) 56 B-4 NDcv Profiles, Sudanian Woodlands (Classes 62,64, and 65) vi Profiles, Sahel-Sudanianl Acacia Wooded Bushland (Class 44) ,-wvi Profiles, Sudanian Woodland (Class 64) Dvi Profiles, GuineanWoodland (Class 74) Nvi Profiles, High-Productivity West Afican Cultivation and Forest'.:.osaic (Class 75) 9-5 Cote d'ivoire Wood Production, 198D C6te d'lvoire Wood Production by Class, C6te d'lvoire Wood Production by Class, Nvi Profiles, Medium-Productivity West African Cultivation and Forest Mosaic (Class 76) Niv Profile, Mangrove. (Class; 81) Nwv Profiles, Bushland and Thicket (Classes 41,44, and 45) NDVI Profile, Sudamwian Woodland (Class 65) NOW Profiles, High Woody Biomass Mosaic (Classes 74 and 77) NEW Profile, Montane Forest (Class 84) Nwm Proffle, Edaphic Wooded Grassland (Clas 25) NOv Profiles, Sudanian Woodland (Classes 62,64, and 65) 11-3 NDvi Profiles, Miombo Woodland (Classe 66 and 67) NDvi Profile, Cultivation and Forest Regrowth Mosaic (Class 73) NOvi Profiles, Forests (Classes 82,85, and 87) NtVI Profiles, Bushland and Thicket (Classes 41 and 43) NDvx Profile, Seasonal Miombo Woodland (Class 66) NDvi Profiles, High Woody Biomass Mosaic (Classes 72 and 73) NDon Profile, Evergreen Forest (Class 82) 129 L3-1 NDvi Profiles, Transitional Wooded Grassland (Class 24) NDvI Profiles, Shrubland (Classes 34 and 35) NDvi Profile, Moist Acacia-Commiphora Bushland and Thicket (Class 43) 147 L3-4 NDW Profile, Acacia Woodland Mosaic (Class 51) NDvi Profiles, Open Woodland (Class 61) Nov Profiles, Seasonal Miombo Woodland (Class 66) Nov Profiles, Wet Miombo Woodland (Class 67) NDvw Profiles, Evergreen Woodland Mosaic (Clss 71) Novi Profiles, Cultivation and Forest/Woodland Mosaic (Class 72) MDvi Profiles, Evergreen Forest (Cla 82) : -. ~-..

9 Conten ts ix Tables 1-1 Final Energy Consumption for African Countries, Woodfuel Consumption in Sub-Saharan Africa, Multistage Strategy for Woody Biomass Resource Assessment CloudCoverinl986lmagery 2! 3-2 Land Cover Classes for Sub-Saharan Africa Growing Stock and Sustainable Yield Data, Distribution of Summary Classes and Estimated Woody Biomass, West African Sahel Distribution of Summary Classes and Estimated Woody Biomass, West African Coast Distribution of Summary Classes and Estimated Woody Biomass, Horn of Africa Distribution of Summary Classes and Estimated Woody Biomass, Central Africa Distribution of Summary Classes and Estimated Woody Biomass, East Afrca Distribution of Summary Classes and Estimated Woody Biomass, Southem Africa Land Cover Classes-Burkina Faso (West African Sahel Region) Land Cover ClassesChad (West African Sahel Region) Land Cover Classes-The Gambia (West African Sahel Region) Land Cover Classes-Mai (West African Sahel Region) Land Cover Classes-Mauritania (West African Sahel Region) 62 R-6 Land Cover Classes-Niger (West African Sahel Region) 62 B-7 Land Cover Classes-Senegal (West Aficain Sahel Region) Areal Distribution, Dry Sudanian Woodland (62), West African Coast Region earal Distnibution, Sudanian Woodland (64), West African Coast Region 66 : 9-3 Areal Distrnbution, Moist Sudanian Woodland (65), West African Coast Region Areal Distribution, Guinean Woodland (74), West African Coast Region Woody Biomass Stocks and Growth Rates, Guinean Woodland Conmmunities, Lamto, Cote d'ivoire a Meal Distnbution, High-Productivity Cultivation and Forest Mosaic (75), West African Coast Region b Areal Distribution, Medium-Productivity Cultivation and Forest Mosaic (76), West African Coast Region Equivalent Names for Cultivation and Forest Mosaic in West African Coast Region.9-8 Degree of Deciduousness, Reserved Moist Forests, Sierra Leone Land Cover Classes-Benin (West African Coast Region) Land Cover Classes-Ghana (West African CoastRegion) Land Cover Classes-Guinea-Bissau (West African Coast Region) Land Cover Classes-Guinea (West African Coast Region) Land Cover Classes-COte divoire (West African Coast Region) Land Cover Classes-Liberia (West African Coast Region) Land Cover Classes-Nigeia (West African Coast Region) Land Cover Classes-Sierra Leone (West African Coast Region) Land Cover Classes-Togo (West African Coast Region) Land Cover Classes-Djlbouti (Hom of Africa Region) Land Cover Classes-Ethiopia (Hom of Africa Region) Land Cover Classes-Somalia (Horn of Africa Region) Land Cover Classes-Sudan (Hom of Africa Region) Land Cover Classes-Cameroon (Central Africa Region) Land Cover Classes-Central African Republic (Central Africa Region) 116..: Land Cover Classes-Congo (Central Africa Region) Land Cover Classes-Equatorial Guinea (ainland; Central Africa Region) Land Cover Classes-Gabon (Central Africa Region) Land Cover Classes-Zaire (Central Africa Region) Land Cover Classes-Burundi (East Africa Region) 132

10 x Contes Land Cover ClIsses-Kenya (East Africa Region) Land Cover Classes-Rwanda (East Africa Region) Land Cover Classes-Tanzania (East Africa Region) Land Cover Classes-Uganda (East Africa Region) Land Cover Classes-Angola (Southern Africa Region) Land Cover Classes-Botswana (Southern Africa Region) Land Cover Classes-Lesotho (Southem Africa Region) Land Cover Classes-Malawi (Southem Africa Region) Land Cover Classes-Mozambique (Southern Africa Region) 170: 13-6 Land Cover Cl_sses-Namibia (Southern Africa Region) Land Cover Classes-South Africa (Southern Africa Region) Land Cover Classes-Swaziland (Southem Africa Region) Land Cover Classes-Zambia (Southern Africa Region) Land Cover Classes-Zimbabwe (Southem Africa Region) 174 Maps (at end of book) Legend for Regional Land Cover Class Maps Regional Land Cover Class Map of West Africa Regional Land Cover Class Map of East Africa Regional Land Cover Class Map of Central Africa Regional Land Cover Class Map of Southem Africa

11 Preface and Acknowledgments Woodfuels play a dominant role in the energy balance John Kirkby, Richard Critchley, and Anthony Mellor. of countries in Sub-Saharan Africa, and shortages of Jon Styles' participation was made possible through a woodfuel exist-or are perceived to exist-in many leave of absence from research training granted by the areas. This volume describes a first attempt to map the Natural Environment Research Council (U.K). Manvaried vegetation and types of land cover in Sub- agement and administration of the project were led by Saharan Africa and to assess the growing stock and Phil O'Keefe of mrc (U.K). sustainable'yield of woody biomass. It is the result of As a first step, it was decided to conduct a contia project undertaken by the Energy Sector Manage- nental overview of woody biomass in the region, nent Assistance Programme (Esmu", a joint program from Advanced Very High Resolution Radiometry of the World Bank and the United Nations Develop- (Avhln) data obtained by satelite. Instrumental in m- ni Programme) to address the lack of data on en- this initial stage were the help of Jim Tucker, who erg -resources for most African countries. This infor- permitted access to data at the U.S. National Aeromation is essential for calculating energy balances. nautics and Space Administration's Goddard Space The work's usefulness, however, goes far beyond Flight Center, and John Townshend, then director of questions of energy sufficiency. Appropriate manage- the Unit for Thematic Information Services at the Natment of Africa's biological resources, to which it is ural Environment Research Councila University of hoped this work will contribute, will help maintain Readin& who provided computational facilities for and perhaps enhance environmental quality on the data interpretation. continent. Moreover, the extent of worldwide forest Drafts of the maps and the project report were cover has come to be recogimzed as a critical variable presented at workshops in Nairobi, Kenya, and in predicting global climate change. Together these Abidjan, C6te d'ivoire, in May An important considerations make the assessment of existing wood part of the collective workwas the peer review led and resources in Africa a timely intervention. - conducted by senior African scholars and profession- The impetus for this project came from an ESMAP als at these meetings. These individuals included Household Energy Seminar held in Harare, Zim- Jean-Marie Ouadba and Cyrille. Kabore (Burkina babwe, in February Energy and forestry officials Faso), Benedict Fultang (Cameroon), Dieudonne attending the seminar reached the conclusion that Dzimasse (Central African Republic), Me Kouame, data on African woodfuel resources, needed as a basic N'Zore Kadja, Jean-Claude Anoh, and Aidara tool for the planning of woodfuel programs, were in Gouesse Largeni (C6te d'ivoire), Waldikidan Nere short supply. With generous funding provided by the and Sultan Tilmo (Ethiopia), Kingsley Ghartey and government of the Netherlands, the mapping of Afri- Gamphi-Aidoo (Ghana), J. Gore, J. Mutie, P. Waciori, can land cover and assessment of its woody biomass H. Mwendwa, and J. Agatsiva (Kenya), Antonio that are the basis of this book were undertaken by nrc Ferrio (Mozambique), Issa Boubacar (Niger), Reg. (UK), under the supervision of Paul Ryan of ESmA. Cine-Cole (Nigena), Isatu Deen (SierraLeone), Ismael The project, which got under way in July 1988,.was Grammadid(Somalia),MustafaElHaqandMohamed a collective effort between a team at the University of Elrady (Sudan), Idris Kikula. and Edward Mlowe Reading, led by Andrew C Mllington and consisting (Tanzania), David Siamuele (Zambia), Peter Gondo of Jon Styles, Richard Saull, Pam Kennedy, and Nick (Zimbabwe), B. K Kaale (Southern African Develop- Drake; and a team at the University of Northumbria, ment Community), and Barry Henrilson (Regional led by Terry Douglas and consisting of Roger Bevan, Remote Sensing Centre, Nairobi). Their local knowl- -.~~x

12 xli Prfacc nd Aclcnowlcdgments edge of vegetation and woody biomass statistics was crucial to the project's success. As a result of their critiques, we were able to refine the maps and the vegetation typing. Thanks also are extended to the librarians, particularly at the World Bank in Washington, D.C., and the Food andagricultureorganization(pao)oftheunited Nations, who diligently pursued our bibliographic inquiries beyond the normal call of duty. Thanks are due also to Gary Haley of the University of Northumbria, who prepared all of themaps, and to Joalle Hivonnet, translator for the French edition of this volume. Revised maps and the final report were presented to ESMAP in January It is fair to say that the project achieved its objectives. Although the accuracy of the assessment is limited by the resolution of the satellite images and by the paucity of existing data on woody biomass In the various vegetation types, the study resultedinaveryusefulmappingreference. Itnot only provides sufficient information on woody biomnss resources to enable broad policy and pldnning decisions regarding woodfuel management but also should help researchers investigating broader issues such as the contribution of caibon sequestering by living plants to the mitigation of global greenhouse warming. Webelieve this collective effort tobe a small step toward understanding, maintaining, and enhancing Africa's wood resources, on which so many peo- ple depend for their well-being......~~~~

13 Authors and Contributors Andrew C Millington, read- m the Department of Geography, University of Reading (UK) Richard W. Critczley, senior lecturer in Ecology, Department of Environment, University of Northumbnria at Newcastle (U.K) Ternr D. Douglas, principal lecturer in Environmental Management, Department of Environment, University of Northumbria at Newcastle (U.K) Paul Ryan,aforestryspecialistwitltheEnergySectorManagementAssistanceProgramme of the World Bank and the United Nations Contributors (Ems&), ajointprogram Roger Bryan, senior lecturer in the Department of Environment, University of Northumbria at Newcastle (U.K.) John Mirkby, senior lecturer in the Department of Environment, University of Noithumbria at Newcastle (U.K.) Phil O'Keefe, professor in the Department of Environment, University of Northumbria at Ne-wcastle (U.CK) Ian Ryle, Education and Trairiing Consultants (U.K) xiii

14 Acronyms and Abbreviations AErFAT Association pour 1'Etude Taxonomique.. Leaf Area Index de la Flore de 1"Afrique Tropicale NASA/GSPC National Aeronautics and Space AVHRR Advanced Very High Resolution Administration/GoddardSpace Radiometer Flight Center CAR Central Afric.n Republic NDVI Normalized Difference Vegetation ccr Computer compatible tape Index cpu Central processing umit NOAA National Oceanic and Atmospheric cinr Centre Technique Forestier Tropicale. Administration DN Digital number - c (Natural Envirorunent Research rers Earth Resource Satellite Council) Unit for Thematic ESmr - Energy Sector Management Assistance InformationServices Programme (of the World Bank rso Polar Stereographic Projection and the undp) SADC Southern African Development EFc( Tuy-) Education and Training Consultants Community (formerly sdcc, South- FAo0. - Food and Agriculture Organition em African Development Coordinaof the United Nations tion Conference) GAc Global Area Coverage SMLMfR ScanningMultichannelMicrowave. GE Geographic information system Radiometer cv Global Vegetation Index SPOT Satellite pour l'observation de la Terre HRFr High Resolution Picture Transmission ToE Tonnes of energy equivalent -ED International Institute. for Environment (metric tons) and Development UNDP United Nations Development MNR Institut de Recherche Agronomique Programme. du Niger. UNMo United Nations Educational, Scientific, IRET Institut de Recherche en Biologie and Cultural Organization et Ecologie Tropicale WARDA WestAfricanRiceDevelopment LAC Local Area Coverage Association. ~ ~. ~ i.. ~~~~~~~.1~~~~~~~~~~~ xiv

15 PART I Application of Remote Sensing for Woody Biomass Assessment- and Mapping

16 Introduction andbackground present in Sub-Saharan Africa. The project was con- ducted byercouc.) under the auspices and supervision of ESMAP, ajoint program of the World Bank and UNDP (United Nations Development Programme). On examining the energy balance for most Sub- Saharan countries, one is struck by the domiinance of woodfuel, induding fuelwood and charcoal This is well illustrated in table 1-1, which shows that wood- Woody bionmass constitutes the main domestic fuel in many parts of Africa. The need for more defintive data on this resource was perceived at a Household EnergySeminarheldbytheWorldBankEnergySector Management Assistance Pogamme (Emw) in Harare, Zimbabwe, in February The present project was conceived as a resull It is a first attempt to produce an analysis by type of land cover of the woody biomass Table 1-L Fnal EnerConsumption for African Countnes, (pent) -- - Woodfidel and a-ldgrigdhraland Total Country Petrolum Coke or coal Ekltrik4 bmass residue (1,000 TeE) Year Benin s Botswana Congo Cote d'lvoire Ethiopia , Gabon n The Gambia Ghana ' , Liberia Malawi , Mozambique , Niger Nigeria , Rwanda , Senegal A 1, Sierra Leone Somalia , Sudan , Swaziland Tanzania , Togo Zambia , Zinbabwe , Weighted average a. Inmany countriesdata are ladcingon the quantityofagricultuiresidues consumedinless-and climates this may amomnttolittle,but inwoodel-poorsaheliancountries it could attain 10 percentof total energy mnsumption. * SourcS : unmp/worldbar&kenergy Sedor a--essments. =...~~~~~~~~~~~~~~

17 2 Estinating Woody Biomass int Sub-Sahmn Afica fuel and agricultural residue represent 67.8 percent of 1-2). Assessment of woodfuel supply, however, has the final energy consumption on a weighted average not been undertaken in most countries. Further, spabasis. The dominance of woodfuel is principaily re- tial extrapolation from the limited reliable data that lated to households, in which the most important use exist is fraught with ecological and environmental is cooking. Woodfuel, however, also is important for interpretive problems. rural and small industries such as tobacco curin& tea Without reliable data on the growing stock and, drying, fish smoking,beerbrewing,brickmaking,and more important, on the sustainable yield of woody commercial balidng. biomass, it is not possible to prepare mnearingful woo d- Accurate estimates of household energy consump- fuel development and conservation plans within tion and overall woodfuel consumption are not avail- an overall energy strategy. Unfortunately, some proable for many Africancountries. Limited surveyshave grams and projects have been designed and imbeen conducted, enablingestimation of woodfuel con- plemented without defining the sustainability of the sumption in conjunction with demographic data (table woodfuel resource in relation to current and pro- Table 1-2. Woodfuel Consumption in Sub-Saharan Africa, 1990 (millionair-drytonnes) Region Country Consumption R-egional total West African Sahel Burkina Faso Chad 2.5 The Gambia Mali 3.7 Mauritania 0.8 Niger 23 Senegal 42 West African Coast Bendin C8te d'ivoire 62 Ghana 8.6 Guinea 2.9 Guinea-Bissau 03 Liberia 3.9 Nigeria 513 Sierra Leone.3.0 Togo 2.8 Hom of Africa Ethiopia Somalia 3.3 Sudan 14.6 Centmal Africa Cameroon Central Afican Republic 2.6 Congo 1.1 Gabon 1.8 Zaire 274 East Africa BurUkdi Kenya 25.3 Rwanda 3.6 Tanzania 34.9 Uganda 17.0 Southern Africa Angola Botswana 12 Lesotho 0.6 Malawi 102 Mozambique 11.1 SouthAfrica 52 Swaziland 0.4 Zambia 6.8 Zimbabwe 63 - Not available- Sourc uwar/world Bank Energy Sector assessments.

18 Introduction and Bnckground 3 Iected consumption, including potential-changes in Bothestinationrmethodsfailtoconsiderbiomasson consumption. farms and around villages. This is a significant limita- Estimatesof woodybiomass derived from field data don in the use of such data for energy planning, becan be divided into two groups, the classical forest cause recent studies have indicated the importance of inventory and estimates of above-ground woody bio- wood produced on farmland (Munslow and others mass. In the classical forest inventory, methodologies -1989). used by foresters have generally concentrated on com- In this book we describe the initial stages of a mulmercial wood in high forests. Consequently, estimates tistage mapping and assessment strategy for produchave focused on the stem volume of potentially com- ing estimates of growing stock and sustainable yield mercial species and within commercial size limits. of woody biomass in Sub-Saharan Africa (table 1-3). Such estimates neglect important sources of woodfuel To achieve this we created a data base of the existing such as branches, twigs, dead wood, woody under-- woody biomass estimates just described and extrapogrowth, shrubs, and noncommercial tree species. These lated these data in a framework of spatial land cover. mensuration estimates give little indication of woody The framework derived from remotely sensed data at biomass stocks and productivity in forest areas, yet a contnental scale. The stages of this methodology are these areas, particularly woodlands, are a principal shown in figure 1-1. woodfuel source in much of Africa. Clearly, only two types of data can be used to map Estimates of above-ground woody biomass are more the woody biomass resource at- this scale: existing useful and have been made in some parts of Africa. maps and remotely sensed data. A compilation of Despite their greater application potential for energy previously published maps would have provided an planning, considerably fewer of these studies have unsatisfactory product for three reasons: the age of been conducted compared with forest inventories, and the maps in relation to the dynamic vegetation and they are restricted mainly to semiarid areas. land cover; the variation in scale of mapping; and the Table 1-3. Multistage Strategy for Woody Biomass Resource Assessment Review of Existing Dab and Maps for Area Low-Spatial-Resolution Satellite Imagery * AdvancedVery HighResolutionRadiometer(AYHR) * Resolution: 1.1 to 15 kilometers * Environment large country or multicountry region * Supported by limited ground verification and existing environmental data * Provides 1:5,000,000-scale maps with broad land cover zones High-Spatial-ResolutionSateitel Iagery a Landsat Multspectral Scanning System (mm), resolution 79 meters * Landsat Thematic Mapper (iu), resolution 30 meters * spar (Satellite pour l'observation de la Terre) High-Resolution Visible (nv) multispectral, resolution 20 meters * Envirornment small country or portion of a country * Supported by ground verification, aerial photography, airbome video imagery, and existing data onbiomass estirates * Provides 1:250,000-to-1:1,000,000-scale maps with more detailed land cover types Biomass Reconnaissance Inventory * s orhrv (panchromatic) or aerial photography, or both * Low-intensity inventory, perhaps with subplots for smaller vegetation types * Supported by cartographic or image maps and established regressions and correlations between measurable para-.meters and biomass * Destructive sampling to establish regressions * May include nonforest as well as forest areas Biomass Management Inventory * High-intensity inventory wvith subplots supportedby maps in forest and nonforest areas Conversion of Industrial Wood Volumes to Total Above-Ground Biomass * Supported by maps or aerial photography, orboth, and established regressions and correlations between industrial wood volume and total biomass Note Elements are not rsarily in sequential order.

19 4 Estitrating Woody Biomass in Sub-Soharant Africa Figure 1-1. Project Methodology Remotdely sensed -. tt dwa~~~~~~~~~~~~~~~~~~~~~~~~~~dt - >Analysis of remolely =uetsed data - I Promsional land Datbase covcr classnmaps elpment w IdentudScion h orurldis an- S Rcvisionl{ - W 3 -~~~~on Lancd very Highs Repsolto Raioetr(Au ata, cors, t wsdcddt s heahrdtrm18 which aspatil ith resoutionof 8ilomeers w vere forkthiswr fteveeaindt frln oe approptc riaterfor this proecthebss fobjctv moft mappingu mappingulthatinwt beca enavitsabltfo the NaVHRRa Aeronland cover. For other studies, such data. have been ery beginning in 1981, the data for 1986 appear to be used successfully to map land cover at the continental thebest quality forarelativelynornalrainfail year over scale in Africa (Townshend and Tucker 1984) and most of Africa. The use of a single year's data leads to Latin America (Townshend, justice, and Kalb 1987), certain problems, which will be addressed later, and and the woody biomass resource in southern Africa the maps produced represent only asituational ov.er-., (Millirngton and others 1989). view of the woody biomass resource for 1986.

20 2 Using Meteorological Satellite Data for Vegetation and Land Use Mapping In this chapter we define remote sensing, stressing its Sensors and Data Products application to the mapping and monitoring of vegetation. We describe the main satellites and sensors, dif- Sensors for the acquisition of remotely sensed data are ferentiate between Earth resource satellites and mete- mounted on board various satellites. These satellites orological satellites, and discuss the nature of digital can be broadly grouped by function into Earth reimagerv. We then examine in detail the Advanced source satellites and meteorological satellites. Earth Very High Resolution Radiometer (AvHmR), the use of resource satellites include the American Landsat Normalized Difference Vegetation Index (Ntw) for series, the French spor satellite, the European Space land cover mapping and vegetation monitoring, and Agency's Es-i, and the Indian Space Agency's *RS-x. the production of Nvim for this project Meteorological satellites include those of the National Oceanographic and Atnospheric Administration (NOAA) Remote Sensing and the European Space Agency's METOSAT. The roster of operational satellite-sensor combinations constantly Satellite remote sensing is the technique of acquiring changes; contemporaneous reviews are available in data (imagery) that describe the Eart's surface and Cracknell and Hayes (1991) and in Drury (1990). atmospherewith the use of satellites. Sensors on board -Data are available either as photographic (optical) satellites provide data at a synoptic scale from a single products or as computer compatible tape (ocr). The image acquisition, thereby making the analysis of re- ccr often is more useful for resource assessment bemotely sensed data a cost-effective tool for resource cause it can be digitally processed to optimize the assessment over large areas. information that can be extracted. Such digital image A further advantage of satellite data is that repeat processing can be carried out effectively only with a imagery for any scene usually can be acquired. The computer-basedimage-processingsystem-interpretarepeatperiodbetween imageacquisitions varies from tion of photographic products is limnited because data less than 24 hours for meteorological satellites to be- cannot be manipulated to highlight particular aspects tween 16 and 26 days for Earth resource satellites. The of the image. In some cases, however, photographic repeat period can be much longer for Earth resource prints provide adequate results. satellites (for example, 18 days for Landsat-4 and Sensors cover the ultraviolet, visible, infrared, and Landsat-5) for various technical and logistical rea- microwave portions of the electromagnetic spectrum. sons. The capability to provide time-sequential data The Earths atmospheric gases act as "windows" to is of great advantage in both biomass resource as- allow the tiansmission of radiation of specific wavesessment and monitoring because of the relatively lengths, and satellite sensors are constructed to correrapid response of plants to changes in the weather, spond to these windows, maximizing reception of the climatic seasonality, changing climates, and other radiation reflected or emitted from the Earths surifce disturbances. Sensors, however, are not built to cover all the avail- 5~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

21 6 Estimating Woody Biomass in Sub-Saharun Africa able windows between ultraviolet and microwave Figure 2-L Characteristic land Cover Reflectance wavelengths. Although it would be valuable for bio- Curves and Relation to AVHRR Sensor Bandwidths mass assessment to have a sensor covering the electromagnetic spectrum from ultraviolet through micro- - 2 wave, thus permitting the simultaneous sensing of all Bnndidth of chans 1 and 2 radiation and absorption features, this is not possible on AVHRR sensor for technical reasons. This situation should improve Healthy with the anticipated launching of mulfisensor plat- vtatio r forms in the mid-to-late 1990s. _-O' Biomass resource assessment using remotely sensed soil data mainly concerns the interaction of electromag- I netic radiation with the vegetation canopy. Radiation ~! _,.L-- - of all wavelengts interacts with this canopy, if one is ' present, and at any wavelength the radiation is either ---- J: Stressed or sneseon absorbed, reflected, or emitted. In addition, if the veg- x_i etation canopy does not completely cover the ground, Watcr some interaction between radiation and the ground surface will occur. To understand how these interactions affect bio- 4 w 6w 700 wo gw mass assessment and mapping, the concept of scale Blue Gaza Rd Mearinfd must be considered. At the scale of arn indidual leaf, radiation interacts with the cells in each leaf of a plant. Although examiiination of individual leaves does not provide information of direct relevance to a biomass Detailed studies relating the physiology and biosurvey, this information is fundamental to understand- chemistry of plants to their spectral characteristics has ing the reflectance from vegetation. Of more direct been undertaken (for example, Sellers 1985, 1986; relevance is the interaction of radiation with all ele- Tucker and Sellers 1986). These studies conclude that ments of a vegetation canopy and with any soil ex- spectralreflectanceofvegetationisielatedto thephysposed through the canopy. It is this kind of interven- iological processes that drive plant growth, namely, tion that occurs at the spatial scale of most sateuite photosynthesis and respiatiorl remotely sensed data and therefore an understanding Shiftng our perspective from the scale of a single of it is relevant to the types of measurements made by- leaf to the scale of a vegetation canopy, it is dear that sensors on board satellites. radiation is reflected fromboth the vegetation and the It is possible to measure the proportion of radiation underlying soil (figure 2-2, bottom). Sensors for both reflected at a particular wavelength and to derive a parts of the spectrum, 0.6 to 0.7 Pm and 0.8 to 1.1 pm,l characteristic curve for green vegetation or a particu- are used in most Earth resource and meteorological lar land cover type. Such curves can be quite distinct sateites. (figure 2-1), but the precise nature of a vegetation curve or land cover curve varies according to the type Advanced Very High Resolution Radiometer of vegetation, the time of year with respect to plant (AvHn) Data development, whether the plant is stressed, and the completeness of the vegetation canopy-that is, how The data used in this work were acquired by the much reflection occurs from the soil surface : Advanced Very High Resolution Radiometer The spectral characteristics of vegetation reflectance (AvIu), a sensor carried on board the American are such that remotely sensed data are most useful for NOAA series of meteorological satellites. Data mapping and monitoring of vegetation and land use derived from the AVIER imagery were supplied in when they are acquired in the visible red and near- ccr format by the National Aeronautics and Space infrared regions of the spectrum. In the visible red Administration/Goddard Space Flight Center region (0.6 to 0.7 micrometer or pm), radiation is ab- (NASA/GSFC). sorbed by chloroplasts for use in photosynthesis. This Of six NOAA satellites launched between 1979 and contrasts with the strong reflection of near-infrared 1988, two remain operational: NoAAj-io, launched Octoradiation that is used for sensing vegetation (0.8 to 1.1 ber 1986, and NOAA-i, launched September The pm); it is caused by reflection and refraction from leaf AvHR sensors on board each satellite. have the followcells (figure 2-2, top). ing spectral channels

22 Channel Wavelength nrnge, zipper to lower (grm) UsingMeteorologicalSatelliteDala 7 equation dimninishes atmospheric effects. The NDVI is produced routinely by NAsA/G5srC and is calculated by: a ND VI - DN (Channel2); -DN (Channell)i DN (Channel2)i + on (Channell)j whereon is the digitalnumberrepresentingthereflec- Each satellitehas an average orbital elevation of 833 tance from channel n for the ith pixel. kdlometers and a period of 102 minutes, affording complete coverage of the Earth's surface every 9 days. Vegetation and Land Cover Mapping Ground coverage is an area of 2,700 Iclometers, with from NDvi Data a ground resolution of 1.1 kilometers at nadir. Data are acquired for any part of the Earth's surface Since 1981 a number of researchers have taken advanby the AviR sensor oanboard each NoAA satellite every tage of such ecological data from meteorological sat-.12 hours. Il every 24-hour period, an image for a. elites for mapping of vegetation and land cover over specific area is acquired once dunng daylight and large areas. These indude Danaher and others (1992); once during the night For example, at the equator, the Eidenshink and Haas (1992); Gatlin, Sullivan,. and local time of image acquisition for NOAA-9 was 1430 and Tucker (1983); Goward, Tucker, and Dye (1985); Gray for motaa4o was and McCrary (1981); Koomanoff (1989); Norwine and For work on vegetation mapping only the daytime rgreegor (1983); Schneider, McGinnis, and Gatlin imagery is used because it reflects the interaction be- (1981); Tarpley, Schneider, and Money (1984); tween incoming solar radiation and the vegetation Townshend and Tucker (1984); Townshend and oththat is the basis for differentiating vegetation types, ers (1991); Tucker, Holben, and. Goff (1984); Tucker, cover, and status. The daily times quoted here are Gadin, and Schneider (1984). This body of work has those when the polar-orbiting NOAA satellites cross the led to a number of natural resource applications in the equator. These times,.however, become later during tropics, includingv the life of each satellite because of orbital drift, affecting the overall reflectance. Gutman (1991) offers fur- * Monitoring Lake Chad (Schneider, McGinnis, and ther discussion of this. Stephens 1985) The short time interval between daily acquisitions * Mapping biomass production for grazing in Seneof imagery creates a high probability that a pixel or gal (Tucker and others 1985) ground resolution element (the area of Earth's surface - Monitoring semiarid rangeland in Africa (Prince viewed by the sensor at any time during its scan) will be free of clouds. 1986, 1991; Diallo and others 1991; Franklin and Hiernaux 1991; Justice and Hiemaux 1986; Prince and Tucker 1986) Normalized Difference Vegetation Index * Monitoring semixarid rangeland in Australia (Bar- (Nnvi) ber 1992; Flemons 1992) * Monitoring cropping patterns in Southeast Asia In the absence of clouds, the AvHR sensors, particu- (Malingreau 1986) lar.y those for Channel 1 (0.58 to 6.8 tm) and.channel * Assessing the woody biomass resource for wood 2 (725 to 1.10 pm), will sense vegetation, bare ground, energyplanninginsouthemafrica (Millingtonand or open water. Consequently, useful ecological infor- others 1989) mation can be obtained about land surfaces- This in- * Drought early warning systems (Hendrickson and formation enables calculation of an index of vegeta- Durkin 1986) tion stacus for each ground resolution element A * Droughtmonitoringin Australia (Brook and others vegetation index is derived to reduce measurements 1992) of the vegetation canopy, ground cover, and biomass * Vegetation management in Australia (Roderick and to a single number (Perry and Lautenschlager 1984). Smith 1992) Many different vegetation indices are available to * Monitoring tropical deforestation in Amazonia scientists who use remotely sensed data, but the Nor- (Malingreau and Tucker 1987) malized Difference Vegetation Index (NDvi) has been * Monitoring land cove:' changes associated with used successfully and extensively with AvHRR data desertification in Palkstan (Saull, Mllington, and principally because the nornalizing nature of the Crosettil991).

23 8 Estimating Woody Biomass in Sub-Saoaran Africa Figure 2-2. Small-Scale and Large-Scale Perspectives of Radiation Absorption and Reflection Near-infrared Visible red ( pm) ( pm) / /C~~~~~~~~~~~~~~~hloroplasts Upper-epidermnis PalisadecelLs Spongy nesophyll ]ith I3 ; ti UUx Leaf cross-section Lower epidermnis C C=0 D C-Z Reflection and absorption of visible red and near-infrared radiation in part of a single leaf Radiation pathways in an open canopy typical of zavanna woodlands Some studies have correlated the Nv with biomass The NDVI is available at different spatial resolutions, parameters such as above-ground biomnass (figure 2-3, includin& for example: left) and vegetation cover (figure 2-3, lower right). Althcaghnone of the studies relates specifically to tree Nwvi at a spatial resolution of 8 kilometers is known or shrub biomass, they do suggest that similar core- as the Global Vegetation Index (GVI)(Tarpley 1991). lations should be discovered for woody biomass. These data also are available at 15 kilometes resolution after resampling to a polar stereographic Production of NOvi Data by NASAIGSFC projection. NUVI at a spatial resolution of 4 kilometers is known Early mapping of vegetation carried out using AVHY} as Global Area Coverage (G;AC) data. derivednrvrdataso vividlyillustratedthepotentialof * NDv at a spatial resolution of 1.1 kilometers is these data that the NDvi is now routinely calculated by ]known as Local Area Coverage (LAC) data or High NAsA/GsrcandtheFoodandAgricultureOrganization Resolution Picture Transmission (wrt) data, deof the United Nations (FAo), using the above equation pending onhow it is obtained. If it is obtained by a

24 Figure 2-3. Relation Between NVVI and Above-Ground Biomass and Vegetation Cover Using Meteorological Satellite Data /0 I * Correlation coefficient a* * ' I- /500 OA- - -/ U / ~~~~~~~~~~~~~~~~~~~1,- w /.. D 03- -, j / - o 0-9 * * C0 -c. _ S _ *00.~~~~~~~~~~~~~~~m /m * ~ ~ ~ ~ 3 * o.s l.t 5 10 IS }Drdt fi sotie eodth aih*zn wl xeti ra whegraed NV daruycous arpev 0~~~~~~~~~~~~~~~~~ of a it gound i recoded tatio byon-bord tae alet(condersion coefficeant 7=0.901) Pobem U * 0 c~~~~~~~~~~~~~ tion iti A aa.diy* aaaecmostdoe 0dyitras ay1 ayeid ahpxli xmie n h 0nkoma MeanDV data abvailable grcen biomss (g haveben dy'ihh Pgrentest Nv fr gtatio covelr susdt groundreeivn gtainneatmiisnwa,pxl ihdus procsd NDn) rpeettedentre This h-as beeno aclcowleded to6) work X -. R f. : : - :~~~~~~~~~~~~~~~~~~~~~~ HnT data.ifitimgsobaineid bfeyoengd te radioiznf welloedurep areducs where thin afromdthe areospreverecodersand thluen transmittedt sthegrsoun whnayi weres endsountre withdouds uaion 4. -;-M fdtahuedvegetatis -Etio;iis, then kndxown asla tempoa. dailyke mm data area copsie ovr1-aitras (ta is th rdco f tepoa copsie of th ne

25 Mapping of Land Cover Class 3~~~~~~~~~~~~~ To identify and map classes of land cover, we interpre- data. In the data set used for this work, errors involved. ted Normalized Difference Vegetation Index (NDvi) only individual pixels. We constructed a digital filter digital data derived from the Advanced Very High to solve the problem of empty single pixels. The filter Resolution Radiometer (AvHi). -The work was per- was a modified 3 x 3 pixel average filter that replaced formed in three phases: any pixel which fetl outside the range of possible DN (digital number) values (120 to 210) with the average Phase I Data inspection and preprocessing of the DN values of its eightneighboring pixels. A pixel Phase II Iitial image interpretation, derivation of alsowasreplacedifitfellwithinthisrangebutthesum the NDvi temporal profiles (Nrov phenolog- of the differences between the pixel and its neighbors ies), provisional mapping of land cover exceeded a certain threshold. This was done so that all casses, and limited field verification of -images that subsequently were processed would not provisional maps of land cover dass have any pixels that lacked data. Phase III Automatic classification and mapping of Larger areas that -lacked data entirely also were land cover classes. observed in the imagery. These occurred in areas where cloud cover persisted during the imaging periods and The followingsections describe the equipment, data meant that, during some months, no pixels free of specifications, errors, preprocessing, identification of clouds could be found from which to calculate the land cover classes, mapping, and interpretation of the NDVI. These areas did not occur in every month of the - mvr data. year at a particular location, but because a continuous sequence of months is needed to define land cover Phase I-Data Inspection and Preprocessing classes, the presence of cloud in pixels clearly could affect the accuracy of the final land cover map. These All image processing was undertaken on mainframe problem areas occurred mainly in the coastal humid and microcomputer-based image-processing systems tropics, in the high mountains of eastem and central at the University of Reading. The image data initially Africa, around Lake Chad, and in the vicinity of Table wereevaluatedon-screen,andhardcopyimagerywas Mountain in South Africa. produced for intermediate stages of the interpretation The smaller areas of cloud were eliminated after process and for field verification, - classification (compare with Phase IIl) using an itera- Global Vegetation Index (GVI) data at 8 kilometers tive median-ffitering method. The original NDvi data spatial.resolution for 1986 were provided bynasmagsfc and the filtered data then were combined using a for 10-day and monthly intervals. These data were logical or operation, so that the original data were registered to one another and were provided on the altered only when a zero value (cloud) occurred in the Hammer-Aitoff Conic Equal Area Projection. Initial original data. Some very large areas of cloud, howinspection of these data disclosed two types of errors: ever, could not be eliminated entirely by filtering; pixels lacling NDvi values, and misregistered data, these occurred in Cameroon, Ethiopia, Nigeria, South The digital data recorded on tape are configured Africa, and Zaire. These areas were exduded from the into pixels by tlhe image-processing system. An ab- final classified images because they would produce sence of data in a pixel can occur because of sensor, classes that were impossible to interpret transmission, or processing errors. Unless attended to Consequently, we produced a cloud mask which before processing, these errors remain in the processed- was overlaid onallof thefinalimages. Weconstructed 10

26 Mappinga/Land CoverClans It1 the mask by extracting information on cloud occur- Difference images were created by subtracting one rence pixel-by-pixel for each month and then sum- set of monthly NoVI data from another and adding a ming the monthly data. lhe mask shows the pixels constant. The constant was added pixel-by-pixel to where cloud restricted the calculation of the mmvi for corresponding pixels to ensure that all values were at least one month during Areasunder this cloud positive and could be displayed. The months used to mask thus lack information on land cover, and these create the difference Images were chosen on the basis areas are not included in the calculations of area, of known phenological characteristics of African veg growing stock, and yield. The extent of cloud-affected etation and land use. pixels is summarized country-by-country in table 3-1, and thedistributionof cloudcover ismappedin figure 3-1. Table 3-1. Cloud Cover in 1956 Imagery Data for one of the 10-day periods in March were misregistered with the other images, offset to the east - o -o. -j-arcn Area by six pixels. To solve this problem, we simply reregistered these data to the other 10-day-perio data. The West African Sake! Burkina Faso 0 0 March NDvimage then was recalculated and these - h-chad 0 0 data were used in the Phase III analysis. The Gambia 0 0 Mauritania 0 0 Phase II-Initial Image Interpretation Niger 0 0 Senegal 0 0 In this phase, we derived images from the data and West African Cost identified and mapped provisional.land cover Benin 0 0 classes. C6te d'lvoire 0 0 -Imagery Denrvedfrom AVHRR Data Ghana 0 0 Guinea Guinea-Bissau Variousimageproductswerederivedfromthemonthly Sberia -o11, NDvi imagery during this initial interpretation phase Siea Leone 263 O n which provisional biomass classes were identified and mapped. The images used in this phase were: - Centr Africa Cameroon 14, Individual monthly images -wvr Central Individual African Republic monthly 0 0 NDv images.cog- Congo 0 0 * Mean annual NVI images Equatorial Guinea 1,634 7 e Difference images between two monthly NDVI values Gabon - o * Unsupervised classification images. Zaire 66,818 3 All of the images were produced at various scales, Southem Africa Angola 1,634 0 ranging from the entire continent to individual coun- Botswana -0-0 tries,tofacilitaternterpretationandimapping.individ- Lesotho uacin images were prepared for each month. Because Malawi 896 Mozambique 5,744 the data were 1 supplied directly from NASA/GSFC, image Namibia 1,475 a production required only tape unloading, inspection south Africa 2,37 0 for errors, and color coding of the image using the Swaziland 0 0 standard NAsA color scale (see the section in Chapter 2 Zambia 2,002 0 titled "Vegetation andlandcovermappingfrom NDVi Zimbabwe a 0 Data"). ~~~~~~~~~~~~~Horn of Afiica A mean annual Gvi image was created through Djibouti 0 0 pixel-by-pixel summing of all monthly Gvi values for Ethiopia 15,071 1 corresponding pixels. The sum of the GVI values for Somalia 0 0 each pixel then was divided by 12 (months). This Sudan a 0 mean-annual image provided a good indication of Enst Africa annual vegetation productivity and the spatial varia- Burundi 1,686 6 tion within it. It is analogous to the integrated NDVI- Kenya 4,216 1 images produced by other workers (see Chapter 2), Rwanda 4,058 2 but differs because this image is calculated from the Tanzania 7,325 1 area under the annual NDvi curve. Uganda 3,636 1

27 12 Estimating Woody Biomass in Sub-Salwran Africa Figure 3-1 Distribution of Cloud Cover in 1986 Imagery.. ;Y... > -U.... j \ ( Martai \--. -? U\ -I~~~~~~~~~~~~~~~~~~~~~~~~~~V - KChad h. n Cite leer.. d,rpaddisa -.-- =~~~~~~~~~~~-ua g 4 * DJitu ti * r X a wlk& 1 aia rse Etip Republic - llaauira *- ' nc, - r - vai' pia E B*ill% ;d ip E i I 7.imbabwe z.... \./: ~~~~~~~~~~~~~~~~~~~(c* r.**;n *....;f Por jdul Bangin ia(zrs«ll 9Nairubi nmndi' Gabo,Sth)NWI r, p 1. ft DodaniaiaDrmSae augoan ~ ~ ~ nznl * - ::::..?. 1P - :* S Arc Lun ~ ~ ~ ~ ~ ' ln s ~~~~~~~~~~~~~~~~. I.e Ca ;. I, 'I.21.t2..~~~~~~DnV>... Clou cover I..... M...pu Two types of difference images are particularly use- or the onset of winter conditions (such as cooler tern- ~~ful in thle interpretation of land cover classes: - peratue or tshe incesd frequency of frosts) Images in which data-for the low-w period 2. Imagesin which dataforthehigh period are -(vegetation senescence) are subtracted fom data for subtracted from dab for the low-wvi period on a the high-wow period(maxiumu greenness)on apixelby-pixel basis. Such images provide a good indication pixel-by-pixel basis. These images provide informa- lion on the greening of vegetation. - of vegetation dieback in response to declining sea- In this phase of the analysis, classification of images sonalrainfall,varyinglevelsofresidualsoilmoisture, was restricted to unsupervised classification. In this

28 Mapping of Land Cover Class 13 context, unsupervised classification is the clustering of Southern African Development Community (sawc) repixels into statistical groups, defined by the temporal gion (Millington and others 1989). Second, the United distribution of their NDVI values. Four months were Nations Educational, Scientific, and Cultural Organichosen for the unsupervised classification. These represented periods of high NDvl and low NDvi and the two intermediate periods during which the NDvI values Figure 3-2. NDVI Profiles, Summary Land Cover were increasing or decreasing at varying rates, de- Classes 0,1, 2 pending on the type of vegetation and land use. These DESIC data then were clustered such that pixels having a (Summay Class 0) statistically similar range of values over the 4 months 0.6 were grouped into the same zlass. o.s -Nonhser MaurirAnin In addition to the images just described, temporal Eastern Nanibia profiles of the NDVI values were plotted; these are 0.4- called NDvi phenologies. These profiles indicate the seasonal fluctuations of the ovi, which can be used to z 0.3- analyze and interpretseasonalvariations inphotosyn- 02 thesis, leaf area index, and biomass production. Each point represents the average NDvi over the same 3 x 3 D.1- pixel square area on each monthly image. Represer.tative temporal profiles are provided in figures 3-2,3-3, O- I J and 3-4. Althoughtheyrelate tosome of the land cover n F Mar Apr May Jun Jul Aug Sep Ot Nov Dc classes mapped inphase m, they also are typical of the profiles derived in Phase IT and were used in the provisional identification of land cover class and in field verificatiorn GRASSLAND Iden tfication and Mapping of Prouisional i (SumLaar(ass 1) Cover Classes - Central Niger South Africa/ \ All of the images were evaluated on-screen, and hard 0.4/ copy imagery was generated. The hard copy images - were used to identify and map land cover dasses on a 03-1:5,000,OO-scale base maps for the six areas where _ field checking was undertaken (a) South Africa and Swaziland, (b) western Zaire and Ccngo, (c) West D.1- Africa-Senegal and the Gambia, (d) West Africa- Mhali (e) West Africa-Cbte d'lvoire and Ghana, and (f) Ethiopia.JaFeIS f) - Ethiopia : o- ', - M. n ~~~~~~~~~~~Jan A p O N.c. Identification of vegetation tpes in these areas employed the criteria of. Feb Mar Apr May Jlm Jul Aug Sep Oct Noy Dec * Vegetation phenology, obtained from temporal profiles of the NDvi values and the seasonality (dif- WOODED GRASSLAND ference) images - Class 2) * Spatialpattemsofvegetationonmonthlyandmean Cenan] Sudan annual GVI images 0.5 Namibia * Secondary data sources, particularly vegetation, land use, and forestry maps, but abo includmg environmental information such as geological, soil, > and topographic maps and statistics on climate. Z D0-. We decided to undertake the initial mapping of provisional land cover classes at the scale of 0D-1 1:5,000,000 for two reasons. First, this scale had been o I _ used successfully for woody biomass mapping in the Jan Feb Mar Apr May Jun Jul Aug Sep Oc Nov Dce 7. ~ ~ ~ ~ ~ ~ ~ -.

29 14 Estimaling Woody io mass in Sub-Salaran Africa Figure 3-3. NDVI Profiles, Summary Land Cover Figure 3-4. NDVI Profiles, Summaiy Land Cover Classes 3, 4,5 Classes, 6,7,8 SHRUBLAND WOODLAND (Summary Class 3) mmary Class ~~~~~~~~~~~~~~~~~0.6 (umr Cas6 - Capc Province South Afric z n ~~~~~~~~~z Southern Burkdna Fao - Northwestem Zambia I I - I -I I - i * - I I I I I Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dcc Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0O4 BUSHLAND AND THICKET HIGH WOODY BIOMASS MOSAICS (Summary Class 4) (SummNry Class 7) _ CenWal Mali 05--Sox lomaia 0.5 OA- z z 02- ~~~~~~~~~~~ Soutem Cftcd Ivoire 0.- I I, I I I Jan Feb Mar Apr Mayun Jul Auj Sep Oct NovDec I - Southem Nigeria I I IAu ' I Jan Feb Mar Apr May Jun Jul Aug Scp Oct Nov Dcc LOW WOODY BIOMASS MOSAICS FOREST 0.6 (Summary Class 5) (Summary Class 8) Northem Botswana Z * : Southcrn Camcroon _n -Nonheastem Zahi 0- p, I * I I I I I 0 O m l i II 1 1,d Jan Feb M nrapr May Jun Jul Aug Sep Oct Nov Dec Jan Feb MarApr May Juan Jul Aug Sep Oct Nov Dec zation (unbsoo) VegetationMap of Africa (White 1983) was produced at this scale. The unssco map is one of the most recent vegetation maps covering all of Africa and provides the most consistent reference source for checking image inter- pretations This map was compiled from previous floristic maps, as well as expert review, but is not a land cover map. Nevertheless, we adopted the broad mapping units devised by White (figure 3-2). Adoam tianrwas possible because the classes of land cover are

30 Mapping of Land Cover Class -15 related to different vegetation communities (forexam- The final classification was based on the 6 months ple, forest, woodland, shrubland, bushland, grass- of January, April, June, July, October, and Decemland, and desert) or types of land use. Consequently, ber. These months were chosen by a process of elimtheir distribution does not exactly correspond to any inating the other 6 months of the year. August was one previous botanical or ecological study of the re- eliminated because it had high-altitude cloud cover gion, although they are broadly comparable to most north of the equator. The months of February, regional vegetation maps. The structure adopted is March, May, September, and November were elimdiscussed in Phase l. inated because correlation coefficients >0.9 were discovered for the adjacent-month pairs of January- Phase 111-Automatic (Supervised) February, April-May, May-June, September-Octo- Classification and Mapping of Land Cover ber, and October-November, creating a virtual data Classes redundancy. A by-product of this decision was that central processing unit (cpu) time for the classifica- The final biomass class maps were produced using tion using only 6 months of data was reduced drasupervisedclassification.supervisedclassificationisa matically compared with that using 12 months of partcularly powerful image-processing procedure be- data. cause the value of each pixel in each spectral channel Previous work on land cover mapping has shown is compared with various sets of pixel values (called that using more months in the analysis inproves the training statistics) established from preselected, classification accuracy when compared with training known areas (called traimnig sites). The individual site statistics (Townshend, Justice, and Kalb 1987). pixels are assigned to the most suitable class, based on Althoughwe donot question this finding, the decision the training site statistics. In this study, the data used to use 6 months of data was based on operational for supervised classification were not from different critera rather than scientific. It should not be assumed spectral changes but were Gcn from different months. that accuracy necessarily would increase if more As a consequence, the classification identified and months in 1986 were used in the classification, but mapped pixels having similar NDVI phenologies. Dif- accuracy would increase significantly if data from ferent algorithms canbeusedinsupervised classifica- other years were included. The reason is that the NDw tion, but in this project a maximum likelihood classi- is very sensitive to rainfall and soil moisture condifier was used. tions, which vary considerably from. year to year Training sites were identified from Phase II inter- across much of Africa. pretations. These were located in areas where provi- The most accurate map of land cover that could be sionally interpreted land cover classes attained nmaxi- produced would use a data set back to 1981, when the mum areal extent, or where they were known to be NDvi data first became available. We strongly recomdistinctive. Training sites were selected from field mend that future use of the NDvi data for application teamreportsandpreviousresearch.trainigstatistics of natural resources employ long-term data sets, were generated for all the types of land cover identi- rather than data sets for single years. fied, and the data were classified on this basis. This produced an automated classification of sixty land Land Cover Mapping cover classes for all of Sub-Saharan Africa. We decided to use data spanning 6 months for the The land cover classes were reinterpreted using the supervised classification. This decision was based on same citeria employed to identify themin Phase IL At two criteria. First, in the woody biomass assessment this stage, classes having strong similarities were carried out for the SAnc region using the NDvi data, the merged, producing forty-three land cover classes. main technique for map production was supervised These are listed in table 3-2 and shown on the four classification (Milhington and others 1989). Evaluation regional maps at the end of this volume. The classes of the imagery showed that a supervised lassification were reviewed by experts in the workshops held in of land cover classes based on the monthly GVI data Abidjan and Nairobi, and recommendations were was more meaningful than other techniques. made for further merging of classes. Secnd, significant temporal autocorrelation exists Most of these mergers are of dasses having statistiin the NDvi data for adjacent months, which could lead cally similar Nv curves which, on the basis of floristic to data redundancy if all 12 months were used in the and ecological criteria, were incorrectly split in the classification. We tried classifying separately two se- Phase Ill analysis.the special case of cultivation mosaquences of data for alternate months-january to No- ics in West Africa and Ethiopia, however, presented a vemberandfebruarytodecember-andfortheentire more difficult problem. In the West African coastal 12 months of data. The visual differences among the zone, areas originally classified as different types of three classified-image maps were insignificant. humidtropicalforestandforestregrowthweremerged

31 16 Estimating Woody Biomass in Sib-Salwran Afric - Table 3-2. Land Cover Classes for Sub-Saharan into two classes: High-Productivity West African Cul- Africa tivation and Forest Mosaic and Medium-Productivity Class nennber Land courr class West African Cultivation and Forest Mosaic. This merger was based on the realization that very 0 Desert little humid tropical forest remains in this region. The I Grssland remaining forest is in small blocks and reserves inter- 11 Vyddrno Grassland spersed with rubber, cocoa, and coffee plantations, 13 EtHidropinmorphicGraslane S farms, and a seral succession forest regrowth. This 13 - Etiopian Montane Step,oe 14 Montane Grassland and Heathland patten of land use occurs at scales that cannot be : - mapped with any accuracy from the con data. 2 Wooded Grassland In the highlands of Ethiopia, a similar problem :21 Semcidese Wooded Grassland exists with small land parcels having different bio- 22 Acacia Woodied Grassland 23 Plateau Wooded Grassland mass levels and NDVI profiles. Here altitude is art 24 Transitional Wooded Grassland additional factor, so we introduced altitude into the 25 Edaphic Wooded Grassland mapping procedure. The 2,000-meter and 3,500-3 Shnabland meter elevation contours were digitized and a 31 Veld Shrubland and Cultivation laid on the classified image. Except for Moncane 32 HllShrubland Forest, all classes occurring between 2,000 and 3, Bushy Shrubland meters in the area weremerged to form a class caled 34 Kalahari Shrubland -35 Woodeld hbland Highland Cultivation Mosaic. Ihose above 3,500 meters were merged to form a class called Ethiopian 4 Bushknd and Tl-cket Montane Steppe. 41 -Dry Axia-Commiphora Bushland 41 Dry A Tbicket COIZ-and Calasses of land cover are defined as areas having 42 - Fynbos dthicket types of vegetation or land use that are comparable in 43 MoistAcacia-Comm47honzBushland overall biomass, productivity, and seasonality. They and Thicket may, however, combine more than one type of ecolog- 44 Sahel-SudanianAcaca Wooded ical structure (for example, forests and woodlands) 45 -EsarpmentWoodedThicdet and many floristic units. In this respect they differ 4- Escarpment Wooded Thicket significantlyinplacesfromtheunuscvegetation Map 5 LowWoody Biomss Mosaic of Africa (White 1983). For the purposes of wood 51 Acuega Woodland Mosaic 5-2 EastAfrican Low Woody Biomass energy planning however, a classification based on o saic parameters that have been correlated with gowing 6Woodland -stock and productivity elsewhere is of more use than 61 Open Woodland a floristic-ecological map. 62 Dry Sudanian Woodland The maps of land cover classes are one of the final 63 Sudan-Ethiopian Woodland and Thicket products of this projectl The maps include: 64 Sudanian Woodland 65 MoistSudanianWoodland * Four regional maps depicting land cover classes, 66 Seasonal MdiowboWoodland f - appearing at the end of this volume. 67 Wet hmombo Woodland A regional summary map of land cover classes, 7 High Woody Biomass Mosaic constructed by categorizing the land cover classes 71 Evergreen Woodland Mosaic 72 Cultivation and Forest/Woodland Mosaic into eight groups on the basis of the dominant 73 Cultivation and Forest Regrowth Mosaic ecological structures (figure 3-5). These groups are 74 GuineanWoodland defined by White (1983). 75 High-Productivity WestAfrican Maps of growing stock and sustainable yield, ap- Cultivation and Forest Mosaic peaning in Chapter 7, figures 7-1 and Medium-ProductivityWestAfrican Culvation and Forest MosaicInterpreting AVR vi Data for Woody 77 Highland Cultivation Mosaic 8 Forest Biomass, Stock, and Sustainable Yield 81 Martgrove 82 Ever Forest As has been shown, we identified classes of land cover 83 Coastal and Gallery Forest by using imagery derived from the AVHRR data to- 84 Montane Forest gether with secondary ecological, environmental, and 85 Mesophilous HumidTropical Forest forestry datal The principal characteristics used to 86 Humid Tropical Swamp Forest distinguish the classes were (a) differences in vegeta- 87 Ombrophilous Humaid Tropical Forest tion cover, as indcated by their monthlynr values;

32 Mapping of Land Cover Class 17 Figue 3-5. Regional Summary Land Cover Classes. 3? 9~ 1). :~~ fl I~~~~~~I.~~~~~~~~~~~~~~~,. - Land ckss.2 cover I - d asslr and -Wd tan n rt to pm p f reaiob e te - a lion catedby areviewland the v ~ nuassof : -: termnthly o wrvaus (8 an th auhr e shwta lsiiaiosfo nlsso.- Low woodybiomass moacs. Thbai : = Woodlland aruet -Il - fo inepetto of th veeato tye of Afia In sumay the. reasons^'. G --. Highwoodybiomassmosaics. 0 0 x,,ff<_- -V- - t~~~~~~~~ (b) dife nle m anua primamy procduction, as indi- Townshed, Jushce, and Kalb (1987).In d-ee works, catedby the annualsums of their monthly NDvivalues; the authiors show that dassifications from analysis of and (c) differeces in phenology, as indicated by their data from satellite rernote sensin g correspond closely temporal profiles of NDVL -. to vegetatio,nmaps and field observations of prndcpal ThebasicargumentsforinterpretationoftheAvnn-. vegetatior, types of Africa. In sunuay, the reasons derived NmV data in relation to primary produc-.for this are (a) the strong relatioribetween the and tion are reviewed by Tukr and- others (1985 andl LeafArea Index (Lm) and (b) the regular acquisition of

33 18 Estimating Woody Biomass in Sub-Soliaran Africa cloud-free imagery using the AVHRR, which enables the imagery have been related to rainfall patterns in the phenology of the vegetation to be observed across the Sahel (Malo and Nicholson 1990). entire continent. 'A very strong seasonality in the vegetation also The most important factor that determines the type exists, caused by the distinct dry and rainy seasons and productivity of vegetation across most of Sub- (Samiento and Monasterio 1983; Rutherford 1978). Saharan Africa is rainfall However, in the far south The seasonalty of rainfaul affects the vegetation diand at high altitude, temperature fluctuations are also recly by affecting primary production (Woodward important (for example, in East Africa and the Ethiop- 1987) and indirectly by the prevalence of fires in ian Highlands). Strong correlations between rainfall the dry season. The latter has selected species that and vegetation often have been observed in Africa in have various forms of dormancy (Sarmiento and relation to vegetation community, composition, and Monasterio 1983). The phenology of African vegetaproduction-in southern Africa by Lamotte and tion is easily detected by the Nay! calculated from the Bourliere (1978) and Rutherford (1978), in Zaire by AvmRR data, and this contributes significantly to the Malaisse (1978), and in West Africa by Menaut and information content of the multitemporal classifica- Casar (1979, 1982). Such patterns depicted by NDWi tions used in this project (Tucker and others 1985).

34 -~~~ Woody Biomass Assessment In this chapter we review problems. in the data base, quently, the quality of the data base is extremely varithe procedure for area calculation, and our method of able. Ideally, data for each land cover class in each interfacing data to produce the maps of growing stock country should be used, but such data either do not and sustainable yield. exist or are not accessible. Even at the regional level, these data frequently are unavailable. Before we could Data Base for Biomass Estimation interpret imagery, we were forced to derive estimates forsomeclasses andcountriesfhroughgeeralization, We define growing stock as air-dried, above-ground extrapolation, or interpolation. Woody biomass, in tonnes per hectare. Sustainable Also,.many of the data collected are not accompayield we define as the mean annual increment of air- niedby sufficient methodological detail to permit satdried, above-ground woody growth, also in tonnes isfactoryevaluation.intheabsenceofotherdata,hoowper hectare.per year. ever, these estimates had to be used. Where no data were found, estimates were based on values in the Problems in the Data Base existing literature, bearing in mind empirical data from similar land cover classes and geographical Numerous problems exist in the data base and we areas. review these diffculties in this section. In summary, Our main sources of Literature were the PAo Forestry the problems include: Departmentlibrary inrome; Department of Forestry Library at Edinburge University; The World Bank - Extremely variable data from secondary sources LibraryandreferencedatafileinWashington;thesADc v Insufficient methodological detail data file held at the Department of Geography, Uni- * Merchantabletiberstudiesexcludefuelwoodand versity of Reading; and the Biomass User Network at understory King's College, University of London. Much of the * Studies of selected species and classes may be material at the first two sources concerns forestry prounrepresentative. jects, both local and countrywide, and consists of for- Studies assess volume, not mass est inventories or otherestinmtes of commercial wood, * Incompatible terms and no classification crteria often to the exdusion of fuelwood and omitting esti- * inconsistent or missing data for classes and regions mates of natural and seminatural vegetation. Further, * UnrepresentatiLve data from limited sampling and these estimates usually are of woody biomass of merno seral information chantable size, such as timber with a diameter greater * Studies do not relate biomass and population than 10 centimeters, or in many cases 30 centimeters. distributions This clearly omits much wood of great significance * Studies do not identify nonfuel uses of selected with regard to fuelwood. speciesr In particular, most studies exclude the understory, * Extensive nonforest vegetation near culture is a resource of great significance in fuelwood supply. excluded. Brown and Lugo (1984) suggest that the total biomass for the understory in closed forests is less than 2 per- The data base used for estimating growing stock cent, whereas in open forest the percentage may be and sustainable yield for each land cover class was much higher. They further point to large underestiderived from a number of secondary sources. Conse- mates that are likely to occur if only stem wood is 19 ''- '';~ ~~ ' 0 '~/' '' ' -- - ;

35 20 Estimating Woody Biomass in Sub-Sahrmn Africa considered. They suggest that the ratio of total woody extensive literature search, data of good quality are biomass to stem wood biomass in closed forest may be lacking for particular land cover classes, such as 1.6andasgreatas2.9 inopenforest,withthe increased high forests and the woody biomass mosaics, and proportion in the lower stories. Clearly, where mer- for some critical regions, such as the Sahel and East chantable timber and stem wood biomass are the Africa. focus, and the understory and standing dead wood are Yet another problem is whether the data are repreignored, a significant underestimate of the woody sentative. Factors such as the successional state of a biomass resource is likely. system can substantially affect growth rates of vegeta- These problems are generauy recognized, as is the tion, but most studies give no indication of seral stage. fact that the studies are concerned with a limited num- Many studies also rely on very small sampling areas, ber of land cover classes and tree species-those capa- creating the possibility of a large sampling error bebleofproducingcommercial-yields,andwhicharenot cause the data do not represent land cover classes necessarily representative of an area. mapped on a continental scale. Also, these estimates generally are concerned with No attempt was made in the present study to relate the volume of wood rather than its mass, and this is the distribution of woody biomass supplies to popuparticularly true for estimates in francophone Africa. lation distribution. (We do note that many of the stud- In these cases, conversion to air-dried mass within tlhe ies have been in areas far from population centers, data base was necessary. This contrasts with the situ- possibly malkng them inaccessible as woodfuel reation noted for the southern African states (Millington sources.) This shortcoming can be addressed by addiand others 1989), in which the commonest data avail- tion of current population estimates at the provincial able are estmates of the mass of dry material Al- or district level to existing biomass data in a geo - though oven-dried biomass is preferred for energy graphic information system (as). studies, so few make reference to it that air-dried There also has been no attempt to distinguish tree biomass is used here. species -that are used locally for specific purposes, Adjustment from volume to mass in the data base which would exclude them from use as fuelwood. used the following conversion factors: This problem needs local research and more data on -the quantit of individual species present. At this time Wood type Aerage air-dried denity little detailed information is available for the data Acac-Conznmphorawoodland 0.6 tonnes/mr3 base. Mliombowoodland it was not possible to take account of the extensive (Brnchystiq&-Julbewanrdia) 0.9 nonforest rees and shrubs that occur in dose associa- Rain forest 0.65 tion with villages and farmland in many areas. The SouthAfiicanwoodland 0.6 spatial resolution of the images was inadequate for their assessment, and conseq7uently almost no mean- These coefficients were derived from the literature ingful data exist on the growing stock and growth and discussion of the problem within amr. In a few rates of surh vegettionl Neverthieless, these trees and studies, stacked volume of wood (stans) is indicated, shrubs may form an important source of woodfuel for for which we used the conversion of 0.25 tonnes per manyruralpeoplewhosustainablycollectwoodyand stat. Volume-to-biomass conversions for commercial nonwoody biomass near their homes (Leach and timber species in high forests do not necessarily rep- Mearns 1988; Munslow and others 1989). resent the mensuration statistics forwoodfuelbecause The combined effect of these factors produces a variations occur due to species differences and their high degree of variability and unreliability in the use. For many species having less-dense wood, no data for each country and each land cover class. The data were available at all. final tabulated data must be regarded with these A further difficulty was the incompatibility of de- limitations. sciptive terms used for vegetation types and the lack of clear definitions and criteria for their lassification. Limitations of Datafrom Literature Sources In some instances this made problematical the assign-- ment of data to a particular land cover class. Table 4-1 indicates the growing stodk and sustainable Inconsistent data recording is another problem in yields deriveed from the data base. The median value creating a standardized data base. Such incorsisten- reported for each class represents the central value of ciesreflectthepartcularinterestsandobjectvesofthe a range of estimates from the literature perfinent to forester or researcher with the result that many refer- that class. Where no estimates were available, values ences provide estimates of, for example, growing of growmg stock and sustainable yield were denved stockbutnotofgrowthrates,orviceversa.despitean from ecologically similar classes. Few estimates are

36 Woody Biomass Assessment 21 Table 4-1. Growing Stock and Sustainable Yield Data, Crowingstockair- Sustinableyieldair- Leand coer cdss dry t hi' range Median dry thayf'range Median Sourc! 1 Grassland (2-27) - [0.101 Huntley [2.27r - [ 01 n [2.27' - [ p na oaj n227ja.a,. 2 Wooded Grassland (3.3) (0.10) Openshaw 1982; Andeke Lengi 1987 _[33F - [0.101 n.a. 23 1[33f _ [0.10]6 n.a (0.62) - [0.10 Rutherford [33f - [0 n.a. 3 Shrubland (855) _ i50o. Huntley 1978; Rutherford [0.50]e Rutherford (10.00) - 050g. Rutherford o _50a.so na (5.60) - - [0.50 Rutherford Bushiand and Thticket (13.9) - [021]' Huntley (1851) - [0.21]' Rutherford 1978; van Wilgen, Higgins,and Bellstedt (16.92) (0.21) Rutherford (1.41) Daus, Guero, and Ada 1986; Andeke Lengui [18.51] - [0.21]6 nma. 5 Low Woody Biomass MOsait-c Rutherford M 0.63 Stomgaard 1985; rrc(u) 1987; Perssn Woodland (29.40) (0.44) FAo/PNuD 1978; Biandhi 1986; Andeke Lerngu (26.70) (0.77) 1Ao1984; FAo/PNTD 1978; C16ment [26.70]' ]' na [26.70]' (0.46) Biandtii986;aement [26.701' (0.48) Cl4ment Z (55-63) (0.89) Persson 1975; Jadckson 1971; ETc (uj4 1987; Guerreira 1966; Chidumayo (119.0) (133) Guy 1970, 1981; Stomgaard- 1985; Lundgren 1975; ETCQUK) 1989; Persson 1975; Kennard andwalker1973;guerreiro High Woody Bwiass Mosaic (5.13) - [0o191 Malleaux [019] a Bianchi [011 na. [k [16.84]- - [0.19]1 ta [684] - [019' ila [16.84)k - [0.19)' na. (Tablecontinues on thefolltoiing page)

37 22 Estimating Woody Bioniass itn Sub-Salarmn Africa Table 4-1 (continu-ed) Growingstockair-- Sustainableyield air- Land cover class d ry li" ruange Median dry ti4 ye ' ran Medianu Soterc 8 Forest A9 Christiansen 1978; Golley and others [4.98]" Baines Christiansen Chapman and White 1970; nrc (U.K) (127.78) [14.8r Pecha 1986; FAO/IfUD 1978; Marsch 1978; Bianchi 1986; Greenland and Kowal 1960; Nye 1961; John [127.78]m - [14.80r (104.89) (14.80) Rdpublique du Togo 1987; Bartholomew, Meyer, and Landelot Not available. n.a. Not applicable. o Median value. [I Value to use in calculations where no data exisl a. Derived from values for Class 11; b. Derived from values for Cass 21; c. Derived from values for Class 21; d. Derived from values for Class 33; e. Derived from data given for differences and stands of 33/34 ibsas heath) in Rutherford (1978) and translated to Classes 31, 32, 34, and 35; f. Derived from values for Class 42; 8; Derived from values for Class 43; h. Derived frorm values for Class 52; i. Derived from values for Class 62 j. Derived from values for Class 61; k. Derived from values for Class 73; 1. Derived from values forclass 73; n. Derived from values for Class 85; rl Derived from values for Class 83; o. Derived fom values for Class 87. available in the summary classes of Grassland (1), Furthermore, only two of the BushLad and Thicket Wooded Grassland (2), Low Woody Biomass Mosaic classes had any data on sustainable yield- (5), and High Woody Biomass Mosaic (7).-The estimates for the Forest classes are usable in the The values for all Grassland classes (11-14) are data base, although four of the seven classes had only based on a sample of estimates for growing stock for one value for growing stock or sustainable yield. Of Veld Grassland (11). The several studies quoted by particular importance are the estimates for Mangrove Rutherford (1978) on grasslands in southern Africa are forests (81), which are very small because they came mainly concerned with herbaceous biomass. No val- from ecologically constrained areas such as Somalia ues for sustainable yield were found for these classes and Guinea-Bissau. These low values dearly do not and therefore values have been inferred from data represent the bulkof mangrove forests in Africa, so we gathered for wooded grassland systems. In practice, decided to use the value of tonnes per hectare given the lesserbiomass present in the grasslands, the with an equivalent sustainable yield value of sustainable yield would probably be lower than that tonnes per hectare per year. These values come from forw6odedgrasslands.asimilarlackofdataexistsfor estimates of mangrove stocks and productivity in these wooded grassland systems, however, and be- Puerto Rico (Golley and others 1969) and Thailand cause these classes have low woody biomass compo- (Christiansen 1978). nents, they are likely to become areas of critical wood Few quantitative data were available on the woody supply if demand should increase. biomass stocks and yields for the Low and High For Shrubland classes (31-35), Bushland and Woody Biomass Mosaics (5,7). Of the two, more data Thicket (41-45), and Woodland (61-67), the levels of exist for Low Woody Biomass Mosaics, mainly bedata available are generally better. Nevertheless, data cause -hey are most widespread in southern Africa on growing stock are notably absent for some import- where significantly more ecological studies have been ant woodland classes (63-65) in East Africa and the conducted and estimates have been made of woody SaheL In addition, far fewer data are available on biomass for planning fuelwood energy. Unfortusustainable yield for Shrubland and for Bushland and nately, the same cannot be said of the High Woody Thicket classes than are available for the equivalent Biomass Mosaics anywhere in Africa. growing stocks. Only one value of sustainable yield We recognize that the ratio between growing stock for Shrubland was obtained from the literature, and and sustainable yield shows a level of inconsistency that had to be estimated from studies of stands of that can be attributed to the use of median (central) varying age of bumedfynbos heath (Rutherford 1978)- figures from different studies.

38 Woody Biomass Assement 23- Area Calculations of Land Cover Class cover class or woody biomass and consequently are omitted from the tables. A map of nationalboundaries for Sub-Saharan Africa 2. Some countries had significant cloud that afwas digitized and warped to the land cover class map fected the classification of land cover (table 3-1). These derived from the supervised classification of AVEIR areas were masked from the final imagery (Chapter 3) imagery. In addition, the outlines of the main lakes and are not recorded in the tables. and reservoirs in Sub-Saharan Africa were digitized 3. Supervised classification establishes a cass of and warped to the cg imagery. This enabled the maps pixels which were rejected because they were outside of the land cover classes and summary classes (see the statistical limits of the land cover classes. These Chapter 3) to be overlaid with national boundaries, pixels were not assigned to any land cover class and lakes, and reservoirs. constitute a further proportion of the area of each A program was written to count the pixels in each country that is not mapped or used in the woody land cover class within each country across the entire biomass database. image. This produced a table of pixel counts by land. cover class for all Sub-Saharan African countries. The Interfacng Area Growing Stock and Data table was imported to a spreadsheet and used to cal- on Sustainable Yield culate growing stocks and sstainable yields. The area of each country was calculated by sum- Data on biomass growing stock and sustainable yield, ming its pixels and multiplying them by the area of assigned to land cover classes, were transferred into each pixel. This was possible because the map projec- the spreadsheet directly from the data b-se. Macros tion used to display the Gn data, the Hammer-Aitoff were written within the spreadsheet to calculate the X projection, is an equal area projection, and therefore following information from the interfaced data base each pixel occupies the same area across the entire files: image. The areas thus calculated were compared to areas of each country taken fromstandardsources (for * Growing stock by biomnass class and summary bioexample, Stonehouse 1985). In no instance were the mnass dass differences between pixel-derived areas and those * Sustainable yield by biomass class and summary given in.the literature greater than 1 percent of the biomass class. country area. These small differences are easily explained by the The growing stock and sustainable yield. for each raster nature of the international boundaries and land cover class were used to produce maps for all of coastlines from the overlays on the remotely sensed Sub-SaharanAfrica.Mappingwasachievedbyexamimagery, which depict "smooth" curves and lines on ining the data on stock and yield (table 4-1) and idenmaps as steplike boundaries. Note, however, that the tifying natural breaks m both, which defined five summed areas of land cover classes for most countries classes of stock and yield. Eadh land cover class in the are less than the recorded area of the countries. These final supervised classification image was assigned to discrepancies arise for three reasons: a color-coded stock class and a yield class to produce 1. The area of lakes and reservoirs is included in the mapsof growingstock(figure 7-1) aid sustainable the total country areas, but these have no data for land yield (figure 7-2).

39 5~~~~~~~~~~~ Description of Biomass Classes On a continental scale, the vegetation of Sub-Saharan In southem.africa, tie Namib Desert stretches Africa forms fairly distinct bands that span the conti- along the west coast from southern Angola through nent latitudinally, although this zonation disinte- Namibia and into South Africa. The other significant grates in the extensive highlands associated with the desert areas are. parts of the Kalahari Desert in Bo- Rift Valley of East Africa, especially in Ethiopia. The tswana. In total, the Desert class covers nearly 4.1 map of growing stock (figure 7-1) shows a similar million square kilometers in Sub-Shran Africa. De- * pattern, with the greatest density of woody biomass. spite the dwarf shrubland on the fringes of the Namib centered on the Guineo-Congolian rain forest belt, Desert andafewwidelyscattered treesinthekalahari which extends through much of coastal West Africa Desert of southwestern Botswana, the overall contriand across northem Zaire as far as Uganda. To the bution of Africa's desert areas to growing stock is south, biomass stocks decrease steadily from the ex- negligible. tensivemiombowoodlandsbetweenabout5%sandls 0 S 11 Veld Grassland covers only small areas over to the open woodlands, shrublands, and eventually much of Africa, but it is extensive in southern Africa, the veld grasslands and deserts of southem Afica coverng nearly 6 percent. The largest expanse covers North of the rain forest belt, a similar decrease in most of Lesotho and much of the Orange Free State growing stock occurs through rather narrower zones and southern Cape Province. Outliers are found of Guinean and Sudanian woodland, dry Acaci- throughout northwestern Botswana and around the Comnm4km bushand and thicket, and the semidesert northeastern border of Angola. Woody biomass is wooded grassland, which eventually merges into the largely restncted to riparian woodlands and to small, Sahara Desert at about 200 N. East of the Rift high- widelyspacedtreesinthekalaiarithomveldofnorthlands,muchofkenyaandthehomarecoveredbydry western Botswana. The total contribution of Veld bushland and semidesert vegetation, giving way to Grassland to growing stock in Africa is negligible. desertincoastalethiopia, Djibouti,andpartsof north- 21 Semidesert Wooded Grassland is most exten-. em Somalia, and to a lesser extent in northern Kenya. sive along the southern edge of the Sahara Desert, In total, Sub-Saharan Afica contains an estimated stretching eastward from southern Mauritania 65,689 million tonnes of growing stock, producing a through the central areas of Mali, Niger, Chad, and sustainable yield estimated at 3,549 -million tonnes a Sudan.ThesouthernmostareasareinnorthernKenya, year-. and the class is also important.in Somalia and the The following paragraphs summarize the growing eastern half of Ethiopia. The vegetation is grassland stock and sustainable yield for the most significant with bushes and small bushy trees (White 1983). The land cover classes listed in table 3-2. grassland is domninated by annual species, and the 0 Desert dominates the northern portion of the cover is ephemeral except where shaded by larger study area, with the Sahara Desert. extending into trees. large areas of Sudan, Chad, Niger, Mal, and covering ThewoodycomponentislargelyofAcaciaspp.,with most of Mauritania. Desert also covers coastal regions a crown cover of less than 10 percent Woody biomass of the Horn as far south as northern Somalia Apart stocks are low-for example, Semidesert Wooded from a few woody species at the southem fringe of the Grassland covers more than 18 percent of the West Sahara,. most biomass is in the root systems of African Sahel region but provides less than 6 percent drought-resistant plants- of the region's growing stocl Population density is.24

40 Dcriplioi of Blomans Clasn 25 low, increasing toward the south in the belt. ln Soma- tatioin, with scattered trees on the dune crests and lia, however, the semidesert area supports more than shrubs restricted to the troughs. In the hills bordering 40 percent of th-e population, and their demand for the Namib Desert, tall succulents and bushy trees up fuelwood has caused severe stress on the- trees to 5 meters are scattered in an open shrub layer of (Kamweti 1984). Semidesert Wooded Grassland has a about 2 meters height. (See Class 35, Wooded Shrubgrowing stock of nearly 364 million tonnes, with a land, for growing stock and sustainable yield statiscs.) sustainable yield of less than 14 million tonnes. The 35 Wooded Shrubland is similar to Kalahari severe social consequences of depleting this resource Shrubland (34). Although its distribution is similar to make sustainable management vital. that of Kalahari Shrubland, it occurs in slightly moister 24 TransitionalWooded Grassland covers ll per- areas, often at hiigher altitudes. Each of these classes cent of southern Africa, including 40 percent of Bo- has a growing stock of less than 150 million tonnes, tswana, 20 percent of Namibia, and areas of northern although the sustainable yield of more than 30 million and central South Africa which add up to 14 percent tonnes for Kalahari Shnaubland compares to only 2.5 of that country. Overmuch of this area, the vegetation million tonnes for Wooded Shrubland. In such barren is similar to that of the Veld Grassland (11). In parts of surroundings, these classes represent a relatively poor Botswan and the Orange Free State, this class consists source of fuelwood, and the present rate of fuelwood of grassland dominated by Aristida, with a number of extraction easily exceeds annual production. Karroid shrub comnmunitiesup to 5 meters in height. In 41 Dry Acacia-Commiphora Bushland and Thicket the Transvaal and southern Mozambique, this class occurs in the countries of East Africa and the Horn, includes Zambezian woodland donmnated by Acacia extending over much of southeastern Ethiopia and in a mosaic with veld grassland. inland areas of southem Somalia and eastem Kenya. Despite its large area, Athe Transitional Wooded Significant outliers are found in central and north- Grassland class includes only 1.23 percent of the grow- westem Kenya, northem Tanzania, northeastern ingstockfor southern Africa. Nonetheless, these areas Uganda, and the extreme southeastem area of Sudan. are heavily cultivated and grazed. Because the major- The bushland is dense, often forming impenetrable itl of biomass is nonwoody, and access is restricted thickets attaining 5 meters in height, with scattered because of land tenure, Transitional Wooded Grass- emergent trees up to 10 meters. With a growing stock land is a poor source of fuelwood. This class has an of nearly 944 million tonnes, this class produces a estimated grwing stock of nearly 146 million tonnes, sustainable yield of only 10 million tonnes. Overproducing a sustainable yield of about 5.6 million population and overgrazing have severely degraded tonnes. large areas of the bushland, often leading to 33 Bushy Shrubland occurs mainly in East Africa desertification around boreholes. The considerable and southern Africa, covering 1.8 percent and 1.96 demand for fuelwood exacerbates the problem. percent respectively. In Kenya, this class consists 43 Moist Acacia-Commiphora Bushland and mainly of small bushes and stunted trees, dominated Thicket is spread in patches across southern and East by Acaia reficiens ss. misera, although it also accounts Africa in a roughly arc-shaped distribution. The most for areas of montane vegetation in the eastern northerly, large area extends down the coastal strip of Aberdare Mountains. In southem Africa, the largest East Africa from about 40 N in Somalia to the south areas are found along the west coast of Cape Province coast of Kenya. A large triangular area of moist bushand on the northern fringe of the Namib Desert along land covers much of central Tanzania as far as Lake the coast of Angola. I-n Cape Province, the vegetation Malawi, and the arc continues through Malawi and is sderophyllous shrubland, with a few taller bushes parts of Mozambique and Zambia, eventually reachbut very few trees. In coastal Angola, this class con- ing the coast of southern Angola Large outlying areas tains a rather open vegetation of small shrubs, with are found in the south of Mozambique and especially occasionaltaawa-conmniphorm shrub communities. in Swaziland and the Transvaal. Once again, grass forms a large proportion of the In East Africa, this class is principally dense bushbiomass, and wood production is low. From a total land up to 7 meters height, with grass cover growing growing stock estimated at 96.7 million tonnes, Bushy to 1.5 meters. hi southern Africa, however, this class Shrublandproduces asustainableyieldof lessthan 1.6 spans a wide range of floristic regions. The major nillion tonnes. Thus, the potentialfor large-scale, sus- similarity between these areas is in their snsonal phetainable exploitation of fuelwood is extremely limited. nology. Productivity peaks in the summer, sometime 34 Kalahari Shrubland covers 4 percent of south- between December and May, depending on the area em Africa, including 10 percent of South Africa, 9 of Africa examined. A decline sets in from May, with percent of Botswana, and 8 percent of Namibia. In the lowest productivity occurring toward September, Botswana, Namibia, andnorthern Cape Prrovince, this the end of the dry season. After this, production inclass consists argely of rather sparse sand dune vege- creases steadily. '9..

