Temporal scaling of moisture and the forest-grassland boundary in western Canada

Transcription

1 ELSEVIER Agricultural and Forest Meteorology 84 (1997) AGRICULTURAL AND FOREST METEOROLOGY Temporal scaling of moisture and the forest-grassland boundary in western Canada Edward H. Hogg I Canadian Forest Service, Street, Edmonton, Alta. T6H 3S5, Canada Received 30 September 1995; revised!o March 1996; accepted I April 1996 Abstract A major challenge in understanding how forests may respond to climate change is that many of the critical processes, such as precipitation patterns and disturbance regimes, are highly variable and operate over a wide range of temporal and spatial scales. Analyses of vegetation-climate relationships have been conducted as a simple scaling technique to identify the most important long-term factors controlling the boundary between boreal forest and grassland in western Canada. In an earlier study, the boundary was found to correspond closely to the zero isoline of a climatic moisture index, based on mean annual precipitation minus potential evapotranspiration (PET) by the lensen-haise method. In this study, similar results were obtained using PET estimates from the more process-based Priestley-Taylor equation or a simplified form of the Penman-Monteith equation, although actual PET estimates were sensitive to input parameters. It is concluded, based on this analysis and other evidence, that forest distribution in the region is controlled by chronic moisture deficits. A simple method of estimating PET is presented that is suitable for assessing long-term changes in moisture regimes from limited climate data. Keywords: Boreal forest; Evapotranspiration; Drought; Climate change; Moisture 1. Introduction The boreal forest occupies nearly 30% of the Canadian landscape, and also extends over large areas of Alaska, Siberia and northern Europe. This forest type occurs in cold climatic regions, and in Canada, is typically dominated by conifers (e.g. Picea glauca, Picea mariana and Pinus banksiana) with only a few hardwood tree species. In eastern Canada, the boreal forest forms its southern bound- I Tel.: (403) ; fax: (403) ; thogg@nofc.forestry.ca. ary with temperate forests, with a high diversity of hardwoods. In western Canada, however, the boreal forest is bounded to the south by prairie grasslands, where coniferous trees are naturally absent, and hardwoods are mainly restricted to stunted groves of aspen (Populus tremuloides) in the more favourable sites. Recent model predictions indicate a warmer and possibly drier global climate in the future, as the levels of atmospheric carbon dioxide and other greenhouse gases continue to increase (e.g. Houghton et ai., 1990, 1992). These effects are expected to be greatest in northern continental regions (e.g. Manabe et ai., 1992), including much of the circumboreal /97/$ Elsevier Science B.V. All rights reserved. PlI SO (96)

2 116 E.H. Hogg / Agricultural and Forest Meteorology 84 (1997) region. This could lead to a prairie-like climate over parts of the southern boreal forest in western Canada (Emanuel et ai., 1985; Rizzo and Wiken, 1992; Hogg and Hurdle, 1995). Scaling issues are fundamental to understanding both (I) ecological responses of the boreal forest to changes in climate and (2) the feedbacks of the forest on atmospheric and climatic processes. BOREAS (Boreal Ecosystem-Atmosphere Study) is a large, ongoing field experiment in western Canada focusing on processes governing the interactions between the boreal forest and the atmosphere, with an emphasis on how these processes may respond to climate change (Sellers et ai., 1995). The experiment is interdisciplinary and has a spatially nested design, which should allow both physical and biological processes to be scaled up, from small spatial scales (e.g. individual leaves and trees) to much larger scales (landscape and region). BOREAS can be expected to lead to major improvements in the understanding of interscale relationships of diurnal and seasonal processes in the boreal forests of Canada and elsewhere. Predicting responses of the boreal forest to future climate change will also require a better insight into ecological responses to climate, many of which operate under much longer time scales (decades to centuries). These include, for example, the effects of climate on tree regeneration and growth, forest succession, fire regimes and the accumulation of carbon in peatlands. Under both northern and semi-arid climatic regimes, weather conditions are prone to high interannual variability at a given location. Within a given month or year, precipitation can be patchy even over relatively flat terrain, especially in regions such as western Canada where thunderstorms are common during the growing season. Often it is only when viewed from longer time scales that regional gradients in climate become apparent. Thus there is a temporal scaling