STANDARD METHODS FOR MEASURING AND MONITORING BIODIVERSITY

Transcription

1 STANDARD METHODS FOR MEASURING AND MONITORING BIODIVERSITY EDITED BY DONAT AGOSTI. JONATHAN D. MAJER. LEEANNE E. ALONSO. AND TED R. SCHULTZ SMITHSONIAN INSTITUTION PRESS WASHINGTON AND LONDON

2 2 by the Smithsonian Institution All rights reserved Copy editor and typesetter: Prineton Editorial Assoiates, In. Prodution editor: Duke Johns Designer: Amber Frid-Jimenez Library of Congress Cataloging-in-Publiation Data Ants : standard methods for measuring and monitoring biodiversity I edited by Donat Agosti.., [et al.]. p. m.--{biologial diversity handbook series) Inludes bibliographial referenes (p. ). ISBN (loth: alk. paper) ISBN (pbk.: alk. paper) 1. Ants-Speiation-Researh. 2. Biologial diversity-measurement. I. Agosti, Donat. II. Series. QL568.F7 A '6-d British Library Cataloguing-in-Publiation Data available Manufatured in the United States of Ameria The paper used in this publiation meets the minimum requirements of the Amerian National Standard for Information Sienes-Permanene of Paper for Printed Library Materials ANSI Z For permission to reprodue illustrations appearing in this book, please orrespond diretly with the owners of the works, as listed in the individual aptions. The Smithsonian Institution Press does not retain reprodution rights for these illustrations individually, or maintain a file of addresses for illustration soures.

3 Chapter 15 Applying the ALL Protool Seleted Case Studies Brian L. Fisher, Annette K. F. M: Ish, Raghavendra Gadagkar, Jaques H. C. Delabie, Heraldo L. Vasor elos, and Jonathan D. Majer Although the ALL Protool put forward in this book was only reently developed and is published for the first time here, it is derived from the experienes of many myrmeologists and from numerous studies of ant diversity over the years. Before the development of this protool, studies of ant diversity utilized a wide variety of methods, as desribed in Chapter 9. The ALL Protool is the result of an extensive evaluation of these methods in different ountries and under a variety of onditions. The studies by Delabie et al. (Chapter 1) provided the strongest data for hoosing methods for the ALL Protool, but a number of other key studies were also influential in its development. Brief desriptions of these key studies, from diverse parts of the world (Madagasar, Malaysia, India, and Brazil), are provided here to illustrate how the two prinipal olleting methods of the ALL Protool (leaf litter extration using the Winkler tehnique and pitfall traps) ompare, and how they have been used suessfully in a variety of biodiversity studies, partiularly to measure ant diversity and to detet habitat hange. Full desriptions of these studies are presented in Agosti et al. (2). Madagasar A series of studies in Madagasar by Fisher and olleagues (Fisher l 996a, I 996b, 1998, I 999a, 1999b; Fisher and Razafimandimby 1997; Fisher et al. 1998) provides the best insight into the usefulness of the ALL Protool and the omparative value of its two prinipal methods, Winkler extration and pitfall traps. In one 27