41 26 Estimating Woody Bioniass in Suib-Saharam Africa The moist bushland is rather more productive than 61 Open Woodland is scattered among the miombo the Dry Acacia-Comnmiphora Bushland andthicket(41), woodland of Tanzania, where its composition is simhaving a sustainable yield of nearly 77 million tonnes ilar to that of the surrounding vegetation, and in on 1,685 million tonnes of growingstock. Nonetheless, Burundi, where the class is represented by moist Acathe pronounced dry season ensures that productivity cia-commiphora bushland and Acaia wooded saremains relatively low, limiting the potential for fuel- vanna. The vast majority of this class, however, lies in wood exploitation. the southem African region. Here it covers much of 44 Sahel-Sudanian Acacia Wooded Bushland southeastem Aingola and northevstern Namibia, forms a belt across North Africa centered on about northem Botswana, southweste,. smbia, much of 1? N, from Senegal in the west to central Sudan and Zimbabwe, and parts of southern Mozambique. This northern Ethiopia in the east. This belt lies between forms an irregular belt running latitudinally between Semidesert 1l ooded Grassland (21) to the north and the miombo woodland to the north and shrubland to Dry Sudanian Woodland (62) to the south. Sahel- the south. Open Woodland encompasses a variety of SudanianAcacia Wooded Bushland is generally found vegetation, including woodlands dominated by in areas receiving between 250 and 500 millimeters of Brachystegia spp., Bakiaea spp., and Burkea africana. rainfall a year. There is a herbaceous layer of mainly The growing stock is estimated at 743 million annual grasses, although perennial species are more tonnes, producing a sustainable yield of nearly 63 common toward the south of the zone. The ground million tonnes. Despite being an important source of layer is more persistent than that of semidesert, sur- fuelwood in South Africa and Lesotho, the low pro viving most of the dry season. The main woody spe- ductivity of this class makes hiigh rates of sustainable des are Acacia, increasing in height from 4 meters in exploitation impossible in most areas. the north to 8 meters in the south. 62 Dry Sudanian Woodland lies in a latitudinal The growing stock of nearly 4,203 million tonnes is belt immediately south of the Sahel-Sudanian Acacia the third largest of any class and produces a sustain- Wooded Bushland (44) previously described. From able yield exceeding 178 million tonnes. Processes west to east, it occurs in southem Senegal and the such as agricultural and pastoral intensification (Gra- Gambia; southern areas of Mali and Burkina Faso; ham 1969) and overgrazing around tube wells have northern Ghana, Togo, and Benin; northem Nigeria, destroyed a great deal of woody biomass. Careful Cameroon, and Central African Republic; southem management is needed if this class is to supply a parts of Chad and Sudan; and in northern Ethiopia. sustainable yield of fuelwood. The vegetation is mainly a rather open tree savanna of 51 Acacia Woodland Mosaic occurs mostly in 15to25metersinheight,orshrubsavanna.BothAcada southern Africa, where it covers large areas of eastern and Combretum species are common. A ground layer Botswana'and-the Namibian interior, as well as north- of annual and perennial grasses exists, which burns ern Transvaal and the fringes of South Africa's veld easily during the dry.season. grasslands. Extensive areas are also, found in the The growing stock for this class is less than 1,050 southern regions of Mozambique and Zimbabwe. million tonnes, with a sustainable yield exceeding 213 Generally this class consists of rather open woodland million tonnes. Agriculture and grazing have dewith trees up to 10 meters height, most commonly graded large areas of woodland. Fuelwood collection Acacia, Commiphora, Combrefum, and Terninalia spe- for cities such as Chartoum has left considerable areas cies. These woodlands have frequently undergone almost treeless (Lewis and Berry 1988). The current scrub invasion. fuelwood shortages and associated environmental E In southem Africa this cluss is dominated by agri- degradation will almost certainly become worse at culture. Pastoralism has destroyed much of the woody present rates of extraction. biomass and ranching hasrestricted access. Productiv- 64 Sudanian Woodland forms another latitudinal ity is fairly low, with 35 million tonnes of sustainable belt across Africa, lying immediately to the south of yield from nearly 1,100 million tonnes of growing the Dry Sudanian Woodland (62). The belt is rather stock, so once again, exploitation requires management. scattered, but broadest in the region of the West Afr- 52 East African Low Woody Biomass Mosaic oc- can coast From the west, this cdass covers areas of curs mostly in a belt from northeastern Uganda to the Guinea-Bissau, Guinea, Mali, Cote d'ivoire, Burlina highland region of southwestern Kenya, although Faso, Ghana, Togo, Benin, Nigeria, Cameroon, southsmall outliers exist in the countries of the Hom. Much emn Chad, northern Central African Republic, southof this vegetation is Combretum small tree savanna, em Sudan, and small areas in the highlands of Ethiobecoming evergreen and semigreen bushland in the pia. This class corresponds to areas that White (1983) Kenyan highlands. With a growing stock estimated at classified as "Sudanian woodland with abundant - 399milliontonnes,producingnearly32milliontonnes Isoberlinia" and a mosaic of lowland rain forest and of sustainable yield, this class is locally important secondary grassland. Much of the area mnay be de-

42 Description of Biomass Classes 27 scribed as derived savanna, ranging from closed can- shrubs tend to Iorm relatively dense stands separated opy savanna woodlands to thicket savannas. by grasslands or herbaceous wooded savanna Sudanian woodland contains a growing stock of (Laclav&re 1980). nearly 1,184 million tonnes, producing a sustainable Guinean woodland has a growing stock estimated yield exceeding 391 million tonnes, the second highest at nearly 3,123 million tonnes, greater than all but four of all the classes. Disturbance has been low and the of the other classes, producing a sustainable yield of prospects for sustainable fuelwood extraction at pres- 256 million tonnes. ent rates are good. 76 Medium-Productivity West African Cultiva- 65 Moist Sudanian Woodland is scattered along tion and Forest Mosaic is the most extensive land the. south of the Sudanian woodland belt in coastal cover class in coastal West Africa. It occurs in every West Africa, spreading from central Guinea into cen- country of the region and forms a broad band from tral Africa. The largest area of this class straddles the Guinea-Bissau to southeastern Nigeria. This class conborder between Sudan and Central African Republic. tains renants of semideciduous humid tropical for- The zone ends in northern Uganda. Vegetation is gen- est in a mosaic with areas cleared for agriculture and erally an open woodland savanna with trees attaining timber. The mosaic typically contains both rain forest 15 meters in height. The most usual domiant tree and savanna species at its northern limits, where it species is Isoberlinia doka, and an understory of shrubs may represent an ecological transition from humid sometimes is present Growing stock is 1,887 million tropical forest and Guinean Woodland (74). tonnes, with a sustainable yield exceeding 282million Generally, forest and woodland dominate in areas tonnes. of low population density and clearance is more ex- 66 SeasonalMiomboWoodlandcovershugeareas tensive in heavily populated regions, especaally of central and East Africa between about 15 S and the around towns and cities. The mosaic has the fourthequator. It is most extensive in northern Angola, largest growing stock of any class,- exceeding 3,641 southern Zaire, Zambia, Zimbabwe, Tanzania, million tonnes, coupled with the largest sustainable -Malawi, and Mozambique, and extends along the In- yield, nearly 561 million tonnes. Ongoing forest deardian Ocean coast from Beira in southern Mozambique ance makes this class an important supplier of fuelto Port Elizabeth on the southern Cape. This class wood both locally and to deficit areas. covers a wide range of woodland types, the most 77 Highland Cultivation Mosaic is confined to common including Brachystegia, Julberardia, and Ethiopia, of which it covers about 17 percent. Found Isoberlinia woodland, often to 20 meters height, and on plateau areas between 2,000 and 3,000 meters elefrequently with a woody understory. vation, it is the result of thousands of years of defores- Seasonal miombo contains the largest growing stock tation due to fuel collectin& grazing, and cultivation. of any class, exceeding 6,836 million tonres. Sustain- Large areas are now treeless, although land uses range able yield is nearly 138 million tonnes. Thisrepresents widely from dense cultivation to tree plantations, as a considerable potential for fuelwood exploitation, well as remnants of forest and woodland. Total grow- 67 Wet Miombo. Woodland is most extensive in ing stock is less than 1,350 million tonnes, but with a souther Lire, northern and centralangola,northern sustainable yield less than 15 million tonnes. The region Z:ambia, southeastern Tanzania, and central Mozam- has long suffered from critical fuelwood shortages, and bique. Generally found in wetter areas with a lower it appears that these will continue and worserl seasonalitythanseasonalmiombowoodland(66),wet 81 Mangrove grows extensively on the coasts of Miombo Woodland has a similar species composition, East, central, and West Africa, especially around river but with rather dense thickets that often form the estuaries. Local pressure on mangrove swamps is inunderstory. Wet Miombo Woodland has a growing tense in many areas. Mangrove wood is excellent fuel stock of nearly 2,540 million tonnes, producing a sus- and is used for smoking fish, so many mangrove tainableyieldofmorethan43mnilliontonnes.withfew stands are heavily exploited and depleted, despite exceptions, this class represents an important fuel- general protection by law. The inability of the satellite wood resource in the countries where it occurs. images to resolve very narrow fringes of mangrove 74 Guinean Woodland has a similar distribution swamp, notably on the East African coast, means that to the Moist Sudanian Woodland (65) already de- the figures for growing.stock (193 million tonnes) and scribed. It is most extensive in three main areas: (1) in sustainable yield (68 million tonnes) are almost cercoastal West Africa, where it occurs in every country tainly underestimates.- except Liberia; (2) in Central African Republic and 82 Evergreen Forest mainly represents montane adjoinng areas of Caimeroon, northern Zaire, and forest vegetation. It occurs in a number of highland southen Chad; and (3) in the border region of south- areas, including the mountains of southem Tanzania em Sudan, northwestern Uganda, and northeastern and eastern Transvaal in South Africa, as well as the Zaire. Rainfall is fairly high in these areas, so trees and rift escarpment in eastern Zaire. These areas typically

43 28 Estimating Woody Biomass in Sub-Saharan Africa experience high rainfall and humidity and encompass and constitutes a growing stock of about 1,540 million a wide range of both montane and lowland rain forest tomes. Sustainable yield is a little more than 39 milspecies. The growing stockfor theclass is l,310million lion tonnes. tonnes, wit a sustainable yield of more than 90 mit- 87 Ombrophilous HumidTropical Forest extends lion tonnes. Clearance has occurred in some areas, along the equator, covering vast areas in Zaire, Congo, with only the steepest slopes remaining forested. Gabon, Cameroon, Central African Republic, and Rio 85 Mesophilous Humid Tropical Forest is most MunLThereislittleseasonality,withthemeanannual extensive on the northem fringe of the equatorial and rainfall of 1,500 to 2,100 millimeters distributed fairly Ombrophilous Humid Tropical Forest (87), where it evenly through the year, and productivity is consiscovers large areas of Cameroon, Rio Muni, Gabon, tently high. The forest has a tall, closed canopy which Congo,CentralAfricanRepublic,andespeciallyZaire. may reach 45 meters in height. Most species are ever- South of the equatorial forest, this class occurs in a green, although a few deciduous species shed their broad swath along the west coast from Cabinda to leaves during the brief dry season. southern Gabon The mesophilous forests are rather The ombrophilous forests have a growing stock of more seasonal than the equatorial forests, showing a nearly 5,932 million tonnes, second only to the much more pronounced dry season, and correspond partly more extensiveseasonalmiombo Woodland (66) to the to the drier Guineo-Congolian rain forest of White south, and a sustainable yield exceeding 150 million (1983). The range of species is wide, as is te degree of tonnes. Much of the remaining forest area throughout species domnmance. Thlis class contains about 2,850 Africa is protected within reserves. Much of the presmillion tonnes of growing stock, producing a sustain- ent equatorial forest area has been cultivated in the able yield of more than 72 million tornes. past, and the vegetation is secondary. Within some 86 HumidTropicalSwampForestformsanarrow reserves, older secondary forest may be indistinstrip near the equator, with areas around the western guishable from primary forest This enormous bioshores of Lake Victoria in Tanzania and Uganda, and mass resource has been largely removed from pdtenin Zaire, Congo, southern Cameroon, and Gabon.- tial fuelwood reserves in the interest of science and Swamp forest covers up to 6 percent of central Africa tourism-

44 6 Regional Summaries by Class of Biomass This chapter summanzes physical characteristics, Physical Characteristics, Population, and Resources population, resources, and the condition of woody biomnass for each of the six regions examined in tis Rainfall in the region is markedly seasonal, with the study. For each region, a table presets (a) the esti-. humid season peaking in July and August and the mated surface area and percentage of the nine sum- principal biological production occung soon after. mary dasses of land cover, (b) the growing stock by In the semudesert regions, annual rainfall is only 100 summary land cover class, and (c) the sustainable to500nmillimetersayear.evaporationrateisrapidand yield by summary land cover class. For a more de- water-holding capacity of the soils is small In the tailed review of the distribution and character of the south of the region, annual rainfall may exceed 1,000 woody biomass in each region, please refer to Part II millimeters. This, combined with high summer temwhere the 43 land cover dasses are discussed. peratures, produces favorable conditions for photosynthetic activity and resultsi m the marked seasonal- The West African Sahel ity of the biomass. The north of the region is subject to periodic fluc- For the purpose of this study, the West African Sahel tuations of climate, of which the great drought of is bounded by the latitudes of IUD N to 2;3 N and the was an extreme. This significantly affects longitudes of 18" W to 24e B. The region includes the water availability. As a consequence, a mosaic of seven countries of Mauritania, Senegal, the Gambia, vegetation patterns exists, reflecting areas in which Burkina Faso, Mali, Niger, and Chad. The area is de- land is stressed. ImiLted to the north by the Sahara Desert, and all In the nortiem countries, distribution of the sedencountres have significant desert territory except.the tary human population is restricted to more produc- Gambia, Senegal, and Burkina Paso. five areas having greater and more predictable rain- Table 6-1 Distribution of Summary Classes and Estimated VWoody Biomass, West African Sahel Arna rozuingstock Sustainableyield- Million Thousand Summary cl kn? Perent tonnes Percent tinesperyear Percent 0 Desert 2,590, Grassland 40, Wooded Grassland 981, , Shrubland Bushland and Thicket 879, , Low Woody Biomass Mosaic Woodland , , High Woody Biomas Mosaic 8, Forest 2, Lakes 18, Total 5, , ,74& Noht Details may not add to totals because of rounding Saun2 Tables 8-1 to

45 30 Esimintting Woody Biomnis in Sub-Saharan Africa fall. Land use in the semidesert regions is largely Resources Institute 1986). (Ihe region doesnotexhibit pastoral, although significant permanent populations the rapid growth rates of other parts of Africa, such as exist along the Senegal and Niger rivers and their Kenya.) This increase, coupled with greater than 40 associated irrigation schemes. According to the PAO, percent of the population being 14 years of age or population density varies throughout the region, with younger, means that demographic pressure can only 1985 statistics showing 1.8 persons per sqqare kilome- increase. The demand for greater space for food proterinmauritania,contrasting with33.2insenegaland duction results in degradation of land already ex- 56 to 59 in the Gambia (World Resources Institute ploited to its full potential. This can be seen in areas of 1986). Tlhese mean figures, however, cannot truly re- Senegal or in extensive migration like that occurring flect the possible demographic stress caused by the in Burkina Faso. population distribution. Discussion with the director of the Institut de Re- Relief and soil dearly have implications for biomass cherche en Biologie et Ecologie Tropicale (imstr), based production. The plateaus of the southern Sahara and in Burkina Faso, confinns that the migration is particthe Manding region of Guinea and Mali are reflected ularly significant. Areas of Sudanian woodland that in the classification produced from the imagery. Red- previously provided a more-than-adequate fuelwood brown soils of the south are endowed with greater resource now are being cleared for agriculture. capability of retaining water, contrasting with desert Research within the Sahelian zone by the Universoils and sands farther north. sity of Kano in northem Nigeria suggests three man- A review of the classes shown in table 6-1 discloses agement strategies for sustainable production of that Desert (Class 0) occupies 49 percent of the region, fuelwood (Cline-Cole and others 1987): (a) farmed including large areas of Mauritania, Mali, Niger, and parkland in which trees are retained, (b) shrubland Chad. Grassland (1) covers less than 1 percent but or fallow, and (c) forest reserves. This work also Wooded Grassland (2) covers 19 percent as a broad notes the diversity of use and calorific values of diflatitudinal belt. No Shrubland (3) is recognized in this ferent species. region but Bushland and Thicket (4) covers 17percent. Accessibility to fuelwood in the most densely pop- Woodiland (6) covers 14 percent of the region and ulatedareasvariesthroughtheregioln.wheretreesare locally is important at the southem edge adjacent to retained to a significant extent, management is possithe West African Coast region. Low Woody Biomass ble that recognizes diverse use for food and fodder, Mosaic (5), High Woody Biomass Mosaic (7), and constmction, pharmaceuticals, heating, and cooking. Forest (8) are almost completely absent from the West An assessment of woody biomass is not necessarily an Afican Sahel region. assessment of fuelwood, although all trees become fuelwood if no alternative exists. Condition of Woody Biomass A significant contribution to growing stock comes from the Bushland and Thicket class, of which the Assessment of woody biomass in the West African dominant type is Saahel-Sudanian Acaca Wooded Sahel is problematic because of several difficulties in Bushland (44). It accounts for more than 130 million interpreting Normalized Difference Vegetation Index tonnes of growing stock and a sustainable yield of (NDvi) values in relation to biological activity. One 30 million tonnes, about 30 percent of the regional problem is that we cannot distinguish between herba- total (table 6-1). The Sudanian Woodlands (62, 64, ceous and woody biomass. In a region like the Sahel, and 65) are highly significant, at more than 1,900 which includes areas of limited tree cover and vast million tonnes of growing stock and 50 million areas of seasonal grassland, this is obviously signifi- tonnes sustainable yield. This group is most vulnercant For example, in a study of the regeneration of able to the present population migration in the degraded Sudanian savanna during a 10-year period, south of the region. Bonkoungou, Bortoli, and Oudba (1988) show that the It should be remembered- that estimates for this grass Andropogongayanus forms 50 percent of the bio- region are particularly crude because of the paucity of mass. Another problem is our inability to distinguish data for herbaceous biomass and noncommercial between seasonal crops and natural wooded grass- wood. The Sudanian Woodlands in particular merit land, which produce similar phenologies. closer, coordinated study, with a more intensive local Human population density is much greater in the examination of woody biomass estimates linked to a southemportion of all the countries, exacerbated both finer resolution of remotely sensed imagery. Much by desertification andpopulation growth UnitedNa- work Is under way within organizations such as mnr, tions figures show that all Sahelian countries except the Centre Technique Forestier. Tropicale (arr), and the Gambia have a population growth rate exceeding the Institut de Recherche Agronomique du Niger 2 percent, and all rates are projected to increase (World (INRn).

46 Regvionil Summaries 31 The West African Coast Condition of Woody Biomnass Tfhe West African coastal region includes all of the The West African coastal states generally have adecountries bordering the. Atlantic Ocean from Guinea- quate woody biomass, particularly for fuelwood and Bissau to Nigeria. Strong ecological and agricultural charcoal Two countries, Liberia and Sierra Leone, arguments exist for this division. First, this region have only local problems of supply, due either to contains the humid tropical forest zone of West Africa dense population concentration, edaphic restriction and wet (Guinea) savannia woodland. Second, cultiva- on tree growth, or land reservation. tion in the southern two-thirds of the region is donmi- The other countries all have a general decline northnated by root and tuber crops, rice of both upland and ward in growing stock and sustainable yield. This swamp varieties, and plantation tree crops such as pattern results from the latitudinal disposition of land cocoa, coffee, rubber, and oil palm. Grazing systems cover classes across the region. The high-productivity and'dryland crops are restricted mainly to the north- Forest and H-igh Woody Biomass Mosaic classes are ermmost parts of the region. adjacent to the coast, the wet (Guinea) savanna woodland is farther inldand, and the drier (Sudarnian) sa- Physical Characteristics, Population, and Resources vanina woodland thrives in the ntorthern part of the region. This pattern is a function of the mean annual Estimated woodfuel consumption in 1990 was ap- rainfall and the length of the wet season. proximiately 81 million air-dried tomnes (table 1-2). TIhe majority of growing stock and sustainable yield. Sustainble yield significantly exceeds this; therefore is in the High Woody Biomass Mosaic adjacent to the the region has a positive overall balance of woody coast, the Woodland, and the highly significant Manbiomass supply. Two problems exist, however, with grove class. Repeated cutting of the wood in the Hfigh both the actual s'upply and the potential supply. First, Woody BiLomass Mosaic means that many trees and the northern part of the zone is dominated by dry shrubs continue to grow rapidly. To counter this rapid savannia woodland, from3 which it is difficult to trans- annual growth, however, forest reservation and planport wood to large cities such as KCano and Sokoto. tations probably restrict access to land more for two Second, muc-h wood from the Sudanian woodland is clamsse than for any others in the region: H-igh-Producconverted into charcoal and transported to the large tivity Wet African Cultivation and Forest Mosaic (75) coastal cities. and Medium-Productivity West African Cultivation In the wetter areas of higher productivity to the and Forest Mosaic (76). south, significant extraction of timber occurs. Such Guinean Woodland (74) includes 7.6 percent of the forest degradation probably has little negative impact growing stock and 1.8 percent of the sustainable yield. on the wood energy market and in fact may lead to a ApartfromGuinea,where extensive clearance of these local surplus of "waste wood that is sold or freey woodlands has been conducted in the Fouta jallon collected. Of course, local supply problems exist, such region, this zone is the least exploited of the main as those in central Sierra Leone, related to particularly biomnass classes in the region. This situation could dense population concentrations, edaphic factors that change because of increasing population density or restrict tree growth, and reservation of forest and other increasing exploitation of the trees for fuelwood and types of land. charcoal in the Guinean woodland. A review of the classes in table 6-2 shows thatdesert Tfhe Sudanian Woodlands (6Zto 65) encompass only (0) and Shrubland (3) are virtually absent from this 40 percent of the growing stoc-k and 17 percent of the region. Grassland (1) represents less than 2 percent sustainable yield (table 6-2). This suggests a currently and Wooded Grassland (2) less than 1 percent of the hiigh level of exploitation, probably attributable to region's land cover. Bushland and Thicket (4) covers 8 charcoal production for cities along the coast and to percent of the regilon, principally in northern Nigeria. local demand. Comparison of growing stock with the Low Woody Biomass Mosaic (5) is relatively unim-. area covered by these woodlands reveals that they are portant and covers just 1 percent. Woodland (6) is an understocked. Consequently, woody biomass supply important category, covering 40 percent of the region problems undoubtedly.exist in these woodlands, parin a broad belt from Guinea in the west to central ticularly where the population is concantrated and the Nigeria in the east Bordering this zone to the south is cultivation is intensive. the mzost extensive summary class, High Woody Bio- TIhe mostnortherly largebiomass class in thisregion mass Mosaic (7), which represents 44 percent of the is Sahel-Sudanian Acacia Wocoded Bushl-and (44), regional land cover and is present in all countries. which is typically.sahelian. Occurring only in north- Finally, Forest (8) covers only 3 percent of the region, em Nigeria, it represents just 0KS percent of the growmainly in small coastal blocks. ing stock and about 2 percent of the sustainable yield.

47 32 Estimating Woody Biomoss in Sub-Saharan Africa Table 6-2. Distribution of Summary Classes and Estimated Woody Biomass, West African Coast Area Growing stock Sustainableyield Million Thlousand Summaryclass km? Percent tonites Percent tonizes peryear Percent 0 Desert , I Grassland 38, Wooded Grassland 17, Shrubland Bushland and Thicket 167, , Low Woody Biomass Mosaic 18, , Woodland 809, , , High Wody Biomass Mosaic 906, , , Forest 64, , , Lakes 17,284 a.s Total 5,252, , , Note Details may not add to totals because of rounding. Source: Tables 9-9 to This area has dense population concentrations (for mals. This is quite apart from its value in environmenexample, the Kano and Sokoto "close-settled" zones), tal maintenance of soil quality, as shade frees, and as and in these areas fuelwood supply is a significant a habitat for economically valuable wildlife. problem, exacerbatedby an overallshortage of woody biomass.thisclassoffersthepoorestwoodfuelsupply Physical Characteristics, Population, in the region. and Resources The Hom of Africa Woody biomass in this region reflects variations of climate. These are caused by seasonal variations in the Four countries constitute this region: Djibouti, Ethi- position of subtropical high-pressure cells; by the latopia, Sornalia, and Sudan. Although the population is itudinal extent of the region; by variations in altitude, overwhelmingly dependent on subsistence agricul- particularly in Ethiopia; and to a. limited extent by ture, commercial agriculture, or pastoralism, woody higher precipitation, which may be caused by proxinbiomass is vital to everyone in these countries, even ity to the sea. Although temperatures significantly those in urban areas. Woody biomass is essential as a influence the highland areas of Ethiopia and the fuel (it is locally available and relatively inexpensive), higher parts of the Equatoria region in Sudan, the as building and construction matenal, as material for main determinants of vegetation patterns are precipithe construction of equipment, and as browse for ani- tation and seasonal variability. Table 6-3. Distribution of Summary Casses and Estimated Woody Biomass, Hom of Africa Area Growingstock Sustainableyield Millior. Tlhousand Sunmary clss. kit? Percent tonnes Percent lonnes per year Percent 0 Desert 1,115, Grassland 133, , Wooded Grassland 761, , Shrubland o.m 4 Bushland and Thicket 1,053, , Low Woody Biomass Mosaic 106, , Woodland 755, Z , High Woody Biomass Mosaic 363, , Forest 52, , Lakes 4, Total 4,347, , , iO. Notc Details may not add to totals because of rounding. Sourcc Tables 10-1 to 10-4.

48 Regional Summaries 33 Except for deserts and small areas of tropical forest, vegetation in a zone radiating from the wellhead, almost all vegetation classes reflect the pronounced possibly extending several kilometers. This phenomseasonality of precipitation, which includes one or two enon has been noted in central Sudan and in Somalia. dry seasons of differing length. In many of the classes, One unfortunate effect of heavy grazing has been to precipitation variations from year to year may pro- encourage development of unpalatable species. foundly influence plant cover, particularly when Clearance of trees for agriculture, either bush falprolonged drought is exacerbated by heavy human lowing or permanent agriculture, has reduced woody interference. biomass in damper areas of Sudan, Ethiopia, and So- Variation in soil and relief is important in some malia. This may lead to virtual elimination of tree classes of land cover because such variations deter- species unless trees are deliberately incorporated into mine the overall nature of the class. For example, the farming system. In bush-fallowing systems, cervegetation is influenced by the lateritic soils of south- tam desired tree species may be protected, but a reern Sudan, the dark, cracking clays and sands of cen- duced penod of fallow may suppress the successful tral Sudan, and the escarpments of Ethiopia. Within regeneration of trees. each class, variations of soil, relief, and drainage also Commercial timber extraction is selectively demay locally affect biomass. structive of larger straight-boled trees which are A review of the classes (table 6-3) discloses that most common in forested areas. Complete removal Desert (Class 0) occupies about 26 percent of the re- of all large trees may follow in accessible areas. gion, including a significant portion of each country. Areas of higher population also may suffer loss of Grassland (1) covers 3 percent of the region and is trees and bushes through the extension of agriculimportant mainly in Sudan. Wooded Grassland (2) of ture; this is notable in areas of coffee plantation in various types occupies nearly 18 percent of the region southwestem Ethiopia. and is present in eadh country, but is most extersive Buning of grassland as a method of encouraging insudan. Shrubland (3) isnearly absenl Bushland and grass growth is widespread inwooded savanna grass- Thicket (4) occupies 24 percent of the area and is lands. Consequently, trees and bushes that are not fire important in Sudan, Ethiopia, and Somalia. resistant are ldlled off, whereas resistant species are Low Woody Biomass Mosaic (5) is confined to the encouraged. extreme south of the region and covers just 2 percent. Fuelwood collection is the other main cause of tree Woodland (6) covers 17 percent of the area and is most cover loss. It occurs in rural areas, but may be greatly important in Sudan and Ethiopia. High Woody Bio- increased by urban demand for wood or charcoal 1 mass Mosaic (7) covers 8 percent of the region and is whichmaybe transportedhundreds of kilometers (for most important in Ethiopia. Forest (8) covers only 1 details, see the discussion of Class 22 in Chapter 10) percent of the region, largely in Ethiopia; it is signifi- Unlessadelhberatepolicyof managementof fuelwood cant because it may be associated with high biomass plantations is introduced near towns, tree removal potential- spreads outward, particularly alongmotorable roads. This has been done in Ethiopia, where Eucalyptushas Condition of Woody Biomass been planted for almost a century, and in parts of Somalia. For some decades throughout the region, concern has Kamweti (1984) has investigated the probable effect grown among both experts and laypersons about the of fuelwood collection up to the year 2000 in the East condition of natural biomass, both woody and non- African region, whidh inmludes Ethiopia and Somalia. woody. In the drier regions of each country, biomass Most of Ethiopia and Somalia already suffer fuelwood conditionrelatestodesertification,detailsofwhichare deficit, with the most serious effect in the desert, dry considered inthe followmng discussion of specific land savanna, and especially the densely populated highcover classes. In general, desertification is the degra- lands of Ethiopia. In Somalia, the problem is serious dation and possible elimination of biomass caused by throughout the country. sequences of unusually dry years, which have oc- In Ethiopia, at the present rate of deforestation, 1.76 curred approximately every 20 years in the present million hectares of woodland will be lost by the year century. 2000, and 2.27 mnillion hectares of shrubland will be Thisclimiaticstressonvegetationiscompoundedby degraded. In Somalia, the main problem is degradaintensified consumption of biomass through grazing tion of shrub; however, 13,000 hectares per year of and browsing by the animals of herders who have productive forest is being lost Comparable informabeen displaced from other degraded areas. Increasing tion is not available for Djibouti and Sudan, but population, both human and animal no doubt con- clearly, what little forestremains in the formerisbeing tributes to the process. Tube wels focus grazing near devastated, and in the latter the effect of wood rewatering points, leading to severe deterioration of movalin the central area is catastrophic (for details see

49 34 Estimating Woody Biornass in Sub-Snianton Africa the discussion of Classes 21,22, 44, and 62 in Chapter biomassdepletionlikely toexceed the locally available 10). -sustainable supply. Whitney, Dufoumaud, and Murdk (1987) calculate that 31,000 square kilometers of woodland are con- Physical Characteristics, Population, and Resources suned as fuetwood each year in Sudan. Additional pressures have occurred in parts of Sudan, Ethiopia, Estimating woody biomass in this large region is difand Somalia as the result of famine and political un- ficult, so it is reassuring that the classes of land cover rest. Refugees from both threats have concentrated in that we describe often are similar to previously idensmall areas of the tree countries, causing additional tified ecological zones. In the humid tropical forest stress on fuelwood resources. In Sudan, areas particu- zone, areas exhibiting both sparse population and larly affected are the eastern plains, the Red Sea Hills, plentiful woody biomass often are a conjunction of east of El Obeid, western Darfur, and the Juba area. cause and effect. Because domestic fuel is the main end The Hom of Africa has experienced a great influx of use of woody biomass in this region, our discussion refugees: 1.5 million people have moved from Ethio- focuses on areas of scarce or potentially scarce pia to Somalia, Sudan, Kenya, and Djibouti since the woodfuel. mnid-1970s. Civil wars in recent years in Sudan and Woody bionass of the region reflects variations of Somalia have displaced more than 800,000 refugees to climate associated with distance from the equator. neighboring countries (Black 1989). Thus, the largest forest areas of Africa occur in this Inthe famine areas of eastemethiopia (Tigre, Welo, region, within a narrow beltextending 2 degrees north and Ogaden) and Sudan (Darfur, Kordofan, and Red and 2 degrees south of the equator. This area receives Sea Hills), additional stress on vegetation is caused by less rainfall than many other tropical rain forests (for increased animal browsing and by human consump- example, Amazonia and Southeast Asia), but still suption of food derived from trees during famiine (for ports extensive tracts of closed forest This narrow example, see FAo/UN 1984). It is unclear if the present equatorial zone hais a brief dry season of about 2 resettlement camps will permanently stress local bio- months. mass, as might occur when host governments attempt North and south of this zone, where the dry season to make the camps self-supportingby giving only land extends to 3 or 4 months, tiinner and usually deciduas a contribution to refugee self-sufficiency. This is, of ous forests and woodlands occur. These are miombo course, an additional pressure on biomass resources. woodlands in the south (part of the ZarbeziarL phy- For example, near Qala 'en Nahal, the largest Sudan- togeographical region) and Guinean and Sudanian ese refugee settlement, the rate of deforestation has woodlands to the north. At the northern extremity of increased dramatically during the past decade. the central African region (10' N in Central African The previous Ethiopian government's policy of re- Republic; 120 N in Cameroon), the dry season lasts at locating people to new settlements, particuarly in the least 6 months, and the vegetation is that of the Sahel southern part of the country, undoubtedly did con- belt Atthesouthernextrenity ofthe regioninthehigh tribute to deforestation of land cover classes which at plateau of Shaba (1f Sin Zaire), where the dry season present have reasonable reserves, such as Ombrophi- also extends to 6 months, the vegetation is open woodlous Humid TropicalForest (87). land, often described as wooded savanna. On this general pattern of woody biomnass imposed Central Africa by variation of the equatorial climate, further varability results from relief, soils, and human interference. Six countries constitute the central African region: The effect of relef is most notable on the high ground Cameroon, Central African Republic, Congo, Equato- of the westem Rift Valley in eastern Zaire and on the rial Guinea (Rio Munil), Gabon, and Zaire. Woody mountains of westen Cameroon. (HeIre, persistent biomass is generally plentiful in central Africa, al- cloud cover precluded some small areas from being though important local exceptions occur, and a strong classified from the satellite imagery.) geographical pattem reveals significant variation be- Because of its varied relief and latitudinal extent, tween the relatively plentiful supplies of the Zaire Cameroon provides a complete spectrum of the vegebasin and the less-endowed areas of northern Camer- tation of intertropical Africa, with humid tropical foroon, Central African Republic, and southern Zaire. est in the south, open woodland in the center, grass- Locally, large urban populations can stress the supply land and bushland in the north, and mountainous of woodybiomass,asituationrecognizedincitiessuch forest (Laclav&re 1980). Relief also plays a role in the as Kinshasa, Brazzaville, and Ponte Noire along the extensive plateau areas to the south of the Zaire Basin, Zaire River and Kolwezi and Lubumbashi in the where temperature reflects the extensive terrain above southern part of Zaire (Malaisse and Binzangi 1985). 1,000 meters. Away from the narrow belt of humid Only in a few areas like these, however; is woody tropical forest. the interplay of the relief with the pre-

50 Regional Summaries 35 Table 6-4. Distribution of Summary Classes and Estimated Woody Biomass, Central Africa Area Growing stock Sustainableyield Million Thousand Suimmary class km 2 Percenit toines Percent lannes per year Percent 0 Desert Grassland 17, Wooded Grassland 41, Shrubland 7, Bushland and Thicket 49, , Low Woody Biomass Mosaic 5, Woodland 1,620, , , High Woody Biomass Mosaic 645, , , Forest 1,550, , ,226, Lakes 20, Total 3,958, , ,378, Not Details may not add to totals because of rounding. Sourmc Tables 11-1 to cipitation variation from year to year can produce of 18.8 percent. They observe that some vegetation much greater spatial and temporal variation in the typeshadalmostdisappearedlocally,indudingdense plant cover. riparian forest Soil types and conditions can influence extensive Detailed studies like this are rare, but the extent tracts where the woody biomass is less than would be of human impact is not Even in Gabon, 80 percent expected for central Africa. Examples include Veld of which formerly was covered with forest, large Grassland (11) on the Kalahari sands of Quaternary areas now are recognized as High Woody Biomass ageinsouthemzaireandedaphicwoodedgrassland Mosaic. Here, small-scale clearance by traditional (25) in Congo on "drought/' plateau soils, despite means of shifting cultivation has significantly recelving more rainfall on average than the adjacent changed the nature of the forest. However, no dishumid tropical forest. quiet need yet be felt about Gabon regarding woody A review of the classes in table 6-4 discloses that the biomass supply (Catinot 1978). Cultivation and lower woody biomass summary classes are either not forest mosaic is an important component of land represented or occur only as very restricted land cov- cover, accounting for 16 percent of the area of ceners. Desert () is not present in the region at all; Grass- tral Africa (table 6-4), commonly occurring at the land (1) covers less than half of 1 percent; Wooded margin of the forest zone where it represents human Grassland (2) covers 1 percent; Shrubland (3) and Low encroachnent and modification. Woody Biomass Mosaic (5) are nearly absent; and Bushland and Thicket (4) covers just over 1 percent. By East Africa contrast, Woodland (6) covers 41 percent of the region and is an important fuelwood reserve in many of the Five countries constitute this region: Rwanda, more populated areas High Woody Biomass Mosaic Burundi, Uganda, Kenya, and Tanzania. (7) covers 16 percent and Forest (8) covers 39 percent in a broad swath along the equator, presenting Physical Characteristics, Population, and Resources Africa's best-endowed woody biomass reserve. Household fuelaccountsfor90percent of all wooduse Condition of Woody Bionass in this region. With an annual average per capita consumption of 1 cubic meter a person, coupled with Even here, in the African region having the most plen- an annual average population growth rate exceeding tiful woody biomass, reference often is made to the 3 percent, it is not surprising that concern exists reprofound influence of human interference on the land garding dwindling woody biomass. At first, attention cover. This is illustrated by data collected by Milaisse focused on semiarid regions of obvious scarcity, parand Binzangi (1985) for an area 1 degree of latitude by ticularly along roadsides. More recently, attention has 1 degree of longitude, centered on Lubumbashi in focused more logically on high-potential arid areas soutiem Zaire. These authors estimate that, since the where the majority of the population lives and where beginning of the century, the retreat of woodland and the greatest demand exists. This refocusing has led to forest involved 21.6 percent of the area (212,162 hect- the design of projects to reduce the problem of fuelares), with an associated diminution of wood reserves wood scarcity within areas of high potential

51 36 Estinmnting Woody Biontass in Sub-Sharani AJfica Table 6-5. Distribution SuJl)mmanj clnss of Summary Classes and Estimated Woody Biomass, East Africa Ana Growihag slock Suaslainablc ield. Million Tlhousand Permet ens Percentt Ionniesperyear Percenpt 0 Desert 14, I Grassland 6, Wooded Grassland 142, , Shrubland 34, , Bushland and Thicket 490, , Low Woody Biomass Mosaic 146, , Woodland 526, , , High Woody Biomass Mosaic 140, , Forest 202, , , Lakes 94, O Total 1,798, , ioo.oo 228, Moae. Details may not add to totals because of rounding. Source: Tables 12-1 to The situation, however, is not simple. First, critical Because the satellite images are intended to give a fuelwood shortages exist primarily around towns in detailed view of vegetation on a regional scale, the the more arid areas, including the rain shadow of the resolution is insufficient to record the often dramatic coast, rather than in any particular rural production changes in forest type caused by changing altitude. system. Second, although wood scarcities are obvious Hence, the images fail to record Montane Forest (84) in landscapes as one drives through them, what one insuch areas as MountlKilimanjaro andmountkenya. sees is wholesale cutting for charcoal production in Although these ambiguities deserve recognition, response to the demand for urban energy-an arterial Montane Forest is of little value as available woody process closely related to principal transport routes. biomass. It not ordy is physically inaccessible, but Third, although fuetwood constitutes most of the remaining Montane Forest generally is strictly prowood consumed, this wood is waste wood-it is a tected by government. Fortunately, in light of their by-product of woody biomass being used for many immense scientific value, these sites have been well other purposes, including fodder. Finally, significant documentedelsewhere. evidence exists that land privatization or greater con- A similar problem arose in tte mapping of extentrol by- village communities leads to significant in- sive M-ngrove (81) forest, which lines much of the caeases in woody biomass stocking because it ceases East African coast. For most of its length, Mangrove tobeafree good accessible to all. This occursin poorly forest is too narrow to be resolved at the mapping controlledandmanagedgovernment"forests,"which scale used here, apart from the area around large often are treated as commons. riverinlets. Although more accessible than Montane Estimating woody biomass production in East Af- Forest, as well as being an excellent source of wood ricais particularly difficult, especially from large-scale for charcoal manufacture, the value of mangrove as inagery, for two reasons. First, the East African bio- a source of fuelwood is also limited because of mass landscape is made up of ecological niches in frequent protection. which altitude, not spatial variation, determines A review of the classes in table 6-5 discloses that growing stock and annual productivity. Second, on many different land covers are important in this rethis altitudinal mosaic, which ranges from equatorial gion. Desert (0) and Grassland (1) together cover only glaciers to desert, a complex series of production sys- about 1 percenl of the region, but Wooded Grassland tens exists, from pastoralism to modern mnonocrop- (2) covers 8 percent and is important in northern ping agriculture. These systems involve sophisticated Kenya. Shrubland (3) covers 2 percent of the region, traditional technologies of dry land farming as well as mainly in eastern Kenya. Bushland and Thicket (4) intensiveirrigationforhighvalue-addedexportcrops. occupies a broad area in eastern Kenya and central The phenology of woody biomass generally shows a Tanzania, covering 27 percent of the region- Low marked seasonality, which reflects the bimodal distri- Woody Biomass Mosaic (5) covers 8 percent, mainly bution of annual rainfall. in western Kenya and Uganda. Woodland (6) forms The use of satellite imagery created problems in the most extensive class at 29 percent and is distribmapping of the forested areas in this region. The most uted in the west and south of the region. High Woody extensive type of forest in East Afica is Evergreen Biomass Mosaic (7) covers 8 percent and Forest (8) Forest (82), mostly situatedin the highlands of Kenya. covers 11 percent of the region; they are confined

52 Regional Simniauics 37 mainly to areas imnediately north and west of Lake Tanzania and significant areas of Rwanda and Victoria. Burundi. One woodland type, however, Moist Sudanian Woodland (65), occurs predominantly in the Condition of Woody Biomnass northern part of Uganda. These open woodland areas have a growing stock of more than 40 million tonnes, At a general level, the figures for growing stock reveal nearly 1 percent of the regional total. The remaining the characteristic vegetation of the three larger coun- woodland classes occur principally in Tanzania. tries of East Africa-woodland in Tanzania, bushland EastemTanzaniahas the main block of Wet Mionzbo in Kenya, and the mosaic of cultivation, forest rem- Woodland (67), with outliers scattered throughout nant, and derived wooded grassland in Uganda. Of Tanzania as well as Rwanda, Burundi, and southwestthe main classes of land cover, the first significant em Uganda. The growing stock of Wet Mionrbo Woodgrowing stock is Semidesert Wooded Grassland (21), land occurs mainly in Tanzania (1,272 million tonnes), which dominates the arid north of Kenya. Despite its followed by Uganda (126 million tonnes), Rwanda (28 large extent, the growing stock of this class is only 34 million tonnes), and Burundi (22 million tonnes). million tonnes, less than1 percent of the regional total. Seasonal Miombo Woodland (66) dominates the An important class in southern Kenya and northern western third of Tanzania, again with outliers Tanzania is Transitional Wooded Grassland (24). It throughout the country and, to a lesser extent, in the has a growing stock of nearly 1.8 million tonnes, more other countries of East Africa. With more than onethan two-thirds of which is in Tanzania. This repre- fifth of the entire regional growing stock, Seasonal sents 0.03 percent of the regional growing stock, and Miombo Woodland accounts for more biomass than the class produces 0.13 percent of the regional sustain- any other class in East Africa. The vast majority of the able yield. growing stock is in Tanzania (1,872 million tonnes), Bushy Shrubland (33) is locally quite important, although the class is locally important in Burundi (60 although the greatest area is largely within the protec- million tomnes), southwestern Uganda (31 miuion don of Aberdare National Park in the highlands of tonnes), Kenya. (25 million tonnes), and Rwanda (12 southern Kenya. These relatively small areas have a million tonnes). combined growing stock of more than 32 million TheHighWoodyBiomassMosaicismostimportant tonnes, with two-thirds in Kenya and most of the in Uganda and southwestern Kenya, although signifremainderintanzania.sustainableyieldis0.7percent icant areas exist in the highlands of Tanzania. The of the regional total Guinean Woodland (74) is especially common in Dry Acacia-Commiphora Bushland and Thicket (41) northwestern Uganda, whereas the center of the councovers huge dry areas of the region, especially eastern try is predominantly a Cultivation and Forest Re- Kenya and the steppes of nortiern Tanzania. Of a growth Mosaic (73). Evergreen Woodland Mosaic (71) growing stock of 359 million tonnes, 298 million are in is restricted to the moister parts of the region, notably Kenya, with 50 million of the remainder in Tanzania. the Kenyan highlands, the mountains north of Lake This class constitutes 5.5 percent of the regional grow- Malawi, the Lake Victoria Basin, and the coastal and ing stock. A rather smaller area is covered by Moist offshore islands of Tanzania. Acacda-Commiphora Bushland and Thicket (43), which Guinean Woodland has a growing stock of 35 milforms a large triangular block in the center of Tanza- lion torunes, producing only 0.45 percent of the susnia, with significant outliers in all of the other coun- tainable yield in Uganda, with 3.25 percent of the tries. Despite its size, the growing stock of this land growing stock. The Cultivation and Forest Regrowth cover class is almost twice that of the drier bushland, Mosaic of central Uganda is a little more productive, with 294 million tonnes in Tanzania and 66 mnillion with 1 percent of the Ugandan sustainable yield on 7.4 tornes in Kenya. Similarly, the sustainable yield of this percent of the growing stock. Uganda holds 72 million moister bushland is much greater, accotuting for a tonnes of growing stock, with the remainder in higher percentage of the regional sustainable yield. Tanzania. The East African Low Woody Biomass Mosaic (52) The forests of East Africa occur largely in the Lake covers much of northeastern Uganda and much of Victoria Basin of southern Uganda and in the highsouthwestern Kenya. This class has most of its grow- lands of the other countries. The Montane Forest class ingstock inkenya-about 11 million tonnes, with the (84) ismostextensive in thehighlands of Kenya, where remaining 70 million tonnes in Uganda. The produc- the growing stock is 29.7 million tonnes. The remaintivity is similar in the two countries, with the com- der of this class is in Uganda, where the Montane bined sustainable yield nearly5 percentof the regional Forest of the Lake Victoria escarpment contains a total. growing stock of 13.5 million tonnes. Productivity is Perhaps the most important biomass resource is the high, as in most of the forest regions, which account extensive miombo woodland that dominates much of for 67 percent of the regional sustainable yield.

53 38 Estimating Woody Biomnass in SUb-Sahnamn Africa Coastal and GaUery Forest (83) occurs mostly in Southern Africa Kenya and Tanzania, with much smaller areas in Uganda and Burundi. The growing stock of nearly 150 For the purpose of this study, southem Africa is that million tonnes includes 93 million in Kenya and 49 portion of the continent,south of Zaire and Tanzania, million in Tanzania. Overall, they represent 3.5 per- excluding the snall Angolan enclave of Cabinda. Its cent of the regional sustainable yield and 2.3 percent land area is nearly 6 miltion square kilometers. of the growing stock. Most Evergreen Forest (82) growing stock is in Tanzania, with 660 million tonnes, Plhysical Characteristics, Population, and Resources although theothercountrieseachhavestocksbetween about 17 million and 41 million tonnes. Much of this All eight of the sununary classes of land cover used in is physically inaccessible, and large areas of East this study occur in southem Africa, where we have Africa's dwindling forests are protected. divided them further into nineteen subdivisions. The majority of Mesophilous Humid Tropical For- These nineteen classes are based on vegetation pheest (85) is in Uganda, with a growing stock of about nology;productivity, andland cover, the floristiccom- 172 miion tonnes.the remaining 10 million tonnes is position being less important. Consequently, some in Kenya, largely within the Kakamega forest north- classes incorporate a variety of vegetation types that east of Lake Victoria, which is uniquebecause i is the have been identified as distinct floristic units by auonly remaining moist tropical forest, in Kenya. The thors such as Werger (1978) and White (1983) but Lake Victoria Basin has been heavily cultivated, and which we have related by phenology, productivity, the indigenous forest, representing the eastem periph- and biomass. Our land cover dasses, however, are ery of the Guineo-Congolian rain forest belt, exists more appropriate for appraisal of fuelwood resources mainly in a-mosaic of secondary savanna and farm- than previously defined floristic mapping units. land. The growing stock is 2.8 percent of the regional A review of the classes in table 6-6 shows that Desert total and sustainable yield is 9.3 percent. (0) occupies nearly 6 percent of the region, being con- Humid Tropical Swamp Forest (86) is restricted to fined largely to the westem coastalstrip fromsouthem Uganda, where it occurs extensively on the western Angola to South Africa. Grassland (1) covers 6 percent shore of Lake Victoria and elsewhere. The growing of the region and Wooded Grassland (2) covers 11 stock exceeds 18 million tonnes, representing 0.2 per- percent; both are important classes in South Africa, cent of the East African total and 0.3 percent of the Botswana, and Naniubia. Shrubland (3) covers nearly sustainable yield for the region. The Ombrophilous 11 percent and is important in the same countries. Humid Tropical Forest (87) is much more extensive, Bushland and Thicket (4)-covers 8 percent and Low with a growing stock of 347 mnillion tonnes in Uganda Woody Biomass Mosaic (5) covers 9 percent Woodand lesser amounts of 37 million in Kenya and 9 land (6) is by far the most extensive class, blanketing million in Tanzania In Uganda, this class is a mosaic 38 percent of the region with a broadly northem and ofcultivated land and remnants of the drier peripheral eastem distribution from Angola to Mozambique. semi-evergreen Guineo-Congolian rain forest of High Woody Biomass Mosaic (7) and Forest (8) each White (1983). cover 5 percent of the region. Table 6-6. Distribution of Summary Classes and Estimated Woody Biomass, Southem Africa Area Growingstock Sustainable yield Million Thousand Summary class A?n 2 Percent lonnes Percent tonnes per year Percent O Desert 350, a.oo 1 Grassland 356, , Wooded Grassland 667, , Shrubland 636, , Bushland and Thicket 471, , Low Woody Biomass Mosaic 519, , , Woodland 2,243, , , High Woody Biomass Mosaic 329, , Forest 317, , , Lakes 41, Total 5,933, , , Note Details may not add to totals because of rounding.. Sourre: Tables 13-1 to

54 Regional SummarIes 39 Condition of Woody Biomnass viding dry-season fodder, construction poles, furniture, tools, habitat for game, mulch to enrich the soil, This study extends the work on biomass in the SADC and protection against wind and water erosion, This area described in titc Foundation (1987) and Milling- exerts considerable pressure on- trees and smaller ton and others (1989). As explained in the methodol- woody plants in all densely populated parts of soutlogy, the NDVI wecreused as thebasis of classification for em Africa. It is especially noticeable in the semiarid both of these studies and in the present work. Th NDVI arc of low woody biomass that stretches from southindicatemorethan justwoodproduction,considering westem Angola through Namibia, across southem all plant material, including leaves, grasses,herbs, and Zambia and Botswana, and into southem Zimbabwe crops, as well as wood. Large areas of southem Africa and South Africa. are cultivated, often by large commercial agricultural Another zone of lesser woody biomass resource, enterprises, and this poses problems in estimating although less obvious, is discemible from westem available woody biomass growing stock and produc- Mozambique through Zimbabwe, Zambia, and tivity. The risk also exists of overestimating fuclwood Malawi into Tanzania. This wide "dry-zone corridor" yields if the accessibility factor is not considered, for has a basis in topography and climate; for example, example in the case of areas reserved by govemment southwestem Mozambique is drier and the Luangwa and national parks. Valley in eastem Zambia is lower and drier. Much of Commercial forestry, such as that in South Africa this zone's raison d'etre, however, seems to be in local and Swaziland, also may give a false impression of the soils and drainage and, more important, in the popuavailability of fuelwood. Ideally, avaiability should lation pressure on woody resources. Therefore, this be evaluated using detailed local or regional informa- z&ne needs to be examined more closely at a national tion and fieldwork. Although some of this typeof data and regional level. is incorporated herein, the regional information islim- The following list synopsizes the distribution of ited, and the scale of this study meant that field check- summary classes in the region: ing could be undertaken only in selected areas of 0 Desert covers nearly 6 percent of southern Africa fuelwood having high and low potential. and occurs in the southwest as the Namib Desert, as On abroad, regional scale, the pattern of land cover part of the Great Karroo of South Africa, and in the groupings in this study resembles that of biogeo- driest parts of the Kalahari Desert in Botswana. graphical studies (for example, Werger 1978) because 1, Grassland also covers nearly 6 percent of the many of the influencing factors-are similar. For exam- region and includes the Veld Grassland (11) of the ple, the regional pattem of land cover groupings interior South African plateaus and the Montane closely resembles the regionalpatternof meanannual Grassland and Heathland (14) of the southern precipitation (figure 7-1). These corresponding pat- DrakensbergMountains. terns range from the Namib Desert (Class 0), which 2 Wooded Grassland is represented by Transihas less than 100 millimeters of precipitation a year, to tional Wooded Grassland (24), covering 11 percent of the extensive biomass and high-productivity areas in the land area but containing only 0.23 percent of the the north and east, where precipitation exceeds 1,000 woody:growing stock (table 6-6). It occurs mostly on millimeters annually. fine-textured soils in areas experiencing approxi- Numerous environmental factors also influence the mately 400 to 800 millimeters of rainfall a year, in patternsdisplayed,however.werger(1978)pointsout widely vaned locations but mainly in Namibia, that climate and topography are the basic determi- eastem Botswana, southern Zimbabwe, southern Monants of vegetation structure, biomass, and productiv- zambique, northern Transvaal, and in patches ity in southem Africa. This is reflected in this study, as throughout the remainder of South Africa, often on the attested by a comparison of the maps depicting south- fringes of the veld. em African biomass, rainfall, and topography. The 3 Shrubland is relatively important in southem subdivision of land cover classes, however, is further Africa, coverng nearly 11 percent of the region but affected on a smaller scale by local factors, of which containing only about 2.8 percent of the growing soils, drainage, and land use are the most important. stock. More significantly, Shrubland accounts for 7.1 The impact of human population growth is evident percent of the region's sustainable yield. It indudes from the different land use pressures and from the the small class Veld Shrubland and Cultivation (31) of clearance of woody vegetation. n rural areas, people Cape Province; the large classes of Fill Shrubland (32) rely mostly on wood from trees and bushes for house- and Bushy Shrubland (33), both of which largely hold fuel, and the demand for urban energy promotes fringe the Namib Desert area; the KCalahari Shrubland cutting for charcoal production on a much more dev- (34), which occurs mostly in southeastern Namibia, astating scale. Woody biomass is also part of the inte- northem Cape Province, and southwestern Botswana; grated production system of the peasant farmer, pro- and Wooded Shrubland (35), which occurs largely in

55 40 Estitmating Woody Biomass in Sibd-Saharan Afilcn central Cape Province, on higher ground east of the greater rainfall along the southeastem coast of South Namib Desert, and on the slightly wetter fringes of the Africa and Mozambique. Kalahari Desert. 4 Bushland and Thicket covers nearly 8 percent of 7 High Woody Biomass Mosaic is represented by two classes, covering about 5 percent of southern southern Africa, and represents about 4 percent of the Africa but including only about 2 percent of the growing stock. It is mostly represented by Moist Aca- subcontinent's growing stock. Cultivatiun and Forcia-Cominiplora Bushland and Thicket (43). This class est/woodland Mosaic (72) includes many highoccurs as a broken arc from southern Angola and productivity (although not always very woody) bionorthem Namibia through parts of Botswana, Zam- mass areas, associated with areas of greater moisture bia, Malawi Zirnbabwe, and Mozambique to South and cultivation zones, often occurring in areas of un- Africa and Swaziland, with its largest continuous oc- usual drainage characteristics such as the Okavango currence in Transvaal and Natal. Bushland and Delta in Botswana. Evergreen Woodland Mosaic (71) Thicket is often related to particular soils or land occurs mainly in south-central Angola and along the forms, or to regenerative phases following exploita- coast of central Mozambique. tion. The much smaller area of Fynbos Thicket (42) is 8 Forest has only one representative in southem confmed to small areas of the coastal ranges of Cape Africa-Evergreen Forest (82), which covers more Province in South Africa. than 5 percent of the land area and includes dense 5 Low Woody Biomass Mosaic covers nearly 9 montane forest as well as commercial forested areas. percent of southern Africa and indudes about 6 percent of the growing stock It occurs in Namibia These areas represent about 10 percent of the growing stock of southern Africa, and 36 percent of the sustainand Botswana and is scattered throughout southem able yield. Angola, eastern and southem Zambia, Zimbabwe, The proportion of woody biomass in these sumand southeastern Mozambique, as well as in South mary classes mostly increases from Class 0 through Africa. Class B, although exceptions exist. Classes Oto3 clearly 6 Woodland dominates nearly 38 percent of the have very limited fuelwood potential. In Clkss 4, the continent south of Zaire and Tanzania and has large rateof exploitationcaneasilyexceedproductbvity.the woody biomass reserves representing 74 percent of Low Woody Biomass Mosaic (Class 5) can be importsouthern Africa's growing stock It extends across the ant locally, often coincident with dense rural populacontinent from Angola to Mozambique, generally tions and in need of careful management. modulating density according to conditions, from Open Woodland (61), through; Seasonal Miombo The large Woodland group (6) clearly includes a largeproportionofthesubcontinent'swoodybiomass Woodland (66), to- Wet Miombo Woodland (67). resources,butgreatvariationoccursinactualproduc- Miombo woodland is related to moist, frost-free or tivity and availability of fuelwood. The High Woody nearly frost-free tropical conditions, and it doniinates Biomass Mosaic (7) includes some cultivated areas of the northern part of this region, with density varying little woody biomass stock, but also includes large according to seasonality ofprecipitation, edaphic con- areas of dense woodland having high productivity dition, and degree of hulman interference. The group, and fuelwood potential Forests are among the most however, also includes wann, temperate woodland (part of Class 67) and tropical coastal woodlands (part productive woody biomass classes of southem Africa, but often include areas that have accessibility probof Class 66). These occur respectively in the areas of lemsbecauseofremotenessorcommercialownership.

56 7 Conclusions and Future Directions riparian vegetation on narrow flood plains, narrow beltsoflandcovercontrolledbyaltitudeineastafrica and the Ethiopian Highlands, and the individual ele- ments of cultivation mosaic (Summary Classes 5. and 7). The level of detail in our study is adequate for an initial situational overview intended for determining policy objectives, but for planning purposes it is not adequate. An extensive search of the secondary literature con- firmed the lack of data on woody biomass supply. Much of this literature has its origin in forestry and therefore concems commercial woody species. But. other end uses in addition to energy and commercial timbe are important in supply-side estimates and should be considered. Also, this work does not consider agricultural resi- dues and herbaceous biomass. Not only are these an energy resource, particularly in arid and semiarid areas such as the SaheL but crops and herbaceous plants contribute to the NDvL Future work to upgrade the quality of data bases such as these should involve such considerations. A positive outcome of our research is that the inte- gration of the data base and land cover maps to produce maps of growing stock and sustainable yield (figures 7-1 and 7-2) does begin to define areas in whichproblemsof supplymayexistorbenascent.the presence of low estmates within or adjacent to areas of greater biomass suggests areas where early action should be taken to confirm the level of supply. Exam- ples of such areas are the West African Sahel and the dryzonecorridoradjoiningpartsofeastemandsouth-* era Africa. Early action could take the form of further mapping from imagery that has both a finer spatial resolution and a continuous time sequence of data. Other early response could mean improving the quality of the field data for the data base. Studies based in regions like those just identified, where improvement in the estimate of wood supply is needed, should include at The products of this study are (a) maps showing classes of land cover (defined by their woody biomass),cb)adatabaseofwoodybiomassgrowingstocand sustainable yield, and (c) maps of growing stock and sustainable yield for Sub-Saharan Africa. Throughout this volume we have stressed that this study has been a first attempt to cost-effectively produce a continent-wide view of the woody biomass situation in Sub-Saharan Africa. The maps and data base are available for strategic planning, including the priority designation of areas requiring woodfuel supply enhancement. This should enable use of such data for maldng better-informed decisions about stressed areas. These areas will require more intensive assessment, coupled with surveys of woodfuel demand, to yield a better understanding of how to manage the fuelwood resource. A benefit of conducting a survey at this scale is the ability to compare areas at reconnaissance level, so that extrapolations based oni ecology can be made from areas of plentiful data to areas where data are scarce or absenl From this study it is clear that a pressing need exists to develop among nations a more harmonized strategy for woody biomass estimation. Another benefit of this study will be inreased awareness of the potential of AVHRR imagery to allow biomass assessment at the regional level, when combined with secondary data or field surveys. This method affords an impression of the woody biomass situation in relation to consumption. It also indicates where more intensive investigations are needed, to support meaningful intervention to manage the woody biomass resource. Clearly, use of AVHRR NDvi data at a spatial resolution of 8 kilometers restricted our identification and mapping to broadly defined classes of land cover. Small classes having a distribution restricted to narrow belts or small parcels of land use were unlikely to be identified. Examples include the very narrow belt of mangrove along the Indian Ocean coast, 41

57 42 Eslintaling Woody Bionias in Sub-Slaiarap; Africn F:gure 7-1 Growing Stock for Sub-Saharan Africa, 1986 ISO I ~~~~~~~~~~~~~~~~~~~.1 0 ~ ~ ~ ~ ~ ~ ~ ~~~~~~3 *~~K' ' /.... S -/X--' ' 1E,<' V -.. Growing slock J'A More han L ' 10,0 20,0 p ICO2D Is, - is, the design stage a common methodology deived cation of the present study usingavhrrdata having from the findings of this projecl a finer resolution. Analysis at the regional level, We have identified four specific issues for future however, requires multisensor land resource satemapping of land cover for woody biomass asses- fite iimagery of finer resolution (for example, Landsment. They are: sat Thematic Mapper) to provide maps of better detail for energy planners. Such work is already Spatial resolution of the maps needs to be improvedfor being carried out in Pakistan by ESMAP in cornuncpla-nningpurposes. We are not advocating the repli- tion with the Ministry of Planning (Energy Wing).

58 Conclusio and Fulire Directions 43 Figure 7-2. Sustainable Yield for Sub-Saharan Africa, 'a-~~~~~~~~~~~~~~~~~~~~~~~~5 -~ ~~~ / r S.,~~.J.0 I -'''1'-'p 7' --. <:' * ~~ -...,~~~* I a. :~~~~~~~~~~~**~ *4' Ia :~~ ;91!ll>I (V Sutaiale yied) M0 :r t0h : \x \7 g., *Reliance solely on NDViprod ucts from theamrrimagery nel Microwave Radiometer (smmrt) data from the -brzooody biomass mapping is questionable- It has been mirowave portion of the spectum should be conshown that thermal infrared data, also available sidered in future studies. fr-om the.avhrr provides ad ditional infonmation *Year-to-year variations in. vegetation cover, photosthaitcanbeused to rf ineland covermaps produced ynthetic activity,and the resultant,mivviresponse should * using oi products. it is particularly usefl for be considered. These were not considered in the preidentifying andmappinglandcoverdisturbance. In sent study, and the present maps are only a situaaddition, the applicability of Scanning Multichan- tional overview of the woody biomass resource for

59 44 Estimating Woody Bionmss in Sub-Salharan Africa They are subject to change caused by human inter- Support Programxs vention and climatic variation. Significant improvement to the maps of land cover, growing * Earlyinvolvementoflocal,regional,andnational stock, and sustainable yield would occur with use organizations of the complete AvHRR data set spanning 1981 to * Technical assistance and support in staffing and the supply of necessary hardware, software, and Medium-term monitoritg of the woody biomass re- training programs in developing countries. soturce, for example, spanning a 5-to-10-year period, requires (a) a highi-quality baseline data set, (b) data of Links withi Studies of tlte Demnand Side tie appropriate spatial resolution to detect changes, and (c) a robust metihodology. We doubt that AVHRR data * Coordination with studies of woodfuel demand of 8-kilometer spatial resolution would allow rec- in the same land cover classes ognition of changes in land cover class caused by * Supporting studies of competing end uses of human activity that spans periods briefer than a wood. decade. The NDvis very sensitive to rainfall and soil moisture variations. It is apparent from research The following areas in Sub-Saharan Africa should already undertaken that, utilizing a phenological receive high priority as areas in which to conduct the strategy for mapping at this scale, changes in land research just outlined: cover classes caused by such variations probably will mask changes caused by human activity. 7he Sudanian Woodlands of the Sudano-Sahelian zone. The work reported here dearly identifies and maps Other areas require further research three belts of Sudanian Woodland that are differentiated by the length of their dry seasons and by their Improvement of the Data Base moisture regimes. These belts stretch from Senegal to Sudan, covering 13 million square kilometers, - Harmonization of assessment techniques for and are a principal source of woodfuel This area woodybiomass currently is experiencing a dual threat to the fuel- = Ground sampling that includes destructive sam- wood supply. First, people are migrating northpling to establish the relation between measurable ward into these woodlands from the environmenparameters and the weight of woody biomass tally stressed areas of the SaheL Second, the large - Preservation of permanent plots for long-term population centers to the south, along the West monitoring to determine annual growth incre- African coast, are increasingly relying on woodfuel ment and sustainable yield from these woodlands to meet their domestic en- - Assessment thatincludes the herbaceous compo- ergy demands. nent of biomass The main factors to be covered by such work are Study to relate the age structure of vegetation to (a) mapping of the land cover classes within the its productivity. Sudano-Sahelian zone at a fimer spatial resolution and (b) a greatly enhanced data base on the woodv Technical Improvement biomass growing stock and sustainable yield. Such work could be accomplished by using Landsat im- * Useofremotelysensedimageryhavingappropri- agery for interpretation of land cover and by estabately finer spatial resolution lishing a local network of biomass assessments Inclusion of continuous remotely sensed data, re- within the woodlands. corded during a span of several contiguous years * The dry-zone corridor of southern and East Africa. This as opposed to a single year, to produce land cover dii -r-. belt, typified by arid and semiarid vegetamaps that are more independent of year-to-year tion and farmnng systems, extends from Ethiopia climatic varations and Somaliasouthward througheast-central Africa * Support of satellite remotely sensed imagery to Botswana and Zimbabwe. The zone comprises with other fom-s of assessment, such as aerial several areas of ecological stress resulting from photography and ground verification scant rainfall and strongly seasonal moisture re- * Integrated use of three powerful tools: assess- gimes. This area is important for woodfuel supply ments of land cover and woody biomass, envi- because of the great rural population density and ronmental data bases for climate, demography, large urban centers. Woodfuel supplies have been and other factors such as soils, and geographical depleted by clear felling for agriculture. Several information systems. local studies of good quality provide benchmark

60 References Every effort has been made to facilitate access to the Australasicn Remote Sensing Conferece (6th: 1992: documents listed here. Some documents, however, Wellington, New Zealand), 2-6 November, Vol. 1, lack full bibliographic information because it was un available; also, some documents ae of limited circula- Brown, S., and A. E. Lugo "Biomass of Tropical ton The word "processed" descrbes informally re- ForestsANewEstimateBasedonForestVolumes." produced works that may not be commonly available Sdence223: drough libaries. Catnot, R "The Forest Ecosystems of Gabon: An Overview." linvum (JrdtedNationsEducational Andeke Lengui, M. A Formulation du projet Scientificand CulturalOrganization), TropicalForest amdnagemt et reboisementfjrestiers A bhts multiples. Ecosystems. Paris uno. ime f 85/009 Dcumentde terrainno. 2Niamey, Chidumayo, E.N "SpeciesStructureinZambian Niger Ministbre de 1'Hydraulique et de l'environ- Miombo Woodland-" Journal of Tropical Ecology 3(2): ment. FAo Baines, A. C "Sierra Leone: Assistance to For- CLristiansen, J. H 'Biomass and Primary Proestry Development." Freetwn. Processed. ductivity of Rhizpho Apiculea in a Mangrove in Barber, Dennis R "The AppLcation of Change Soufterm T and. Aquatic Botany4:43-5 Detction Techiques to Rangeland Monitormg in Climent, J "Estimation des volumes et de la South Australia." Proceengs, Avstrasian Remte productivit6 des formations mixtes foresti&res et.sensing Conference (6W 1992: Wellington, Newo Zea- grinues tropicales v ue Bois et Forts des land), 24 November, VoL 1, Trpiques, 198(4): Bartholomew, W. V., J. Meyer, and - Landelot Cline-Cole, Regiald, J. A. Falola, IL A. C. Main, Mi- -Minerdl Nurient Immobilization Under Forest chael E Montimoe Janet E. Nichol and Frank D. and Grass FalUow in the Yangambi (Belgian Congo) 0. O`Reilly Woodfu in Kano. Department of Region." Sie Scientifique 57 lnstitut National. GeogVaphy, Bayero University, Kano, Nigeria. Final Bianhi, I-L Assistace, au developpen sntfiwstier, report ofthe rural energy researchprojectsubmitted Guinec Bissau; Planification foresti?re. Tc;a'/s to the UN. UnniyDeparnert Studies Division. 4506(A). Document de terrain. Rome: FO. Cracknell, Arthur P., and Ladson D. Hayes In- Black, R "Africa's Refugee Crisis." Geognphical troduction- to Remote Sensing. London: Taylor and Magezine 61(1)12-1& Bonkoungou, L, L. Bortoli, and J. M. Oudba Francis. Danaher, T. J.- J. 0. Cart, D. Bok A. Peacock, "Note an Burkina Faso Ecosystems: Woody and and G. S. Dudgeon "Broad-Scale Vegetation IHerbaceous Vegetation Characteristics of a Pro- Mapping Using NOAvRR Imagery.- Proceedings, tectedfallowin DegradedSudaian Zone.lInstitut' Australain Remofe Sensing Cifnere (Wth 1992: de Recherche en Biologie et Ecologie Tropicale Wedlington, New Zaland), 2-6 November, VoL 3, (ier) Ouagadougou, Burlina Faso. Processed Brook, K D.,J. 0. Carter,T. J. Danaher,G. NE MdxCon, Daus, Steven J., Mamane Guero, and Lawally Ada. N. R Flood, and A. Peacock "The Use of "A Remote-Sensing-Aided Inventory of Fuel- SpatialModellingand RemoteSensingforMonitor- wood Volumes in the Sahel Region of West Aficac ing and Forecasting of Drought-Related Land A Case Study of Five Urban Zones in the Republic Degradation Events in Queensland." Proceedings, of Niger." Paper presented at Symposium on Re- 46

61 References Every effort has been made to facilitate access to the Australasian Remote Sensing Conference (6th: 1992: documents listed here. Some documents, however, Wellington, New Zealand), 2-6 November, Vol. 1, lack full bibliographic infornation because it was un available; also, some documents are of limited circula- Brown, S., and A. E. Lugo "Biomass of Tropical tion. The word "processed" describes informally re- Forests ANew EstimateBased onforestvolumes." produced works that may not be commonly available Science223: through libraries. Catinot, R "The Forest Ecosystems of Gabon: An Overview." InuNEsco (United Nations Educational, Andeke Lengui, M. A Formulation du projet Scientific and Cultural Organization), Tropical Forest ambtagement et reboisementforestiers a buts multiples. Ecosystems. Paris: unesco. ivtnm/ 85/009 Document de terrain No. 2. Niamey, Chidumayo, E. N "Species Structure inzambian Niger Minist&re de 1'Hydraulique et de l'environ- Miombo Woodland." Journal of Tropical Ecology 3(2): ment. FAO Baines, A. C "Sierra Leone: Assistance to For- Christiansen, J. H "Biomass and Primary Proestry Development." Freetown. Processed. ductivity of Rhizopho7a Apiculatea in a Mangrove in Barber, Dennis R "The Application of Change Southem Thailand." Aquatic Botany 4: Detection Tecbniques to Rangeland Monitoring in Clenent, J "Estimation des volumes et de la South Australia." Proceedings, Australasian Remote productivite des formations mixtes forestieres et Sensing Conference (6th: 1992: Wellington, New Zea- gramnmennes tropicales." Revue Bois et Forets des land), 2-6 November, Vol. 1, 1A Tropiques, 198(4): Bartholomew, W. V., J. Meyer, and IL Landelot Cline-Cole, Reginald, J. A. Falola, IH A. C. Main, Mi- "Mineral Nutrient Immobilization Under Forest chael E. Montimore, Janet E. Nichol, and Frank D. and Grass Fallow in the Yangambi (Belgian Congo) 0. O'Reilly Woodfuel in Kano. Department of Region." S&rie Scientifique 57 Institut National. Geography, Bayero University, Kano, Nigeria- Final Bianchi, H Assistance au d6veloppementforestier, reportoftheruralmergyresearchprojectsubmiitted Guinee Bissau; Planification forestiere. TCP/GEIS/ to the U.N. University Department, Studies Division. 4506(A). Document de terrain. Rome: FAO. Cracknell, Artiur P., and Ladson D. Hayes, In- Black, R "Africa's Refugee Crisis." Geographical troduction to Remote Sensing. London: Taylor and Magazine 61(1):12-1& Francis. Bonkoungau, L, L. Bortoli, and J. M Oudba Danaher, T. J., ' J. Carter, K D. Brook A. Peacock, "Note on Burkina Faso Ecosystems: Woody and and G. S. Dudgeon "Broad-Scale Vegetation Herbaceous Vegetation Characteristics of a Pro- Mapping Using NoAA-AvHRR Imagery." Proceedings, tectedfallowindegradedsudanianzone."institut Austnlasian Remote Sensing Conference (6th: 1992: de Recherche en Biologie et Ecologie Tropicale Wellington, New Zealand), 2-6 November, VoL 3, (=ET), Ouagadougou, Burkina Faso. Processed Brook K D.,J. 0. Carter,T. J. Danaher, G. lvi McKeon, Daus, Steven J., Mamane Guero, and Lawally Ada. N. R. Flood, and A. Peacock "The Use of "A Remote-Sensing-Aided Inventory of Fuel- SpatialModelling and Remote SensingforMonitor- wood Volumes in the Sahel Region of West Africa: ing and Forecasting of Drought-Related Land A Case Study of Five Urban Zones in the Republic Degradation Events in Queensland." Proceedings, of Niger." Paper presented at Symposium on Re- 46

62 References 47 mote Sensing for Resources Developmnent and En- eds., Environmen t and Land Use in Africa. London: vironmnertal Management, August 1986, Enschede, Methuen. Netherlands. Gray, T. L., and D.CG. McCrary "Meteorologicakl Diallo, 0., A. Diouf, N. P. Hanan, A. Ndiaye, and Y. Satellite Data--a Tool to Describe the Health of the Prevost "AvwwR Monitoring of Savanna Pri- World's Agriculture." AGWusTARs Report rw.ni mary Production in Senegal, West Africa, NASA/USC (National Aeronautics and Space Admiin " International Journal of Remote Sensing 12(6): istration/goddard Space Center), Houston Greenland, Denis T., and Jan M. L. Kowal "Mu- Drury, Steven A A Gutide! to Remote Sensing: trient Content. of Moist Tropical Forest of Ghana." Initerpreting Images of the Earth. Oxford: Oxford Unti- Plant and Soil 12(2): versity Press. Guerreiro, M. C "A floresta Africana e os Eidenshink, J. C., and R. H. Haas "Analysing factores bioticos. Primeiras observaq8es de urn en- Vegetation Dynamics of Land Systems with Satellite saio emmocambique." Paper in M. J.A.Werger,ed., Data." Geocarto International 7(1): , Biogeography and Ecology of Sou thern Africa. The rrc (Education and Training ConsultaLnts) Foundation. Hague: Junk Wood Energy Dvlpet: Biomass Assessment, Gutmnann, Garik. C "Monitoring Land E-cosysa Study of the sadcc Region. Leusdien, Netherlands. tenms Using the NJOAA Global Vegetation Index Data FAo (Food and Agricultural Organization of the United Set!' Paleaegeography, Palaeoclimnatology, Palaeoeco- Nations) "Etudes sur les volumes et la pro- logy 90: ductivitd des peuplements forestier.s tropicaux. 1. Guy 1 P. R "A ndansonia Digitata and Its Rate of Formation forestiares s&ches." Etude MAo For&ts Growth in Relation to Rainfall in South Central 51/1. Rome. Africa." Proceedings and Transactions ofrpjzdesian Sci- FAo (Food andagriculturalorganizationof theunited ence Association. 54: Nations) World Food Report. Rome. Guy, P. R "Changes in the Biomass and Produc- PAo/pND (Food and Agricultural Organization of the tivity of Woodlands in the -Sengwa Wildlife Re- United Nationts/Programme des Nations Unies search Area, Zimbabwe." Journal of Applied Ecology pour le D&veloppement) "La conservation 18(2): des dcosyst&mes foresti6res dans la Re-gion des Henderson-Sellers, A., G. Seze, F. Drake, and M. Marts Kouffe." Rome. Debois "Surface Observed and Satellite- Flemons, P. K. J "Estimating Pasture Variables Received Cloudiness Compared for the 1983 isca Using AvNHR Data. The Effect of Soil Colour anld (International'Satellte Cloud Climatology Pro- Cover Extent." Proceedings, Australasian Remote gram) Special Study in Europe!' journal of Geophys- Sensing Conference (6th.~ 1992: Wellington, New Zea- ical Research 22:04,4019. land), 2-6 November, Vol 3, Hertrickson, B. L., and J. W. Durkin "Growing Franklin, J., and P. H. Y. Hiernaux "Estimat- P eriod and Drought Early Warning in Africa Using ing Foliage and Woody Biomiass in Sahelian and Satellite Data!' International Journal of Remote Sens- Sudanian Woodlands Using a Remote Sensing ing 7(11): Mo del" Internatio nal Journal of Ramote Sensingl12(6): Holben, Brent N "Characteristics of Maximum Value Composite Images from Temporal AThER Gatlin, J. A., R. J. Sullivan, and Compton J. Tucker. Data." International Journal of Remote Sensing 7(11): "Monitoring Global Vegetation Using NoAA AvHRRData." Proceedings of the IGAP.RSymposium, San Huntley, B. J 'Characteristics of South Afric-an Francisco, Fart I, FF2, 7.1. Biomass.' In P. de Booysen and N. M. Tainton, eds., Golley, F. B., J. T. McGiunis, K. G. Clements, G. Child, Ecological Effects of Fire in South Afrian Ecosystems. and M.L J. Duever "The Structure of Tropical Ecological Studies 48. Berlin: Springer-Verlag. Forests in Panama and Colombia." Bioscience 19: Jackson, J. K "Productivity of Natural Wood land and Plantations in the Savanna Zones of Nige- Cofley, F. B., and others "La biomasa y la es- nia.-savannaforestryresearchstation,federalde-. tructura midneral de algunos bosques de Darien, partment of Forestry Research, Samuru,. Zaria, Panama." Turrialba 21: Nigeria. Coward, Samuel N., Compton J. Tucker, and N. A. Justice, Christopher 0., ed "Monitoring the Dye "North American Vegetation Patterns Grassland of Semiarid Africa Using NoAA-AvHBR Observed withl No.'AA-7 Advanced Very HEigh Reso- Data." International Journal of Remote Sensing 7(11):..lution Radiometer." Vegetation 64t Graham, A "Man-Water Relations in Fast Central Justice, Christopher O., and P. H.L Y. Hiernaux Sudan." In KA F. Thomas a-nd C. W. W'hittington,. "Monitoring the Grasslands; of the Sahel UsingNoAA

63 48 Estimatinig Woody Biotmss ia Sub-Salwran Africa AVHRR Data: Niger." International Journal of Remote Huntley and B. H. Walker, eds., Ecology of Tropical Sensing7(11): Savannas. Berlin: Springer-Verlag. Kamweti, D. M Fuelwood in Eastern Africa: Pres- Millington, Andrew C., John R. G. Townshend, Pam ent Situation and Future Prospects, Rome: FAo. A. Kennedy, Richard Saull, Steven D. Prince, and Kennard, D. G., and B. H. Walker "Relationships Robert Madams Biomass Assessment. Woody Between Tree Canopy Cover and Panicum maximum Bio;mass in the sadcc Region. London: Eartbscan. in the Vicinity of Fort Victoria." RJ:odesian Journal of Munslow, Barry, Yemi Katerere, Adrian Ferf, and Agricultural Research 11: Philip O'Keefe Thte Fuclwood Trap: A Study of Koomanoff, V. A "Analysis of Global Vegeta- the sadcc Region. London: Earthscan. tion Patterns: A Comparison Between Remotely Norwine, J., and D. H. Greegor "Vegetation Sensed Data and a Conventional Map." Biogeogra- Classification Based on AVHRR Satellite Imagery." phy Research Series, Report , Dept. of Geog- Remote Sensing of Environment 13(1): raphy, University of Maryland. Nye, P. H "Organic Matter and Nutrient Cycles Laclavere, G Atlas of the United Repuiblic of Cam- Under Moist Tropical Forest." Plant and Soil 13(4): eroon. Paris: Les editions Jeune Afrique Lamotte, M., and F. Bourliere "Energy Flow and Openshaw, Keith Somalia: TheForest Sector. Prob- Nutrient Cycling in Tropical Savannas." In F. lems and Possible Solutions. Mogadishu, Somalia: Bourliere, ed., Ecosystems of the World 13. Tropical World Bank/Energy/Development Interna- Savannas. Amsterdanu Elsevier. tional/ministry of Mineral and Water Resources. Leach, G., and R. Mearns Beyond the Woodfuel Pecha, Albert Developpementforestier. Document Crisis-People, Land, and Trees in Africa. London: de terrain No. 18, Republique Populaire du Congo. Earthscan. Rome: FAO. Lewis, L. A., and L. Berry African Environments Perry, CharlesR.,Jr., and LyleF. Lautenschlager and Resources. Boston: Unwin Hyman. "Functional Equivalence of Spectral Vegetation In- Lundgren, B., ed Land Use in Knya and Tanzania. dices." Remote Sensing of Environment 14 (13): 169- Stockholm: Royal College of Forestry Malaisse, P. P "Phenology of the Zambezian Persson, Ri Forest Resources ofafrica, Part I, Coun- Woodland Area with Emphasis on the Miombo Eco- try Descriptions. Stockholm: Royal College of Forsystem. In He Leith, ed., Phenology and Seasonal estry. Modelling. Berhlin Springer-Verlag. Prince, Steven D "Monitoring the Vegetation of Malaisse, F. P., and Binzangi, K "Wood as a atropical Rangelands with the NoA-7 Ad- Source of Fuel in Upper Shaba." Commonwealth For- vanced Very High Resolution Radiometer." In MI J. estry Review 64(3): Eden and J. T. Parry, eds., Remote Sensing and Tropi- Malingreau,J.-P "Monitoring Tropical Wetland cal Land Management. Chichester, England: John Rice Production Systems: a Test for Orbital Remote Wiley. Sensing." In M. J. Eden and J. T. Parry, eds., Remote Prince, Steven D 'Satellite Remote Sensing of Sensing and Tropical Land ManagemenL Chichester, Prmary Production: Comparison of Results for Sa- England: John Wiley. helian Grasslands " International Jounal Malingreau, J.-P., and Compton J. Tucker of Remote Sensing 12(6): "Large-Scale Deforestation in the Southeastern Am- Prince, Steven D., and Compton J. Tucker "Satazon Basin of BraziL" Ambio 17: ellite Remote Sensing of Rangelands in Botswana I Malleaux, H The Forest Inventory of Mozambique. NOAA AVHRR and Herbaceous Vegetation." Interu- Rome: FAO. tional Joura of Remote Sensing 7(11): Malo, A. R., and Sharon E. Nicholson "A Study R4publique du Togo "Document de terrainof Rainfall and Vegetation Dynamics in the African Systemes d'exploitation rurales et la situation Sahel Using Normalized Difference Vegetation actuelle de la forot" Republique Togolaise TOG/ Inde" Journal ofarid Environments 19: /008. Rome: FAo. Marsch, H. E Inventaired'amAnagemtendeforEtde Roderick, M. L., and t C G. Smit "Use of NOAA Lama. Document de terrain No. 3. Rdpublique Pop- Derived Seasonal Vegetation Data Within a GlS for ulaire de B6nin. Rome: FAO. Broad Scale Vegetation Management in Westem Menaut, J. C., and J. Cesar "Structure and Pri- Australia." Proceedings, Australasian Remote Sensing mary Productivity of Lamto Savannas, Ivory Conference (6th: 1992: Wellington, New Zealand), 2-6 Coast." Ecology 60(6): November, VoL 3, Menaut, J. C., and J. C&ar '"he Structure and Rutherford, M. C "Primary Production Ecol- Dynamics of a West African Savanna." In B. J. ogy in Souther Africa.1" n M. J. A. Werger, ed.,

64 References 49 Biogeography and Ecology of Souttlern Africa. The Land Cover Classification by Remote Sensing: Pres- Hague: Junk. - ent Capabilities and Future Possibilities." Remote Ryan, Paul, and Keith Openshaw "Biomass As-. Sensing of Environ ment 35: sessment: A Discussion on its Need and Methodol- Townshend, John R. G., and Compton J. Tucker ogy." Working paper. World Bank, Industry and "Objective Assessment of AVHRR Data for Land Cover Energy Department, Washington, D.C. Mapping." InternAtional Jouirnal of Remote Sensing Sarmiento, G., and M. Monasterio "Life Forms 5:492. and Phenology." In F. Bourliere, ed., Ecosystems of Tucker, Comptonj.,J. A. Gatlin, andstanley R. Schneithe World 13. Tropical Savannas. Amsterdam: Elsevier. der "MonitoringVegetation in the Nile Valley Saull, R. J., Andrew C. Millington, and M. Crosetti. with NoAA.6 and NoAA-7VHRR." Photogammetric En "Pakistan: A Land Cover Zonation from Multi- gineering and Remote Sensing 50(1): temporal AVHRR and Envirorunental Data Using cis Tucker, Compton J., Brent N. Holben, and Thomas E. Techniques." Proceedings of the 5th AVHRR Data Goff "Intensive Forest Clearing in Rondonia, Users Meeting, LUMEISAT, Tromso, Norway, Brazil as Detected by Satellite Remote Sensing." June, Remote Sensing of Environment 15(3): Schneider, Stanley R., David F. McGinnis, Jr., and J. A. Tucker, Compton J., and Piers Sellers "Satellite Gatlirn "Use of NOAA AVHRR Visible and Near- Remote Sensing of Primary Production." Interna- Infrared Data for Land Remote Sensing." NOM (Na- tional Joural of Remote Sensing 7(11): tional Oceanographic and Atnospheric Administra- Tucker, Compton J., John R. G. Townshend, and tion) Technical Report NESS 84 Washington: usdc. Thomas E. Goff "African Land Cover Classi- Schneider, Stanley R., David F. McGinnis, Jr., and fication Using Satellite Data." Science 227(4685): George Stephens "Monitoring Africa's Lake Chad Basin with Landsat and NoAA Satellite Data." Tucker, Compton J., C. L. Vanpraet M. J. Sharman, International Journal of Remote Sensing 6(1): and G. Van Ittersum "Satellite Remote Sens- Sellers, Piers "Canopy Reflectance, Photosyn- ing of Total Herbaceous Biomass Production in the thesis, and Transpiration." International Journal of Senegalese Sahel: " Remote Sensing of En- Remote Sensing 6(8): vironment 17(3): Sellers, Piers "Canopy Reflectance, Photosyn- van Wilgen, B. W., K B. Higgins, and Du Beilstedt. thesis, and Transpiration I. The Role of Biophysics "The Role of Vegetation Structures and Fuel and Their Interdependence." Remote Sensing of En- Chemistry inexcluding Fire from Forest Patches in vironment 18: thefire-proven FynbosShrublandsofSouthAfrica." Stonehouse, Bernard Pockt Guide to the World. Journal of Ecology 78(1): London; George Philip and Co., Ltd. Werger, M. J. A Biogeograp1iy and Ecology of Stomgaard, Peter "Biomass Estimation Equa- Southern Africa. The Hague: Junk. dons for Miombo Woodland, Zambia." Agroforestry White, F "The Vegetation of Africa." Natural Systems 3(1):3-13. Resources Research Series 20. Paris: UNEsCO/AEIlAT/ Tarpley, D. R "The NoAA Global Vegetation Index unso (United Nations Educational, Scientific, and Product-A Review-" Palaeogeography, Palaeo- CulturalOrganization/Associationpourl'Etude Taxo- - climatology, Palaeoecology 90: nonique de la Flore de 'Afrique Tropicale/United Tarpley, J. D., Stanley RI Schneider, and R. L. Money. Nations Sudano-Sahelian Office). - 19S4. "Global Vegetation Indices from NoAA-7 Me- Whitney, J. B. R., C. MI Dufournaud, and B. W. Murck. teorological Satellite." Journal of Climate and Applied "An Examination of Alternatives to Tradi- Meteorology 23:491. tional Fuelwood Use in Sudan.' Journal of Environ- Townshend,John R.-G., Christopher O.Justice, andv. mental Managenent (25) Kalb "Characterization and Classification of Woodward, F. Ian Climate and Plant Distribu- South American land Cover Types Using Satellite tion. Cambridge, England: Cambridge University Data." Internatimal Journal of Remote Sensing 8(8): Press World Resources Institute/International Institute for Townshend,JohnR.G.,ChristopherO.Justice,LiWei, Environment and Development World Re- Charlotte Gumey, and Jim McManus "Global sources New York Basic Books.