problem in linking short-term measurements of forest-atmosphere interactions (e.g. BOREAS), with longer-term relationships between climate and ecosystem functioning. One of the simplest approaches to address the temporal scaling problem is to examine the spatial relationships that have developed over time between climate and natural vegetation zonation (e.g. Holdridge, 1947; Tukhanen, 1984). A previous analysis (Hogg, 1994) showed that the southern boundary of the boreal forest in western Canada corresponds poorly with thermal characteristics of climate, including July mean temperature and the number of growing degree days. In contrast, this boundary was found to fit very closely with the zero isoline of a climatic moisture index, calculated as mean annual precipitation (P) minus mean annual potential evapotranspiration (PET). The close fit between the P-PET climatic moisture index and vegetation zonation suggested that in western Canada, the southern boundary of the boreal forest is governed by moisture rather than temperature (Hogg, 1994). The Jensen-Haise method (Jensen et ai., 1990) is relatively simple to use, requiring only temperature, altitude and mean monthly global solar radiation. However, it is empirically derived from measurements taken over well-watered grasslands and agricultural crops in the western United States. In contrast, the Penman-Monteith equation (Monteith, 1981) is based on physical processes, and can be parameterized for any vegetation type, including forests (Jones, 1992). It is also a suitable basis for interscale, spatial comparisons (e.g. McNaughton and Jarvis, 1991) and has been used to estimate evapotranspiration in forest ecosystem process models (e.g. Running and Coughlan, 1988). Unfortunately, the Penman-Monteith equation normally uses detailed input data (e.g. net radiation, vapour pressure) that are not generally available from the long-term climate record, particularly in remote regions of the boreal forest. The equation also needs vegetationspecific estimates of surface conductance (both physiological and aerodynamic). Another approach to estimating PET is the Priestley-Taylor method (Priestley and Taylor, 1972), which is physically based but has fewer input requirements. However, it also uses net radiation and the parameter a, both of which are influenced by vegetation characteristics. The objective of this study was to develop a framework in which we could begin to link shortterm, intensive measurements of forest-atmosphere interactions to the issue of how moisture affects forest distribution over longer time scales. For this purpose, a simple climatic moisture index (P-PET) was developed, where PET is based on the principles of the Penman-Monteith equation, yet requires only

3 E.H. Hogg / Agricultural and Forest Meteorology 84 (1997) climate data that are readily available over long time periods (monthly mean daily maximum and minimum temperatures). Estimates of PET using the simplified equation are then compared against the Jensen-Haise and Priestley-Taylor methods, in terms of how well the zero isoline (of P-PET) corresponds to the southern boundary of the boreal forest in western Canada. For the purposes of this study, PET and climatic moisture indices are calculated by the three methods using generic parameters, and are thus independent of the specific vegetation type at each location. 2. Estimates of PET for climate moisture indices In the analysis, PET is defined as evapotranspiration from a well-vegetated landscape with adequate soil moisture. Using this definition, there is no requirement for wet leaf or soil surfaces, which usually have a transient occurrence in upland forests and grasslands. Mean monthly PET based on three different methods was determined for each of the 12 months using the long-term climate normals (Environment Canada, 1982a) from 254 stations in the region (station locations shown in Hogg, 1994). Mean annual PET was then determined as the sum of the monthly values. Finally, a climate moisture index was calculated for each station as P-PET (annual precipitation minus PET, with units in cm). The procedure used for estimating PET by the Jensen-Haise method has been described previously (Bonan, 1989; Jensen et ai., 1990; Hogg, 1994). The method is empirically based on evapotranspiration of reference crops in the western United States. For the Priestley-Taylor method, the following equation based on the parameter a, net radiation (Rn, in MJ m-2) and mean monthly temperature (Tmean) was used to estimate PET in mm month-i (ground heat flux was assumed negligible over long time periods): PET = a R n A (.1 + y ) (I) Rn was calculated from Tmean, global solar radiation and clear sky radiation using the procedure described by Jensen et al. (1990). Mean global solar radiation was obtained from a digitized version of the maps produced by McKay and Morris (1985). Albedo was estimated based on Tmean, as