4 28 B. L. ASHER ET AL. partiularly informative study, Fisher surveyed ground-dwelling ant diversity at elevation zones separated by 4 m at four rainforest loalities: the Reserve Naturelle Integrale d' Andohahela, the Reserve Naturelle Integrale d' Andringitra, the Reserve Speiale d' Anjanaharibe-Sud, and the Western Masoala Peninsula in eastern Madagasar. At eah site, 5 pitfall traps were used and 5 leaf litter (Winkler) samples were taken, in parallel lines I m apart along a 25-m transet. Pitfall traps were plaed and leaf litter samples gathered every 5 m along the transet. Pitfall traps onsisted of test tubes (18 mm internal diameter by 15 mm long), partly filled to a depth of about 5 mm with soapy water and a 5% ethylene glyol solution, inserted into polyvinyl hloride (PVC) sleeves, and buried with the rim flush with the soil surfae. Traps were left in plae for 4 days. Ants were extrated from samples of leaf litter using Winkler extrators over a 48-hour period (Chapter 9; Fisher 1998). The observed and predited number of speies sampled by the Winkler, pitfall, and ombined methods for 1-, 2-, 3-, 4-, and Table 15.1 Observed Number of Ant Speies Evaluated at Different Sample Sizes for Winkler Saks, Pitfall Traps, and Both Methods for Eah 8-m Zone Site in Madagasar" Observed Speies Rihness after: Estimated Speies Rihnessb All (5) Jak- Methods Samples Samples Samples Samples Samples ICE lnife M-M 8 m Andohahela Winkler 39.6 (59.3) 49.5 (74.1) 55.3 (82.8) 59.5 (89.l) 63 (94.3) Pitfall 14.8 (43.6) 2.5 (6.3) 24.3 (71.5) 27.4 (8.6) 3 (88.3) Both methods 44.5 (59.6) 55.5 (74.4) 62.1 (83.2) 66.9 (89.7) 71 (95.2) m Andringitra Winkler 52.2 (66.8) 63.3 (81.) 68.8 (88.1) 72.8 (93.1) 76 (97.2) Pitfall 1.2 (45.6) 14 (62.4) 16.4 (72.9) 11.6 (78.6) 19 (84.7) Both methods 53.8 (68.6) 64.1 (81.8) 69.8 (89.) 74. (94.4) 77 (98.2) m Andringitra Winkler 43.4 (67.6) 51.7 (8.6) 56.4 (88.) 6.5 (94.4) 64 (99.8) l Pitfall 8.7 (39.) 12.2 (54.9) 14.9 (67.) 17. (76.7) 19 (85.6) Both methods 44.9 (66.5) 53.3 (79.1) 59.4 (88.1) 63.5 (94.1) 67 (99.3) m Masoala Winkler 62.2 (57.1) 79.8 (73.3) 91.1 (83.7) 99.3 (91.3) 16 (97.4) Pitfall 8.9 (38.2) 12.7 (54.5) 15.8 (67.7) (78.6) 2 (86.) Both methods 62.5 (55.4) 81.2 (71.9) 93.4 (82.7) 12.4 (9.6) 19 (96.5) m Anjanaharibe-Sud Winkler 54.4 (58.4) 69.1 (74.1) 78.9 (84,7) 86.7 (93.l) 92 (98.8) Pitfall 8.5 (37.8) 11.7 (52.1) 14.6 (65.) 17.4 (77.2) 2 (88.8) Both methods 56.2 (57.6) 71.9 (73.7) 82.? (84.7) 91.1 (93.3) 97 (99.3) "Number or speies represents the mean or 1 randomizations or sample pooling orde. 6 JCE, inidene-based overage estimator; jakknife, first-order jakknife estimator; M-M, Mihaelis-Menten asymptote (the perentage of the M-M asymptote is given in parentheses in the first five olumns).