65 PART II Regional Distribution of Land Cover Classes Part II of this book examines the regional distribution of In eachregion, we describe only large classes of land landcoverdasses(figure3-5).broadecological,environ- cover (those having an areal extent greater than 1 mental, and economic criteria defined our six regions: percent of the region), with the exception of smaller areas that are important to an individual country. * West African Sahel (Mauritania, Senegal the Gam- We present each class in relation to its geographical bia, Mali, Burkina Faso, Niger, and Chad) distribution within the region, describe the main ele- * West African coast (Guinea-Bissau, Guinea, Sierra ments in the natural vegetation and land use, and Leone, Liberia, C6te d'ivoire, Ghana, Togo, Benin, address the current and future fuelwood resource, as and Nigeria) appropriate. * The Hom of Africa (Sudan, Ethiopia, Djibouti and For some important classes in each region, NDVI * Somalia) profiles are plotted to afford an impression of how * EastAfrica(Uganda,Burundi,Rwanda,Kenya,and land cover can be differentiated using the methods Tanzania) described in Chapter 3. At the conclusion of each * Central Africa (Cameroon, Central African Republic, regional chapter are tables that provide national data mmnland EquatorialGuinea,Gabor,Congo,andZai.e) on the area, growing stock, and sustainable yield for * Southern Africa (Angola, Zambia, Malawi, Mozam- each land cover class. References are also supplied. At bique, Zimbabwe, Nanibia, Botswana, South Africa, the end of Part 11 are the four regional maps depicting Swaziland, and Lesotho). the land cover classes.

66 -~ The West African Sahel Richard W. Critchley This chapter presents a detailed description of the exist, consisting of the extensive rooting systems of most importantlandcoverclassesinthisregion.help- drought-resistant geophytic plants. This, of course, ful figures in other chapters include figure 3-1 (cloud cannot be considered a sustainable fuel resource. cover); figures 3-2, 3-3, and 3-4 (Now sununary land Other species are ephemerals, surviving by the procover profiles); figure 3-5 (regional summary map of duction of seed during favorable periods. Rainfall in land cover classes); figures 7-1 and 7-2 (continental the more northern areas of the region may be exmaps of growing stock and sustainable yield); and the tremely rare, as little as one event in 10 years, and "Regional Land Cover Class Map of West Africa" at may occur as brief rainfall events of 10 to 30 millithe end of this volume. -meters. Helpful tables in other chapters include table 3-2 Following a rainfall, a brief explosion of vegetative (land cover classes); table 4-1 (data and sources for activity forms a herbaceous carpet, creating a system growing stock and sustainable yield); and table 6-1 referred to as acheb. The plants that emerge are mostly (West African Sahel estimated woody biomass by tropical, although overlap exists with more Meditersummary class). ranean species in areas of greater relief, such as the Air Mountains in MNger and the ribesti Plateau in Chad. Class 0-Desert Herbaceous species also are present, such as the hissock-forming grass Stipagrostis pungens and the grass Within the region that we designate West African Panicum turgidum, which responds to water availabil- Sahel in this study, the dominant land cover class in ity by rapid vegetative growth from dormant buds at relation to surface area is desert, about 50 percent. The ground level, providing an important food for antecountries of Mauritania, Mali, Niger, and Chad extend lope. The annual grass Coelachyrum compressa can eswell into the northern Sahara. In these areas, where tablish itself on the banks of wadis, and the perennial precipitation is scant and evaporation is rapid, the stolon-forming grass, Eleusine compressa, also repreflora is restricted to more favorable locations where sents the tropical desert flora (Gillet 1986). Aristida groundwater may be available, such as dried-up river pungens, A longflora, Cornulaca monocantha, Cymbopobeds or wadis. gon monocantha, Eragrostis sp., Stipagrostis uniplumis, Rainfall in the western half of the Sahel region has and Tribudus sp. also are described in these areas and been less than average in every rainfall season since farther south (no 1989). Because much of the region is mar- The southernmost edge of the class approximately ginal for agriculture, an extended reduction in precip- coincides with the 20-millimeter isohyet. In this area, itation is likely to be disastrous. The contribution of or in depressions farther north where runoff may augthis dass to the regional growing stock of woody ment the available water supply, woody species do biomass obviously is negligible, although locally it occur. The commonest is Acacia tortilis ss. raddiana, may be significant to nomadic people in the southern which has a restricted but uniform distribution. Leptaportion of the area. denia pyrotechnica is restricted to dunes and Balanites The vegetation is not obvious for most of the year, aegyptiaca and Commiphor africana occur locally in the although significant below-ground biomass may low peneplains (Le Hou&ou 1980). 53

67 54 Estimating Woody Biolass in Sub-Salmren Africa The south Saharan plateaus of Ar (Niger), Adrar Figure 8-2. NDVI Profile, Hydromorphic Grassland des Iforas (Mali), and Ennedi (Chad) provide areas of (Class 12) increased runoff and slightly greater rainfalls. They are characterized by the Semidesert Wooded Grass land Vegetation (Class 21), and in general, woody -ZChmd biomass gradually increases as total precipitation, ire- os5 quency, and regularity increase toward the south of 0.4- the zone. Because of the supervised classification technique 0.3- used to map land cover (Chapter 3), some small parcels of land in Niger appear to be Class 0, but are surrounded by Class 21. These are believed to represent degraded areas. The causes of this degradation are.considered to be of two kinds. The first cause is o, cyclic climatic vanration, leading to an increase in arid- Jan reb Mar Apr May Jnn Jul Aug Sp Oct Nov Dec ity spanning a period of years. In such areas, the vegetation would be expected to recover in years of greater rainfall The second cause is believed to be Class 21-Semidesert Wooded Grassland anthropogenic activity, and such areas could return to greater production levels with appropriate land The Semidesert Wooded Grassland class forms a relamanagement tively narrow belt, at itsbroadest about 400 kilometers, nnig from Mauritania in the west through central Class 12-Hydromorphic Grassland Mali, northern Burkina Paso, southwestem and central Niger, central Chad, and extending into East Af- This class has a very restricted distributionin the West rica. It covers about 19 percent of the surface of the African Sahel, covering less than 1 percent of the re- West African Sahel region. A.more limited distribugion. Its contribution to the regional woody biomass tion occurs north and south of this zone within the growing -stock and sustainable yield is negligible Desert (Class 0) to the north and the Sahel-Sudanian (table 6-1).. Acacia Wooded Bushland (Class 44) farther south The The biomass activity produces a. relatively small Semidesert Wooded Grassland class corresponds NDvi response (for example, in Chad, figure 8-1) but largely to mapping unit 54a of White (1983).IThe more with a marked seasonality. It is small in March and northerlyextensions are associated withthemassifs of April at 0.35, but smallest in September at Peaks Adrar des Iforas (Mali), Air (Niger), and Ennedi (Chad), in production are indicated in December and January but the floral constitution of these areas is different with values of 0.55, somewhat later than the peak due to the inclusion of Saharan desert species. precipitation of the wetseason (uly and August). This Rainfallinthe regionis about250 millimeters a year, lag relates to the drying out of waterlogged soils. with a marked seasonality. Rainfall peaks in July and Hydromorphic Grassland occurs as small patches August and coincides with summer temperatures to witiin the Dry Sudanian Woodland (Class 62) in the produce warmth and humidity that can cause rapid westem and southeastern portions of Mali and to a and vigorous plant growth. The monthly Nmm curves lesser extent in eastern Senegal and in Chad, south- show a marked seasonality, with peak values of 0.44 west of N'Damena. Its distribution in Mali is associ- in September -in response to the increased rainfall, ated with the Manding Plateau, and similarly with falling to 0.10 in May (figure 8-2). The northern limit minor plateaus in SenegaL In Chacf it represents a of the class coincides approximately with rainfalls of grassland vegetation formed on Quaternary clays in 100 to 150 millimeters a year, whereas the southern areas susceptible to flooding. White (1983) considers limit coincides with the 500-millimeter isohyet We this to be the most widespread kind of edaphic grass- estimate the class to contribute about 6 percent of the land, similar to that forming on seasonally or perma- growing stock and 13 percent of the sustainable yield nently waterlogged soils in the upper Nile Basin. More in the region. extensive areas occur in Sudan and northern Nigeria The vegetation consists of grassland with a crown and complete an east-to-west pattem wiffiin Clas 62, cover of less than 10 percent woody species. In the Dry Sudanian Woodland. northem extreme of the zone, annual plants are asso- For a more complete description, please refer to the ciated with wadis or plateau regions, but are typically regional reports for the West African coast (Chapter 9) tropical rather than Mediterranean in origin. Annual and Hom of Africa (Chapter 10). grasses such as Codlachyrum bremifolium and perennial

68 - W-The WetAoran She 55 Figure 8.2. NDVI Profile, Semidesert Wooded A. tortilis attaining 5 meters height, with total woody Grassland (Class 21) biomass rarely exceeding 3 percent of the abovepgound biomass. This species also provides shade 0.6- within which herbaceous vegetation may thrive and - 21 Mauritania persist long after the and of the dry season. Also "S described among the more important woody species Oh - _ reported by White (1983) are Acacia lacta, A. ehrenbergi- /anat(flw), Balanitesaegaptiaca, Boscia senegalensis, Camffi 0.3- / mipborafricana, and Laptadenia pyroteclrn ea. Anogeissus leiocarpus, an important fuelwood tree, 0.2- also increases to the south in this class. The size to o. l- > - -.which semidesert trees being grow varies, stunted those and in attaining the northem full stature Sahel o- 4only - with increased rainfall in the south of the region. ionxfeb MAr AprMay ion Jul Aug sep dat NoC ; Calligonum comosum, a typical Saharan shrub, often is associated with Panicum turgidum in areas where rainfall is locally greater. species such as Eleusine compressare cited by Gillet Human population density is low in the region but (1986) as representing the basic Sahelian flora. generally increases along a nort-south transect. The WhererainfallexceedslOOmillimetersayear.andis great drought of saw extensive movement predictable, a more permanent steppelike vegetation southward (Granier 1980), intensifying the pressure develops. Indeed, on a north-south gradient, a general on a system that has a low carrying capacity. In. this imcrease occurs in the proportion of trees. This more region, the most characteristic activity is pastoralism, with seasonal movements of herds and southerly region coincides with the second Sahelian zone described by: Boudet (1972). Local conditions associated flocks. vary and considerable differences exist in the proportionof woody species, but White's description (1983) of bushes and small bushy trees within grassland Class 44-Sahel-Sudanian Bushland Acacia Wooded seems appropriate, with more favorable areas developing sufficient woody biomass to be described as. This class lies south of the Semidesert Wooded Grassbush. land (Class 21) and forms a parallel belt of similar The species represented in the herbaceous layer width across the region, extending into East Africa, vary considerably from east to west, with local varia- with significant representation in all nations in the tions related to the proportion of sand and clay within Sahel except the Gambia. In Mauritania, only the areas the soil. Generally, however, tussock-forming grasses along the Senegal River Valley, the escarpments of gradually give way to perennial, steppelilke species 'Assiba and Tagant, the Massif of Afolle, and the southward along a north-south transect. Annual Hodh Basin in the southeast are included. All other grasses such as Aristida mutabilis, A. adscensionis, A. countries in the West African Sahel include significant stipoides, Cenchrus biflorus (cram-ram), Scdoenefeldia areas that add to 17 percent of the surface of the region. grcilis, and Thzgus racemnosus are widespread and typ The class is associated with greater rainfall than Class ical of the northern Sahelian semidesert, whereas 21, with annual precipitation between 250 and 500 perennial grasses such as A. mutabilis and Panicum millimeters occurring largely-in July and August turgidum extend quite far south. Regionally, this class is important for woody biomass, On unstable soil systems such as dunes, gasses contributing an estimated 5 percent of the growing such as Stipagrostis pungens are more common; such stock and 31 percent of the sustainable yield (131 species also are associated with lesser rainfall. The million tonnes and 30.7 million tornes, respectively). instability of tall dunes farther south favors the pres- The Nxvi phenologies of this type of land cover ence of Aristida sieberana and Panicum turgidum and exhiibit the greatest value dunng September (figure their distribution is increased at the expense of woody 8-3). The greater NDvi period has a duration of 2 to 3 species by processes that encourage desertification. months, and throughout the rest of the year.nxvx val- Of the woody biomass component, the Acaa genus ues are approximately The growing season is is representedbyanumberof species. Widespread are longer in the west, and there herbaceous. vegetation Acacia tortilis ss. rddiana, A. senegal, A. seyal, and, in persists through a large part of the year. the more southerly limits of the zone, A. nilotica In The herbaceous layer consists mainly -of annual. the northiem part of the region, White (1983) reports grasses, including species represented farther north

69 56 Es1tinia lg Woody Biomtass in Su b-sclarant Africa Figure 8-3. NDVI Profiles, Sahel-Sudanian Mcacia A case study of two traditional systems of consump Wooded Bushland (Class 44) tion in the Sahel-Sudanian zone of Burkina Faso (MArz 1986) suggests that 75 percent of all wood collected is 0.6- for fuel and that farmers very carefully select different - 44 Sciegl tree species for different uses. The area required to 0,5-44 Nigr bupply the wood demand of each person is 1.4 to 2.8 OA- hectares from a system producing 0.16 to 0.48 tonne a hectare. In areas seen during field visits in December , a great intensity of land use was noted for many kilometers around both Dakar in Senegal and Bamako \ / / - \in Mali, with agricultural plots and fallows in regular o1 l '-pattems, serving both the cities and numerous villages. -i -In recent years, in Burkina Faso in particular, large 0 I, M M J, A, p N, numbers of people have moved from the north to the Jn rcb Mar Apr May Jun Jul Aug Scp Oct Nov Dee southwest in response to population increase. The associated agricultural activity has led to degradation of the environment and the effect is apparent on the such as Aristida stipoides, A. mutabilis, Cenchnrs biflones land cover class maps, where areas of Semidesert (cram-cram), Chlorisprieurii, and Schoenoeldiagracilis. Wooded Grassland are surrounded by Sahel-Sudan- Perennial grass spedes such as Andropagon gayanus ian Acacia Wooded Bushland. Areas of wooded bushoccur farther south. The ground vegetation is more land also extend farther south into the Dry Sudanian persistent than in Class 21 to the north and survives Woodland (Class 62). almost to the end of the dry season, attaining heights A recent study of the fuelwood resource in Niger greater than a meter. (Daus, Guero, and Ada 1986) indicates varying de- Of the woody species, Acacia tortifis ss. mddiana is grees of vegetation degradation associated with areas the most common in the north of the zone. According of greatpopulationdensityin the south and southwest to White (1983), it varies in form throughout the re- of the country. In study areas of 200 kilometers diamgion, being short and bushier farther north with a eter aroundfive urban zones, the greatest degradation maximum height of 4 meters, whereas in the south it occurred around Tahoua, where the ratio of woody is a bushy tree of up to 8 meters. Regional descriptions biomass to surface area is smallest This juxtapos: ion identify differences in species com.position across the of two classes, the class associated with lesser woody Sahel. Of widespread occurrence are Acacia senegal, A. biomass occurring against a background of greater seyal, members of the Combretaceae, especially Com- woody biomass, suggests that extensive vegetation bretum glutinosum in the Sahelian zone of northern landcoverdegradationcanbe mappedfrom8-kilome- Burlina Faso, and Ziziphus mauitiania. ter-resolution AVHRRNDVI imagery. Adansonia digitata (baobab) is locally abundant, particularly around villages or on more saline soils. Buty- Class 62-Dry Sudanian Woodland rospermum spp. and Parkia biglobosa, more Sudanuan species, are particularly important as fuelwood in This class lies south of the Sahel-Sudanian Acacia Burkina Paso. Bush species include Acacia mellfrra, Wooded Bushland (Class 44), form-ting a narrow belt Commiphora africanm, and Euphorbia balsamifmra.. through southern Senegal, the Gambia, southern and Asignificantdifferenceexistsbetweenvegetationin western Mali, southern and western Burkina Faso, the the north of the region, where the mnain human land southerrunost area of Niger, and southern Chad. Aluse is nomadic and pastoral, and that in the south, though forming quite extensive areas-about 10 perwhere agricultural activity is more sedentary. This is cent of the West African Sahel-it is penetrated by related largely to the ability to grow crops such as nmany other vegetation classes from farther north and millet in the south. In Niger, this difference is associ- south, forming complex mosaics in western Mali and ated with the 350-millimeter isohyet It may also be Chad in particular. related to seasonal transhumance in response to the This class corresponds to the Sudanian Undifferengrowth of vegetation in the wet season (Peyre de tiated Woodland mapping unit 29a of White (1983) Fabregues 1980). Granier (1980) describes how people and lies within the Sudanian regional center of endein Mali use indigenous woody and herbaceous spe- mism (M). Inmany instances, Whiteconsidersthe-area cies, andshowsthe intimaterelationshipbetweenpeo- to be a product of agriculture, consisting of land under ple and plants that must be borne in mind when cultivation orbush fallow, the original vegetation posconsidering the potential fuelwood resource in this sibly having been dry forest Certainly in Senegal, region. - - Mali, and Burlina Faso it coincides with areas of dense

70 Tie West African Sahel 57 permnent settlement. RegionaLly, 57 percent of the short-trunked, thick-barked, fire-resistant species, growing stock and 41 percent of the sustainable yield with thin-barked species surviving in areas that esare estimated to come from this class (1,439 million cape regular burning or agriculture where they form tonnes and 41.5 million tonnes, respectively). areas of closed woodland. Farther south, more water- The area is associated with annual rainfalls of 750 to demanding species are in evidence. 1,250 millimeters with a dry season of 7 to 9 months. Butyrospermumparkii, which traditionally marks the In areas of greater rainfall, such as lower Casamance start of the Sudanian zone, and Parkia biglobosare in Senegal, the region merges with the Cultivation and more common, particularly around villages. Both spe- Forest Regrowth Mosaic (Class 73). Phenology curves cies are used by the local population. They probably for sites in Burkina Faso and Senegal (figure 84) show represent the remnants of dry forest cover eliminated increases in NDVI beginning in June or July, peaking in by human activity. Butyrospermum spp. also exhibit August and September at 0.48 in the Burkina Faso site good regeneration from a root crown and many of the before fallingback tobaseline values of approxumately Combretaceaeshow cryptogeal germination,enabling This contrasts with an earlier start to the growing them to recover from fire damage. season for the vegetation in Classes 64 and 65, for Other tree species typically present around villages which growth begins in March or April. include Acaci albi/a, Adansonia digitata (baobab), Bom- The vegetation is typically open woodland with bax costatum, Borassus aethiopum, Man gifera indica, and mainly deciduous trees, 15 to 25 meters in height, with Tamarindus indica. Khaya seneglensis, Parkia biglobosa, a herbaceous carpet dominated by grasses, although and Pferocarpus erinaceus are reported as forming dry shrubs, bushes, and climbers are present The peren- woodland in Senegal (Ndidye 1983). Bosweilia dalzieli, nial herbs include Andropogon gmjanus on sandy soils; Commiphora spp., and Ziziphu spp. also are present Cymbopogon giganteus, Ctenium newtonii, and Hyparr- Areas that Steentoft (1988) descnibes as "savanna rehenia dissoluta on wetter soils; and Laudetia simplex on growth" are vegetated by Adansonia digitata, Acacia slopes or rocky outcrops. Among the annual grasses albida, Bajanites aegyptiaca, Butyrospernum sp., Ceiba are Andropogon pseudopricus, Eragrostis tremnula, Pen- sp., and Tamarindus sp. nisetum pedicellatum, and Utenium elegans, the latter Bush and shrub species include Hyphaene thebaica, a considered by Steentoft (1988) to be an indicator of palm used for building, together with members of the disturbed ground. During the dry season, the herba- Combretaceae, Rubiaceae, and Euphorbiaceae. In ceous layer becomes highly combustible, and frequent moister regions of Senegal such as river valleys, Borassus fires bum when the harmattan blows. The merger flbdlifr (rhun palm) and Oxytenantffr abyss/nica (a with the Cultivation and Forest Regrowth Mosaic bamboo) occur (Ndiaye 1983). Gallery forests including (Class 73) in southem Senegal is marked by the ap- Colalaurfolio and Mitragyna inermis exist in Niger (Peyre pearance of herbaceous species such as Daniellia oliveri de Fabregues 1980). and Ejythrophteurn aineense. The woody layer contains a significant number of thomy species typical. of the Sahel-Sudanian Acacia Class 64-Sudanian Woodland Wooded Bushland (Class 44) to the north. Large num- This class occupies 3 percent of the WestAfrican Sahel bers of Combretaceae, including Anogeissus leiocarpus and occurs in southeastern Senegal, western and and Combretum glutinosurn, can form distinct wood- southwestem Mali southwestern Burkina Faso, and lands, and Acacia spp. are common. Many trees are southerrmost Chad. It corresponds approximnately to mapping zone Ila of White (1983) as a mosaic of Figure 8-4. NDVI Profiles, Sudanian Woodlands lowland rain forest and secondary grassland, lying in (Classes 62,64, and 65) the center of the Sudanian regional center of endemisim It incorporates species from the drier northern 0.6- part of the Sudanian region, including Acacia spp. and,-* - o.s- -,, ;, -members of the Combretaceae, together with species more typical of the humid southern parts such as 0.4-, " 1aya senegalensis and Lophira lanceolata, both Sudan- -, A.ian endemic species. > 03-,, : j -*Class 64 is intimately associated in the south of the z :,-, ' Av - >, region with Dry Sudanian Woodland (Class 62), and ' -,, e Xl- 2 S is considered in some areas to be a degraded form of 62 Burldna Faso drywoodland.itmayrepresentparcelsof groundthat ~ / Burkdna Faso - are less accessible or are forest reserves. (The imagery Chad 65 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dc appears to show areas of Sudanian woodland coinciding with "foras dassees" around Bobo Dioulasso in : ,. -.,

71 58 Estimating Woody Biomass in Sub-SaI:aron Africa westemrn3urkinafaso).lntheregion,sudanianwood- The class is more fully and typically described in land is estimated to be 15 percent of the growing stock Chapter 9 on the coastal West African region. and 6.5 percent of the sustainable yield (378 million tonnes and 6.5 million tonnes, respectively). Class 65-Moist Sudanian Woodland Climatically, this zone has a precipitation of at least 1,100 millimeters a year with a dry season of 6 to 7 Class 65 has a very limited distribution, only 1 percent months from October to May. Monthly rainfall values of the land area in the West African Sahel region, during the wet season of July and August are 250 to occurring to a significant extent only in southeastem 350 millimeters, with mean daily temperatures of Chad, with isolated small pockets in Burkina Faso. It about C. This vegetation is clearly the northem corresponds to mapping unit 27 of White (1983) or the limit of the coastal WestAfricanregion.The factor that northern Guinea savanna of Steentoft (1988). limits the growth of woody species is the length of the Climatically, the region is similar to that in which dry season, and characteristically the trees are decid- Sudanian Woodland (Class 64) develops, having a dry uous. Tie canopy is seasonally open and, in combina- season of 5 to 7 months. The vegetation is similar in tion with the relatively wide spacing of the trees, light structure to that class, but is dominated by Burkea is able to penetrate, allowing tussocky grasses such africana, Daniellia sp., Erythrophleum africanum, Isoberas Andropogon gayanus, Hyparrhenia cyanescens, and linia doka, and Lophira lanceolata. It represents the most Schdzachyrium sanguineum to attain heights greater than northerly extension of a zone of vegetation occurring 1.5 meters. in the coastal West African region and spreading Granier (1980), in describing the biogeography of eastward to Cameroon, Central African Republic, and Mali, classifies-this area as partof the Sudano-Guinean Sudan. domain with a savanna-woodland mosaic, subject to Where it occurs in the Sahel, this class forms a regular fires and with gallery forest in the valleys. mosaic with Sudanian Woodland (Class 64) and Dry Ts&tse is common and this may explain somewhat the Sudanian Woodland (Class 62). The contribution to lesser exploitation of this area, compared to more the regional growing stock is significant, estimated as degraded examples of Sudanian Woodland (Class 64). 4 percent of the grdwing stock and 1.7 percent of the Characteristic species are Isoberlinia doka, a fire-resis- sustainable yield (100 million tonnes and 1.7 million. tant species, Afzelia africana, Danillia oliven, Erythro- tonnes, respectively). phleumguineense, and U4paca somon. Phenologies appear to be similar to those of Ndiaye (1983) similarly identifies the equivalent re- Classes 62 and 64, but with a maximum vn value gion in southem Senegal with a number of dominant of 055 occurring earlier in the year in May and June woody species, including senealensis gaaya and Lophim and remaining at approximately 0.45 until September lanceolaat as well as ParkaM biglobosa and Pterospus (figure 8-4). erntaceus. In valleys, thebamboo Oxytenantheraabyssinica Moist Sudanian Woodland is an open woodland and the rhun palm Bomssus flabeldfer occur. The rhun savanna, considered by some to be a reduced form of. palmalsooccursasplantationsinsouthwestemburkina miombo because fewer species are present and trees Faso. Sanford and Ischei (1986), in describing "Guinea rarely attain heights greater than 15 meters. Typically, savanna," identify Danidl1a oliveri, Isoberlinia doka, r.1 it includes Monotes kerstingii on dry eroded slopes tonentosa, and Parkia cappertina as important species, (Steentoft 1988). In the northemmost part of the area, but in the more northern areas the Isoberlinia genus is commonspecisarecussonahbarten,enadafi7ana,lsoberconsidered by White to be resticted to small podcets on limna tfosa, Lannm mncrocarpa, Terminalif avicennioides, rockyhills - and Xmn ameaa, but taces of Sahelian vegetation The term "derived savanna" is used for large parts occur in the form- of Aadt spp. (particulaly A albida), of this area, but the effect of human activity is much Burkea aficana, and ViBostigma thonningi. lessthan intheneighboringclass62.agradientof tree In areas where the woody biomass is regenerating, cover probably exists, with the cover diminishing as species such as Isoberlinia sp., Piliostigma sp., and Terone moves from Sudanian Woodland (Class 64) into minalia avicennioides exhibit the ability to regrow from areas more properly described as Dry Sudanian underground suckers and hence have an enhanced Woodland (Class 62). ability to withstand fire. The appearance of Isoberlinia The phenological curves (figure 8-4) indicate that is considered to correspond with the 1,000-millimeter the maximum No value of 0.43 occurs in August and isohyel An understoxy of climbers and shrubs may be September with the smallest value of 0.13 in January, present, and a herbaceous carpet of perennial grasses February, and March. These indicate the distinct sea- occurs, including Andropogon spp., Eragrostis spp., sonality of the vegetatiorl Schizaclyrium spp., and Pennisetum spp. This class in the Sahel is probably a mixture of a Cass 65 is largely the product of anthropogenic number of elements of Sudanian and Sahelian ongin, influences of burning, grazing, and agriculture, with

72 - te West African Sahel 59 regeneration of vegetation occurring during fallow Gillet, H In C. W. Lawson, ed., Plant Ecology in periods. Consequently, this class is unlikely to offer West Africa. London: John Wiley and Sons, Ltd. fuelwood shortages, considering the fairly low human Granier, C In M. Traore, ed., Atlas du Mali. Paris: population densities that exist. Las Editions Jeune Afrique. -m-(international Institute for Envirornment and De- Land Cover Class Tables velopment) "Rainfall in the Sahel. Paper No. 10, September, iucn (International Union for Con- Tables 8-1 through 8-7, beginning on page 60, present servation of Nature) /Sahel Programme. summaries for each land cover class of the area, show- Le Hou4rou, R N "The Rangelands of the Sahel." ing growing stock and sustainable yield for the West loumal of EWmronmental Manageent 33(1): African Sahelian nations of Burkina Faso, Chad, the Miz, U 'Wood Consumption in Traditional Gambia, Malt Mauritania, Niger, and Senegal. Systems in- the Sudano-Sahelian Zone of Burkina Faso." Quarterly Journal of International Agriculture References 25(1):4858. Ndiaye, P Atlas du Snidgal. Paris: Les Editions Every effort has been made to facilitate access to the Jeune Afrique. documents listed here. Some documents, however, lack Peyre de Fabregues, B In E. Berniss and S. A. fill bibliographic information because it was unavail- Hamidou, eds., Atlas du Niger. Paris: Les Editions able; also, some documents are of limited irculatiorl Jeune Afrique. Sanford, W. W., and A. 0. Ishei In G. W. Law- Boudet, G "Disertification de l'afrique son, ed., Plant Ecology in West Africa. London John tropicale skche." Adansonia, ser. 2,12(4): Wiley-and Sons, Ltd. Daus, Steven J., Mamane Guero, and Lawally Ada. Steentoft, M. A Flowering Plants in West Afica "A Remote-Sensing-Aided Inventory of Fuel- Cambridge, England: Cambridge University Press. wood Volumes in the Sahel Region of West Africa: White, F "The Vegetation of Africa." Natural A Case Study of Five Urban Zones in the Republic Resources Research Series 20. Paris: UNESCO/AEJFATI Of Niger." Paper presented at Symposium on Re- UNSO (United Nations Educational, Scientific and mote Sensing for Resources Development and En- Cultural Organization/Association pour l'etude vironmental Management, August, Enschede, Taxonomique de la Flore de l'afrique Tropicale/ Netherlands. United Nations Sudano-Sahelian Office).

73 60 Estimnaling Woody Biomass in Suk-Salaran Africa Table 8-1. Land Cover Classes-Burkina Faso (West African Sahel Region) Area Growingstock Suistainableyield -Thosand Thousand tonnas Land cover class kmz Percent lonnes Pcent peryear Percent 0 1, , , , , , ,321A , , , , , , , , , , B, , , , , , , , , , , Total 271, , , (Percentage of region) (5.17) (17.96) (14.74) Note:. In the fodowing tables, details may not add to totals because of runding. Source: Authors' cakulations from data bases derived from land cover dassification and table 4-1. Table 8-2. Land Cover Classes-Chad (West African Sahel RegiDn) Area Grozingstock Sustainable yield Thousand Thousand tonnes Land cover class - kn? Percent Ionnes Percent peryear Percent 0 494, , , , , , , , ,849 OA6 1, , , , , , , , , , , , , , , , , , ,214A , , , , , , , , , A , , , , LAkes 15, Total 1,277, , , (Percentage of region) (2431)- (34.40) (31.36) Sourc: Authors' calculations fiom data bases derived from land cover dassification and table 4-1.

74 Thc West Afrienji Saliel 61 Table 8-3. Land Cover Classes-The Gambia (West African Sahel Region) Area Growixg stock Su4slainable yield Tlaoussand Thouisand lontnes Land comr class km 2 Pecenit lonones Percent per year Pcrcent Z , , , , , , , , Total 16, , ,657A (Percentage of region) (0.31) (2.18) (3.67) Source. Authors' calulations from data bases derived from land cover classiflcation and table 4-1. Table 8-4. Land Cover Classes--Mali (West -African Sahel Region) AreaGoiigtd Sustainable yield Thousnd Thousand tonnes Land comer chas km Percent tonnes Percent per year Percent 0 586, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ~~~~~ B 0.06' 18, , , , Total 1,269, , , (Percentage of region) (24.16) (29.61) (26.20) SourT e Autors' calcudaons from data bases deived from in cover classifiation and table 4-1.

75 62 Estimating Woody Bfomass in Siab-Saharan Africa Table 8-5. Land Cover Cassew-Mauritania (West African Sahel Region) Area Growing stock Sustainableyield Tlhousnd Thousand tonnes Land cover class km 2 Percent toiznes Percent per year Percent 0 797, , , , , , , , , , , , , Total 1,035, , , (Percentage of region) (19.71) (3.01) (3.18) Soumv: Authors' calculations from data bases denved from land cover classification and table 4-1. Table 8-6. Land Cover Claisses-Niger (West Anrican Sahel Region) Area Growfng stock Sustainableyield Thousand Thousand tonnes Lae. coaer class kn? Percent tonnes Percent peryear Percent 0 710, , I 2, , , , , , , , , , , , , , , , , , , Lakes 3, Total 1,184, , , (Percentage of region) - (22.55) (5.26) (11.36) Sounce Authors' calculations from data bases derived from land cover classification and table 4-1.

76 The West African Coast Andrew C. Millington 9~~~~~~~~~~~~~~~ This chapter presents a detailed description of the in northeastern Ghana (100 persons per square kldomost importantland cover classesin this regiorl Help. meter) and lowest in Benin and Nigeria (2 persons per ful figure in other chapters include figure 3-1 (cloud square kilometer). In areas of greater population dencover); figures , and 3-4 (Novm summary land sity, fuelwood demand undoubtedly exceeds woody cover profiles); figure 3-5 (regional summary map of biomass stock, which is relatively low in tiese domiland cover classes); figures 7-1 and 7-2 (continental nantly grassland areas. maps of growing stock and sustainable yield); and the "Regional Land Cover.Cass Map of West Africaw at Class 44-Sahel-SudanianAcasia Wooded the end of this volume. Bushland Helpful tables in other chapters include table 3-2 (land cover classes); table 4-1 (data and sources for Aazcia wooded grassland and deciduous bushland growing stock and sustainable yield); and table 6-2 occua nly in the very north of the region, with small (West African Coast estimated woody biomass by areas in northeastem Ghana and the adjacent parts of suxnmaxy class). Togo and northeaster Benin. The class has its maximum developmentinnorthernni whereitforms Class 12-Hydromorphic Grassland a broad, continuous belt 50 to 300 kilometers wide across the country, north of 9 N latitude. It accounts Only small patches of Hydromoxphic Grassland are for approximately 7.9 percentof the region, or 163,145 mapped in this region. They form a discontinuous belt square lilometers; the largest proportion is in Nigeria at the foot of the Gambaga Scarp and along the Niger (161,27 square Idlometers). Valley in Benin, and in northern Nigeia. This area In Nigeria, this area is dominated by three vegetaextends across the inselberg-dominated landscape of tion types-"burkea africana Savanna," "Moist SudannorthwesternNieria as far as the Jos Plateau. Furtier ian Woodland," and "Wooded Savanna.- All tiree small pathes occur to the east of the Jos Plateau at a types are grass and tree savannas, the latter two clas.- similar latitude. Togetlher these areas account for ap- sified as Sudamian and the former as Guinean (Areola proximately 1.86 percet of the region (38,415 square 1982a4 In Togo and Ghana, this class falls into the kilometers), the larest area by far being in Nigeria three areas mapped as (a) "Sudanian Savanna".(33 461sqae kilometers). -,(Brunel 1981; Brookman-Arissah 1987), (b) the north- These areas mainly occur in the "Burkea fiana errmost parts of the 'Dry Forest Zone" in Togo Savanna Zone" as defined by Areola (1982a), although (Brunel 981), and (c) the northenmost parts of the some areas also occur in the norternmost parts of the Guinea Savanna Zone in Ghana (Brookman-Amissah "Isoberfinfa Savanna" and 'Wooded Savanna-' These 1987). A typical Nw curve for this class (figure 9-1) areas are sparsely populated with an attendant low shows a strongly develped seasoriaity with genercultivation density; the graslands themselves are ally lower NMv levels than the Sudni Woodlands used mainly for cattle grazing. (Casses 62,64, and 65) to the south. The fuelwood demand in these areas is widely van- The drier areas are domnnated by a short, fealthey, able. On the basis of population denity, it is greatest condnuous grass sward whiwh attains a height of only 64

77 The West African Coast Andrew C. Millington This chapter presents a detailed description of the in northeastern Ghana (100 persons per square kdlomostimportantlandcoverclassesin thisregion.help- meter) and lowest in Benin and Nigeria (2 persons per ful figures in other chapters include figure 3-1 (cloud square kilometer). In areas of greater population dencover); figures 3-2, 3-3, and 3-4 (NDwI summary land sity, fuelwood demand undoubtedly exceeds woody cover profiles); figure 3-5 (regional summary map of biomass stock, which is relatively low in these domiland cover classes); figures 7-1 and 7-2 (continental nantly grassland areas maps of growing stock and sustainable yield); and the 'Regional Land Cover.Class Map of West Africa" at Class 44-Sahel-Sudanian Acacia Wooded the end of -this volume. Bushland Helpful tables in other chapters include table 3-2 (land cover classes); table 4-1 (data and sources for Acacia wooded grassland and. deciduous bushland growing stock and sustainable yield); and table 6-2 occurs only in the very north of the region, with small (West African Coast estimated woody biomass by areas in northeastern Ghana and the adjacent parts of summary class). Togo and northeastem Benin. The class has its maximum developmentin northern Nigeria where it forms Class 12-Hydromorphic Grassland a broad, continuous belt 50 to 300 kilometers wide across the country, north of 9g N latitude. It accounts Only small patches of Hydromorphic Grassland are for approximately 7.9 percent of the region, or 163,145 mapped in this region. They form a discontinuous belt square kilometers; the largest proportion is in Nigeria at the foot of the Gambaga Scarp and along the Niger (161,827 square kilometers). Valley in Benin, and in northern Nigeria. This area In Nigeria, this area is dominated by three vegetaextends across the inselberg-dominated landscape of tion types-"burkea africana Savanna," "Moist Sudannorthwestern Nigeria as far as the Jos Plateau. Further ian Woodland," and "Wooded Savanna." All three small patches occur to the east of the Jos Plateau at a types are grass and tree savannas, the latter two clas-- similar latitude. Together these areas account for ap- sified as Sudanian and the former as Guinean (Areola proximately 1.86 percent of the region (38,415 square 1982a). In Togo and Ghana, this class falls into the kilometers), the largest area by far being in Nigeria three areas mapped as (a) "Sudanian Savanna" (33,461 square kilometers). (Brunel 1981; Brookman-Amissah 1987), (b) the north- These areas mainly occur in the "Burkea africana ernmost parts of the "Dry Forest Zone" in Togo Savanna Zone" as defined by Areola (1982a), although (Brunel 1981), and (c) the northernmost parts of the some areas also occur in the northernmost parts of the Guinea Savanna Zone in Ghana (Brooknan-Amissah "Isoberlinia Savanna" and "Wooded Savanna-" These 1987). A typical Novi curve for this class (figure 9-1) areas are sparsely populated with an attendant low shows a strongly developed seasonality with genercultivation density; the grasslands themselves are ally lower Nrow levels than the Sudanian Woodlands used mainly for cattle grazing. (Classes 62,64, and 65) to the south. The fuelwood demand in these areas is widely vari- The drier areas are dominated by a short, feathery, able. On the basis of population density, it is greatest continuous grass sward which attains a height of only 64

78 77Te Wesl Afrca i Coast 65. Figure 9-1. NOVI Profiles, Sahel-Sudanian Acacia remaining trees have multiple use, and their potential Wooded Bushland (Class 44) as fuelwood is limited. This, combined with the relatively low sustainable yield for trees and shrubs in this 0.6 class (0.35 tonne per hectare per year), means that 0.5 D.5- these areas are suffering from a shortfall in woody biomass. z Class 62-Dry Sudanian Woodland Pry Sudanian Woodland occurs in a belt covering the 0.2 / - 44 Notherstem Ghana northemmost parts of Guinea-Bissau, Guinea, Ghana, Northeastern - Togo, Benin, and northem Nigeria. The main areas in (Hoo State) Nigeria this belt are:.. ~~~~~~~(Borno Stale) -*. Jan Fcb Mar Apr May J.'n Jul Aug Scp Oct Nov Dcc * In Nigeria, a discontinuous belt runs between 7a N and 13" N, attaining its greatest areal extent in 1.5 meters. Interspersed in this grass sward are short Bauchi, Bomo, and Sokoto states. savanna trees dominated by Acacia spp. Och-er com- * In Ghana, the class extends southward in a strip mon trees are Adansonia digitata, Burkea africana, But along the Black Volta as far as Dochire and over a rospermnum parkii, Capparis spp., Conimiphor africana, large area in northeastern Ghana, which extends as Diospyros thespilifjrmis, Maytenus senegalensis, and far south as Tamale. Tamarindus indica. The dry woodlands that are * In Togo and Benin, an almost continuous belt runs mapped within this class are dominated by trees such across the northern part of these countries. In Togo as Adansonia digitata and Isoberlinia doka. These latter it occurs only north of 100 N, but in Benin it occurs savanna types covered by this class range from tree- as far south as Banikoara and KCandi as well as in a less grass savanna to savanna woodland with a more- strip along the Chaine de l'atakora. or-less closed canopy and occur mainly in the south of the mapped area. Dry Sudanian Woodland covers 192,338 square kllo- Land use in this class is dominated by extensive meters (about 93 percent of the land area), making it grazing and intensive cultivation of both food crops the fifth-largest land cover class in the region (table 9-1). (beans, millet, and sorghum) and cash crops (cotton, This is one of the driest savanna woodland classes groundnuts, and tobacco). The population density in in the WestAfrican coastal region. The only drier class parts of this zone can be great-for example, 20 to 39 that occurs to any significant extent is Sahel-Sudanian persons per square kilometer in CirconscTiption Ad- MAccia Wooded Bushland (Class 44). Although the ministrative de Dapadong in northern Togo (Gu- latter class is restricted mainly to Nigeria, Dry Sudan- Konu 1981a). Typical of these areas of dense-popula- ian Woodland is far more widespread. Brooknantion are the Kano and Sokoto Close-Settled Zones in Amissah (1987) maps Dry Sudanian Woodland as the northem Nigeria. Between the population concentra- most northerly of the "Guinea Savanna Woodlands" tdons are areas with very low densities and conse- in Ghana. In the adjacentparts of Togo, Dry Sudanian quently a greater frequency of woodlands. Woodland falls within the dry northem parts of Conversely, population pressure leads to great cul- Brunel's "Foret Seche" zone (1981). In Nigeria, it tivation density; an example is parts of Bauchi and Kaduna states in Nigeria and the Savanes Region of Table 91. Areal Distribution, DrY Sudanian Togo (Afolayan and Barbour 1982; Gu-Konu 1981b). Woodland (62), West African Coast Region Population pressure also creates high stock numbers; C an example is the more than 80 million cattle in the Clasn -Cla.si Savanes Region (Gu-Konu 1981b; National Atlas of Country (km?) (percent) (percent) Ghana). Asaconsequenceofthesehighcultivationandstock Nigeria 154, densities, the woody biomass resource in this class is Ghana 18, underseverepressureinsomeareas.landalreadyhas Togo 11,435 4, been extensively cleared for cultivation and grazing Guinea 2, over much of this class in Benin, Ghana, Togo, and in Guinea-Bissau 1, many parts of Nigeria. In areas of low population Thb1 192, density, particularly northem Migeria, the woody bio- _ mass condition is much more favorable. Many of the S- o Authors' data bases.

79 66 Eseimafiing Woody Biomass in Stb-Snaharant Africa corresponds to the northern parts of these types of Figure 9-2. NDvl Profiles, Sudanian Woodland savanna: "Moist Sudanian Woodland," "Wooded Sa- (Class 64) vanna," and "Burkc afticana Savanna" (Areola 1982a). - - The area is dominated by very open tree and shrub 6 savanna, and it is a drier variant of the closed-canopy 0.5- savanna woodland to the south. The main trees and shrubs include Adansoulia digitata, Burkea africana, 0.4. Butyrospermian parkii, Cardenia spp., Comnbretum spp., Entad africana, Hypl#aene tmebaica, Isoberlinia doka, Parkia 0.3\ spp., and Terniinalia spp. The understory grasses are 02 - dominated by Andropogon spp. and Hyparrltenia spp. - - In Nigeria, land use in this class varies from intensively 64 Stem go cultivated and grazed savanna in the east to - 64 Guinea-Mall areas of very low cultivation density and only moder-.bd ate grazing on the Benin border. (Areola 1982b). The an Feb Mar Apr May Jun Jul Aug-Sep Ocl Nov Dc intensity of cultivation and grazing reflects population- distribution (Afolayan and Barbour 1982). Crop- from the Gabti Region in eastem Guinea-Bissau to the ping is dominated by subsistence sorghum cultiva- Nigeria-Cameroon border. Sudanian Woodland is tion, but cotton and groundnuts are the main cash very extensive in Guinea, occurring on the plains becrops. Similar crop combinations exist in the other tween the Sankarani, Niger, and Tinkisso rivers, and parts of the region covered by this class. on the higher ground along the Mali border. Its occur- Population density in this zone is generally low: <19 rence in C6te d'lvoire is restricted to areas north of persons per square kilometer in Togo (Gu-Konu N where mean annual rainfalj is less than 1, a) and 2 persons per square kilometer in Nigeria millimeters. (Afolayan and Barbour 1982). Tnis suggests that fuel-. Sudanian Woodland covers large areas in northwood demand is probably quite low except around western Ghana and between Tamale and the Togo the main settlements. Nevertheless, standing stock Hills. In northern Togo and Benin, it occurs between and sustainable yield in this zone are low and, 9030' N and 1130' N. In Nigeria, it covers most of the although they probably meet current demand, any Nigerian "Middle Belt," demarking a broad tract of increase in population could reverse this favorable high ground from the Benin border in the west to the situation. Jos Plateau and then southward to include the middle Benue River Valley. Class 64-Sudanian Woodland White (1983) classifies this and a number of other land cover classes as "Sudanian Woodland with abun- This is the second most extensive land cover class in dantlsoberlinia" and a "Mosaic of LowlandRainForest the region and occurs in all rine countries except andsecondary Grassland."TheNr MOcurve(figure9-2) Liberia (table 9-2). It covers 381,040 square kilometers, shows a seasonality similar to the Guinean Woodland approximately 18.5 percent of the area, and extends (Class 74, figure 9-3) but with slightly lower NVI values during the wet season. In Guinea-Bissau, ttis and other land cover classes Table 9-2. Areal Distribution, Sudanian Wood- havebeen classified as (1)"Dry ForestTypes including land {64), West African Coast Region Tree Savannas" (2) "Foret Demi-s&he Dense et Class in Class in Claire" and "Foret Claire Degradee et Savane Boisee" Area region country Country (kitm) (percent) (percent) by scer (Bianchi 1986), and (3) "Savane Boise" and "Savane Tres Claire" by Atlanta Consulting Gmb.K Nigeria 153, (Bianchi 1986). In northern C6te divoire, Monnier Guinea 103, (1983) classifies this land cover type as "Sudanian Ghana 47, Savanna." In Ghana, this land cover is classified in Benin 40, A conjunction withvery large parts of central and north- C8ted'Tvoire 16, em Ghana as "Guinea Savanna Woodland" (Brook- Ginea-Bissau 15,480 L4 182 man-amissah 1987). In Togo, Sudanian Woodland Sierra Leone : 2, ) 0. corresponds to the southern part of Brunel's "Foret Total 381, In Nigeria, this class covers much of the 'Middle Note Details may not add to totals because of rounding. Sour=c Authors'data bases. Belt" and represents in part the "Isoberlinia Savanna," "Mixed Leguminous Wooded Savanna," "Afzelia

80 -Te WVest Afircati Coast 67 Figure 9- NDvI Profiles, Guinean Woodland and geomorphological characteristics of the area, (Class 74) which in tum are controlled by the underlying lithol- 0.6.ogy (Stobbs 1963; Millington, Helinisch, and Rhebergen, 1985). 0.5A The area is classified as "Riverain Grassland" by Cole (1968) and the "Lopliri Tree Savanna/ Boland Swamps OA f4. % Complex". by PAO (1981). The "Lophira Tree Savanna" is developed on the thin, infertle soils of the intervening z / / R. high ground and is dominated by stunted, fire-resistant Lophira lanceolata in a grass sward of Cliasmopodium - 74 Northem Guinea-Bissau caudatum.the ecological development of this vegetation O.l-. \~ 74 Northcm Slirm Lgonc is controlled by cultivation, annual burning, and inten- (Kalnadugu disuiet) sive coppicing of the trees. The seasonally flooded grass OJan Fb Mar Jun Jul Aug ' ' and herb swamps between the interfluves (known as bofis) are almrost treeless and have been extensively cleared for swamp rice cultivation. Savanna/Semideciduous Forest," and "Leg- ninous Apart from the Sierra Leonean "bolilands," this class Wooded Savanna of the Jos Plateau" (Areola 1982a). represents mature Sudanian Woodland with low-to- The savanna formations in this class vary from closed moderate levels of disturbance. Therefore, at present canopy savanna woodland (for example, "A ia- levels of fuelwood demand, wood supplies are proba- Savanna/Semideciduous Woodland" in Nigeria and bly assured in the short-to-medium tenn. However, "Foiet lde-s&he Dense' in Guinea-Bissau) to thicket the Sudanian woodlands in the West African coastal savanna. The spatial variation in savanna formation is region are coming under increasing pressure from caused by variation in both rainfall amount and disti- charcoal producers. Trees are being cut, converted to bution, and in the -moister parts, woodland is well charcoal, and then transported by road to the coastal developed. In such woodland, trees and shrubs form cities such as Abidjan and Accra, where charcoal is an a more-or-less closed canopy over grasses. important domestic fuel. The Sierra Leonean Moisture availability becomes more of a problem "bolilands" represent a worse situation where local farther north ir. this land cover class, and in such areas demand is relatively great and the locally available open-canopy savanna woodland and shrub savanna woody biomass resource is severely limited. occur. There appears to be an overlap with Dry Sudanian Woodland (Class 62), but the latter class generally Class 65-Moist Sudanian Woodland represents much drier Sudanian savanna woodland than woodlands in this class. The dominant trees are Moist Sudanian Woodland occurs in scattered patches Aftelia nftkicana, Anogeissus schimperi, Daniellia oliveri, throughout the Sudanian zone of West Africa These Isoberlinia doka, Monotes kerstingii, and Pterocarpus en- patchesoccur westward and eastward fro Omlongitude naceus. Dominant grasses include Andropogon spp., in two groups: (a) a mome-or-less continuous belt from Cteniumelegans,Hyparrheniaspp., and Loudetiasuperba. eastern Guinea to north-celtral Ghana and (b) an area The area covered by this class varies in population from central Togo to the Cameroon-Nigerian border. density. But, apart from the area around principal The first area includes the northern flanks of the towns, density rarely exceeds 50 persons per square Guinea Highlands (the Dongoroma, Going, Kourankilometer and is mostly <10 persons per square kilo- dou, and Tourou ranges). It also encompasses much meter. Cultivation density is therefore moderate, as of C6te d'ivoire north of 8 N, including a number of are stock numbers. The main subsistence crops are areas above 400 meters elevation (for example, the sorghum and yams. Important cash crops irclude cot- Chaine de Tiemd, Chaine de Madinani, and the Mont ton, groundnuts, and beniseed (Berron and Vennetier Yv6l&-to-Mont Bowe de Kiendi axis) in the Regions 1983; Gu-Konu 1981b; and Nwafor 1982b). du Nord and de l'est. In the northem region of Ghana, The only large area mapped as this class that does the area extends as far east as Lake Volta. not correspond to the general pattem of Sudanian The second area contains many smaller areas. Two Woodland isin SierraLeone. Here, a tractin thenorth- main concentrations of these occur in Benin: (a) a thin em interior plains is placed in this land cover class. strip of land between Tchaourou and Kalale, which This area corresponds to the "bolilands." This is an includes the towns of Parakou, Pr6r6, andniki, and extensive area of seasonally flooded hydromorphic (b) the densely populated area to the southeast of grasslands and swamps, with degraded Guinean sa- Chaine de l'atakora, which includes the towns of vanna woodland on the intervening high ground. The Natitingou, Ouake, Kopargo, Dougou, Pehonko, and vegetation pattem is controlled by the hydrological Kouandc.

81 68 Estitma4ting Woody Biomna.ss in Sub-Saanmri Africa Table 9-3. Areal Distribution, Moist Sudanian for grazing. However, around both Enugu and Ilorin Woodland (65), West African Coast Region in Nigeria, edaphic and derived-savanna grasslands, Class ill Class inl respectively, are extensively cultivated. This area is A rea region couitntry one of mixed-grain and root crop agricultural systems, Colunitry. (knila) (percell) (percenl) the main subsistence crops being maize, millet, sor- Nigeria 121, ghum, and yams (Berron and Vennetier 1983; Cu- C6ted'lvolre Guinea 43,474 24, onu1981b;Nwafor1982) In thesouthem areasof thisclass, tree and rootcrops Ghana 22, are more common, whereas in the north, cereals pre- Benin 14, dominate. Cash cropping is less common than in Togo - 6, many other savanna woodland zones in West Africa, SierraiLane 53u > although significant tobacco cultivation occurs in Sierra Len3.northwestern Cote d'ivoire (Berron and Vennetier Total 233, ). Crazing is common, although herd sizes gener- Notr: Details may not add to totalsbecauseof rounding..ally are much smaller than in the drier savanna wood- Sotmre: Authors' data base. land zones to the north. Population density varies from <4 persons per Much of central Nigeria also falls into this class. In square kilometer in many parts of C6te d'lvoire the south, Moist Sudanian Woodland is limited by the (Lecomte and Monnier 1983) to 100 persons per square coastal plains, and it does not extend north of 10 N. It kilometer in parts of Togo (Gu-Konu 1981a). The spais the fourth-largest land cover class in the region, tial variation in both woody biomass and population covering 233,707 square kilometers. It accounts for withinthemoistsudanianwoodlandzonemeansthat approximately 11.3 percentof the land area and occurs the firewood supply is extremely variable, ranging in all countries except Liberia (table 9-3). from significant surplus to severe shortage. Moist Sudanian Woodland is equivalent to a number of vegetation classes mapped by Areola (1982a) in Class 74-Guinean Woodland Nigeria-"M[ixed Leguminous Wooded Savanna," "Afrelia-Savanna/Semideciduous Forest," "Moist The area covered by Guinean Woodland in this region Sudanian woodland," and parts of the "Forest-Savanna is very extensive-243,241 square kilometers or about Mosaic." These are mainly savanna communities, the 11.8 percent of the land area. It occurs in all countries exception being the "Forest-Savanna Mosaic," which of the region except Liberia (table 94). is a transitional class between the rain forest and the Asmapped, thisclass isprobably synonymots with savanna woodland to the north. White's "Guineo-Congolian Secondary Grassland In Cote d'lvoire, this class corresponds to thenorth- and Wooded Grassland" (1983) and occurs in a belt ermmost parts of the "Sub-Sudanian Savanna" (Monn- varying between 50 and 950 kilometers wide from ier 1983).These areas are mainly grass and herb savan- the Guinea-Bissau-Senegal border to the Nigerianas with either trees, shrubs, or thicket formations. In Cameroon border. In Guinea-Bissau, Guinean Wood- Ghana, this area forms part of the "Guinean Savanna land dominates the higher ground away from the Woodland," but is not differentiated from other savanna types by Brooknan-Amissah (1987). In Togo, Moist Sudanian Woodland is mainly restricted to hills Table 9-4. Areal Distribution, Guinean Woodland above 400 meters elevation, and is classified as "Forat (74), West African Coast Region Sche" (Brunel 1981).. Class in Class in Despite thefactthatthiszoneisnamedafterasingle Area region country tree, Isoberlinia doka, by many workers (including Country (knr) (percent) rperceno White 1983), it is not a single-dominant community. Guinea 63, Other equally important trees occur, such as Afzelia C6ted'lvoire 59, A africana, Anogeissus schimperi, Daniellia oliveri, Khaya Nigeria 53, senegalensis, Monotes kerstingii, Pterocarpus erinaceus, Sierra Leone 30, A and Uapaca togoensis. Important shrubs and small trees Guinea-Bissau 14, in drier areas include Butyrospermum parkzi, Cardenia T;ao Ghana 8,958 7, spp., Combretum spp., Entada afrcana, ParTia spp., and Benin 5, Terminalia spp. Cultivation density is much lower in this zone than Total 243, inthe woodland tothe south and nortlh In many areas, Note: Details may not add to totals because of rounding. extensive tracts of savanna woodland are used only Source: Authors' data bases.

82 Tlic West Afnicani Coast 69 coastal plains (for example, in Bafatd, Gabia, and Oio At a small scale, the distribution of trees and shrubs. regions). In Guinea and Sierra Leone, it occursini three is controlled by edaphic factors (Clayton 1958, 1961;. distinct areas: Hambler 1964; Jones 1963; Keay 1959; Menaut and Cesar 1979,1982; Morison, Hoyle, and Hope-Simpson * From the northem coastal plains of Gtuinea inland 1948; Ramsay 1964; Ramsay and dc Leeuw 1964,1965; as far as Koundara. White 1965). Trees and shrubs usually are best devel- * In an area stretching northward from Bo in Sierra oped on the most fertile soils and here the woody Leone to the Fouta Diallon in Guinea and then back biomass is greatest; however, in these areas, competisoutheastward to Kissidougou in Guinea. This area tion from grnsses is also imnportant and fire is an imincludes the interior plateaus and hill region -of portant factor in savanna formation (Menaut and Sierra Leone and Guinea. Cdsar 1982). Stands of palmns, dominated by Borassus * An area extending from Nzerdkord in Guinea into act hiopuim, are widespread and are tolerant of a wide eastern C6te d'ivoire. range of environmental conditions. Menaut and Cdsar (1979,1982) have reported exten- In Cote d'lvoire, Guinean Woodland covers most of sively on Guinean savannas from the Lainto study the ground between 300 and 400 meters above sea level area in central Cote d'lvoire (the UNUSC Man and Binorth of 7 40' N, with the exception of the Mt. Bendi- sphere or MAD Programme). At Lamto, a number of ML Ko axis and.some areas in the extreme northeast. vegetation communitieshavebeen identified that typify The vegetation is poorly developed in Ghana and. the potential diversity of woody biomass stocks in this Beninbutisextensive, albeit ratherpatchy, intogoand and similar land cover classes (table 9-5). These var- Nigeria. The main occurrence in Nigeria is on land iations are mainly related to topographic and soil higher than 300 meters elevation on the Abeokuta- corditions. lbadan-ondo axis, along the Niger Valley as far north On the highest ground at Laamto, freely draining as the Jos Plateau, and on the Bamenda and Mandara ferralsols are dominated by "Andropogoneae Sava=nhills. The NOVI profiles (figure 9-3) show the seasonality nas,' of which five types are recognized. These are all of the vegetation and its intermediate positionbetween dominated by Andropogon spp. grasses. They are clasthe forest zone and the drier woodlands. sified by woody biomass stock and, in order of de- Rain forest trees within these savannas are either creasing stock, they are: remnants or invasive. The communities themselves are extremely heterogeneous, ranging from grassland Svanna a* Woodland, in which the trees form a conthrough shrubby thicket to well-developed savanna tinuous canopy between 2 and 6 meters, although woodland and gallery forest. Three strata canbe iden- palms often emerge to heights of 10 to 12 meters. tilled in thesavannas, although they are not present in The trees are underlain by a dense grass sward. all of the communities:. * Dense Shrub Savanna, which has a lower overall tree and shrub cover varying between 20 and 25 * A lower stratum of grasses, herbs, young saplings, percent, with shrubs attaining a maximum height and shrubs attaining 2 meters in height of 8 meters. * A middle stratum of shrubs and small trees which * Open Shrub Savanna, which occurs in areas with vary between 2 and 8 meters better soil condition than the transitional shrub and * Emergent trees and palms that exceed 8 meters. grass savannas. Consequently, the shrub species Table.9-5. Woody Biomass Stocks and Growth Rates, Guinean Woodland Communities, Lamto, Cote d'ivoire Intermediate Open Dense Savanna Parameter shrubsavanna shrubsavanna shrub savanna woodland Individuals Canopy cover (percent) Leaf area index CA 1 Woody biomass Above ground (kg/ha) 7,400 21,900 32,600-54,200 Below ground (kg/ha) 3,600 9,200 1,430 26,600 Woody biomass growth rate Above ground (kg/ha/yr) Below ground (kcg/ha/yr) SOur7!. After Menaut and Cdsar 1982.

83 70 ElsimaIing Woody Biomu in Sub-Saharan Africa are better developed, attaining ground covers of 15 cattle owners (between 1951 and 1963), and by dryto 20 percent. season farming promoted by small irrigation schemes. * Intermediate Shrub Savanna, which has a very low The best examples of moist closed-canopy woodland woody cover (less then 10 percent), with individual now exist around villages, where they are reserved for shrubs ranging from 3 to 6 meters in height. a variety of reasons. * Grass Savanna, which occurs where shallow In the "Southern Guinean Savanna" of Sierra Leone ferricretes restrict tree and shrub growth, consisting and adjacent areas of Guinea, land use is dominated by exclusively of grasses up to 2.5 meters in height. Fula pastoralism, and thegrassunderstory isextensively grazed. In northwestern Sierra Leone, the unusual On the hydromorphic gleysols, which characterize "Lophiim Tree Savanna" occurs in this class (Cole 1968). the slopes down to the rivers, Loutdetia simplex grass- Here, a combination of infertle thin soils, intensive culdominated savannas occur. These have a very poor tivation, and fierce annual burning have created an ecowoody element (mainly palms and shrubs on termite logical community completely dominated by Lophira mounds) because of various edaphic constraints re- lanceolata in a grass understory of Chasmopodiunt caulated to waterlogging in the gleysols. datum. The trees have been intensively coppiced, both Menaut and Cdsar (1979, 1982) also have investi- dunng agicultural clearance and for fuelwood. Only gated the role of burning on the evolution of tihese shrubby regrowth remains in most areas, In slightly less savannas. On the one hand, Loudetia savannas have a disturbed areas, however, the trees attain 7.5 to 10 meters sharp boundary between savanna and forest, espe- in height, and at the edge of the "Lophira Tree Savanna" cially in burned areas. On the other hand, An- othersavanna trees are invasive. dropogoneae savannas have a gradual increase in Many parts of the Guinean Woodland are sparsely woody component from open grassy areas to savanna populated, and thus have a low cultivation density. woodland. The foret.t-savannabourdary is dependent Other areas have a high cultivation density, mainly for on the date, duration, and intensity of fires and there upland (rain-fed) rice in the wetter areas and sorghum is no one definite regional succession, and millet in the drier areas. Vegetation is cut, dried, In some areas, forest trees establish themselves in and bumed to create fields that are commonly cultiswamps, on termite mounds, or on rocky outcrops. vated for up to 2 years. The woodland is then left to These communities then act as precursors for forest regenerate as a woody fallow. invasion, once burning ceases. In other areas, gallery Although older woodland regrowth resembles the forest trees may invade the adjacent savannas after climax savanna conmmunities in structure and floristic herbs have invaded. As the trec canopy closes, the relations, the derived savannas are characterized bv shading effect restricts grn.s growth, less fuel for the the frequent occurrence of the economically important annual fires is produced, and fires therefore do not fire-resistant oil palm (Elaeis guineensis) and a greater penetrate the invasive forest stands. proportion of fire-resistant trees because of the fre- Cole (1968) recognizes a moist savanna woodland quent burning. Fire-resistant and fire-tolerant succesin northeastem Sierra Leone which he tenns the sional trees (for example, Lophiralata, Parkia biglobosa, 'South Guinean Savanna." It grades from moist Piliostigma thonningii, Pterxarpus erinaceus, and Syzygwoodland on the wetter soils to a shrub or grass ium guineense) first occur in the younger regrowth savanna on mountain slopes. Moist woodland and stages and become subdominant in older successions. open tree savanna are the dominant ecological struc- They are always an important component in the Guintures: In the moist woodland and tree savanna, the ean Woodland communities, however, because of the treesaregnarled,stunted,fire-resistantdanielliaolivei annual burning of the understory grasses to create and Lophtira lanceolata, attaining heights of 10 to 17 better grazing for cattle in the woodlands. meters, growing in a dense grass understory. On thin- The Guinean Woodland class is dominated mainly nersoilswithlesscapacityforholdingmoisture,shrub by well-developed closed-canopy woodland with a and tall grass savannas dominate. In the "South Guin- moderately high sustainable yield. This, combined ean Savanna" zone, gallery forest mixes savanna trees with the variable butgenerally low populationdensity (for example, Cussonia longissima, Parkia biglobosa, and (2 persons per square kilometer), suggests that few Terminaliaglaucescets) with rainforest trees (especially woody biomass supply problems exist, although Chlorophr.a regia, Erythrophleum guineense, Parinai ex- localized problems undoubtedly occur (for example, relsa, and ltapacaguineensis). in the "Lophira Tree Savanna' in Sierra Leone). In Cole argues tihat moist semideciduous woodland is many West African states, this situation could change the climax comnmunity on the best soils, bult that deg- with a switch in charcoal production for coastal radation of these woodlands in norhern Sierra Leone towns from Sudanian Woodland (64) to Guinean has been caused by dry-season buming to promote Woodland. The high-density wood theat is characterbetter grazing, by settlement schemes for nomadic istic of trees in the Guinean Woodland would pro-

84 The West African Coast 71 duce high-quality charcoal if the supply from Sudian- tains on the Sierra Leone-Liberia border and central ian Woodland became scarce. Ghana, and (b) in southern Nigeria, mainly in Bendel State and the Cross River Basin. Small Overview: West African High Woody patches also occur in eastern Guinea, central Cote Biomass Mosaics d'ivoire, and in the Atacora Mountains on the Ghana-Togo border (table 9-6a). The class is mainly The humid tropical forest zone in West Africa is restricted to the lowland plains and plateaus of the densely populated, and only in a few areas do exten- coastal belt. For example, in C6te d'ivoire, it occurs sive tracts of primary or mature secondary forest re- mainly within 150 kilometers of the coast at elevamain. Much of the area is clultivated, the two main tions below 200 meters, although it also exists on agriculturalsystemsbeingbushfallowingandplanta- Mont Bowe de Kiendi and Mont Nangbion. Other tion cropping. Consequently, the land cover of the occurrences on higher ground are, for example, in region is a complex mosaic of primary forest and the Loma-Man Dorsale. mature secondary forest reserves, various stages of The class covers 101,438 square kilometers, about forest regrowth, cultivated fields, and plantations. The 4.9 percent of the total area. Itsgreatestextent is inc8te fragmentary nature of land use is such that individual d'ivoire (47,584 square kilometers), followed by Libeelements cannot be routinely identified and mapped ria (24,925 square kilometers) and Ghana (18,812 using 8-kilometer resolution data. square kilometers). After extensive discussion with foresters and land Rainfall is great over most of the area mapped in this. use planners from West Africa, we mapped the area class, ranging from about 1,400 to more than 4,000 as a high woody biomass "Mixed Agriculture and millimeters a year in most areas. In Ghana, however, Forest Fallow Mosaic." We divided the mosaic into the class occurs in areas with as little as 1,000 millimetwo phases, based on overall level of productivity: ters annual rainfall To a large extent, the greater productivity of vegetation in this class is due to rainfall High-Productivity West African Cultivation and But evidence also exists for a minor depression in Forest Mosaic (Class 75), corresponding roughly to photosynthetic activity in the wet season, approxithe evergreen or ombrophilous humid forest zone mately in August, in Nigeria, Togo, and Ghana, a * Medium-ProductivityWestAfricanCultivation and period known as the "little dry season." In the more Forest Mosaic (Class 76), corresponding roughly to westerly areas covered by this class, rainfall is rethe semideciduous or mesophilous forest zone. stricted to a single wet season with a dry season ranging from 4 to 6 months. The pattern of vegetation Class 75-High-Productivity West African response to rainfall is reflected in the NDvi profiles Cultivation and Forest Mosaic (figure 9-4). Much of the humid tropical forest that once covered This mosaic occurs along the coastal belt of West Af- this area has been cleared. White (1983) reported three rica in two main areas: (a) between the Gola Moun- types of wet lowland rain forest in this area; 'Hygro- Vable 9-Sa. Areal Distribution High-Productivity Table 9-6b. Areal Distribution, Medium- Cultivation and Forest Mosaic (75), West African Productivity Cultivation and Forest Mosaic (76), Coast Region West African Coast Region Class in Class in Class in Class in Ara region cuntry Area region country Country (km 2 ) (percent) (percent) Country (rm 2 ) (percent) (percent) Cote d Ivoire 47, Cote d'ivoire 150, Liberia 24, Nigeria 132, Ghana - 18, Ghana 99, Nigeria 7, Liberia 60, Togo 1, Benin 38, Guinea 1, Guinea 31, Sierra Leone 369 OA 0.5 Sierra Leone 27, Benin Togo 18, Guinea-Bissau Guinea-Bissau 1, Total 101, Total 560, Notr Details may not add to totals because of rounding. Note: Details may not add to totals because of rounding. So. Autlohos' data bases. Source: Authors' data bases *t d. -.0

85 72 -Esiiirnlng Woody iottmass in Sub-So/laran Africa Fig,ure 9-4. NDVI Profiles, High-Productivity West "Mixed Moist Sendi-Evergreen Guineo-Congolian African Cultivation and Forest Mosaic (Class 75) Rain Forest" 0.6 This type is less developed in West Africa than -Hygrophilous Coastal Evergreen Guineo-Congolian 0.5- / Rain Forest." White (1983) attributes this to the abrpt 0.4- DA t \ transition from very wet coastal conditions to the dry 0,4r \ - interior. Areola (1982a) describes "Moist Lowland 0.3- / - - \, - Forest" in southern Nigeria as typically having the -thre layers associated with tropical rain forest: the 0,2- upper layer of emergents 40 to 50 meters high, a mid- 75 Southwestem Ghana Southwestern MCe d'lol dle layer of trees betw e.n 16 and 40 meters, and a lower layer of trees varying between 10 and 16 meters. 0 an MrArMy(EindSdQus)The -Jan Fb Ma Apr May Jun Jul Aug Sep Oct Nov Dc uipper and middle layers fodrm one continuous canopy and the lowerlayerformsasecondcontinuous canopy. Underneath are shrub and herb layers with saplings of the taller trees. philous Coastal Evergreen Guineo-Congolian Rain In this class, the mean deciduousness in Sierra Forest," "Mixed Moist Semi-Evergreen Guineo- Leonean forest reserves was 23.5 percent (Cole Congolian Rain Forest," and "Single-Dominant Moist 1968), although other forests classed as evergreen by Evergreen and Semi-Evergreen Rain Forest.' Little Cole had deciduousness values from 16.5 to 26.2 forest now occurs in the region because of extensive percent. cultivation of food and plantation crops. Forest still This type of forest corresponds to Monnier's occurs, however, in less densely populated areas and "Foret Ombrophile" (1983), which occurs south of in forest reservations. the 1,600- millimeter isohyet in C6te d'ivoire. It is characterized by emergents such as Lophira alata, -Hygrophilous Coastal Evergreen Guineo- which attain a height of 50 meters. hn Sierra Leone, Congolian Rain Forest" this type of forest occurs in the Cola Mountains on the Liberian border and spills across the border into Of the three forest types, this was the most common Lofa County in Liberia. Other interior evergreen in West Africa. Hall and SwaLte (1976) describe humid tropical forests in Liberia (in Montserrado, 'Hygrophilous Coastal Evergreen Rain Forest" as Grand Bassa, Grand Gedeh, and Sinoe counties) shorter than "Mixed Moist Semi-Evergreen Rain For- probably fall into White's "Mixed Moist Semi-Everest," with an upper canopy attaining 30 meters and a green Forest" class. Cole (1968) notes that when the few emergents attaining 40 meters. Gola Forest was first enumerated in 1923, "...most The evergreen characteristic of the forest is caused of it was virgin primary rain forest and even areas by most species shedding their leaves intemrmittently, which had been disturbed by man, had secondary although in areas with a pronounced dry season (for forests of considerable age." example,liberia and Sierra Leone), manyspeciesshed Dominant canopy trees are Ant/sonotha fragrans. their leaves simultaneously and are immediately re- Combretodendron spp., Cynometra leonensis, Chrysophylplaced withnew ones. This type of forest is equivalent lum spp., Diospyros spp., Entandrophragmaspp., Guarea to "ForAt Hyperombrophile" in Cote d'ivoire (Mon- spp., Lophira alata, Nauclea diderrichii, Oldfieldia africnier 1983), which occurs only in areas with a mean ana, and Tarietia utilis. annual precipita don exceeding 1,800 millimeters and a reduced dry season. "Single-Dominant Moist Evergreen and Semi- This type undoubtedly occurs in Liberia, but Jansen Evergreen Guineo-Congolian Rain Forest" (1972) does not differentiate it from other evergreen humid forest types. He does, however, record "Wet This type is interspersed within the other two types. Coastal Rain Forest" vegetation dominated by Tetra- Single-dominant forest is much rarer in West Africa berlinia tubmaniana in a belt that stretches inland about than in the Zaire Basin and is mainly restricted to the 80 kilometers between River Cess and Greenville in wetter evergreen forest along the coast White (1983) Liberia. The canopy dominants of this forestare Crudia notes that only two of the dominant trees in the Zaire gabonensis, Didelotia unifoliolata, Gilbertiodendron pre- Basin extend intc southem Nigeria: Gilbertiodendran ussd, Gluma ivorensis, Lophimalata, Mapania spp., Tar- dewevrei dominates some fringing forests or swamp ietia utilis, Terminalia utilis, Tieghemella (Dumonria) fric- forests, and eulbernardia seretii occurs within the ana, and Sacoglottis gabonensis. coastal evergreen forest. Single-dominant forest is

86 Tire WesI African CoasI 73 Figure9-5. Cbted'lvoireWoodProduction, biomass reserve mean that fuelwood supply problems are, at worst, only localized. 5 \ The main problem facing the woody biomass resource in this area is exploitation of hardwood tim- 4 - lber by local contractors and under concession to foreign companies. Valuable hardwoods such as Oldfieldia - 3 -afjrcana, Khaya spp., Enlandropbrgrngna spp., Turraeanthus africana, Cluloroplrora excelso, and Tiegherneila hockelii all are harvested from these forests (Cole 1968; Arnaud 1983). The trade in timber is of great importance in _ Ghana, Cote d'ivoire, and Liberia (for an example in I - Uxpoit Cote d'lvoire, see figure 9-5). Comparison of the qual- 'R - E 'ity of timber felled in Cote d'ivoire in 1980 and Torn shows a slight deterioration in quality (figures 9-6 and Industral 9-7) that might be expected with overexploitation. 0-~ : Class 76-Medium-Productivity West African Cultivation and Forest Mosaic dense, typically 35 to 45 meters high, and is composed almost entirely of one species. This mosaic occurs in all nine countries in the region Some areas in the high-productivity zone still have (table 9-6b), extending from the Tombali Region in extensive mature secondary evergreenhumid tropical Guinea-Bissau to the Nigeria-Cameroon border. In forest These areas always occur where population Guinea-Bissau, it is restricted to the Tombali Region. density is low. For instance, in southwestern Cate An extensive area in Guinea runs from the Fouta d'ivoire (Region du Sud-Ouest), population density is Djalion foothills to the Forecariah Plains. The coastal <10 persons per square kilometer, and most of these and interior plains of southern Sierra Leone and Libepeople live in a few coastal towns or along roads. ria form another large contiguous area, including the Consequently, cultivation density is low andlivestock forested southern interior plateau, parts of the Gola, herds are small in these areas. Gon, and Nimini hills, and the Upper Moa Basin of In Sierra Leone, Liberia, Cote d'ivoire, and Ghana, Sierra Leone. population densities are much greater and forest clear- The mosaic continues across much of the southern ance for arable and tree crop cultivation is common. half of C6te d'ivoire into Ghana, where three main The main crops are nce and cocoa, although some areas exist: (a) on high ground to the west of Sunextensive rubber plantations in Liberia are mapped in yani, (b) around Lake Volta in the center of the hisciass.thelowpopulationdensityandhighwoody country, and (c) on the Accra Coastal Plains. This Figure 9-6. Cote d'lvoire Wood Production by Figure 9-7. Cote divoire Wood Productionf by Class, 1980 Class, O-a. Mixed woodsofsuperor 2 la Mixedwoods orsuperior quality including rcdwoods: quality including redwoods: lb. Othermibxedwoods or b.- Other mixed wood.sof supeirualy quality..spror 27 M3wtomwry, 2. Mixed woods; iney. 2 ixedywoods: faier plywood. furnitur1--el. e I Supedier tliiyjoinery woods. 3. Supe nery woods. 4.Jinrwod. 4. iie~wos S. 5ttecrsandmouldings. 5 -necrs ndmouldings. o. Hrdwods (reainsandjoits). 6. Hardwds (fiaames and joists). 7. Other c.7. Other clses. ID E1000E C 1D ~ ~ ~ ~ ~ ~ lb. ~ 2. ~ ~~~~I n.i ,: E , -;~~~~~~~~- R - ~ ~ ~ ~ ff lb. 2X, ; la. l. :2. 3-4X X-. : :: - g f i ~ ~ *.

87 Tlx West African COast 75 Figure 9-A NovI PrOfileS, Medium-Productivity narrowcoastalstripattabou,no-mer,lagunetalo, West African Cultivation and Forest Mosaic Lagune Ebrie, dnd Lagune Ehi. These savannas are (CIass 76) dominated by a dense grass cover of Bnichiana brachy- -lapa, Hyparrlaenia cdiryjsara, aand Loudetia piragmitoides. Trees occur in the grasslands and, in a few places, are remnants of large forests with typical rain forest trees such as Brideliaferginea and Ficus capen- OA sis, although Adjanohoun (1962) ascibes thir origin to wooded savannas Two hydromorphic graland communites have been recognized in Sierra Leone: 'Riverain Grass- 02 tpidehn duufflsa c lands" and "Grass-Herb Swamps" (Cole 1968). River ain grasslants occur extensively along the lower 7 (W du Ceitmi ck - -OanFdar Ap M nl.g e courses of the Sewa and Waargje rivers and are cominated by grasses, sedges, and herbs, -the only commonly occurring shrubsbeinganthostemasegalens, C-..l rto iafic#wia, Croton scardesii, Dissotis cornifolia, shrubs. However, witiin the grasslands, shmbby and Hyptis spzcige,. The wetter parts of the coastal thckets occur on old trmite mounds. These thickets. savanna zone of Liberia around Greenville, Harper, form a dense, cdsed anopy of Capparis aytnocpr, and River Cress are also mapped in this class. These Flcirtia zndiaz fvescns), Grewianpulwa, Seawqa are assumed to be areas of denved savanna (Voorwwuz, Uaria c1wa, and Anmiylnm xmllwxylods hoeve 1968) or hydromorphic grassland. about 5 meters high with airirts up to 10 metec of Elaphwibia dnqpxenz and Dkymos mspiljbrmis. Origin On irgs, patches of evergreen and semi-evergren coastal forest have developed since the cessation In its northeumnost occurrenes, this land cover class of farming (Swame, liebemian, and Hall 1990) These forms one of the main ecotones between the humid aremultlayeredlike therain forest, althoughtheyrarely tropical forest and the wetter savanna woodlands to exceed 10 to 12 met in heigh The dominant canopy the norti aracteristically, this area contans both trees are CynomeU megaphylla, Diaspyns abyssinac, D. rain forestand savanna elements. The former occur as espimnnis, Mnlhla obouxl, and Miettia thoiningil small areas of senidedduous rain forest and gallery The lower layer is chay- composed of Dnjpetaflori- forestalongnversthataresetinlargetractsof wooded bunda, D. pv*lia* and Vepris heiero* la Emergents savanna. such as Anti aristoxicaria, Ceiba pentandra,, Cellis One body of researches says that much of the samldbmalki and Nesogordmia papea also exist. vanna is maintained by fire and terms it "derived The ecology of a forest on an inselberg differs from savanna.' They argue that, if bumrig ceased, the rain that of the rain forest because irselberg forest is forest would invade dtese areas and they cite evishorter, lacks evergreen tree species, and has a poor dence of remnant humid tropical forest areas in the regeneration rate from seed. he agricultural pattem. savanna and the exisene of fire-tolerant and fireis similar to that elsewhere in the land cover lass. resistant trees to support this hypothesis- Anotherbody of researchers questions this interpre- Grassland tation on the basis of the sharpness of the forestsavanna boundary and the lack of penetration of fires *Guineo-Congolian Edaphic Grassland," which oc- into roist forests This lends credence to the hypothcurs on hydromorphic soils throughout the region, is esis that,beforecultivationandburningintheseareas, alsomapped withinthis land cover cass.wlite (1983) a continuum of vegetation existed from rain forest to suggests that most of these grasslands represent tran- dry savanna (lioplins 1974; Keay 1952; Swaine, Hall, sitions between aquatic vegetation and forest, -the and Loc 1976). grasslnd structurebeing maintained by annual bumin& Meraut (983) co Qsiders these to be 'Coastal Population and Agriculture Grass Savanas"k in C6te dlvoire and suggests that many are.fire-prodimax communities derived from Population density varies in the area covered by this forests or wooded savannas (Adjanohoun 1962). class. Where it is low, forest and woodland areas Nevertheles, truly hydromorphic grasslands rec- daminate, but in areas of greter -pulation density, ognized in Coe d'lvoire (7Savanes de Basse C6te" or forests are cleared extensvely. M.L i large cities and "Savanes Lagunaims) occupy small tracts along a towns in the region are situated in this class, and.. -,*.. -,..