follows: Months warmer than 2 C were assumed to be snow-free with an albedo of 0.15, while months colder than - WOC (continuous ground snowcover) were given an albedo of Albedo was assumed to vary linearly for Tmean between - WOC and 2 C. Estimates of net long-wave radiation were based on Eqs. (3.16) and (3.17) in Jensen et al. (1990), using general coefficients recommended for semi-humid areas. PET estimates were then made using different values of a, including the value 1.26 for well-watered surfaces (Priestley and Taylor, 1972). The following procedure was used to develop a simple equation for PET based on the Penman Monteith equation. Normally, the Penman-Monteith equation includes terms for both available energy (net radiation) and vapour pressure deficit (Monteith, 1965). However, evapotranspiration in forests is typically more sensitive to vapour pressure deficit than to net radiation because of high aerodynamic conductances (g,). Using ga = 200 mm S-I, applicable to coniferous forests (Running and Coughlan, 1988), the net radiation term of the Penman-Monteith equation was found to contribute only 5.0 ± 0.4% of total PET for three stations examined in the region (Bad Lake, Saskatchewan; Churchill, Manitoba and Edmonton, Alberta) where long-term net radiation data are available (Environment Canada, 1982b). Thus for this application, the Penman-Monteith equation (as shown in Jensen et ai., 1990) could be simplified to the following, which requires only the mean maximum and minimum temperature for each month: PET = 1.05 kpcpdga A(.1 + y (1 +ga /gc)) ' (2) Methods for calculating most of the input variables and parameters for Eq. (2) (including the effect of altitude on mean atmospheric pressure) are given by Jensen et al. (1990). Vapour pressure deficit (D), was estimated from Tma x and Tm in using the following equation: D = 0.5 ( e * - e * ) - e * ( 3) T max T min Tdew As a first approximation, e;d< W was assumed to equal the saturation vapour pressure at a temperature of 2.5 C less than mean minimum temperature, corre-

4 118 E.H. Hogg / Agricultural and Forest Meteorology 84 (1997) cause g cs is expressed in standard units (mm s - 1 ), it is affected by the purely physical effects of temperature and pressure (altitude) at a given degree of stomatal opening. Corrected conductance (g J was thus calculated from conductance in standard units (g cs) according to Jones (1992, pp ). 3. Results The analysis showed a close correspondence between the southern boundary of the boreal forest in western Canada and the zero isoline of the P-PET climate moisture index, based on the simplified Penman-Monteith Eq. (2) with gcs = 5 mm S-I (Fig. I). The position of this isoline is very similar to that obtained previously using the Jensen-Haise method (Hogg, 1994). The Priestley-Taylor method (Eq. (1)) could also be parameterized (with a = 1.0) to give a zero isoline in nearly the same position (Fig. I). It is interesting to note that a similar result can be obtained using PET estimates based almost entirely on available energy (Eq. (I) or on estimates of 55 N 500N sponding to the mean nightly maximum relative humidity (May-September) of 84% at Saskatoon (Iocation shown in Fig. I). Canopy conductance (gc) is highly variable, being controlled largely by stomatal responses to environmental conditions, but typically ranges from I to 10 mm s - 1 in both forests (coniferous and deciduous) and native grasslands (Jones, 1992, p. 116). However, the apparent sensitivity of transpiration to changes in g c decreases as the spatial scale under investigation increases; McNaughton and Jarvis (1991) and their earlier analyses provide a more comprehensive theoretical framework for spatial scaling of the processes governing transpiration. In the present, highly simplified analysis, a standard mean canopy conductance (g cs) of 5 mm s - 1 was used for mean monthly temperatures > 10 C, but other values of gc s were also tested. In all analyses, a linear reduction in g cs to 0 mm s - 1 was assumed to occur as mean monthly temperature decreases from 10 C to - 5 C. This is meant to allow for the effects of cold or frozen soils, subfreezing night temperatures and dormancy on evaporation and conductance (Running and Coughlan, 1988). Bel1li CORDILLERAN FOREST BOREAL FOREST ZERO P-PET ISOLINES: PRIESTLEY -TAYLOR D GRASSLAND PENMAN-MONTEITH LAKES JENSEN-HAISE Fig. I. Correspondence of the southern limit of forest distribution in the interior of western Canada with the zero isolines of P - PET moisture indices based on the simplified Penman-Monteith equation (gc, = 5 mrn S-I), the Priestley - Taylor method (a = 1.0) and the lensen-haise method. Political boundaries of three Canadian provinces are shown (Alberta, Saskatchewan and Manitoba, from left to right). Forest outlier near 50oN, IIO"W belongs to the Cypress Hills (P-PET = + 5 cm by Penman-Monteith method, - 1 and -3 cm by the other methods).