5 Applying tht! AU Protool J 14 l E 8 ::I ID 6 ~ II 4 3 E ::I Number of stations sampled Figure Assessment of leaf litter ant sampling tehnique at 825 m on the Masoala Peninsula, Madagasar. The lower speies-aumulation urve (thik line) plots the observed number of speies as a funtion of the number of stations sampled. The upper urves display the nonparametri first-order jakknife (dashed line) and the inidene-based overage estimator (ICE; solid line), estimated total speies rihness based on suessively larger numbers of samples from the data set. Curves are plotted from the means of 1 randomizations of sample aumulation order. SO-sample sizes are shown in Table Within the area of the survey, the Winkler tehnique olleted the majority of ants foraging and living in the leaf litter. Most speies olleted by pitfall traps were also sampled by Winkler extration, indiating that whereas pitfall samples in the same area would most likely add additional speies, these speies would probably have already been obtained by the Winkler method. Although this may hold true for most rainforest sites, it may not apply to all habitats. For example, Fisher and Razafimandimby (1997) found that in dry forest habitats, whih ontain more areas of open or bare ground, pitfall traps may ollet a greater number of unique speies. Speies-aumulation urves for the 825-m site on the Masoala Peninsula, the most speiesrih site (Fig. 15.l), indiate that within the area of the survey the tehniques employed olleted the majority of ants foraging and living in the leaf litter in the area enompassed by the 25-m transet, and that with inreased sampling effort using the same methods in the same area, only marginal inreases in speies rihness would be attained. Although additional olleting methods, or a survey in a different area or season at the same elevation, would most likely ollet additional speies, these results show that the ALL Protool provides suffiient sampling for statistial estimation and omparison of speies rihness, and for omparison of fauna) similarity and speies turnover. Malaysia Studies of ant diversity in the Pasoh Forest Reserve of West Malaysia by Malsh provide interesting data on the effets of plot size sampled in ant diversity studies. Situated in Negeri Sembilan, West Malaysia, about 14 km southeast of Kuala Lumpur (2 59'N, 12 19'E), the Pasoh Forest Reserve is a typial example of a Southeast Asian ever-wet lowland rainforest, with primary lowland dipteroarp forest situated between 75 and 15 m above sea level. A total of nine leaf litter plots (eah 25 m 2 ) were investigated. Eah plot omprised a 5 x 5- m2 area with an additional 3 x 3-m 2 area nested in the middle of the plot. Eah of the two nested areas (16 m 2 and 9 m 2 ) was sampled separately, and the sum of the two equaled a 25-m 2 area. This approah enabled the omparison of all nested areas within plots. Ants from the leaf litter were extrated by the Winkler method after 24 hours and then again after an additional 24 hours (Chapter 9). The nested sampling area design revealed that the size of the leaf litter sample (plot size) an influene the number of ant speies olleted. The number of speies olleted per square meter for eah 9- and 25-m 2 plot is shown in Table On average, one more speies was

6 21 B. L. FISHER BT AL. Ill 6 GI 5 "ij 8. 4 Ill 3 ii 2 E i 1 25m 2 16m 2 a 9m Frequeny of apture Figure Frequeny of apture for speies in nine 9-m 2, 16-m 2, and 25-m 2 plots in Malaysia. found per m 2 on the 9 m 2 plots ompared to the 25 m 2 plots. The similarity in ant speies omposition was also affeted by plot size, with Sorensen s similarity values of the 25 m 2 plots ranging from 37.5% to 63.8%, and those of the 9 m 2 plots from 28.5% to 66.7%. The mean values are 52.7% for the 25 m 2 plots and 43. % for 9 m 2 plots. These differenes are highly signifiant (P <.1, Mann-Whitney U-Test). The higher speies turnover in the 9 m 2 plots resulted from more single aptures and a smaller number of repeated aptures (Fig. 15.2). The speies-aumulation urves for eah plot size reveal that none of the plot sizes samples all ants of the area sine none of the urves Table 15.2 Number of Ant Speies per Square Meter for 9- and 25-m 2 Plot Samples in Malaysia Plot 9m 2 25 m 2 Pl P P P PS P P pg "" P Mean _; Ill! 8 E 6 :I GJ >,89 '. 4 ~.,. ~ O 25m2 i 16m2 :; 2 9m2 E :I (J Area sampled (m2) Figure Speies-aumulation urves for the three plot sizes of 9 m 2, 16 m 2, and 25 m2 in Malaysia levels off (Fig. 15.3). In addition, eah plot size produes a different estimation of the overall ant speies rihness. This is an important onsideration when omparing the results from studies using different plot sizes; it is not possible to ompare them diretly. These results emphasize the importane of using a onsistent, standard plot size aross studies in order to make omparisons. The l-m2 litter plot of the ALL Protool provides this standardization. Western Ghats, India Ant diversity was investigated using a variety of sampling tehniques in the state of Kamataka, Western Ghats, India (Gadagkar et al. 199, 1993). A total of 36 1-ha plots from 12 habitat types were sampled in sites representing elevations from sea level to 6 min forested habitats, in three monoulture plantations, and in a forest that was regularly harvested to produe leaf manure. At eah of these sites, sampling was arried out in three 1-ha plots. Five sampling methods for ground-dwelling ants were employed at eah site: vegetation sweeps, pitfall traps, light traps, sented traps, and diret (hand) olleting. Light traps use a luminesent light soure to attrat insets that