88 Thie West African Coot 75 Figure 9-a NDVI Profiles, Medium-Productivity narrow coastal strip attabou,nero-mer, Lagune Taio, West African Cultivation and Forest Mosaic Lagune Ebrie, dnd Lagune Ehi. These savannas are (Class 76) dominated by a dense grass cover of Brachiiaria bnrcly- - lopha, Hyparrhenia clirysargyrea, and Loudetia phragmitoides. Trees occur in the grasslands and, in a few places, are remnants of large forests with typical rain : 1 - DA/forest trees such as Brideliaferrugi'nea and Ficus capen- OA - < 7 \ / / \\ sis, although Adjanohoun (1962) ascribes their origin 0.3 sc3' \tr.."' to wooded savannas ii 'Two hydromorphic grassland comnmunities have outheastern. S-erra A So(Pujhurn disiact) - -been recognized in Sierra Leone: "Riverain Grasslands" and "Grass-Herb Swamps" (Cole 1968). River Western Cenirni C6te divoiro ain grasslands occur extensively along the lower 76 Westem Ccntml Cbto dlvivde (Rcgion du Centre-Oucst). courses of the Sewa and Waanje rivers and are domio-. -1-,J,J,, J,Novnated I by grasses, sedges, and herbs, the only com- Jan Fe bw Ma pr day Jlma Jual.Aug Sep Oct Nov Dec..n.eb Mar A.r May Aug Sep Dee monly occurring shrubs beingant hostema senegulense, Clappertoniaficifolia, Croton scarciesii, Dissotis cornifolia, shrubs. However, within the grasslands, shrubby and Hyptis spicigera. The wetter parts of the coastal thickets occur on old termite mounds. These thickets. savanna zone of Liberia around Greenville, Harper, form a dense, closed canopy of Cparis erythmrpos, and River Cress are also mapped in this class. These Flacourtia indica flgvesces), Grewia c imfua, Securinega are assumed to be areas of derived savanna (Voor- - irosa, Uvana chamae, and Zanthoxylumn xanthoxylokdes hoeve 1968) or hydromorphic grassland. about 5 metes high with emergents up to 10 meters of Ek-kwphorbia drup*fr and Diospyros mespilifoniis. Origin On inselbergs, patches of evergreen and semi-evergreen coastal forest have developed since the cessation In its northermnost occurrences, this land cover class of farming (Swaine, Liebemnan, and Hall 1990). These forms one of the main ecotones between the humid are multilayered like the rain forest, although they rarely tropical forest and the wetter savanna woodlands to exceed 10 to 12 metes inheight The dominant canopy the north. Characteristically, this area contains both trees are Cynometra megalophylla, Diospyros ayssinica, D. rain forest and savanna elements. The former occur as mespiifjbrnis, ManilAm oborzta, and Miffettia thonningii. small areas of semideciduous rain forest and gallery The lower layer is chiefly composed of Dnjpetes flori- forest alongrivers that are set inlarge tracts of wooded bunda, D. paruifolla, and Vepris heterophyl Emergents savanna. such as Antiarfs toxicaria, Ceiba pentandra, Celtis One body of researchers says that much of the samildbraedii, and Nesogordonia papavejfera also exist vanna is maintained by fire and terms it "derived -The ecology of a forest on an inselberg differs from savanna." They argue that, if burning ceased, the rain that of the rain forest because inselberg forest is forest would invade these areas, and they cite evishorter, lacks evergreen tree species, and has a poor dence of remnant humid tropical forest areas in the regeneration rate from seed. The agricultural pattem. savanna and the existence of fire-tolerant and fireis similar to that elsewhere in the land cover class. resistant trees to support this hypothesis. Another body of researchers questions this interpre- Grassland tation on the basis of the sharpness of the forestsavanna boundary and the lack of penetration of fires "Guineo-Congolian Edaphic Grassland,' which oc- into moist forests. This lends credence to the hypothcurs on hydromorphic soils throughout the region, is esis that, before cultivation and burningin these areas, also mapped within this land cover class. White (1983) a continuum of vegetation existed from rain forest to suggests that most of these grasslands represent tran- dry savanna (Hopkins 1974; KIeay 1952; Swaine, Hall, sitions between aquatic vegetation and forest, the and Lock 1976). grasslandstructurebeingmaintainedbyannualburning. Menaut (1983) considers these to be "Coastal PopulationandAgriculture Grass Savannas' in Cote dlvoire and suggests that many are fire- prodimax communities derived from Population density varies in the area covered by this forests or wooded savannas (Adjanohoun 1962). class. Where it is low, forest and woodland areas Nevertheless, trly hydromorphic grasslands rec- dominate, but in areas of greater.opulation density, ognized in COte divoire ("Savanes de Basse Cote" or forests are cleared extensively. M. zy large cities and "Savanes Lagunaires") occupy small tracts along a towns in the region are. situated in this class, and

89 76.sliinatiung Woody B i(amss in Suib-Safiaram Africa around them the anthropogenic impact is severe. Due zygia, Dic)lirostacltys glomerata, Hartungana adacaneinto the continued forest clearance and fuelwood de- sis, Tmrna guincensis, and Xylopia quintasii grow very fast mandwithinthezone,thismosaicisaprincipalsource in the early successional s t ages in response to large of fuelwood in the region. Not only is local demand amountsofsunlight. Othercommon foresttreesatthis from within the class great, but the land cover class is stage are Ceiba pentandra, Chlarophora regia, Cola nitida, probably a net exporter of fuelwood to the cities and Eiaissguinwnsis, Mangifera indica, and Steramui tragacautba. wood-deficit areas of the region as a whole. Similar agricultural systems to that described for Extensive subsistence and cash crop agriculture Sierra Leone exist throughout this mosalc, and can be occurs in the area covered by this class. Subsistence termed bush-fallowing agricultural systems. In these, agriculture is dominated by upland and swamp rice the forest regenerates in the fallowing period, allowcultivation in Sierra Leone, Liberia, and the south of ing soil nutrient levels and structure to recover. For CBte d'ivoire. In Guinea and the central areas of instance, in Cote d'ivoire, the "Secteur Prdforestier' C6te d'ivoire, the main subsistence crop is rice, al- (Monnier 1983) is part of this class; it falls entirely into though significant areas of bananas, plantains, cas- the "Zone de Iigname." In Togo, this area is very sava, and yams are grown. These latter four crops extensive,stretchingfromthecoastalplainsto8 30'N, are also typical of the mosaic in Ghana, Togo, and and falls into four agricultural zones mapped by Nigeria. Gu-Konu (1981):(a) the southern parts of the maize- The main cash crops in these forest zones are cocoa, sorghum-yam zone, (b) the bean-maize-yam zone, (c) coffee, oil palm, and rubber, often grown on extensive the cassava-maize-yam zone, and (d) the cassavaplantations owned by companies or wealthy farmers. maize zone. However, extensive peasant cultivation of these crops Similar subsistence cropping pattems exist in Benin also occurs at a much smaller scale. In Sierra Leone, and Nigeria, where the main cash crops are cocoa many riverain grasslands are important rice-growing (Liberia, Nigeria, Togo, and Sierra Leone); coffee (C6te areas, whereas in Cote divoire and Liberia, important d'ivoire, Liberia, Togo, and Sierra Leone); cotton (Cote agricultural areas are fewer but produce a broader d'ivoire, Liberia, Togo, and Sierra Leone); oil pahn range of crops. (Liberia, Nigeria, and Sierra Leone); rice (Cote -The main control on the distribution of forest re- d Ivoire, Liberia, and Sierra Leone); sugarcane (Liberia growth vegetation adjacent to and within the forest and Togo); and tobacco (Cote divoire) (Berron and zone is primarily the continual clearance and burning Vennetier 1983; Gu-Konu 1981; Nwafor 1982; von during the dry season, which is part of the local farm- Gnielinski 1972). ing systems. Clearance and burning are carried out for three reasons: to prepare fields, to promote grazing, Class 81-Mangrove and to flush game. In Sierra Leone, such areas are mapped as "Forest Mangrove swamps are common along the West Afri- Regrowth and Farmland" (Cole 1968; FAo 1981; Clarke can coast, but are particularly important in these 1966). This area was formerly closed forest but was areas: opened to timber exploitation in the nineteenth century. Since then it has been extensively cleared and * From the Casamance River in Senegal to the Rio burned to provide land for upland rice, coffee, cocoa, Nunez in Guinea, a coastal strip that includes the and oil palm cultivation (Millington 1987). extensive mangrove of the Bijag6s Islands and the The ecological communities witiin this class range Cacine, Cacheu, Corubal, and Geba rivers in from recently cleared farms to relatively mature sec- Guinea-Bissau and the Komponi river in Guinea ondary forest regrowth. Secondary forest and bush * Around Conakry in Guinea regrowth is cleared between January and April, and * In Sierra Leone, especially in the estuaries of the the wood is dried and burned just before the seasonal Scarcies, Rokel, and Jong rivers, and around rains start in May. The dominant crop, upland rice, is Sherbro Island grown in a mixed cropping system and harvested * As a senes of small isolated occurrences in Liberia from October to December. and Cote d'ivoire The fields are sometimes used for a second year, but * In the Volta Delta in Ghana often are allowed to revert to bush after a single year's * Along the Nigerian coast, especially in the Niger cultivation. After 2 to 3 years, a dense, low thicket and Cross deltas. forms, composed of herbs, razor grass (Scleria barteq,) climbers, and coppicing stumps left dunng clearance These areas account for 63,814 square kdlometers, (particularly Canthium glabrffolium, Craterispermnum about 3.1 percent of the region. The largest areas are launitun,wandausangaccropiodes).treessuchasalbizia in Nigeria (35,200 square -kilometers), Liberia (5,533

90 TIe West Afilcat Cmst 77 square kilometers), Guinea (5,480 square kilometers), Figure 9-9. NoVI Profile, Mangrove (Class 81) and Ghana (5,111 square kilometers). Mangrove swamps occur only in tidal and brackish 0.6 water along the coast in places where adequate shelter from storm waves usually exists. They comunonly 0.5 occur fringing wide estuaries, in shallow creeks, andbehind islands. Mangrove is swamp forest in which a DAdense network of tree trunks, stilt roots, and pneu- > 03 matophores usually form an impenetrable thicket. The z - J trees root directly in the mud of the tidal flats. The 0.2 canopy cover is high and mostly evergreen, generally with no ground cover. 0.1 The mangrove forests of West Africa are dominated 0-1 Niger Delta, Southrn Nigeria by three genera: Avicennia, Lagunazdaria, and Rhizophora Spatial relations among the genera, however, are not Jan Fcb Mar Apr May Jun Jul Aug Sep Oet Nov Dcc always clear. Mangrove trees of the genus Avicennkn generallyoccurinlandof Rhi20phora. However, along the mately 150 millimeters, mangrove again is quite Gambia River, for instance, no obvious zonation occurs stunted (Barbosa 1970). (Giglioli and Thornton 1965), and the Niger Delta is Many of the mangrove swamps in the region have dominated by a nixture of Rizq#opwmra harisonfi, R. man- beencleared forswamp rice culti vation. This clearance gle, R. racemosa, and Avicennia gerninans. has had an important impacton woody biomass stock, In Sierra Leone, mangrove swamp zonation is as is well illustrated by Sierra Leone. Although Temne thought to be controlled by soil conditions, topogra- farmers cultivated swamp rice in northern Sierra phy, and water salinity (Cole 1968). In Sierra Leone, Leone estuaries in the nineteenth century, the main Rhizophora racemosa is the pioneer species on silty, impetus to clear mangrove swamp vegetation for culfibrous soil, while on sandy soils the pioneer species tivationcame with twentieth centurycolonial penetraareavicennia africana,r mangle, and R. harrisonii Trees tion of the interior (Richards 1985; Millington 1987). at the mouths of rivers attain maximum heights of 7.5 Clearance of mangrove swamps leads to a buildup of to 10 meters, while upriver, at the limit of tidal incur- sulfuric acid to toxic levels in the thionic fluvisols of sion, they grow to 20 meters. the swamps. This acidity initially restricted rice culti- Inland from the area of pioneer trees, the open vation, but research by the West African Rice Develwoodland changes to a dense, thicketlilce swamp for- opment Association (warda) in Sierra Leone on soil est cften dominated by Conocarpus erectus and Lagun- management techniques and the introduction of specularta racemosa. In these areas, grasses and sedges cific rice varieties has led to the widespread use of characteristic of freshwater swamps invade to form an mangrove swamps for rice fanming. undergrowth, whereas other areas display a mixture Due to the concentration of population along the of mangrove and freshwater swamp forest tree and West African coast, pressure on mangrove swamps shrub species. can locally be very great. Pressure arises from land Mangrove productivity and biomass are very clearance for rice cultivation, from fuelwood exploitaclosely correlated with mean annual rainfall. The tion, from the preferential use of mangrove wood for greatest productivity exists in humid tropical for- smoking fish and baking bread, and from its use in est areas, although they show only moderate sea- building construction. Areas of mangrove adjacent to sonality (figure 9-9). This relation can be. towns suffer severe exploitation, and in these areas illustrated using tree height as an indicator of the fuelwood supply problems undoubtedly exist, accengrowing stock In the Niger Delta, where the mean tuated by the value placed on mangrove wood beannual rainfall is more than 4,000 millimeters, cause of its high calorific value. Rhizophora mangle stands attain 45 meters in height (Rosevear 1947). But at the northern and Land Cover Class Tablessouthern limits of mangrove on the west coast of Africa, tree heights are much lower. At the north- Tables 9-9 through 9-17, beginning on page 80, pres-, ern limit on Ile Tidra, Mauritania, 19 50' N, where ent summaries for each land cover class of the area, mean annual rainfall is 100 millimeters, Avicennia showing growing stock and sustainable yield for the spp. trees are only a few meters high (Chapman West African coast nations of Benin, Ghana, Guinea- 1977). At the southem limit in Benguela, Angola, Bissau, Guinea, Cote d'ivoire, Liberia, Nigeria, Si ' S, where mean annual rainfall is approxi- erra Leone, and Togo.

91 78 Estirnaling Woody Biornass in Sib-Sgaharan Africa References Giglioli, M. E. C., and I. Thornton "The Manrgrove Swamps of Keneba, Lower Gambia River Every effort has been made to facilitate access to the Basin. L Descriptive Notes on the Climate, the documents listed here. Some documents, however, Mangrove Swamps and the Physical Composition lack full bibliographic information because it was of Their Soils." Journal of Applied Ecology 2(1): 81- unavailable; also, some documents are of limited 103. circulation. Gu-Konu, Y. E. 1981a. "Population." In Y. E Gu-Konu and G. Laclavere. Togo. Paris: Les Editions Jeune Adjanohoun, E "Etude phytosociologique des Afrique. savanes de basses Cote divoire (savanes lagunaires)." Gu-Konu, Y. E. 1981b. "Agriculture." In Y. E. Gu-Konu Vegetatio 11:1-38. and G. Laclav&e. Togo. Paris: Les Editions Jeune Afolayan, A., and K M. Barbour 'Population Afrique. Distribution and Density." In K M. Barbour, J. S. Gu-Konu, Y. E., and C. Laclavere Togo. Paris: Les Oguntoyinbo, J. 0. C. Onyemelukwe, and J. C. Editions Jeune Afrique. Nwafor. Nigeria in Maps. London: Hodder and Hall, J. B., and M. D. Swaine "Classification and Stoughton. Ecology of Closed-Canopy Forestin Ghana."Journal Areola, a. "Vegetation." In K. M. Barbour, J. of Ecology 64: S. Oguntoyinbo, J. 0. C. Cnyemelukwe, and J. C. Rambler, D. J "The Vegetation of Granite Out- Nwafor. Nigeria in Maps. London: Hodder and crops in Westem Nigeria." Journal of Ecology 52(3): Stoughton t Areola, b. "Land Use." In. K. M. Barbour, J. S. Hopkins, B. J Forest and Savanna. 2d ed. London: Oguntoyinbo, J. 0. C. Onyemelukwe, and J. C. Heinemann. Nwafor. Nigeria in Maps. London: Hodder and Jansen, J. W. A "Vegetation." In S. von Stoughton. Gnielinski. Liberia in Maps. London: University of Arnaud,J.-C "Economie du bois." Atlas de la COte London Press. d'lvoire. Paris: Les Editions Jeune Afrique. Jenik, J., and J. B. Hall "Plant Communities of Barbosa, L. A Carta fitogeogrdfica die Angola. the Accra Plains, Ghana." Folia Geobotanica et Phiyto- Luanda: Inst. Invest Cient. Angola. taofnomica 11: Berron, H., and P. Vennetier "Agriculture." Jones, E W "The Forest Outliers in the Guinea Atlas de la COte d'ivoire. Paris: Les Editions Jeune Zone of Northern Nigeria." Journal of Ecology 51(2): Afrique Bianchi, H Assistance au developpementforestier, Keay, Rt W. J "Isoberlinia Woodlands in Nigeria Guibn-Bissaw Planificafion forestiere. Tcp/cs/4506(A). and Their Flora." Lejeunia 16: Rome: FAO. Keay, K W. J "Derived Savanna-Derived from Brookman-Amissah, MultipurposeMaLnagement What?"- Bulletin d'institut Frangais d'afrique Nord, of Woody Vegetations in the Northern Regions ofghana. Serie A 21: Kumasi. Lawson, G. W "Ghana." Acta Phzytogeographica Brunel, J. F "Vegtation." IT Y. E. Gu-Konu and Suecica 54: C. Laclavere. Togo. Pads Las Editions Jeune Afrique Lecomte, G., and N. Monnier 'Population." Atlas Chapman, V. J. (ed.) Wet Coastal Ecosystems. Am- tde la aote d'lwire. Paris: Les Editions Jeune Afrique. sterdamn Elsevier Scientific. Menaut, J. C "The Vegetation of African Savan- Clarke, J. I "Vegetation." Atlas' of Sierra Leone. nas." In F. Bouliere, ed., Tropical Savannas. Amster- London: Hodder & Stoughton. dam: Elsevier. Clayton, W. D "Secondary Vegetation and the Menaut, J. C., and J. Cesar "Structure and Pri- Transition to Savanna Near ibadan, Nigeria." Jour- mary Productivity of Lamto Savannas, Ivory Coast." -nal of Ecology 46(2): 'ZA Ecology60(6): Clayton,W.D "DerivedSavannainKabbaProv- Menaut, J. C., and J. C6sar "The Structure and ince, Nigeria" Journal of Ecology 49(3): Dynamics of awest African Savanna." In B.J. Hunt- Cole,N. H.A The VegetelionofSierra Leone. Njala, ley and B.'H. Walker, eds., Ecology of Tropical Savan- Sierra Leone: Njala University College Press. nas. Berlin: Springer-Verlag- FAo (Food and Agriculture Organization of the United Millington, Andrew C "Environmental Degra- Nations) Land Systems of Sierra Leone. FAO- dation, Soil Conservation and Agricultural Policies Government of Sierra Leone Land Resources Sur- in Sierra Leone, ' In David D. Anderson vey Project, Report i, Freetown. and Richard Grove, eds., Conseraion in Afria People,

92 The West Afiican Coast 79 Policies, and Practice. Cambridge: Cambridge Uni- Swaine, M. D., J. B. Hall, and J. M. Lock "The versity Press. Forest-Savanna Boundary in West-Central Ghana." Miilington, Andrew C., Felix Helmisch, and Gaest Ghana Journal ofscience.16: Rhebergen "Iand Valley Swamps and Bolis in Swaine, M. D., D. Lieberman, and J. B. Hall SierraLeone:HydrologicalandPedologicalConsider- "Structure and Dynamics of a Tropical Dry Forest ations for Agricultural Development.' Zeitsdchftffr in Ghana." Vegetatfo88:31-1. Geomorphologie, Supplemnent Band 52, von Gnielinski, S Liberia in Maps. London: Uni- Monriier, Y "Vdgdtation." In Atlas de la CMte versity of London i'ress. d'ivire. Paris: Les Editions Jeune Afrique. Voorhoeve, A. G "Liberia." Acta Phlytogeo- Morison C. G.T.,A. C. Hoyle, andj. F. Hope-Simpson. graphica Suecica 54: "Tropical Soil-Vegetation Catenas and Mosa- White, F "The Savanna Woodlands of the Zamics." Journal of Ecology 36(1): bezian and Sudanian Domains. An Ecological Nwafor, J. C. 1982a. "Agrcultural Zones." In K M. and Phytogeographical Comparison." Webbia. 19: Barbour, J. S. Oguntoyinbo, J. O. C. Onyemelukwe, and J. C. Nwafor. Nigeria in Maps. LondonE Hodder White, F "The Vegetation of Africa." Natural and Stoughton. Resources Research Series 20. Paris: LINESCO/AETFAT1 Nwafor, J. C. 1982b. "Major Cash Crops and Planta- unso (United Nations Educational, Scientific and tions." In K Iv Barbour, J. S. Oguntoyinbo, J. 0. C. Cultural Organization/Association pour ['Etude Taxo- Onyemelukwe, and J. C Nwafor. Nigeria in Maps. -nomique de-la Flore de ['Afrique Tropicale/United London: Hodder and Stoughton. Nations Sudano-Sahelian Office)..~ ~~~~~~~~~~~~ , ,.A:.

93 80 Estimatinig Woody Biomnss in SUb-Saharan AfricM Table 9-9. Land Cover Classes-Benin (West African Coast Region) Arma GrowiugRslock Sntslainableyvield -os.rliosad Th1oulsand o0n ies Laid cover class ait 2 Percel f-ojiiis Prcenl per year Perceni ,108 2, , ,531.AS , , , , , , , , , , , , , , , , , , Total 114, , , (Percentage of region) (5.60) (5.08) (2.53) Note: In the followirg table details may not add to totals beuse of roundi _g. Source: Authois calculations from data bases derived from land cover classification and table 4-1. Table Land Cover Classes-Ghana (West African Coast Region) Area G.irongstock Sustainable yield Thousand TIousand tonnes Land cover class knm Percent tonnes Percent Iper ear Percent ,054 1, ,001 1, , , , , , , , , , , ,;02.8D , , , , , , , , , , , , , , , , , , ,093.3B Lakes 13, Total 236, , , (Percentage of region) (11.59) (10.78) (8.66) Source Authors' calculations from data bases derived from land cover classification and table 4-1.

94 Tbe West Africas Coast 81 Table Land Cover Classes-Guinea-Bissau (West African Coast Region) - Arm Gro7uhigstock Susio;mablcyield -Thlousand- - Thomsand Jonnies Lannd couer class kl7:2 Percri -o ionnies _ Percenit peryear Pcrcent 22. 3, , , , , , , , , , , , , , , , , , , , , , , Total 30, , , (Percentage of region) (1.47) (2.18) (3.49) Source: Authors' calculations from data bases derived from land cover classification and table 4-1. Table Land Cover Classes-Guinea (West African Coast Region) Area Grouing stock Sustainable yield Thousand Thousand tonizes Land cover class k-n 2 Percent tonnes Percent per year Percent O.0) , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Total 239, , , (Percentage of region) (11.71) (12.15) (9A5) Sour=c Authors' calculaos frm data bases derived from land cver dassificaion and table 4-L

95 82 Esiimaling Woody Bioniass in Sib-Snlmnran Africa Table Land Cover Classes-COte d'ivoire (West African Coast Region) Area -Grwiugstock Sustainableyield TIhousand Tlhousand lop nes Land cover class kn9 Percemn lonies Percent per year Percen , , , OA , , , , ,474 13A1 116, , , , , , , , , ,131A , , , , , , , , , , , , Total 324, , , (Percentage of region) (15.87) (12.86) (7.34) Source Authors' calculations from data bases derived from land cover classification and table 4-1. Table Land Cover Classes-Libeda (West African Coast Region) Area Growing stock Sustainable yield Thousand Thousand tonies Land cover class km 2 Percent onnes Percent peryear Percent =1 2 15B , , , , , , , , , , , , ,R5? , , OA5 4, , , , Total 94, , , (Percentage of region) (4.61) (5.41) (7.10) SoDure: Authors' calculations fom databases derived fiom land cover clwfiqakiorad table4-1.

96 77wc West Africa: Coast 83 Table Land Cover Classes-Nigeria (West African Coast Region) - - ~~~Arcn Growin.g stack Sstasinanble yield - otosand TIIouisan-d lcanes Land cowvr clanss k Paen? fauna Percent per ear Percenl ,461 33, , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,153, , , , , , , , , , , , , , , , , , , , , Lakes 3, Total 879A Z411, , (Percentage of region) (43.06) (44.98). (54.17) Source: Authors' calculations from data bases derivp' from land cover dassirication and table 4-1. Table Land Cover Classes-Sierra Leone (West African Coast Region) Area Crowing stock Suisfainabl yield R! ll -- - Thotlsand TZl~~~~~~~~~~~~~~outsanid toiines Lrn!i0 kpercen Pcnt To gsa 65 ndt perayear Percent ,002 Z , , ' , , , , , , , , , , , , , , , , , , , , , Total 70, , , (Percentage of region) (3.44)-(423) (5.81) Source Authors' cakulaons from data bases derived from anld cover dasfiation and table 4-1.

97 84 Estimating Woody Biomnass in Sub-Salmtran Africa Table Land Cover Classes-Togo (West African Coast Region) Area -Grouingstock Sustairtable yield Tlousand Thjousand tonnes Land covr class km 2 Percent tonines Perent peryear Percent 12 1, :, , : = , , , ,96& , , , , , , , , , , , , ,71& , , , z, Total 54, , X00 3, (Percentage of region) (2.65) -C231) (1.46) Som0 Authors' calculations from data bases denved from land cover dasification and table 4-1.

98 The Horn of Africa, John Kirkby This chapter presents a detailed desaiption of the event Soils are at best skeletal and completely absent mostunportantlandcoverclassesinhisregion.help- from the two large areas of erg (mobile sand dunes) in ful figures in other chapters indude figure 3-1 (cloud the southwestern Libyan Desert and on part of the cover); figures 3-2, 3-3, and 3-4 (NMvI summary land Egyptian frontier in the Nubian Desert Extensive. cover profiles); figure 3-5 (regional sununary map of areas of bare rock and rock debris without any soil land cover classes); figures 7-1 and 7-2 (continental formation (reg) exist in the Libyan and Nubian demaps of growing stock and sustainableyield); and the serts, the Red Sea Hills of Sudan, inland areas of "RegionalLand Cover Class Map of EastAfrica" at the Eritrea, the Danaldl Alps, and on the northwestern end of this volume. coast of Somalia. Saline soils occur in the Danakil Helpful tables in other chapters include table 3-2 Depression of Ethiopia and on parts of the northeast- (land cover classes);. table 4-1 (data and sources for em coast of Somalia. growing stock and sustainable yield); and table 6-3 Phenological curves for the Danakil and Nubian (Horn of Africa estimated woody biomass by sum- deserts show expected low rates of activity, with figmaryclass). ures for the Danalal being less than those for the southern-nubian Desert Minima of -4102(Danakil) Class O-Desert and (Nubian) occurfromjuly to September, with maxima of +01)2 and in March and November Occupying one-quarter of this region, with more than (Danakl), and in April and December (Nubian). 1 million square kilometers, Desert is the largest of all Vegetationisvirtuallyabsent frommuch ofthe area, the land cover classes. Desertcovers a significant area particularly the Libyan Desert; nevertheless, plant of all four countries in the region-sudan, Ethiopia, cover does exist, if ordy temporarily, in some situa- Djibouti, and SomaLia. The northern third of Sudan is tions. After rare rain episodes, and even after long almost entirely desert, and this continues into north- droughts, grasses (zzu) may bnefly flourish in norerneritrea and the Danakil of Ethiopia. Virtually all of mally unvegetated areas. More substantial plants that Djlboutiisdesert,andthisareaextendsintothenorth- are adapted to desert conditions may survive in a em coastal strip of Somalia. Extensive areas of desert dormant state until rainfall occurs. occur on the eastern coast of northem Somalia, partic- White (1983) identifies semidesert conditions in the ularly in the Bad and Mudug regions. In Ethiopia, northeastern desert of Sudan. Hemming (1961) finds areas of desert occur within Ogaden and further more advanced vegetation than is generally implied patches exist on the borders with Kenya and Somalia. by the term "desert' in the coastal zone of northem Estimated growing stocrk and sustainable yield are Eritrea. Mangrove occurs in a discontinuous strip of both 0 percent of the regional resource (table 6-3). coastal swamp along parts of the Red Sea coast, for Most of the desert is below 300 meters altitude; example, a community of Avicennia and Rhizophora, indeed, extensive areas of the Danakil are below sea the latter particularly valued as fuelwood, exists for level Annualprecipitation is less than 150millimeters about 12 kilometers along the coast of Djibouti. and is extremelyvariable fromyear to year. Inparts of Farthernorth,intheRedSeacoastaldesertofSudan, the Libyan Desert of Sudan, rainfal is a very rare particularly to the north of Port Sudan, vegetation 85

99 86 Estimating Woody Biomass int Sub-Saharan Africa occurs on the coast. Salt marshes have an almost com- Although the area is described as grassland, patches plete vegetation cover. Saline areas of wadis support of trees, woodland, and thicket exist on drier soils, Juncus arabicus and Taniarix mannifera and nonsaline particularly toward the margin of the class, on the areas of wadis have Acaca tortilis and Zil1a spinosa infrequently flooded zone. Bari (1968) identifies three (White 1983). flooded zones, based on frequency and duration of Inland, small areas of trees exist along the larger flooding (seasonal, frequent, and permanent). The desert wadis along terraces of the River Nile in north- area of seasonal flooding is mainly grassland. Swamp em Sudan and on sites where surface water concen - species such as Phagmites communis and Ilyparrhenia trates. Acacia, usually of low stature, occurs on such rufi dominate the frequently or permanently flooded sites. In higher areas of the Red Sea Hills, such as the areas. Trees are mainly Acacia species; Acacia mellifera Karora Hills, small relict areas of Juniperus procera is common at up to 570 millimeters of precipitation forestoccur.inthellbyan Desert too, more substantial and Acada segal occurs where precipitation is greater vegetation is associated with moister areas of higher than 570 millimeters. Where flooding is deeper or land. Elsewhere, at lower elevation, a scrubby vegeta- more frequent, trees are absent Grasses occurring tion dominated by Acacia glaucophylla (circummar- with the Acacia trees include Cymbopogon nervatus, gtinata) and A. etbaica occurs (Ban 1968). Along the Hyparrhenia anthistirioides, Schoenefeldia gracilis, River Nile, and at a number of places near the Red Sea Schuma ischaemoides, and Sorghum purpureo-sericeum coast in Sudan, irrigation schemes are associated with (White 1983). Hypanhenia rufa and Setaria. incassata much greater NDvi values (see Class 22). are, however, the domiinant grasses (Badi 1968). Fuelwood is very scarce and threatened by stock Itisprobablethatgrasslandandthicket haveacyclic browsing, particularly during sustained drought relation through time. Cultivation and burning of the (Mogbhaby,Ali, and Seed 1987), when drought stress, grass have combined to reduce the tree cover of the overgrazing, and cutting may destroy the limited Sudanese Hydromorphic Grasslands, and because wood stockl Where sand dunes migrate toward oases grassland is by far the largest component of the ecoor toward the desert margin, as in parts of Kordofan, system, fuelwood is limited. The estimated growing woody species may be overwhelned by blowing stock and sustainable yield are both 0 percent of the sand. Because herdspeople must use oases for water- regional resource. ing animals, the vegetation in these areas is at particula risk from overgrazng and firewood cutting. In Class 13-Ethiopian Montane Steppe the Danakil Desert, an indication of the severe shortage of woodfueiis the widespread use of dung as fuel This class occupies 3 percent of Ethiopia, including all KCamweti 1984). land above 3,500 meters on both the easter and westem plateaus. This class is notable ir. the northern area Class 12-Hydromorphic Grassland of the western plateau in the SimienRegion and Choke Mountains, and in the mountains overlooking the This class occupies 2 percent of the region, and exists east-facing escarpments of the Welo, lgre, and Shewa mainly in Sudan, where it constitutes more than 3 regions. In the easter plateau, the main areas are in percent of the area. Principal areas of this class lie im the Bale Mountains and Arsi Mountains. The area is southeastern Sudan, particularly in Jonglei and coincident with Von Breitenbach's "Mountain Steppe" eastern Sobat, but with discontinuous areas in El art of his 'Mountin Savanna' classes (1963). Bukeyral and near the Blue Nile on the Ethiopian -L.Re terrain of these areas is high mountain and border, mainly below 500 meters. pla teau, formed on volcanic rock, with poorly devel- Almost all hydromorphic grassland in Sudan is oped skeletal soils and, in many cases, with steep, dearly within the flood region of the upper Nile, an -unstable slopes. Strong winds, cold temperatures, and area where the annual flood affects the flat terrain to frost are limiting factors in the development of vegedifferent extents, depending on small vanations of tation (White 1983). Environmental variables such as altitude. Terrain is everywhere extremely flat, with soil depth, protec-tion from wind, slope, and precipislopes of less than 1 on the infilled Quatemary lake tation are reflected in vegetation cover, but generally basin, which consists of dark cracking clays associated thebiomass and theheight of vegetation decrease with with hydromorphic soils (vertisols). The relative im- altitude. The lower limit of the Ethiopian Montane perviousness of the soils whenmoistcontributes to the Steppe hasremnants of aformerlymore extensive tree severity of flooding. Precipitaton increases from 400 cover, including Aada xiphocnpa (abyssmica), Apodytes millimeters a year in the north to 1,000 millimeters in spp., -Hagenia abyssinica, Junrpenus procera, and Olea the south, and at more than 700 millmeters the soil is afiicana. These may form elfin thicket, particularly unable to absorb all the rainfall in less-accessible places such as gorges. Most of the

100 The Horn of Afrca 8V landscape, however, is covered by Cyperaceae and Balanites rmcemosa. Near the lakes of western Eibouti Graminieae. a small area is domidnated by dom palm, Hyphaene It is not clear to what extent the grass is prilmary or the baica. the result of repeated burning. At the highest levels, in Ethiopia, Class 21 extends between the Danakil widely spaced tussocks of Agrostis scierophylla, Carex Desert and the lower plateau steppes through the monostachya, Deschampsiacaespitosa, and Limosell aftri.- areas of Keren,Asmera, Adigrat, Korem, Dese, Ankovam are able to withstand high winds and cold temper- ber, and northern Hdrergd. attires. In protected sites, scattered shrubs such as Transition from desert may be gradual or sharp, Erica arborea and Lobelia rhynchapetalum add diversity depending on edaphic conditions. Where water is to the vegetation, but an% susceptible to burni'ng, available near the surface, and in topographic situa- Browsing& burnin& grazing& and fuelwood collection tions. that enc-ourage water-gathering& such as 'wadis, have all contributed to loss of woody bioniass from well-de veloped Semidesert Wooded Grassland may this region, where the rate of tree g-rowth is very slow occur in nearly pure desert conditions. Similarly, the and virtually no fuelwood remains. Consequently, dry limit of this class may be as little as 80 millimeters estimated growing stock and sustainable yield are pfecdpitation (White 1983) on sands, although the desnegligible. ert may extend as far as the 150-millimeter isohyel Generally, Semidesert Wooded Grassland gives way Class 21-Semidesert Wooded Grassland to denser woodland at about 250 millimeters. The rainfall. regime varies between 1 and 3 months This class is the second most extensive in the region, with s-ignificant rainfall. The phenolog shows a limoccupying 15 percent of the area. Semidesert Wooded ited seasonal variationi, with two NOW maxima of ap- Grassland, which we will refer to generally as, semui- proximately 0.11 and 0.12 in November-December desert, is present in all four countries of this region, andi Mayt-june and N'io minidma, of 0.06 and 0.07 in and is particularly significant in Sudan and Somalia. April and September, reflecting the dual precipitation In Sudan, where this class makes up 14 percenlt of maxima. the' area, an irreguar wedge extends from Kassala Most of the class is on land below 600 meters, on on the Ethiopian frontier, broadening to the west in soils that are thin and poorly developed. On the Norther-n Darf ur. Ithas significant extensions north- edge of the Danakil Desert, they are saline, and in ward into the Libyan Desert, along larger~ wadis parts of northern Somialia, theyhave gypsumn crusts. such as El Milk and Hiwa, with outliers such as the ln western Sudan, extensive areas are developed on El'Atrun Oasis and ont higher ground that receives fixed sand dunes, although incursions of mobile more precipitation. Small, outlie rs also occuir in the dunes from the Libyan -Desert are widespread and.- winter rain areas of the Red Sea Hills, for example destroy vegetation. on the Erkowit Hils south of Port Sudlan and in the Much of the dass has longbeen useclfor grazingand Karora Hlills. White (1983) and Sudan Survey De- this has strongly influenced the grse,herbs, and partment (1954) indicate that all of the Red Sea Hills ItreeS. Because of browsing and grazing, manty of the and much of Kassala Province is "Acacia torliis- mare palatable Ancin species have been replaced by Macrun cra ssifolia Desert Scrub," which mustbe tram- unpalatable ones such as Acauda nubica, Cnssia acutili i, sitional to "Saheijan Savanna.". Wickens, in Ban' and Cal otropis procei he vegetation is subject to brief (1968), suggests that the Red Sea Hills were rapidly fluctuations of precipitation and occasionalprolonged deteriorating at that time (probably as a result of droughts. Although the plants are resilient, the corngrazing pressure), and that subsequent droughts bined effect of drought and overraig has reduced would further stress the vegetation, the tree cover, even of valued species such as Acacia Most nondesert areas of northern Somalia and ex- senesal that normally would be protected as a source tensive areas of the Wadi Shebelle Valley of central of gum arabic. Somalia are Sahelian Semidesert Class 21 covers one- Crown cover is usually less than 10 percent (White third of Somalia..1983) and continuous areas of taller trees are almost The nondesert area of Djibouti also is in this clasr. It entirely limited to rocky outcrops and water-receiving is particularly importantnorth of the Gulf of Tadjoura. site.s. Where rainfall is about 100 miillimeters, woodly Here, wilthin the savanna and associated with greater plant cover rarely exceeds 3 percent Annual grasses rainfall on Mount Coda between 1,300 and 1,80)0 me- dominate and trees tend to have thin stemis and to be. ters, is a small area.(870 hectares) of Juni penus procema of limited height. hin Somalia, for example, the drier. and Bunts forest in the For&i du Day (Haauounil1984). scrub is rarely.more thkan 3 meters high. Woody spe- Surrounding the forest are various forms. of woody cies such as Acacin ehienbergiana, A. laeta, A. ftoles, savanna, with Acacia arabica, A finva, A tortilis, and Balanites aegyptiaca, Boscia senegalensis, Commiphora

101 88 Estimating Woody Bionmss in Sub-Saharan Africa afrcana, and Leptadenia pyrotechnica (White 1983) are south of Port Sudan, including the Tokar Delta. The more common and better developed toward the wet- Gash Delta, Khashim El Girba, and Gezira-Managli ter margin of the class, but are usually less than 5 irrigated areas clearly are included in the class. A large meters in height Sudan Survey Department (1954) complex also isassociated with the edaphic grasslands and Bari (1968) found that Acacia tortilis and MAerua of north Sobatand anisolated occurrence on the slopes cassifolia dominate in the east of Sudan and Acacia of Jebel Gurgei in Darfur. mellifera dominates in the west. Grasses such as Ecologically, the class covers a wide range of situa- Panicum turgidum dominate the more mobile, drier tions, and in some areas it reflects strong human influareas of dune crests, whereas Centdzrus bizlou covers ence on the landscape. In fact, in the northern half of lower, more stable sands. Sudan, almost every occurrence of Class 22 is directly The estimated growing stock is 2 percent of the related to irrigation and crop production. Monoculregional total (139 million tonnes), and estimated sus- Wmre of cotton donimates much of this area, and trees tainable yield is 5 percent of the regional. total (5.3 may be virtualy absent In contrast, trees form a conmillion tonnes). Woody biomass is already grossly spicuous element of the Class 221andscape occurring overused in this class. Growth rates are low, and the in Sobat and western Ethiopia. social consequences of overuse are severe for the pas- White (1983) associates "Acacia Wooded Grasstoralists of this region. Overgrazing may lead to dune land' with precipitation between 250 and 500 millidestabilization and to desertification. The class also is meters, but in Sudan, Class 22 also occurs in drier susceptible to deterioration during drought years, for and wetter areas. In the Nubian Desert, for example, example and pumped-water irrigation allows the development Desertification in this area has been examined in of this class in extremely dry areas, whereas in the Dibouti by Hamnrouni (1984); in Somalia by Okafo humid area of Sobat, the greater rainfall is effec- *(1987); in Sudan by Lewis and Berry (1988), by tively increased by Nile floodwater. Much of the Moghraby, Ali, and Seed (1987), and by Booth (1984); western plateau of Ethiopia receives up to 1,000 and in Ethiopia bykebbede and Jacob (1988). Human millimeters of precipitation. population may be very dense in this class, consider- A variety ofspeciesofacaciaisassociated withclass ing the limited resource base. hn Somalia, for example, 22. In Ethiopia, Von Breitenbach (1963) identifies this t-his class supports 40 percent of the population in an area as "Lowland Woodland" with an upper story of environment in which cultivation is not possible, so trees of 5 to 12 meters, forming a more-or-less closed that grazing puts great stress on the trees (Kamweti canopy that is sparse enough to admit light to a thick- 1984). Hamilton and Maizels (1989) predict that, etlike lower story of shrubs of 1 to 3 meters. Characwith increasing population, the dearance of wood- terstic trees areacuciaamythetophylla,a ask A. rnel0frn, land in northern Ethiopia in this class will continue A senegal, A. sieberna, A tortilis, A. rnosa, Ficus aco- Moghraby, Ali, and Seed (1987) attribute severe dete- carpa, Cassia singuaa, Combretum moll, and Ziziphzs rioration to the northward spread of cultivation dur- abyssinica. Toward 2,000 meters and in drier condiing the late 1970s. tions, Euphorbia thicket8 to 12 meters high, dominated by Euphorbi abyssinica, replaces Acacia trees. Class 22-Acacia Wooded Grassland A range of Acacia spp. also is characteristic of Class 22 along the Nile and Atbara; these include Acacia This class occupies approximately 2 percent of the nilotica, A. senegal, A. seyal, and A. tortilis. But in those region. It is present in smal patches on the lower areas where gravity irrigation, pump irrigation, or slopes to the north of the limestone escarpment in flush irrigation is possible (Grove 1978), the dominant northern Somalia. In Ethiopia, it occurs as numerous vegetation is cropland, particularly cotton. In such disconfinuous areas north of the western plateau in areas, woodland may be completely absent. Welega, Gojam, and Begemder, with further small Acacia seya is associated with a mosaic of grassland patchesonthelowereasternslopes of the Eritreanhills and trees in the flood region of Sobat Locally, Acacia near Asmera. A string of small patches exists along spp. form almost impenetrable thicket, separated by the lower Awash River, almost to the border with areasofgrasslandontheannuallyfloodeddarkcrack- Djibouti. ing clays. Acyclicrelationappears to existbetween the The main area of 68,000 square kilometers is in thiicket and grassland of Cymbopogo nervatus, Hypa- Sudan, where it occurs in many environmentally con- haia anthlrkles, and Sorghum pwpui-se= (White trasting areas of the country. Significant occurrences 1983). Areas of shallow flooding are associated with are along the Nile, conspicuously in the Libyan Desert A. seyal communities; areas of deeper flooding are and in the flood area of the south. Other stretches are associated with open grassland, particularly Setria along the White Nile, Atbara, and Khor Abu Habi incrassatal Extensive areas of swamp vegetation such tibutaries of the Nile. Two areas occur on the coast, as Cypenrs papyrus cover severely flooded areas.

102 *Ihe Horn of Africa 89 Bushlandandthicket2to3metershighcoversrocky 0.10, but according to the length and precipitation outcrops and water-receivingsiteson the lowerslopes during the wet season, marked increases can occur. of Jebel Gurgei. Elsewhere on the pediment of Jebel Figure 10-1 shows an example from the Ogaden in Gurgei, Acacia melliferm, Boscia senegalensis, Combretum Ethiopia that demonstrates consistently small values. africanum, and Euphorbia candelabrum form the princi- Deciduous bushland and thicket covers much of pal elements of the plant cover.. this class, normally in the form of dense bushland 3 to The estimated growing stock is 0.5 percent of the 5 meters high, but with scattered emergent trees to 9 regional total (20 million tonnes), and the estimated meters. Across extensive areas it is virtually impenesustainable yield is 0.5 percent of the regional total trable because of the density of trunks and branches (770,000 tonnes). Burnin& clearance for cultivation, and thorniness of many plants. White (1983) believes and grazing pressure have severely reduced woody that, it is more appropriate to describe the' class as biomass in much of the class in Ethiopia, Sudan, and thicket than as savanna.-in rocky areas, or where more Somalia. In central areas of eastern Sudan, agriculture water is available, emergent trees may form a moredominates the class, particularly on the cracldng clays or-less full cover with touching crowns. Toward the where Graham (1969) notes that removal of trees is west of Ogaden, the class is less thicketlike and Acacia associated with clearance for commercial agriculture. seyal is more dominant Along wadis and rivers, river- A further pressure on trees in this zone is the demand ine forestmay occur, and along the course of thejuba, for fuelwood for the three towns of Khartoum, Khar- a gallery forest about 200 meters wide once was prestoum North, and Omdurman (LewiL and Berry 1988). ent over considerable stretches (Mooney 1959). Bally Kassas (1970) notes the progressive removal of tree (1968) notes that Hyphaene is present along both the cover southward from Khartoum (see Class 44). The Juba River and Wadi Shebelle. extension of agriculture and need for fuelwood. Again, awide range ofaacia spp. is present; White aroundrefugeesettlementshassignificantlyincreased (1983) notes thirty Acacia spp. and sixty Commiplwra the rate of tree remova in Sudan since spp. as endemic. Shrubs are short-stemmed, multibranched, and underlie scattered umbrella-like trees Class 41-DryAcacia-Commiphora Bushlad such as Acacis nubica, A melifera, A. se negal, A. seyal, and Thicket and Commiphora boiviniana. The thicket-forming scrub is short-stemmed and multibranched (Von TIhough isolated areas of this class occur within the Brietenbach 1963). Perennial grasses occurintussodcs Semidesert Wooded Grassland (Class 21) of central between shrubs and trees, and after rainfall, bare Sudan, the main occurrence is in the souther and groundiscoveredby annualgrasses.vonbreitenbach southeastern parts of Harerge Province, southern (1963) describes the class as "Shrub Steppe." Bali and Gamo-Gofa in Ethiopia, extending into the Most of thiis class is used by migrant pastoralists, larger portion of southern Somalia. Several patches many of whom travel in an annual cycle between occuronthecoastandslopesof thenorthenesomalian Somalia and the Ogaden. Bosmellia spp. is used for hifls, extending as a discontinuous line into the low- frankincense, and Commipho spp.- for myrrh, and lands of north Hareri and the Rift Valley of Ethiopia. because individual trees are owned by families, they Small areas follow the lower escarpmentin Eritrea and tend to be carefully manage i and not overused. Plan- Welo, and a large section occurs in the drer area of tations of Conocarpus spp. near Berbera, Somalia are Eastern Equatoria Province of Sudan. The class ooc- used for urban firewood (Finlayson, Child, and Van pies one-qarter of the area of Ethiopia and one-third of the area of Somalia, or 12 percent of the entire. Figure 10-L.NDv Profiles, Bushland and Thiclcet region. (Classes 41,44, and 45) Most of the class is below 1,000 meters and grades into semidesert at its drier limits. Soils are skeletal or 0.6- weakly developed. Although precipitation may be as -41 Edilopia great as 500 millimeters in southeastern Sudan and Ethiopia southern Somalia, it is as slight as 200 millimeters in ma- 4Eip the Ogaden of Ethiopia and below 100 millimeters in ' northern Somalia. Evaporation is rapid and patches of 03-. true desert occur witiin this class in parts of the Z '/ -, * Ogaden. Precipitation varies greatly from year to year, 0.2- and prolonged drought lasting several years is dcaracteristic of this zone. Phenological curves show some variation in this. a,,, *,, *, class, with generally small values of approximately Jan Feb Mar Apr May Jun Jul Aug Scp Oct Nov Dcc

103 90 Estimaing Woody Biomnss in Sub-Saharan Afticr Rensburg 1972) and charcoal from this region is ex- to 5 meters tall with scattered emergent trees up to 9 ported to other parts of Ethiopia and exported from meters. (White 1983), although Mooney (1959) denorthern Somalia to Aden. These researchers found scribes the vegetation as scrub less than 3 meters high. that trees had been planted near vilages and were Acacia and Commiphora are spiny,. so impenetrable being carefully tended, and Mooney (1959) reported a thickets may forn, particularly because many species range of planted trees near Margherita, Somalia. are multistemmed and some species of Commiphlora As nlight be expected, the pressure of grazing has havearadialprostrateform.inthesecasesasingletree led to some deterioration of trees. Finlayson, Child,. may be 12 meters across, although less than 5 meters and Van Rensburg (1972) found that Acacia spp. and high. Commiphora spp. were degraded near Kisimaya, al- A few species have well-defined trunks and form though in the same area dunes were being fixed by emergents, occasionally reaching 10 meters high. Aaplanting of Commiphora spp. This class has suffered cia tortilis, Adansonia digitata, and Euphoia robecchii extreme degradation through overpopulation, partly are in this category, although the baobab here rarely because of the movement of refugees, and partly due exceeds 8 meters in height. -Emergent trees are more to overgrazing. Hemming (1966) states that overgraz- common in thewetter areas, andmayevenformatype ing has occurred since the nineteenth century. The of woodland. In the Holawaiir Forest near the Kenyan most severe overgrazing is near deep boreholes, border, emergent trees reach 30 meters, with a middle around which desertification occurs. Demandcfor fuel- story higher thn 10 to 15 meters. Acia bussei domiwood increases the rate of depletionl The estimated nates large areas, with few other spemes under its growing stock is 15 percent of the regional total (729 spreading crown (Bird and Shepherd 1988). nillion tonnes). The estimated sustainable yield is 8 Most species are deciduous, although a small perpercent of the regional total (11 million tonnes). centage are evergreen, and succulents are widespread. Kamweti (1984) condcuded that fuelwood shortages in Grasses form an insignificant part of the above- Somalia are extreme. ground- biomass, although Bally (1968) notes that where stock are excluded, a much greater profusion of Class 43-MoistAcacia-Commiphora vegetation quickly develops. Acacia spp., Cappari- Bushland and Thicket -dwceaspp., Commiphom spp., and Grewiiaspp. are very -common. Toward the south of Somalia, Euphorbia This class covers 2 percent of the region, mainly to- robeeccii, which is unusual in having a straight bole, ward the coast in south Somalia (Shabeeflala Dhexe, may also form a conspicuous and taller element, al- Shabeellaha Hoose, Jubbada Dhexe, and Jubbada though it has been used extensively in box-making Hoose) atelevations of up to 200 meters, and includes because it is one of the better timber tres (Mooney thevalleys of thelowerjubaand WadiShebele.About 1959), and numbers at that tim were much reduced.. percent of Somalia is in this class. Precipitation In a narrow zone near the Juba River, a number of varies between 300 and 600 millimeters a year, with taller trees that form a type of gallery forest are retwo wet seasons. Considerable fluctuation in precipi- ported by Firlayson, Child, and Van Rensburg (1972). tation occurs from year to year, however. This forest, however, is being encroached on for both Class 43 in south Somalia is almost completely sur- permanent and shifting cultivation and has been cut rounded by Dry Acaca- C.- nmmiphora Bushland and -for commercial timber (Luhini 1986). The largest of Thicket (Class 41), with a gradual transition between LhelowerJuba foests was about l50 square kdlometers the two. A number of patches also exist in southem in extent in 1948, but had been reduced to less tan Ethiopia at altitudes up to 1,500 meters. The largest 2,000 hectares in 1987, and less than 700 hectares rearea is near Ginir in the Shebelle Valley and in the main in the lower Shebelle (Douthwaite 1987). Domiupper Geneale Valley. Other small areas are on the nant tres in the riparian Juba. forests indude Aca slopes of the southern Rift Valley and in the lower seyal, Afea quanzensis, Anidesm?m7enosm, Mifmusops OmoValley. PreCipitationmEtiopia mayattain1,000 fru wiosa, Sorina madagascariensis, Spfrostadcys renimiimeters, with two rainfa maxima. ll ferz, Thesia danis, and TrichrTcR emetca (Luchini Compared with Class 41, wivt -values are generaly 1986). Douthwaite (1987) identifies about 50,000 hectgreater, although June, thfe month with the greatest ares of forest in the Ltolawajir Depression bordering value of 0.45, is almost tre same as values frou Class Kenya,butpredicts thatatthepresentrateof clearance 41 in inland southern Somalia. Minimum mmvi values most will have been removed by the year of 0.18 and 0.19 for March and April are, however, Speciescharacteristicof themain canopyofclass 43 considerably greater than those for Class 41. in Somalia are Aacia bussei, A horrida, A. mdlifera, A. InSomaliathelassissilartobutdenser tanthe nilotica, A. refias, A thomasi, Baamnites orticularis, Dry Acacia-Commiphora Bushland and Thicket (Class Boscia coriaca, Boswellia neglecta, Commiphora africana, 41). Generally, it forms a two-story dense bushand 3 C. boivinima, C campestris, Cadaba spp., Dobera spp.,

104 Tite Horn of Africa 91 Euphlrbia spp., Salvadora spp., Sterculia spp., and Ter- that even poorsources such as Commiphora were being minalia spp. (White 1983; Mooney 1959). Smaller bushes used for charcoal Much of the tree canopy in the Bay include Bauhinia titensis, Bridelia taitensis, Cae.salpinia area east of the Juba River has been lost, causing soil trotwae,combretumaculeaturn, Ecboliumnspp., Grewia spp., erosion and land degradation and greatly reducing Maerua spp., and Sericocomopsis spp. (White 1983; the grazing potential. Mooney 1959; Bird and Shepherd 1988). Cultivation also is increasing in the Juba Valley. The small areas of this class in the lower Omo Valley Douthwaite (1987) stresses the significance of farrning, and the Rift Valley of Ethiopia are similar to the Soma- logging, mining, and water resource development in lian area described previously, but much of the reduction of the lowland forest resource of southern Ethiopianoccurrenceof thisclassisathigher altitudes, Somalia. Uses of timber in the remaining Shebelle above 1,500 meters. Dry Acacia-Commipphora Bushland gallery forest (approximately -700 hectares) include and Thicket (Class 41) adjoins this class at lower alti- timber for beehives, charcoal, fuelwood, building timtudes, and.cultivation and Forest Regrowth Mosaic ber, agricultunal implements, and goat browse. The (Class 73) lies at the same altitude to the south. Beals estimated growing stock and sustainable yield are (1968) describes Class 43 as "Deciduous Woodland about 2 percent of the regional total (118 million and Savanna," dorninatedbyacaciawoodlanidwitha. tonnes and 1.5 nuillion.tonnes, respectively). senegal, A. seyal, and A. tortilis, locally domninated bv other Acacia spp. At higher elevations,acraciazetbaica is Class 44 Sahel-Sudanian Acacia Wooded more common. Bushland Von Breitenbach (1963) considers the area to be part of the lowland savanna woodland with a closed can- This class, which occupies 10 percent of the region, opy at 5 to 12 meters, but allowing sufficient light to occurs in Sudan and Ethiopia. It is particularly extenpenetrate and allow development of shrubs 1 to 3 sive in Sudan, where it covers 16 percent of the counmeters high. This forms Acaia thicket with Euphorbia try. It lies between the Semidesert Wooded Grassland thicket at higher elevations. (Class 21) to the north and Dry Sudanian Woodland Species listed by Von Breitenbach include Acacuia (Class 62) to the south. It forms a band about 500 amythetophya,a.asak,a mellitera,aseyal,a sieberana, kilometers wide, stretching from southem Darfur A. torhlis, A. venosa, Albizia amra, Cassia singueana, through central Kordofan, White Nile, and. Blue Nile Combreturn inlle, Entadopsis abyjsinica, Eupharbia spp., provinces, and continuing into the northem part of Ficus spp., Greania spp., Maera angolensis, Pterolobium Eritrea. Small areas of Sahel-SudanianAcacia Wooded stelatum, and Ziziphus abyssinica. Parasites, including Bushland occur on the Red Sea coast of Sudan, and Loranthus and Viscum, are comnmon in tree crowns. isolated areas exist along the Mile and in small areas Kamweti (1984) identifies the coastal region of So- of the Libyan Desert More substantial outliers are malia as a region. of fuelwood deficit. White (1983) scattred to the south of the main belt in both Sudan describes the dass as having suffered extreme degra- and Ethiopia. dation through overgrazing, particularly intensified Values of- N.aforasiteinnorthermEthiopiaconfirm by the provision of boreholes used as watering sites, that growth activity is greater than in the Dry Acadaand by the increase in cattle allowed by the improve- Commiphora Bushland and Thicket (CLass 41). These mentof veterinaryservices. Fmlayson,Child,andVan values also reflect the double wet season with peaks Rensburg (1972) found that shifting cultivation and in May and October (figure 10-1). commercial woodcutting were reducing the remain- Most of the class lies on plains below 500 meters, ing areas of good-quality nparian woodland and that although in Ethiopia it extends to more than 1,500 sedentary agriculture in the Juba and Wadi ShebeUe meters. Precipitationrangesbetween 250 and500 milvalleys and near Mogadishu was causing further re- limeters, and most of the area covered by this class ductioninvegetationcover. It is almostcertain that the experiences 2 months with at least 100 miimeters. effects of grazing and agriculture are significantly Soils vary significantly to the west of the Nile, where reducing the quality and quantity of tree cover in theclassisbestdeveloped. Quatemrarysanddunesup Class 43. to 40 meters high and up to 100 kilometers long, now Kamweti (1984) identifies fuelwood demand from mainly stabilized by vegetation, overlie kaolinitic Mogadishu as a cause of tree loss, extending for a days. These form the qoz landscape, which is well. considerable distance (some 20D kilometers along nv- developed near El Obeid (Booth 1984). These sandy ers) to supply the 250,000 tonnes needed each year. soils cover about 65,000 square kilometers of Sudan Bird and Shepherd (1988) found thatacacia busseiwas (White 1983). Acacia senegal is characteristic of this extensively exploited for charcoal used in towns, par- substrate, and may form almost pure stands. Because ticularlymogadishu,butthata senegal wasbecoming it is used for gum arabic production, it has been proimportant as A bussef was being overexploited, and tectedandincorporatedintheagriculturalsystem,but - --, X

105 92 Estimating Woody Biornss in Sub-Salsaran Africa in the past two decades it h-s been extensively cut for. severely depleted within 5 kilometers of wells (Lewis fuelwood, and pressure on the land has led to its and Berry 1988). A similar effect has been observed in omission from the cropping cycle. Somalia in Dry Acacia-Commiphora Bushland and To the east of the Nile, the central clay plains are Thicket (Class 41). doninated by soils with significantlevels of montmo- I Desertification, reflecting both climatic fluctuations rillonite and illite (Graham 1Q69). These clays are im- and anthropogenic factors, has led to the deterioration permeable, becoming dry, dusty, and deeply cracked of extensive areas of Class 44. Grass burning and in the dry season, but muddy when wel Because of selective browsing have very considerably.altered the theimpermeabilityof thecrackingclays, watersupply tree species present. (Bari 1968). Moghraby, Ali, and forvegetationisless thaninthe qoz areas. For example, Seed (1987) attribute deterioration of biomass in this Acacia senegal needs 280 to 400 millimeters of precipi- zone to the southward migration of livestock-based tation on sandy soils, but 500 millimeters on the crack- communities. ing clays. Acacia seyal forms almost pure stands in The fuelwood potential of this class has been seareas of 50D millimeters precipitation.on flat areas of verely depleted by the processes described, anrd if clay, but in depressions and along streams it occurs deterioration is to be halted, remedial action is essenwith much less precipitation (Smith 1953). Acaci tor- tial In the Ethiopian areas of Class 44, few trees now tilis ss. raddiana extends along streams into areas -with remain. Booth (1984) investigated restocldng with as little as 150 millimeters precipitation, but on flat Acacia senegal in the area of El Obeid to prev'ent clays needs about 450 millimeters. desertification. Throughout the clay plains, trees and shrubs are - The estimated growing stock is just 1.2 percent of more common along drainage channels. In this class, the regional total (59milliontonnes), butthe estimated on favorable sites, tree crowns may be almost touch- sustainable yield is 11 percent of the regional total (15 ing but usually are several crown diameters apart million tonnes).- The main tree species include Aacia astda, A. astrin- - gens,a aeta, A nilotica,a. sengal,a. sejal, A tomento, Class 45-Escarpment Wooded Thicket A. tortilis, Balanites aegyptiaca, Boscia senegalnsis, Cornbretum cordof anmu. Commiphor africana, DalTrgia spp., Class 45 is most extensive in Ethiopia, particularly in Maenia crassifolia, Leptadenia pyrotechnica, Terminalia a long arc from Eritrea, following the eastern edge of spp., and Aoogeissus schimperi (White 1983; Bari 1968). the western plateau through TiLgre, Welo, and Shew-a, In general, trees are of small stature, on the order of 7 along part of the Rift Valley floor, along the upper to 10 meters. Grass species are mainly annuals such as escarpment of Ch'erch'er and the Harer Hills. Small Cenchrus bij7orus, but perennials become more com- relict patches exist in Shewa, Begemder, and Gojam mon in wetter areas. and in small areas in Bora. The Hirer escarpment The zone is identified by Berry (1983) as suffering section continues into northern Somalia on the high severe deforestation because of both direct and indi- north-facing escarpment, inmduding Daloh Forest In redt physical effects of severe drought, such as that.djibouti, a small patch of 870 hectres persists in the between 1968 and Drought in itself stresses the Foret du Day. Sudan has areas in the Boma, Didinga, trees, and, when other vegetation is lacking, herders and Dongotona hills of Equatoria, and a small relict cut much more fodder from the trees than normal area in the Karora hills on the Red Sea coast. Fuelwood collecton for urban areas, particularly for Almest all of the areas covered are relatively high; the three towns of Khartoum, Omdurman, and Khar- for example, the Daloh Forest at 2,000 meters is atop a toumnorth, had extended to theethiopian frontierby 500-meter vertical limestone scarp. In Ethiopia on the By 1980, fuelwood was being transported from east-facing scarp, Junipen's woodland occurs to 3,200 more than 600 kldometers south. Much of this fuel- meters, although in DThbouti the Fort du Day on wood is from the Sahelian-Sudanian zone. Mount Goda is only between 1,300 and 1,800 meters Kassas (1970) records that Acacia spp trees were (Hamrouni1984). Precipitationis approximately 1,000 common near Khartoum in 1955, but that by 1972 the to 1,500 millimeters in Ethiopia, although exampjes of nearest trees were 90 kilometers south of the city. Juniperus exist as emergents in scrub forest with only Graham (1969) identifies agricultural andpastoral in- 650 millimeters of rainfall (White 1983). In the Daloh tensificationas an importantcause ofwood lossin this and Day forests, precipitation is much less, about 700 zone. On the clay areas, mechanized farming has en- mnillimeters in the former and 350 millimeters in the tailed complete removal of trees. In the qoz sands, latter. Both examples owe their existence to occult digging of high-yield tube wells since the early 1960s precipitationthatcreateshighhumidities fora considhas concentrated grazing pressure around them. By erable part of the year (Hamrouni 1984; Hemming 1962, little vegetation remained within 2 kilometers of 1966). Values of NDvi, however, generally reflect the these wells-a piosphere-and vegetation had been double rainy season and, in the case of an Ethiopian

106 T-he Horn of Africa 93 example from the rift escarpment, exhibit apeak value million tonnes). It seems unlikely that accessible greater than 0.4 in May (figure 10-1). lunipenes spp. and Podocarpus spp. will survive degra- Though the class is identified as Woodland Thicket dation by grazing, agriculture, lumbering, and cutting in its climax state, above 2,000 meters much of it for fuelwood, but Acacia spp. too are suffering from originally was forest. IJnipents procera formed im- the same pressures in this class. Our impression is that pressive, almost pure stands up to 3,200 meters the rate of loss significantly exceeds sustainable yield. (Rochetti 1961), whereas Podocarpusgracilior occurred The natural regeneration of Juniperus procera appears in damper areas at about 2,000 meters. In such forests, to need particular conditions that are unlikely to be bothjuniperus spp. and Podocatpusspp. form an upper met in extensively used forest, so the present generaevergreen story 40 to 50 meters high, with an under- tion of trees may-be the last to survive.. story of Apodytes dimidiata var. acutifolia, Cuissonia spp., and Ekebergia spp. at 10 to 20 meters, plus a lower Ciasl 52-East African Low Woody Biomass shrub layer. Epiphytes and lianes may be present. Mosdic Junipenesprocera is light-demanding, and White (1983) suggests that, because it does not regenerate in its own Covering 3 percent of the region is this class. It is shade, it probablv depends on fire (either natural or particularly extensive in Ethiopia, where it covers anthropogenic) for its survival. about 5 percent-of the country. Scattered patches occur In the Forat du Day, Torcdonanthus camphoratms in in southern Somalia, but the main areas of this class more open forest follows the degradation of Junperus are in the hills of northem Harerge and on hilly areas spp., and Acacia etbaica, A. seyal, and A. tortilis succeed of the Gamt-Gofa and Sidamo provinces in southas more open woodland. In Daloh Forest, Junipenms western Ethiopia. Other areas exist on the edge of the procera forms a relatively open forest with widely lava plateau of Welo Province. A discontinuous zone branch. -4 crowns. Associated with Jiniperus is a bush occurs on the lower slope of the mountains of Eastern. vegetation of Buxus lildebrandftii Cadia purpurea, and Equatoria and in the northernarea of Western Equato- Dodonaea viscosa. Iniandjutnipenis forest is replaced by dia between Dry c ladanian Woodland (Class 62) and Acaciaascrub (Hemming 1968). On the lower partof the Moist Sudanian Woodland (Class 65). escarpment, Buxus spp., Boswilia cariert Dracena schi- In Sudan, the dass occurs at about 600 meters altizantha, and Superbia grndisform an evergreenscrub, and tude,butin Ethiopiaitoccurs above 2,000 meters. Soils lower still is a zone of Acacid-Corimiphora scrub. In Soma- associated with this class are extremely varied, indudlia, Hemming (1966) identifies tree types of juniper ing those of the lava plateaus and the lateritic soils of forest-closed, open degraded, and climatic relicts.. southern Sudan. Rainfall exceeds 500 millimeters, apm Throughout the class, the Juniperus and Podocarpus proaching 1,000 nillimeters in western Harerg2, and woodland is rapidly declining, partly through com- the dry season lasts between 5 and 7 months. The mercial exploitation of timber, partly through over- pattem of NDv increases to 0.27 in May and June, withl grazing, and partly through a failure to regenerate. On a second slight increase to 0.15 in October, but for the the scarp between DireDawa and Hirer, considerable remaining 9 months of the year, values are as small as areahad been deared for agriculture by Much of 0.11 and the central Ethiopian plateau may have had this type The EastAfrican Low Woody Biomass Mosaic is an of woodland; for example, it was common near Addis intermediate zone between Acacia-Commiptbora Bush- Ababa at the end of the nineteenth century, but very land and Thicket (Classes 41 and 43) and Escarpment little remains, apart from that which is protected. Wooded Thicket (Class 45). It is far from uniform in Much of the deforestation occurred in the nineteenth structure or constituents, and may be either floristicentury, when timber was used for buding and as fuel cally rich or relatively poor. Certain species are almost for towns on the plateaul The Emperor Menhlik encour- always present (White 1983): Acokant hera spp., Car;ssa aged the planting of Eualptus spp., particularly Euca- edulis, Dodonaea viscosa, Eudtea spp., Olea africana, Sanhyptusgiobus as a replacement, so that Eualdyptus spp. are semierna spp., Tarchonanthits canmphoratus, and Teclea probably the most important trees in Ethiopia today. spp. Succulent species of Aloe and Euphorbia are nor- Approximately 15,000 hectares of Eucalyptus spp. plan- mallypresent The canopy is fairlylow, between3 and tation exist around Addis Ababa, but many small 7 meters being common, but emergents such as EudlumpsalsoexistinShewa, andthetreehasbeenplanted phorbia candelabrum may rise to 10 meters, and toward widely as a road margin plant for use as fuel and poles. the upper limit tall trees such as Juniperus procem start Scattered blocks of Eucalyptus spp. elsewhere cover to appear. 20,000 to 25,000 hectares (FAO 1963). In Somalia, dominant species include Acokanthzera The estimated growing stock is IA percent of the schimperi, Buxus hildebrandtii (which may form scrub regional total (69 million tonnes) and the estimated forest), Cadia purpurea, and Dodonaea viscosa. Other sustainableyield is 0.6 percent of the regional total (0-7 important species indude Aloe eminens, Barbeya oloides,

107 94 Estimating Woody Biomanss in Siub-Salnara;, Africa GCssonia holstii, Dracae,aa sciizant ha, Euphorbia grandis, properties and have a profound influence on drainage and Sideroxiylon spp. (White 1983). A narrow zone of conditions and soils, and thus on vegetation. this class lies. below the Junipenrs forest of the escarp- Toward the west, the dominant cover is "Combrehinm ment in northern Somalia (Hemming 1966). cordofanutn-dazbergia spp., Albizia seicoceplmla Wood- Grazing, fuelwood collection, charcoal produc- land" and "Anogeissts-Combretutn hartmannian utr Woodtion, and clearance for agriculture have reduced the land.' In the east, "Acacia melliferm Thomland' and fuelwood poterntial of this class. In the Harer region, "Acaciaseyal-Balanites Savanna" are the most importclearance for agriculture. has been the main influ- ant formations, the former on clays, the latter on sands. ence on woody biomass, and extensive areas now In both formations, woodland alternates with patches have a low tree density and are almost entirely of grassland (Sudan Survey Department 1954). Ancropped. Roadside trees are common; these are ogeissus leiocarpus may occur in pure stands on thicker often Eucalyptus spp., which also occur in small soils in Darfur, whereas stony ground is dominated by plantations and farms. In Ethiopia, the class covers Boswelliapapynfera (White 1983). On flooded soils near a complex mosaic of land uses, with, for example, the White and Blue Niles, Acacia astringens, A. nilotica, extensive coffee production and remnants of A. subalat, A. seyal, and A. tomentosa are comlmon. Podocarpus forest The estimated growing stock is 4 Much of the woodland that had continuously covpercent of the regional total (230 million tonnes), ered extensive areas has been cleared for agriculture, and the estimated sustainable yield is 5 percent of particularly in central, eastem, and western Sudan. the regional total (6 million tonnes). and inethiopia, because the treeswere easily removed with simple equipment. Pastoralism is widespread in Class.62-Dry Sudanian Woodland. the south and west, which are free of tsetse; overgrazing and. browsing has almost everywhere led to thin- Of the total area of the region, 9 percent is in this class. ning of the tree cover. Extensive smallholder cultiva- It is particularly important in Sudan, with more than tion occurs in Darfur and throughout Ethiopia (Berry, 330,000 square kilometers. The class forms abeltup to Taurus, and Ford 1980)..500 kilometers wide in south-central Sudan, south of In the past, vegetation was allowed to recover as about 8 N and between Sahel-Sudanian Acacia part of a bush-fallow cycle. More recently, cultivation Wooded Bushland (Class 44) and Sudanian Isoberlinia has become almost continuous. The effect of agrculsavannaandsudanianwoodland(class64).southern ture and grazing has been to degrade the woodland,. Darfur, Southern Kordofan, and parts of Jonglei also which on the slopes of Jebel Marra, for example, is are included in this zone. replaced by shrub or thicket Above 1,800 meters, the Dry SudanianWoodland almostsurrounds the Nile upper plains and peaks are now covered by montane floodzone, butpartoftheclassis wiffiin thesoutheast- grassland, swept annually by fire. This is almost cerem part of the flood zone. This class covers 15 percent tainly a fire-climax vegetation, although relict trees of the surface of Sudan. Extensive areas of the northern remain. Ripanan forest remains in inaccessible gorges section of the Ethiopian plateau, including much of as an indication of widespread former conditions Begemder, western ilgre, Goiam, and western Welo (White 1983). are within this class, and significant expanses exist in Firewood collection for Khartoum has caused the Rift Valley. considerable reduction in forest cover in Eastern In Sudan, precipitation is between 500 and 800 mil- Kordofan and Blue Nile (Lewis and Berry 1988). Fuellimeters; in Ethiopia, up to 1,000 millimeters. Gener- wood collection has severely depleted woodland in ally, areas of this class in Sudan lack pronounced relief Ethiopia around all towns, but particularly Asmera and lie below 500 meters, but in Jebel Marra the land (deforestationandovergrazingareprobablymorein- * rises to 3,057 meters. In Ethiopia, much of the class portant in this part of Ethiopia than elsewhere in the occurs on land exceeding 2,500 meters, and the topog- country). raphy is more pronounce& The Precambrian Base- As a result of these pressures, the landscape has ment Complex and Nubian (Cretaceous) sedimentary changed from mainly woodland with some grassland rocksunderliemuchof thesudanese section, although to parkland or very thinly wooded farmland, and in Jebel Marra, Jebel Gurgei, and part of the Ethiopian some places even to extensive treeless steppe. Rocky frontier section are on Tertiary basic lavas. Much of areas, or those where water for human use is scarce, this, however, is covered by the qoz (stabilized Quater- are the only areas where relatively intact Dry Sudannary aeolian sands) in the west, or by deep, dark, ian Woodland survives. Consequently, the estimated.impermeable cracking clays (descnibed. in the section growing stock is 23 percnt of the regional total (1,098. on Sahel-Sudanian Acacia Wooded Bushland, Class million tomnes), and the estimated sustainable yield is 44). These sands and clays are vastly different in their 23 percent of the regional total (31.7 million tonnes).

108 The Horn of Africa 95 It is almost certain that rates of wood depletion'now exceed the sustainable yield of this class and thiat into "Comlrnetum Thicket," "Croton Thicket," and "Ba lanifes Thicket." fuelwood shortage will became an even more serious. As in much of the Sudanese and Ethiopian woodcause of envirornmental deterioration arid of human land', clearance for. agriculture and collection of firesuffering, wood has considerably reduced the woody biomass stock of this class. Currently, the estimated growing Class 63-Sudan-Ethiopian Woodland stock is 5 percent of the regional total (259 million and Thicket tonnies), and the estimnated sustainable yield is 3 percent of the regional total (4.3 million tonnes). AO. This class, which is equally extensive in Sudan and (1963) comments that these woodlands are of value Ethiopia, occupies 2percent of the region. Most of it is for fuelwood and charcoal, but, reports aerial stiron the edge of the central Sudan platin and the foothills veys which suggest that; bamboo stands -may be and west-facing slopes of the Ethiopian plateau. Ex- more limited than was previously believed. Western tensions occur along river valleys, particularly the Sudarian areas have been severely thinned, and in Blue Nile and Atbara and their tributaries in Gojam Ethiopia, permanent cultiation doiates th area and western Welega provinces. Discontinuous excten- covered by this class. sions exist into the Blue Nile and Sobat province-s of Sudan. A small area in Bahr al Ghazal adjoins Undifferentiated Dry Sudanian Woodland, and other areas Class 64-Sudanian Woodland occur in less-frequently flooded areas of the Sudd. This class covers 2 percent of the region. in Sudan, a in Sudan, most of the clas lies below 600 meters, large stretch of Sudanian Woodland extends parale but in Ethiopia large areas exist above 1,000 mtrs to the southwestern border through El Buheyrat into. with some above 1,500 meters. Precipitation-in Sudan Equatorcia to the south of the flood region of the upper is about 600 to 800 millimeters and in Ethiopia. aboyut 1,000 millimeters. Both countries experience 5 or 6 Nile..Other areas exist on mountains in. Eastern Equatoria.lnElthiopia,numerouspatchesoccuronthe monthis with more thian 100 moiimeters Of rain. Thsis east of the plata bv h atfcn sapet reflected in the phenology for this class, which shows The largest area lies to the north of Addlis Ababa. a miarked seasonality, with a major contrast betweent isolated areas occur near Akcsum.and Sek'ot'a, and a thersmallindvr,values of 0.08 to 0.15 between December small area occurs anthe Hirer Hills. The class.occurs and May and values of 0.55 and 0.56 for August and at altitudes below the Moist Sudanian Woodland September. Values of NDow exceed 0.27 from June (Class 65) in Sudan; in Ethiopia, it is above the Escarpthrough November. TIhe class is developed on dark ment Wooded Thicket (Class 45). cracking clays in Sudan and on complex Precambrian Sudanian Woodland lies mainly above 2Z)00 meters, basement rocks and Tertiary lavas in Ethiopia. with considerable area above 3,000 mreters in Ethiiopia. Von Breiternbach- (1963) includes the class in his In Sudan, however, the class mainly occurs belowv 500 "'Lowland Woodland" and "Lowland Savanna" meters. Soils differ too, the Ethiopian soils beingmauinly classes. It forms the central section of White's Class 25b formed on basi volcanic lavas, whereas those in Sudan (1983), which extends along virtuallythe wholeborder are formed on the edge of the ironstone plateau and are between Sudan and Ethiopia. White identifies among lateritic. For mare than 6 monfts of the year, the Sudantthe prindpal constituents.: Anogeissus leiocarpus, Balan- ian area receives in excess of 100 milflimeters ofprip ifesaegptiacaboswelliapapyriferacombretumcollinum, tation each mnonth. But in Ethiopia, the dry seasn is C. hartmannianum, Commiphora ajricana, Dalbergia shorter (4 monthfs in Addis Ababa and Dea4) In both meknoxylon, Eryfhrina abyssinica and Terminala browvnii. areas, annuual rainfall. exceeds 1,000 milimeters. Von Breitenbach (1963) identifies an upper story at 5 This class is intermetdiate between Moist Sudanian to 12 meters of mainly deciduous trees, forming an W-oodland (Class 65) arid Dry Sudanian Woodland almost completely dosed canopy, bu loing suffi- (Class 62). It is best described as "Arogeissus-Khayadent light for a shrub layer of 1 to 3 meters height. Isoberlinia DeciduousWoodland"in Sudan (Bani1963), Rochetti (1961) records a zone. similar to WVhites, because it has fewerlisoberhinia spp. than the "Sudanian extending from Eritrea to southemn Ethiopia "le bois A Isoberlinia Savanna "proper. Trees are rarely more than feuilles caduques"-with.an almost identical list of 15 mneters high and species in thec south of Sudan tree spedies. Like Von Breitenbach, h-owever, he also include Acacia dudgeonff, A. gourmaensis, A-ntidesma. finds extensive areas of bamboo thicket, particularly v'enosum, Faunea salignza, Lophira lanceolata, Maprou-nea Oxytnnanthera abyssinica and 0. borzii in the higher africana, Maranthes pol yandra, Monotes kerstingii, Ochna areas. He describes the trees as quite. small with thick afzelii, 0. schweinflurfhiana, Protean madiensis, and Tertrunks. Von Breitenbach ([963) subdivides the class minaliaglaucescens (White 1983).

109 96 Estimathin WmWy Biotmiss i Sid'-SalSaran Africa In its natural state, tree cover is extensive in this Figur e NDVI Profile, Sudanian Woodland class, but cultivation has profoundly modified much (Class 65) of Class 64 through bush fallowing in less densely populated areas and by permanentcultivation in more 0c6 densely populated areas. In some areas, trees may be - )5-65 Sudan completply eliminated (White 1983). This is particularly true in Ethiopia, where the term "woodland" is 0.4- appropriate only in a historical sense. In Ethiopia, the woodfuel scarcity is comparable to that in Class 77 c 0.3- x (Highland Cultivation Mosaic), and the estimated growing stock is only 4 percent of the regional total D- (259 million tonnes). Further, the estimated sustain- 01 able yield is only 3 percent of the regional total (4.3 - million tonnes). - 0n, M,, A,, S -Jan Fb Mar Apr Mny Jan Jul Aug Sep 'da Noy Dc Class 65-Moist Sudanian Woodland bach's "Mountain Savanna" category (1963), although In this region, 4 percent falls within this category, he classifies it as "Lowland Savanna" in parts of northwhich covers 6 percent of Sudan. A broad band up to em Harerge, the Rift Valley, and the southwest. Von 300 kilometers wide follows the southwestem frontier Breitenbach identifies Acnda abyssinica, Agauia salicifolia, of Sudan. Isolated discontinuous patches occur farther - Buddiea polystadiya, and Ofinia usambarensws as characeast and in El Buheyrat as "islands" in the Sudd. In teristic of the more humid parts of the "Mountain Ethiopia, numerous patches occur on the central and Savanna." southwestem plateaus of Gonder and Shewa, in trib- Near Awash Station, field investigations in 1988 utary valleys of the Blue Nile and Atbara, and to the showed a mosaic of cultivation and grazing with north of Addis Ababa. Larger areas occur toward the some evidence of buming and a range of densities western boundary of the rift, and a series of Linear of crown cover of Acacia spp. varying from approxareas follows the foot of the: Hirer Hills near Dire imately 20 percent cover to large areas with rela- Dawa. Another area lies on the upper dipslope of the tively few trees. Near the railway, trees had been cut Hkrer Hills and numerous small areas exist in the for fuel. Over extensive areas, Opuntia spp., or valleys of Ilubabor and Gamo-Gofa. prickly pear, formed the understory or was the main The main extension of this class in Sudan is at about component of vegetation. 500 meters above sea level on the lateritic soils devel- As in the case of Dry Sudanian Woodland (Class 62), oped on the ironstone plateau. In Ethiopia, most of the human activity has severely altered climax Moist class is at considerably higher altitude, more than Sudanian Woodland, particularly in the highland areas 2,500 metes in places, and mainly occurs on volcanic of Ethiopia. In Sudan, both pastoralism and smallrocks having a great variety of soils. Precipitation in holder faming have reduced tree cover in the west, both cases is 1,000 millimeters or greater, with a wet whereas to the southeast smallholder farming has had seasonof 4monthsinSudanand7monthsinEthiopia. a majorinfluence (Berry, Taurus, and Ford 1980). Bush This is reflected in a phenology that reveals marked fallowing hasbeenanimportantfeatureofcultivation, seasonality and NDvi values that exceed 0.45 at the. and has reduced biomass, favored some species, and maximum (figure 10-2). probably altered soil conditions. When the fallow sea- Intact Moist Sudanian Woodland rarely exceeds 15 son is shortened, however, significant ecological meters in height It is taller than Dry Sudanian Wood- changes occur, and tree cover is much reduced. land (Class 62), but it is identifiable mainly through Sudan Survey Departnent (1954) shows a considerthe presence of Isoberlinia spp. which may locally dom- able area of Western Equatoria, which is now within inate this class. It is restricted, however, in the drier this category, as having been recently derived from northernwoodlands, where it may exist on rocky hills rainforest. Fuelwood demand in this area of Sudan is (White 1983). Other species present include Acacia less an it is farther north because of the lack of urban dudgeonit, A. gourmaensis, Antidesma enosum, Faurea settlements, and the productivity of the woodland is saligna Lophira lanceolata, Maprounea afiana, Maranthes greater. Consequently, more possibility of fuelwood polyandra, and Monotes kerstngit. productionfrom this class exists thanfromother types White (1983) argues that "Isoberlinia Sudanian Sa- of woodland. However, in Ethiopia, on the western vanna" is significantly different in stature and flors- plateau, the class is severely depleted. Field investigatics to enable its separation frbm miombo woodland. In tions in 1988 suggested, however, that a reasonable Ethiopia, much of this class falls within Von Breiten- reserve exists in the Awash Station area.