5 E.H. Hogg / Agricultural and Forest Meteorology 84 (1997) Table I Estimates of annual potential evapotranspiration (PET) and climate moisture index (P-PET) based on three methods (see text for explanation), for locations in the grassland, along the forest -grassland boundary and in the boreal forest. Values used in determining the zero P-PET isolines in Fig. 1 are shown in bold Grassland Saskatoon a P (annual precipitation, mm) 349 PET (mm) lensen-haise method 518 Priestley-Taylor (a = 1.0) 473 Priestley-Taylor ( a = 1.26) 596 Penman-Monteith (g cs = 4 mm s - 1 ) 418 Penman-Monteith ( gcs = 5 mm s - I) 516 Penman-Monteith ( g cs = 6 mm s - 1 ) 611 P-PET (in cm) lensen-haise method -17 Priestley-Taylor (a = 1.0) -12 Priestley-Taylor (a = 1.26) -25 Penman-Monteith (g cs = 4 mm s - I ) -7 Penman-Monteith (g cs = 5 mm s - 1 ) Penman-Monteith ( gcs = 6 mm s - I) Forest-grassland boundary Forest Edmonton a Winnipeg a Flin Flon a a Station name. vapour pressure deficit (Eq. (2». However, it should also be noted that these estimates of PET are sensitive to the parameterizations used. In the simplified Penman-Monteith method, PET estimates increase in nearly direct proportion to values of g c s between 4 and 6 mm S-I (Table I). Similarly, PET estimates using the Priestley-Taylor method are directly proportional to a (Table I), and are strongly influenced by estimates of net radiation (especially albedo). Thus the position of the zero-line is also highly dependent on input parameters in both cases. In contrast, the PET estimates based on the more empirical lensen-haise equation are uniquely determined by a station's elevation, global solar radiation and temperature. For the simplified Penman-Monteith method, a subsequent analysis indicated that Eq. (2) could be further simplified to the following set of equations (where g c s = 5 mm S-I and ga = 200 mm S-I ), which gave nearly the same PET estimates (within ± 10 mm year-i for all stations): For Tmean> 10 : PET = 93 D exp( A/9300) For 10 > Tmean > - 5 : (4a) PET = (6.2 T + 31) Dexp( A/9300) (4b) For Tme n < - 5 : PET = 0 (4c) where PET is in mm month - 1, Tmean is mean monthly temperature (OC), D is vapour pressure deficit (kpa) as calculated from Eq. (3) and A is station altitude (m). 4. Discussion The results suggest that in the western Canadian interior, the boreal forest is generally restricted to areas where there is a surplus of precipitation over potential evapotranspiration. The boundary between the cordilleran forest and prairie near Calgary also corresponds to the zero isoline where P = PET using all three methods (Fig. n. It should be recognized that this preliminary analysis is based on long-term mean climate data only, and thus does not include the role of disturbance, extreme climate events, or differences in evapotranspiration among different vegetation types. At a minimum, however, it appears that the forest-grassland boundary in western Canada follows an isoline of climatically similar moisture regimes. Is there any independent evidence to support the concept of a long-term zero isoline of P-PET, that hydrologically separates forested regions from regions that are predominantly grassland? In western

6 120 E.H. Hogg / Agricultural and Forest Meteorology 84 (1997) Canada, the boundary between forest and grassland regions (postulated zero isoline) corresponds to a sharp regional gradient in the quantity of annual water runoff. Runoff increases rapidly from < 2.5 cm year-1 in the grassland to > 10 cm year-lover most of the boreal forest (Fisheries and Environment Canada, 1978). In the grassland, lakes tend to dry up during periods of drought, while freshwater lakes with relatively constant levels over the long-term predominate in the forested region (Campbell et ai., 1994). The forest-grassland boundary also corresponds to a sharp discontinuity in several landscape characteristics