7 Applying the ALL Protool 211 LightTrap3 Net Sweep Pitfall Trap Sented Trap Distane= 1- Morishita-Horn index Light Trap ~ Net Sweep -----~ Pitfall Trap All Out Searh Sented Trap ' Distane= 1- Jaard index Figure Dendrograms omparing different sampling methods by ant speies trapped in India. Data pooled from 36 plots for eah sampling method. are ative at night. Although light traps are typially used to sample flying insets, they an oasionally be useful for attrating flying alate ants and some noturnal ant speies. Sented traps are essentially a ombination of two standard ant sampling tehniques (pitfall traps and baits); in this study they onsisted of 2.5-liter plasti jars that were baited with unrefined sugarane and hung at about I m from the ground on wooden pegs. Intensive hand olleting was performed in eah I-ha plot to ollet representatives of as many speies of ants as possible. Two persons made the searh for I hour between 14 and 15 in every ase. In addition to providing the first estimates of ant diversity and abundane for any forest loality of India, the results of this study provide informative omparisons of five different methods of ant sampling. The ombination of the four trapping methods used was somewhat more suessful than hand olleting, yielding 12 speies from 31 genera while hand olleting yielded 11 speies from 27 genera. More signifiant is the fat that the traps and hand ol- leting yielded different speies: while 78 speies were obtained by both methods, the traps yielded 42 unique speies and hand olleting yielded 2 unique speies. It appears. therefore, that in spite of the effiay of the traps, a ombination of trapping and hand olleting may be desirable if a more omplete list of ant speies at a site is desired. Of the four trapping methods used. pitfall traps sampled the most speies, followed by vegetation sweeps, sented traps, and light traps in that order. The fat that pitfall traps and vegetation sweeps were more suessful is not surprising; indeed the fat that sented traps and light traps yielded as many ants as they did is surprising. Not only did the sented traps and light traps yield more ants than expeted, they yielded an ant fauna rather different from that obtained by the other methods (Fig. 15.4). Had this study inluded leaf litter extration as in the ALL Protool, many more ant speies would likely have been olleted. However, the ombination of several sampling methods used in this study, inluding hand olleting. illustrates that different tehniques usually ollet different omponents of the ant fauna. Therefore, if a more thorough inventory of grounddwelling ants is desired, it is reommended that a few additional methods he used along with the ALL Protool. Brazil Two studies in Brazil, one in the highly fragmented Atlanti rainforests of Bahia and the other in the Brazilian Amazon, reveal the utility of the ALL Protool as a means of deteting habitat disturbane. Atlanti Forests In Bahia, ten 11-m transets were established from the enter of a botanial reserve of seondary rainforest in the Centre for Cooa Researh, ltabuna, Bahia (Majer et al. 1997).

8 212 B. L. FISHER ET AL. 4. CD I ! 2.5 : 2. :E a 1.5 Pitfall Trap Samples Grassland Forest ~ Distane (m) CD l ! E :I CD :E " b.!!! "' Grassland Winkler Sak Samples Forest 3.._ Distane (m) u 8.5 CD ~ j E 7. :I 6.5 " CD 6. :E Pitfall and Winkler Samples Combined Grassland Forest ~ Distane (m) Transets were marked at 1-m interval points and were sited so that seven points extended into the reserve, one ourred in the middle of the planted edge 5 m outside the forest, and two were in an adjaent field. The ALL Protool was followed, with leaf litter samples olleted along the transet and extrated using Winkler saks (for only 24 hours) and pitfall traps set out along the transet and left out for 48 hours. Figures 15.5a- illustrate the variation in the mean number of ant speies olleted by pitfall traps, by Winkler saks, and by both methods along the transets. The mean number aught by pitfall traps ranged from 1.8 to 3.4, and there was little apparent trend in numbers along the transets, although the maximum rihness was enountered at the point farthest into the forest (Fig. 15.5a). The mean number of speies extrated by the Winkler saks ranged from 3.8 to 7.3. It is noteworthy that the highest density was one again reahed at the point farthest into the forest and that the lowest density was enountered at the point 25 m into the field (Fig. 15.5b). The results of this survey reveal an abrupt differentiation in forest and field ant ommunity omposition. Five assemblages of ant speies were distinguished along the transet. The largest grouping ontained speies that were ubiquitous along the transets or that were ubiquitous exept at points outside the forest. The seond group omprised ants that showed a tendeny to our around the outer forest margin, while the third and fourth groups ontained ants that were generally found in deeper forest. The fifth group ontained 12 speies that were found Figure Mean number of ant speies (and standard error) sampled by (a) pitfall traps, (b) Winkler saks, and () both methods ombined along ten transets extending from the field into the rainforest in Bahia, Brazil. The vertial line indiates the position of the fene around the forest reserve.