110 -he Hlan- of Africa 97 The estimated growing stock is 10 percent of the regional total (468 million tonnes), and the estimated (1963) recognizes a number of associations In this class, where the upper story is composed of trees 5 to sustainable yield is 6 percent of the regional total (8 million tonnes). 12 metershigh, with a thicketlike lowerstory of shrubs 1 to 3 meters high. Characteristic associations Include "Aa7ciaThicket," 'CombmhrtmhMicket," "Balanits'ThIcketf" Much of this class has, however, been profoundly Approximately 3 percent of the region is in this class. modified by grazing, burning, and clearance for agri- Class 73-Cultivation Mosaic and Forest Regrowth and "Craton Thiccet," with "Euphorbia Thicket" in drier areas. InEthliopia, where most of tis-classlies, and in Sudan, culture, so that cleared sections around farmed areas this class occurs in two distinctly different environ- may be effectively grasslands. Less 'disturbed bushments. Most of the Sudanese area is in a strip along the land occurs at a greater distance from settlements. floodzoneoftherivernilesouthof12 0 N. InEthiopia, Field investigations in 1988 in the area between Dire the class is widespread on the lower areas of the plateau, between 1,000 and 2,000 meters above sea Dawa and Awash Station showed that agriculture, using irrigation water from streams originating in the level. The largest single area, however, about 1,500 Hlrergi Hills, has localiy modified the land cover square kilometers, is on the dipslope of the eastem pattern. Significantly larger trees with trunk diameters plateau, on the slopes of mountains in Bale and up to 1 meter occur. in these better-watered areas. S5idamo provinces. Another series of irregular patches Prickly pear (Opuntia spp.) forms a common compoof Cultivation and Forest Regrowth Mosaic lies to the nent of the understory in this area, where tree density west of the southem end of the Rift Valley. Other significant areas occur toward the west of southem varies considerably. Consequently, the class describes a mosaic of differ- Gojam and Welega. In Somalia, small areas exist in the ent types of agriculture with various stages of relower Juba Valley. growth vegetation. Woody biomass stock varies con- Geologically, the areas are disparate. In Sudan, the siderably spatially, but it is estimated that the growing class occurs on the plains, dominated by cracking stock is 3 percent of the regional total (149 million dlays of Quaternary age and overlying Tertiary sedimentary strata. In Ethiopia, it exists on a variety of tomes) and the sustainable yield is estimated to be about 1 percent of the regional total (1.6 million extrusive volcanic rocks. Precipitation in Ethiopia is tormes). about 1,000 millimeters a year, with 6 or 7 dry months, whereas in Sudan precipitation is significantly less at 600 to 800 nillimeters. In the latter area, however, Class 74-Guinean Woodland considerable soil moisture is available because of the In this region, 2 percent lies within this class. It is water-holding properties of the clays and the flooding particularly extensive in Ethiopia, where it covers 3. of the Nile. percent of the country. A considerable area in south- The phenological curve for this class shows greater em Equatoria Province of Sudan also is within this NDVI values from April to November, exceeding 0.43 class, and scatteredpatches existwithin the flood zone for 8 months with the exception of September, when of the upper Nile, including some along the river itself. the value drops to Maximum values of 0.57 and LargerareascoverpartsofwesternandhighlandEthi occur in May and June. During the 4-month dry opia between ZO, and 3,000 meters above sea level season, the NDVI value falls to The class is particularly important on the western Most areas of Cultivation and Forest Regrowth Mo- plateau, in Gojam, Welega, Shewa, and Kefa, and a saic would fall within Von Breitenbach's "Lowland Woodland' class (1963), but parts of it, particularly small area occurs in Welo. An almost continuous strip follows the highest edge of the escarpment of the near the Kenyan border and in Gamo-Gofa, fall into his "Lowland Savanna" class. In Beal's description eastem plateau from the Harer Mountains, through the Gogu Mountains, to the lower sections of the (1968), most of this class occurs in the "Acacia Woodland-Savanna" zone, with AcaCia senegal, A. seyal, and Arsi Hills. This area also includes an outlier as far south as the hills overlooking Karsa Dek A. tortilis being widespread. On stonier soils, particu- Precipitation regimes in Ethiopia vary; rainfall is larly in western Ethiopia, combretaceous woodland is almost everywhere greater than 700 millimeters, but more common, with Combretumn molle or Terminalia approaches 1,400 millimeters on the edges of the rain brownii the domiinant species. Also in western Ethio- forest on the southwestern plateau. The wet season pia, Oxytenanthera abyssinica bamboo may be locally lastsbetween 6and 8months. In Sudan, elevations are important lower, but precipitation is still between 700 and 1,000 The grasses Themeda triandr and Hyparrhenia spp. millimeters for almost all of the area of this class. A are widespread below the tree layer. Von Breitenbach sample NDVI curve for southem Sudan shows the

111 98 Estimating Woody Biomassin Sub-Saimran Africa marked seasonality of vegetation growth in this class lyptus spp., Euphorbiaspp., and Ficus sycomorus, and in (figure 10-3). higher areas, grassland. This class does not fit easily into existing classifica- In such a varied environment, fuelwood varies in tions such as those of White (1983) or Von Breitenbach availability across a short distance. Reserves of wood (1963) in Ethiopia, although it accords with Bari's remain,andinsomecasesareobviouslyprotected.but '"High Rainfall Savanna Woodland' in S;,A' An (1968), woodfuel has become a marketed good rather than a which is characterized by Anogeissus-Khayn &negaln- free good. This, together with lopped and trimmed sis and Isoberlinia spp. Much of this ciasss is on the trees, is certain evidence of woodfuel shortage. As ferricrete ironstone plateau of Equatoria, which is else- population increases, the remaining wood will be where occupied by the remains of Isoberlinia wood- even more stressed, because it will be even more exland. Sudan Survey Department (1954), however, tensively used for fuel and because it is the only place shows that much of the area of this class in Equatoria where cultivation may be extended. Currently, the was recently cleared rain forest, where Cola cordcfolia, estimated growing stock is 2.4 percent of the regional Erythrophleum guineense, Khaya grantdoli, Mi?ragyna total (113 million tonnes). The estimated sustainable stipulosa, and Syzygium guineense were common spe- yieldislessatlpercentof theregionaltotal(l.3million cies. This interpretation also may be valid for some tonnes). areas along the upper Nile. In Ethiopia, however, it is unlikely that the class originally would have been Class 77-Highland Cultivation Mosaic rain forest, into which it grades on the southwestern plateau. Highland Cultivation Mosaic occurs on the plateau Elsewhere, because of precipitation and altitude on areas of Ethiopia between 2,000 and 3,500 meters both the western. and eastemr plateaus, much of the. above sea level and occupies approximately 17 perclass consists of degraded and largely cleared centof Ethiopia. ImportantareasareinShewa,Gpjam, Iuniperus procera and Podocarpus gracilior woodland. This has been reduced to a complex mosaic of farmed Welo, Gonder, and TiLgre, with smaller extents in Harerge, Bale, Eritrea, Sidamo, Welega, Kefa, and land with discontinuous areas of secondary woodland flubabor. This class coincides with Von Breitenbach's and thicket The amount of wooiland remauung re- 'Mountain Savanna" (1963) and. forms substantial flects the pressure of cultivation and grazing. areas of the drier sections of White's Class 19a, "Un- OntheEthiopianplateau,moreextensivelywooded differentiated Afromontane" (1983). It lies in the area areas are common on steeper slopes, although some of "Woing Dega" (Von Breitenbach 1963), or temperhills are almost completely denuded of trees and have ate highlands, the moderate temperatures reflecting beenreplacedbygrassland.fieldinvestigationsonthe the effect of altitude. plateau to the west of Addis Ababa in 1988 showed a Average precipitation is between 500 and 1,000 omlcomplex pattern: within a short distance occurred a limeters, increasing locally to 1,400 millimeters, and a landscape of extensively wooded escarpments, par- 7-monthdryseasonexists. Valuesof Nvm idicate high tially wooded and treeless hills, parkland savanna, productivity between June and November, with a and small copses. Indeed, every variation between full peakvalue of 0.6 (figure 10-3). The mostimportant soil woodland and treeless grassland was observed. Tree types on the lava plateaus are calcareous black soils species varied greatly and included Acaca spp., Euca- with a tendency toward desiccation and leaching on flatterareas, andbetterred soils onsteeperslopes (Von Breitenbach 1963). Figure NDvi Profiles, High Woody Biomass A dominant characteristic of this class is the severe Mosaic (Casses 74 and 77) effect of deforestation sparning thousands of years m the most densely populated part of the country. Here 0.6- the effects of cultivation, grazing, and fuel collection 0A Sudan - -n Et.biopia / / \ \. - combme to make difficult the identificationof prmary and secondary vegetation. Little tree cover now remains over large areas, although Von Breitenbach 0-3 / / \(1963) and White (1983) believe that forest and wooda / \land cover were formerly much more extensive. They -2 z / - / - R identify remnants of the former cover from Acacia 0iphocarpa (abyssinica), Apodytes dimidiata, HIagenia * 0.1- uabyssinica, juniperus procea,.nujxia spp., Olea afiiacn, and Prunus africana in mixed woodland or from o- - single-dominants such as Hagenia abyssinica,juniperus - Jan Feb Ma Apr May Jun Jul Aug Sep Oct Nov Dce praca, or Widdringfonia cupressoides.

112 Tte Horn of Africa 99 White (1983) believes that the single-dominant Class 84-Montane Forest stands may depend on fire, either natural or anthropogenic. Von Breitenbach (1963) identifies "Grass Sa- This class covers 1 percent of the region. It occupies 3 vanna" with Cyperaceae and Gramineae and "Scrub. percent of Ethiopia. Montane Forest is almost entirely Savanna" as a more arid variant A moister "Shrub restricted to southwestemethiopia,with the two larg-- and Tree Savanna" grades upward to an "Afro-Alpine est areas divided by the Rift Valley. The largest area is Savanna" with, for example, Lobelia rhtynchopetalum. In to the east in the Mendebo Mountains, with an outlier the same area, White (1983) describes areas of on the Gogu and Badda mountains. It has extensive afromontane bamboo dominated by impressive sections above 3,000 meters and occurs above 4,000 stands of Arundinariaalpina. meters in three areas. To the west, the other more Field investigations near Addis Ababa in 1988 re- dissected area exists in the mountavis of llubabor, vealed the complexity of landscapes of this class, with Kefa, and southern Shewa, with an outlier rising to cultivatedlanddominatingflatterareasandremnants 3,300 meters at Wallel. In Sudan, small areas occur in of a more extensive tree cover evident on some hills, the Didinga Hills and Dongotona Mountains of Nearby hills either were completely treeless or had EasternEquatona. odnly a partial tree cover. In a short distance, the land- Precipitation isbetween 1,000 and 1,500 millimeters scape encompasses treeless steppe, dense cultivation, in the west of Ethiopia, and much of the eastern area parkland savanna, small patches of farm frees, and receives more than 1,400 millimeters. In the western plantations, particularly of Eucalyptus spp. (see Class zone, only 2 months are dry, but in the east this in- 45). Eucalyptus globus was introduced in the late nine- creases to 5 months. Values of Noviare variable for this teenth century to provicle fuel and building wood for class(figurel 4)withanOctoberpeakvalue of nearly urban areas, particularly Addis Ababa. 0.6 The presence of trees in the landscape, however, In Von Breitenbach's classification (1963), the highmay give a false impression of the adequacy of fuel- est classes above 3,200 meters are mapped as 'Monwood supplies. In 1988, near Addis Ababa and in the tane Steppe," whereas most of the remainder is city itself, there were signs of a severe fuelwood prob- "Humid orsubhumnidmountainwoodlandl"accordlem. The indicators were extensive cutting of trees, ing to White (1983), the individual trees are usually parxzularly Eucalyptus spp.; the use of low-grade shorter than in the montane rain forests that exist at wood such as twigs for fuel; the carriage of fuelwood lower altitude. The majority of tree species are wideacrosslongdistances; the remarkablysmallindividual spread and not distinctive to this zone. Induded are parcelsofwood andcharcoalavailable forsaleinsome Apodytes dinidiata, Halena lucida, 1ex mitis, Nuxia constreets; the fact that men were involved in fuel coliec- gesta, Ocote bullata, Podocwpus fizlcatus, P. gracilior, tion as well as women and children; and that consid- Prmnus afrima, and Rapanea melanophloeos. Hagenia erable time was spent in wood collection. Dung was abysni may form dominant stands, as may Juperus sold as fuel even in Addis Ababa, a clear indicator of procra, creating Hagenia thicket and Juniperus thicket fuel shortage. Even in the country 50 kilometers to the Von Breitenbach (1963) refers to Junipues procera westofaddisababa,twigsanddungwerebeingused forest 45 meters high, but in much of the forest, the as fuel upper story does not exceed 20 meters, although Eucalyptus plantations on the edge of Addis Ababa Ekebergia spp. may reach 35 meters. On wetter sites, showed evidence of severe cutting, andfuelwas being Arundinariaalpina (bamboo) may form dense thickets. imported from at least 50 kilometers away; at this "Subalpine Forest," often festooned with lichens and distance, numerous coppiced plantations exist The high price'of a donkey load of small branches (5 bin) Figure NDVI Profile, Montane Forest (Class 84) implied a severe fuel shortage: in companson, the same load in Dire Dawa on the edge of the desert 0.6- would cost only 3 birr, confirming Kamweti's sugges- 84 Ethicoa tian (1984) that fuelwood shortage is most severe in 0.5- the highlands. Historically, this has been a region of severe fuel- 0.4 wood shortage and there is everyreason to expect that 03/ the shortage will become worse- Kamweti (1984) Z writes of "a worsening energy balance"' in the area of 02- this class. The estimated growhig stock is still 7.5 percent of the regional total (350 million tonnes), how- - ever, and the estimated sustainable yield is 3 percent of the regional total (3.9 million tonnes). Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dcc

113 100 Eslittiatitng Woody Bionaass hi Su b-sahmran Africa mosses, occurs in the highest areas (Beals 1968). "Side- Beals, E. W "Ethiopia." In 1. Hedberg and 0. roxylon Forest," with a dense canopy 10 to 20 meters Hedberg, eds., "Conservation of Vegetation in Afhigh and "Galiniera Forest" 10 to 15 meters high may rica South of the Sahara." Acta Phytogeographica be precursors to, ordegradedforms of,juniperusforest Suecica 54: (Von Breitenbach 1963). After fires, floristically mnixed Berry, L East Africa Country Profile-Sudan. Prostands may be replaced by almost pure stands of gramme for International Development Worcester, Juniperus procera, Hagenia abyssinica, or Widdrinlgtonia Mass.: Clark University. cupressoides (White 1983). Berry, 1, T. Taurus, and R. Ford East Africa Much of the natural forest in this class has been CountryProfiles.ProgramforInternationalDevelopremoved during the past century with the extension ment. Worcester, Mass.: Clark University. of agriculture, andconsiderableareasare now used for Bird, N. M., and G. Shepherd. 198B. Charcoal in Somalia: coffee plantations. This is particularly true of the west- A Woodfuel Inventory in the Bay Region of Somalia. em plateau of Ethiopia, where virtually no montane London Overseas Development Administration. forestremains. Elsewhere, crops are grown within the Booth, G. A Assessment of Restocking of the Gum remnants of the forest. Tumber extraction for commer- Beltfor Desertification Control. Phase I. Rome: rao. cial purposes has led to the removal of a wide range Douthwaite, R. J "Lowland Forest Resources of the larger trees. Of those previously mentioned, and Their Conservation in Southern Somalia." En- Apodytes dimidiata, Ekebergia spp., Juniperus procem, vironmental Conservation 14(1): and Hagenia abyssinica are used for sawn timber and.fo Forest Resource Reportfor Ethiopia. Rome FAO. plywood (FAo 1963). Firilayson,W.,G. S. Child, andj.j. VanRensburg Removal of timber commercially, or during land Forestry and Wild)ife Development Survey Mission, 12 clearance for agriculture, has greatly reduced the fuel- June July Sonalia. Rome: FAO. wood potential of this zone. This is one of the areas of Graham, A "Man-Water Relations in East Cen- Ethiopia where Kamweti (1984) concludes that, as- tral Sudan!' In NL F. Thomas and G- W. Whitsuinnig suitable access, fuelwood supplies should be tington, eds., Environment and Land Use in Africa. adequate to the end of the century, although that isnot London: Methuen. very long. At the present time the estimated growing Grove, A. T Aficat Oxford: Oxford Uraversity stock is 7.6 percent of the regional total (356 million Press. tonnes), and the estimated sustainable yield is 2.8 Hamilton, P., and J. Maizels "Introducing Flow percent of the regional total (14.9 million tonnes). Charts to Forecast Famine.' Geographical Magazine 61(7)34. Land Cover Class Tables Hamrouni, A. E Rapport au gouvernement de la Rdpublqupe de Djflbuti sur le ddveloppementforestier et Tables 10-1 through 10-4, beginning on page 102, pres- la lutte contre la d&rtficcatron. Rome FAo. ent summaries for each land cover class of the area, Hemming, CF "The Ecology of the Coastal Area showing growing stock and sustainable yield for the of Northern Eritrea." Journal of Ecodlogy Hom of Africa nations of Djibouti, Ethiopia, Somalia, Hemming, C R "The Vegetation of the Northem and Sudan. Region of the Somali Republic." Proceedings of the Linnaean Society of London 177(2): References Hemming, C F "Northern Somalia." In I. Hedberg and 0. Hedberg, eds, "Conservation of Every effort has been made to facilitate access to the Vegetation in Afnca South of the Sahara." Acta documents listed here. Some documents, however, lack Phytogeographica Suecica 54t full bibliographic information because it was unavail- Kamweti, D. M Fuelwood in Eastern Afica: Presable; also, some documents are of limited cuaatilon. ent Situation and Future Prospects. Rome: FAo. Kassas, NC "Desertification Versus Potential for Bally, P. K "Somali Republic South." In I. Recovery in Ciircum-Sahan Territories." In H E. Hedberg and 0. Hedberg, eds. "Conservation of Dregne, ed., Arid lands in Transition. Washington: Vegetation in Africa South of the Sahara." Acta Phy- American Institute for the Advancement of Science. togeographica Suecica 54f Kebbede, G., and M. J. Jacob 'Drought, Famine Bari, E. A "Sudan." In L Hedberg and 0. andthepoliticaleconomyofenvironmentaldegra- Hedberv eds., "Conservation of Vegetation in Af- dation in Ethiopia." Geography 73(1): rica South of the Sahara." Acda. Phytogeographica Lewis,L A., and Leonard Berry African Environ- Suecca 54:59-" ments and Resources. Boston: Unwin Hyman.

114 T--e Horn of Africa 101 Luchini, R "Useful Trmbers of Italian Somalia." Ethiopie. Florence, Italy: Istituto Agronomico per Worldng paper number 6, Ministry of Uvestock, For- l'oltremare. estry, and Range, National Range Agency, Mogadishu. Smith, D Tree Growth in Sudan. Publication 3. Moghraby, A. I., 0. M M. Al, and-m. T. Seed Khartoum: Sudan Department of Forestry. "Desertification in Western Sudan and Strategies Sudan Survey Department Vegetation of the for Rehabilitation." Environmental Conservation 14 Anglo-Egyptian Sudan. Khartoum. (Map scale (3): :4,W,OOAD.) Mooney, H. F Report on tfie Scopefor Forestry in Von Breitenbach, F The Indigenous Trees of Ethio- Somalia. Addis Ababa British Middle East Develop- pia. Addis Ababa: Ethiopian Forestry Association. mentdivision. White, F "The Vegetation of Africa.' Natural Okafo, 0. A Northern Rangelands Development Resources Research Series 20. Paris: UNESCO/AEMFAT- Project (NRDP). UrFN/soM/022, Forestry component, unso (United Nations Educational, Scientific and Terminal Report. Rome: FAo. Cultural Organization/Association pour l'etude Rochetti, G ProbThnes de reboissementforestier et Taxononiique de la Flore de l'afrique Tropicale/ de conservation des sols dans les pays d'otfre mer: United Nations Sudano-Sahelian Office).

115 102 Estimating Woody Biomass in Sub-Saiaran Africa Table Land Cover Classes-Djibouti (Horn of Africa Region) Area Growingstock Sustainableyield Thousand - llwusand tonnes Land cover cass kn? Percent tonnes Percent peryear Pcrcent 0 22, U Total 22, (Percentage of region) (0.53) (0O.D) (0.01) Note. In the following tables, detals may not add to totals because of rounding. Saurc& Authors' cakulations from data bases derived from land cover classification and table 4-1. Table Land Cover ClassesEthiopia (Hoom of Africa Region) Anrn Growingstack Sustdnable yid Thousand Thusand tonnes Landl cover dkos kin? Percent tonnes Percent per year Percent 0 134, O.DO , , , , , , , , , Q , , , , , , , , , , , , , , , , , ,38Z , , , , , , , , , , , , , , , , , , , , , , , , , , O.O 7 306, , , , , , , , ,281D , , , , , , , , , Lakes -4, Total 1,211, ,998, , (Percentage of region) (27.88) (42.78) (42.84) Sourer: Authors' calculations from data bases derived from land cover classification and table4-1.

116 Thie Horn of Africa 103 Table Land Cover Classes-Somalia (Horn of Africa Region) Area Growtingstack Sustainable yield Land cover closs kme Percent Thtousand tonnes Percent Thousand tonnes per year Percent 0 ~~~~~~~118, , , , , , *70, , ,049 64, , , , , , , , , , , , , , , , , , * 6 4, , , , * 7 1, , , or , Total 634, , , (Percentage of region) (14.5)(12) (6.64) Sam= Authiors' calculations from datat bases derived fromn land cover dassification and table 4-1.

117 104 Estimatipng Woody Biiovnass in Suib-Sriarmn Africa Table Land Cover Classes-Sudan (Hom of Africa Region) Area Growduig stock Sistainaiableyield 77toutsand Thjousand tontes Lantd cover class km2 Percent tonnes Percenft perilear Percent 0 840, ,478 90, , , , , , , , , , , , , , , , ,B , , , , ,360 36, , , , , , , , , , , , , , ,0; , , , ,652, , , , , , , , , , , , , , , , , , , , , Total 2,478, ,146, (Percentage of region) (57.01) (45.95) (50.51) Source: Author5' calculations from dala bases derived from land cover dassifioation and table 41.

118 Central Africa Terry D. Douglas This chapter presents a detailed description of the In places the grasslands are very lightly wooded, mostimportantlandcoverclassesinthisregion. Help- with shrubby and stunted species such as Burkea ful figures in other chapters include figure 3-1 (cloud africana and Hymenocardia acida. In Kaniama, where cover); figures 3-2, 3-3, and 3-4 (Nrvr sunmnary land the grasslands occur on the Plateau of Kasai 600 to cover profiles); figure 3-5 (regional summary map of 900 meters above sea level, the principal grasses are land cover classes); figures 7-1 and 7-2 (continental Andropagon schirensis and Hyparrhenia confinis. The maps of growingstockand sustainable yield); and the woody flora is Zambezian in affinity and very "Regional Land Cover Class Map of Central Africa" at sparse, although Acacia spp. have been reported. ihe end of this volume. Woody biomass is virtually nonexistent in this class Helpful tables in other chapters include table 3-2 and there is pressure on surrounding land cover (land cover classes); table 4- (data and sources for classes (miombo woodland, Classes 66 and 67) for growing stock and sustainable yield); and table 6-4 fuelwood. (centralafricaestimatedwoodybiomassbysummary class). Class 25-Edaphic Wooded.Grassland Class 11-Veld Grassland This wooded grassland class represents.1 percent of the land area of central Africa, occurrng largely in This class is of only minor importance in central Af- Congo (22,922 square kilometers) with a contiguous rica, occurring only in Zaire where it covers 0.6 percent area in Gabon (6,323 square kilometers) and a few of the land area (14,017 square ldiometers). The class isolated occurrences in Zaire (1,686 square kldometakes its name from the extensive (but floristically ters). It is just 0.04 percent of the region's growing different) grasslands of South Africa. Many Belgian stock The majority of this class occurs in one specific writers have descrbed the Zaire examples as "steppes." area between f1 S and 3S in the Plateaux Province of These are distibuted m areas along the Angola border, Congo and extending into southeastern Gabon to the most notably south of Kaananga, on interfiluves in the east of Francevillp. Other small occurrences are im- Kasai Basin, and in the upper Kwango Basin. These mediately to the north of Brazzaville. locations lie between 6 S and? S. Bounding these areas is Ombropbilous Humid Trop- Debate continues about the origin of these grass- ical Forest (Class 87) to the north, Seasonal Miombo lands. According to Mullenders (1954) and Devred Woodland (Class 66) to the south and west, and forest (1958), they are secondary grasslands, replacing wood- mosaics (Classes 72 and 73) to the east The class is land degraded by fires that maintain the grassland. The almost entirely restricted to the Batek6 Plateau, which dry season in these areas lasts 90 to 120 days (June to is composed of soft sandstones. The plateau surface is September) and annual rainfall ranges from 1,600 to gently undulating at about 600 to 800 meters above sea 1l800 millimeters. In Kwango, this class occurs on the level and is dissected by tributaries of the Zaire River, extensive Kalahari Sand-covered plateau where edaphic notably the L.finL factors are very influential The principal grasses here are A remarkable feature of this land cover class is that Ars!ida ven rystii and EDudetiz demesi it exists in the wettest part of Congo, where annual 105

119 106 Estliantitng Woody Biornass in Sleb-Salwrao Africa Figure NDVI Profile, Edaphic Wooded Grass- Class 62-Dry Sudanian Woodland land (Class 25) This class represents 1.6 percent of the land cover of central Africa, but only 0.3 percent of the estimated 0.s Congo growing stock. It occurs at the very north of Cameroon and Central African Republic (CAR) around 10 N. In OA - - Cameroon, it existsbetween the slopes of the Mandara Mountains and the lowlands south of Lake Chad. In a>0b3- ) / CAP, it occupies the lowland of the northern tip around Birao. It coincides with White's "Sudanian Undiffer entiated Woodland" dass; it is regarded as being within the Sahelian domain (Vennetier and Laclavare 1984). Within Cameroon, it covers 35,000 square kilo- - - meters (7.7 percent. of the land cover) and 28,000 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dcc square kilometers (45 percent) incar. The dry season is at least as long as the wet, and in northern CAR it lasts from November to May. Rainfall is precipitation is between 1,800 and 2,200 millimeters, less tian 1,000 millimeters a year, even in the elevated yet NDVI values are remarkably small for a site so near areas of northem Cameroon, and is less than 700 milithe equator within the Guineo-Congolian zone. meters in norhenicar. he rainfall is sufficient for cil- White's map(1983)doesnotdistinguishthisar-eafrom tivation, but is unable to support a dense vegetafion the surroundinghigherbiomass groups, althoughitis cover. For northemr CAR, KNw values are less than 0.15 particularly distinctive. from January to June, ir_easing to a little more than 0.3 Fora stationneardjambalainthecenterof this class, inaugustandseptember(figurell-2).thephenological NDvi values are consistently below 038, falling to 0.17 curves show dearly the basis on which the various in the driest month (iuly) and to 0.26 in December and Sudanian woodlands have been diferentiated. January (figure 11-1). Although the prime reasor. for This class is encroaching southward as the Sahel the low productivity of this class is the droughty na- undergoes desertification in response to cultivation, ture of the plateau soils (hence the name Edaphic grazing, and climate change. Inc,u much of this zone Wooded Grassland), regular burning has helped to has been designated a "Zone d'alarme" by the authormaintain the grassland area. ities in recognition of the desertification threat The mix of grassland and trees is varied. On the Characteristic tree species include Acacia seyal, Complateau, large areas are devoid of trees or have only brefum aculeatum, and Ziziphus abyssinicz. -The thorn a thin scattering, and have been descrbed as savan- woodlandissparseandinplacesdiscontinuous,much nas. Elsewhere, on steeper valley slopes in the Lfini of it being replaced by secondary thicket. Cultivation and Mpama catchinents and along the water- inthisclassispredomninantlyofsorghumandgroundcourses, dense woodland and gallery forest exist. In nuts. Withincreasing population and shftng agriculaddition; small areas of forest occur on the plateau, ture, the fuelwocd resource is treatened. The estiusually in patches of less than 1 square kilometer. The grasses grow to heights exceeding 2 meters and include Andropogon spp., Hypanrrnia spp., and Loudetia FLgure NDVI Profiles, Sudanian Woodland demensii. Many of the tree and shrub species are (Classes 62, 64, and 65) Sudarian in nature; of partiular importance are Acacia spp., Annona senegalensis, Hymenocardia acida, and 0.6- Khaya senegalensis.-in the islands of relict forest there - 62 Cennl African Republic exist Gilbertiodendron dewevrei, MMusanga cecropioides, 5-64CentrlAfrican Republic.. ' "., and other representatives of the Guineo-Congolian Cen Afdcan : Domain. Republic This class coincides with scant population density, in 023- particularly in the Plateaux Province of Congo. The. Z growing stock estimate for the region is lniflion tonnes with a sustainable yield of 0.3 million tonnes. Near towns such as Djambala in Congo,. some pres-. sure forwoodfuel clearly exists on the stock, although. o. this is localized. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

120 CentraeAfrica 107 mated growing stock in CAR is7s million tonnes and Estimates of growing stock are 133 million tonnes in Cameroon 93 million tonnes, with sustainable for the region and a sustainable yield of 2.3 million yields of 2.2 miulion tonnes and 2.7 million tonnes, tonnes. respectively. Class 65-Moist Sudanian Woodland Class 64-Sudanian Woodland This is the most extensive woodland class in central This class covers just 1 percent of central Africa but Africa north of the equator and represents 9 percent of accounts for only 0.5 percent of the growing stock. It the land area of the region. The growing stock is is restricted to the northernmost fringes of the region. estimated to account for just 4.58 percent The class InCameroon, it occupies 21,974 square kilometers (4.9 occurs largely within White's "Sudanian Isoberlinia percent of the area). In car, it occupies 27,000 square andrelatedwoodlands"class,butoverlapsthenorthkilometers (4.5 percent of the area). It occurs as a belt em edge of the Guineo-Congolian mosaic. It occupies between the Dry Sudanian Woodland (Class 62) to the a broad latitudinal belt from 20 N to 9? N. It is by far north and the Moist Sudanian Woodland (Class 65) to the largest land cover class in CAR, covering 263,000 the south. square kilometers (42.1 percent) and is important in It falls within the class described by White (1983) as Cameroon (65,000 square kilometers, 14A percent) "Sudanian Woodland with Abundant Isoberlinia." In andnorthern Zaire (43,157square kilometers, 1.9 per- Cameroon, it occurs up to 1,800 meters altitude in the cent). In Cameroon, it impinges on the northern edge Alantika Mountains and up to 1,400 meters in the of the Mesophilous Humid Tropical Forest (Class 85). Mandara Mountains. It stretches as a belt into the In CAR, it occupies most of the central area on plateau south of Chad and thence into northemcarandsouth- surfaces between 500 and 900 meters, where it exists em Sudan. In CAR, it exists in the Sudan-Sahelian cli- alongside Guinean Woodland (Class 74). Tn northern matic zone to the north of Ndel, with annual rainfall Zaire, it occurs in Oubangui and Haut Uele in the of 1,000 to 1,300 millimeters. Here the topography is extreme northeastnear the Sudan border. largely below 500 meters and is developed on the This broad belt of woodland is characterzed by Quatemary deposits of the basin, which drains north- annual rainfal}s of 1,400 to 1,600 millimeters and dry ward into Lake Chad. The dry season extends from seasons usually lasting 3 to 4 months from November November to March and thus seasonality is quite pro- to at least March. The values for NDvi increase to 0.5 in nounced. For this class in eastern CAR, NDvi values are the wettest month (figure 11-2) and show smaller valsinilar to Dry Sudanian Woodland (Cass 62) durng ues during the dry season. thedryseason,butincreasetomorethan DAinAugust As is the case with many of the Sudanian classes in and September (figure 11-2). central Africa, human activity has played an import- This class is intermediate between "Isoberlinia ant role in altering the natural Moist Sudanian Wood- Woodland" and the "Dry Sudanian Woodland" to the land, with pressure on tree cover and considerable north. It is deciduous and the woody species in CAR evidence of burning and the creation of secondary include Acacia spp., Combretum spp., Comrniphora spp., savanna The natural woodland often is dominated by Isobalina dok, Thaya segakensis, and Tennwlia Ia4flr Isoberlinaspp.,whichformadiscontinuouswoodland At a local scale, the vegetation may be linlked to cate- cover with a canopy 10 to 15 meters high. Other tree narypatterns(barber,buchanan,andgalbreath1980), species include Burkea africana, Daniellia oliveri, Erwhich are dependent on the distribution of laterite, throphleumafricnum, and Monotes kersingil floodplains, -and drainage. Cultivation has substan- Jongen and others (1960) studied an area south of tinaly modified much of this class, with tree degrada- the River Oubangui in northwestem Zaire (Bangala). tion occurring in the more populated areas. The They identified several communities at a small scale, French term for such areas is "Mosaique de Savanne with ripanan forests containing Guibourtia demeusei.arboree, de Savanne Arbustive et de Culture." In the and U.apaca heudelotii They obseved more open woodmore elevated parts of the class in northwestern Cam- land on the sandy interfluves and yet other variants eroon, Isoberlinia doka and other typical frees of the on the laterite crusts. In this region, Isoberlinia spp. are Sudaniansectoroftenhavebeenreplacedbydomesti- rare, and many of the tree species, such as Gilbercated tres, which include Acacia albida, Cdi itts nterfia, tiodendroi dezexrei, are derived from the humid trop- Khaya senegalensis, and Parkia biglobosa (Ladavre ical forests to the south. 1980). In many areas, for example in western CAR, this class The fuelwood potential of this class is partly threat- contains a quite dense rural population and, although ened by the extensive cultivation within it, notably woodybiomass growing stockisestimatedtobequite sorghum with some maize, cassava, and groundauts. substantial, in some areas of greater population its

121 108 Estinmatinig Woody Bionauss in Sub-Salomrn Africa sustainability may be threatened. In eastem CAR, along Figure NDVI Profiles, Mionbo Woodland the border with Sudan, thepopulation density is much (Classes 66 and 67) lower and fuelwood depletion is not currently a problem. Growing stock and sustainable yield are esti mated as follows: 0.5 Growvingstacrk Sustainableyicld Coinlry (million ton nres) (milli lonrcs) CAR z Cameroon Zaire (northern) OA Zairc 67 Z2irc Class 66-Seasonal Miomnbo Woodland - M Apr J Jul. -3an eb Mar Apr May lun Ju Aug Scp Oct Nov Dcc SeasonalMiomboWoodland is themostextensive land cover class recognized in this study in central Africa. the northern edge of this class is within the Guineo- It covers 21 percent of the region and accounts for 12.4 Congolian region. Hence, trees such as MAarquesia percent of the growing stock This class exists in a macroura are dominant to the north and Brachystegia broad belt stretching from Gabon (19,000 square kilo- spp., Isoberlinia spp., and Julbenardia are dominant in meters) through southem Congo (62,000 square Ilo- southernzaire. Where an understory is present, Proten meters) and across Zaire (768, square kilometers) from spp. and Uapaca spp. exist. Malaisse (1984,1985) has Bas Zaire through Kasai and Shaba. The class has been -discussed the structure of this woodland in the Shaba widely studied in Zaire (Malaisse 1978) and has been Province of Zaire near Lubumbashi. He describes a identified by Milington and Townshend (1989) in the transect at Luisnishi, 28 kilometers northeast of SADC region from 1984 AvHRRND>I imagery. Lubumbashi, where the tree layer attains 9 to 19 me- Because this land cover class spans such a wide area ters in height with a few dominant species reaching 22Z in central, southern, and East Africa, it does not coin- meters and, exceptionally, 29 meters. An inventory of cide with a particular climatic type, although its treespecieswithdiametersgreaterthanlocentimeters marked seasonality is a diagnostic feature. Rainfall showed Brachystegia spicijbnnis to be the most abuncomnmonly is between 900 and 1,300 millimeters a dant species, with. 150 individuals per hectare, folyear, but this disguises considerable year-to-year vari- lowed by Aidia micrantia (63), Syzygiumgttineense (52), ability, which in the Lubumbashi area of southern Porinari excelsa (49), Brtchlystegia taxifolia (43), and Zaire may be as much as 700 to 1,500 millimeters twelve other species with densities of more than six (Malaisse 1978). Miombo woodlands have been di- perhectare. vided into wetter and drier types. White (1983) sepa- The SeasonalMiomnboWoodlands area veryimportrated these at the 1,000-millimeter mean annual rain- ant source of fuelwood. Malaisse and Binzangi (1985) fall isohyet, whereas Chidumayo (1987) used 1,100 have considered the depletion of trees by fuelwood millimeters. gathering in Upper Shaba. For principal centers such The values of NDVI vary according to latitudinal as Lubumbashi, surrounding woodland has been position and other climatic factors. Southern Gabon clear-cut throughout a circle of radius 30 kilometers, woodlands exhibit the least seasonality in the region, and this zone is estimated to be expanding at about 1 with values between 027 and 0.5. In southern Zaire, kilometer a year. Projections indicate complete deforthe dry season is longer (May to September), with NDVI estation of more than 10,000 square kilometers survalues reducing to 0.22, but increasing to as much as rounding Lubumbashi by the year 2050, although just 0Q57 at the wet season of November to March (figure 20 percent of this area applied to forestry would be 11-3). suffiaent.to satisfy the woodfuel needs of the popula- Seasonal Miombo Woodland exeibits a structure of tion. Growing stock in this class in Zaire is 4,270 milspaced trees, which sometimes nearly interlock to lion tonnes, with a sustainable yield of 68 million fornmumbrellalike canopies. Where agricultural clear- tonnes. ance and burning are extensive, the woodland is often more open. The trees are deciduous in response to the Class 67-Wet Miombo Woodland marked dry-season lull in productivity. The undergrowth is rarely as well developed as in savannas- The Wet Mionmbo Woodlands form an intermittent band Seasonal Miombo Woodlands are mainly restricted to largely to the north of the Seasonal Miombo Woodland the Zambezian phytogeographical region, although (Class 66). They occur almost entirely in Zaire where

122 Ccntral Africi 109 they cover 268,000 square kilometers and stretch from miombo woodlands (Classes 66 and 67). Within central the Atlantic coast in Bas Zaire to the Rift Valley. This Africa it represents 5 percent of the land area and class covers 7 percent of central Africa and accounts accounts for an estimated 6.7 percent of the growing for an estimated 4.7 percent of the growing stock. It stock. It is particularly important in Congo (152 peroccurs in Kwango, on the Angolan border, and as cent, 51,000 square kilometers) and Zaire (5.3 percent, several patches throughout Kasai and Shaba on the 120,000 square kilometers);butitalso occurs in Gabon. plateau at elevations of about 1,000 meters. It also is The greatest extent of this land cover class is in central prominent on the hills near Kalermie to the west of Congo, to thenorthof Brazzaville and intheneighbor- Lake Tanganyika, where it extends over 1,500 meters ingparts of Bandundu Province in Zaire. The class also above sea level. In southern Zaire, near the border occurs in scattered pockets from central Gabon with Zambia, Wet Miombo and Seasonal Miombo eastward throughout central Zaire to Rwanda and Woodlands are intermixed on the higher plateau area Burundi. This class exists within the area mapped by to the south of Lubumbashi. White (1983) as "Guineo-Congolian Mosaic of Low- This class often occurs on light sandy soils of the land Rain Forest and Secondary Grassland." Kasai and Shaba plateaus and on the well-drained The climate of this belt is characterized by a dry soils of the hills near Lake Tanganyilka. Such edaphic season which extends for 90 to 120 days between June factorshave created anopen woodland with extensive and December (Laclav&re 1978) and has a rainfall begrassland cover. Human interference has meant that tween 1,200 and 1,800 millimeters a year. The values woody species are under pressure, and a well-devel- of Ncvx are consistently about 0.35 in the growing oped shrub layer is rare within these woodlands. season, diminishing to 0.23 in the dry season. Wet Miombo Woodland in Zaire generauy exists in The high woody biomass of this class results from wetter areas than the SeasonaLMiombo Woodland (Ciass being a forest remnant, secondary forest, or well- 66) and in areas where the effect of seasonality is less. wooded agricultural land with large patches of culti- Rainfallisgenerally1,000tol,Y0Dmillimetersayearand vation. The main crops include manioc, maize, the dry season is usually shorter than 3 months, al- groundnuts, bananas, and vegetables. The trees are though it is longer in Bas Zaire between Cabinda and varied, with some species occurring in the forest belts Angola. As a result, annual productivity is greater than to the north-cdtis zenkefi, Clorophorm exceisa, TerforSeasonalMiombo Woodland, with mv values falling minalia superba, and Triplochiton scleroxylon, and some below 0.35 for only Zmonths (usually July and August). thathave been introduced, such as the oil palm (Elaeis InthewetterpartsofcentralZaire,NvivaluesreachO-58 guineensis). The presence of many light-demanding during the wet seasonl The phenology shows marked species is attributed to the open nature of the forest m seasonality, with a big reduction in productivity during this mosaic, which is probably a function of disturthe dry season. The values of wiw for a site in eastern bance and regrowth. Zaire show wet season values weu above 0.5 from Octo- Woody biomass productivity and sustainable yield ber to April, but a sharp reduction in mom to 015 in are quite significant and fuelwood shortages are likely August (figure 11-3). to occur only m areas of concentrated cultivation. The The species occurringat thenorthem limitof this class reglonal growmg stock estimated for this class is 320 often include many from the humid tropical forest, but million tonnes, and the sustainable yield is 3.6 million Brartystia spp., kiobernia spp., Juibernardia spp., and tonnes. Marquesia macroura are dominant Also cominmon, however, are Aacia spp., Annona spp., Phitosfigma hnningii, Class 73-Cultivation and Forest Regrowth and Terminalia seicea. Quite dense thidcets exist in the Mosaic understory together with a mixed herbaceous layer. This class accounts for an estimated 3,191 million This land cover class occurs in two extensive areas and tonnes of growing stock in Zaire alone. The class is as a series of smaller patches. The area of greatest associated with quite high humana population densi- extentexistsingabon,where itextendsfromriomuni ties in some areas, especially near Kananga, Mbuji- (Equatorial Guinea) in the north, occupying a sector Mayi, and Kalemie. Maize, cassava, groundnuts, and from the coastnear Libreville and through much of the some cotton are the main agricultural crops. center of the country in the basin of the Ogooue River. In t-his area, commercial timber has been exploited Class 72-Cultivation and Forest/Woodland (Walker and Sillars 1967), and much of the area is now Mosaic cultivated. It is the second most important land cover class in Thisclass occurs south of the equator, generally at the Gabon, covering 21 percent of the land area (56,000 southerly edge of the Ombrophilous Humid Tropical square kilometers). In Cameroon, 47,000 square kilo- Forest (Class 87) and toward the northern limit of the 3meters are covered with this mosaic, some 9.4 percent

123 110 -Estimuiing Woody Biomnuss in SuibSahmarm: Africa of the land area. It occurs in the coastal lowlands to the Cultivation throughout this class includes large areas south of Douala and on the border with Nigeria to the of manioc. Other crops often.are grown in areas of north of Motnt Cameroon, where it attains altitudes secondary forest. to 1,500 meters. Elsewhere it exists in patches of vary- Fuelwood is relatively plentiful, as indicated by the ing size in eastem and southeastem Cameroon on the large growig stock of 285 million torrnes for the region. plateau at 500 to 700 meters and in patches throughout Sustainable yield is estimated at 3.1 million tonnes. the Guinean-Sudanian transition in CAR and northemmost Zaire. In car it covers 34,000 square kilometers Class 74-Guinean Woodland and in Zaire 32,000 square kilometers. Within central Africa as a whole, it covers 4.percent This class occurs in a belt to the north of the Mesophiof the land area and accounts for 2 percent of the lous Humid Tropical Forests (Class 85) where it shares growing stock. As a class, it does not correspond to a broad zone between 4e N and 8 N with Moist any of White's mapping units (White 1983). It occurs' Sudanian Woodland (Class 65). This zone runs from in conjunction with the forest classes and other high the west of Cameroon, where the Guinean Woodlands woody biomass mosaics (most notably Guinean are rather fragmented, through central car It contin- Woodland, Class 74, in car) and at the southern mar- ues to its southern extremity ini northern Zaire to the gin of tue Sudanian Woodland. The values of NDVI for south of the Oubangui River and in the vicinity of the this class in Gabon near the equator show a range of Parc National de la Garamba on the border with moderate values attaining 0.5.(figure 114). Sudan. The class constitutes 6 percent of the land area The key to understanding this extensive mosaic is of central Africa and is 1.6 percent of the growing that in most locations it has been greatly affected by stock. Thisincludes26.3 percent of car (165,000 square human interference. In West Africa, this land cover kilometers) and 13 percent of Cameroon (60,000 class is sometimes known as "Farm Bush." Elsewhere square kilometers). it corresponds with the "secteur Preforestier" (see With the exception of occurrences in Cameroon. on Chapter 9, Classes 75 and 76). It is largely evergreen in Massif de l'adoumaoua above 1,000 meters, the Guinthe wetter, western parts of central Africa and semi- ean Woodlands generally exist at altitudes between deciduous at the drier end of the range in eastern CAR 400 and 800 meters. The rainfall of this area exceeds and northeastern Zaire. Extremely high rainfalls of 2,000 millimeters a year in central Cameroon but is 2,000 to 4,000 millimeters a year are associated with typified by annual ranges of 1,300 to 1,600 nillimeters this class on the Cameroon-Nigeriaborder, whereas in in CAR. The wettest months are July to October and the the east, annual rainfall is only 1,300 millimeters. driest December to March. This is reflected in the In a class that covers many ecological zones, the tree vegetation productivity with rcvz values m central car species are of course quite vared. In Cameroon, spe- ranging from 0.16 in February to 0.52 in August. des from the "Atlantic Evergreen Forest' dominate, Ecologically, this class occupies a transitional zone notablycaesalpinaceae, whereas in Gabon andnorth- between the Guinean-Congolian floristic region to the em Zaire, species from the Mesophilous Humid Trp- south and the Sudanian floristic region to the north ical Forest (Class 85) occur. The remaining forest in (White 1983). As a result of the greater rainfall, woody this class is inevitably secondary, with light-demand- vegetation is quite dense in places, with trees and ing tree species such as Cfiorophora excelsa, Khaya an- shrubs separated by grasslands where buming for Eliotheca, Musanga cecropioides, and Terminalia superba. agriculture often has created local patches of herbaceous wooded savanna (Laclavr&e 1980). Gallery for- Figure NDVI Profile, Cultivation and Forest ests often exist along water courses, being more fre- Regrowth Mosaic (Class 73) quent at the southern edge of the zone. The main tree : 0.6 and shrub species include Burkea africana, Daniellia oliveri, Hymenocardia acida, Lophira lanceolata, and Gabon Uapaca heudelotii Many of these species have high calorificvalues and are importantsources of fuelwood OA/\ and high-quality charcoal (2 / \Growing C stock for this class is estimated at 440 In 0.3- /million tonnes and sustainable yield at 4.8 million zonnes. The combination of closed canopy woodland 0.2- os :and high sustainable yieldwith low to moderate pop ulation density indicates few woody biomass supply problems. Only when supplies from the neighboring o,,,,,,,, -,,,*, SudanianWoodlandsbecomescarceisthisclasslikely -Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec to come under pressure.

124 CentralAfricm 111 Class 82-Evergreen Forest occur. On thehigherslopes, the forest often exists only on the steepest rocky slopes wlere It has escaped This class occupies less than 2 percent of central AfriCd clearance. and represents an estimated 1.9 percent of the growing In eastern Zaire, this zone Is moderately well popustock. It occurs almost exclusively in eastern Zaire, lated and much of the forest has been cleared. Growwhere it represents a transition between the lowland ing stock is estimated for Zaire at 434 million tonnes, humid tropical forests of the Zaire Basin and the mon- with a sustainable yield of 36 million tonnes. tane classes with less woody biomass on the westem Most of the areas occupied by this land cover class wall of the Rift Valley. Leonard (1965) identifies this are rural, with few large urban ce.ters. Wood is not as area as representing the eastern extremity of the a rule exported from here and the only threat Is further Guineo-Congolian phytogeographic zone. cleara ;ce for agriculture. The class is represented by a scattering of small areas among woodlands and areas of secondary forest Class b6-mesophilous Humid Tropical Forest and cultivation; no one area is particularly extensive. This pattem probably results from the widespread Mesophilous Humid Tropical Forest is marginal to the clearance this class has undergone because it occurs narrow belt of the Ombrophilous Humid Tropical on accessible terrain between 1,100 and 1,800 meters Forest (Class 87). It occupies 10 percent of the land above sea level. The greatest concentrations exist on cover of central Africa and is an estimated 18 percent the Dorsale du Kivu and have been described in some of the growing stock. Its distribution in central Africa detail by Pecrot and Leonard (1960) and mapped by is best described with reference to those areas north Devred (1960). and south of the equator. The climate of this area is moist with the high hu- To the south of the equator, a coastal belt stretches midities typical of cloudy montane regions. Rainfall is from Cabinda in an unbroken swathe across Congo greatest on the upper slopes and temperatures highest and reaches its broadest development in southem toward the lowland margin, so a complex mix of forest Gabon, where it stretches from the coast for 200 kilotypes that changes rapidly with altitude is encoun- meters inland. A few other small patches exist in Zaire tered; this pattern is largely controlled by climate and to the south of the main area of Ombrophilous Humid altitude. The phenology shows a wet season of 6 Tropical Forest (Class 87). months from November to April, with mmv values To the north of the equator, this class is much more readcing 0.58 and a drier period from June to October extensive. Here it is an eastward continuation of the with vn values below 0.3 (figure 11-5). West African peripheral domain of the Guineo- The canopies of these forests are rarely as high as Congolian phytogeographic region (Leonard 1965). It thoseinthezairebasin,withtheupperstratumattain- extends from the Nigerian border, where the pattern ing20 to25 meters orlessathigheraltitudes. The lower is broken by the uplands of the Bamileke Plateau, slopes are characterized by contact with the lowland throughout central Cameroon into the north of Rio rain forest and species such as Cynometra alexandr, Muni, Gabon, and Congo. It continues via the south- Julbernardia seretii, Lebrunia bushaie, Pentadesma em tip of car to northemzaire, where it occurs as a large lebrwuii, and Symphonia globulifera are commonly en- area in Haut Zaire along the tributaries of the OubanguiL countered. Higher up,aningeria adoli-friedericii, Fical- Zaire has by far the, largest area within this land hoa laurifolia, Ocotea usambarensis, and Podocarpus spp. cover class of any country in Africa-more than 200,000 square kilometers (8.9 percent), whereas Cam- Figure NDVI Profries, Forests (Classes 82, 85, eroon has 19.6 percent of its area within t class, and 87) about 88,000 square kilometers. Gabon has 36,000 square kilometers; Congo 26,000 square kilometers; 0.6- car 40,000 square kilometers, and Rio Muni Z,500 square kdlometers. Phenologically, thits class demonstrates greater sea- 04- V \&t. /,,t' / : sonality than Ombrophilous Humid Tropical Forest (Class 87). Because some of the area of this class corren 0.3- \ sponds to White's "Drier Guineo-Congolian Rain 0.2- Eas em zaim 82 Forest," it is not surprsing that NDV values are lower. For a Southern Henmisphere site in southern Gabon, 0.1- I 8s airc Bsi, NDVI values increase from 0.32 in January to 0.46 in 87 Cetr air June, reducing dramatically in the dry season to about a-, 0.12 between July and Septeniber, and increasing to Jan Feb Mar Apr May Jun J'l Aug Sep Oct Nov 0.35 again by November.

125 112 Estimating Woody Biomnss in Sub-Saharan Afican In Haut Zaire, along the border with CAR in the It is principally surrounded by Ornbrophilous Oubangui Basin, these forests occupy land above 500 Humid Tropical Forest (Class 87) and includes much meters altitude. Here the NDVI values increase to of the area mapped by White as "Swamp Forest." greater than 0.48 from September through to May, and Many authors (for example, Lcbrnm and Gilbert 1954) reduce to 0.36 in July and August (figure 11-5). In havesubdividedswampforesttodistinguishbetween central Cameroon, to the east of Yaounde, on the south riparian or floodplain forest and forest that occupies Cameroon plateau above 500 meters, a similar NovI drained sites (often with waterlogged, clay soils). pattern exists to that in Haut Zaire, but with slightly Many swamp forests are periodically inundated with increased seasonality, indicating 2 months with low water, especially during the wet season. productivity-june (0.35) and December (0.27). Climatically, this class occurs in conditions very These Mesophilous Humid Tropical Forests, being similar to those of Ombrophilous Humid Tropical farther from the equator than either the Humid Trop- Forest (Class 87), with annual rainfall exceeding 1,500 ical Swamp Forests (Class 86) or the Ombrophilous millimeters.the NDvi values show little seasonality, as Hurnid Tropical Forests (Class 87), are generally char- would be expected of a class occupying an area so acterized by a more definite dry season, often with a dose to the equator. distinct period of 2 dry months. Rainfall for this land Phenologically, these forests are separated from the cover class ranges from 1,200 to 1,60 millimeters a Ombrophilous Humid Tropical Forests (Class 87) of the year, although some parts of westem Cameroon are drier soils by ow levels that are slightly smaller overall wetter. Most tree species are briefly deciduous, al- The depression of No values may be attnbuted in part though few trees are leafless at the same time; hence to the presence of standing surface water, which may the widely applied term "semideciduous. - appear through gaps in the even tree canopy. Its The broad extent of this forest type demonstrates a low reflection coefficient will depress NDcv values in wide range of species. Some areas have dominant.mixed pixls of standing water and tree canopy. species, whereas others are dharacterized by a consid- Wetter sites are host to species such as Guibourtia erable mixture In the Mayumba Forest of westem devneusetmitragynaspp.,raphiaspp.,symphoniagtobu- Congo and southern Gabon, the most important trees lijfera, and Ulapaca guineensis. Badly drained sites supare Chlorophora excelsa, Klainedoxa gabonenszs, Peter- port similar species, as well as Phoenix recdinata and sianthus macrocarpus, Sarcocephalus diderrihi, and Ter- Xylopia rubescens. minatlia superba To the north of. the equator, these Accessibilityisverylimited and population density forests often are floristically better characterized than in Gabon-Congo, with the occurnce of species such is generally very low in this class. Growing stock is estimated at 2,830 milion tornnes for the region, with as Celtis zenueri, Chlorophora excelsa, Musanga cecro- a sustainable yield of 327 million tonnes. pioides, and Piptadeniastrum afrcanum. Most of these more common species are widely distributed, from Class 87-Ombrophilous Humid Tropical Cameroon irough southernmost CARt in northern Forest Congo, and in the north of the Zaire Basin toward the border with Sudan. This class represents areas with consistently greater These forests often are inaccessible and dearly rep- wno values and corresponds to White's "Wetter resent a significant woody biomass resource. Agicul- Guineo-Congolian Rain Forest" and "Les Forets Omture is practiced in many areas of this land cover class, bropbilessempervirentesequitoriales" of Lebrun and but forest clearance has been local rather than exten- Gilbert (1954). It occupies a belt centered on the equasive.growingstockfortheregionisanestimated5,033 tor and largely within 30 or 40 of it, although a large milion tonnes, with a sustainable yield of 583 million outlier exists in central CuR at 8" N. Its broadest extent tonnes. is in the center of the Zaire Basin on terrain between 300 and 500 meters above sea level. It is a particularly Class 86-Humit Tropical Swamp Forest important class for woody biomass, covering 20 percent of central Africa and providing an estimated 30.9 This class is restricted to a narrow belt near the equa- percent of the growing stock tor, running eastward from the coast of Rio Muni It formsasignificantelement in thevegetationof all through northern Gabon, southernmost Cameroon, central African countries, with more than half a mil- Congo, and into Zaire in the province of Equateur. lion square kilometers in Zaire (24 percent of the area); Here it extends to the north of Mbandaka between the 95,000 square kilometers in Congo (28 percent); 55,000 Oubangui and Zaire rivers. Other isolated areas exist square kilometers in Gabon (21 percent); 63,000 square within the Zaire Basin- It is an important class, repre- kilometers in Cameroon (14 percent); 52,000 square senting 6 percent of the area of central Africa and 10.1 kilometers in Ct (8 percent) and 5,000 square kilomepercent of the growing stockl ters in Rio Muni (22 percent).

126 Central Afica 113 To the east, this class merges with Evergreen Forest Land Cover Class Tables (Class 82), to the north and west with the Mesophilous Humid Tropical Forest (Class 85), ae:d to the south Tables 11-1 through 11-6, beginning on page 115, preswith the two minombo woodlands (Classes 66 and 67). ent summaries for each land cover class of the area, Ombrophilous Humid Tropical Forest often occurs on showing growing stock and sustainable yield for the better-drained soils of the Zaire Basin. Within it, on central African nations of Cameroon, Central African riverine sites, patches of Humid Tropical Swamp For- Republic, Congo,. mainland Equatorial Guinea, est (Class 86) exist Gabon, and Zaire. The phenology of Ombrophilous Humid Tropical Forest reflects the generally even distribution-of rain- References fallof the equatorialbeltand thelackofenvironmental stress on plant growth. The values of NDVI are consis- Every effort has been made to facilitate access to the tently above 0.35 and usually are about 0.5 in the documents listed here. Some documents, however, wettest months. The least variation is displayed by lack full bibliographic information because it was sites in the vicinity of Kisangani (Zaire), whereas with unavailable; also, some documents are of limited increasing distance from the equator, a measure of circulation. seasonality occurs, with smaller values notable injuly and January. This class as a whole, however, is differ- Barber, K. B,S.A. Buchanan, and P. F. Galbreath. i980. entiated from surrounding land cover classes on two An Ecological Survey of the SL Flonis National Park, criteria,namelythelackofappreciableseasonalityand Central African Republic. Washington, D.C.: Internaconsistently large NDvi values (figure 11-5). Rainfar tional Park Affairs Division, National Park Service, within the region covered by Ombrophilous Humid US. Department of the Interior. Tropical Forest is evenly distributed throughout the Chidumayo,E.N.1987."SpeciesStructureinZambian year, with a mean annual total in the range 1,500 to Miombo Woodland." Journal of Tropical Ecology 3(2): 1,900 millimeters, although it may be locally greater The ecology of this type of rain forest is dominated Devred,R "Lav6g6tation forestiaredu Congo et by tall, closed-canopy tree species which typically du Ruanda-Urundi." Bulletin ie la Soci&6 Forestibre attain. 35 to 45 meters in height The fairly evenly Belge65(6): distributed rainfall with only a short dry season cre- Devired, R "La cartographie de la veg6tation au ates a semi-evergreen forest with a mixture of ever- Congo Belge." BulletinAgicolede Congo Beige 51(3): greens and some species thatbriefly shed their leaves Few detailed descriptions of this forest have been Fahem, A. K "Vegetation." In G. Laclav&e, Atlas published from the area of its largest extent in central de la Republique du Zaire. Paris: Les Editions Jeune Zaire. Here, as in much of the Zaire Basin, Brachystegia Afrique. laurentii and Gilbertiodendron dewevrei often are domi- Jongen, P., M. Van Oosten, C. Evrard, and J.-M. Berce. nant in the upper stratum of the forest Notice explicative de Ia carte des sols et de la Fahem (1978) reports that Gilbertiodendron spp. in vdtatidom No. 11, Ubangi [regioni. Brussels: ineac northeastern Zaire (Ituri) sometimes represent 80 per- (L'InstitutNational pour ['Etude Agronomique du cent of the canopy species. Cynometra alexandri, Congo Belge). Julbernardia seretii, Oxystigma ozyphyllurn, and Scoro- Laclavxe, G Atlasdela RipubliqueduZaire. Paris: *dophloeus zenkeri also are important, but gradual. Les Editions Jeune Afrique. changes occur in composition throughout the large Laclavere, C Atlas of the United Republic of Camarea covered by this class. For instance, in the area of eroon. Paris: Les Editions Jeune Afrique. this land cover class occurning in southwestern Cam- Lebrun, J., and G. GilberL 1954L "Une classification eroon, Cesalpinaceae are dominant. A detailed study ecologique des forets du Congo." Sei Scientifique of species in southern Cameroon is orovided by No. 63. Brussels. WEmAc (Llnstitut National pour Vivien and Faure (1985). The pror-lent northerly l'etude Agronomique du Congo Belge). outlier of this class in centralcar is less dense than that Leonard, J "Contribution a la subdivision phytoof the Zaire Basin and includes species such as geographique de la r6gion Guineo-Congolaise Enythrina tomentosa and Ochthocosmus africanus. d'apres la repartition g6ographique d'euphorbiacees Clearly, this belt of Ombrophilous Humid Tropical d'afrique tropicale." Webbia 19(2): Forest currently is one of the world's greatest re- Malaisse, F '"he Mtiombo Ecosystem. In unesoo sourcesof woodybiomass.growingstockisestimated (ljnited Nations Educational, Scientific and Culhural to be 8,644 million tonnes, with sustainable yields of. Organization), Troicl Forest E&systems. Paris UN!o. 1,218 million tomn as. Much of this forest remains gen- Malaisse, F "Structure d'une foret dense seche erally intact with very low population densities. Zambezienne des environs de Lubumbassi" Bullt-

127 114 Estimating Woody Bimass in Sub-Saharan Africa tin de la Soci6tA Royale de Botanique de Belgique 117(2): -Kivu [region]. Brussels: INEAc (L'Institut National 42&-58. pour L'Etude Agronomique du Congo Belge). Malaisse, F "Comparson of the woody struc- Vennetier, P., and G. LaclavLre Atlasde la Rpubture in a regressive Zambezian succession." Bulletin lique Populaire du Congo. Paris: Les Editions Jeune de la Soci*te Royale de Botanique de Belgique 118(2): Afnique Vivien, J., and J. J. Faure Arbres desforhts denses Malaisse, F., and K Binzangi "Wood as a Source dafrique Centrale. Paris: Agence de Cooperation of Fuel in Upper Shaba.' Commonweath Forestry Culturelle et Technique. Review 64(3): Walker, A., and R. Sillars Les plantes utiles du Millington,Andrew, and J. Townshend Biomass Gabon. Paris: Editions-Paul Lechevalier. Assessment. London: Earthscan. White, F "The Vegetation of Africa." Natural Mullenders, W "a vegftafion dexaniama" Sbie Resources Research Seres 20. Pars: UNESCO/AErFT/ Scintiuque No. 61. Bmrussels: w (LInslt National unso (United Nations Educational, Scientific and pour iretude Agronornique du Congo Belge). CulturlCrganizatin/Associationpourl EtudeTaxo- Pecrot, A., and A. Leonard Notic explicative de la nomique de la Flore de l'afrique Tropicale/United carte des sols et de la V$gitation. No. 16, Dorsale du Nations Sudano-Sahelian Office).

128 Central Africa 115 Table Land Cover Clasines-Cameroon (Central Africa Region) Arem Grouingstock Sustainableyield Thousand Thousand tonnes Land cover class km 2 Pernt tonnes Percent peryear Percent , , ,65 a , , ,0i , , , , , , , , , , , , , ,289 A , , , , , , , , , , , , , , , , , ,129, ,7B , , , , , , ,592, , Lakes 2, Total 451, ,099, , (Perenlge of region) (11.42) -(11.07) (13.75) Note: In the following tables, details may not add to totals because of roundin&g Sw= Authors' cakulatons from data bases derived frm land cover dassification and table 4-1.

129 116 Eslimnating Woody Biomnass in Sub-Salwnzn Africa Table Land Cover Classes-Central African Republic (Central Africa Region) Area Growing stock Suslfiizable yield Thousand Tlhotsaid tonnes Land cover cdass b,,? 2 Percent Ioniztes Percenzt per wear Percent 1 12 los S , ' , ioj ,212 1, , , , , , , , f , , , , , , A 6 321, , , , , , , , , , , , , , , , , , , , , ,183, ,986f Total 624, ,380, , (Percentage of region) (15.77) (8.50) (7.19) Souroc Autiois' calculations from data bases derived from land cover lassification and table 4-L Table Land Cover Classes-Congo (Central Africa Region) Area Grouwing stock Sustainable yield Th-ousand 7housand tonnes Land cover class kn? Percent tonizes Percent per year Percent , , , , ,423 1, , , , , , , , , , , , , , , , , , , , , , m , , , , , i 93, , , , , , , , , , , , !95 38, , , , , , , , ,083, , Total 338, ,600, (Percentage of regon) (85) (9-29) (11.44) SOUrc: AuthOrS' calculatins from data bases derived from land cover classifiton and table 4-1.

130 Central Afiica 117 Table 114. Land Cover Classes-Equatorial Guinea (Mainland; Central Africa Region) Area Groudng stock Sustainable yield Thousand Thtouswnd tonnes Land over class km 2 Percent tonnes Percent per war Percent , , , , , , O ,581 7.m 2, , , , , , , , , , , , , , , , , Total 22, , , (Percntage of region) (0.57) (0.96) (1.34) Saw=ct Author' calclibtions from databases derved from land cover dasification and table 4-L: Table Land Cover Classes-Gabon (Central Africa Region). Area Growing stock Sustainble yield Thousand Th1ousand tonnes Land cover cdss km2 Percent tonnes Percent per Vear Percent O.DO , , Z, , Z ,005 4, ,005.W0 4,005.) , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,719, , Total 261, ,993, , (Percentage of region) (6.61) (7.12) (9.04) Source: Aufhorst calcalaions fm databases derived fromnland cover dassificationand table4-1.

131 118 Estimating Woody Biomass in Sub-Saharan Africa Table Land Cover Classes-Zaire (Central Africa Region) Anra Growing slock Sustainable yield Thiousand P- Tiousand lonnes Land coverctass km' Pcrce7at tonnes Prcent per year Percent 1-14, , , , O AO , , , , , ,W , , , , , , , , , , l , , , , , , , ,270, , , ,191, , ,084, ,593, , , , A , , , , , , , , , , , , , , , , , , , ,559, , , , , , ,819, , , ,682, ,250, Lakes 18, Total 2,259, o00 17,659, ,361, (Percentage of region) (57.07) (63.07) (57.24) Sourc Authors' calculations from data bases derived fron land cover dassification and table 4-1.

132 East Africa ff2 Phil O'-Keefe, Ian Ryle, and John Kiarkby Thiis chapter presents, a detailed description of the year but is unreliable. A slight tendency exists for motmortantiland cover classes inthis region. Help- rainfall maxima in April and November. ful figures in other chapters include figur 3-1 (cloud Much of the area. classified.as Desert in thiis study cover); figures 3-2, 3-3, and 3-4 (NDvi summary land has a surface largely of bare rock. Other areas in this cover profiles); figure 3-5 (regional summary map of land coverdclssprobably represent White's-"semidesland cover classes); figures 7-1 and 7-2 (continental. ert Annual Gralssland" (1983) which, along with true maps of growing stock and sustainiable yield); and the desert, occupies the driest areas of Kenya, nearly al- "RegionalLand Cover ClssMap of East Africa"atthe ways below 1,000 meters. These areas are domidnated enid of this volune. b ythe grdsses Arisda adscensionis and A. mntabil is, but Helpful tables in other chapters incude tbe32 these may disappear completely during drou ghtperi- (lad cover classes); table 4-1 (data and sources for ads lasting several yearsgrowing stock and sustahinable yield); and table 6-5 Even here, however, woody plants are rarely ab- (East Africa estimated woody biomass by summary sent, and theyprovide 2 to 20lpercentof the vegetatio class). cover. Domninants may be either shrubs, such as Duospermn eremphilum or bushes and small trees, notably Class 0-Desert Acacia Jwomda, A. reficens, A seyal, A. fortilis, A senegal, and Comzmiphora spp. It is worth remembering that the The only significant areals of this land cover clas in the ftrnsitionfrombusbland.todesertthrough semidesert East African region occur in northern and eastern is very gradual, and many so-called desert areas may Kenya, especiallyinmarsabitdistrct andon the west- differ only in vegetation densty from the land cover em side of Lake Turkana, with smaller areas running clas that surrounids; them.l Thims maybe a result of local parallel to the KCenyan coast Desert covers 2.4 percent and climatic falctors at thte time. of Kenya, an area of 14,439 square kidlometers. Some vegetation is present in most areas classed as Accordingto Morgan (1973), only the ChalbiDesert -desert, and this presents opportunities for mnigrant of MarsabitDistrictisdlassed as true desert Itisa rock pastoralists,who use grasses and leaves as fodder. The desert of a type unique in East Africa and forms part vegetation is resident, adapted to survive irregular Of a closed drainage basin. The edge of the desert is periods of severe drought and to respond quickly to marked by n-umerous springs, suppliedby subsurface any water that bec-omes available. Locally, overuse water from the slurrounding mountains. The desert is may lead to at least the early stage of desertification. liable to occasional flooding, and subsequent evaporation of the floodwater results in an accumulation of Class 21-Semidesert Wooded Grassland salt, which inhibits plant growfth Locally, however, outlets of significant tributary streams formed by sea- Semidesert Wooded Grassland is restricted to the arid sonal floods may support a variety of ainnual herbs areas of north Kenya, where the annual -rainfall is and grasses, such as the halophytic grass Dnzkebrock- less than 250 millimeters, with maxima in April and mania somalensis. Rainfall is less than 100 milimeters a November. The greatest area is around Lake Turkana 119

133 120 Estimnaling Woody Biomass in Sub-S ahamn Africa in Kenya, extending southward and eastward toward country. The largest blocks of this vegetation are in the the border with Somalia. This class accounts for 17.7 northern provinces of Arusha and Tanga, but signifipercent of Kenya, an area of 104,600 square kilometers. cant outliers are scattered throughout the country, Growing stock for the region is 34 million tonnes and notably in Dodoma, Morogoro, and Tabora districts. sustainable yield is 1 million tonnes a year. The dry season here lasts about 6 months. The transition is gradual from Dry Acadca-Commi- Transitional Wooded Grassland also occurs in phora Bushland and Thicket (Class 41) to Semidesert Burundi, especially in the northwest, and covers 2.6 Wooded Grassland (Class 21). The latter consists percent (658 square kilometers) of the country. Growmainly of widely spaced dwarf shrubs and bushes ing stock for the region is about 2 million tonnes and with a grass cover that is barely discemible except in sustainable yield is 300,000 tonnes a year. the wetseason.the importantwoody species areaca- In Tanzania, this class is probably similar to the cia mellifera, A. reficiens, A. senegal, and A. tortilis ss. "Acacia-Commiphorn Deciduous Wooded Grassland" spirocarpa, along with Commiphora spp. Other genera described by White (1983). Canopy cover is less than indude Balanites, Boscia, Euplorbia, Jatropha, and San- 40 percent, with the woody vegetation dominated by sevieria. Various succulents also grow in these areas, Acacia spp. or Commiphora spp. thorn trees that form a and the grasses include Aristida spp. and Chrysopogon single, open stratum. Trees are usually between 4 and aucheri var. quinqueplmris (Trapnell and Langdale- 7 meters in height, although they may attain between Brown 1972). 9 and 20 meters in a few species. Very few bushes and Crown cover is usually less than 10 percent (White shrubs exist, although scattered bushes or small 1983) and continuous areas of taller trees are almost groups of Grewiafallax and Cordia ovalis occasionally entirely restricted to rocky outcrops and sites where form a very open understory. water is available, for example near streams. In drier The grass cover is fairly well developed and ranges areas, with about 100 millimeters of rainfall, woody from 0.5 to 1.5 meters in heighl Principal species plantcoverislessthan3percent, andindividualplants include Digitaria macroblephara, Eustachys paspaloides, usually are less than 3 meters high(white 1983). Grasses and Themeda triandra on well-drained soils and Pensuch as Cendirus biflorus and Panicum turgidum domi- nisetum mezianum on poorly drained soils. The small nate the drier areas. amount of Transitional Wooded Grassland in north- Much of the area of Class 21 has long been used for western Burundi lies in an area of 'Wooded Savanna" grazing and browsing. This has strongly influenced dominated by Acacia spp. (MHirit 1986) and will be the grasses, herbs, and trees, because the more palat- described more fully under Open Woodland (Class able species such as Acacia sencgal are suppressed. 61) in this chapter. Fluctuations in rainfall and occasional droughts can The other areas probably correspond to the most be tolerated by the resistant vegetation, but overuse important East African savanna type, the "small Tree of. the biomass by pastoralists has caused general Savanna." This is dominated by various broad-leaved degradation of the environment, with severe social Combretum species, most commonly C. binderanum, C consequences. ghasalense, C molle, and C zeyheri. In wetter areas, the "small Tree Savanna" also includes large-leaved Ter- Class 24-Transitional Wooded Grassland minalia spp., especially T. glaucescens and T. mollis. These are replaced in drier areas by smaller-leaved Transitional Wooded Grassland occurs at heights of species-t.browniiin Kenya and T.sericeainTanzania. approximately 1,000 meters, mainly in north Tanzania The ground cover between the trees is mainly tall and south Kenya, but occurs also in scattered areas of grassland, dominated by Hyparrhenia spp. It is possicentral and southern Tanzania. ln Kenya, it grows ble that the savanna flanking the Kenya Highlands extensively on the southern and eastern flanks of the includes certain other Acacia spp. observed in southhighlands in the south, and forms a distinct band em Kenya (Trapnell and Langdale-Brown 1972); spealong the coastal hinterland separating the MoistAca- cies involved include Acacia gerrardii, A. nilotica, A. cia-commiphoma Bushland and Thicket (Class 43) to- senegal, and A. seyal ss. subalata, in a grass layer domiward the coast from the Dry Acacia-Commiphorm Bush- nated by Themeda spp. land and Thicket (Class 41) of the interior. Land use in Transitional Wooded Grassland is Rainfall in this area is about 250 millimeters, with a more variable than in the groups previously de- November-December mnaximurn' This class accounts scribed. Pastoralism dominates, although extensive for 1.1 percent (6,376 square kilometers) of Kenya. The areas are unused. Extensive smallholder farming is Transitional Wooded Grassland of Tanzania is more carried on in more favorable areas (Berry, Taurus, scatteredthanin Kenya,althoughitdoescoveralarger and Ford 1980). The pressure of use in this class is area-23,607 square kilometers, or 2.5 percent of the fairly light-

134 East Africa 121 Class 33-Bushy Shrubland where else in EastAfrica. The "MontaneMoistForest" itself is dominated by Ocotea usanibarensis and Most Bushy Shrubland in the East African region oc- Podocarpus milanjianus, among.other species. All of curs at elevations between 500 meters and 1,000 meters these zones fall within the boundaries of the Aberdare along the equator in Kenya, where it grows extensively National Park. on the lower eastern and southern slopes of the Central On the lower slopes of the-aberdare Mountains, as Highlands, and to a lesser extent at lower elevations in well as the area covered by this class in the highlands the lower Tana Valley. Elsewhere, significant areas of southwestern Kenya, the vegetation is probably of exist on the eastern slopes of Mount Kilimanjaro and evergreen and semi-evergreen bushland, forming a in the highlands on the Tanzanian border, east of Lake transition between the montane vegetation above and Victoria. In all, this class accounts for 3.7 percent of Acacia-Cornrniphora bushland below. This transitional Kenya, an area of 21,552 square kilometers. vegetation has been observed on a number of drier Elsewhere in the region, Bushy Shrubland is con- mountain slopes in East Africa, and is described more fined to Tanzania, where its distribution is more scat- fully in the section on East African Low Woody Biotered. Patches tend to occur within and around the mass Mosaic (Class 52). A similarvegetation occurs oni areas of Moist Acacia-Cormiphora Bushland and the lower slopes of Mount Kilimanjaro, although the Thicket in central Tanzania, notably in Arusha, Kili- moister conditions here give rise to upland rain forestmanjaro, Tabora, and Tanga districts, with outliers in type vegetation at slightly higher altitudes (see Mbeya District on the eastern shore of Lake Rukwa. Coastal and Gallery Forest, Class 83). The area of Bushy Shrubland in Tanzania is 10,065 The remaining areas of Bushy Shrubland are rather square kilometers, 1.1 percent of the country. small, and scattered throughout the Moist Acacia- The large block of Bushy Shrubland in eastern Commiphora Bushland and Thicket (Class 43) of Tan- Kenya lies at the southern fringe of the. Semidesert zania. The vegetation in these areas is probably of the Wooded Grassland (Class 21), which dominates the scrub forest type, more open than the surrounding arid interior of the country. It is likely that this vege- bushlanid and thicket. A dense understory of 3 to 5 tation is intermediate between Semidesert Wooded meters height exists, dominated by Commiphora spp., Grassland (Class 21) and Dry Acaca-Commiplora among others. The main emergents are Adansonia Bushland and Thicket (Class 41) in areas receiving digitata and Euphorbia spp. annual rainfall between 200 and slightly less than 250 Land use in Class 33 is limited by the sparse and millimeters. Reaching only 2 to 3 meters in height, the seasonal rainfall. Grazing is of poor quality and the vegetation consists of small bushes and stunted trees, carrying capacity low. Given the small population of dominated by Acada reficiens ss. misera, forming a the areas of Class 33, little degradation of the biomnass sparse canopy over a ground layer of small shrubs. occurs. Regional growing stock is 34 million tonnes The grass cover is ephemeral, with annual species and sustainable yield is 1.7 milion tonnes a year. appearing only briefly after rain. The largest single area of this vegetation is in the Class 41-Dry Acacia-Commiphora Bushland Aberdare Mountains of south Kenya, mostly concen- and Thicket trated around the lower eastern slopes but attaining higher altitudes. The recording of Bushy Shrubland at This class is most characteristic of Kenya, of which it such altitudes may result from amalgamation of a covers 36.4 percent (214,312 square kilometers). Domwide variety of montane vegetation types, ranging inating the eastern third of Kenya, it also extends west from heatliland to forest, an artifact of the coarse spa-' along the north of the CentraL Highlands as far as tial resolution of the AVHRR imagery used in the study. Uganda, whereitcovers8,115squarekilometersinthe The Aberdare Mountains host a well-developed northeast of that country. In Tanzania, itis confined to montane vegetation, with the highest altitudes sup- the north, occurring mainly in Arusha and Shinyanga porting Afro-alpine communities characterized by districts. Considerable areas also exist as inclusions giant Senecia and Lobelia species. Farther down is the within the Moist Aacia-Commiphomi Bushland and "Shrub and Moorland Zone," dominated by Erica spp. Thicket (Class 43) of Tabora and Singida districts and Next comes the uppermost layer. of the montane. in the border districts of Tanga and Kilimarjaro. Dry rainforest belt, the"hagenia-hypericumzone," consist- Acaca-Commiphora Bushland and Thicket covers ing of vegetation stands between 9 and 15 meters in 36,044 square kilometers in Tanzania, 3.9 percent of height, dominated by agenia abyssmica and Hypenium the country. rewlutum. The middle of the forest zone is character- Values of NDvi are quite small and reflect the semiized by giant bamboo, Arundinaria alpina, which is annual rainy season (figure 12-1). Class 41 occurs at more extensive on the Aberdare Mountains than any- elevations between 200 and 1,000 meters.

135 122 Eslimnting Woody Biomnass in Sub-Sal arami Africa Figure NDVI Profiles, Bushland and Thicket addition to cutting for fuelwood. In consequence, (Classes 41 and 43) some degradation of the quality of the biomass occurs. Because the woody vegetation is often spiny, this Kcayaf - - vegetation class provides problems, both of access to - 41 Kenya fuelwood and its use. Both the growing stock and -Central -43 Tmwanh - - sustainable yield of the woody biomass resource are OA- - small, and considerable clearance for fuelwood and 5.3 < / charcoal already has taken place-even in protected a 0.3- > /\ - / areas like Serengeti National Park. Although biomass *02- destruction by-elephants and other gamehas occurred 0.2 / \ / \ //in recent years, this problem is more serious in the 0.1- V - f bushlands of Kenya than in Tanzania. a- * * a -T-. Class 43-MoistAcacia-Commiphora Jan Fcb Mar Apr May Jun Jul Aug Sep Oct Nov c Bushland and Thicket This area coincides with Moore's "semidesert' cat- Moist Acacia-Commiphora Bushland and Thicket covegories (1971), being characterized by deciduous ers a large triangular area of central Tanzania, extendbushland and thiket of markedly seasonal growth ing from Lake Victoria in the north almost as farsouth and extensive grassy plains with no woody vegeta- aslakenyasa,andreadhingtheusambaram ountains lion. The bushland is dense, varyingbetween 3 and 5 in the east Elsewhere in Tanzania, outliers exist in the meters in height, with very occasional emergent trees southeast and around Lake Rukwa in the southwest, to 10 meters. Tree species are dominated by Acacia giving an area of 174,052 square kilometers, or 18.8 spp.andcommphoraspp,orsatbush(suaedamowoic), percent of the country. In Kenya, where Dry Acacdain vegetation that has been described as 'Acacia- Commiphora Bushland and Thicket (Class 41) predom- Commiphora thoom savanna" (Lucas 1968). mates, this moister variant grows in areas of greater However, studies of this vegetation class on the rainfall-notably along the coast and in the southen Serengeti Plain of northern Tanzania indicate a much and western highlands. greater dominance of grassland communities than Moist Acacia-Commiphora Bushland and Thicket is suchaname would suggest, especially on soils formed much less extensive in Kenya than in Tanzania, coveron young volcanic ash, hardpan soils, and black crack- ing only 6.6 percent of the country (38,836 square ing soils (mbuga). The grasses provide only 15 to 45 kilometers). Other important areas exist in central and percent ground cover and hees are rare, restricted to southern Rwanda (5,796 square kilometers, or 30 per- Acaia mellifera bushes on young volcanic ash. These cent of the country) and on the central plateau of secondary grassland communities are controlled by Burundi (2,319 square kilometers, or 8.7 percent of the burning and, to a lesser extent, by browsing and graz- land area). ing. When the interval between burning increases be- Values of NDvi are invariably greater than those for yond 5 years, Acacia spp. are able to regenerate, be- Dry Acacia-Commiphora Bushland and Thicket (Class coming fire resistant 3 to 4 years later. 41) and reach OA during the wet seasons (figure 1-1). The areas of "Acacia-Comnmiphorm thorn savanna" In Tanzania, extensive areas of Class 43 occur between (Lucas 1968) are dominated by Acacia spp., especially 1,000 and 2,000 meters, but the dlass also exists at low A. melif ua, A. seyal, and A. tortilis, and CoWJmiphora elevation on the Kenyan coastal plain. spp., especially C madagascariensis and C merkeri. In I. n Tanzania, the vegetation is mostly dense bush- Kenya, the transition to semidesert vegetation is land of 3 to 7 meters height, with occasional emergent marked by Acacia reficiens ss. misera Other species trees to 20 meters, and is characteristic of the semiarid commonly indude Boscia spp. and Grewmia spp. The regions of the country. The trees are generally fire most common emergent trees include Adansonia resistant, with evergreen species making up only 2.5 digitata and Euphorbia spp. Within the wooded areas, to 10 percent of the trees and shrubs. Grass cover is thegroundcoverisofscattered tussoclsofgrass,often thinner than in more open bushland, but may still dominated by Aristida spp., and occasionally reach a height of 1.5 meters. It is dominated by tall Sporobolus robustus.. Hyparrhenia spp. with shorter Panicum, Setaria, and Growing stock for the region is 359 million tonnes Thlemeda species. and sustainable yield is 55 million tonnes a year. The principal tree species are Acacia, notably A Migrant pastoralism is the dominant land use, with gernii, A hockii, A. mellifera, A. nilotica, A. seyal, and burning of grasses to encourage fresh, palatable A tortilis, and Commiphora, especially C. africna, C growth and, locally, concentrated browsing of trees, in cuenilea, C mollis, and C schimperi. Other common tree

136 -East Africa 123 species include Adansonia digitata, Boscia coriacca, pylacantiha, A. hebecladoides, A. seyal ss. rnultijuga, and CadabafArinosa, Cadia spp., Delonixelata, Lannea spp., A.sieberana.Theareasof'WoodedSavanna"aredom- Sterculia spp., and Terminalia spp. A shrub layer also inated by Conbretum elaeagnifolium, and include Acacia exists, dominated by Combretum aculeatum, Greruia campylacantha, A. hebecladoides, and, more occasionspp., and Maerun spp. These regions also contain suc- ally, Dombeya madiensis and Gardenia jovis-tonantis. culents and climbers. In central Burundi, where this class also occurs, In areas where branches interlace and impede pas- most of the area is taken up by coffee, tea, and sage, a thicket is formed within the bushland. These tobacco cultivation. However, the vegetation also may be small patches around old termite mounds or may include Exotheca abyssinica savanna, replacing mayextendoverhundredsof square kilometers.inthe the original humid forest, and Ericaceous and Afrodrier areas of Tanzania, thickets often are fully decid- alpine vegetation dominated by Agrostis spp., uous and contain a rich variety of plant species. The. Deschampsia spp., Festuca spp., Koeeria spp., Pentasbest such example is the "Itigi Thicket" coverng 62.0 chistis spp., and Poa spp. square kilometers of the central plateau. The principal This is regionally one of the more important producspecies here are Baphia massaiensis, B. burttii, Bussea massaiensis, and Pseudoprosopsis fischeri, forming a tive classes; growing stock is 378 million tonnes and sustainable yield is 4.5 million tonnes a year. dense thicket 3 to 6 meters in height, occasionally The greater precipitation associated with Class 43 broken by emergent evergreen and semi-evergreen allows a wider range of land uses than in Class 41. trees to 8 meters. Bush fallowing allows the regeneration of tree cover, Other species associations within the thicket in- butwithabiastowardfire-resistantspecies.sedentary dlude taller stands of Crnibia bmvicaudata ss. burtii, agculture entails almost complete loss of trees, with Combretum trothae, Grewia burtii, and Tapiphyllumflori- a tendency for deforestation to spread outward from bundumn Other important types of thicket include: settlements. The "Commiphora-CordylaThickets" ofeasterntan- Class 51-Acacia Woodland Mosaic zania, dominated by Croton spp., Hippocratea spp., Lannea spp., and Stnjchnos spp.- This class, which covers 3.4 percent of the region, is The "Commiphora-Cordyla Thickets" of central Tan- almost entirely restricted to Kenya (43,105 square kirama, with species including Aacfa circummarginata, lometers) and Tanzania (18,970 square kilometers). Commiphomaedea,Chombyi,Cn*ed,andConlyla In Kenya, the main occurrence is a large arc to the de flora south and east of the Central Highlands at an altitude * "Euphorbia Thicket' of about 1,000 meters, with Dry Acacia-Commiphora. * A thicket occurring on rockyhills that is dominated Bushland and Thicket (Class 41) adjacent to it at lower by Dalbergia spp., Diospyros spp., Dombeya spp., altitudes. Another area forms an arc on the southem Markhamia spp., Strychnos spp., and Teclea spp. margin of Tsavo National Park The third Kenyan area is around the equator on the Somalian border. In drier upland areas, the transition between Moist Northern Tanzania has two main occurrences: an Acaci-Commiphora Bushland and Thicket and Mon- area running east-west around Arusha, and a smaller tane Forest (Class 84) is marked by an evergreen to arc on the eastern side of Serengeti National Park. semi-evergreen bushland 3 to 7 meters in height This Growingstock for this class is 138 million tonnes, with rather dense vegetation is dominated by Acokanthera a sustainable yield 3.9 million tonnes a year. spp., Carissa edulis, Dodonaea viscosa, Euclea spp., Olea Rainfall may be as highas 750 millimeters, but to the africana, Tarchonanthus camphoratus, and Tecea spp. south of Tsavo National Park and on the Kenya-Soma- The presence of this vegetation class around Lake lia border, it is below 500 millimeters. These two areas Rukwa in southwestern Tanzania is the result of in Kenya nave two wet seasons, as does the main miombo woodland clearance. Repeated clearance for Kenyan area. The two areas in Tanzania have one wet cultivation here has led to a decline in the original season. miombo species, which was replaced by fire-resistant. Land use has contributed significantly to the rharspecies common to Combrelum savanna, especially acteristics of the vegetation in much of the area, with Brachystegia spiciformis, Combretum mechowianum, pastoralism dominating in the drier areas, moreorless Diplorhynchus condylocarpon, and Syzygium guineense. intensive smallholder farming in moister areas, but In Rwanda, this class consists mostly of "Herba- with sections in Tanzania unused. Small parts of ceous Savanna," but also includes areas of "Wooded TsavoandSerengetinationalparkslieinthisclass.The Savanna," especially in the north (Am 1987). The 'Her- main modification to the biomass in Class 51 has been baceous Savanna" is dominated by T1emeda triandra, through clearance of trees for agriculture, grazing, but includes scattered individuals of Acacia cam- browsing, and buming of grassland. Considerable

137 124- Estirnalin,gWoody Biomass in Snb-Salmomn Africa tree cover, has been lost in the main Kenyan extent of Class 61-Open Woodland this class, so that cultivation and grazing land dominate except in the less-favorable areas of higher, Open Woodland exists throughouttanzania, where it steeper, rockier slopes. Charcoal buming also has occurs as small patches scattered among larger blocks caused conversion to grassland (White 1983). of Seasonal and Wet Miombo Woodland (Classes 66 Acacia spp. dominate the tree cover of Class 51, with and 67) and Dry and Moist Acacia-Comndplwora Bushsome Commiphorn spp. Evergreen species such as Car- land and Thicket (Classes 41 and 43). Together these issa edulis, Dodonaea viscosa, Euclez divinorunr, and E. blocks add up to an area of 21,025 square kilometers, racemosa may be present, with a tendency for Acacia covering 2.3 percent of the country. The remainder of spp.to invade areas degraded by concentrated brows- this woodland exists in. Burundi with 1,212 square ing. Under severe browsing, woody plants may be kilometers along the northwestem border and in the absent or reduced to bushes, such as Boscia spp. (White southeast This covers 4.5 percent of Burundi. 1983). South of Nairobi, cultivation and pastoralism The ratherrandom distribution of this vegetation in have completely removed the tree cover over exten- Tanzania makes accurate description difficult. It is sive areas. probably true that, inmany cases, the species structure of small patches of Open Woodland is similar to that Class 52-East African Low Woody Biomass of the surrounding vegetation, the major difference Mosaic being in density. In central Burundi, this class is probably similar to the savanna already described under The East African Low Woody Biomass Mosaic covers MoistAcacia-Commiphora Bushland andthicket (Class much of the northeastem quarter of Uganda, extend- 43).fheextensiveareainnorthwestemBurundiincludes ing eastward over the highlands of southwestem expanse of "Wooded Savanna," dominated by Acacia Kenya. It is of considerable importance in. Uganda, albida and A. gernrdii This region also includes 'Tatsicoverng 13.7percent of the country, or 32,671 square tional Rain Forest," containing Albfzia spp., Hyplaene kilometers. In Kenya, it occurs largely in inaccessible benguelns, and Nemtonia budanuan (M'Hirit highlands and covers 51,905 square kilometers, or 8.8 Growing stock of Class 61 is 68.9 million tonnes and percent of the country. The class is associated with a sustainable yield is 1.0 million tomes per year. 6-month dry season. In Uganda and Kenya, much of the mosaic is likely Class 65-Moist Sudanian Woodland to consist of "Combretum Small Tree Savanna," dominated by C. binderanum, C gihasalense, C moalk, and C Moist Sudanian woodland in the East African region zeyheri, among others. In Uganda, this "Combretum is restricted to northern Uganda, where it forms the Small Tree Savanna" is combined with occurrences of eastern edge of a land cover belt extending as far west Acacia spp. and Ahlbizia zygia and the "Terminalia Sa- as Mali It is classified by White (1983) as "Sudanian vanna" in the north of Uganda. This latter community Isoberlinia and Related Woodlands." This land cover is dorninatedbyt.glaucescensalongwithalbiziazygia, class covers 15,334 square kllometers (6.4 percent) of Combretummolle, and other deciduous species; itprob- Uganda. Climatically, the area has a dry season of ably is derived from denser woodland. about 6 months and annual precipitation of about In the highlands of southwestern Kenya, it is 1,000 millimeters. likely that the natural vegetation in the mosaic con- The vegetation may be described as Open Woodsists of evergreen and semi-evergreen bushland. landsavanna.itseemstobeadegradedmiombowood- This vegetation type occurs on drier mountain land, although it does not possess either of the two slopes throughout East Africa, often as a transition genera most characteristic of miombo, Brachystegi and between "Montane Forest" (Class 84) and Moist Acia- lulbernardia. There occurs, however, a single species in Commiphora Bushland and Thicket (Class 43). Al- each of the other two commonmiombo genera,monotes though the species structure is rather variable, a num- and Uapaca. This class also is shorter than true miombo ber nearly always are present, including Acokantherm woodland,rarelyexceeding15metersinheightatthe spp., Carissa edulis, Dodonaea viscosa, Euclea spp., Olea southern edge of the mnain Isoberlinia belt, including africana, Sanseuieria spp., Tarchonanthus camphoratus, Uganda, Isoberlinia doka becomes more important, aland Teclea spp., as well as succulents such as Aloe spp. though it is scattered. and Euphorbia spp. Isoberlinia spp. have the ability to regrow from un- Smallholder cultivation with some pastoralism are derground suckers and thus are able to withstand fire. the dominant land uses, although the land is of poor An understory of climbers and shrubs maybe present, quality and little anthropogenic pressure threatens the with a herbaceous carpet of perennial grasses includbiomass. Regiornal growing stock is 181 million tonnes ing Andropogon spp., Eragrostis spp., Pennisetum spp., and sustainable yield is 53 million tonnes a year. and Scdizachyrium spp.