and processes, all of which appear to be sensitive to climatic moisture regimes. In the grassland region where chronic moisture deficits occur (P < PET), conifers and peatlands are generally absent (Zoltai, 1975; Hogg, 1994; Vitt et ai., 1994), while aspen is usually restricted to small patches and has a stunted growth form (Hogg and Hurdle, 1995). Soil zonation also corresponds closely to climatic moisture isolines (Sly, 1970). Disturbance regimes are linked to moisture, either directly or indirectly through long-term changes in vegetation and soils. Prior to European settlement, the grassland region was strongly disturbed by frequent prairie fire and by large populations of bison; this region continues to be frequently disturbed, but mainly by human activity (cultivation and livestock grazing). Although human settlement has greatly reduced the extent of fire in the grassland region, it continues to be the major natural agent of vegetation change in the forested regions. Under the predicted future climate, the zero P-PET isoline may move northward (Hogg and Hurdle, 1995), leading to a dramatic increase in the incidence of severe drought and forest fires (Flannigan and Van Wagner, 1991). The drought of , for example, caused fires to bum over 3 million ha of boreal forests and peatlands in northern Manitoba (Hirsch, 1991). Unusually poor regeneration of jack pine (Pinus banksiana) has been noted, following one of the large 1989 fires in Saskatchewan. This drought, coupled with defoliation by insects, has since led to massive dieback of aspen over large areas near the forest-grassland boundary in western Canada (E.H. Hogg, personal observation, 1995). These effects seem likely to increase in the future, if the predicted climate changes occur. They could also lead to major feedbacks on atmospheric COo levels (cf. Tans et ai., 1990) and in the energy balance of the regional landscape (e.g. Bonan et ai., 1992). Other feedbacks include immediate impacts, such as those produced by large-scale fire or drought, and longer-term responses, including chronic reductions in forest regeneration and growth, increased decomposition of forest soils and peatlands, and melting of permafrost. Such processes cannot be adequately assessed solely through intensive, shortterm ecophysiological, hydrologic and meteorological studies. Interscale analyses of past ecosystem responses to climate (e.g. Delcourt and Delcourt, 1991), coupled with simulation modelling, can be a powerful approach to understand these longer-term processes. This approach forms part of the Climate Change Initiative of the Canadian Forest Service. Current studies include dendrochronology, palynology and peat stratigraphy, surveys of decomposition rates, tree growth and biomass and seedling regeneration across regional climate gradients. Modelling initiatives include a carbon budget model (Kurz and Apps, 1994), and several forest growth models (e.g. Price et ai., 1993). One of the problems in assessing long-term climate effects, even on a time scale of years, is the scarcity of detailed weather data. Some data are available for this region, but early data are mainly limited to observations of temperature and precipitation. Increases in mean temperature over the past century are relatively simple to document (e.g. Gullett and Skinner, 1992). However, precipitation data alone are insufficient to document long-term fluctuations in moisture at the regional scale because temperature effects on PET are not considered. The simplified Penman-Monteith approach presented here can use the limited weather information to estimate PET on a monthly or annual time-step. Temporal changes in the climate moisture index (P-PET) can then be estimated and applied, for example, to studies of tree-ring responses to climatic moisture deficits. Other questions concerning long-term, regional processes remain to be answered. For example, is the southern boundary of the western Canadian boreal