9 Applying the ALL Protool 213 only in the field or planted edge. These ant assemblages an be used to monitor ant ommunities in these different land uses and to detet further hanges, even if fairly subtle, in habitats and their rnirolimates. Amazon In the Brazilian Amazon, ground-dwelling ants were olleted in three 1-ha forest fragments, in three 1-ha fragments, in two 1-ha fragments, and in one ontinuous forest area. In eah of these nine fragments a 1-ha plot was delimited and, within this, a total of 36 sampling points, distributed at intervals of 2 m, were established. Three methods of ant sampling were used: litter extration, pitfall traps, and soil samples. Of the three methods, litter sampling was the most effiient in terms of the number of speies olleted. The mean number of speies olleted per plot was signifiantly greater in the litter than in the pitfall traps and soil, whereas the number of speies olleted in the pitfall traps was greater than that in the soil (ANOVA, F 2, 16 = 29.87, P <.1; Table 15.3). Although the number of speies reorded per fragment was greater in the litter samples than in the pitfall traps, the total number of speies reorded by eah of these two methods in all nine forest plots studied was quite similar, and both yielded greater numbers than olletions from the soil samples (Table 15.3). Litter sampling was also the best method to predit overall ant speies rihness (number of speies olleted using the three methods ombined) in eah of the study plots. The number of speies that were unique to eah method ranged from 2 to 43 speies, a number that usually represented more than 2% of all speies olleted by that method (Table 15.3). This observation indiates that these methods are omplementary. Their use in ombination, therefore, better haraterized the ant fauna of the fragments. Speies of Cerapahy- Table 15.3 Number of Ant Speies Colleted Using Three Different Sampling Methods in Forest Fragments near Manaus, Brazil Litter Pitfall Soil Subfamily Samples Traps Samples Myrmiinae Ponerinae Formiinae II Dolihoderinae 3 5 Eitoninae l 5 I Cerapahyinae 2 5 Pseudomyrmeinae l 2 Leptanilloidinae 2 Total (unique) 147 (43) 142 (39) 16 (2) Mean±SD 54.2 ± ± ±5.6 inae, for instane, were reorded mostly in the soil samples, whereas those in the Eitoninae were mostly reorded in the pitfall traps. On the other hand, many Myrmiinae were only reorded in the litter samples (Table 15.3). No onsistent hanges in speies diversity were found in response to variations in fragment area. It must be stressed, however, that these results reflet a lak of relationship between the density of ant speies (number per unit area) and forest area, not in overall speies number and forest area, as the latter relationship is learly positive and signifiant. Within two of the three sites studied, the density of ant speies inreased as forest area inreased, whereas in the third site the opposite trend was found. Differenes in the history of fragment isolation (resulting in different matrix habitats) may have aounted, at least in part, for these onfliting results (Vasonelos and Delabie 2). Ordination of the study plots aording to their similarities in speies omposition indiated that forest fragmentation does affet the omposition of the ground-dwelling ant ommunity. A "site effet" on speies omposition was also deteted, indiating some degree of

10 214 B. L. FISHER ET AL. heterogeneity in speies distribution among the three sites studied, even though these sites were only 1-25 km apart. The results of this study strongly suggest that forest fragmentation affets the struture of ground-dwelling ant ommunities. The diversity and omposition of the ant ommunity would thus be useful to inlude in a monitoring program of forest fragments to follow and predit future hanges. Conlusion These five ase studies, plus those onduted by Delabie et al. (Chapter 1), provided the omparative basis for seletion of methods for the ALL Protool. They also illustrate the appliation of the ALL Protool to address a wide range of researh and applied onservation questions in a variety of loations. We hope that these studies will inspire and guide the use of the ALL Protool, and the inlusion of grounddwelling ants, in biodiversity studies aross the globe. ACKNOWLEDGMENTS The Malagasy projet was funded in part by grants from the World Wide Fund for Nature, Madagasar; the National Geographi Soiety ( ); and the National Siene Foundation (INT ). Researh in India was supported by grants from the Ministry of Environment and Forests, Government of India. The Amazon study was supported by the Biologial Dynamis of Forest Fragments Projet (a projet of the Smithsonian Institution and the Brazilian Institute for Amazon Researh) and the Brazilian Counil for Sientifi and Tehnologial Development.