138 EasI Afnic' 125 To a considerable extent, the form of Class 65 has Figuire NDVI Profile, Seasonal Miomnbo Woodbeen determined by human activity such as rotational land (Class 66) bush fallowing (which entails clearing and buming), sedentary smallholder cultivation, and grazing and. browsing by domestic animals. Because population densities are low, fuetwood is in.good supply. Growing stock is 48 miluion tonnes and sustainable yield is O-4 just less than 1 million tonnes a year. Class 66-Seasonal Miombo Woodland Seasonal Miombo Woodland is very extensive in the. west of Tanzania, where it occurs from Kagera in the north to Lake Malawi in the south, covering much of 0 the regions of Shinyanga, Kigoma, Tabora, Rukwa; Jan Fe Mar Apr May Jun 3,I A 15 Sep OC Nov Dee Mbeya, and Iringa. Large areas of Seasonal Miombo Woodland occur along escarpments flanking the Lake where mean annual rainfallexceeds 1,000 millimeters, Tanganyika trough, with smaller fragments extending the Seasonal Miombo Woodland tends to be fairly wet north along the border with Rwanda and Burundi, and therefore floristically richer than drier miombo and into the extreme southwest of Uganda. In Burundi variants. Evergreen species are more common and itself, this class accounts for much of the northeastemr strips of riparian forest may develop around permaand southern regions. nent streams. -Apartfromthenorthern ipofzar, themaiig The woodland canopy usually is between 15 and 20 areas of Seasonal Miombo Woodland are scattered meters in height, with dominant species including through Tanzania, especially along the border with Mo- Isoberliniia angolensis, L panicuzata, and Tenninalia mollis. zambique, east of Lake Victona, in the Central High- AlsocommonareBrwchystegiaspp.,althoughtheiroccurlands, and in the eastern districts of lindi and Pwani. rence varies with local conditions and mean annual Scattered outliers also exist witiin the Wet Miombo precipitation. Much of the Seasonal Miombo Woodland Woodland (Class 67) of southeastern Tanzania. along the border with Mozambique also is rather wet, The figures for area and percentage cover of Sea- andinadditiontothespeciesjustmentioned,julbemardia sonalmiombo Woodland are: Tanzania, 336,565 square magnistiulata is comunon. Similarly, the former species kilometers (36.4 percent); Burundi, 10,855 square kilo- also appear with Isaberlinia tomentosa where Seasonal meters(40.6percent);uganda,5,63bsquarekilometers Miombo Woodland coincides with the greater rainfall (2.4 percent); and Rwanda, 2,10B square kilometers areas of the eastern Plateau. (10.9 percent). Although slightly open, the woodlands just de- Values of NDVI reach 0.5 in the wet season and show scribed generally lack a shrub layer, having instead a the marked seasonality characteristic of these wood- ground layer of the grasses Andropogon spp. and Panilands (Millingtun and others 1989), with dry-season cum maximum, and saplings of the main tree species. minima of about 0.2 (figure 12-2). It is interesting to In Burundi, the principal species in this class are note the close similarity of the NDVI temporal profiles Brachystegqia spp. and Maesopsis spp., reflecting the for Seasonal Miombo Woodland in Tanzania and Zaire transition toward forest vegetation. (figure 11-3). The dry season in this class lasts about 6 The largest block of Seasonal Miombo Woodland months. extends south from Lake Victoria to just north of Lake As White (1983) points out, "where climate changes Malawi It is drier than the Seasonal Mi,owbo Woodrapidly, as on the escarpments flanking the Lake Tan- land on the western border of Tanzania and is siriganyika and Lake Malawi troughs... ithas not always larly dominated by Brachystegia spp. (notably B. boehmii been possible to map wetter and drier miombo sepa- and B. spicifbnnis) and Julbernardia globiflora. However, rately." Much of this vegetation class is distributed on some species associated with dier vegetation also are the steeply sloping shores of Lake Tanganyika, along present,forexample,acacia breuispiaz,acalyhafruticosa, the coast of Zanzibar, andas scattered patches in other Euclea divinorum, Grewia bicolor, Lantana camami, Orblocks of fairly uniform vegetation, suggesting that it mocarpum trichocarpum, and Terminalia m&llis. In genoften is a product of local conditions. Nonetheless, in eral, however, the woodlands are floristically poor, mnany cases it is possible to suggest a probable conpo- with the canopy rarely atining 15 meters. In espesition for this vegetation class. cially poor conditions, the canopy may be as low as 3 Along Lake Tanganyika and the border regions of meters; in such cases, it often is-dominated by Monotes Rwanda, Burundi, and southwestern Uganda, an area spp. and Uapaca kirkiana.

139 126 l2sliintling Woody Biornass in Snb-Salrnran Africa Regionally, this is one of the most significantclasses. shifting agriculture, especially near the south of Lake Growing stock is 1969 million tonnes and sustainable Tanganyika. Here a mosaic of woodland and cultivayield is 31.6 million tonnes a year. tion plots exists, where woodland is cleared and Human population density throughout Class 66 is burned prior to a cultivation cycle of 2 to 4 years. In a generally low, partly because tsetse limits the keeping study of the regrowth of abandoned cultivation plots of cattle. Bush fallowing entails some renewal of bio- in Zambia, Stomgaard (1985, 1986) identified three mass, but the intensity of land use is low and vegeta- successional stages in the regrowth of cleared miombo tion is substantially intact. Few problems exist with woodland: the fuelwood supply. 1. In the first year after cldeoo ice, crops still dominate, although plots are be,.ng to be invaded by Class 67-Wet Miombo Woodland shrubs especially Euphorta tincalli and Smilax kwaus. 2. In the2 to 6 years afterclearance,furtherinvasion Wet Miombo Woodland dominates the large block of by shrub and grassspecies occurs. The woodyvegetawoodland in. the southeast of Tanzania, which extends tion at this time is dominated by Euphorbia tirucalli and over a roughly triangular area from Tanga District in Smilaxlkaussiana. By the end of this phase, the ground the north to Lake Malawiand the Mozambique border cover is dominated by grasses, notably Rynchelytm in the south. Smaller patches are found in the western repens. half of the country, especially along Lake Tanganyika 3. From 6 to 25 years after clearance, the reinvasion and the borders with Rwanda and Burundi, and to the of woody species is sufficient to form a canopy woodnorth and east of Lake Rukwa. In all Wet Miombo land, although it usually lacks a shrub layer. Woodland cova..s 11.6 percent of Tanzania, an area of However, the original Bracdystegia-Julbernardia 106,919 square kilometers. woodland is replaced by fire-resistant and fire-toler- Elsewhere, significant outliers occur in the north- ant trees of the "Combretum Savanna," most notably east and southwest of Rwanda (2,319 square Idlome- Combretum mechowianum, Diplorhynchzus condyloters, or 12 percent), the southwestern corner of carpon, and Syzygium guineense. The tree canopy en- Uganda (10,644 square dilometers, or 4.5 percent), and ables Uapaca spp. to invade, and the main components to a much lesser extent in westem Burundi (1,844 of the Bramhystegia-ulbemnardia woodland may reapsquare kilometers, or 6.9 percent). In Tanzania, Wet pear toward the end of this stage Generally, however, Miombo Woodland is the equivalent at lower altitude the increasing pressure on land has shortened recovof Seasonal Mkiombo Woodland, with extensive areas ery time of abandoned plots, so the cultivation savanna below about 500 meters. is spreading at the expense of zionbo woodlands. Large areas of southeastem Tanzania covered by Thernemaining areas of WetMiombo Woodland have thisclasswerepreviouslylabeled"drymiombowood- a varied composition. Along the Tanzanua-Burundi land" (White 1983). However, the evergreen nature of border, this class appears in a fairly high rainfall area the entire vegetation community is more apparent at an altitude of more than 1,200 meters, indicating a here, and some areas appear to have virtually no moist type of montane forest vegetation. In this case, period without growth The woodland is dominated the dominant genera include Aningeria, Cassipourea, by Bnchystegia boehmi, B. spicifonnisjulbernardiaglobi- Chnrsophylum, Macaranga, Neoboutonia, Porinari, Polyflom, and J. magnistipulata, formning a canopy that rdrely scias, andtabernaemontana. Shrub andherb layers tend exceeds 15 meters in height. to be fairly discrete and dominance among the tree In many eastern districts, including areas of tle species is low. coastal lowlands, the Wet Miombo Woodland is In the moister climate of Rwanda, Wet Miombo probably derived from the 'Dry Lowland Evergreen Woodland often consists of dense stands of "Acacia Forest" that once may have covered wide expanses Savanna." The dominant species are A. campylacantha, of the eastern plateau (Polhill 1968). Hence, Wet A webweladoides, A. syal spp. multijuga, A. senegal, and Miombo Woodland contains a number of tree species A. sieberana, often with an understory of A seyal In the from this older vegetation community, notably southwest of Rwanda, extending down the westem Drdbergia spp., Ostryodernis spp., Pleurostylia spp., side of Burundi, this class also is represented by Sclerocarya spp., and Tamarindus spp. In the extreme "Acacia Savarna," but the dominant species here are southeast, the Brachystegia-Julbernardia woodlands A. albida and A gerardii On the eastern edge of this are replaced by structurally similar woodlands zone, toward the wetter highlands of Burundi, the dominated by other leguminous tree species, espe- vegetation may take on the character of Open Woodcially Dalbergia spp., Lonchocarpus spp., and Millettia land (Class 61) rather than savanna, with Brachystegia spp. - spp. becoming more common (MHiit 1986). Some areas of Wet Miombo Woodland in the south- Growing stock for the region is 1,555 million tonnes west of Tanzania have been affected by clearance for and sustainable yield is 16.2 million tonnes a year.

140 East Aftrrca 127 Land use in the area of Wet Miombo Woodland There are fragments along the northem border of varies from sedentary smallholder cultivation in more Burundi and larger areas in north and central Rwanda. densely populated areas to extensive rotational fal- The class is important in both countries, covering 17.4 low. In general, few problems of overuse occur. percent (3,372 square kilometers) uf Rwanda and 7.7 percent (2,055 square kilometers) of Burundi. Class 71-Evergreen Woodland Mosaic The greater productivity of this class is confirmed by NDVI curves. An example from west Tanzania Despite its considerable extent, the Evergreen Wood- shows values consistently above 0.3, with no lengthy land Mosaic is difficult to describe due to its scattered dry season (figure 12-3). distribution. The area and percentage cover- in each In the Lake Victoria Basin, where virtually no dry country are: Tanzania, 23,291 square kilometers (2.5 season occurs, the mosaic is made up of wooded grasspercent); Kenya, 9,485 square kilometers (1.6 percent); land with relics of the original peripheral "Guineo- Uganda, 1,423 square kilometers (0.6 percent); Congolian Forest" A description of this vegetation is Rwanda, 843 square kilometers (4.4 percent); and presented under the Forest classes (82,85,86, and 87). Burundi 790 square kilometers (3 percent). The-area to the north and west of Lake Victoria is the The carpet areas of Evergreen Wo,odland Mosaic largest area of anthropogenic savanna in the region, occur on the coast and up to about 500 meters altitude stretching from the humid areas near the Kakamega in northeastern Tanzania, scattered within larger Forest in Kenya around the lake through southem blocks of vegetation, especially the Miombo Wood- Uganda, and into the northwest of Tanzania. lands (Classes 66 and 67). This mosaic frequently ex- Near Dares Salaam, the mosaic is similar to the East ists in mountainous areas and near lakes. African Low Woody Biomass Mosaic (Class 52) de- The word "evergreen" in the class name needs tobe scribed for the same area. However, both this area and qualified because the tree species in this vegetation, those on Zanzibar probably have suffered more degmainly common miombo species such as Brachystegia radation due to ovenise than the high woody biomass spp., Isoberlinia spp., and Julbemardia spp., are deddu- mosaics such as those typified by this class. Conous. The evergreen nature of the woodland mosaic versdly, the woodland patches within the High appears to result from an evergreen understory that is Woody Biomass Mosaic in Tanga District probably recorded by the AVwRR sensor whether or not the can- consist of Wet Miombo Woodland (Class 67), the pninopytree species are inleaf. This suggests smallpatches cipal species being Brachystegia bohmii, B. spiciformis, of vegetation with a tree flora similar to the surround- Julbernardia globiflora, and J. magnistipulata, and possiing woodland, but with a lush understory resualting bly older forest remnants (see Class 67). from favorable local conditions. These conditions un- In Rwanda, the mosaic area in the north includes doubtedly are related to moisture availability, as indi- "Montane Forest" and "Wooded Savanna with Forest cated by the more humid climate of lakeside and Regrowth" (AID 1987). The "Montane Forests" are mountainous locations. characterized by Podocarpus usambarensis and Syzyg- Growingstockinthisclassisl&3miUiontonnesand ium paruifolium, among other species. The open areas sustainaole yield is 680,000 tonnes per-year. Popula- of the mosaic consist of "Wooded Combretum elaeagtion density is relatively sparse and no senous short- nifoliurn Savanna," includingacacia spp. In southeastage of fuelwood exists at present. ern Rwanda, and across the border into Burundi, this Class 72-Cultivation Mosaic and Forest/Woodland This land cover class occurs between sea level and 2,000 meters altitude in all five countries of the East 0-6- African region. In Kenya and Uganda, it is restrcted to the Lake Victoria Basin, where much of the original -- Figure NoVI Profiles, High Woody Biomass Mosaic (Classes 72 and 73) rain forest hasbeen deared for cultivation. The mosaic OAcovers 2.9 percent (6,850 square kilometers) of - Uganda,butonlyO.3percent(2O02squarekilometers) a 0.3- of Kenya. Z In Tanzaia, blocks exist around Lake Victoria, es- 02 peially to the west, and on the coast around Dar es o - 72 Western Tanzmia SalaanL The area covered is 9,327 square kilometers, 73 Was Ug:mda or 1 percent of the cou,ntry. The mosaic also occurs on 0 i I I I the eastern side of Zanzibar. Jan Feb Mar Apr May Jun JUl Aug Sep Oct Nov Dec

141 128 Estimatinig Woody Biomass ifn Stb-Saulwnmi Africa class consists largely of "Bradcystegia Wooded Savanna" western Uganda (figure 12-3). These relatively large and areas of "Gallery Forest." In Buyenzi and values indicate both the ability of the remaining tree Mugamba districts in northem Burundi, this class con- species to "green up" in response to the wet-season tains "Ombrophilous Mountain Forest," with Albizia conditions and the growth of crops within this mosaic. gummifera, Prunus africana, Polysciaspp., and Sympi1on - Growing stock is 101 million tonnes and sustainnia spp. The associated altimontane prairies are dom- able yield is 1.1 million tonnes per year. Fuelwood mated by Agrostis spp. and Erica spp. supplies will become more difficult to obtain as This class is to a large extent the product of defores- forest is removed for agriculture. tation for smallholder settlement, either as rotational fallow, or increasingly for permanent settlement As Class 74-Guinean Woodland population increases, it is likely that the woody blomass in this class will be further reduced. In the East African region, Guinean Woodland is re- Crowing stock for Class 72 is 40 million tonnes and etricted to Uganda, where it occupies much of the sustainable yield is 448,000 tonnes per year. The fuel- northwestern sector of the country. It is similar to the wood supply is threatened by the removal of wood- "Comnbretum Small-Tree Savanna" that grows extenland as agriculture is extendet. sively over East Africa. This is characterized by a number of broad-leaved Combretinm species, notably Class 73-Cultivation and Forest Regrowth C. binderanum, C ghiasalenlse, C. miolle, and C. zeyheri, In Mosaic Uganda, however, the Combretum savanna combines with Acacia spp. and Albizia zygia. In wetter areas of Cultivation and Forest Regrowth Mosaic occurs ex- Uganda, the vegetation also contains Terminalia mollis tensively around the north of the Lake Victoria and T. glaucescens, replaced by the- smaller-leaved T. Basin, from central and western Uganda to south- brownii in drier regions (Trapnell and Langdalewestern Kenya. It covers 18 percent (43,105 square Brown 1972). In northem Uganda, the Coombretutm spekilometers) of Uganda and 2.9 percent (16,915 cies are replaced by Butyrospennuim paradoxtrn ss. square kilometers) of Kenya. Regional growing niloticum. stock is 101 million tonnes and sustainable yield is Generally, the trees of the savanna are rather scat- 1.1 million tonnes a year. tered, reaching about 5 to 10 meters in height. They Some uncultivated area of this mosaic includes grow in a sea of tall perennial grasses of I to 2 meters, Guinean Woodland (Class 74), described in the next often dominated by Hyparrhenia spp. Guinean Woodsection. However, this land cover class is most exten- land covers 7.9 percent of Uganda, an area of 18,918 sive on the northern fringe of the degraded "Periph- square kilometers. eral Guineo-Congolian Humid Tropical Forests" Growing stock in Class 74 is 35 million tonnes and north of Lake Victoria Therefore, small areas of sec- sustainable yield is 394,000 tonnes per year. Because ondary forestregrowthmightbe expected onformerly the population is relatively low, no fuelwood shortcultivated land. As an example of forest regrowth in age exists yet the drier and floristically poorer margins of the Guineo-Congolian rain forest belt, the following de- Class 82-Evergreen Forest scription is adapted from White's account (1983) of the invasion of secondary wooded grassland by forest Evergreen Forest is scattered throughout the region species at Olokemeji in Nigeria. and, with the exception of some coastal areas of Tan- After 6 years without fire, the site had been invaded zania, occurs in highland areas between 1,000 and by Antiaris toxicara, Ceiba pentandra, Celtis brownii, 2,000 meters. Generally, it occurs along the Lake Tan- Diospyros mespil4bnnis, Hildegardia baterh Holarrhenaflo- ganyika escarpment in the highlands of north Tanzaribunda, MAlacanthalnifolia, Manilkara obovata, and nia and south Kenya, and in the highlands and coastal Zantl}oxylum xanthoxyloides. After another 25 years, a areas of southeem and eastern Tanzania between Lake canopy 8 to 11 meters tall had formed, dominated by Malawi, the Mozambican border, and Tanga District. forest species such as Afzelia africana, Diospyros The area and percentage cover for each country are: mespilmbmis, Hildegardia barter, and Manilkara obovata. Tanzania, 109,97T square kilometers (119 percent); Even so, more than a dozen savanna species still Kenya, 6,580 square kilometers (1.2 percent); Burundi, persisted. However, the pressure on agricultural land 5,111 square kilometers (19.1 percent); Uganda, 3,056 in Uganda makes it unlikely that succession would square kilometers (13 percent); and Rwanda, 2,846 be allowed to continue undisturbed for such a long square kilometers (14.7 percent). This is one of the period. most significant classes for woody biomass. The rdmvcurve shows values in excess of 0.5 for the Itisdifficultto describeatypicalvegetation for such wet season but reduces to 0.2 in January for a site in a wide-ranging group of montane forest areas, given

142 East Africa 129 the diversity of Afromontane vegetation. It is de- Andropogon spp., Hyparrhenia spp., Pennisetuin spp., scribed for a number of these areas by Lucas (1968), and Setaria spp. In the "Heathland" and "Afro-alpine" Osmaston (1968), and Polhill (1968). The following zones, the most common genera in secondary.grassgenerally describes Afromontane vegetation types in land are.agrostis, Descl)ampsia, Festuca, and Pon. the region. Along the coast of Tanzania, the Evergreen Forest Afromontane rain forests grow in a fairly wide class is probably derived from the "Dry Lowland Evrange of local conditions, but rainfall generally ex- ergreen Forest" that formerly covered much of the ceeds 1,250 millimeters-a year, and often is more than coastal plateau. The vegetation consists of Brchy- 2,500 millimeters. Most forest lies between 1,200 and stegia-julbernardia miombo woodland containing a 2,500 meters altitude, although this too is variable. number of forest species from the old lowland forest. Typical tree species include Aningeria adolfi-friedericii, It is more fully described under Wet Miombo Wood- Ocotea usambarensis, Olea capensis, Parinari excelsa, land (Cass 67). Podocarpus latifolius, Prs afiraa, Syzygium guineense The likelihood that fast-disappearing forest will be ss. aftomontanumn, and TabrnaeronUtnajolrnstonii. protected by law makes this class an unprom-lising "Afromontane Bamboo" (Arundinaria alpina) is source of fuel In fact, much of the Evergreen Forest widespread in the highlands of East Africa. Occurring exists in protected areas, including the Mount Kenya mostly between 2,380 and 3,000 meters, it is extensive and Nairobi national parks in Kenya, the Ruwenzori on the Aberdare Mountains and Mount Kenya, and on National Park in Uganda, and the Kilimanjaro Forest the gentler slopes of the Ruwenzori Mountains. It is Reserve in Tanzania. Those areas lacking protection almost absent on Mount Kilimanjaro. Tree species are rapidly being cleared forsmallholdings. scattered among the bamboo include Dombeya Growing stock in Class 82 is 767 million tonnes and goelzenii,hageniaabysnica,lexmitfis,uniperusprocera, sustainable yield is 64 million tonnes per year. The Nuii congesta, Podocarpus latifolius, Prunus africana, fuelwood supply is likely to become a problem in and Tabernaemontanajohnstonii. areas being cleared for agriculture. The phenology for this vegetationdclass shows consistentlyhighvaluesof mvi (aboveo.4) andno marked Class 83-Coastal and Gallery Forest seasonality (figure 124). Another feature of most high mountains, and the Coastal and Gallery Forest is most extensive along the summits of many smaller mountains, is "Evergreen Kenyan coast, although significant areas are dotted Bushland and Thicket"-particular.y near the coast along the northem coast of Tanzania. Elsewhere, and large lakes (White 1983). Unburned thicket often smaller areas are on Mount Kilimanjaro and in the is between 3 and 13 meters tall, and typical genera Cater Highlands and west Usambara Mountains of include Blaeria, Erica, Philippia, and Vaccinium. The northern Tanzania. This class accounts for 1.8 percent most extensive vegetation in the mountains.of East of Kenya (10,539 square kilometers) and 0.6 percent of Africa is, however, grassland; the vast majority of it is Tanzania (5,586 square kilometers). secondary, resulting from destructive human activity. The southernmost block of this forest lies between Secondary grassland in the montane forest belt typi- Kisii and Kilwa Kivinje, in the Tanzanian district of cally includes Exothe abyssinica, Loudetia simplex, The- Pwani. Here, the forest is made up of extensive stands meda friadra, and anumber of other species including of mangrove. The main genera are Avicennia spp. and Rhizophonr spp.., although distinct mangrove zones have developed along the East African coast, and Figure NDVI Profile, Evergreen Forest other. species, especially Bnouiena gnarrhiza and (Class 82) Heritier littoralis, also are presenl Along the coast of Kenya and northern Tanzania, as 0.6- far south as Dar es Salaam, this class is a mosaic of "Moist Forest" with "Transitional Evergreen Bushland and Scrub Forest.' This is a diverse forest vege tation, its 15 to 20-meter canopy composed mainly of Afzelia quanzensis and Erythna sacuxii. Emergent 3 03o- species, some reaching 30 to 35 meters, include Albizia z adiantholia, Balanites wilsoniman, Combretum scizuman niijulbernardia magnistipulata, Lannea spp., and AUmlkara sansibarensis. The coastal fringe of the mosaic, 82 Ugmda especially along the northern Kenyan coast, consists 0.- * ofmangroveswampdominabedbyavicenniaspp.and Jan Feb Mar Apr May Jun Jl Aug S Oept Nov Dec Rhizphora spp.

143 130 Estimating Woody Biamanss in Sub-Snlhran Arica In the remaining mountainous regions, the gallery Class 84-Montane Forest; Class 86-. forest component of this class covers small areas at high altitudes. The greater rainfall of Mount Kiliman- Humid Tropical Swamp Forest; Class 87- Ombrophilous Humid Tropical Forest jaro and the Crater Highlands gives rise to "Upland Rain Forest," dominated by Aningeria spp., Parinari Much of this region's forested area is concentrated spp., and Ocotea spp., together with Cassipourea spp., around the northem and western shores of Lake Vic- Ckrysophylhum spp., Macaranga spp., Neoboutonia spp., toria, in southem Uganda. Much smaller areas exist in Podocarpus spp., Polyscias spp., and Tabemaemontmna the north of Uganda and in the highlands of Kenya spp. A similar vegetation community could be ex- and noriern Tanzania. Ombrophilous Humid Tropipected for this class on the slopes of the Usambara cal Forest (Class 87) is the most extensive of the forest Mountains in Tanga District, although here it proba- classes, covering 33,145 square kdlometers (IBA perbly merges with "Moist Lowland Forest," containing cent) of Uganda, 3,531 square kilometers (0.6 percent) genera such as Allanblackia spp., Cephalosphaerm spp., of Kenya, and 1,370 square kilometers (0.2 percent) of Isoberlinia spp., and Newtonia spp. Tanzania. Growing stock in Class 83 is million tonnes Two otherrelated classes, MontaneForest (Class84) and sustainable yield is 8.1 million tonnes per year. and Humid Tropical Swamp Forest (Class 86), are far The fuelwood supply is good generally, although smaller, with an area of less than 6,000 square kldomemangrovecuttingiscausingundesirableenviroranen- ters for the entire region. Because of the difficulty in tal effects in the coastal zone. distinguishing among these three forest types, the following descrption covers all three classes. Class 85-Mesophilous Humid Tropical Forest The area around Lake Victoria has been extensively cultivated, so the forests of south Uganda exist, for the This landcovercass occurs alongthenorthernmargin most part, in a mosaic with other vegetation and cropof Lake Victoria in an area thathas no dry season. Most land. This taes the form of a grassland about 2 meters of the class occurs in Uganda, where the area of 13,490 tall with a varying density of fire-resistant-trees. The square kilometers constitutes 5.6 percent of the colu. prncipal gasses indude Andropogon spp., Hyparrhenia try. Hereitprobablyrepresents White's"DrierPeriph- spp., and Loudetia spp. Tree species include Burkea eral Semi-Evergreen Guineo-Congolan Rain Forest" africana, Combretum collinum, Strydmos spp., and Ter- (1983).Ablockof MesophilousHumidTropicalForest minalia spp. lies in the extreme southwest of Uganda, extending This region is at the easten edge of the "Guineointo both Tanzania and Rwanda. This is an area of CongolianRainForest"zoneandisthusadriervariant highlands exceeding 1,400 meters in elevation, with a of this rain forest type, containing most of the importvegetation similar to that of the Wet Miombo Wood- ant species. The domninant trees include Albizia spp., land (Class 67) on the Burundi border-that is, domi- Aningeria altissima, Celtis spp., Chrysophyllum albidum, nated by Aningeria spp., Parinari spp., and Ocotea spp. Entandrophragma spp., and Khaya spp. At the upper limit other genera tend to dominate, The shores of Lake Victoria host extensive swamp notably Hagenia spp., Myrica spp-, N?Jia spp-, and forests. Around the mouth of the Kagera River on the Rapanea spp., especially on soil derived from volcanic westernshore,thesearedominatedbybai!aeainsignis rock (emini-i) and Podocarpusfakatus. Western Uganda also hn Kenya, this class makes up a single block of 843 contains forest dominated by Parinari excelsa. Another squarekilometers,correspondingimparttokakamega important group of forests in this region are those Forest on the Nandi escarpment east of Lake Victoria. dominated by Cdtis spp., especaly C. mildbraedii, The forest contains both lowland species (includ- with a variety of other species. Related forests in the ing Aningeria altissima, Cordia millenii, En tan- Mount Kenya area are dominated by Newtonia buchadrophragma angalense, Maesopsis enunii, and Monodora nanii In Tanzarua, forests mapped in this class include myristica) and Afromontane speces (sudh as Apodytes Allanblackta stuhlmannii and soberlinia scheffleri, with dimidiata, Macaranga kilimandscharica, Neoboutonia Newtonia spp. and Parinari spp. macrocalyx, Prunus aficana, Strombosia scheflri, and The montane forests of East Africa mapped in Class Turraea holstit). 84haveavarietyof dominantspecies,aningeriaadolfi- A number of types of land use occur in this class: friedericii, Entandrophragma excelsum, Ocotea permanent smallholdings, rotational fallow, and usambarensis, and Podocarpus mlanjinus being the mixed farming. Considerable areas of the forest have most commonu Kenya hosts some related forests that been cleared for agriculture; this threatens the long- indudealbizia spp., Polysciasspp., and Olea weluitschii term fuelwood supply. Growing stock of Class 85 is as dominants. A general description of the Afromont million tonnes and the sustainable yield is 21.8 ane vegetation in the region is in the Evergreen Forest million tomnes per year. (CIass 82) section.

144 East Africa 131 In Lake Victoria Basin, a few fragments remain of Berry, L., T. Taurus, and R. Ford East Africa the "scrub Forest" that once covered much of the Country Profiles-Somalia. Worcester, Mass.: Clark area. These relicts exist in south Uganda, notably University, Program for International Development. around Lake Edward, and in Burundi. Scrub forest Lucas, G. L "Kenya." In I. Hedberg and 0. usually is dominated by Euphorbia dawei, forming a Hedberg, eds., "Conservation of Vegetation in canopy 12 to 15 meters in height. Often associated Africa South of the Sahara." Acta Phytogeographca with this canopy is Cynometra alexandri, a canopy Suecica 54: and emergent species oi rain forests that here rarely M'Hirit, Besoins en mati&re d'education et de exceeds 10 meters in height. formationforesthereau Burundi.R6sum6nintroductoire A common characteristic of these three high- &la reunion de la commission tripartite. Rome: FAo. biomass classes is that they are threatened by exploi- Moore, J. E "Vegetation." In L. Berry, ed., Tantation for cultivation and this jeopardizes the fuel- zania in Maps. London: English Universities Press. wood supply, particularly in Uganda. The growing Morgan, W. T. W EastAfrica. London: Longman. stock of Class 87, which is the most extensive of the Osmaston, H. A "Uganda." In L. Hedberg and three classes, is 399- million tonnes. The sustainable 0. Hedber& eds., "Conservation of Vegetation in yield is 56.3 million tonnes. Africa South of the Sahara." Acta Phytogeographica Suecica 54t Land Cover Class Tables Polhill, Ri "'Tanzania." In I. Hedberg and 0. Hedberg, eds., "Conservation of Vegetation in Af- Tables 12-1 through 12-5,beginning on page 132, pres- ricasouth of the SahaxamAcfaPhytogeographica Sueecs ent summaries for each land cover class of the area, 54: showing stock and sustainable yield for the Stomgaard, Peter "Biomass, Growth and Bum- EastAfricannationsofBurundLKenya,Rwanda,Tan- ing of Woodland in a Shifting Cultivation Area of zania, and Uganda. South Central Africa." Forest Ecology and Management References Stomgaard, Peter "Early Secondary Succession in Abandoned Shifting Cultivator's Plots in the Every effort has been made to facilitate access to the Miombo of South Central Africa." Biotropica 18(2): documents listed here. Some- documents, however, lack full bibliographic information because it was Trapnell,C. G., andl Langdale-Brown "Natural unavailable; also, some documents are of limited Vegetation.' In W. T. W.Morgan, ed., East Africa, Its circulation. PeopleS and Resources. Rev. ed. Nairobi: Oxford University Press. Am (U.S. Agency for International Development) White, F "The Vegetation of Afric" Natural "Draft Environmental Profile on Rwanda." AD Of- Resources Research Senes 20. Paris: UNEscO/AETAT/ fice of Forestry, Enviromnent and Natural Re- unso (United Nations EducationaL Scientific and sources, RSSA A/TOA Prepared in cooperation CultulOrgae zation/associationpourl'etudetaxowith U.S. Man and Biosphere (MAB) Program of the nomique de la Flore de l'afrique Tropicale/United Department of State. Nations Sudano-Sahelian Office).

145 132 Estimiting Woody Biomrass in Slub-Satlaran Africn Table 12-1 Land Cover Classes-Burundi (East Afdca Region) Area Gro'wing stock Sustainableyield Thousand Thloutsand tonnes Lantd cover clas ksn Percent tonines Percent peryear Percnt , , , , , , , , , , , , , , , , , , , , , , , Lakes 1, O.0 Total 26, , , DO (Percentage of region) (1.49) (1.93) (1.72) Note: In the following tables, details may not add to totals because of rounding. Sourcm Authors' calculations from data bases derived from land cover classification and tabwe 4-L Table Land Cover Classes-Kenya (East Africa Region) Area Growing stock Sustainableyield Thousand Thousand tonnes Land cover class km 2 Percent tonnes Percent per year Percent 0 14, , , , , , , , , , , , , , , A , , Z29 1, , ,85B , , , , , , , , , , , , , , , , , , , , (Tabl coninues on thefollowing page.)

146 East Africa 133.Table 12-2 (continwed) Area Grawing stock Sustinmableyield T7housand Thousand tonnes Land cover class km 2 Perment totnes Percent peryear Percent , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Lakes 15, O.oW Total 589, , , (Percentage of region) (32.78) (14.50) (13.69) Source? Authos' calculations from data base derived from land er classifation and able 4-1. Table Land Cover Classes-Rwanda (East Africa Region) Area Growing stock Suistafivubleyiedd Thousand Thfousand tonnes Lanid cover clas km Percent tonines Percent per year Percenzt or , , , , , , , , , , , , , , , , , , , , Lakes 1, Total 19, , , (Percentage of region) (1.0) (1.13) (0.93) Sourcc Authoraecalculations from dababasesderived from lndcover classification and table 4-1.

147 134 Estimating Woody Biomass in Sub-Sahqran Africa Table Land Cover Classes-Tanzania (East Africa Region) Area Growing stock Sustainableyield Land cover class knt Percent Thousand -7tousand tonnes tonnes Percent per year Percent , , , , , A , , , , , , , , , , , , , , , , , , , , , , ,565 36A3 1,872, , , ,272, , , ,209, , , , , , , , , , , , , , , , , , , , Lakes 41, Total 923, ,360, OD 111, (Percentage of region) (51.36) (67.30) (48.80) -Sour= Authors' calulations from data bases derived from land cover classication and table 4-1.

148 East Africa 135 Table Land Cover Classes-Uganda (East Africa Region) Area CroWingstock Suistainable yield Thouisand.. 7Thousand tonnes Lanid cover class k,n Percent tunnes Percent per year Percent 23 Z, , , , , , , , , , ? , , , A , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Lakes 34, Total 239, , , (Percentage of region) (13.29).(15.15) (34.86) SDurct Authors'calculations from data bases derived from lnd cover dassificationtand table4-1.

149 Southern Africa Roger Bevan 13 This chapter presents a detailed description of the It peaks at 0.06 in northern Namibia and souther mostimportantlandcoverclasses ithisregion.help- Angola in March and July, but has a slightly greater ful figures in other chapters include figure 3-1 (cloud peak of 0.08 in September in southem Namibia. Precover); figures 3-2, 3-3, and 3-4 (NDvi sumnmary land cipitation is mostly less than 100 millimeters in the cover profiles); figure 3-5 (regional summary map of Namib Desert at the core of this class, but may exceed land cover classes); figures 7-1 and 7-2 (continental 250 millimeters in the wettest parts. maps of growing stock and sustainable yield); and the The coastal belt is largely frost-free, but there may "Regional Land Cover Class Map of Southern Africa" be only 5 or 6 frost-free months in southem Namibia at the end of this volume. and interior Cape Province. The slightly wetter areas. Helpful tables in other chapters include table 3-2 of this class largely coincide with White's "Bushy Kar- (land cover classes); table 4-1 (data and sources for roo-namib Shrubland" and "semidesert" vegetation growing stock and sustainable yield); and table 6-6 (1983) and the drierparts of the "Dwarf Karroo Shrub- (southern Africa estimated woody biomass by sum- land," together with areas of the 'Kalahari/Karroomary class). Namiib Transition." These classes all represent transitions at the desert edge to areas of greater biomass. Class 0-Desert The Namib Desert occupies a coastal peneplain, an extensive area of bare rock often covered with mobile This class occupies approximately 6 percent of the sand. In southwestern Angola, the eastern part of the southern Africa region, including 20,446 square kilo- Desert of Moc,amedes may rarely and locally support metersinangola,29,457squarekilometersinbotswa- scattered low bushes and dwarf trees in depressions na, 221,478 square kilometers in Namibia, and 79,148 that receive water. In Namibia, true desert areas supsquare kilometers in South Africa. It comprises most port very little woody biomass However, on the desof the Namnib Desert and its immediate environs, a ert fringes in the Namaqualand foothills, in the lower large triangular area stretching along the Atlantic Orange River Valley on the Namibian-South African coast from near Namibe in southern Angola to south border, and on higher ground to the east, this class of Alexander Bay in South Africa. It stretches inland may contain shrubland with widely scattered small about 100 kdlometers in southern Angola, but widens bushy trees and larger shrubs. Succulents are abunsouthward to about 500 kilometers in Cape Province. dant, especially in the south. It is broken only by the slightly better vegetated Welwitschia bainesii is widespread from central Namaland Mountains in southern Namnbia. Outliers Namib northward, especially in the transition zone occur in the Great Karroo in Cape Province of South between the inner and outer Naminb. It is particularly Africa, the Etosha Salt Pans in north Namibia, the abundant in the shade of riparian bushland domi- Makgadikgadi Pan in northeastern Botswana, and in nated by Colophospermu mopane and Terminaila pndegraded parts of the Kalahari Desert of south-central nioides. On the stony flats between watercourses, and southwestern Botswana. Welwitschia bainesiis the only woody plant, associated The phenology of this class shows a consistently low with grasses. On the margins of larger rivers, taller NDvi, mostly between 0.02 and 0.06 for the entire year. shrubs and small trees such asacacia albida, A erioloba, 136

150 -. Sailhcrn Africa 137 Euclea psaeudebenus, Ficts sycomorus, Ritus lancea, and Although this latter area was previously classified Salvadora persica occur locally, as "Woodland-Bushland" (irc Foundation 1987; Mil- In the outliers of this class in Cape Province, a lington and others 1989), it can have a more-or-less "Dwarf Karroo" vegetation dominates, including continuous grass sward and has a sufficiently similar dwarf shrubs, mostly Compositae, and larger shrubs, -Novi phenology to be classed with the veld grasslands such as Rlziguz0Um2 trichotomum. Bushes and trees are of South Africa. In semiarid areas, classes may vary almost entirely absent and succulents are less promi- from year to year because of fluctuations in rainfall. nentthangrasses. Soilsoftenareslightlysaline,so that Although somnewhat ecologically dissimilar, the halophytes such as Salsola tuberculata are widespread. various regions of this class all have moderately high In the very arid Karroo, large, brackish flats or vloere productivity in the wet season, with NDVI values exist, sometimes covered by Salsola aphylla and other greater than 0.35, reaching peaks of 0.43 in December halophytes, but often these areas are bare. and January. The values are sustained at greater than The Etosha Salt Pans of Namibia and the Makga from February to April, but decline to approxidikgadi Pan contain no woody vegetation. Outliers of mately 0.2 from June to September. A quick transition this class in the Kalahari may contain limited woody from greatersummer NOVI values to the smaller values biomass stocks. Southwestem Botswana's rolling of the winter months largely reflects seasonality in sand dune areas contain a few widely scattered trees precipitation and evapotranspiration. such as BosciaalbitruncandAcaciamelltferm ss.detinens, The highveld grasslandsof South Africa are a typical occurring mainly on dune crests, with shrubs such as continental interiorhigh plateau ranging from 1,000 to Rhigozu7n trichotomum confined to the interdune 1,800 meters altitude. The grassland vegetation usutroughs. Farther northeast, the rolling sandy country ally is considered the result of hot, wet summers and with wide plains, depressions, andpans includes trees cold, dry winters, which restrict tree growth (Taylor such as Acacia erioloba, A. nellihera (detinens), Boscia 1972). The growth of woody vegetation also is realbitrunca, Dichrostadrys cinerea, and Terminalia sericea. stricted by other factors, such as the more frequent Low shrubs indudeacacia hebedada, Bauhinia macrantha, incidence of frost, which can be expected at least 80 Boscia albitrunca, Dichrostadlys cinea, Grewia spp., and days a year for most of this class and up to 120 days in Ziziphus mucronata. the highest areas. Rainfall increases from approxi- This class is clearly of almost no value as woody mately 600 millimeters in the west to more than 800 biomass. The very limited standing stock would not millimeters in the east. The topography is a plateau, be replaced for many years if it were removed as a flat to gently rolling. It has predominantly black result of severe environmental constraints on biologi- montmorillonitic soils in some of the more northem cal production parts. In more southerly areas, soils are mainly add to neutral yellow anld grey, sandy and loamy, and over- Class 11-Veld Grassland lie sandstones (Scheepers 1975). Most authors-for example, Werger and Coetzee High Veld Grassland is typical of the interior plateau (1978) and Acocks (1975)-distinguish "true" grassareas of western Lesotho, Orange Free State, parts of lands from "false" grasslands, the latter being largely southwestern Cape Province, andsoutherntransvaal anthropogenically induced. The former include the It constitutes nearly 6 percent of southern Africa and so-called sour grasses, tall grasses with wide, coarse occurs largely in South Africa (214,364 square kilome- leaves that have less nutitive value in winter than ters). It coincides largely with White and Moll's "High "sweet" grasses. They usually occur at slightly higher Veld Grassland" (1978) and Werger and Coetzee's attitudes and form uniform-looking sward of And a- "Moist Cool Temperate Grassland" (1978). * pogon spp., Coris, Ergrosis spp., and Panicum Small outliers occur in Bophuthatswana, just north coloratum. Themeda triandra dominates the less-acid of Pretoria, and in small tongues within the coastal "mixed" and "sweet" grasses, which have short, narmountain ranges north of Port Elizabeth. Other small row leaves and often intermingle with forbs. outliers occur in the extreme southwest and northeast The "false" grasslands occur in potentially more of Angola (22,922 square kilometers) and in south- woody areas that have suffered excessive burning, central Mozambique (6,271 square kilometers). The overgrazing, or cultivation (Werger and Coetzee largest outlier of this class occurs in western and north- 1978). Overgrazing often results in replacement of western Botswana (53,380 square kilometers) and some of these species by tough and unpalatable northeastern Namibia (15,282 square kilometers). It Aristida spp. and Cidoris virgata. covers large areas of central Chanzi, Ngamiland, and Riparian woodland commonly exists along water- North East districts in Botswana and coincides with courses, although the trees usually are stunted as a the "Kaukau Veld" (White 1983) on the Namibian- result of climatic severity and exploitation. Partially Botswanaborder. wooded are the Mohokare and Sengu valleys in

151 138 Estimating Woody Biomnass in Suib-Salnamn Africa Lesotho, the Orange, Vaal, Madder, and srnaller tributary valleys in South Africa, plus small valleys in the bushes include Acacia ciolobl, A,fleckii, A. hebeclada, A. leuderitzii, A. me!lvera, A. tortilis, Boscia albitrunca, coastal ranges of Cape Province. The deep, sandy Dichrostachys cincrca, and Terininalia sericca, In the levees fringing theorangeriversupport riparien for- north, broad-leaved trees are more common, includest 6 to 10 meters tall, composed principally of Acacia ing Combrehtn collinum, Comniphiora africana, C. onkarroo, Celtis africana, and Diaspyros lycioides. Else- golensis, Oclina ptilcira, and ZfziplIus mucronata, but where, these trees have a bushlike form and may with Acacia spp. still dominant. In both areas, the trees include Rhus laucea and Zizipi/us nmucrnata, as well as are always less than 7 meters tall (usually much A. karroo, C. africana, and D. lycioides. smaller) and are quite widely spaced, with dominant In dry, low-lying, undisturbed or protected areas, grassy sward beneath them. low bushland and scrubland less than 5 meters high An outlier of this class occurs in northwestem Anoccurs. This is dominated by A. knrroo, Buddleja saligna, gola and extends into Zaire. It is described by Belgian Celtis africana, Cussonia spp., Diospyros spp., Eluretfa writers as "steppes" and probably is the result of rigida, Euclea crispa, Grewia occidentalis, Olea africana, degradation of woodland (see Chapter 11, Central Osyris spp., Rlus spp., and Ziziphus mucronata. South- Africa, Class 11). The other outlier in south-central ward from Lesotho, Aloeferox occurs on north-facing Mozambique takes the form of an "Open Shrubby slopes. Much of the southwestern Cape Province is Savanna Grassland." This was mapped by i-rc Founclassified as "High Veld/Carroo Transition" by White dation (1987) and Millington and others (1989) as Dry (1983), with overgrazing having converted much of Miomnbo Woodland but is, in fact, a degraded woodthis area to a secondary dwarf shrubland with grassses land area, consistingof scattered low trees and bushes, such as Aristida congesta, Cynodon kirsutus, Eragrostis including Uapaca spp., Monotespp., and Proteaspp. in spp., and Themeda triandra. a well-developed grass sward about 0.5 to 1.5 meters Other areas of greater woody biomass occur in the. high. higher parts of eastem Orange FreeState and adjacent The fuelwood resource in this land cover class is areas of Lesotho. These are probably transitional to small, both in growing stock and sustainable yield. Transitional Wooded Grassland (Class 24) and Open More important, it is very patchy in distribution. In Woodland (Class 61). Northwest of Kimberley, this Lesotho and South Africa, fuelwood supply is further class occurs over coarse, stony soils where trees have restricted by the land tenure pattern. Consequently. been cleared for mining and where grazing has re- the potential fuelwood supply is poor in these areas, placed "sweeter" Themeda-dominant vegetation by the which include a number of important settlements such more "sour" Aristdaominant communities. as Teyateyaneng, Maseru, Mafeteng and Qu thing in Over much of South Africa, this class coincides with Lesotho; Virginia, Bethlehem, and Bloemfontein in the good ranching country, and in more temperate areas Orange Free State; L.dock and Middelburg in southis used extensively for maize and wheat cultivation. western Cape Province; quite densely settled parts of The southern part of the so-called Maize Triangle Bophuthatswana, both north of Pretoria and west of coincides with the northern part of this land cover Lesotho; the north of Transkei near Matatiele; and the class in South Africa. Maize cultivation is extensive, north of both Ciskei and Transkei near Queenstown. with fairly small yields of 0.9 to 3.2 tonnes per hectare The pressure on limited woody biomass stock is (Christopher 1982). Wheat is grown in southeastern particularly marked near black townships and in the Orange Free State as a winter crop. Farther south, in densely settled rural areas of the homelands. Here, the "High Veld/Karroo Transition" of the southwest- pastoralism and subsistence agriculture are more em Cape Province, the land is largely devoted to common than the commercial agriculture which dompastoralism and subsistence agriculture. inates much of the heart of this land cover class. In northwestem Botswana and northeastern Namibia, this class is represented largely by the "Kalahari Class 14-Montane Grassland and Heathland Thornveld/Zambezian Broad-Leaved Transition" (White 1983) and is included as 'Plains Bushveld" Montane Grassland and Heathland covers only 15,440 class by Werger and Coetzee (1978). It is an area of. square kilometers of southern Africa, which is less coarse-textured soils on tropical plains with 500 to 700 than 1 percent of the land area. Nevertheless, this class millimeters rainfall a year- This wooded grassland is is locally significant in the Drakensberg Mountains the characteristic vegetation of the thick mantle of the and in Lesotho. True subalpine and alpine grassland Kcalahari Sands (White 1983). and heathland occur mainly in three areas in Lesotho In Botswana, the almost continuous grass sward is (Mokhotlong, Qacha's Nek, and Thaba-Tseka disless thai1rneterbighandconsistsof Anthephoraoargentea, tricts) which total 7,535 square kilometers and four A. pubescens, Eragrosfis spp., Panicum kilaharense, and small areas in adjacent South Africa and Transkei, Schmidtia spp. In the south, the principal trees and mostly in the Drakensberg range, between Kopshorn

152 -. SOIer11 AfrIca 139 and Scobell's Kop near Barclay East, which total 6,587 squareldlometers.smalloutliersalsooccurto thewest. Mozambican. border into southern Mozambique (16,388 square kilometers are in Mozambique) and above Rouxviile and Sterkstrom in eastern Cape Prov- into northern Orange Free State. ince, also at high altitude. All area In this class Is above 1,500. meters, and the majority of it is above 2,00D The other area stretches north-northwest to southsoutheast adjacent to the drier classes of land cover meters. The phenology of this class exhibits a fairly marked that border the Namib Desert, from southwestern Angols (15,967square kilometers) across centralnamibia seasonal pattern, with a moderately high NDVI of 0.35 (161,458 square kilometers), into northem Cape Proyin the warmer period of December to April, rapidly ince. Outliers occur in South Africa in the Langeberg giving way to values of about 0.19 in the cold period Mountains, Asbestos Mountains, Mount Gakarosa, from June to October. and Kaap Plateau of northem Cape Province, near The subalpine belt in Lesotho exists between 1,830 -Kimberley on the border between Orange Free State meters and 2,895 meters. Soils are very thin and stony and the climate is severe. Rainfall is variable (500 to and Cape Province, and in a few small areas in the coastal ranges of southern Cape Province, on.the 1,200 millimeters) and much of it fals as snow..tem- north-facing slopes of the Langeberg and Kougaberge peratures colder than 0 C are common in winter, but in January mean temperatures of about 15 C are comranges. Altogether this class covers 171,365 square kilometers in South Africa. Small areas totaling 25,294 mon. The frost-free period is about 185 days. square kilometers also occur in the extreme west of Subalpine vegetation is dominated by fire-controlled Th1emeda-Festuca grassland, the species of which differ Zambia. This class coincides largely with areas receiving 200 with aspect and altitude. Chrysocoma dominates the grassland over about 13 percent of this zone, and this to500 millimeters rainfall a year, although a minority of areas receive substantially more..the phenology is believed indicative of overgrazing. The scant woody varies slightly be tween eastern and western represenvegetation in the subalpine belt indicates a very small tatives, as follows. fuelwood potential. The alpine vegetation occurs in more severe condi- In eastem Botswana, the seasonality is less marked because of a low productivity in wet season. The greatdions than the subalpine grasslands and is dominated by homogeneous low, woody heathlands. The domiest NMVI values occur in April, May, and November, when levels of 0.21 to 0.25 are recorded. For the rest of nant species, Erica spp. and Helichrysum spp., are in- the year, a low level of between 0.12 and 0.18 is mainterspersed with grassland. All are evergreen woody tained (figure 13-1). plants adapted to the cold, dry climate and all show a hi northwestern Narnibia, seasonality is somewhat reduction in height with increasing altitude. Various greater,-buthe overall annual levels are siml. Peak vegetation types correspond to different enviromnen- Nrvi values of 0.33 and 0.30 are experienced in March tal situations, such as bogs, pools, stream banks, wet and dry meadows, and cliffs, but almost all are domiand April, declining to 0.12 in December and January (figure 13-1). nated by grasses, mosses, and herbs. Only the heath communities have a large propor- In eastem Botswana, this land cover class is floristically similar to the Acacia Woodland Mosaic (Class 51) tidon of woody species, but even in the best developed and the Open Woodland (Class 61) areas to the north, of these communities, which are restricted to the sum- dominated by Colophosperinumopane, with Acacia mits of the Drakensbergs, the woody species rarely attain more than 1 meter in height. Isolated patches of scrub, often only up to 2 meters in height, and domi- Figure NDVI Profiles, Transitional Wooded nated by Buddlja corrugata, Leucosidea sericea, and Grassland (Class 24) Passerina montana, exist in undisturbed areas on the lower mountain slopes (Millington and others 1989) However, no fuelwood resource exists over much of - 24 Eastem Botswania this part of the class Northwestem Namibia Class 24-Transitional Wooded Grassland 0. cz0.3-: / \: Two large areas of this class, together with a number 2 of smaller areas, cover more than 650,000 square kilo meters in southerm Africa. One area is in eastern Botswana (240,607 square kilometers), southwestern Zimbabwe (25,610 square kilometers), and northern m;, TransvaaL Outliers extend along the South African- Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

153 140 Estirmating Woody Biormss in Sub-S liarali Africa spp., Combrefinn spp., Tertninalia prunioides, and Ziziphus inated by Grewiaflava, Ormocarpum#1 triclocarpum, Rits mucronata and including shrubby Colopbospmerum mop- pyroides, and Ziziphus mucronata. ane, Grewiaflava, and Teninnalia sericea. A strong ecolog- In other areas, vegetation is restricted to thorny and ical gradient exists to the west, and once the sandier spinyshrubsscattered amonggrasslands. Woody shrubs soils developed on the Kalahari Sands are encoun-- andstuntedtreesreachabout5meters.typicalwoody tered, it grades rapidly into bushland and shrubby constituents are Bfca spp, Cormi*ora spp., Dichrobushland. Floristically and structurally, it is therefore stad:ys dinema, Grewiaafava, Lycium spp., Rhigozuin breviquite diverse, but nearly always has a well-developed spin asum, Ternminalia prunioides, and Ziziphus. grassy sward. -mucronata. The only common emergent is the low, Eastward, the class continues into southwestem shrubby Colophospennum mopane, which here rarely Zimbabwe and northem Transvaal, mostly on the exceeds 5 meters in height and often reaches only 2 flanks of the Limpopo Valley, being most extensive in meters. It occurs as scattered individuals in a shrubby Matabeleland South. It occurs along the hot, dry ground cover interspersed with grasses. Limpopo Valley and along the northern tributary val- On escarpments overlooking the river valleys is a leys of the Mwenezi, Shashe, Tuli, and Umzingwani very open savanna of small trees and bushes. The rivers (as well as in the middle of the Sabi Valley in the structure varies from dense woody stands reaching 4 east) in Zimbabwe, and the southem tributary vaueys to 8 meters in height, through a variety of shrub types, of the Mogol, Palala, Mogalakwena, and Sand valleys to the most arid form, in which trees rarely exceed 3 in the Transvaal. The class also includes a large part of meters and vegetation is very open withgrass and fem the lower Limpopo Valley in northern Venda. layers dominant. The species are related to those in the In these regions, it also contains a variety of similar lowlands,andthemainwoodyspeciesareacaciaspp., vegetation types-"arid and Dry Mountain Bush- Androstachys johnsonii, Boscia albitrunca, Colophosvelds," "Arid Spiny and Thom Plains Veld,' "Ter- permum mopane, Combretum apiculatum, Commiphora minralla Sandveld,' and low "Colophospermu mopane glandulosa, Kirkia acumizata, Plerarpus rotundfolius, Bushland" (Werger and Coetzee 1978). All the vegetzr Terminalia prunioides, and Ximenia americana. uon types are dominated by low trees and bushes, - stribution of the various species is in. :nly depensome of which are emergent trees above a well-devel- dent on soil properties. On sandy parts of the Highoped ground layer of grasses and herbs. They are veld, Terminalia sericea-dominated savanna occurs, similar in phenology, biomass productivity, and with an open canopy and with slightly taller trees, growing stock usually 8 to 10 meters high, underlain by a well-devel- Especially in the dry valleys & southern Zimbabwe, oped grass layer with few shrubs. These areas may be a deficit in soil moisture exists during the long dry extensively grazed (Mllington and others 1989). season. This seasonal drought is accentuated by two In Mozambique, this class coincides largely with factors. First, soils have low infiltration capacities and White's "Halophytic Communities" (1983). Rainfall of high runoff rates; consequently, little rainfall during 400 to 600 millimeters a year with moderately saline the wet season infiltrates and replenishes the soil soilsproduces grassland withacacia nilotica ss. kraussimoisture reserve. Second, rooting depth is very -m- ana with extensive bare areas. However, much of this ited in many cases. Th-e vegetation is adjusted to these area is agriculth ral land or degraded land, characterinhospitable dry-season conditions. ized by low productivity and marked seasonality. Slight variation in soil properties is reflected in These areas aredominated by agriculture, which has slight variation in the vegetation communities- On the destroyed the vegtation to such an extent that occaalluvial soils of the river valleys, low shrubby vegeta- sionally only very small patches of thicket and grass tion is domiinanl Variation in the vegetation conmmu- savanna are left between farns. These usually are nities reflects variation in both soil texture and fertil- areas either of scmbby vegetation less than 3 meters y, as well as disturbance. The latter can be quite high or of open, grass-dominated savanna with isoextensive, as many of these soils are quite fertile and lated trees which attain 5 to 10 meters. suitable for grazing or cultivation. A few areas of In northern Transvaal, this class occupies a signifibadly degraded savanna do occur in these valleys, and cant area along the Limpopo Valley, as described earthe land grades into low, unproductive Wooded Shrub- lier, but it also extends south of the Soutpansberg land (Class 35). Mountains onto higher areas northwest ofpietersburg Vegetation in the Limpopo Valley is open grassy and southward along the Mozambican border in Krusavanna with a smaller woody biomass component ger National Park. It largely consists of a dry savanna This is mainly restricted to scattered emergent trees type of vegetation-"south Zambezian Undifferentisuch as Acacia spp., Dichroslachys cinerea, and Sclero- ated Woodland and Scrub Woodland," according to carya caffjra, which attain heights of about 10 meters, White (1983). It usually occurs on alkaline soils of and scattered shrubs varying from 1 to 3 meters, dom- hotter, drier areas of lowland valleys and rejuvenated

154 Soau hermi Africa 141 upland regions. Calcretes are common. Rainfall varies Termilnalia sercea. become more frequent. This also is from 325 to 1,000 millimeters and altitude from 150 to the case in the small outliers of this class in the west of 1,525 meters. Botswana and the extreme northwest of Cape Prov- Well-developed woodland more than 9 meters tall ince. Occurrences of this class in Namibia largely cois localized; elsewhere, it is mostly scrub woodland. incide with the more densely populated areas, includ- Most of the larger woody plants are about 7 meters ing Windhoek, and with areas of privately owned high with occasional taller emergents. In its natural farmland, which is principally devoted to cattle ranchstate, the vegetationcan bequite dense, or even closed, ing (van der Merwe 1983). but never impenetrable. Some of this is now preserved Other outliers occur farther south in Cape Province. in the KrugerNationalPark,but elsewhere innorthem The area just east of Springbok coincides with the Transvaal the area is extensively grazed. A wide vari- "False Succulent Karroo" of Acocks (1953) and is very ety of species includes Acacia spp., Combreturn spp., similar to the areas in the north innamibia, exceptthat Coimiphora miaollis, Dichzrostachzys cinerea, Diosptyros succulents are more abundant. The large areas north miespiljformis, Sclerocarya cafra, Ternuinalia pruinioides, T. west of Kimberley in the Langeberg and Asbestos sericea, and Ziziphls imucronata. Commonly referred to mountains, and northern Kaap Plateau, as well as as the 'Transvaal Bushveld' (Christopher 1982), the those southeast of Kianberley in Orange Free State, drier parts are suited only to exter'sive cattle grazing, correspond with the "Kalahari Thornveld and Shrub and cultivation depends on irrigation. Bushveld' of Acocks (1953) and White's "Bushy Kar- The other significant area of this class stretches roo Shrubland" (1983). A similar vegetation ecology north-northwest to south-southeast, from southwest- occurs in the coastal range outliers, described by em Angola to the north of Cape Province. The most Acocks (1953) as "Succulent Mountain Scrub" and northerly of these areas is on the lower slopes of the "Karroid Broken Veld.- coastal ranges inland from the Desert of Moqhmedes. All these areas have a distinctive shrub layer with a Examples occur along a distinct north-south-oriented scrubby succulent vegetation. Small trees and shrubs altitudinal zone along the coast between Namibe and include Aloe dichotoma,- Ceraria namaquensis, Euclea Tombua, and on the high ridges of the Serra da Chela. tomentosa, and Ficus ingens, with smaller shrubs often Here, as farther south, this class is largely adjacent to of Euphorbia spp. Shrubs usually are less than 2 meters the Desert (Class 0) and Bushy Shrubland (Class 33). tall, but larger woody plants, either arborescent suc- The vegetation is mainly low, open, and shrubby, culents ornonsucculentbushy trees, are scattered over with some taller shrubs to 7 meters in height (for the landscape. These rarely exceed 5 meters in height, example, Acacia mellifera, A. reficiens, and Commiphora the most notable being Acacia erioloba, A. mellifera ss. zngolensis), passing into Colophospermum mopane- detinens, Aloe dichotoma, Boscia spp., Cerria spp., Cotjdomnmated communities with grasses on the higher ledon panicaata, Ehretia dgida, Euphorbia spp., and Grem7a parts. Around the Angolan towns of N'Giva, An- flav. hanca, Evale, Namacunde, Nehone, and Moygua, this This vegetation is floristically similar to Wooded land cover class is a degraded form of Open Woodland Shrubland (Class 35),butincludes areas of commercial (Class 61) to the north. This is a secondary vegetation agriculture, such as the irrigated zone along the Ordominated by stunted tree thickets on grassland, ange River near Upington. caused by clearance near settlements and, in part, by Although the growing stock is greater than in many military activities and rural population movement adjacent semiarid areas, the annual productivity is Very extensive areas of this class occur in the high quite small, which limits fuelwood extraction from plateaus of Namibia, including large areas innorthem this land cover class. It is a tranrsitional class between Ovamboland, around Windhoek in central Namibia, woodland and grassland, but more the latter, and and in the higher parts of Namaland, including parts commnonly is used for extensive grazing. Although of the Hanam Plateau. The communities often are some floristic overlap with other land cover classes dominated by Colophospermum mopane, with Cenaria probably exists, this class is phenologicaly distinct in longepedunculata, Commiphora spp., and Sesamotham- the 1986 AvmHRRNVI data. nus spp- occurring as trees to 15 meters height over a Many areas in this class already have been extengrassy sward- Farther south through Namibia, the sively exploited, as in southern Angola, in the principal bushes and trees gradually become Acacia Limpopo Valley in Zimbabwe, and in the Cliangane spp., Combretum apiculatum, Dombeya rotundtfolia, Eu- Valley in Mozambique. In many more-open areas, ceia undulata, Ficus spp., Rhus marlofii, and Tarcho- growing stock not only is small but is scattered. In nanthus camphoratus. southemr Zimbabwe, some of these areas coincide In the more eastem parts on the fine red Kalahari withcommunalareas, where deforestationis common Sands, Acacia melli#ra ss. detinens, Catophractes alexan- because of fire and land needed for grazing, cultivadri, Combretum apiculatum, Dichrostakys cinerea, and tion, construction timber,-and fuelwood. In other

155 -142 Eslimoaing Woody Biomanss in Sib-Soaran Africa areas, such as Kruger National Park, woody vegeta- Diospyros austro-africana, Euclea spp., Maytenus leterotion is protected from exploitation. Overall, it is a land pitylla, Osynis spp., and Rhus spp. These occur as scatcover class of limited fuelwood potential. Its woody tered individuals 2 to 3 meters high, or locally as small bioniass growing stock is estimated at 41 million thickets. tonnes for the whole of southern Africa, with a sus- This densely settled and intensively cultivated area tainable yield of 6.1 million tonnes. is not a significant source of woodfuel and access to the limited stock of woody biomass undoubtedly is Class 31-Veid Shrubland and Cultivation restricted by pattems of land ownership. This class occurs only in South Africa. It covers 4,005 Class 32-Hill Shrubland square kilometers of westem Cape Province, partly in lower, flatter land around the Great Berg River, south- This land cover class is relatively limited, present only eastward from St. Helena Bay and partly inland from in South Africa, covering 24,978 square kilometers. It Piketberg to Malmesbury, including some of the low is limited largely to the lower slopes of western Cape hills up to about 800 meters. It occurs mostly on the Province, occurring mainly at the foot of the north- Malmesbury Shales and part of the Quatemary and northwest-trending subparallel coastal ranges, in- Tertiary Sands nearer the coast. It coincides with the cluding the Cedarberg Ranges in the south, the Bok- "Coastal Renosterbosveld" of Acocks (1953),. now re- keveldberge and the Hardeveld in the center, and the ferred to as "Renosterveld' by Taylor (1978) and as slopes around Springbok, Nababeep, and Steinkop in "Coastal Macchia" by Moll and Bossi (1984). It also the north. overlaps in the north with the "strandveld" of Acocks The ranges are composed of the Table Mountain (1953). Sandstone or the Wttteberg Quartzites of the Cape This class has a phenology distinctive from the System, while the Cape Granites frequently form the "Bushy Karroo-Namib Shrubland" that mostly sur- foothills and lower slopes, leading down onto the rounds it, having a much more marked seasonal pat- Bokkeveld Shales and sandstone of the Cape System tem. Precipitation occurs almost entirely in winter, on the lower ground. The soils have little water-retaintotaling 570 millimeters a year. A low summer NDv1 of ing capability, are acidic, and generally are infertile. about 0.12 occurs from November to May. This in- Deeper, brown or reddish sandy loamns that are more creases quite rapidly to approximately 0.58 in July, ferile occur c:er granites and shales. August, and September in response to the winter Precipitation varies considerably with relief from rams. 150 to 250 mimeters in the north to more than 500 This class coincides with an area of shales that millimeters in the south, with a distinct winter maxiweather to form fine-grained soils, denser and more mum. This class includes "Mountain fynbos" areas in fertile than the sands of the mountains or the coast on the south. The term fynbos has been used to describe which Hill Shrubland and Bushy Shrubland (Classes the Mediterranean-type evergreen sclerophyllous 32 and 33) occur. This is the so-called swartveld, a shrubland, but it equally may be classed as a form of gently undulating landscape which has been exten- heathland (Moll and Jarman 1984). Acocks (1975) desively plowed for wheat cultivation, so that only traces scribed these areas as "Macdhi; others, such as Moll of natural vegetation remain. This has been one of the and Jarman (1984), refer to some parts as 'Mountain most important wheat-growing areas of South Africa Renosterveld." Farther north, the class includes the since early white settlement and, although yields are more southerly parts of the "Karroo-Nimnb Domain" small, it remains a prime producer of South Africa's of Werger (1978), also referred to by White (1983) as wheat Wheat is grown in rotation with oats and other "Bushy and Succulent Karroo Shrubland!" cereals, with some sheep farming, using stubble or The NDVI phenology is similar to Bushy Shrubland fallow land. These virtuary continuous wheatlands (Oass 33), reflecting a Mediterranean-type seasonal (Christopher 1982) contrast markedly with the sur- pattem,butwithaslightlyhighernrvrlevel,especially rounding shrubland. in winter as a result of the greater precipitation. Be- Natural communities that do remain are dominated tween December and June, NDvi levels are 0.12 to 0.15, by the shrub Elytropappus rhinocerotis (renosterbos), increasing to 0.22 to 0.3 from July to October, pealdng usually densely branched and about 05 to 1.5 meters at about 03 in September. high. Grasses often form a significant component, and In the southern true Mediterranean area of thisclass, scattered patches of tail scrub 3 to 4 meters high may the "Mountain fynbos" is very complex floristically. It occur, often introduced plants such as the Australian changes from tall, proteoid shrubs 1.5 to 2.5 meters Acacia cyclops, A. saligna, and Leptospermum keviga tum. high on the lower slopes to shorter, ericoid forms on Semisucculent and broad sderophyllous plants also the upper slopes, with restioids often dominating the occur, as do species of tropical affinity, principally exposed ridges. Protea nereifolia is the most common

156 Southern Africa 143 shrub on the lower slopes. Only rarely are small trees (1983), and some classified as "Coastal Renosterveld" observed, although Hieea argente and Mayenus oklides and "strandveldc by Acocks (1975). Much of this latter occur on rocky hillsides. Where surface rock is exten- area has been disturbed and degraded to a "Coastal sive, Maytenus acurninata, Olea africana, Olinia ventosa, Macchia,' according to Moll and Bossi (1984). and Podocarpus elongatus may form a closed scrub Precipitation varies between 150 and 250 milimeforest 7 to 8 meters in height. Ola capensis and Wid- ters in the north but exceeds 300 millimeters in the dringtonia cupressoides also can occur as scrub forest, south of this area; most falls in the winter months. usually less than 7 meters tall. Mean daily temperatures vary from about 13 C in Above the proteoid zone, the shrubs are smaller, July to about 22 C in January. The topography trends usually only 1.5 meters high, and are dominated by parallel to the Atlantic coast (northwest to southeast) ELic spp. and Phylica spp. Trees.are absent except in and southern coast (west to east) with the two series the Cedarberg ranges, where Widdringtonia cedar- of folds meeting near Ceres. This class dominates the bergensis persists on rocky scarps and screes. Upper lowerlandaroundthecoastandbetweentheseranges. parts of these mountains exhibit a low restioid vege- In the valleys, this class overlies Bokkeveld Shales tation, usually less than 0.5 meter high, such as that in and sandstones of the Cape System and the the northern Cedarberg Mountains. There, Restio Malmesbury Shales of the Archaean Complex. Along curziramis can form dense, low, rounded, cushionlike the coast, it overlies sands, conglomerates, and limetufts 20 centimeters high, with Cannamois nitida grow- stones of Tertiary and Recent age. Soils from the ing in dense tufts 1 meter high. Bokkeveld series are fairly fertile, but coastal soils are Farther north in western Cape Province, this mon- often shallow and clayey. tanevegetationgrades into the southern end of Wergers The other significant occurrence of this class is at the 'Western Cape Domain" (1978) of the Karroo-Namib northern edge of the ldesert of Mofimedes in Angola. region. This area was classified as "Western Mountain It extends along much of the Angolan coastal plain Karroo" by Acocks (1953). At its northermlimnit, succu- and along the foot of the escarpment inland from the lent dwarf shrubs are fairly common, but farther desert areas in Benguela and Namibe provinces. This south, as precipitation increases, succulents are less area coincides with the northern end of the "Namacommon and the dwarf shrubs grow only 1 meter land (or Namaqualand) Domain" of Werger (1978) high. These includeasparagus capensis, Cotyledon wallt- and the northern extreme of the "Bushy Karroo Shrubchi, Euphorbia mauritanica, Galeni aft cana, Pentzia spp., land" of White (1983). Patches also occur in northern Pteronia glauca, Ptemothrix spinescens, Ruschia ferox, Namibia, including an area on the fringe of the Etosha Zygophyllum gilfillani, and others. Higher up the Pan. Precipitation levels are less in this area, often less mountain ridges, between Calvinia and the southern than 100 millimeters, and mean daily temperatures Cape Mountains, this karroid vegetation merges into reach 15 C in July but only 20 C in January. "Mountain Renosterveld," with dwarf shrubs such as The phenology of this class in western Cape Prov- Elytropappus rhinocerotis, Eriocephalus afiitanus, Pentzia ince demonstrates the effect of the winter rainfall maxincana, ReLhania spp., and Ruschia multifora. imum, with NDvi values of about 0.19 from July to Woody vegetation decreases northward through- October (ate winter to sprng), decliiing to slightly out this class, but nowhere is it of great significance. It lower values of approximately 0.12 from December to is dominated again by small growing stock and poor May. Although of similar level, the phenology curve sustainable yield. for the Angolarn extent of this class shows a near reversal of this pattern, with values of 0.04 between Class 33-Bushy Shrubland July and January, and slightly greater values above 0.12 between February and May. The generally lower This class covers an extensive lowland area along the NDvi values in Angola probably reflect less precipitawestem coast of Cape Province from Cape Town to tion. The different seasonal pattern reflects the slight the Namibian border, stretching along the southern summer maximum of rainfall in this northern part of coast as far as Cape Barracouta, and reaching into theclass. interior Cape Province as far as. the Roggeveld. It Floristically, the southem area of the class is one of covers 117,405 square kilometers, with 70,032 square extreme diversity (Taylor 1972), but with few woody kildmetersinsouthafrica,33,725square kilometersin plants or trees. Typically, thejynbos occurs in the form Angola, ard 7,483 square kilometersinnamibia- of sderophyilous shrubland of 1 to 3 meters height, It coincides largely with the 'Western Cape Do- with some scattered taller bushes and, rarely, widely main" of the "Karroo-Namib Realm" of Werger (1978), spaced trees. In many parts of the Cape lowlands, this although it also includes some areas dassified as class has been replaced by secondary shrubland dom- "Cape Shrubland" (part of the fynbos) by Werger inated by Elytropappus rhinocerotis (renosterbos). (1978), some classified as "Capensis Realm" by White Streams may be fringed with riparian thicket and

157 144 Estirnali,ig Woody Biomnass in Sub-Slalart Africa scrub forest. The Protea spp. bushes and other taller areas of bare saline flats and grassy sand dunes where plants that occur in bushierfinbos may thicken to form the only woody species is the shrubby Strychnos thickets 4 to 6 meters high, but large areas are de- spinosa. However, it is similar to the South African graded by fire, overgrazing, and cultivation. occurrences in form and growing stock, and the The varied fora is characterized physiognomicauy woody component rarely exceeds 2 meters in height by three elements: restioid, ericoid, and proteoid. except near the escarpment. Restiold usually is tufted, 20 centimeters to 2 meters Land use in this class is mostly limited to grazing, high, with nearly leafless tubular or nonwoody stems. except around Cape Town where cultivation, olive- Ericoid is characterized by small, narrow, rolled growing, and commercial stock-rearing occur. This leaves. Proteoid is represented by taller bushes to 6 land cover class is of very restricted fuelwood value, meters, with leaves moderate in size, but still with dominated by a small growing stock, a small wood-torelatively small woody elements. Species common to grass ratio, and a fairly small sustainable yield. Little both western and southern coast areas include the potential for fuelwood exploitation exists except loericoids Anthospermum aetitiopicum, Eriocephalus race- cally. The woody biomass growing stock is estimated ntosus, E. umnbeflulahus, and Metalasia muricata. A con- at 116 million tones, with a sustainable yield of 5.7 spicuous restioid is Thamnoclwrtus erectus. Proteoids million tomnes. are represented by Leucadendron coniferum, L muinni, Protea obtusifolia, and P. susannae. Class 34-Kalahari Shrubland In typicalfyibos, true trees are virtualy absent The only species with well-defined boles are Leucadendron This class covers 4 percent of southem Africa, an area and W. schwar- of nearly a quarter of a million square kilometers. It argenteum,widdringfoniacedarbergensis, zii. OIea capensis and W. cupressoides,which often occur mostly coincides with the E41ahari Desert of southas tees elsewhere, here exhibit a bushy habit and are western Botswana (52,643 square kilometers), southless than 7 meters tall Dense thickets up to 6 meters eastem Namibia (63,603 square kilometers), and sometimes occur on stream banks and are dominated northem Cape Province (129,367 square kilometers), by Berzdia lanuginosa or Leucadendron salicafolium. The induding parts of the and Karroo. Outliers occur 'm latter protects its seeds from fire, and many other the higher areas to the east of the Naniib Desert, the fjnbas plants have many protective characteristics southern Great Karroo of Cape Province, and in small areas in central Botswana. It mostly coincides with aginstfire. Because of intensive land use in areas covered by White's 'Kalahari Deciduous Aaz Bushland and tiis class, many areas have been invaded by alien Wooded Grasland" (1983) and with the ' ualahari/ species, especially Accia cyanophylla, A. cyclops, A. Karroo-NamibTrarsiion:' melanoxylon, Hakea acicularis, and Pinus pinaster. The Precipitation in these areas ranges f rom 80 to 250 lower and more gradual water courses are fringed millimeters, with a summer maximum. This class with dense thickets, 5 to 7 meters tall, of Brabeium. shows only a slight seasonality, with a fairly uniform. slellatifolium,freylinia oppositifolia, and Metrosideros an- NDVI tiroughout the year that varies from 012 to 0.17 gustifolia, with some Cunonia capensis, Harfogia Wpen- foreverymonthexceptfebruaty,whenitpeaksato.21 sis, flex miis, and Maytenus oleoides. (figure 13-2). This is somewhat lower than for the In the northem part of this class, in southwestern Wooded Shrubland (Class 35), which ocors on its Angola, the vegetation forms part of Werger's "Namaland Domain" (1978), a low, open, shrubby vegetation Figure NDVI Profiles, Shrmblands (Classes 34 withoccasionalsucculents,andwithisometalershrub communrtities induding, for example, Acaca meuifera, and 35) A reficiens, and Commiphora angolensis. On higher areas inland, this shrub vegetation grades into Col- r. ophospermum mopane communities, with some grasses - 34 Botswana. on sites that are less rocky and that have deeper soils Along streams and valleys, the communimtes may con- OA tain several species of Commiphora spp. and Euphorbii spp. as well as Catophractes alxndr, Phaop!uMm sp no a 3- z sum, and Rhigozum vwrgatum. Muchof thecoastalvegetationin thispartofangola 02- was classified by Millington and others (1989) as "Dry Coastal Savanna" and 'Arid Coastal Thicket." A thin - band of deserdike vegetation is included in this dass, O, stretching along the coast north of Lobito, induding Jan Feb Mar Apr May Jun Jul Aug Sep Oct No c

158 Soutihrn Africa 145 slightly wetter margins (figure 13-2). On its drier mar- Karroo areas of Cape Province. Here it occupies large gins, this class often is adjacent to the Namib Desert. areas to the north of the coastal ranges, covering In southwestern Botswana and southeastern Nami- 123,781 square kilometers. In Namibia, it occurs as a bia, the vegetation is very sparse on the rolling sand series of broken patches in the higher ranges, trending dunes with few, widely scattered trees, always less north-northwest to south-southeast, on the inland side than 7 meters tall and normally much shorter. They of the coastal desert areas, covering 50,061 square mostly occur on dune crests and include Boscia al- kilometers. bitunca and Acacia mellifera ss. detinens. Shrubs such as A few smail areas occur in southwestern Angola, Rhigzunm trichotomum are confined to the troughs be- north and east of the Desert of M5imedes covering tween the dunes. Shrubs restricted to this area include 2,002 square kilometers. In Botswana, it occurs largely in Acacia haematoxylon and Monedhma spp. Farther to the wetter areas to the east and northeast of the Kalahari northeast, in Botswana, the small outliers of this class Desert,covering60,547 square kilometers Small outliers may contain a slightly greater tree component with occur on the border with Zimbabwe near Tuli, in north- Acacia spp., Burkea africana, Combretum spp., Commi- emtransvaal,andindrierpartsofinteriormozambique. phora spp., Peltopharum acum, and Terminalia serice, This class has fairly uniform NDVI values throughout but again they are mostly confined to sand ridges and the year, between 0.15 and 0.18 every month except have limited annual productivity. Similar parallel JanuaryandFebruary,whenitincreasesto.22(figure ridges of sand dunes occurinnortherncape Province, 13-2). This profile reflects the slightly greaterpotential where the commonest scattered trees are Acacia emoloba, for woody biomass production in this class than in the A. reficiens, Albizia anthelmintica, Bosda albitnrunc, and drier Kalahari Shrubland (Class 34). Precipitation is Tenrinalia sericea. Similar species occur in the Great mostly 100 to 250 millimeters a year, with a distinct Karroo outlier of this class, but the woody component summermaximum, butmay exceed 400 millimeters in is limited to bushes and small trees; most are less than the northeastern outliers. 3 or 4 meters highl. Wooded Shrubland mostly coincides with White's In the outliers bordering the hills above the Namnib "Bushy Karroo Shrubland" (1983) in SouthAfrica and Desert, the vegetation is a very dry form of White's Namibia. This is a shrubland dotted with smallbushy "Bushy KarrooShrubland" (1983). Arborescentsuccu- trees and large shrubs. It occurs in the KCarroo, where lents or nonsucculent bushy trees, usually less than 5 soils are mostly derived from Dwyka tillite and dolermeters tall are scattered above an open shrub layer 2- ite, and often are clayey and have a tendency to accumetershigh Succulent species include Aloe dichotoma, mulate salts. The class also occurs on rocky, well- Ceraria spp., Cotyledon paniculata, Euphorbia spp., and drained mountain slopes in Namibia. Pachypodium spp., with some water-stonrng species, Succulents are abundant, often dominating nonsuch as Cyphostemma currlii. A wide variety of non- succulent smaller shrubs. Shrubs are usually less than succulent bushes, bushy trees, and tall shrubs also is 2 meters tall, but larger, woody plants, either arborespresent, including Acacia erioloba, A. mellifera ss. centsucculents ornonsucculentbushy trees, are tinly detinens, Boscia spp., Ehretia rigida, and Grewiaflava. scattered throughout the landscape. They rarely ex- Land use in this class is mostly very extensive graz- ceed 5 meters in height and often are confined to ing. It indudes the Gemsbok National Park of south- rockier areas where water supply is enhanced by runwestern Botswana and northern Cape Province, as off or seepage from surrounding slopes. well as other game reserves. Of the arborescentsucculents, Aloedichotoam (which Althoughthe fuelwoodresource issomewhatbetter grows to 5 meters) is the most abundant, with Ceraria than in the desert that this class fringes, this class has spp., Cotyledon paniculta, and Pachypodium spp., all of a small sustainable yield. Where extraction takes which attain 3 to 5 meters in height Toward the north, place, it greatly exceeds the annual rate of wood pro- the succulents Euphorphia currorii and E. eduardoi ductivity. Growing stock is esfimated at 145 million occur; also present may be species with water-storing tonnes, with half of this in South Africa. The sustain- stems, such as Cyphostemmna currorii, Moringa oualifolia, able yield is estimated at 12 million tonnes. and Sesamothamnuspp. (White 1983). Succulent arborescents are less common in the south on the Karroo. Class 35-Wooded Shrubland Awidevanetyofnonsucculentbushes,bushytrees, and tall shrubs are likely to occur with varying den- Wooded Shrubland covers 4 percent of southen Af- sity, including Aacia erioloba, A melzifera ss. detinens, rica, occurring on 244,644 square kilometers. It is Bosiaspp., Ehretia rgida, and Grewiaflava. In the north, largely adjacent to, and very similar to, Bushy Shrub- Acacia montis-usts, A. robynia, Adenolobus peckueliti, land (Class 33), but usually occurs on the wetter mar- Colophospermu mopane, Commiphora spp., Euphorbia gins, often at higher altitude, although not always Its guerichiana, and Rhigozumuiz'rgatum occur in scattered type area is in the central Upper Karroo and Great clusters, 2.5 to 4 meters tall