7 E.H. Hogg / Agricultural and Forest Meteorology 84 (1997) forest determined directly by climate, through drought effects on tree regeneration and survival, or is it a result of a high, drought-induced fire frequency on the prairies in the past? If the present forest distribution is strictly climate limited, then even slight future changes in climate could have a major impact (Hogg, 1994). Feedbacks of vegetation on regional climate may also be important, as suggested, for example by models of Amazon deforestation (Shukla et a\., 1990), forest effects on mesoscale rainfall (Blyth et a\., 1994), and models of land use feedbacks on global desertification (Schlesinger et a\., 1990; Savenije, 1995). For western Canada, this leads to the question of how the observed climatic moisture gradient at the regional scale is affected by the distribution of forest and grassland across the landscape. Recent results from the BOREAS experiment in this region (Sellers et a\., 1995) also indicate major differences in transpiration rates among boreal vegetation types, being much lower in coniferous forests than in deciduous forests and fens. How might such differences be manifested over the long term, through mesoscale feedbacks on regional hydrology, climate and ecosystem functioning? In terms of vegetation-climate relationships, there are many such questions remaining, which can probably be addressed only through continued interdisciplinary research efforts, with a much greater emphasis on temporal scaling. 5. List of symbols A A p station altitude (m) parameter for Priestly-Taylor equation (dimensionless) psychometric constant (kpa C - 1 ) slope of saturation vapour pressure (kpa C-I) at Tm ean latent heat of vaporization (MJ kg - 1 ) at Tm ean density of air (kg m -3) at station pressure and Tm ean heat capacity of air (1.012 kj kg - 1 c - 1 ) vapour pressure deficit (kpa) saturation vapour pressure at Tmax saturation vapour pressure at Tmin saturation vapour pressure at mean dewpoint temperature g a aerodynamic conductance (mm s -1 ) g c corrected canopy physiological conductance (mm S-I ) g cs standard canopy physiological conductance (mm S-I ) k proportionality constant: s month - 1 /106 or 2.64 Rn net radiation (MJ m -2) Tmax mean daily maximum monthly temperature (OC) Tmin mean daily minimum monthly temperature COC) mean monthly temperature (OC) Tmean Acknowledgements This study was supported by Green Plan funding within the Canadian Forest Service, Department of Natural Resources Canada. I thank A. Robertson for encouragement of this work, and LD. Campbell, D. Halliwell, P.A. Hurdle, D.T. Price and S.C. Zoltai for useful discussions and comments on the manuscript. References Blyth, E.M., Dolman, A.J. and Noilhan, J., The effect of forest on mesoscale rainfall: an example from HAPEX MOBILHY. J. Appl. Meteorol., 33: Bonan, G.B., A computer model of the solar radiation, soil moisture, and soil thermal regimes in boreal forests. Ecol. Modell., 45: Bonan, G.B., Pollard, D. and Thompson, S.L., Effects of boreal forest vegetation on global climate. Nature, 359: Campbell, c., Campbell, LD. and Hogg, E.H., Lake area variability across a climatic and vegetational transect in southeastern Alberta. Geogr. Phys. Quatern., 48: Delcourt, H.R. and Delcourt, P.A., Quaternary Ecology. Chapman and Hall, London, 242 pp. Emanuel. W.R., Shugart, H.H. and Stevenson, M.P., Climatic change and the broad-scale distribution of terrestrial ecosystem complexes. Clim. Change, 7: Environment Canada, 1982a. Canadian Climatic Normals Temperature and Precipitation. Prairie Provinces. Atmospheric Environment Service, Downsview, Ontario, Canada, 429 pp. Environment Canada, 1982b. Canadian Climatic Normals Vol. I. Solar Radiation. Atmospheric Environment Service, Downsview, Ontario, Canada, 57 pp.