11 Literature Cited Agosti D., Majer J., Alonso L., Shultz T.R. & (Eds.) 2. Sampling Ground-dwelling Ants: Case Studies from the World's Rain Forests. Perth, Western Australia: Curtin University of Tehnology. Delabie, J. H. C., B. L. Fisher, J. D. Majer and I. W. Wright. 2. Sampling Effort and Choie of Methods, Pp. pp In D. Agosti, J. D. Majer, L. E. Alonso and T. R. Shultz (Eds.), Ants: Standard Methods for Measuring and Monitoring Biodiversity. Smithsonian Institution Press, Washington and London. Fisher, B. L. 1996a. Ant diversity patterns along an elevational gradient in teh Reserve Naturelle Integrale d'andringitra, Madagasar. Fieldiana: Z.Oology (n.s.). 85: b. Origins and affinities of the ant fauna of Madagasar, Pp. pp In W. L. Louren~o (Ed.), Biogeographi de Madagasar. Editions ORSTOM, Paris Biogeography and eology of the ant fauna of Madagasar (Hymenoptera: Formiidae). Journal of Natural History. 31: Ant diversity patterns along an elevationa gradient in the Reserve Speiale d'anjanaharibe-sud and on teh western masoala Peninsula, Madagasar. Fieldiana: Z.Oology (n.s.). 9: a. Ant diveristy patterns laong an elevational gradient in the Reserve Naturelle lntegrale d'andohahela, Madagasar. Fieldiana: Z.Oology (n.s.). 94: b. Improving inventory effiienty: A ase study of leaf litter diversity in Madagasar. Eologial Appliations. 9: Fisher, B. L., H. Ratsirarson, S. Razafimandimby et al Les fourmis (Hymenoptera: Formiidae ), Pp. pp In J. Ratsirarson and S. M. Goodman (Eds.), Inventaire Biologique de la Foret Littorale de Tampolo (Fenoarivo Atsinaanana) Reherhes pour le Developpement, Serie Sienes Biologiques. Centre d'information et de Doumentation Sientifique et Tehnique and World Wide Fund for Naure, Antananarivo, Madagasar. Fisher, B. L. and S. Razafimandimby Les fourmis (Hymenoptera: Formiidae), Pp. pp In. Langrand and S. M. Goodman (Eds.), Inventaire Biologique Forets de Vohibasia et d'/soky-vohimena. Reherhes pur le Developpement, Serie Sienes Biologiques. Centre d'lnformation et de Doumentation Sientifique et Tehnique and World Wide Fund for Naure, Antananarivo, Madagasar. Gadagkar, R., K. Chandrashekara and P. Nair Inset speies diversity in tropis: Sampling methods and a ase study. }.Bombay Nat.Hist.So. 87: Gadagkar, R., P. Nair, K. Chandrashekara and D. M. Bhat Ant speies rihness and diversity in some seleted loalities in Western Ghats.India. Hexapoda. 5: Majer, J. D., J. H. C. Delabie and N. L. MKenzie Ant litter fauna of forest, forest edges and adjaent grasslands in the Atlanti rain forest region of Bahia, Brazil. Insetes Soiaux. 44: Vasonelos, H. L. and J. H. C. Delabie. 2. Ground ant ommunities from entral Amazonia forest fragments, Pp. pp In D. Agosti, J. Majer, L. Alonso and T. R. Shultz (Eds.), Sampling Ground-dwelling Ants: Case Studies from the World's Rain Forests. Curtin University Shool of Environemental Biology, Bulletin No. 18, Perth, Australia.