159 146 Estisnaling Woody Bionwss in Sib-Saolran Africa In Botswana, this class coincides with White's Cape Province in South Africa, between Cape Town "Kalahari Deciduous Acacia Bushland and Wooded and Port Elizabeth, 17,021 -square kilometers in all. Grassland" (1983) and was classified as "Southern These areas were classified by Acocks (1953) and Moll Kalahari Bush Savanna" by Weare and Yalala (1971). and Bossi (1984) as "Macchia" or "False Macchia," but This vegetation largely occurs on the rolling sandy they also contain large cultivated areas. country of the slightly wetter parts of the Kalahari The class occurs on south-facing parts of the coastal Desert, an area of wide plains, depressions, and pans. ranges, which are composed mostly of Table Moun- The main tree species are Acacia erioloba, A. nellifera ss. tain sandstones and shales, and on the coastal lowdetinens, Boscia albitrtnca, Diclirostachys cinerea, and lands. The south-facing aspect promotes greater pre- Termninalia sericea. Low shrubs include Acacia hebeclada, cipitation and somewhat more even distribution than Bauliinia macrantha, Grewia spp., and Ziziphus mucronata. in the extreme southwestem Cape Province. Precipi- Generally, the densest tree and shrub growth is on the tation is altitudinally linked across a wide range from rises of sandy ridges; the depressions are more open. 500 to 1,500 millimeters, moving from winter maxima Farther north, the vegetation becomes a tree and in the west to spring and autumn peaks in the east bush savanna with a slightly greater tree component This results in a fairly consistent phenology. It is lowin the vegetation, including Acacia spp., Boscia albi- est in June, July, and August, at about 0.38, and peaks trunca, Burkea ofricana, Combretum spp., Commiphora at about 0.5 in February and March. This is a somespp., Croton zambesicus, Loncitocarpus nelsii, Ockna what greater NDVI value than the otherfynbos vegetapzldzra, Peltophiorum africanum, Rius tenuinervis, and tion areas in Hill Shrubland and Bushy Shrubland Terminalia sericew. These trees also are more common on (Classes 32 and 33), perhaps because in this clissfiynbas sandridges,whereaslow-growingshrubsarecommonon enjoys partial protection from fire. Some of the floral the plains and include Croeon subgnztissimus, Grewia spp., and biomass changes with elevation are described by Cminiphoa pyrcanthides, and Xmenia cuffiu Rutherford (1972, 1978). These areas also are more The small outlier in southern Zimbabwe is parallet woody, with somewhat better growth conditions. to and just north of the Limpopo River, stretching They seem to be transitional from the Bushy Shrubacross the border between Botswana and Zimbabwe land (Class 33) to the "Warm Temperate Woodland" close to Beitridge. It is a degraded shrub savanna. in Class 67. Similar outliers exist in northern TransvaaL Here, the White (1983) suggests that Protea spp. bushes and seasonal winter drought is accentuated by soil of low other tal plants that often occur scattered in bushy inifitration capability and shallow rooting depths, fnbos may produce dense, impenetrable thickets 4 to plus pressure from nearby settlements such as Tuli in 6 meters high if they are protected from fire for long Zimbabwe or from land clearance and grazing, as is periods. This may apply in these areas. A wide vanety the case in the TransvaaL The vegetation mainly con- of species occurs, including restioid (for example, the sit of very low, shrubby Colophospermum mopane up large Wildenounaa siata) and ericoid forms. while proto 5 meters in height, occurring as scattered individu- teoid shrubs also are commorl In these areas of greater als, with a variety of Acacia spp., Boscia spp., Dichros- NDVI values, certain shrubs and trees that are not true tachys cinerea, and ScLerocarya caffra. The outliers in ]ynbos species locally formt thicket or scrub forest up to Mozambique can be classed as "Dry Riparian Wood- 10 meters tall These include Cassine peragua, Euclea land," with shrubs and small trees up to 5 meters tall tomentosa, E. undulaa, Maytenus heterophylla, Myrsine scattered in a grass-dominated savanna (Millington afrkcna, CiomnithWrusj itus, Pkmelastrutrcuspidatus, and others 1989). Putterlixd pyrantha, Rhus spp., Tardonmnthus camp- Much of this area, especially in South Africa, is honmtus, and Zygophyllumorgsana. Some encroachment given over to extensive cattle ranching. The carrying of Elytropappus spp. may occur. capabilityislow,however,commonlyasinglecowper The most westerly part of this land cover class oc- 6 to 25 hectares (Christopher 1982). curs in the vicinityofstellenbosch,paarl,andwefling- The growing stock in this land cover dass is esti- ton, where vineyards are a dominant feature, together mated to be 137 million tonnes for southern Africa. with some olive growing. The more easterly parts The sustainable yield is estimated at 12 million tonnes include some sheep and cattle grazing and wheat a year, seriously limiting the potential for extraction. cultivation. The valley floors support citrus orchafds. It is probable that even a small rate of fuelwood extrac- For example, the Gamtoos Valley, possessing the most tion would exceed the annual rate of wood production. valued land of the region, is intensively cultivated. The thicket areas in this class have some fuelwood Class 42-Fynbos Thicket potential, but access islimited by land tenure patterns, and areas of commercial cultivation obviously have This land cover class is confined to a few small areas little fuelwood potential. Thus, despite an esthmated that trend east-west in the southern coastal ranges of 47 million tonnes of growing stock on this small area

160 Sotlwhern Afiica 147 of South Africa, the sustainable yield is estimated at to"highveldgrasslndandfrom"afromontaiescrub/ only about 0.5 million tonnes, so this remains an area Forest" to "Coastal Forest." The area lies chiefly beof limited woody biomass resource. tween 800 and 1,700 meters. The vegetation is mainly grassland, but originally was probably bushland with Class 43-Moist Acacia-Commiiphora patches of scrub foresl The relicts of this woody veg- Bushland and Thicket etationhaveafromontaneaffinitiesathigheraltitudes and include Apodytes dimidiata and Halleria lucida. The This class covers 441,586 square kilometers of south- lowland remnants, whichare related to coastalforests, em Africa, more than 7 percent of this part. of the include Acacia spp., Aloe arborescens, Celtis africana, continent. It crosses many floristic regions, but they Commiphora harueyi, Ficus spp., and Syzygium cordatum. are classified together by phenology and exhibit con- In South Africa, this land cover class borders the siderable ecological and structural similarity. The Highveld grasslands in the west and high-altitude class occurs in patches that formanarc extendingfrom woodlands in the east (for example, on the Lubombo the coastal plain of Angola, through southern Angola, Hills and forest plantations on the South Africanparts of Zambia, Zimbabwe, and Malawi, to a south- Swaziborder). The grassveld on the higher areas often erly extension in southern Mozambique, southern is an open, grassy savanna with areas of bushes and Transvaal, and Swaziland. smaller trees, represented in moister areas by Acacia South Africa contains 111,767 square kilometers of karroo, Maesa lanceolota, Syzygium cardaium, and Verthis class, Mozambique 75,038 square kilometers, An- nonia ampla, and in drier areas by Acacia m7otica, Dichrogola 71,224 square kilometers, Zimbabwe 63,498 stachys cinerea, Ficus capensis, Maytenus senegalensis, square kilometers, Zambia 55,067 square kilometers, and Sclerocarya birrea. and Malawi 32,408 square kilometers, with smaller This class now- is largely an area of intensive comoccurrences in Namibia, Botswana, and Swaziland. mercialagricultureinthewest,especialyfromennelo All areas exhibit the broad characteristics of an open to Harrismith. This is mainly pastoral agriculture, but bushland or thicket, frequendy above a dense grassy with some maize and sorghum cultivation. Farther sward, but often this is considerably degraded or al- east and south, in Kwazulu and inland Natal, are tered by agicultural activity. considerable areas of subsistence agriculture. As a The phenology of this class has a distinct seasonal- result bushes and trees are very restricted by cultivaity, with summer Nm maxima (between December tion and grazing,-and woody vegetation now is conand April) of 0.35 to 0.47, followed by a fairly gradual fined to watercourses,poorersoils,andsteeper slopes. decrease to September (0.12), and a rapid increase In many areas, most notably Swaziland, lower reid fromnovember to December (figure 13-3). Most areas areas have been overgrazed and invaded by Acacia coincide with a mean annual precipitation within the nigrescens, with secondary Combretum zeyheri, Dichrorange of600 to 900 millimeters,usually withasummer stachys cinerea ss. africana, and Sclerocarya birrea. The maxinum. trees can attain 10 meters height, and between them The largest contiguous area of this class occurs in are areas of low bushes, the most common being Corsouthern Transvaal, southern Swaziland, and inland dia g}haraf, Onnocarpum tuichocarpum, and Sclerocarya Natal This area coincides with two of White's transi- caffmr These frequently are shorter than 4 meters, but tionalareas(1983):from"afromontanescrub/forest" tree density is great, ranging frorn.25 to 625 trees per hectare. Eradication of this thomybushlandis difficult and expensive. This sort of bushy savanna- is both Figure13-3. NDvlProfile,MoistAcac-Conmiphorm widespread and currently extending, especialy in Bushland and Thicket (Class 43) Swaziland (ErC Foundation 1987). ln the Highveld and Middleveld of Swaziland, graz ing and cultivation have led to reduced woody bio Malawi mass and an increase in the proportion of grasses. This also is true of the more western parts of this class in z 0A4 \ - South Africa, including the small outliers on the margins of the Drakensbergs to the south and east of c 03- \ - / Lesotho. This class also occurs in Mozambique, mostly on.0.2- lowerhills facingthe coastinlnhambaneprovmce and on the slopes of the upper Save Valley, extending into southeastern Zimbabwe on south-facing slopes in - - Southern Matabeleland. Most of these areas are a low- JOan Feb Mar Apr May Jun Jul Aug Sep Oct Nov DCC woody-biomass tlicketlike form of the "Dry Minmbo

161 148 Estitttlig Woody Biomass in S:ab-Saluaran Africa Woodland" areas that surround them. Some are very Adansonia digitata, Hyphtaen ventricosa, and Stercilia degraded, whereas others possibly are edaphic cli- spp. On low-fertility soils, either thicket or savanna max, dry thicket vegetation. The proportion of shrubs vegetation forms. Tllickets occur on alluvial soils to to trees varies from high thicket vegetation, in which the south of Lake Malawi; here the species are similar trees dominate and attain 10 meters in height, to low to those in wooded areas, although they are more thicket vegetation, in which shnibs up to 3 meters are stunted. On thin, stonysoils, a wooded savanna occurs more prevalent than the trees, which occasionally which is dominated by Bracihystegia spp., Combrettun reach 8 meters. spp., Colopliospennun in mopane, and Pterocarpus spp. With decreasing height, the canopy becomes more Edwards (1982) quantitatively analyzed disturbed open and grasses and herbs increase. They are floris- savanna woodland near Blantyre. Compared with matically related to mionbo woodland and the main tree htre miioa1bo woodland, this disturbed savanna woodspecies are Acacia nigrescens, Albizia spp., Brachlystegia land (a) lacks stratification, (b) has a significantly boehmii, and Izldbermardia globiflora. Some of these as- lower basal area (4.87 square meters per hectare sume a shrublike habit, but other shrub species also compared with 11.8 square meters per hectare), and invade the thicket vegetation. In northern Mozan- (c) has a better-developed grass layer. The main bique, this class largely occurs in the middle Zambezi dominants are Brachystegia boehrnii, Burkea africana, Valley, following closely the 800-mirimeter isohyet, Jzlbernardia globiflora, Monotes aftricanias, and Pseuidobut small outliers also occur on the slopes of the upper laclinostylis maaproi neifolia. and middle Lurio Valley in the northeast. Areas of this class that occur in Zimbabwe are again A similar vegetation occurs in Malawi in the Shire largely on valley slopes, mostly to the north and west Valley north of Blantyre, on the Phalombe Plains, on of Harare (for example, on the flanks of the Hunyani the Shire Highlands, and over a large area north and Range). Patches also exist near the Mozambican borsouth of Lilongwe. Most of these areas are character- der, mainly on the sides of the Sabi Valley. As in other ized by very open woodland or a "Cultivation Sa- areas, such "Dry Savanna Woodland" (as it is called in vanna' (a term commonly used in Malawi to describe Zimbabwe) is floristically poor and dominated by a savanna of mnaize cultivation and grassland with BracJystegiaspicjformis and ulbernardia globiflora, with very scattered miombo woodland and exotic tree spe- a number of smaller trees. cies). It generally grades from very open woodland to At its drier margins and in more disturbed areas, a grassy savanna in which the tree component often is however, the canopy can be as low as 3 meters and the represented onlybymangoes. Itmainly developsafter class takes the form of an open shrubby savanna. Here woodland clearance for cultivation and fuelwood the domninant tree is Brachystegia boehmnii, with other extraction. small, invading trees and shrubs such as B. spicjformis, On very thin, stony soils, a dry, open-canopy Colophospennum mopane, Julbernardia globifora, Kirkia miombo woodland is the natural vegetation. An exam- acuminata, and Sclerocurya cafra. Some of these areas ple is on the Rift Valley escarpment, where it is known may include "Escarpment Thicket," which is common locally as "Msuku Woodland." The main canopy along the entire length of the Zaambezi escarpmenl doniinants are Bracdystegia spp., especially B. boehmii. This thicket rarely a: ains 10 meters height and forms In most areas, however, dry, open-canopy miombo a dense mixture of tree and shrub species, dominated woodland results from clearing well-developed by Combretum spp., Commipiora spp., and Pterocarpus woodland, and under increased population pressure antunesii. it grades into "Cultivation Savanna." Typical of this is In Zambia, a compact representative area of this the Lilongwe Plateau in Malawi, where the only rem- class exists in the Luangwa Valley and around Choma nants of open-canopy woodland are sacred grave- in the Southern District, together with scattered yards among grasslands and maize cultivation. The patches on the sides of the upper Zambezi Valley. In woodycomponentof the"cultivation Savanna' away the Luangwa Valley, the vegetation is largely altered from these remnants is dominated bymangoes (Mangi- "Mopane Woodland." "Mopane Woodland" is usually fera indica), Acacia spp., Combretum spp., Piliostigma a single-story, open, deciduous woodland attaining 6 spp., and Lapaca kirkiana. It is particularly common on to 18 meters in which the donminant tree is Colophosthe Lilongwe Plain and in the Namwera and Malindi pernm mopane. Although it often occurs as pure stands, regions (Nillington and others 1989). occasionally "Munga Woodland" trees invade, es- In the Southern Region of Malawi, population pres- pecially Acacia nigrescens, Adansonia digitata, Comsureinthis class hasbeen far greaterthanfarthernorth bretum imberbe, Kirkia acuminata, and Lannea stuhland "Cultivation Savanna' and woody thickets are mannii. Shorter trees and shrubs usually are absent common. Fertile soils are characterized by grassland and the grass and herb cover is only locally dominant. interrupted by copses of relict woodland as well as It is restricted to alkaline soils on valley floors which isolated tees such as Acacia albida, Cordyla afiicana, flood in the wet season and are dry in the dry season.

162 Southlern Africa 149 Destruction of "Mopane Woodland" leads to a drastic Class 51-Acacia Woodland Mosaic reduction in woody biomass because it is replaced by coarse tussock grassland. This class covers 519,892 square kilometers or 8.7 per- Farther west in Zambia, this class is represented by cent of southem Africa. Although best developed in areas on the slopes of the upper Zambezi Valley and Botswana, this class is distributed in an arc from northits tributaries. It exhibits a similar Colophospermumn western Namibia, through Botswana, into southern mopane-dominated open woodland, but stands mixed Cape Province. It occurs in northern Botswana in the with CommipFora spp. and Terninalia sericea are com- Chobe region to the east of the Okavango Delta, exmon and sometimes grade into the "Baikiaea Wood- tending north into the Caprivi Strip and a ound to the lands" on deeper alluvial soils near drainage lines. west of the Okavango Delta near Tsau. In southern The other compact outlier of this class around Botswana, it is strongly developed along the Bo- Choma consists partly of "Munga Woodland" tswana-south African border and in the Kalahari, (Fanshawe 1969). This is a type of savanna woodland with a number of significant patches south of the with an open, parklike appearance, with either one or Makgadikgadi Pan. It covers 97,012 square kilometers two woody layers, dominated by Acacia spp.. Com- in Botswana. bretum spp.,and Terminalia spp. in the upperstory and It also occurs extensively in Namibia (166,464 a dense woody understory reaching 4.5 meters. The square kilometers), with patches to the east of the bushes are deciduous to semiideciduous and are again Etosha Pan, in the Great Karas Berg, in the hilly parts dominated byacacia spp. and Combretum spp.in some of Namaland, in the mnountains surrounding Windareas of "Munga Woodland," woody thickets exist. hoek, and northward as far as the Angolan border. These are related to specific soil and water conditions, In South Africa, 142,277 square kilometers are and usually include Commiphora molffs, Euphorbia can- mapped in this dass; the principal areas are in dabr,markhamiaobtusifola,andschrebern trichockla northern Cape Province and Bophuthatswana. Although it e-xhibits variations, "Munga Wood- Patches also occur in the Transvaal near Pretoria, to land" generally is an invasive secondary woodland. It the north and east near Potgietersrus and Pietersappears to have developed by the invasion of wood- burg, and between the Drakensberg Ranges and land trees onto alluvial grasslands and all the tree Kruger National Park. Further small areas occur speces are fire resistant. Many other types of wood- scattered on the fringes of Veld Grassland (Class 11) land in Zambia are currently being invaded by in Orange Free State and southwestern Cape Prov- "Munga Woodland," in particular the riparian wood- ince. The southernmost representatives of the class lands. occur on the slopes of the coastal ranges, including In Angola, small areas occur in the southeast, sin- some areas in TranskeL ilar to those of western Zambia. The main areas are in Other small areas occur in southern Angola (25,768 the southwest, between Chiange and Otechinjan; on square kilometers), eastern and southern Zambia the west-facing slopes northwest of Lubango; and on along the Zambezi Valley (9,116 square.jiometers), the slopes above Benguela and Lobito. These include and in northern Zimbabwe on the north-facing slopes the driest variants of the Angolan miombo woodlands of the Zambezi Valley. The class covers 47,637 square and the "Eafidaea and Mopane Woodlands." The can- kilometers in Zimbabwe. Other areas occur in opy trees are generally low, about 10 meters, and are Matabeleland and into eastem Mozambique on the sides mainly. Brackystegia spp. and Julbernardia spp. On of the Limpopo Valley (31,143 square ldlometers). higher ground, a variety of shrubby and stunted trees Nearly all of these areas are transitional between invadetoformabushythicket. Theareasfarthernorth Transitional Wooded Grassland (Class 24) and Open are floristically richer than the mionbo and "Mopane Woodland (Class 61), although in the south and west Woodlands," and dominantspeciesbecome Adansonia they often are adjacent to Wooded Shrubland (Class digitafa,dichrostachysspp., Euphorbiaconspicua,Setarha 35). This mosaic of Acacia woodland, Transitional welwitschii, and Sterculia setigern. Wooded Grassland (Class 24), and Wooded Shrub- Land use in this class ranges from commercial cul-- land (Class35) is notasproductive as OpenWoodland tivation (for example, in parts of Malawi and South (Cliss 61). Its occurrences mainly coincide with those Africa) to subsistence farming (for example, Zambia areas that have an annual rainfall of 400 to 700 milliand Swaziland). Pastoralism is inhibited by tsetse in meters, although it occasionally occurs in both drier many parts of Zambia and Zimbabwe. and wetter areas, depending on local soil-moisture The fuelwood potential of all these types of open holding conditions. woodland and thicket is significant, although limited Values of NDvi. are high, more than 0.36 between by the seasonal drought Woody biomass growing DecemberandMarchandpeakingat 0.42inFebruary, stock for southern Africa is estimated at 747 million with a slow decline from March to September when tonnes, with a sustainable yield of 9.1 million tonnes. the minimum of 0.17 occurs (figure 13-4).

163 150 Eslivnalisig Woody Biiornass in Sutb-Salaamn Africa Figure NOVI Profile, Acacia Woodland better developed and tree density is much grater. Mosaic (Class 51) Nevertheless, the canopy is generally open arid, espe cially in the Colophlospennumopane-dominated areas, 0.6- the grass cover is usually poor. The trees usually are SI5 Nonliem Botswana 0.5- less than 12 meters high, with average heights between 7 and 9 meters. OA- The more open woodland in the north of Botswana is composed chiefly of Acacia spp., Cornbreturn spp., z 0.3 Tenninalia pnunioides, and Ziziphius mitucronata, with 0.2. shrubby Colophospernum mopane, Grewiaflava, and T. sericea. In the northwest, scrub woodland occurs along 0. - the Ghanzi Ridge, which is rich in quartzite and lime- -* stone and where vegetation is denser than in the sur- 0J F M A M J J A S O N C rounding.areas. Grasses are relatively unimportant Jan Feb Mar Apr Ma,y iu'n J'ul Aug Scp Oct Nov DeccI here, and shrubs are more inportant Trees include Acacia spp., Albizia anthelmintica, Boscia rehmanniana, In Botswana, bushland along the eastern side of the Combretum spp., Lonchocarpus nelsii, Montinia caryocountry comprises trees of small to medium height, phyllocea, and Pdtophorum africanum. A distinct grasswith a well-developed shrub layer. In the south, the land transition zone between dense and open wooddominant tree is Peltophorum africanum with Acacia land occurs around the Mabebe Depression; it is spp., Batuhinia macrantha, Boscia spp., Burkea africana, dominated by Cenchrus ciliarus and Chloris gayana. Ocdmapulchra, and Terminaliasericea. Farthernorth, the This land cover class continues into northeastern bushland is dominated by scattered trees of Acacia Namibia, including a large part of what White (1983) spp., Combretuim imberbe, Peltophorum afiicanum, -and classifies as the "Transition to Zambezian Broad- Sclerocaryn caifra. Among these trees are numerous leaved Woodland." ItisdominatedbyAcacia spp.., but smaller ones, such as AcaCia spp., Albizia spp., Boscia also includes Combretum collinum, Commiphora aialbitrunca, Combreturn apiculatum, Commiphora schim- cana, C. angolensis, Ochna pulchra, and Ziziphus mucropen, Terminalia sericna, and Ziziphus mucronata The nata, forming a widely spaced woodland with trees main shrubs are Dichrostachys cinerea and Grewia spp. normally less than 7 meters tall. To the north of the Tswapong Hills, the vegetation Farther. west, the largest occurrence of this class in changes both floristically and structurally as Colophos- Namibia is between Etosha Pan and Grootfontein, and permum mopane becomes more important. This species along the Kaokoland Escarpment in the north (as well occurs both asa medium-height tree and in shrub form as other west-facing escarpments throughout Nami- It usually is the domrilnant species wherever it occurs. bia, which become drier at lower altitudes). Here Col- Associated trees include Aacia nigresens Burkea icana, ophospermum mopane dominates the class at 7 to 10 Combretun spp., Commiphora nmssmnbiensis, Sclerocarya meters height and forms a light, stunted woodland caffra, and Terminalia prunioides. The most common with a shrubby understory on stony or sandy soils. shubs are Acada spp., Commiphoa pynrmdlmides, Dichro- Here and in Angola, where a number of small outliers stachys cinerea, and Grewia spp. occur, the main associated plants are Acacia spp., - The scrub woodland to the north of the Tswapong Boscia micropphylla, Combretum spp., Commiphora spp., Hills in Botswana is the vegetation community with Greuna tillosa, Rhigozum spp., Terminalia spp., and the greatest woody biomass potential in this class. It Ximenia spp. has floristic similarities to Open Woodland (Class 61) In the Windhoek Mountains, a vegetation with simand is dom-linated by Colophospennum monpane. The as- ilar structure consists of tall bushes and small trees sociated trees have variable geographical distribu- over a grassy sward. The principal bushes and frees tions and may become locally dominant. They include are Acacia hebeclada, A. hererocusis, A. reficens, ATbizia Acacia nigrescens, Burkea africna, Comnbretum spp., Corn- anthelmintica, Combretum apiculatum, Dombeyn rotunmiphora mossambicensis, Kirkia acuminata, and Sclerocarya difolia, Euclea undulata, Ficus cordata, F. guerichiana, Heecaffra. Accianigrescens and Burkeaafricana are the most ria (Oorez), Rhus marlotii and Tarchonantius camphoraus. common over large areas. The shrub layer consists of -Other areas of this class in Angola with a similar varying proportions of Acacia spp., C mopane, Dkhro- phenology and productivity ocur in the upper tribustachys cinerea, Grewia spp., Tenninalia prunoides, and taries of the Zambezi Valley. These continue along the Ziziphus mucronata. flanks of the Zambezi Valley in Zambia. In Zambia, it The portion of this land cover class in easten Bo- is again "Colophospermum mopane-dominated Woodtswana differs-from that in the KalahariDesert and the land" with Acacia nigrescens, Adansonia digitata, Comnnorthwest of the country in that the shrub layer is bretum imber, Kirkiaacuminata, and Lannea stuhhmnmnfi.

164 Sonilieni Africa 151 C. nopanepreferssodium-rich soil in wetter areassuch fontein, Bethulie, and Wepener, and in the Orange as this, because it disperses the clay particles which Riverand Caledon valleys. These areas reflectmosaics then accumulate in the deeper horizons to form an of agricultural land within the natural veld vegetation. impervious layer, resulting in a small water storage Further small outliers occur in southwestern Cape capability and poor depth penetration. These occur- Province and Transkei as a transition zone between rences continue along the Zambezi Valley beyond the veld and the coastal range woodlands. Here, spe- Maramba and into northern Zimbabwe, south of Lake cies are more varied than those of Transvaal. In some Kariba. Further outliers occur in Matabeleland as, for cases they represent degradedwoodland, especially in example, south of the Matopo Hills facing the Transkei. Limpopo Valley. Most of these areas in South Africa have consider- In Mozambique, this class occurs on well-drained able agricultural activity, mostly cultivation of maize. sites above the floor of the Lirnpopo Valley and is In addition, these areas include sdme of the most represented by a mixture of "Mopane Woodland" and densely populated parts of the country. The woody. "rdry Miombo Woodland," often adjacent to Transi- component of the vegetation may be an important tional Wooded Grassland (Class 24). It also includes source of fuelwood for some rural populations. Howthe dry riparian woodland along the Limpopo, where ever, this woody biomass often exists on privately the vegetation is essentially a grass-dominated sa- owned land, thereby restricting access. Elsewhere, vanna with shrubs (3 to 5 meters) and small trees (5 to these areas provide a significant fuelwood source, 10 meters). The species are related to the secondary most notably in northern Botswana and northwestern invasive tree and shrub species of the surrounding Namibia. woodlands. Fuelwood exploitation needs to be carefully moni- In South Africa, this class transcends a number of tored and controlled because of small sustainable floristic regions, but tends to occur on the margins of yield. Cattle grazing is an important consideration in the Veld Grassland (Class 11). In northern Transvaal, Botswana, where a carrying capability as small as one it coincides with the "Sour and Mixed Bushveldt of beast per 25 hectares (Millington and others 1989) Acocks (1975) and with White's "South Zambezian makes overgrazing commonplace. The drier southem Undifferentiated Woodland and Scrub Woodland' and western parts of this class are most prone to (1983). The soils of these flatter areas at altitudes of 500 excessive exploitation. to 1,000 meters are nutrient-poor, leached, and occasionally waterlogged during the wet season. Precipi- Class 61-Open Woodlandtation always exceeds 500 millimeters a year. A mosaic of dense woodland with grass understory is interspersed with more grassy patches in which This land cover class occupies more than 8 percent of southem Africa, nearly a half a million square kilome- Burkea africana, Termimalia sericea, and Ocmna pulchra ters. It occurs extensively in southem and-southeastdominate the woody component, and Eragrostis pal- ernangola (127,944 square kilometers), southwestern lens and Panicum maximum dominate the herb layers. Zambia (72,561 square kilometers) and northeastern The tall perennial grasses are mainly Andropogoneae Namibia (99,278 square kilometers), including the and havepoor nutitivevalue, especially in the winter; Caprivi Strip into the northof Botswana (24,767 square hence the term "sour veld." Werger and Coetzee (1978) kilometers). It then becomes a broken but extensive refer to these areas as part of the "Broad-Orthophyil band through central Zimbabwe (115,455 square ilo- PlainsBushveld,'whichincludesthespedesjustmen- meters), curving southward into southern Mozamtioned and Combretum apiculatum as dominant on bique (26,295 square kilometers) and into small areas sandy soils derived from granite. They also refer to in northem Transvaal and Orange Free State (19,023 "Terminalia Sandveld'"communitiesinthedeepsandy square kilometers). Significant, although smaller, areas of the Transvaal, which are largely on the pla- areas occur in Swaziland (2,319 square kilometers) teaus and on small outliers in the low veld on granite and Lesotho (369 square ldometers). Although they soils. transcend a number of floristic groups (White 1983), Patches of this class also occur in northern Cape all these areas are classified together by productivity Province, and these have similar characteristics to oc- and phenology. currences in southern Botswana. Other outliers occur Open Woodland occurs mostly where precipitation along the south-f-acing slopes of the Kaap Plateau and ranges from 600 to 800 millimeters a year with a sumin the middle of the Vaal Valley between Prieska and mer maximumi. It also occurs in areas outside this Barclay West. Here, Acacia erioloba in particular has precipitation range, depending on soil drainage conbeen removed from large areas to provide fuel for the ditions. In southeastem Angola, the phenology exhibmines at Kimberley. Similar vegetation occurs in its NDVI peaks of 0.42 in December and 0.38 in ApriL southem Orange Free State, in the va-inity of Spring- dedining slowly to a mirimum of 0.21 in September.

165 152 Eslimoting Woody Biornass ;i Swb-Sahamn Africa In central Zimbabwe, a similar pattem is displayed, nant species characteristic of this class are Baikiaca with a peak of 0.36 in March and low of 0.2 in Septem- plurijuga, Surkea africana, Combretumn din teri, Guibourtia ber (figure 13-5). colcosperma, and Plerocarpus angolensis. Tree densities In the south and east of Angola, the seasonal of the dominant species ranged from 1.5 to 47 percent drought is manifest in scant annual rainfall and the for all trees, with 1.4 to 48 percent for trees 20 centimelongdry season. Thisseasonality isaccentuatedby the ters dbh (diameter at breast height). B. plurijuga forms soils developed on the KIalahari Sands and the thin, a canopy which can reach 20 meters, but is commonly stony soils of the Bid Plateau. The inability of these lower in drier areas; beneath the canopy is a rich, soils to hold water throughout the dry season subjects dense, shrubby understory (Millington and others the trees to a severe seasonal drought. This woodland 1989). differs from woodlands on the Angolan plateaus: it is The topography of ancient dunes on the Kalahari more open than Seasonal Miombo Woodland (Class Sands producesacharacteristievariationinvegetation 66),structurallydifferentfromWetMiomboWoodland patterns throughout the border areas of. Angola, (Class 67), and often floristically distinct from Ever- Namibia, Zambia, Botswana, and Zimbabwe. The green Forest (Class 82). ridges are tens of meters high and several kilometers The canopy trees are generally low (about 10 me- apart and are dominated by B. plunjuga, P. angolensis, ters) and are interspersed in a grass layer of up to 2 and G. coleosperma. Interdune areas are characterized meters height. On the Bie Plateau, the dominant trees by Combretunt spp. and Tenninalia sericea. A mixture of are Brachystegia boehmii, B. gosswideeri, B. spicifornis, these occurs on the intervening slopes together with Julbernardia globiflora, and J. paniculata. On the high Erythrophie un spp. "Mopane Woodland" occurs on baparts of the plateau, between 1,900 and 2,200 meters, saltic areas and on alluvial soils such as those in the the canopy is much lower (less than 5 meters) and is Cunene Valley in southwestern Angola. dominated by B. floribunda, B. spiciformis, and J. pani- "Mopane Woodland" has the smallest woody bioculata. However, a variety of other shrubby and mass potential of al Angolan woodlands, although stumted trees invades the high ground to formabushy the open woodland of the Kuando-Kubango District thicket. Above 2,200 meters, the tree species die out continues into northwestern Namibia and the Caprivi and the open thicket is replaced by "Montane Grass- Strip. In these areas, this class is composed chiefly of land." Acacia spp., Baikiaea spp., Burkea africana, Combreaum The fuelwood potential of this woodland is far spp., Terminalia prunioides, and Ziziphus mucronata, lower than others in Angola, because both growing with some Colophospernumr mopane shrub. It occurs to stock and productivity are restrictcd. On the highest the north of the Acacia Woodland Mosaic (Class 51) parts of the plateau, fuelwood potential is extremely and has a somewhat greater productivity, although it small, but fortunately these areas are quite localized in is floristically very similar. extent In southeastern Angola, on the Kalahari Sands, In southwestern Zambia, these woodlands are rewhere the.rainfall varies between 500 and 1,000 mii- ferred to as "Kalahari Woodlands" by Fanshawe meters a year, "Baikiaea Woodland" and "Burkea afiic- (1969), and include several types. "Guibourtia Woodana Savanna" occurs. land" is a two-story open woodland with a deciduous- In a forest feasibility study of Kuando-Kubango to-semideciduous, floristically rich upper canopy 18 District, Coelho (1967) identified seven forest zones in to 24 meters high which includes G. coleospenna, varithe area covered by this land cover class. The domi- ous deciduous trees, and invasive "Miombo Woodland" species. The understory is composed of a high Figure NDVI Profiles, Open Woodland thicket (13 to 2.6 meters) of small trees and shrubs. (Class 61) Climbers and scramblers are scarce, and the grass element is of variable density "Burkea-Erythrophleuim Woodland" is a "Kalahari -61 SoutheasternAngola Woodland" that has a more open canopy than - 61 Central Zimbabwe -GuibourtiaWoodland,"and an unstratifiedunderstory It is floristically less diverse and is dominated by Burkea africarnand Erythzrophzeum africanum. Both of a cd3- these trees also exist in Guibourtia woodland, and G. -col.spema D - also is common. Other trees common to both woodland types ae Amblygonocarpus andongen sis, Combretum mechowianum, Cryptosepalum exfoliatum, Dialiumengleranum, andmonotesspp. "MiomboWood-. a- * I. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dcc Woodland." Other types of "Kalahari Woodland" are I X t land" elements are rare in "Burkea-Erythrophleum

166 Souithern Africa 153 mainly"burkea-diplorhyncliuswoodland 0 and"diplo. Dirichictia rogersii, Pamspin barzcana, Pticlopsis anrhynchus Scrub." isoptem, and Pterocarptus anilu1csii. lhe density of the These woodlands are all related, each'securing a shrub layer increases with repeated burning, and beslightly different ecological situation. In general, they neath shrubby understories, grasses are poorly develhave developed because of the destruction of natural oped. Baikiaca forests which now are restricted to the extreme The monlane vegetation in Zimbabwe is complex. southwest of Zambia. The Baikinea forest area has de- True montane forest, sirnilar to that in Malawi and in clined rapidly in recent yeare. due to cultivation, burn- East Africa, is rare, for two reasons. First, the Ziming, competition from Cryptosepalurn forest, and, most important, timber extraction. B. pluinjuga ("Zambezi babwean mountains are generally lower. Second, many areas of southwestem Zimbabwe are dry leeteak") is a valuable timber tree. The destruction of the forest leads to "Chipya" and other types of "Kalahari ward slopes of mountains that form the border with Mozambique. Consequently, Zimbabwean inontane Woodland" or to a secondary "Baikiaea Wood]and" forests suffer greater drought stress than the equivawith elements from adjacent "Munga and KCalahari lent forest in Mozambique. Because of this, savanna Woodlands." In Zimbabwe, this class includes the same "Open woodland occurs to 2,100 meters. The canopy rarely exceeds 6 meters in height and is underlain by sparse Savanna and Baikiaea Woodland" vegetation de- shrub, fem,and grass layers. The canopy is dominated scribed above, plus some montane vegetation adjacent by Brachiystegia spicifonnis, with occasional areas of B. to areas of Evergreen Forest (Class 82). Much of this glaucescens, B. taxifolin, and Uapaca kirkiana. classgenerally exists onwell-drained soils above 1,350 In southem Mozambique, this class occurs in the meters, although it occurs as low as 675 meters on the Liinpopo Valley where it is a "Dry Miombo Wood- LimpopoEscarpment. The canopy is between 6 and 13 land" with a maximum canopy height of about 8 to. meters tall and is dominated by Brachystegiaspiciformis 10 meters. As it becomes drier it can be restricted to and fulbernardia globiflora. The tree canopy varies be- only 3 meters height and forms an open, shrubby tween 50 and 80 percent; shrub cover is open and usuallybelow50percent,and grasscoverranges from savanna. This latter form is the more common. "Dry Miombo Woodland" is floristically poor and domi- 50 to 80 percent. An increase in the proportion of nated by Bracitystegia boehlmtii, B. spiciformis, and Julbemnardia spp. is associated with increased distur- Julbernardia glabflorn. Scattered among the canopy bance, usually afteragriculturalclearance. Witiin this w oodland are patches of grasslands and savannas of are smaller trees of Monotes spp., Protea spp. and Uapaca spp. Where the canopy is very disrupted, Acacia spp. and Parinari curatellifolia. Allof thesediver- shrubs commonly invade and a grass layer 0.6 to 1.2 gentvegetationtypesmaybeindicativeofdisturbance meters high develops. In these cases, the dominant or local environmental conditions. The fuelwood potential of these vegetation inclusions is much smaller tree is B. boejmii, but other small trees and shrubs invade the canopy, such as Burkea africana, Faurea than the B. spiciformis-1. glabifiom woodland enclosing speciosa, Hymenocardia acida, Ocdna sclhweinfitrthiana, them. Parinari curatellifolia, Swartzia rnadagascarienstis, Syzygl in In Matabeleland North, west of Hwange on the guineense ss. gineense, Termdnalia bmchystenuna, and Vandeep Kalahari Sands, the class is dominated by an guenopsis lancafolia. open, dry deciduous Baikiaca forest. The open nature In South Africa, this land cover class is a version of of the forest is again a function of its exploitation for Acacia Woodland Mosaic (Class 51) having slightly agriculture, fuelwood, and timber. The B. plunju,ga greater productivity. It represents the denser woodwoodland canopy is extremely variable in height, land areas, with dominant species being Burkea afriranging from a dwarf form (only 1 to 1.5 meters) to cana, Ochna pulcdra, and Terminalia sericea, along with regenerated specimens about 20 meters high. Other Combretum apiculatum on sandy soils. Much of this canopy trees are rare unless the area has been exten- class includes a mosaic of agricultural activity anrd sively disturbed. At its edges, Baikiaea forest grades into associated irrigation developments. Examples include intensive orchards north of Nelspruit and around Zebwfiela in woodlands and bushy savannas which contain a mix- the Transvaal and irrigation developments along the ture of tree species. Associated with B. p!arijuga on Vaal River in Orange Free State. deep, sandy substrates are Burkea africana, Entandro- In Lesotho, this land cover class is represented by phragma caudatum, Erythrophleum africanum, Ptero- pockets of bushy woodland along escarpments and carpus antunesii, and Ricinodendron rautanenii. The in- riparian areas. Along escarpments, the dominant trees vasive Acacia erioloba and Combretum collinum also are are evergreen such as Diospyros whyteana, Euclea spp., widespread. A floristically rich, dense, shrubby under- Halkeria lucida, llex mitis, Maytenus spp., Olinia enarstoy 5 to 8 meters tall exists, in which the main species ginata, and Podocarps latifolius, with a few deciduous are Baphia nassaiensiss. obovata, Combretum engeni, species. In the riparian woodland component, the

167 154 Eslinmtiig Woody Botimss in Sub-Sal)auan Afrien main woody species are Acacia karroo, Ce1tis africana, are ecologicallysomewhatdifferent, buthaveasimilar Diospyros lycioides, Populusspp., Rtus fancea, Salixspp., phenology and productivity to Seasonal Miombo and Ziziphius niucronata. Generally, these are poor- Woodland. These coastal areas in South Africa and quality scrubby woodlands, known as shallahalila, and Mozambique coincide with White's "Eastern Africa are a product of overgrazing; however, they are an Coastal Mosaic" (1983) and with White and Moll's important source of fuelwood in Lesotho. "Tongaland-Pondoland Regional Mosaic" (1978). In Lesotho, this land cover class contains the largest These areas occurmostly whererainfall ranges from woodybiomassstock, but much of itis on escarpments 800 to 1,200 millimeters a year, with a summer maxior along rivers and is inaccessible or has restricted mum. The phenology reflects the rainfail seasonality: access. In Mozambique, the sustainable yield is re- minimum winter NDVI values occur in August in Anstricted by seasonal drought. Although the class pro- gola (0.19), September in Zambia (0.22), and October vides readily accessible woody biomass in areas with in northern Mozambique (0.18). All three curves (figotherwise poor potential, the class is not a significant ure 13-6) show a rapid increase in NDVX values through long-term resource. the sprin& staying mostly above 0.40 through the In most of the other areas-angola, Botswana, summer months. Peaks are 0.54 in Angola (Decem- Namibia, Zambia, and Zimbabwe-these woodlands ber), 0.55 in Zambia (December andjanuary), and 0.52 have moderately high growing stock but smaller sus- in northern Mozambique January). tainable yield, so that only slow rates of exploitation Tropicalcoastal woodland areas of SouthAfrica and are realistic if widespread vegetation destruction is to Mozambique have a peak NDVI value of 0.45 in March. be avoided. Open Woodland does provide a poten- For the rest of the period from November to June, NDVI tially important fuelwood source, however, in many values range from 0.38 to After June, levels-diareas where surrounding lands are less endowed. The minish to approximately 0.25 from August through most favored sites are the "Montane Woodlands" of October, followed by a rapid rise through November- Zimbabwe, although many of these areas are reserved. December (figure 13-6). A potential conflict also exists between fuelwood and In Angola, Seasonal Miombo Woodland resembles timberextractioninmanyareasonthekalaharisands. Wet Miombo Woodland (Class 67), but has a more This class has an estimated 1,446 million tonnes of marked seasonality, a more open structure, and a woody growing stock throughout southern Africa, slightly lower productivity. Undisturbed Seasonal with a sustainable yield of 21 million tonnes. Miombo Woodland has a clear structure, a closed canopy to 30 meters in height, and is floristica}ly rch. The Class 66-Seasonal Miombo Woodland main species are Brachystegia boehmii, B. gosswieleri, B. (including Tropical Coastal Woodland) spiciformis, B. wangermeeana, Combretum spp., Isoberlinia angolensis, and Julbernardia paniculata. A well- This is the largest of all land cover classes of southern developed tree layer at 5 to 10 meters occurs above a Africa, accounting for more than 1.3 million square dense- herbaceous undergrowth. Apart from rainfall kilometers, or about 22 percent of southern Africa. It amount and seasonality, the strongest controls on consists of well-developed seasonal forest to the south miombo woodland distribution in Angola appear to be of the Mesophilous Humid Tropical Forest (Class 85) vegetation disturbance, depth of soil, and altitude. and Ombrophilous Humid Tropical Forest (Class 87) In the dissected river valleys of the north, which of the equatorial areas. It is characterized by strong trend north-south, a rapidly regeneratingriparianforseasonality and asemi-evergreenordeciduousnature. Seasonal Miombo Woodland covers nearly the Figure NDVI Profiles, Seasonal Miombo northemhalf of Angola and along theangolanescarp- Woodland (Class 66) ment to the south of Huambo. Consequently, it xccurs in all districts except Namibe and Kuando-Kubango, 0.6- covering 450,070 square kdlometers. In Zambia, it is widespread, although more scattered, occurring in 0> every province and covering 344,891 square kilome ters. It occurs in Malawi (46,425 square kilometers), across the north of Zimbabwe (78,252 square kilome- a 03 v ters), and large areas of Mozambique (360,646 square Z K Noribem Angola kilometers) Smaller areas of mostly tropical coastal woodland in southern Mozambique, Natal, eastern Cape Prov- o Nordem Mozambique ince (54,065 square kilometers total for South Africa) Mozambique coast andsmallpartsofswaziland(4,479squarekilometers) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

168 Souithern Africa 155 est occurs, 35 to 40 meters tall. On the drier interior globiflora, and Terminalia sericea appear with greater plateau, however, the forest is less productive and frequency in the canopy, The shrub and grass layers more sensitive to disturbance. Existing in northwest- also may be better developed. In the hills along the em Lunda Norte Province is a semi-evergreen wood- border of Zaire, in Northwestem Province, along the land of Marqucsiacurninata-Pteleopsis diptera, 35 to 40 Mozambican border, and on the Zambian extension of meters tall, although Marquesia rnacrazira-brachystegia the Nyika Plateau, the trees come under severe taxifolia woodland is more common. This is a succes- droughtstressduetothesmallerwater-holdingcapacsional mnfombo woodland with a dense evergreen can- ities of the thin, stony soils. Here the canopy is domiopy of 30 meters height, dominated by the two trees, natedby Brachystegia glaucescens, 8. microphylia, B. taxiplus Brachiystegia spp., Daniellia alsteeniana, Pterocarpus Wlia, and Cryptoseplum exfoliatum. angolensis, and Uapaca spp. A small tree and shrub The drier end of this class in Zambia is a "Dry layer also occurs, dominated by Bridelia spp., Erythro- Miombo Woodland" featuring both natural and dislphleum spp., Faturea saligna, Parinari curatellifolia, and turbed elements. It is common on alluvial sands and Uapaca spp. It is likely that the miombo woodland inrbie flats, along the Zambezi Valley, the Cafue Flats, and, and Hula districts are similar to those in LundaNorte in part, on the Kalahari Sands. In the first two areas it District, is closely associated with "Munga Woodland." The In western parts of Bid, Cuanza Norte, Cuanza Sul, canopy of such woodlands is dominated by Ery- Malanje, Uige, and Zaire districts, wooded savanna- tlhrophleum africanum, comnmonly with Brachystegialtype vegetation is more common than woodland, with lenii, B. bussei, BW,ea africana, Isoberlini angolensis, patches of bushy thicket and grassland, depending on -Jutbernrdia globi.lom, and4 Terminalia sericea. 'he woodthe environment. This is altogether drier and more land has a more open canopy of widely spaced deciddeciduous than the surrounding woodlands. The dom- uous trees, enabling a diverse understory-to develop, inant species are Cochlospermum angolense, DiploJhyJuCs of which the common elements are Baphia massaiensis, condylocarpon, Piliostigna thonningii, and Terminalia seri- Combretum elaeagnoides, Crossopteryx febrifuga, cea. The shrub layer can develop into a thicketlike Dalbergiella nyasae, Diospyros kirkii, Diplorhynchus consavanna, but usually only a well-developed tall grass dylocarpon, and Pseudolkchnostylis maprouneifolia. layer exists. "Munga Woodland" is common on the Kafue Flats, To the east, much of this land cover class is charac- in Southern District, and along the Lunsemjwa, lower terized by an open-canopy miombo woodland that is Luangera, and Zambezi valleys. It is a type of savanna floristicauy poorer than wetter miombo woodland woodland with an open, parklike appearance. One or areas. In particular, Brachystegiafloribunda is far less two woody layers exist, both deciduous, with emercommon or even absent locally. Other species of gents reaching 18 meters in height. Particularly imn- Brachystegia that occurs in the wetter miombo wood- portantinthecanopyareacaciaspp.,combretumspp., lands are less important In most areas, particularly on and Tenninalia spp. trees, although the woodland can thin or freely dramiing soils, Bracltystegia gosswieleri, B. be floristically diverse. Sometimes a dense understory spiciformis, and Julbernardia paniculata dominate. reaching 4.5 meters occurs. The bushes are deciduous The stunted, shrubby understory of trees also in- or semideciduous and floristically diverse; but nonecreases because more light penetrates through the theless, they are dominated by Acacia spp. and Comncanopy. The grass and herbaceous ground layer also bretzm spp. Climbers occur occasionally, and a tall, is better developed. Nevertheless, in wetter areas, the thick grass layer always exists. It often is bumed canopy trees are very similar to the wetter miombo annually. woodlands. This land cover class also includes a degraded InnorthernBieDistrict,SeasonalMiomboWoodland miombo woodland, notably in the Kafue Flats and in is very poorly developed. The canopy varies between Luapala, Northem, and Northwestern provinces. In 6 and 30 meters height and the tree canopy cover these areas, woodland is destroyed by a type of shiftvaries from 15 to 70 percent. The canopy is dominated ing agriculture practiced by the Bemba in Zambia by Brachystegiq tanxiindoides, Maniuesia )oadensis, Syzy- known as chitemene. It usually occurs in "Miombo or giumguineense, and Lapaca spp. The opennature of the Combretum Woodland." Mainly cassava, cowpea, canopy leads to a low tree and shrub layer (reaching 5 maize, and millet are cultivated, with groundnuts meters) and an open, grassy undergrowth- planted in the secend year. In Zambia, Seasonal Miombo Woodland is again The regrowth of abandoned plots has been studied closely related to Wet Miombo Woodland (Class 67) by Stomgaard (1985,1986), and is discussed in detail (Chidumayo 1987). The woodland is similar in struc- in Chapter 12 (EastAfrica). Overall, the sitaationis one ture, but it is more open and leaf loss in the dry season of declining miombo woodland species as they are is far greater. Species such as Bracdystegia allenii, B. replaced by fire-resistant and fire-tolerant trees charbussei, Burkeaafricana,Isoberliniaangolensis,Julbernardia acteristic of "Combretum Savanna.' The dominant

169 156 Estinialigng Woody Bbamnass hi Siib-Salhra: Africa trees are Bracltystegia spiciformis, Combrehtm mecdio- south is largely a result of extensive clearance for uianiiuii, Diploriiyncdiuis condylocarpon, and Syzygium fueiwood and cultivation. guineense. The canopy these form enables Unpaca spp. Undisturbed seasonal open-canopy rnioinbo woodto invade, and, much later in the succession, the main land occurs in northern and central Malawi, with an componentsofthebmchystegia-julbernardiawoodland open canopy dominated by Bmchystegia boelzmii, B. may reappear. Other vegetation types are common in mnanga, B. stipulata, B. spiciformis, Julbernardia globl4lora, the degraded SeasonalMiom1bo Woodland; in northern m. paniculata, and Isoberlinia to:nentosa. Once cleared, Zambia, these include "Itigi Forest' and "Lake Basin the nature and composition of the woodland changes Chipya Woodland." drastically. In the south of the country, three different "Itigi Forest" exists along the Zairian border near types of vegetation can be seen, which are described the Mweru Swamp. It is composed of deciduous trees in the following paragraphs. 6 to 12 meters high, mainly Bapltia massaiensis, Boscia Slightly denser wooded areas occur on previously wngustjfolla, Burttia pnrnoides, Bussen nmassaiensis, Dios- cultivated soils that remain somewhat fertile, so that pyros mweroensis, and succulents. Underneath is a de- relict niiombo woodland and isolated trees occur in ciduous-to-evergreen thicket 3 to 4.5 meters high, patches. These contain mainly Acacia albida, Adansonia dominated by Boscia mossambicenmis and Teclea isleri. It digitata, Cordyla africana, Sterculia spp., and palms, paris a preclimax forest unable to attain dry deciduous ticularlyhyphaeneventricosa, between grassland areas. forest forms because of poor drainage conditions. In less-fertile areas, the vegetation degrades into Disturbance of 'Itigi Forest" results in the formation either a wooded or thicket savanna. Thicket savanna of "Lake Basin Chipya Woodland and Scrubland." occurs on alluvial soils in the lower Bwanje and Shire This is a thre-e story woodland with an open decidu- valleys and around the southem shores of Lake Malawi. ous canopy which can attain 27 meters height. The The tree species are similar to those in wooded areas, but main canopy dominants are Albizia antunesiana, Burkea. all are much more stunted. rfticaa, Combretum collinum, Erythrophleum af icanumn, On thin or stony soils, a very open wooded savanna Parinari curatellifolia, Pterocarpus angolensis, and Ter- forms, dominated by Brachystegia spp., Colophosperminalia sericea. This is underlain by a floristically di- mum mapane, Combretum spp., and Pterwcarpus spp. verse, evergreen-to-semideciduousunderstory,6 to 12 Considerable areas in this class are devoted to cultivameters high. The main dominants at this level are tion, including cotton, which is grown mainly along Cormbretum spp, Diplarhyndhus condylocarpon, Markha- the Lower Shire Valley below 600 meters, and tobacco mia obtusifolia, Piliostigma thonningii, Pseudoladcnostylis which is generally grown at altitudes of 500 to 1,000 rnaprouneifolia, and Syzygium guineense. Under the meters in the Shire Highlands around Blantyre and middle story are a shrub layer, 2 to 3 meters high, and Limbe and on the Lilongwe Highlands. Other subsisa rich ground flora, 0.6 to 2 meters high. It occurs tencecropsindudemaize,.cassava,millet,andrice,the extensively in Luapula Province along the Zairian last along the lake shores. border and around the Bangweulu Swamp. In Mozambique, this land cover class is an interme- In the north of Zimnbabwe, this class coincides with diate miombo woodland type, between the wet and dry a "Seasonal Savanna Woodland." In less-degraded phases This relation is clearly reflected in its geoand moister areas, the canopy trees reach 8 to 10 graphicaldistributioninthe ZambeziValley.Innorthmeters and are dominated by a mixture of typical em Mozambique, it is representative of slightly drier Zimbabwean savanna woodland species, mainly Bra- conditions than in the surrounding Wet Miombo chystegia boehmii, B. spicifornis, and Julbernardia globi- Woodland (Class 67), and in the south it represents flora, although other trees are invasive. Tree canopy wetter conditions than the surrounding "Dry Miomnbo cover varies between about50 and 80 percent, and the Woodland" and scrubby vegetation. shrub and grass cover is better developed. The under- In less-degraded areas, and in moister areas, the story is dominated by shrubs and smaller trees such as canopy trees attain 8 to 10 meters height, and, because Diospyros kirki, Faurea saligna, Protea gaguedi, Pseudolatch- of seasonality, are more open than WetMiombo Woodnoslis nuwmb/ila, and Psorspenmumftbn4fum. Many land (Class 67). Small trees, shrub layers, and grass of these areas include tobacco cultivation, often grown layers are better developed. In more drought-prone in rotation withmaize, especially in the Harare regiorl areas, this woodland degrades into an open savarna In the north of Malawi, Seasonal Miombo Woodland with a well-developed herb and grass layer and is dominated by a denser, open-canopy woodland widely spaced trees to 8 meters; the canopy is domiwhich exhibits a strong seasonality caused by climatic nated by Brachystegia spiciformis and Julbernardia globior pedological factors. In the south, the woodland is flora. This class also includes the extensive tea plantafar more open and commonly grades into thicket and tions of Zambezia. wooded savanna, although some of it can be termed In southern Mozambique and South Africa, areas of open-canopy woodland. This poorer vegetation in the woodland along the coast and in some inland areas

170 Soiltiernt AfriM 157 have a different floristic composition but similar pro- In the mist belt forests of parts of the coastal and ductivity and phenology. These are tropical coastal interiorhighlandsof Natal, the maincanopycan reach woodlands, divided into five types by White and Moll 18 to 25 meters and emergents may attain 37 met-ers. (1978). Four of these-"sand Forest," "Dune Forest," Lianas and epiphytes are rare, but a bush and herba- "Swamp Forest," and "Fringing Forest"-occupy ceous layer encroaches in more open woodland areas. edaphically controlled locations. The fifth, "Undiffer- On the coast, the wind-trimmed coastal thicket soon entiated Lowland Forest," is more widespread and gives way to a scrub forest with Cordia caffra, Ekebergia consists of mixed evergreen and semideciduous spe- capensis, Euclea racenosa, Mimusops caifra, Sideroxylon cies of a height and stratification depending on local inenne, and Trichilia dregeana forming an open canopy site factors. over a dense, bushy understory. Canopy heights range from 10 to 30 meters with a In places on the rocky Transkei coast, Eurphorbia large number of woody lianas. Characteristic canopy triangularis forms a narrow fringe of scrub forest 10 trees include Albizia adianthifolia, Bractylaena spp., meters tall at the mouths of rivers. Scrub forest in the Cassipourea spp., Celtis spp., Chiaetacme aristata, Comn- interior valleys is sometimes dominated by Aloe brethm kraussii, Drypetes gerrardii, Ficus spp., Millettia bainesii (12 to 15 meters tall) associated with a wide grandis, Mimnusops obovata, Protorhus longgifolia, and variety of other species. The original forests would Trichilia dregeana. A widespread subcanopy layer ex- have extended along this entire coastal belt of South ists, with a wide variety of species; tail subwoody Africa, with the most luxuriant stands approaching plants up to 23 meters may occur. Many of these areas rain forest in stature and structure. The canopy would have beencleared, especiallyforsugarcane cultivation have been semi-evergreen to evergreen, with up to 30 along the coast of NataL - species represented in any one place out of approxi- "Sand Forest," which largely occurs to the south of mately 120 species for the entire region (White 1983). Maputo, has a deciduous habit occurring on sandy However, the remnants of such forests are few, and soils which are pale orange to grey, with rainfall be- those that exist are classified as Wet Miombo Woodtween 700 and 900 millimeters a year. Here the forest land (Class 67) and Evergreen Woodland Mosaic is 10 to 25 meters high, with a canopy of Balanifes (Class 71). maughamii, Cleistanthus schleclteri, Newtonia hilde- Much of this class is a mnixture of cultivation and brandtii, and Ptaeroxylon obliquum, with a well-devel- woodland. Cultivation ranges from commercial, as in oped subcanopy with Croton spp., Grewia spp., and the sugar plantations in Natal and tobacco estates of Ochna spp. northeem Zimbabwe, to subsistence, most notably in "Dune Forest" is well developed north of Cape St. parts of Angola and Zambia. In Malawi much of the Lucia, with common pioneering species such as woodland is owned by tobacco estates, providing an Scaevola tiunbergii giving way to a full-canopy forest important accessible fuel source for curing tobacco. Of which includes Diospyros rotundifolia, Eucea natalensis, course, this limits access to fuelwood for others. and Mimusops caffra. In South Africa, mnany forest reserves exist in the "Swamp Forest" occurs on wet sites near streams tropical coastal woodland, both on the coast and and lakes, but is now very limited in extent. The gen- inland. Some small areas have been given over to eral structure is a closed canopy of even height (about commercial forestry. Elsewhere, Seasonal Miombo 30 meters) with a sparse woody understory and a Woodland contains extensive growing stock, dewell-developed herbaceous layer. spite the strong lull in dry-season productivity. "Fringing forests" follow the Limpopo VaUley in Where Seasonal Miombo Woodlands occur in associsouthern Mozambique. Common species include Eke- ation with drier woodland areas, they are important bergia capensis. Fis spp., Ranolfia caffia, Syzygium spp., source of fuelwood; but in wetter regions they may Trichilia emehica, and Xanthocercis zambesiaca (White be less important than the surrounding Wet Miombo and Moll 1978). Woodland. Other patches of tropical coastal woodland occur Significant variations in growing stock occur besouthwest of Durban along the Umkomaas Valley tween different regions; for example, between inland of thedrakensbergs;around the coastnearport eastern and westem Angola; between the smaller- Shepstone, extending inland to the Transkei border; potential "Munga Woodland" in Zambia and the along the "Wild Coast" at the mouth of the Umtata wetter Seasonal Miombo Woodland; and between the river in Transkei; and along the high ridge that forms more degraded areas of southern Malawi and the the border between Orange Free State and Natal, be- greater potential in northern MalawiL Estimated tween Ladysmith and Harrismith. These represent woody growing stock of this class throughout southsome of the better-preserved forest areas of the Tonga- em Africa is 7,463 million tonnes, with an estimated land-pondoland floristic area recognized by White sustainable yield of 119 million tonnes, making it a (1983). very important resource, but one in need of careful

171 158 Estittaling Woody Biotiass inl Stib-Samnrain Africa management. It is about one-third of the woody is a response to cooler temperatures rather than resource of southem Africa. drought. In South Africa, these areas were classified by Class 67-Wet Miotnbo Woodland (including Acocks (1953) as "Valley Bushveld" and by White and Warm Temperate Woodland) Moll (1978) as transitional between "Afromontane" and "Undifferentiated Lowland Forests of the Tonga- In southem Africa, this land cover class occurs over land-pondoland Mosaic." The natural woodlands about 7 percent of the land surface, an area of 409,916 vary considerably in luxuriance. Some have closed square kilometers. Two distinct and widely separated canopies with many trees at 20 to 30 meters in height, areacexist. WarmTemperate Woodland occurs on the but the majority of these temperate woodland areas coast of South Africa, covering about 23,500 square have a much more open and lower canopy and have kilometers. Wet Miombo Woodland stretches across been considerably altered by agricultural and other the SADC states and covers more than 385,000 square land uses. At their driest end, which is usually accofnkilometers. These two areas are discussed in the fol- panied by warmer temperature, they grade into lowing sections. Fynbos Thicket (Class 42). The tree flora is fairly uniform buthasa widevariety Warm Temperate Woodland of species, including Apodytes dimidiata, Combretu1m kraussii, Cryptocarya Iatifolia, Curtisia dentata, Halleria The South African area mainly occurs between licida, flex mitis, Kigelia aftricana, Nuxia floribunda, Oc- Knysna in Cape Province and Kokstad in westem olea billata, Podocarpus spp., Prunus africana, Ptaero- Natal, although small areas also occur along the xylan obliquzun, Scolopia mundii, and Xyrnalos nionospora. coastal ranges of Natal and in the Drakensbergs of These arboreal communities exist today largely on Transvaal. This latter area includes one large expanse steeper slopes in valleys and in higher areas because stretching from just west of Uitenhage to east of Cra- much of the land, especially in Transkei, has been hamstown; it also includes significant highland areas severely degraded as a result of great population presin Ciskei and Transkei, between King William's Town sure. In addition, on rocky sites, sandstone outcrops, and Umtata and east of Umtata, from Mt. Frere down and unstable slopes, bushland is more common. With the UImzimvubu Valley to the "Wild Coast" near Port increasing altitude, ericaceous forms, especially Passe- St Johns. nna spp., Philippia spp., and Widdringtonia spp., dom- Annual precipitation in this class varies widely inate in a scrub forest of only 5 to 7 meters height. within the range of 500 to 1,500 millimeters a year, but The most southwesterly occurrence of this land usually is between 650 and 750 millimeters. The NDVI cover class is the Knysna Forest along the coastal area is close to 050 in the period from November to Febru- to the west of Port Elizabeth. This varies greatly in ary, exhibiting a slow decrease to a minimum of 0.27 stature and floristic composition in relation to slope, in August, followed by a rapid increase in the spring aspect, altitude, and soil moisture (Taylor 1978). At (figure 137). The summer maximum is a vegetative lower altitudes, where temperatures are warmer and growth response to precipitation and warmer temper- droughts are more common, this class grades into atureincoastalsouthafrica,andthewinterminimum scrub forest, 1-ushland, and thicket, with dominant species being Diospyros dichirophyjla, Eudea racemosa, Grewia occidentalis, Maerma aifra, Sydemoxylon inerntc, and Rhus Spp. Figure NDVI Profiles, Wet Miombo Woodland At higher altitudes, with decreased temperatures (Class 67) and increased humidity, canopy height again falls off rapidly, and forest gives way to scrub forest and 0.6 thicket. The main species are Berzelia intennedia, Diospyros glabra, Leucadendron eucalyptifolium, Protea 0.5.4/- cyntrnides, and Virgilia oroboides. Many areas of this 0.4 % \ - Z - -class are mosaics of agriculture, orchards, afforested areas, and natural woodlands. This is true of areas o 0.3- north of Nelspruit, in the coastal ranges of Natal, and z 0o2 - near East London, where the mosaic of agricultural land and warm temperate coastal forest extends to the Zambia "Wild Coast." -67 Southem Afuican cont This dass is an important source of fuelwood in the 0- I I I I -,,Ir-,T,-,-, densely settled Transkei uplands, where it is exten- Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dcc sively exploited. There is a marked contrast between

172 SoIlJIIrn Afnicn 159 the accessibility of fuelwood in the Transkei and floristically rich, with a dense, closed canopy, often exceeding 15 meters but capable of attaining 30 memercially managed, in adjacent Cape Province. The tets. It is dominated by Brachtystegia boehmnii, B. gos- Ciskeicomparedwiththeequivalentareas,oftencomwoody biomass growing stock in these regions is esti- wuieleri, B. spiciforntis, B. wangerinecana, Combretum spp., andjulbernaryield of 1.3 million tonrnes. mated at about 74 million tonnes, with a sustainable Cussoniciangolensis,Isoberliniaangolensis, dia paniculata. Other canopy trees are less common, but a welldeveloped stunted tree layer exists at between 5 and Wet Miombo Woodland 10 meters. A herbaceous undergrowth also is well The Wet Miomba Woodland part of this class occupies developed. Within the woodland, stands of Uapaca approxinately 6 percent of the land area of southem forest occur, 4 to 10 meters high. These indicate a Africa, an area of more than 385,000 square kilometers. successional vegetation following agricultural clear- Itcovers large areas in Angola (150,234 squarekilome- ance. On the higher ground in the Bie Plateau, the ters), especially in northern Angola, where it includes woodland canopy is lower (6 to 25 meters) and more the Dernbos cloud forest in the northeast of the coun- open (cover varies from 30 to 90 percent). Here it is try and large areas on the border between Bi and dominated by Brachystegia bakerana, B. Iotigifolia, B. Moxico, in Lunda Sul, southeastern Malanje. It occurs spiciformis, Copaifera baumiana, and Guibourtia colin eastem Moxico and extends across the Zambian eosperma. There is a low (0.5 to 2 meters) woody layer border into Northwestem Province. with 50 to 60 percent ground cover and a dense, tall Elsewhere in Zambia, it occupies a considerable grass layer (2 to 3 meters). portion of Copperbelt, Central, Latlipula, and Lusaka The Dembos cloud forest on the northeastern esprovinces, with smaller areas scattered in Northem, carpment of the central plateau of Angola between Southern, and Western provinces, totaling 102, ' N and 9030' N also is included in this land cover square kilometers. Malawi has significant occur- class. Cloud forest vegetation exhibits less moisture rences, mainly along the shore of Lake Malawi and at stress than that on the adjacent lowlands and plateau high altitude, totaling 7,694 square kilomieters. Im- due to the frequent mists and increased precipitation vortant areas also exist on either side of the Zambezi (1,100 to 1,500 millimeters a year) from the moisture- River in Zambia and Zimbabwe. Smaller patches laden onshore winds. The forest canopy reaches beoccur in westem Zimbabwe, in Matabeleland, and tween 30 and 50 meters and is floristically very rich. along the higher slopes of the Sabi cathdment, extend- The main canopy trees are Albizi spp., Celtis spp., Ficusspp., and Mous spp. The area is used extensively ing into Mozambique. The class covers 22,606 square kilometers of for coffee cultivation, with the coffee grown under the Zimbabwe. In Mozambique, it occurs extensively in shade of the large canopy trees. the north, particularly along the Luganda River and The woodlands in this category in eastem Moxico along the northern coastal areas of Cabo Delgado and Province are ecologically and floristically related to Nampula provinces. It also occurs on the east-facing those of southern Zaire and western Zambia. These slopes in Manica and Sofala, with one smal occur- woodlands are highly variable and responsive to soil rence on the coast near Inhambane. It covers 10,861 depth and quality. On well-drained soils, various types of "Miombo and Cnyptosepalum Woodland" desquare kilometers in Mozambique. The phenology indicates that fairly high Nvi values velop. On the thinnest soils, Brachysfegia macropkylla (0.42 to 0.54) are sustained from October to June with and B. utilis dominate. Closed woodlands of B. spiciapeakof approximately 0D4inNovember.Thelowest -fiis also exist, and thicker, well-drainedsoils exhibit vtvi values (026) occur in August and Septenber, either open "Marquesia calonerus-uapaca pilosa Woodincreasing rapidly in October (figure 13-7). This re- land' or, more commonly, dense "Cryptosepalum maraflects the high rainfal, usually greater than 1,000 mil- viense Woodland." All of these woodlands bave high limeters with a summer maximum, and a limited dry canopies of 15 to 20 meters, but they vary considerably period in the winter. Chidumayo (1987) divides the in canopy openness and understory vegetation. On miombo woodland of Zambia into wetter and drier the poorlydrained soils, the height of the woodland is types at about 1,100 milimeters rainfall a year. less. It is dominated by Bracdystegia boehmii, Marquesin In Angola, these areas were classed by Erc Founda- kialngensis, and Uapaca spp. In many places, however, tion (198) and Millington and others (1989) as "Dense this low woodmand develops into a grassland with Medium Height Miambo Woodland." It differs from very poor fuetwood potential. Evergreen Woodland Mosaic (Class 71) slightly in I' Z am'bia, Wet Miombo Woodland is generally a seasonality and in less overall productivity, although two-story, closed, semi-evergreen woodland. The it is denser and wel wooded, with similar species main upper-story dominants are Bracstegia spp, Isocomposition. It has a well-defined structure and is berlinia spp., ul bemardia spp., and Pteroacorpus angolen-

173 Soul/tmr Afrco. 161 Outliers of the class exist in Mozambique, in Tete (1967), which is describedin more detail in Mllington and Niassa Provinces in the north, in the extreme and others (1989) This area differs from that farther northeast, and along the coast. of Inhambane and north because the trees are related to Open Woodland Gaza, especially near the mouth of the- impopo. (Class 61), with which this class interdigitates in the Other outliers occur on the border between Mozam- south. The most common tree from Open Woodland bique and Zimbabwe in Inyanga National Park. Still (Class 61) is Burkea africana, and generally, the Everother small outliers occur in Malawi (,635 square green Woodland Mosaic here is dominated by Idlometers) and in parts of Northern Province in Zam- Brachystegi bakevkna, Cryptosepulum pseudotaxus, Erybia (6,060 square kilometers). thtrophleurn afrlcanum, Guibourtia colcosperma, and Evergreen Woodland Mosai suffers little, if any, Isoberlinia spp., especially L. baumii. The canopy beseasonal moisture stress This. may be the result of comes more open on the shallower soils of the Bie reliable precipitation and local soil moisture condi- Plateau and as the Open Woodland (Class 61) bo the tions. Its "evergree" nature, however, is more likely south Is approached. the result of understory leaf growth during the short In Mozambique, the Evergreen Woodland Mosaic periods when upper-canopy trees are relatively leaf- is best developed adjacent to the coast in Zambezia less. This "evergreen behavior is illustrated in the and Sofalaprovinces. The coastal plain isflat and often phenology.photosynthetic activity is high throughout marshy, with large areas periodialy flooded. Litle if the year, mostly varying between mmvt values of ap- any seasonal moisture stress occurs. The canopy proximately 0.35 and High productivity is sus- reaches to more than 15 meters, but more commonly tamedbetweennovemberandjulywithabriefperiod is about 10 meters and is underlain by a layer of low. of lowproductivity between August and October (fig- trees (5 to 10 meters high) and a poorly developed ure 13-8). herbaceous layer. The canopy is dominated by In Angola, this class was termed Dense High Brachystegiaf m and Juiberiar globufiam, but Miombo (sc Foundation 1987; tmingon and others the whole woodland is floristically very rich. 1989). Ithas a very well defined strucure and is flors- The coastal forest belt is characterized mainly by a ticady richl Ihe canopy, which often extends to 25 dry, deciduous forest on the lowland plains, with meters, is dense and often closed, although not al-. moister evergreen and semideciduous forest on the ways. It is dominated by Bractystegia kruna, Crypto- windward mountain slopes. The evergreen and semisepaiwn exliatum, Dialum engemnwn, Guib&ta col- deciduous component of the coastal forest complex is eospma ss. pseudotam, and Julbernardia paniculata. grouped in this land cover class. Much of this forest Few other canopy trees occur, but stunted trees grow has now been converted to low woodland (MIliington to between 5 and 10 meter This lower layer is domi- and others 1989) with an upper canopy of 10 to. 12 nated by Baphia massaensis, Copaftm bawniana, Dios- meters and a lower canopy of 5 to 7 neters and is pyros spp., Paropsia brwna, PaHnari spp., Tichilia dominated by Acacia nigrescens and Albizia spp. In qwdrensis, and Xylopia odoratissi ma, many areas it has become so degraded that it now In addition to thuse two tree layers, a very poorly fornis a tudketike vegetation with shrubby species developed grass and herbaceous gromd layer exists. varying between 3 and 7 meters in height and emer- Detailed data on the composition of Evergreen Wood- gells of up to 10 meters. landmosaicinkuando-kbangodistrictare available Not all of te lower forest and thidket can be attrin the forestry feasibility study carried out by Coelho uted to degradation, because low, open vegetation occurs naturally on sandy soils of low water-holding hgure 13-8 mmv Profiles, Evergreen Woodland capabilty along the coast (for example, Inhambane Mosaic (Clasc 71) Province). Scattered pockets of higher, nondegraded coastal forest still exist, although much is degraded These forests have a canopy of about 20 meters, with -_S 05- \ X-, emergents reaing 40 meters. The main species are Crossoneh&s (Melanodiscs) oblgs, La= szyne- 0.4 / tonji, and Maranthes goetzniaa In Manica and Sofala provinces, and on the wind- S \ //. ward slpes of the Macondes Plateau, the moister forest is semideciduous rather than evergreer It has a V.-- canopy of about 15 to 20 meters with emergets up to - 71 CCII - -A. 35 meters, the mean height of the larger frees being - 71 Wzambiwe wm about 25 meters. The main canopy species are -- Jan Fcb Mar Apr May J., Jul Aug Sep Oct Nu Do Eryhropkwn surneolens, NeaJtora buchananj, and Pachystela breipe A well-developed shrb layer con-

174 SoLin(trer Africa 161 Outliers of the class exist in Mozambique, in Tete (1967), which is described in more detail in Millington and Niassa Provinces in the north, in the extreme and others (1989). This area differs from that farther northeast, and along the coast of Inhambane and north because the trees are related to Open Woodland Gaza, especially near the mouth of the Limpopo. (Ciass 61), with which this class interdigitates in the Other outliers occur on the border between Mozam- south. The most common tree from Open Woodland bique and Zimbabwe in Inyanga National Park. Still (Class 61) is BIirkca africana, and generally, the Everother small outliers occur in Malawi (2,635 square green Woodland Mosaic here is dominated by kilometers) and in parts of Northern Province in Zam- Bradcystegia bakerana, Cryptosepalurn pseudotaxus, Erybia (6,060 square kilometers). thtrophleurni africanumi, Guibourtia colcosperma, and Evergreen Woodland Mosaic suffers little, if any, Isoberlinia spp., especially L bautnii. The canopy beseasonal moisture stress. This may be the result of comes more open on the shallower soils of the Bid reliable precipitation and local soil moisture condi- Plateau and as the Open Woodland (Class 61) to the tions. Its "evergreen" nature, however, is more likely south is approached. the result of understory leaf growth during the short In Mozambique, the Evergreen Woodland Mosaic periods when upper-canopy trees are relatively leaf- is best developed adjacent to the coast in Zambezia less. This "evergreen" behavior is illustrated in the andsofalaprovinces.thecoastalplainisflatandoften phenology. Photosynthetic activity is high throughout marshy, with large areas periodically flooded. Little if the year, mostly varying between NDVI values of ap- any seasonal moisture stress occurs. The canopy proximately 0.35 and High productivity is sus- reaches to more than 15 meters, but more commonly tainedbetweennovemberandjulywithabriefperiod is about 10 meters and is underlain by a layer of low of low productivity between August and October (fig- trees (5 to 10 meters high) and a poorly developed ure 13-8).. herbaceous layer. The canopy is dominated by In Angola, this class was termed Dense High Brachystegia spiciformis and Julbernardia globiflra, but Miombo (EIc Foundation 1987; Millington and others the whole woodland is floristically very rich. 1989). Ithas a very well defined structure and is floris- The coastal forest belt is characterized mainly by a tically rich. The canopy, which often extends to 25 dry, deciduous forest on the lowland plains, with meters, is dense and often closed, although not al- moister evergreen and semideciduous forest on the ways. it is dominated by Brachystegia bakena, Crypto- windward mountain slopes. The evergreen and semisepalum exfoliatum, Dialium engleranum, Guibourtia col- deciduous component of the coastal forest complex is eosperma ss. pseudotaa,- and Julbernardia paniculata. grouped in this land cover class. Much of this forest Few other canopy trees occur, but stunted trees grow has now been converted to low woodland (Millington to between 5 and 10 meters. This lower layer is domi- and others 1989) with an upper canopy of 10 to 12 nated by Baphia massaiensis, Copafera baumiana, Dios- meters and a lower canopy of 5 to 7 meters and is pyros spp., Paropsia barzzeana, Parn?ari spp., Trichilia dominated by Acacia nigrescens and Albizia spp. In quadrensis, and Xylopia odoratissima- many areas it has become so degraded that it now In addition to these two tree layers, a very poorly forms a thicketlike vegetation with shrubby species developed grass and herbaceous ground layer exists. varying between 3 and 7 meters in height and emer- Detailed data on the composition of Evergreen Wood- gents of up to 10 meters. landmosaicinkuando-kubangodistrictareavailable Not all of the lower forest and thicket can be attribin the forestry feasibility study carried out by Coelho uted to degradation, because low, open vegetation occurs naturally on sandy soils of low water-holding Figure NDVI Profiles, Evergreen Woodland capability along the coast (for example, Inhambane Mosaic (Class 71) Province). Scattered pockets of higher, nondegraded coastal forest still exist, although much is degraded. 0.6 These forests have a canopy of about 20 meters, with emergents reaching 40 meters. The main species are o.s- Crossonephelis (Melanodiscus) oblongus, Lovoa swynner kn/ toii, 1 andmaranthesgoetzeniana. In Manica and Sofala provinces, and on the wind- > 03- ward slopes of the Macondes Plateau, the moister forest is semideciduous rather than evergreen. It has a 02- canopy of about 15 to 20 meters with emergents up to 71 Centl Angola 35 meters, the mean height of the larger trees being -71 Mozanbique coast about 25 meters. The main canopy species are o- Jan Feb MarApr May Jun JuE Aug Se Oct Nw D Eryropzhleum suaveolens, Newtonia buchananti, and Pachysteta brevpes. A well-developed shrub layer ccon- p yj.m~g 'pw o e -Ja Fc.