8 122 E.H. Hogg / Agricultural and Forest Meteorology 84 (1997) Fisheries and Environment Canada, Hydrological Atlas of Canada. Surveys and Mapping Branch, Dept. Energy, Mines and Resources, Ottawa, Canada. Flannigan, M.D. and Van Wagner, C.E., Climate change and wildfire in Canada. Can. J. For. Res., 21: Gullett, D.W. and Skinner, W.R, The state of Canada's climate: temperature change in Canada A State of the Environment Report, Environment Canada SOE Report No Supply and Services, Canada, 36 pp. Hirsch, KG., A chronological overview of the 1989 fire season in Manitoba. For. Chron., 67: Hogg, E.H., Climate and the southern limit of the western Canadian boreal forest. Can. J, For. Res., 24: Hogg, E.H. and Hurdle, P.A., The aspen parkland in western Canada: a dry-climate analogue for the future boreal forest? Water, Air Soil Pollut., 82: Holdridge, L.R, Determination of world plant formations from simple climatic data. Science, 105: Houghton, J.T., Jenkins, GJ. and Ephraums, JJ, (Editors), Climate Change. The IPCC Scientific Assessment. Cambridge University Press, Cambridge, UK, 365 pp. Houghton, J.T., Callander, B.A. and Varney, S.K, (Editors), Climate Change The Supplementary Report to the IPCC Scientific Assessment. Cambridge University Press, Cambridge, UK, 200 pp. Jensen, M.E., Burman, RD. and Allen, RG. (Editors), Evapotranspiration and Irrigation Water Requirements. Manuals and Reports on Engineering Practice No. 70, Am. Soc. Civ. Eng., New York, 332 pp. Jones, H.G., Plants and Microclimate, A Quantitative Approach to Environmental Plant Physiology. 2nd edn., Cambridge University Press, Cambridge, UK, 428 pp. Kurz, W.A. and Apps, MJ., The carbon budget of Canadian forests: a sensitivity analysis of changes in disturbance regimes, growth rates and decomposition rates. Environ. Pollut., 83: 55-6\. Manabe, S., Spelman, MJ. and Stouffer, RJ., Transient responses of a coupled ocean-atmosphere model to gradual changes of atmospheric CO2, Part II. Seasonal response. J. Clim.,5: McKay, D.C. and Morris, RJ., Solar Radiation Data Analyses for Canada Atmospheric Environment Service, Environment Canada, Ottawa. McNaughton, KG. and Jarvis, P.G., Effects of spatial scale on stomatal control of transpiration. Agric. For. Meteorol., 54: Monteith, J.L., Evaporation and environment. Symp. Soc. Exp. BioI., 19: Monteith, J.L., Evaporation and surface temperature. Q. Rev. R Meteorol. Soc., 107: Price, D.T., Apps, MJ., Kurz, W.A., Prentice, I.c. and Sykes, M.T., Simulating the carbon budget of the Canadian boreal forest using an integrated suite of process-based models. In: C.-H. Ung (Editor), Forest Growth Models and Their Uses. International Workshop, November Canadian Forest Service, Ste.-Foy, Quebec, pp Priestley, C.H.B. and Taylor, RJ., On the assessment of surface heat flux and evaporation using large-scale parameters. Mon. Weather Rev., 100: Rizzo, B. and Wiken, E., Assessing the sensitivity of Canada's ecosystems to climatic change. Clim. Change, 21: Running, S.W. and Coughlan, J.C., A general model of forest ecosystem processes for regional applications I. Hydrologic balance, canopy gas exchange and primary production processes. Ecol. Modell., 42: Savenije, H.H.G., New definitions for moisture recycling and the relationship with land-use changes in the Sahel. J. Hydro!., 167: Schlesinger, W.H., Reynolds, J.F., Cunningham, G.L., Huenneke, L.E, Jarrell, W.M., Virginia, RA. and Whitford, W.G., Biological feedbacks in global desertification. Science, 247: Sellers, PJ., Hall, E, Margolis, H., Kelly, B., Baldocchi, D., den Hartog, G., Cihlar, J., Ryan, M.G., Goodison, B., Crill, P., Ranson, KJ., Lettenmaier, D. and Wickland, D.E., The Boreal Ecosystem-Atmosphere Study (BOREAS): an overview and early results from the 1994 field year. Bull. Am. Meteoro!. Soc., 76: Shukla, J., Nobre, C. and Sellers, P., Amazon deforestation and climate change. Science, 247: Sly, W.K, A climatic moisture index for land and soil classification in Canada. Can. J. Soil Sci., 50: I. Tans, P.P., Fung, LY. and TakahaShi, T., Observational constraints on the global atmospheric CO2 budget. Science, 247: Tukhanen, S., A circumboreal system of climatic-phytogeographical regions. Acta Bot. Fenn., 127: I-50. Vitt, D.H., Halsey, L.A. and Zoitai, S.c., The bog landforms of continental western Canada in relation to climate and permafrost patterns. Arctic Alpine Res., 26: Zoltai, S.C., Southern limit of coniferous trees on the Canadian prairies. Information Report NOR-X-128, Canadian Forestry Service, Northern Forest Research Centre, Edmonton, Alberta, 12 pp.