175 162 Estimuling Woody Biomiass in Sub-Sn hnmn Africa tains Albizia adianthlifolias the main species. Varia- resource. Representatives of the class in the Chamtions to this general pattern occur, particularly around beshi Valley remain an important source of fuelwood. the Zambezi Delta, where "Hirtella Forest" develops as a response to greater precipitation (1,200 to 1,400 Class 72-Cultivation and Forest/Woodland millimeters) and a high water table. "HirtelIa Forest" Mosaic is a mosaic of Miombo Woodland and Semideciduous Ptelcopsis-Erythmopileutm Forest, and is dominated by This class covers nearly 3 percent of southem Africa, Hirtella zanzibarica. Rice and sugarcane are grown ex- an area of 163,250 square kilometers. It occurs mainly tensively in these coastal plain areas. - in areas with special drainage characteristics and in- In Malawi, this land cover class coincides largely cludes considerable cultivated land. The most distincwith a mosaic of miombo woodland and cultivation tive occurrence is the Okavango Delta area in northalong the lower land adjacent to the westem shore of westem Botswana, with an associated small area on Lake Malawi, but it also includes montane areas such the Chobe River on tthe Botswana border, together as those around Mount Mulanje. FloristicaUy and covering 16,441 square kilometers of Botswana. structurally, it is similar to the miomnbo woodland de- Most other areas are less compact and scattered to scibed in Classes 66 and 67, with the canopy being the north and west The majority occur in a circle of dominated by Brachystegia boehmii, B. manga, B. patches fringing the Evergreen Woodland Mosaic stipulata, B. spiciformis, Julbernardia globiflora, J. (Class 71) incentral Angola (99,172squareldlometers). paniculata, and Isoberlinia tomentosa. B. boehmii is the The largest of these areas is in Kwanyama in southern dominant tree species in many areas. This land cover Angola, with another large area on the southeastem class is intermediate between Wet Miombo Woodland slopes of the Morro de Moco, northwest of Huambo. (Class 67) and Evergreen Forest (Class 82). Other smaller areas are associated with south- -This land cover class is also represented innorthern ward-flowing tributary streams and occur in associ- Zambia by the vegetation associated with swamps, ation with the Zambezi and its tributaries in western lakesides, and wet valley floors. Many of these areas Zambia. Small areas exist in the extreme northwest are grassland, but some are areas of swamp forest. of Zambia around Mwinilunga on the River Lunga, These are mainly three-story dosed evergreen forests so that altogether this class covers 29,193 square with a canopies readcing 27meters, dominated by flex kilometers of Zambia. Small areas occur in northmitis, Syzygium spp., and Xylopia spp. They are under- western Nairibia, associated with the ephemeral lain by a discontinuous evergreen understory of 9 to drainage of the Omatako (9,801 square kilometers). 18 meters and a dense evergreen shrub layer which Other areas are on the Zambian-Zimbabwean borreaches4.5meters.theforestflooriseitherbareorhas der on the shores of Lake Kariba (Zimbabwe has stands of herbaceous vegetation. 2,582 square kilometers in this class) and on the All swamp forests are controlled by high ground- border between Botswana and Transvaal on the water levels and are small, varying from 1 to 120 Limpopo, near Martin's Drift (South Africa has hectares. Fanshawe (1969) estimates that only 380-5,270 square kilometers in this class). sq uare kilometers of swamp forest exist in Zambia. The phenology of this land cover class exhibits a Otherareasofthislandcoverclassoccurinvalleyfloor limited seasonality, with NDwvr peaks of.27 to 0.35 sites, notably the Chambeshi. These are floristically between February and May, followed by a slow deand structurally similar to those described in the adja- crease to below 0.2 between July and October. The cent Wet Miombo Woodland (Class 67), but with a Okavango Delta area shows somewhat greater seasomewhat different phenology. sonality, with a marked peak of 0.37 in December and For most areas in this class, growing stock is high, January, slowly decreasing to 0.12 in September. This as is sustainable yield. Overall, in southern Africa, the is followed by a sharp increase through October and esbtmated growing stock is 83 million tormes, with a November to December (figure 13-9). Moisture availsustainable yield of 3 million tonnes. The high grow- ability seems to be the key to this seasonality; this can ing stock and sustainable yield mean that, in most be influenced by the level of the water table as well as places, the fuelwood potential of these forests is great precipitation. This is the case for most areas in Angola and Mozarn- A more compact area of this class occurs in the bique, but areas in Malawi and Zambia are more vari- Okavango Delta, with small outliers in the Chobe able. In many parts of Malawi the areas are either Valey on the Namibian border, notably around the associated with lakeside cultivation or are confined to large southen meander of the Chobe River. It is partly less-accessible montane woodlands. In Zambia, al- a fringing forest with floristic similarities to the surthough they have plentiful growing stock, the swamp rounding woodland types, as well as having species forests are important only locally as a woody biomass adapted to the edaphic conditions. The class exhibits

176 Southlern Africa 163 FigUre 13-. NDVI Profiles, Cultivation and Forestf White (1983) refers to thickets dominated by Woodland Mosaic (Class 72) Brachystegiahbakera-na, somnetimes no more than 1.3 me- 0.6 ~~~~~~~~~~~~ters in height although usually taller, occurring as an -72 Angola Snulham ~ecotone between hydromorphic grasslands.and o- 72woodlands on better-drained sites. Well-developed. - 2Okovoago. Bmwwana forestinaydevelop along the largerwatercourses,and D4- most of the. tree species are deciduous for at least 2 months of the year. The floodplain of the. upper zzabzinbrteadifloeeahyerfo 0.3-~ ~ ~ ~ ~ ~ ~~~abz nl3rteadi loddec erfo 0.2- ~~~~~~~~~~~mid-february to mid-june to a depth of 2 meters or more. Few tree species can withistand such a pro nountced fluctuation in water level,-arnd the outer fringe of riparian forest, 9 to 12 meters tall, is domi- Jan My eb Jn ar Ju Ar Au Se OctNovD nated by Syzygium guiineense ss. barorsense, with an understory of RJius qua rtintiana. Some thickets in this class occur as regenerative the greatest productivity of any class ibotswana, but stages after disturbance from overgrazing, shifting is also characterized by some seasonal vrliation in cultivation, or fire. Or, they may occur on certain soils, productivity. Around the Okavango Swamp, grass- such as loamy sands that have clayey. impervious land is fringed by a belt of large trees such as Acacia layers at depth. erioloba, A. galpinil, A. xanthophoioea, Cola plwspermumn Woody biomass stock and productivity may appear mopane, Combretum imberbe, C. megalobotbys, Croton ample in this clans, but accessibility problems exist in megalobotnjs, Diospyros mespiljfonnis, Ficus sycomorus, many areas. Many areas are remote arnd, for much of and lonchocarpus capassa. The grassy Chobe floodplain the year may be flooded. Some areas are given over to has a similar fringe of woody vegetation, but some cultivation. Some areas, like Chabe and Okavango, lie areas are given over to cultivation, partly in national parks. This class is, therefore, of Although not as dlearly linked to a particular soil. miixed potential for fuetwood. moisture regime such as the above, areas with similar biomass and phenology in Angola and western Zam- Class 82-Evergreen Forest bia owe their existence to seasonal flooding. Large areas are periodically flooded by the Zambezi and its This class covers about 5 percent of southern Africa, tributaries, as well as many other rivers that drain an area of 313,906 square kilometers. It consists of two north and south from the central highlands of Angola. principal types of forest. The larger category consists During several months of the year, the water table is of high-altitude miombo woodland that extends in close to the surface, promoting a short, tufted grass- patches across Angola, Zambia, Zimbabwe, and Moland with thicket on nearby nonflooded sites; the most zambique. The smaller category consists of afforested abundant grass is Loudetia simplex. In Kuando- areas and natural woodlands in subtropical South Kubango in southern Angola, such grasslands, inter- Africa. It includes large areas of west-facing slopes on rupted only by thickets, cover large. areas. coastal mountains in Angola, stretchling from Zaire These areas often are flonistically transitional be- Province in the north to the Serra da Che-la escarpment tween "'Baikiaa and Miombo Woodlands," but struc- in Mogfhmedes Province in the south. turally and phenologically they are controlled by sea- Farther inland, this class includes. the highest sonal precipitation and flooding. Baikiaea plurijuga' areas of the country on the Bi6i Plateau to the northwoodland isdonminantinmnany areas, alternating with west of Huambo, extensive high areas in Huila Provgrassland and mixed savannas containing Acacia ince, and smaller areas in Kuan do-icubango Provsiebemana, Diospyos mespiliformis, and sometimes Aca- ince. The coverage in Angola is 82,573 square cia erioloba along the rivers. In some higher areas, kilometers. Other occurrences are in Zambia, in the however, a woody commnunity of Combretumn imberbe, upper Zambezi and tributary areas of Northwestern D. mespiliformis, and Ficus sycomnorus occur, although and Western provinces, including West Lunga Nathis is sometimes domiinated by Acacia albida (an ex- tional Park, totaling 86,420 square kilometers. It also ample is along the upper Cunene River). Along drain- is an important class in Malawi, where it covers age lines in Kuando-Kubango, savanna communities 11,751 square kilometers. occur, dominated by Acacia spp., along with islanid In Mozanbique, extensive areas occur on higher thickets of savanna species, sometimes dominated by land in Niassa Province. A few small high-altitude the palm Borassus athaiopum, on old termitaria. areas are present in Namnpula Province and on Planalto

177 164 Estimaintig Woody Bomass ill Siub-SaImmn Africa do Mavia in Cabo Delgado Province. The area of this soils enable sufficient moisture to be reserved through class in Mozambique is 76,039 square kilometers, In the dry season to allow plant growth. It also includes Zimbabwe, the most significant areas are in the higher some areas of intensive agriculture, such as tobacco parts of Matabeleland North, totaling 29,931 square kilometers. In South Africa (22,237 square kilometers) and sugarcane near Lusaka. The "Dry Evergreen Forests" of the Zambian Plaand Swaziland (2,687 square kilometers), this class is represented by high-altitude forested areas in southteau now are limited in extent but include "Parnhiri Forest," which has two main canopy trees, Painmari eastern Transvaal and Swaziland, and at the northem excels and Syzygiwn guineense. Common understory end of the Drakensbergs in northern Transvaal. trees in this type of forest are Aidia micrntha, Chryso- In the tropics, this class demonstrates a similar phenology to that of Wet Miombo Woodland (Class 67), phylluin rnegalismontanum, Olea capensis, Tabemaeinon- tana angolensis, and Teclea nobilis. All of the "Dry Everbut with slightly lower overall NDVI values. It exhibits green Forests" show signs of invasion by iniomnbo high levels from November through June, mostly woodland trees at their edges, suggesting a great ecogreater than 0.42, decreasing throughjuly and August to 0.25 in September. This behavior reflects the inlogicalfragility underpresentclimaticconditions. Dis- turbance by anthropogenic causes leads to formation creased effect of cooler temperatures at higher alti- of "Chipya Woodland." Some areas in westem Zaitudes in winter, followed by a very sharp increase in productivity through November (figure 13-10). The bia are protected reserves; for example, West Lunga National Park. subtropical areas in South Africa and Swaziland demonstrate a similar seasonal pattern but with sig- In Malawi, this class represents "Montane Rain For- est," which now is rare and occurs only on the wetter nificantly lower NDvx values. Those above 025 are eastern slopes of higher mountains between 1,200 and attained from November through April, but decline to 2,500 meters. Mean annual rainfall usually exceeds 0.12 in August and September (figure 13-10). 1,250 millimeters, and anydryseasoneffects are coun- I Angola, the vegetation in these areas is similar to tered by mists. The. forest is well structured, with that described as Wet Miombo Woodland (Class 67) canopy species reaching a height between 24 and 45 and Evergreen Woodland Mosaic (Class 71), but with meters. A mniddle tree stratum, between 6 and 15 meslightly less overall productivity. This is caused ters, forms a dense closed canopy with a poorly develmostly by the effect of higher altitude on growth, but opea herb and grass layer. The rich flora includes in some areas by varying degrees of interference. The- Aningeria adolft-friedericii, Chrysophyllum gorungoclass includes, for example, Pare Nacional da Kisama sanum, Colagreenwayi, Diospyrosabyssinica,Myriant bus to the southeast of Luanda, parts of Parc Nacional da holstii, Odcna hoistii, Olea capensis, Prunus africana, and Biknar in Huila Province, and high, less-accessible Syzygiumguineense. areas such as those on the west-ftcing Serra Upanda Apart from the general types of moontarne forest in and the Serra da Chela escarpment In Zambia, this class is again miombo woodland, Malawi, some forest areas are dominated by a single tree species. On the Nyika Plateau, pure stand forests with vegeta ion types similar to those of woodlands of muziperus procera and Hagenia abyssinica exist. The described in Classes 67 and 71. Here, however, the class also includes parts of the "Dry Evergreen Forest" J. procera forest is formed on slightly drier slopes be- tween 1,800 and 2,900 meters altitude and reaches 30 (Er Foundation 1987), in which moisture-retaining meters in height. Some evidence suggests that it is controlled by fire. The Hagenia forest has similar alti- Figure NDVI Profiles, Evergreen Forest tudinal and precipitation controls and forms a canopy (Class 82) 8 to 15 meters tall on the forest margin, but it is not fire tolerant. 0.6 On the Mulanje Massif, between 1,525 and 2,135 meters altitude, forests of Mulanje cedar (Wid dringtonia cupressoides) exist, 25 to 40 meters high. The / forest reserves in Malawi are mainly reservations of 0.4/ existing areas of "Montane Rain Forest" or doseda 0.3- canopy woodland on the lowlands- Locally important z - exotic timber plantations exist, mostly of pine. Some 0.2o of these areas are classified in this land cover class; - 2 Swazaand -~i~ - -otherwise they are classified in Classes 67 and l 82 Wcstern Lia In Mozambique, this class occurs mostly in Manica :2 Wester Zambia Province on the Zimbabwean border and extensively Jan IFeb N;r.Ap, any lun Jul Aug Scp Oct Nov Dec inniassa Province (for example, near Lichinga). Afew

178 Soutizena Africa 165 outliers exist at higher altitude near Nampula and on western Swaziland Highveld, extending across the the Planalto do Mavia in Cabo Delgado Province. The border into South Africa near Piet Retief. The plantashort dry season often is offset by mists, because the tions are domcinated by pines, in particular Caribbean areas of forest all occur above 1,200 meters. imports such as Pinus elliottii, P. patula, and P. taeda, They range from well structured to poorly struc- which are mainly destined for South African pulp tured and are floristically diverse, including conifers mills. Some of the southern SwaziJand and South Afas well as broad-leaved tree species. In the high-altitude rican plantations are of gum, especially Eucalyphts areas, where precipitation is greater, a well-structured granolis and E. saligna, of black wattle (Acacia mearnsii), forest can form with a well-developed, not dense can- or of poplar (Populus deltoides). opy between 14 and 30 meters high. Emergents reach Black wattle has been an important source of tannin, heights of 25 to 45 meters, with a typical height being but because demand has declined, some areas in Swa- 30 to 38 meters. A lower, dense, small-tree canopy ziland have degenerated into "Wattle Jungle" (Miloccurs between 6 and 15 meters, and beneath the tree lington and others 1989). Eucalyptus plantations have strata is a shrub layer of 3 to 6 meters and a sparse grass expanded, however, the wood being extensively used and fem herb layer. The main tree species are Ocotea for poles and mining timber. In Swaziland, the conifspp., Podocarpuspp. (especially P. latifolius), Prunus erous plantation area has expanded steadily during africana, and Xymalos monospora. the past three or four decades. Timber statistics for In many areas, the montane forests are nov, much 1982 show that the ratio of planted to natural forest is shorter than they were, and commonly form canopies 45.5 to 1 (Millington and others 1989). Natural forest between 8 and 15 meters. This height reduction usu- consists mainly of Cussonia umbelifera, Podocarpus ally is accompanied by a significant influx of second- latifolius, Rawsonia lucida, and Xymalos monospora, usuary deciduous species, which take on a low-tree-and- ally orly on sheltered slopes. shrubby habit. Large proportions of this land cover Woody biomass growing stock and sustainable class in Mozambique are reserved, thereby restricting yield are ample in most parts of this land cover class, accessibility. but many places do not have any exploitable fuel- En Zinbabwe, the montane vegetation is complex wood. In parts of Angola, Zimbabwe, Mozambique, and forest is rare for two reasons. First, the Zimbab- and South Africa, this class is in various reserves and wean Mountains are generally lower than elsewhere thus is generally inaccessible. In most of Swaziland, in the region, and consequently stunted ericaceous and over much of South Africa, the plantations are forests exist only in the Chimanimani Mountains. Sec- inaccessible for fuelwood exploitation, except for the ond, the wet, windward slopes of the "Zimbabwean" "Wattle Jungles" of Swaziland, which provide a valu- Mountains are in Mozambique, for example near In- able local wood resource. It Zambia, much of this class yanga National Park, and only the dry leeward slopes is in ecologically marginal situations (Erc Foundation occur in Zimbabwe. Consequently, much of this class 1987) anci any exploitation mnay lead to a degraded consists of otherwoodland forms (see Class 61) rather "Chipya Woodland." Generally, many of the montane than true montane forest. forest areas are at high altitudes and far from the Above the miombo woodland or on wetter sites at centers of population. similr altitudes, montane forests do occur. In these This class has an estimated growing stock of 1,877 forests, trees rarely exceed 15 meters, but a canopy million tonnes, with a sustainable yield of 156 million more commonly forms between 9 and 13 meters. tonnes. It is about 4 percent of the woody biomass Below the canopy, a number of other smaller trees resource of southern Africa. exist, varying in height between 3.5 and 9 meters. The most common species are Agauria salicifolia, Aphloia Land Cover Class Tables inyrtiflora (theiformis), Apodytes dimidiata, Cussonia spicata, Diospyros whyteana, Dombeya enfthroleuca, Hagenia Tables 13-1 through 13-10, beginning on page 167, benguelensis, Pexmitis,Junmixrusprocera, Kiggelariaafiic- present summaries for each land cover class of the one, Myrica salicfqlia, Nuxia congesta, N floribunda, Olinia area, showing growing stock and sustainable yield for usambarensis, Philippi benguelensis, Pittosporum zri-- the southern African nations of Angola, Botswana, diforum, Podocarpus latifplius, Prunus africana, Ptero- Lesotho, Malawi, Mozambique, Namibia, South Afcelastruspp., Rapanea spp., Trichocladus ellipticus, Wid- rica, Swaziland, Zambia, and Zimbabwe. dringtonia nodyifora, and Xymalos monospora. Parts of this land cover class in Zimbabwe are reserved. References In Swaziland and South Africa, the majority of this class exists in plantations, although there are a few Every efforthas been made to facilitate access to the natural forests. They are exemplified by those of the documents listed here. Some documents, however,

179 166 Esltinrating Woody Biom0ss in SblSa)mra,i Africa lack fuli bibliographic information because it was Rutherford, M. C "Ka roo-fynbos Biomass Along unavailable; also, some documents are of limited anelevationalgradientinthewestemcape.'"bothalia circulation. 12(3): Scheepers, J. C "The Plant-Ecology of the Kroon- Acocks, J. P. H "Veld Types of South Africa." stadandbethlehemareasofthehigh VeldAgricul- Memoirs of tire Botanical Survey of Soutlh Africa-28: tural Region." Ph.D. diss., University of Pretoria, South Africa. Acocks, J. P. H "Veld Types of South Africa." Stomgaard, Peter "Biomnass Estintation Equa- Memoirs of tile Botanical Survey of South1 Africa 40: tions for Mionmbo Woodland, Zambia." Agroforestry Systems 3(1):3-13. Chidumayo, E. N "Species Structure in Zambian Stomgaard, Peter "Early Secondary Succession in Miomibo Woodland." Journal of Tropical Ecologly 3(2): Abandoned Shifting Cultivator's Plots in the Miombo of South Central Africa." Biotropica 18(2): Christopher, A. J Soutih Africa. London: Longman. Taylor, H. C "Fynbos." Veld and Flora 2: Coelho,H.V.P 'ZonagemflorestalnoDistrictodo Taylor, H. C "Capensis." Chapter 8 in M. J. A. Cuando Cubango." Agronomicas Aigolana 26:3-27. Werger, ed., Biogeography and Ecology of Southern Edwards, I. D "A Quantitative Description of an Africa. The Hague: Junk. Area of Savannah Woodland at Nichira Mountain van der Merwe, J. H National Atlas of South West Conservation Area, near Blantyre." Forestry Research Africa. University of Stelienbosch, South Africa: In- Institute of Malawi, Lilongwe.' stitute for Cartographic Analysis. ETc (Education andtrainingconsultants) Foundation. Weare, P. R., and A. Yalala "Provisional Vege Wood Energy Development: Bioinass Assessment, tation Map of Botswana." Botsrwana Notes and Rea Sktdy of the SADcc Region. Leusden, Netherlands. cords 3. Fanshawe, D. B mthe Vegetation of Zambia." Werger, N. J. A Biogeography and Ecology of Forest Research Butletin 7. Southern Africa. The Hague: Junk. Millington, Andrew C., John R. G. Townshend, Pam Werger, M. J. A., and B. J. Coetzee "Me Sudano- A. Kennedy, Richard Saull, Steven D. Prince, and Zambezian Region." Chapter 10 in M. J. A. Werger, Robert Madams Biomass Assessment. Woody ed., Biogeography and Ecology of Southern firica. The Biomass in the SADCC Region. London: Earthscan. Hague: Junk. moll, E. J., and L BossL "Assessment of the White, F "The Vegetation of Africa." Natural Extent of the Natural Vegetation 3f the Fynbos Biome Resources Research Series 20. Paris: UNESCO/lArPATI in South Africa." South African Journal of Science 80 unso (United Nations Educational, Scientific and (August): Cultural Organization/Association pour l'etude MolL E.J., and Id. L. Jarman "Clarification of the Taxonomique de la Flore de l'afrique Tropi- Term Fynbos.n South African Journal of Science 80 cale/united Nations Sudano-Sahelian Office). (August): White, F., and E. J. MolL "The Indian Ocean Rutherford, M. C "Notes on the Flora and Veg- Coastal Belt" Chapter 13 in M1. A. Werger, ed., etation of the Omuverume Plateau, Mountain Biogeography and Ecology of Southern Africa. The Waterberg, South West Africa." Dinteria Hague: Junk

180 Sooithe ni AfrIca 167 Table Land Cover Classes-Angola (Southern Africa Region) - Aren- Growin stock SusInilmableyield Tiousasnd. Tiwusand tomines Land couer clanss km 2 Percelnt lionnes Pement per ear Pcrceui 0 20, II 22, , ' 23, , , U , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,503, , , ,787, , , ,667, , , , , , , , , , , , , , , , , A6 87 1, , , , , , Total 1,213, ,64B, , (Percentage of region) (20.45) (30.24) (26.90) Note: In the following tables, details may not add to totals because of rounding. Source: Authors' calculations from data bases derived fron land cover classification and table 4-1.

181 168 Eslimating Woody Biomass in Sub-Salmran Afric Table Land Cover Classes-Botswana (Southeem Africa Region) Are Growinjgstock S-slainablcyicld Thousatnd Titousand tonses Land cocr class km 2 Parcenl lannes Percenl peryear Percent 0 29, ,00 0, , , , , , , , , , , , , , , , , , , , , , , ,012 16A9 216, , ,012 16A9 216, , , , , , , , , , , , , , , , , , Total 588, , , (Percentage of region) (9.91) (2.48) (3.80) Source: Authors' Calcu.dtions from data bases derived from land cover classification and table 41. Table Land Cover Classes -Lesotho (Southern Africa Region) Area Growinig stock Sustainableyield Tihousand Thousaond tonnes Land co ecl'ss km 2 Percent tonnes Percent peryear Percenf 11 25, , , , , , , , , , Total 34, , (Percentage of region) (058) (0.05) (0.08) Source: AuthorS calculatons from data bases derived fron land cover dassification and 'able4-1.

182 Souwherm Africa 169 Table Land Cover Caasses-Malawi (Southern Africa Region) Area G QWti, slock -Sustainnble yield -Tousand Thousand lonnes Land cover class km Percent Lonns Percent per year PeceCCt : , , , , , , , , , , , , , , , , , , , , , , , OAI , , , , , , , , , Lakes 20, Total 129, , , (Percentage of region) (2.18) (2.65) (2.72) Source Authos' cakubtions from datw bases derived from land cover dassification and table j4

183 170 Esimnaling Woody Bionuss itn Sb-Snl:arani Africa T-ible Land Cover Classe--Mozambique (Southern Africa Region) Area Growing slack Sutstainable yield Thnousand Tlaousaizd fauines Lnnd cover class km - Percnt lanncs Percent peryear Percent 11 6, , , , , , ,8B , , , , , , , ,03B , , , , , , , , , , , , , , , ,006, , , ,224, , , ,307, , , , , , , , , , , M , , , Lakes 4, Total 767, ,000, , (Percentage of region) (12.93) (21.42) (20.10) Source Authors' calcubtions from data bases derived from land cover classification and table 4-1.

184 Southern Africa 171 Table Land Cover Classes-Namibia (Southem Afnca Region) Area GCwing slock Sustainable yield Lantd cover clanss ku Percenti Tliousand tonnes Percent riouisand tonnes peryear Percent 0 221, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Total. 816, , , (Percentage of region) (13.76) (4.49) (534) Sourw Authors' calculations rom data bases derived frxn land cowver dassification and table 4-1..*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ r:'>: , r

185 172 Estimating Woody Biomass in Sub-Salarmn Africa Table Land Cover Classes-South Africa (Southern Africa Region)._ Arcom - -. GCrouii! stotc Sustainableyield T-houisand. -Touisatnd tonnes Land cover class km 2 Percent lonnes Percent per year Percc , , , Z , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , : 19, , , , , , , , , , , , , , , , , , , , , Total 1,220, ,664, , (Percentage of region) (2058) (8.91) (12.01) Scurce Authors' calculations from data bases derived from land cover classification and table 4-1.

186 SoUI1hen1 Africa 173 Table Land Cover Classes-Swaziland (Southem Africa Region) Area Crowing stock Suslainabie yield Land cover class kn 2 Percent *- Thousand tonnies Percrnt 7Thousand.lonncs peryear Percenit * , , , , , , D , , , , , , , , , , , Total 17, , , (Percentage of region) (0.30) (0.34) (0.45) So2J Authors' calculations from data bases derived from land cover dassification and table 4-1. Table Land Cover Classes--Zambia (So2them Africa Region) Araa Growing stock Sustainableyield -whousand Tousand tonnes Land cover class bn 2 Percent tonnes Pecnt peryear Percent 11 1, , , , ,1O , , o.m 43 55, , , , , , , ,328.6B , , , , , , ,918, , , ,225, , , ,357, , , ,108.7B , , , , , , , , , , Lakes 10, Total 747, ,046, ,36iŽ, (Perettage of region) (1259)- (21.66) (20.2) Source. Authorscakulations from data bases derived from land cover classification and table f. X ' ' X-. ~ ; 0 -

187 174 Estimating Woody Biomass in Sub-Sulhamn Africa Table Land Cover Classes-Zimbabwe (Southern Africa Region) Area Growingstock Stustainableyield Land cover class -km,, 2 Percent 7housand tonnes -7tousan1d Percent tonnes peryear Pcrcent 11 1, , , , , , , , , , , , A3 1,333A , , ,343A , , , , , , , , , , , , , , , , ,043, , , , , , ,997.D , , , , , , Lakes 6, Total 39B, ,446, , (Percentage of region) (6.72) (7.75) (7.77) Source Authors calculaions from data bases derived from land cover cassification and table 4-].

188 Glossary - - acheb Local West African term forherbaceous vegetation growth following rainfall in arid areas. - alfisol(s) A relatively productive, and therefore agriculturally important, type of soil commonly occurring in West, East, and southern Africa; characterized by downward movement of water, day minerals, and cations. anthropogenic Vegetation that has formed or is controlled mainly by human activity (for example, anthropogenic savannas and anthropogenic grasslands). See also DERavD SAVANNA. arborescent Archaean complex Having a treelike form..precambrian basement rocks; some occur in various parts of Africa. See also PRECAMBRIAN. basement complex See ARCHAN COMPLEX. base-rich Soils that have high cation content (for example, caldum, magnesium, potassium) and are, therefore, usually highly productive. boliands The region in central Sierra Leone where the landscape is domlinated by bolis. bolis SeasonalLy flooded, low-lying areas in central Sierra Leone. The edaphic climax vegetation of such areas is grassland and swamp forest, but they often are cleared for rice cultivation. They are somewhat similar to the dambos of East and central Africa bush fallowing A farming system common in humid tropical Africa in which forest or woodland is cut, dried, and burned, then cultivated for one to four years and allowed to revert to a fallow of secondary woody vegetation. See also CHmvENs calcrete A sedimentary pedogenic deposit composed of rock fragments cemented together by calcium carbonate. Cape Domain A regional center of endemism covering the southwestern and southern part of Cape Province, South Africa (after White 1983). Capensis Realm The geographic area recognized as the Fynbos Biome, equivalent to Cape Floristic Region or Cape Floral Kingdom (after Werger 1978). chipya A local Zambian name for a variety of woody vegetation regrowth. chitmene A type of bush-fallowing farming system practiced by the Bemba in Zambia. See also sush FALLowING. cirrus cloud A high-level, white, wispy cloud occurring at an altitude between 5,000 and 13,700: meters. commercial wood Wood useful in construction or manufacturing, which gives it a value such that it is not used as firewood. Cretaceous Geologic period extending from 65 million to 135 million years ago. 175

189 176 Estimnaliiig Woody Biomnss in Slib-Shllarai Africa dambo dbhderived savanna edaphic elfin thicket ephemeral ephemerals erg ericoid fire resistant fire tolerant -Jnbos A southem African term for a seasonally waterlogged streamless hollow with vegetation differing from the surrounding woodland or savanna, commonly occurring in East, central, and southem Africa. Diameter at breast height: a measurement of tree girth at 1.3 meters above the ground. Savanna vegetation formed and sustained by human activity, resulting from disturbance of a different type of vegetation, usually forest. See also ANTHROPONDC. Vegetation that is adapted to, and controlled by, the soil and groundwater conditions of a site; for example, edaphic grasslands. Thickets of stunted and gnarled woody vegetation found at high altitudes. Vegetation which grows only occasionally in response to favorable conditions, usually wet. Common in arid and semiarid areas. Ephemeral plant species. An arid (desert) landscape consisting of extensive sand cover. A descriptive tenn for plants having a shrubby growth habit, like heathers. Plants with thick bark or other adaptations which enable them often to survive fire. Plants with a limited tolerance to fire, better able to survive fire than nontolerant plants, but less able to survive fire than fire-resistant species. A Mediterranean evergreen sclerophyllous shrubland occurring in South Africa. gallery forest A dense growth of tropical forest that follows the course of a river. - i-.z A local term for the flush of grass-dominated vegetation that grows in the Libyan Desert following rainfall events. ground resolution element The smallest area of the Earth's surface that a sensor can resolve. For each ground resolution element, one measurement of reflection per channel occurs on a sensor. growing stock Air-dried, above-ground woody biomass, expressed in tonnes per hectare. Guineo-Congolian Domain A regional center of endemism found as a broad band north and south of the equator from the Atlantic, through the Zaire Basin, to the western slopes of the dorsale du Kivu (after White 1983). halophyte Hammer-Aitoff Conic Equal Area Projection hardpan(s) hanattan high forest illite isenga iligi thicket -Kalahari Sands kaolinite (kaolinitic) A plant that is adapted to and favors growth in soils with a high content of soluble salt. The map projection used in the construction of AVHRR NDV data. A subsurface soil horizon that is often hard and rocklike. In the humid tropics, hardpan often forms by precipitation of iron and aluminum, which then harden to form a ferricrete (laterite). Hardpans usually form impermeable horizons in the soil A dust-laden wind blowing off the Sahara from December to February. Closed-canopy climax forest A clay mineral derived from the weathering of mnicas and feldspars under alkaline conditions. Sand patches on the Zambian Plateau. They lead to the growth of a distinct form of miombo woodland. AnEast African vegetation type containing many deciduous woody species which form a dense, intertangled thicket. Areas of Quatemary aeolhan sands from southem Africa northward into southern Zaire. A clay mineral produced by the weathering of feldspar-rich rocks under add conditions,particularly common in humidl tropical soils.

190 Classarry 177 Karroo Domain kaukau wld little dry season macchia mguba miombo mnistbelt montmorillonite munga Namaland Domain Namib Domain occult precipitation piosphere Precambrian proteoid qoz Quatemary raster reg restoid Ihe area of summer rainfall east of the Westem Cape Domain and south of the Namaland Domain (after Werger 1978). A veld grassland type occurring on the Namnibia-Botswana border (Kaokoland). A. period of approximately 4 to 6 weeks, usually in August, during the wet season in West Africa when rainfall amounts decline markedly. The phenomenon is related to the northward movement of the Intertropical Convergence Zone (rrcz) and is restricted to the southerunost parts of th West African coast. A term used in South Africa to describe heathlands and evergreen scierophyllous shrubland. A local East African name for a VImRsOL An open woodland land cover occurring in southern Africa. An area characterized by oultprecipitamton. A clay mineral derived from the weathering of basic rocks. Montmorillonite-type clay minerals absorb moisture and swell during wet periods and contract during dryperiods. Open, parklike savanna woodland occurring in Zambia. A region of southem Africa covering the narrow escarpment belt inland of the Namib, broadening southward to include the southern Kalahari Plateau (after Werger 1978). A region of southern Africa comprising the arid coastal strip of Namibia between the Atlantic and the escarpment (after Werger 1978). Precipitation that occurs directy on vegetation in the form of condensation. It commonly occurs when humid onshore winds are forced to rise up high escarpmments along the coast. An area of degraded soil and vegetation around a waterhole or wells, occurrng in arid and semiarid areas. The entire span of geological time before the start of the Cambrian Period at approximately 600 million years before the present A term applied to plants thathave leaveswhich are thickened, as in the genusprotea. Stabilized wind-blown sands of Quatenary age in Sudan. The current geological period, which began approximately 1.8 nillion to 2 million years before the present. A means of representing spatial data in the form of a grid. An and (desert) landscape consisting of extensive tracts of gravel. Having a low, cushionlike form similar to that of the southem African genus Restio. sclerophyllous A term describing evergreen shrubs and trees thathave adapted to lengthy seasonal droughts by prolucing tough, leathery leaves. senescence Ripening and/or dieback of vegetation seral succession Sequential development of a plant comnmunity or communities over time. shalahaila - A term used in Lesotho to describe shrub woodland growing on overgrazed, degraded pasture. hiftingcultivation See ish FALLowncG. stem volume Volume of main woody stem derived from height and diameter measurements, expressed in cubic meters. st2r A stacked volume of wood, expressed in cubic mete! subcanopy The layer or layers of vegetation beneath the canopy

191 178 Estimatiing Woody Biorniss iin SUb-SalIarmt Africa succulent suffrutex sustainable yield : swvartveld Tertiary thionic fluvisols tonne veld vertisol vloere wadi wattle jungle Western Cape Domain Wild Coast Woing Dega woodfuels woodybiomass Zambezi teak ZambezianDomain Zone de l'igname A plant with thick, fleshy leaves or stems that retain large amounts of moisture; common in arid areas. An understory shrub. Mean annual increment of air-dried, above-ground woody growth, expressed in tonnes per hectare per year. A gently undulating landscape in Cape Province, South Africa, extensively cultivated for wheat. Geologic period exteniding between 2 million and 65 million years before the present. Alluvial soils, occurring mainly in estuaries, in which subsoil sulfuric acid builds up when the soils are dried. Metric ton; 1,000 kilograms or 2,20462 avoirdupois pounds. (1) A southem African term describing almost treeless grassland. (2) A southem African term used to define a region regardless of land cover (for example, the Highveld of Swaziland). A black soil commonly found in semiarid areas which, because of its high montmorillonite contani, swells during the wet season and shrinks and cracks open in the dry season. Often described as having 'cracking clays." A local Afrikaans name for brackish, low-lying areas in the Karroo. A valley in a desert that is wet only after heavy rain and will briefly support vegetation. A dense growth of previously well-managed, but now abandoned, plantations of Aca&a mearnsii in Swaziland. A region of southern Africa including the coastal strip and escarpment mountains in South Africa south of the Namib (after Werger 1978). A local name for a remote stretch of the Transkei coast between East London and Port Alfred. Local name given to the Ethiopian highlands that experience a temperate climate. Fuelwood and charcoal. Ligneous plant material. Name used by foresters for the important timber tree BaikiaeA plurijuga in southern Africa. A region extending from the Atlantic to the Indian Ocean including Zambia, Malawi, and Zimbabwe, as weli as large parts of Angola, Tanzania, and Mozambique. An agricultural zone extending across the central C6te d'ivoire.

192 Index of Botanical Names Acacia, 30,31,37,40,54-58,64,65,86-94,96-98, 105- Albizia, 124,130,148, 150, 159, 161; adiantlifolia, 129, 07,109,120-24,126,127,138-41,144-50,152, 153, 157,160,162; amiaam, 91; anthelmintica; 145, 150; 155,163; abyssinica, 86, 96;;albida, 57,58,92,107, antunesiana, 156; gummifera, 128,160; sericoceplula, 124,126,136,148, 156,163; ainythetophylla, 88, 91; 94; zygia, 76,124, 128 rambica, 87; asak, 88, 91; astringens, 92, 94; brevispica, Allanblackia stuhlimannii, ; bussei, 90, 91; campylacant ha, 123, 126; circum- Aloe, 93,124; arborescens, 147; bainesii, 157; dichotoma, nmrginata, 123; cyanophyl a, 144; cyclops, 142,144; 141, 145; eminens, 93; ferox, 138 dudgeonii, 95,96; ehrenbergiana, 87; etrenbergiana Amblygonocarpusandongensis, 152 (flava), 55; erioloba, ,141,145,146,151,153, Ancistrophyliumsecundiflorum, ; etbaica, 86, 91, 93;flava, 87;fleckii, 138; Andropogon, 58, 66,67,69,106,124,125, 129,130,137; galpinii, 163; gerrardii, 120, 122, 124,126; gayanus, 30,56-58; pseudopricus, 57; schtirensis, 105 glaucophylla (circumiarginata), 86; gourmuensis, 95, Androstachysjohnsonii, ; haematoxylon, 145; hebeclada, 137,138,146,150; Aningeria, 74,126,130; ado45iedericii, 111, 129, 130, hebecladoides, 123, 126; hererocusis, 150; hockii, 122; 164; altissima, 130 horrida, 90, 119; karroo, 138, 147,154; laeta, 55, 87, Annona senegalensis, 106,109 92; leuderitzii, 138; mearnsii, 165; melanoxylon, 144; Anogeissus, 94,95,98; leiocarpus, 57,94,95; schimpeni, melliffra, 56,86,88-91,94,120,122,138,141,144; 67,68,92 meiffrn ss. detinens, 137,141,145,146; montis-usti, Anthephora: argentea, 138; pubescens, ; nigrescens, ,161; nilotica, 55,88,90,92, Anthonothafragrans, 72 94,120, 122,147; nilotica ss kraussiana, 140; nubia, Anthospermum aethiopicum, ,89; reficiens, 90,119,120,141,144,145,150; Anthosiema senegalense, 75 reficiens ss. misera, 121,122; robynsiana, 145; satigna, Antianis toxicaria, 75, ; senegal, 55,56,87-89,91,92,97,119,120,126; Antidesma venosum, 90,95, 96 seyal, 55,56,86,88--94,97,106,119,122,126; seyal Aphioia rnyrtiflora (theiformis), 165 ss. multijuga, 123, 126; seyal ss. subalata, 94,120; Apodytes, 86; dimidiata, ,130,147,158,165; sieberana, 88, 91,123, 126, 163; thomasii, 90; dimidiata var. acutifolia, 93 tomentosa, 92, 94; toriilis, 55,86-88,90-93,97,119, Aristida, 120,122,137,138; adscensionis, 55,119; 122,138; tortilis ss. raddiana, 53,55,56,92; tortilis congesta, 138; longiflora, 53; mutabilis, 55,56, 119; ss. spirocarpa, 120; venosa, 88,91; zantkopholoea, 163; pungens, 53; sieberana, 55; stipoides, 55,56; xiphocarpa, 86; xiphocarpa (abyssinica), 98 vanderysiii, 105 Acalyphafruticosa, 125 Arundinarialpina, 99,121, 129 Acokanthera, 93, 123,124; schimperi, 93 Asparagus capensis, 143 Adansonia digitata, 57,65.66,90,121-23,148,150,156 Avicennia, 77,85,129; africana, 77; germinans, 77 Adenolobuspechuelii, 145 Afzelia, 66-67; africana, 58, 67,68,74,128; quanzensis, Baikie, 149,152,153,163; insignis (eminii), 130; 90,129,160 plurijuga, 152,153,163 Agauria salicifolia, 96, 165 Balanites, q5,97,120; aegyptiaca, 53,55,57,87,92,95; Agrostis, 123,128,129; sclerophylla, 87 maughamii, 157; orbicularis, 90; racemosa, 87; Aidia miraniha, 108,164 wilsoniana,

193 180 Eslirt7aling Woody Bionass in SSb-Salnam in Africn Baphia: burttif, 123; inassaiensis, 123, 155, 156, 161; Catophractesalexanidri, 141,144 -massaiensis ss. obovata, 153 Ceiba, 57; pentandra, 75, 76, 128 Barbeya olcoides, 93 Ceitis, 74, 130, 157, 159; africa na, 138, 147, 154; Bauldinia, 160; nacranthat, 137,146, 150; taitensis, 91 brownnii, 128; integrifolia, 107; rnildbraedii, 75, 130; Berzelia: intermedia, 158; larniginosa,.144 zenkeri, 109, 112 B1ena, 129 Cenchmus biflorus, 55, 56,88,92,120; ciliarus, 150 Bo1bax costartum, 57 Cephalosplaerra, 130 Borassus: aetluiipurn, 57, 69;flabell(fer, 57, 58 Ceraria, 141, 145; longepedunculafa, 141; narnaquensis, Ioscia, 120, 122, 124, 140, 141, 145, 146, 150; 141 albitrunca, 137, 138,140, 145, 146,150; angus Ifolia, Chaetacrne aristata, ; coriacca, 90, 123; microphylla, 150; mossambi- C nasmopodiim caudatum, 67,70 censis, 156; rclnmanniana, 150; senegalensis, 55,87, Chtionalithusfoveolatus, , 92 Cliloris: gayana, 150; prieurii, 56; virgata, 137. Boswellia, 89; carted, 93; dalzieli, 57; neglecta, 90; CIhoroplhora, 74; excelsa, 73, 109, 110, 112, 160; regia, papyrifera, 94,95 70,76 Brabeium stellatifoliumn, 144 Chrysocoma, 139 Brachiiaria brmchylopha, 75 Clirysophyllum, 72,126,130; albidum, 130; Brackylaenta, 157 gorungosanum, 164; inegalisnontanum, 160,164; Brachystegia, 108,109,124-29,148,149,155, 156,159; perpulchrum, 74 allenfi, 155; bakemrna, 159,161,163; boehmii, ; Chrysopogon aucheri var. quinquepluris, , , 156,159, 160, 162; bussei, 155;floni- Clappertoniaficrfolia, 75 bunda, 152,155,160;glaucescens, 153,155,160; Cleistanthusschleckteri, 157 gosswieleri, 152, 154, 159; laurentii, 113; longifolia, Cochlospernum angolense, , 160; macrophylla, 159; manga, 156, 160,162; Coeiachyrum: brevifalium, 54; compressa, 53 microphylla, 155, 160; spiciformis, 108,123, , Cola, 74; cordbfolia, 98; greenwayi, 164; laurifolia, 57; 148,152-56,159-62; stipulata, 156,160,162; nitida, 76 tamarindoides, 155; taxifolia, 108,153,155,160; Colophospermumopane, 136,139,141, , utilis, 159,160; wangennecana, 154, ,156,163 Bridelia, 155;ferruginea, 75; taitensis, 91 Combretodendron, 72 Bruguieragynnorrhiza, 129 Combrehum, 66,68,95,97,107,120,123,124,126,128, Buddleja: cornugata, 139; polystachya, 96; saligna, ,141,145,146,148-50, 152,154,155,159; Burkea, 152,153; africana, 58, 6466, 105, 107, 110, aculeatum, 91, 106,123; africanum, 89; apiculatum, 130, 145, 146, 148, , 155,156, ,141,150, 151,153; binderanum, 120,124, 128; Burttia prunoides, 156 collinum, 95,130,138, 150, 153, 156; cordofanum, 92, Bussea massaiensis, 123, ; dinteri, 152; elaeagnifolium, 123,128; Butyrospermum, 56,57; paradoxurn ss. niloticrum, 128; ehzeagnoides, 155; engleri, 153; ghiasalense, 120,124, parkii, 65,66,68 128; glutinosum, 56, 57; hartmannianum, 94,95; Buxus, 87,93; hildebrandtii, 93 imberbe, ,163; kraussii, 157,158; mechowianum, 123,126,152,156; molle, 88,91,97, Cadaba, 90;farinosa, ,124,128; schumannii, 129; trothae, 123; zeyheri, Cadia, 13; purpurea, ,124,128,147 Caesalpinia trothae, 91 Commiphora, 37, 40,57,89-93,107,119-24,140,141, Calanmus deeratus, ,14850; afriicna, 53,55,56,65,87,90,92, Calligonum comosum, 55 95,123,138, 150; angolensis, 138,141, 144, 150; Calotropis procera, 87 boiviniana, 89,90; caerulea, 122; campestris, 90; Canarium schweinfurthii, 74 glandulosa, 140; )harveyi, 147; hornbyi. 123; Cannamois nitide, 143 -madagascariensis, 122; merkeri, 122, 123; mollis, 122, Canthium: burttii, 160; glabrifolium, ,149; mossambicensis, 150; pyracanthoides, 146, Capparis, 65; erythrocarpos, ; schimpri, 122,150 Carapa procera, 74 Conocarus, 89; ereclus, 77 Cardenia, 66,68 Copaifeni baumiana, 159,161 Carex monostachya, 87 Cordia gharaf, 147; millenii, 130; ovalis, 120 Canissa edulis, 93,123,124 Cordyla, 123; africana, 148,156; densmy7ora, 123 Cassia: acutifolia, 87; singuena, 88, 91 Cornulaca monocantha, 53 Cassineperagua, 146 Cotyledon: paniculata, 141,145; wallichii, 143 Cassipourea, 126,130,157,160 Cmhfitd bf#tkaudata ss. burti, 123

194 Inidex of Boloniicol Nantes 181 Craterispcnmnnii laturinumn, 76 Ely! wpappus, 146; rhiinocerotis, 142, 143 Crossonephaelis (Meatinodiscu$s oblonigus, 161 Enitadafricana, 58, 66, 68 Crossopteryxfebrifiega, 155 En tadopsis abyssinica, 91 Croton, 9'5, 97,123, 157; inwgolobotrys, 163; scorciesii, Entanldrophraguw, 72, 73, 130; ongtolense, 130; 75; subgrmttissimus, 146; zamtbesicuis, 146 catidaturn, 153; cxcelsum, 130 Crudia gabonensis, 72.Eragrost is, 53,58, 124, 137, 138; pollens, 151; Cryptocar'ya latijolia, 158 trentuda, 57. Crypt osepalutm, 153, 159; exfoliatunm, 152, 155, 160, 161; Efrica, 121, 128, 129, 139, 143; arbo rea, 87 nwaraviense, 159; pseuidotoxus, 161 Eriocaplhalus: africanzus, 143; mocemosus, 144; Cteniuun: e)egans, 67; iwewtonii, 57 un:bellulatus, 144 Cunonia copensis, 144 Fry!l hrina: abyssinica, 95; socleuxif, 129; tomentosa, 113 Curtisia den Ita,a 158 Fiytlfrnophleum, 152, 153, ; africanump, 58, 107, Cussonia, 93, 138; angolensis, 159; borteri, 58; holstii, 152, 155, 156, 161; guineense, 57, 58, 70, 98; 94; lo'gissimo, 70; spica to, 165; umibellifero, 165 suaveolens, 160, 161 Cymbopogont: giganfteus, 57;?nionocan2t ha, 53; nenat us, Eucalyptus, 33, 93, 94, 98, 99, 165; globus, 93, 99; 86,88. granolis, 165; saligna, 165 Cynodon hirsutus, 138. Euclea, 93, 123, 124, 142, 153; crispa, 138; divinorum, Cynomebra: alexandri, 111, 113, 131; leonenisis, 72; 124, 125; notolensis, 157; pseudebenus, 137; macenosa, megalophylla, , 158; tomentosa, 141, 146; undulata, 141, 146, 150 Cyperuls papyrus, 88 Euphorbia, 88, 91, 93, 97, 98, , 141, 144, 145; CyphostemanurrorCIiT, 145 abyssinica, 88; balsamitpm, 56; ca-ndelabrum, 89,93, 149; conspicua, 149; dowei, 131;grondis, 94; Dalbergia, 92,94,123, 126; melanoxylon, 95. guercidiona, 145; mauritanica, 143; mobecchii, 90; Dalbergiella nyasae, 155 tiru calli, 126; trionguloris, 157 Daniellia, 58; alsteeniano, 155; oliveri, 57,58, 67,68,70, Eupharphzia: currornt, 145; eduordol, ,110.Eustachys paspaloides, 120 Delonixelata, 123 Exotheca abyssinico, 123, 129 Deschompsia, 123, 129; caespitosa, 87 Dialium engleranum, 152, 161 Faurea: sotigno, 95,96,155, 156; speciosa, 153 Dichtrostadnjs, 149; cinerea, 137,138, 140, 141,146, Festuca, 123,129, , 150; cinerca ss. afiicana, 147; glomerata, 76 Ficalhioa laurifolial 111 Didelotia unifoliolata, 72 Ficus, 91, 141, 147, 157, 159; acro carpo, 88; capensis, 75, Digitaria macroblepharno, ; cordata, 150; guerichiana,.150; ingens, 141;. Diospyros, 72, 123, 138, 161; abyssinica, 75, 164; sycomonus, 98,137,163 oustrno-africana, 142; dichrophylla, 158; glabra, 158; Flacourtia indira (flavesce ns), 75 kirkit 56~f; lydioides, 138, 154; mespiliformnis, 75, Frey/inta opposibjfolia, , 'it4toeonsis, 156; rotundifolia, 157; 444L?Ws#Whyteana,153,165 Galenio africona, 143 E1fllo/4yg4 1j$4W3' 6 ndylocarpon, 123,126,155,156 Ga/intemt, 100 bzn~~~~h4g~~~p,ti~~~s ~Gardenia jovis-tonantis, 123 Dissolts corn ~~~~~~~~~Gilbertiodendron, 113; dewevnei, 72, 106, 107, 113; Dobera, 90. ~~~~~~~~~preussi,7 D doac isco2, 9S, Xfl, 124 Gluma ivorensis, 72 Domnbeya, 123; eryt hrozeu ica, 165; goetzenii, 129; Grew ia, 90,91, 122, 123, 137, 146, 150, 157; bicolor, madiensis, 123; rotund ifo lia, 141, ; burtii, 123; ca rpinifolia, 75;fa Ilax, 120;fiava, 140, Drcacena schizantha, 93,94 141,145, 150; occidenital is, 138; villosa, 150 Drokebrockmnina somakensis, 119 Guanoa, 72 Drypetes:flonlrunda, 75; gerrardii, 157; parv4folia, 75 Guibourtia, 152; coleosperma, 152,159,161; coleosperma Duos perma enemo philum, 119 ss. pseudotaxa,l 161; demeusei, 107,112 Ecbolium, 91 Hogenia, 99,121, 130,164; abyssinica, 86,98-100,121, Eh-retia rigida, 138, 141, , 164; benguelni,165 Eke bergia, 93, 99; copensis, 157, 160 Hizre aciculanis, 144 Elaeis guineensis, 70,76,109 Halleria lucida, 99,147,153, 158 Elaeophorbia drupifera, 75 Hartogia capensis, 144 Eleusine compresa, 53, 55 Horungana mtadagascariens is, 76, 160

195 182 Estimating Woody Biomnss in Sub-Salramn Africa Heenca (Ozorea), 151; argentea, 143 Lowua szuynnertonii, 161 Helichyysum, 139 Lyciun, 140 Heritiera littoralis, 129 Hildegardia barteri, 128 Macaranga, 126,130; capensis, 160; kilimandsclarica, Hippocratea, Hirtella, 162; zanzibarica, 162 Mamnra, 91, 123; anolensis, 91; caffra, 158; crassifolia, Hohmrrhenaflonlbunda, , 88, 92; lanceolota, 147 Hymenocardiaacida, 105,106,110,153 Maesopsis, 125; eminii, 130 Hyparrhenia, 66, 67, 97, 106, 120, 122, ; Malacanthadnifolia, 128 anthistirioides, 86, 88; chrysargyrea, 75; confinis, Mangifera indica, 57,76, ; cyanescens, 58; dissoluta, 57; rufa, 86 Manzilra: obovata, 75,128; sansibarensis, 129 Hypericum, 121; revolutum, 121 Mansoniaaltissima, 74 Hyphaens, 89; benguelensis, 124; thtebaica, 57,66,87; Mapania, 72 ventnicosa, 148, 156 Maproune africana, 95, 96 Hyptis spicigera, 75 Maranthes:goetzeniana, 161; polyandra, 95, 96 Markhamia, 123; obtusifolia, 149,156 Ilex mitis, 99, 129,144,153,158,162,165 Marquesi acuminata, 155; calonerus, 159; katangensis, Ionchocarpus capassa, ; loadensis, 155; macmura, 108,109,155 Isoberlinia, 58,64,66, 94-96, 98,107-09,124,127,130, Maytenus, 153; acuminata, 143; heterophyia, 142,146; 160,161; angolnsis, 125, 154,159; baumii, 161; doka, okoides, 143, 144; senegalen t ;, 65, , 65-68,107,124; paniculatai, 125; scheffleni, 130; Melaodiscus see Crossonephelis ftmentosa, 58, 125,156,160,162 Metaksia municata, 144 Metrosiderosangusfifolia, 144 Jatropha, 120 Milettia, 126; gmndis, 157; stuhimannif, 160; lbernardia, 108, 109, 124, 126, 127, 129, 149, 153, 156, thonningii, ;gloflora, ,148, 152, 153,155,156, Mimusop: coffra, l57;fuifosa, 90; obovata, ; magnistipulata, , 129; paniculata, 152, Mitragyna, 112; cliata, 74; inermis 57; stipulosa, 74,98 154, 156, ; seretii, 72, 111, 113 Monechma, 145 luncusarabiats, 86 Monodona myristirc, 130 Iuniperus,92-94,99, 100; procera, 86,87,93,98-100, Monotes, 124,125, ; africanus, 148; kerstingii, 129,164,165 58,67,68,95,96,107 Montiniacaryophyllc, 150 Khaya, 73, 95, 130; anthoffieca, 110; grandifolia, 98; Moninga oalifolia, 145 nyasica, 160; senegalensis, 57,58,68,98,106,107 Morus, 159 Kigelka afiicana, 158 Mucronata, 137 Kiggelari africana, 165 Musanga cecropioides, 76, 106, 110, 112 Kirkia acuminata, 140, Myrianthus holstfi, 164 KIzinedoXa gobonensis, 112 Myrica, 130; salicifolia, 165 Koeleria, 123 Myrsineafricana, 146 Laguncularia, 77; racemosa, 77 Nauclea, 74; dideichii, 72 Lannea, 123, 129; microcarpa, 58; stuhlnmannii, 148, 150 Neoboutonia, 126,130; macnocalyx, 130 Lantana cama, 125 Nesogordonia papaver Ifera, 75 Lebrunia bushaie, 111 Newlonia, 130; buchananii, 124,130,160,161; Leptadenkapynotechnica, 53,55,88,92 hldebmandtii, 157 Leptospernmum: laeoigatum, 142; argenteum, 144; Nuia, 98,130; congesta, 99,129, 165;florlbunda, 158, coniferum, 144; eucalyptifolium, 158; salicifolium, 144; 165 muinii, 144 lxucosideasericea, 139 Ochna, 157; afzelii, 95; ho!stii, 164; pulchra, 138,146, Limosell afrina, ,151,153; schwveinfurthiana, 95,153 Lobelia, 121; rhynchopetalum, 87,99 Ochthocosmusafricanus, 113 Lonchocarpus, 126; nelsii, 146,150 Ocotea, 130, 165; bullata, 99, 158; usambarensis, 111, Lophira, 67, 70; alata, 70, 72; lanceolata, 57,58, 67, 70, 121,129,130 95,96,110 Oldfweldia africana, 72,73 Loudelia, 70, 130; demonsii, 105,106; phragmitoides, 75; Ole0n africana, 86, 93,98,123, 124, 138, 143; capensis, simplex, 57,70,129,163; superba, ,143,144,164; weltwitschii, 130

196 Index of Botanical Names 183 Olinic emarginata, 153; usambarensis, 96, 165; ventosa, 143 Pterocelastrus, 165; tricuspidatus, 146 Opuntia, 96,97 Pterolobium stellatum, 91 Ormocarpum tichocarpum, 125,140, 147 Pteroniaglauca, 143 Ostryoderris, 126 Pterothrixspinescens, 143 Osyris, 138, 142 Putterlickia pynicantha, 146 Oxystigma ozyphyllum, 113. Oxytenant'-era: abyssinica, 57, 58, 95, 97; borzii, 95 Ranolfia caffrn, 157 Ozoren see Heeria Rapanea, 130, 165; melanophloeos, 99 Rap/lia, 74, 112 Pachypodium, 145 Rwsonia lucida, 165 Pachystela brevipes, 160, 161 Rlhania, 143 Pandanus candelabrum, 74 Resiio cwmirumis, 143 Panicum, 122; coloratum, 137; kuahurense, 138; Rhigozum, 150; brevispinosum, 140; tridciotomum, 137, naximum, 125, 151; turgidum, 53,55,88, ; virgatum, 144, 145 Parinari, 126,130,161,164; curatellfiflia, 153, 155,156; Rhizophora, 77, 85, 129; harrisonii, 77; mangle, 77; excelsa, 70,74,108,129,130,164 racemosa, 77 Parkia, 66, 68; biglobosn, 56-58,70,107; -Rmus, 138,142, 146, 158; lancea, 137, 138, 154; marlotii, clappertoniana, ,150; pyroides, 140; quartiniana, 163; tenuinervis, Paropsia barzzeana, 153, Passerina, 158; montana, 139 Ricinodendron rautanenii, 153 Pdtophorumnafricanum, 145,146,150 Ruschitferox, 143; muletflora, 143 Pennisetum, 58, 124,129; mezianum, 120; Rynchelytrum repens, 126 pedicellalum, 57 Pentadesma lebrunii, 111 Sacoglottisgabonensis, 72 Pentschistis, 123 Salix, 154 Pentzia, 143; incana, 143 Salsola: aphylila, 137; tuberculata, 137 Petersianthus macrocarpus, 112 Salvadora, 91; persica, 137 Phaeoptilum spinosum, 144 Sansevieria, 93,120, 124 Philippia, 129,158; bengueensis, 165 Sarcocephalusdiderrichii, 112 Phoenix reclinata, 74, 112 Scaeola thunbergir, 157 Phragmites communis, 86 Schizackyrium, 58, 124; sanguineum, 58 Phylica, 143 Schmidtin, 138 Piliostigma, 58,148; thonningii, 58,70,109,155,156 Schoenefeldia gracilis, 55,56,86 Pinus. elliottii, 165; patula, 165; pinaster, 144; taeda, 165 Schrebema trichoclada, 149 Piptadeniastrumafrianum, 112 Schuma ischaemoides, 86 Pittospoium viridcflonnm, 165 Scleria barteri, 76 Pleurosty ia, 126 S-lerocarya, 126; birrea, 147; caffri, 140,141,146-48, Poa, 123, Padocpus, 93,94, lit 130,158,165; elongatus, 143; Scolopia mundii, 158 faluatus, 99,130; gracilior, 93,98,99; latifolius, 129, Scorodophloeus zenkeri, ,165; milanjinus, 121, 130; usambaren1sis, 127 Securinega virosa, 75 Polyscias, 126, 128,130 Senecio, 121 Populus, 154; deltoides, 165 Sericocomopsis, 91 Protea, 108, 138,144,146,153; cynaroides, 158;gaguedi, Sesamotlamnus, 141, ; madiensis, 95; nere folia, 142; obtusifolia, 144; Setaria, 122,129; incrassata, 86,88; welwitschii, 149 susannae, 144 Sideroxylon, 94,100 Protorhus longifoli, 157 Smihaxkraussiana, 126 Prunus afrcana, 98,99,128-30,158,164,165 Sorghum purpureo-sericeum, 86, 88 Pseudolacdnostylis naprouneifolia, 148,155,156 Sorindei madagascariensis, 90 Pseudoprosopsisfischeri, 123 Spirostachys veifera, 90 Psorospermum, 156 Spondianthuspreussii, 74 Ptaeroxylon obliquum, 157,158 Sporobolus robustus, 122 Pteleopsis, 162; anisoptem, 153; diptera, 155 Sterculia, 91, 123, 148, 156; appendiculata, 160; setigem, Pterocarpus, 148,156; angoensis, 152,155,156,159; 149; tragacantha, 76 antunesii, 148, 153; erinaceus,57,58, 67, 68, 70; Stipagrostis: pungens, 53,55; uniplumis, 53 rotundifolius, 140 Stroinbosiaschefe, 130

197 184 Estinating Woody Biomnass in Sub-SahNamn Africa Sthychnos, 123,130; spinosa, 144 Trichflia: dregena, 157; emefica, 90, 157; quadrensis, 161 Suaeda monoica, 122 Trichodadusellipticus, 165 Superbiagrandis, 93 Triplochiton sckroxylon, 74, 109 Swartzia madagascarmensis, 153 Turm holstii, 130 Syderoxylon inerme, 158 Turrueanthus afiicana, 73 Symphonia, 128;globuliJera, 74,111,112 Syzygium, 157,162; condatum, 147; guineense, 70, 98, ULapaca, 108,124,126,138,153,155,156,159; 108,123,126,155, 156, 16C, 16t; guineense ss. guinensis, 70,74,112; heudblotii, 74,107,110; afromontanum, 129;guineensess. barotsense, 163; kirkiana, 125,148,153; pilosa, 159; somon, 58; guinaense ss. guineense, 153; pamifolium, 127 togoensis, 68 Utenium elegans, 57 Tabernuemontana, 126,130; angoensis, 164; Uria dhamae, 75 johnstonii, 129 Tamarinidus, 57, 126; indca, 57,65 Vaccinium, 129 Tamarix manniram, 86 Vangueriopsis lancifzlia, 153 Tapiphyllumflonlundum, 23 Vepris heterophylla, 75 Tarchonanthus camphoratus, 93,123,124,141,146,150 Ventonia ampla, 147 Tarietia utilis, 72 Virgilia oroboides, 158 Teclea, 93, 123,124; isheri, 156; nomb7is, 164 Tennridia, 66,68,74,91,92,120,123,124,130,140, We!wiftsha bainesii, ,155; avicennioides, 58; bnzchystemnm, 153; Widdringtonia, 158; cedanrbgensis, 143,144; brownii, 95, 97, 120, 128; glaucescens, 70,95,120, cupresoides, 98, 100,143,144,164; nodjffoni, 165; 124,128; axkjlora, 107; moolis, 120,125,128; schwa rzi, 144 prunoides, 136,140,141,150,15Z, seric, 109,120, Wilkenowia srh*ta, ,13B,140,141,145,146,149-53, ; superba, 109, ; utilis, 72 Xanthocerdszambesia, 157 Tetraberlinia tubznanina, 72 Ximenia, 150; amercana, 58,140; caifra, 146 Thamnochonruserectus, 144 Xylopu, 162Waethiopica, 160; odoratissinm, 161; Themeda, 120,122,138,139; triandra, 97,120,123,129, quitasii, 76; rubescens, ,138 Xymalos monspora, 158,165 Thepesidanis, 90 Tieghemeila (Dumoria) african, 72; hockelif, 73 Zanthoxylum xanthoxyloides, 75,128 Torchonanthus camphoratus, 93 ZilIa spinosa, 86 Tragus racemosus, 55 Ziiphus, 57; abyssinica, 88,91,106; mauritiania, 56; Trema:guinensis, 76; onentaits, 160 mucronata, 137,138,140,141,146,150,152,154 Tribdus, 53 Zyguphyllunvgiqlya=ni 143; morgsana, 146

198 Index of Place Names Abeokuta (Nigeia), 69 Banikoara (Benin), 65 Aberdare Mountains, 25,121,129 Barclay East (South Afica), 139 Aberdae National Park (Kenya), 37,121 Barclay West (South Africa), 151 Abidjan (Cbte d'lvoire), 67 Ban Region (Somalia), 85 Accra (Ghana), 67,73 Barotseland (Zambia), 163 Addis Ababa (Ethiopia), 93,95,96,98,99 Bas Zaire Province (Zaire), 108,109 Adoumaoua, Massif of, 110 Batek6 Plateau, 105 Adrar des if oras Plateau, 54 Bauchi State (Nigeria), 65 AfoU4, Massif of, 55 Begemder Province (Etlhiopia), 88,92,94 Air Mountains and Plateau, 53,54 Beira (Mozambique), 27 Aksum (Ethiopia), 95 Beitridge (Zimbabwe), 146 Alantika Mountains, 107 Bendel State (igeria), 71 Alexander Bay, 136 Benguela (Angola), 149 Anambra State (Nigeria), 74 Benguela Province (Angola), 77,143 Anhanca (Angola), 141 Benue River and Valley, 66 Ankober (Etiopia), 87 Benue State (Nigeria), 74 Ars! Hills, 97 Berbera (Somalia), 89 Arsi Mounis, 86 Bethlehem (South Africa), 138 Arsha CTanzania), 123 Bethulie (South Africa), 151 Arusha Region (Tanzania), 120,121 Bie Plateau, 152, ,163 Asbestos Mountins, 139,141 Bie Province (Angola), 155,159,160 Asmera (Ethiopia), 87,88,94 Biagds Islands, 76 'Assiba Escarpment, 55 Biknar, Parc Nacional da (Angola), 164 Atcora Mountains (Chaine de latakora), 65, 68,71 Birao (Central African Republic), 106 Atbara River, 88, 95, 96 Black Volta River, 65 Awash River, 88 Blantyre (Malawi), 148,156 Awash Station (Ethiopia), 96 Bloemfontein (South AErica), 138 Blue Nile Province (Sudan), 91,94,95 Badda Mountains, 99 Blue Nile River, 86,94,95, 96 Bafata Region (Guinea-Bissau), 69 Bo (Sierra Leone), 69 Bahr al Ghazl Province (Sudan), 95 Bobo Dioulasso (Burlina Paso), 57 Bale Mountains, 86 Bokkeveldberge Range, 142 Bale Province (Ethiopia), 89, 97, 98 Boma Hills, 92 Bamako (Mali), 56 Bophuthatswana, 137,138,149 Bamenda Hills, 69 Bora (Ethiopia), 92 Bam-ileke Plateau, 111 Bomo State (NiLgeria), 65 Bandundu Province (Zaire), 109 Brazzaville (Congo), 34,105,109 Bangala (Zaire), 107 Buyenzi District (Burundi), 128 BangweuLu Swamp, 156 Bwanje Valley,

199 186 Estirnating Woody Biomass in Snb-Salwamn Africa Cabinda (Angola), 28,38, 109, 111 Dembos cloud forest, 159,160 Cabo Delgado Province (Mozambique), 159, 160, 164 Dese (Ethiopia), 87,95 Cacheu River, 76 Desert of Moq&medes, 136, 141,143, 145 Cacine- River, 76 Didinga Hills, 92,99 Cale-Jon Valley, 151 Dire Dawa (Ethiopia), 93,96,97,99 Calvinia (South Africa), 143 EJambala (Congo), 106 Cape Barracouta, 143 Djougou (Benin), 67 Cape Mountains, 143 Dochire (Ghana), 65 Cape Province (South Africa), 24,25,39,40,136-39, Dodoma Region (Tanzania), ,149,151,154,158,159 Dongoroma Range, 67 Cape St Lucia, 157 Dongotona Hills, 92 Cape Town (South Africa), 143,144,146 Dongotona Mountains, 99 Caprivi Strip (Namibia), 149,151,152 Dorsale du Kivu, 111 Casamance River and Valley, 57,76 Douala (Cameroon), 110 Cedarberg Ranges, 142,143 Drakensberg Mountains, 39,138,139,147,149, 157, Central Highlands (Kenya), 121, ,164 Central Highlands (Tanzania), 125 Durban (South Africa), 157 Central Province (Zambia), 159 Ceres (South Africa), 143 East London (South Africa), 158 Chaine de l'atakora (Atacora Mountains), 65,67,71 Eastern Equatoria Province (Sudan), 89, 93, 95, 99 Chaine de Madinani, 67 Ebrie, Lagune (Cote d'ivoire), 75 Chaine de Tlem6, 67 Ehi, Lagune (C6te d'ivoire), 75 Chalbi Desert, 119 EI'Atrun Oasis, 87 Chambeshi Valley, 162 El Buheyrat Province (Sudan), 95,96 Changane Valley, 141 El Bukeyral (Sudan), 86 Chela, Serra da, 141,163,164 El Milk, 87 Cl'erch'er (Ethiopia), 92 El Obeid (Sudan), 34,91 Chiange (Angola), 149 Ennedi Plateau, 54 Chimanirnani Mountains, 165 Enugu (Nigeria), 68 Chobe Region (Botswana), 149,163 Equateur Province (Zaire), 112 Chobe River and Valley, 162,163 Equatoria Region (Sudan), 32,89, 92, 93,95-99 Choke Mountains, 86 Eritrea, 85, 88, 89, 91, 92, 95, 98 Choma (Zambia), 148,149 Erkowit Hills, 87 Circonscription Administrative de Dapadong Ermelo (South Africa), 147 (Togo), 65- Est, R4gion de l' (Cote divoire), 67 Ciskei, 138,158,159 Etosha Salt Pans, 136,137,143,149,150 Conakry (Guinea), 76 Evale (Angola), 141 Copperbelt Province (Zambia), 159 Corubal River, 76 Forecariah Plains, 73 Cradock (South Africa), 138 Foret du Day,87, 92,93 Crater Highlands, 129, 130 Fouta Djailon, 31, 69,73 Cross River and Basin, 71,76 Franceville (Gabon), 105 Cuanza Norte Province (Angola), 155 Cuanza Sul Province (Angola), 155 Gabu Region (Guinea-Bissau), 66,69 Cunene River and Valley, 152,163 Gambaga Scarp, 64 Gambia River, 77 Dakar (Senegal), 56 Gamo-Gofa Province (Ethiopia), 89, 93, 96, 97 Daloh Forest, 92, 93 Gamioos Valley, 146 Danakil Alps, 85 Garamba, Parc National de la (Zaire), 110 Danakil Depression, 85 Gash Delta, 88 Danakil Desert, Gaza Province (Mozambique), 161 Dapadon& Circonscription Administrative Geba River, 76 de (Togo), 65 Gemsbok National Park (Botswana), 145 Dar es Salaam (Tanzania), 127,129 Geneale Valley, 90 Darfur Province (Sudan), 34,87,88,91,94 Gezira-Managli (Sudan), 88 Day, Forat du, 87,92, 93 Ghanzi District (Botswana), 137

200 Index of Place Names 187 Ghanzi Ridge, 150 Juba River and Valley, 34,89,90,91,97 Ginir (Somalia), 90 Jubbada Dhexe Region (Somalia), 90 Gogu Mountains, 97,99 Jubbada Hoose Region (Somalia), 90 Going Range, 67 Gojam Province (Ethiopia), 88,92,94,95,97,98 Kaap Plateau, 139,141, 151 Gola Hills, 73 Kaduna State (Nigeria), 65 Gola Mountains, 71,72 Kafue Flats, 155 Gonder Province (Ethiopia), 96,98 Kagera River, 130 Gori Hills, 73 Kakamega Forest, 38,127,130 Grahamstown (South Africa), 158 Kalahari Desert, 24,35,39,40,105,136-38,140,141, Grand Bassa County (Liberia), ,149,150, Grand Gedeh County (Liberia), 72 Kalale (Benin), 67 Great Berg River, 142 Kalemie (Zaire), 109 Great Karas Berg 149 Kananga (Zaire), 105,109 Great KCarroo, 39,136, 144, 145 KCandi (Benin), 65 Greenville (Liberia), 72,75 Kandoli Mountains, 160 Grootfontein (Namibia), 150 Kaniama (Zaire), 105 Guinea Highlands, 67 Kano (Nigeria), 31,65 Gulf of Tadjoura, 87 Kaokoland Escarpment, 150 Karora Hills, 86,87,92 Hanam Plateau, 141 Karsa Dek (Ethiopia), 97 Harare (Zimbabwe), 148,156 Kasai Basin, 105 Hardeveld Range, 142 Kasai Plateau, 105,109 Harer (Ethiopia), 93,94 Kasai Province (Zaire), 108,109 Hirer Hills, 92, 95, 96 Kassala (Sudan), 87 HIrer Mountains, 97 Kassala Province (Sudan), 87 HIerg Hills, 97 Kefa Province (Ethiopia), H:rerge Province (Ethiopia), 87,87, 93,96,98 Keren (Ethiopia), 87 Harper (Liberia), 75 Khartoum (Sudan), 26,89,92,94 Harismith (South Africa), 147,157 Khartoum North (Sudan), 89,92 Haut Uele (Zaire), 107 Khasbim El Girba (Sudan), 8B Haut Zaire Province (Zaire), 111, 112 Khor Abu Habi River, 88 Ilighveld, 140,147,165 Kigoma Region (Tanzania), 125 Hiwa, Wadi, 87 Kilimanjaro Region (Tanzania), 121 Hodh Basin, 55 Kilimanjaro Forest Reserve (Tanzania), 129 Holawajir Forest, 90 Kilwa Kivinje (Tanzania). 129 Huambo (Angola), 1,54,162,163 Kimberley (South Africa), 138,139,141,151 Hufla Province (Angola), 155,160,163,164 King William's Town (Soutli Africa), 158 Hunyani Range, 148 Kinshasa (Zaire), 34 Hwange (Zimbabwe), 153 Kisama, Parc Nacional da (Angola), 164 Kisangani (Zaire), 113 lbadan (Nigeria), 69 Kisiji (Tanzania), 129 l:le 1-dra, 77 Kisimaya (Somalia), 90 leorn f(dira a), 68 Kissidougou (Guinea), 69 ilubabor Province (Edtiopia), 96,98,99 Kivu, Dorsale du, 111 Imo State (Nigeria), 74 Knysna (South Africa), 158 Lnhambane (MozaIbique), 159 Knysna Forest, 158 Inhambane Province (Mozamnbique), 147,161 Kokstad (South Africa), 158 Inyanga National Park (Mozambique), 161,165 Kolwezi (Zaire), 34 hinga Region (Tanzania), 125 Komponi River, 76 Kopargo (Benin), 67 Jebel Gurgei (Sudan), 88,89,94 Kopshom, 138 Jebel Marra (Sudan), 94 Kordofan Province (Sudan), 34, 86, 91, 94 Jong River, 76 Korem (Ethiopia), 87 Jonglei (Sudan), 86, 94 Kouand6 (Benin), 67 Jos Plateau, 64,66,67,69 Kougaberge Range, 139

201 188 Estimating Woody Biomass in Sub-Salmran Africa Koundara (Guinea), 69 Lunda Norte Province (Angola), 155 Kourandou Range, 67 Lunda Sul Province (Angola), 159 Kruger National Park (South Africa), 140,141, Lunga River, ,149 Lunsemjwa Valley, 155 Kuando-Kubango Province (Angola), 152,154, Lurio Valley, ,163 Lusaka (Zambia), 159,164 Kwango (Zaire), 105,109 Kwango Basin, 105 Mabebe Depression, 150 Kwanyama (Angola), 162 Macondes Plateau, 161 Kwara State (Nigeria), 74 Madder Valley, 138 Kwazulu, 147 Madinani, Chaine de, 67 Mafeteng (Lesotho), 138 Ladysmith (South Africa), 157 Makgadikgadi Pan, 136,137,149 Lagos State (Nigeria), 74 Malanje Province (Angola), 155,159 Lagune Ebrie (Cote djivoire), 75 Malindi Region, 148 Lagune Ehi (Cote d'ivoire), 75 Malmesbury (South Africa), 142 Lagune Taio (Cote d'ivoire), 75 Mandara Hills, 69 Lake Chad, 10,106,107 Mandara Mountains, 106,107 Lake Edward, 131 Manding Region, 30,54 Lake Kariba, 151, 162 Manica Province (Mozambique), 159, 161,164 Lake Malawi, 25,37,125,126,128,148,156,159,162 Maputo (Mozambique), 157 Lake Nyasa, 122 Maramba (Zambia), 151 Lake Rukwa, ,126 Margherita (Somalia), 90 Lake Tanganyika, 109,125,126,128 Marsabit District (Kenya), 119 Lake Turkana, 119 Martin's Drift (Zimbabwe), 162 Lake Victoria, 28,37,38,121,122,125,127, 128, 130 Maseru (Lesotho), 138 Lake Victoria Basin, 37,38,127,128,131 Massif de l'adoumaoua, 110 Lake Victoria Esrpment, 37 Massif of Afoll, 55 Lake Volta, 67,73 Matabeleland (Zimbabwe), 140,147,149, 151,153, Lamto (Cote d'ivoire), ,164 Langeberg Mountains, 139,141 Matatiele (South Africa), 138 Laupula Province (Zambia), 159 Matopo Hills, 151 L6fini River and Valley, 105, 106 Mavia, Planalto, 164,165 Libreville (Gabon), 109 Mayumba Forest, 112 Libyan Desert, 85-88,91 Mbandaka (Zaire), 112 Lichinga (Mozambique), 164 Mbeya Region (Tanzania), 121,125 Lilongwe (Malawi), 148 Mbuji-Mayi (Zaire), 109 Lilongwe Highlands, 156 Mendebo Mountains, 99 Lilongwe Plateau, 148 Middelburg (South Africa), 138 Limbe (Malawi), 156 Middleveld, 147 Limpopo Escarpment, 153 Moqfmedes, Desert of, 136,141,143,145 Limpopo River and Valley, 140,141,146,149, 151, Mofimedes Province (Angola), ,157,162 Mogadishu (Somalia), 91 Lindi Region (Tanzania), 125 Mogalakwena Valley, 140 Lobito (Angola), 144,149 Mogol Valley, 140 Loma-Man Dorsale, 71 Mohokare Valley, 137 Ltolawajir Depression, 90 MokhotlongDistnit (Lesotho), 138 Luanda (Angola), 164 Mont Bowe de Kiendi, 71 Luangera Valley, 155 Mont Nangbion, 71 Luangwa Valley, 39,148 Mont Yev6l, 67 Luapula Province (Zambia), 155,156 Montserrado County (liberia), 72 Lubango (Angola), 149 Morogoro Region (Tanzania), 120 Lubombo Hils, 147 Morro de Moco, 162 Lubumbashi (Zaire), 34,35,108,109 Mount Bend, 69 Luganda River, 159 Mount Cameroon, 110 Luisnishi (Zaire), 108 Mount Gakarosa, 139

202 Index of Place Names 189 Mount Frere, 158 Oio Region (Guinea-Bissau), 69 Mount Goda, 87,92 Okavango River and Delta, 40, 149, 162 Mount Kenya, 36,129,130 Okavango Swamp, 163 Mount Kenya National Park (Kenya), 129 Olokemeji (Nigeria), 128 Mount Kilimanjaro, 36,121,129,130 Omatako River, 162 Mount Mulanje, 162 Omdurman (Sudan), 89,92 Moxico Province (Angola), 159,161 Omo Valley, 90,91 Moygua (Angola), 141 Ondo (Nigeria), 69 Mpama catchment, 106 Ondo State (Nigeria), 74 Mudug Region (Somalia), 85 Orange Free State, 24,25,137-39,141,149,151, Mugamba District (Burundi), ,157 Mulanje Massif, 160,164 Orange River and Valley, 136,138,141,151 Mwenezi River, 140 Otecbinjan (Angola), 149 Mweru Swamp, 156 Ouake (Benin), 67 Mwinilunga (Zambia), 162 Oubangui River and Basin, 107,110,111,112 Ovamboland (Namibia), 141 N'Djamena (Chad), 54 Oyo State (Nigeria), 74 N'Giva (Angola), 141 Nababeep (South Africa), 142 Paarl (South Africa), 146 Nairobi (Kenya), 124 Palala Valley, 140 Nairobi National Park (KCenya), 129 Parakou (Benin), 67 Namacunde (Angola), 141 Parc Nacional da Biknar (Angola), 164 Namaland (Namibia), 141,143,149 Parc Nacional da Kisama (Angola), 164 Namaland Mountains, 136 Parc National de la Garamba (Zaire), 110 Namaqualand (Namunbia), 136,143 Pehonko (Benin), 67 Namib Desert, 24,25,39,40,136,139,144,145 P&6r (Benin), 67 Namibe (Angola), 136,141 Phalombe Plains, 148 Namibe Province (Angola), 143,154 Piet Retief (South Africa), 165 Nampula (Mozambique), 165 Pietersburg (South Africa), 140,149 Nampula Province (Mozambique), 159,163,164 Piketberg (South Africa), 142 Namwera Malawi), 148 Planalto do Mavia, 164 Nandi Escsrpment, 130 Plateau of Kasai, 105,109 Natal Province (South Africa), 40,147,154,157,158 Plateaux Province (Congo), 105,106 Natitingou (Benin), 67 Pointe Noire (Congo), 34 N&d6l (Central African Republic), 107 Pondoland Region (South Africa), 157,158 Nehone (Angola), 141 Port Elizabeth (South Africa), 27, 137,146,158 Nelspruit (South Africa), 153,158 Port St. Johns (South Africa), 158 N4ro-Mer (C6te d'ivoire), 75 Port Shepstone (South Africa), 157 Ngamiland District (Botswana), 137 Port Sudan (Sudan), 85, 87,88 Niassa Province (Mozambique), 161,163,164 Potgietersrus (South Africa), 149 Niger River and Valey, 30,64,66,69,74,76,77 Pretoria (South Africa), 137,138,149 Nikldi (Benin), 67 Prieska (South Africa), 151 Nile River and Basin, 54,86,88,91, 92,94, 95,97,98 Pwani Region (Tanzania), 125, 129 NimIrui Hills, 73 Nord, Region du (C8te d'ivoire), 67 Qacha's Nek Distict (Lsotho), 138 North East District (Botswana), 137 Qala 'en Nahal (Sudan), 34 Northem Province (Zambia), 155,159,161 Queenstown (South Africa), 138 Northern Region (Malawi), 160 Quthing (Lesotho), 138 Northwestem Province (Zambia), 155,159,164 Nubian Desert, 85,88 Red Sea, 85,91,92 Nyilca Plateau, 155,164 Red Sea Hills, 34,85-87 Nzdr&kor6 (Guinea), 69 R4gio'. de lest (COte d'ivoire), 67 Region du Nord (COte d'ivoire), 67 Ogaden Region (Ethiopia), 34,85,89 Region du Sud-Ouest (C6te d'ivoire), 73 Ogooue River, 109. Rift Valley, 24,34,89-92,94,96,97,99,109,111,148 Ogun State (Nigera), 74 Rio Muni, 28,34,109,111,112

203 190 Estimating Woody Biomass in Stab-Saliran Africa Rio Nunez, 76 Sud-Ouest, Region du (Cote d'ivoire), 73 River Cess (Liberia), 72,75 Sunyani (Cote d'ivoire), 73 Roggeveld Region (South Africa), 143 Rokel River, 76 Table Mountain, 10, 146 Rouxville (South Africa), 139 Tabora Region (Tanzania), 120, 121,125 Rukwa Region (Tanzania), 125 Tabou (Cote d'ivoire), 75 Ruwenzori Mountains, 129 Tagant Escarpment, 55 Ruwenzori National Park (Uganda), 129 Tahoua (Niger), 56 Taio, Lagune (C6te d'ivoire), 75 Sabi Valley, 140,148,159 Tamale (Ghana), 65,66 Sahara Desert, 24,53 Tana Valley, 121 St. Helena Bay, 142 Tanga Region (Tanzania), 120,121,126-28,130 Sand Valley, 140 Tchaourou (Benin), 67 Sankarani River, 66 Tete Province (Mozambique), 161 Savanes Region (Togo), 65 Teyateyaneng (Lesotho), 138 Save Valley, 147 Thaba-Tseka District (Lesotho), 138 Scarcies River, 76 TIhyolo Mountains, 160 Scobell's Kop (South Africa), 139 Tibesti Plateau, 53 Sekl'ot'a (Ethiopia), 95 Tieme, Chaine de, 67 Senegal River and Valley, 30, 55 Tigre Province (Ethiopia), 34, 92, 86, 94, 98 Sengu Valley, 137 Tinkisso River, 66 Serergeti National Park (Tanzania), 122,123 Togo Hills, 66 Serengeti Plain, 122 Tokar Delta, 88 Serra da Chela, 141,163,164 Tombali Region (Gabon), 73 Serra Upanda, 164 Tombua (Angola), 141 Sewa River, 75 Tongaland Region (South Africa), 157,158 Shaba Plateau, 109 Tourou Range, 67 Shaba Province (Zaire), 34,108,109 Transkei, 138,149,151, Shabeellaha Dhexe Region (Somalia), 90 Transvaal Province (South Afica), 25-27,39,40, Shabeellaha Hoose Region (Somalia), , , ,149,151,153,158, Shashe River, ,164 Shebelle, Wadi, 87,89-91 Tsau (Botswana), 149 Sherbro Island, 76 Tsavo National Park (Kenya), 123 Shewa Region (ECtopia), 86,92,93,96-99 Tswapong HIlls, 150 Shinyanga Region (Tanzania), 121,125 Tuli (Zimbabwe), 145,146 Shire Highlands, 148,156,160 Tuli River, 140 Shire VaUey, 148,156 Sidamo Province (Ethiopia), 93,97,98 Uige Province (Angola), 155 Simien Region (Ethiopia), 86 Uitenhage (South Africa), 158 Singida Region (Tanzania), 121 Umkomaas VaUey, 157 Sinoe County (Liberia), 72 Umtata (South Africa), 158 Sobat Province (Sudan), 86,88,95 Umtata River, 157 Sofala Province (Mozambique), Umzimvubu Valley, 158 Sokoto (Nigeria), 31,65 Umzingwani River, 140 Sokoto State (Nigeria), 65 Upanda, Serra, 164 Southern Province (Zambia), 148,155,159 Upington (South Africa), 141 Southern Region (Malawi), 148 Upper Karroo, 145 Soutpansberg Mountains, 140 Upper Moa Basin, 73 Springbok (South Africa), 141,1-42 Usambara Mountains, 122,129,130 Springfontein (South Aftica), 151 Vaal River and Valley, 138, 151,153 Steinkop (South Africa), 142 Venda, 140 Stellenbosch (South Africa), 146 Viphya Massif, 160 Sterkstrom (South Africa), 139 Virginia (South Africa), 138 Sudd, the, 95, 96 Volta Delta, 76

204 Index of Place Names 191 Waanje River, 75 Windloek Mountains, 150 Wadi Hiwa, 87 Wadi Shebelie, 87,89-91 Yaounde (Cameroon), 112 Wallel (Etiiopia), 99 Welega Province (Ethiopia), 88,95,97,98 Zaire Province (Angola), 155, 163 Wellington (South Africa), 146 Zaire River and Basin, 34, 72,73,105, Welo Province (Ethiopia), 34,86,89,92-94,97,98 Zambezi River and Valley, ,155,156,159, Wepener (South Africa), , 163 West Lunga National Park (Zambia), 163 Zambezia Province (Mozambique), 156,160 Western Equatoria Province (Sudan), 93,96 Zambian Plateau, 160 Western Province (Zambia), 159, 164 Zanzibar, 125,127 White Nile Province (Sudan), 91 Zei 2diela (South Africa), 153 White Nile River, 88,94 Zimbabwean Mountains, 165 Windhoek (Namibia), 141, 149 Zomba Plateau, 160

205 Legend for Land Cover Class Maps DESERT LI] H Desert GRASSLAND Veld grassland Hydromorphic grassland Ethiopian montane steppe Montane grassland and heathland WOODED GRASSLAND WOODLAND Open woodland Dry Sudanitn woodland SudanlEthinpian woodland and thicket Sudanian woodland Moist Sudanian woodland Scasonal miombo Wet miombo HIGH WOODY BIOMASS MOSAICS Semi-desert wooded grassland Acacia wooded grassland Plateau wooded grassland Transitional wooded grasland Edaphic wooded grassland SHRUBAND shrublanhighland VclId shrubland Hill shrubland Bushy shrubland Kalahari shrubland Wooded shrubland FOREST Evergreen woodland mosaic Cultivation and forestwoodiand mosaic Cultivation and forest regrowth mosaic Guinean woodland 1High productivity West African cultivation and forest Medium productivity West African cultivation and romst Mangrove Evergreen forest cultivation mosaic Coastal and gallery forest Montane forest BUSHLANDry AcacTHICKETni s bushland and thicket * Mesophilous humid tropical forest.. Dry Acacia-Commiphara bush_and and thicket Humid tropical swamp forest Pynbac thicket Moist Acacia-Commiphora bushland and thicket Sahel-Sudanian Acacia wooded bushland Escarpment wooded thicket Ombrophilous humid tropical forest acoud cover LOW WOODY BIOMASS MOSAICS Ae not mapped Acacia woodland mosaic CEastAfrican low woody biomass mosaic

206 T f /-St I /: :: 3,.-o~~ X :/D : : 1\, ;'- Vl /I II.0;:. * z _l_ s;xes I. -.,:S >t-'m1i -- M 2 li] i!: _ S~~~ - S~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~;

207 La,,n Zanahn! EPW4lAieia ftajeccion, 30 Lmrg/rnd, En 40-5D ARAB REPUBLIC_OjEGYPT 'LIBYA - k j - + iporlsudan P 2 l,,,;, D - U i D' A N10 SUDAN I j_ A N 0- ij = CliADI a~~~~~~~~~~~~~~~~~~~~~~~~ c"giz..& I~~~~~~~~~~~~~~~~~~I n <7; r ~~~~~~~~~~I...!r i 7Chsm ik t, mbru ; 11 d i pb d, 1 essainam.\ Nll R - O c an <si4, :,S; 1 1wt -A L-g! d, E.0~~~~~ad! A. Id. ~ ~ ~ ~ Alf'%, E h6,,' ' "14 S. b 1t;FW,M.J I ;

208 i~~~~~~~~~~~~~i EQATORIAT i -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Ch1Wi ArZI a.n *c O c e a.z_t 91a k.: 0 IT 10 L--ZdlE. A-10-L--,E

209 10~1 1. k A t w w 1 r. Swakrap wtotel* ;s r _7eprcolCaprgcon ~~~~~~~~~~~~~~~~~~~~~~s 041 Whmbn Larlbolt 20up ZclrRhal IAr a ProjecliOn 20 / D-ArlA?lrltZhiE.I.dio /JD 30~~~~~~~~~~~~~- 5 1 _, 0 Cqr,sInow}lr IS r{tryitlwb- M! ~~i, -er ii / i LU&,L~~ L.inbrt Za1ith.1 0San A-~ P,o.1mi.. 2O, 10-,glrif &,a 140