* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Temperate Grassland Responses to Climate
Climate change in Tuvalu wikipedia , lookup
Global warming wikipedia , lookup
Pleistocene Park wikipedia , lookup
Attribution of recent climate change wikipedia , lookup
Solar radiation management wikipedia , lookup
Surveys of scientists' views on climate change wikipedia , lookup
Citizens' Climate Lobby wikipedia , lookup
Climate sensitivity wikipedia , lookup
Climate change and agriculture wikipedia , lookup
Climate change in the United States wikipedia , lookup
Effects of global warming on humans wikipedia , lookup
Climate change and poverty wikipedia , lookup
Climate change feedback wikipedia , lookup
General circulation model wikipedia , lookup
Climate change, industry and society wikipedia , lookup
Climate-friendly gardening wikipedia , lookup
Effects of global warming on human health wikipedia , lookup
Instrumental temperature record wikipedia , lookup
Annals of Botany 80 : 205–221, 1997 Temperate Grassland Responses to Climate Change : an Analysis using the Hurley Pasture Model J. H. M. T H O R N L E Y and M. G. R. C A N N E L L Institute of Terrestrial Ecology, Bush Estate, Penicuik, Midlothian EH26 0QB, UK Received : 6 January 1997 Accepted : 24 March 1997 The Hurley Pasture Model is process-based and couples the carbon, nitrogen and water cycles in the soil-grass-animal system. It was used to examine the responses of grasslands in southern, lowland and northern, upland climates in Britain. Short-term response to step-wise increases in CO concentration (350 to 700 µmol mol−") and temperature # (5 °C) were contrasted with long-term equilibrium (the term ‘ equilibrium ’ is equivalent to ‘ steady state ’ throughout this paper) responses and with responses to gradually increasing [CO ] and temperature. Equilibrium responses to a # range of climate variables were also examined. Three conclusions were drawn regarding the interpretation of experiments : (1) initial ecosystem responses to stepwise changes can be different in both magnitude and sign to equilibrium responses, and this can continue for many years ; (2) grazing can drastically alter the magnitude and sign of the response of grasslands to climate change, especially rising temperatures ; and (3) effects of changes in climate, especially temperature and rainfall, are likely to be highly site-specific. It was concluded that experiments should try to lessen uncertainty about processes within models rather than try to predict ecosystem responses directly. Three conclusions were also drawn about the operation of grasslands as carbon sinks : (1) increasing [CO ] alone # will produce a carbon sink, as long as it continues to accelerate photosynthesis and increase net primary productivity ; (2) by contrast, increasing temperatures alone are likely to produce a carbon source, because soil respiration is accelerated more than net primary productivity, even when assuming the same temperature function for most soil and plant biochemical processes ; and (3) the net effect of projected increases in [CO ] and temperature is # likely to be a carbon sink of 5–15 g C m−# yr−" in humid, temperate grasslands for several decades, which is consistent with the magnitude of the hypothesized current global terrestrial carbon sink. # 1997 Annals of Botany Company Key words : Grassland, climate change, carbon dioxide, temperature, ecosystem, model, carbon sink. INTRODUCTION The problem of predicting the impact of gradually increasing temperature and [CO ] on the productivity of grasslands, # and on whether grassland soils are carbon sinks or sources, will not easily be solved by experimentation. Firstly, most experiments are necessarily short-term (less than a decade) compared with the time constants of many ecosystem processes, notably those in the soil which determine N supplies to the grass and carbon storage in the soil. Consequently, there is a risk that experiments observe transient responses, unlike the real-world response of ecosystems to gradual climate change, which is a process of continual adjustment towards an unattained new equilibrium. Secondly, few experiments include grazing, although several include periodic cutting (Newton et al., 1995, 1996 ; Jones, Jongen and Doyle, 1996), which in fact makes the climate responses difficult to interpret owing to transient responses to defoliation (Thornley, 1998). Thirdly, all experiments must be constrained to a limited number of treatments and no two experiments are exactly alike, making comparisons difficult and leading to apparently conflicting results which may or may not be valid. 0305-7364}97}07020517 $25.00}0 Ecosystem modelling provides a feasible way forward. Several workers have used models to predict climate change effects on grasslands, notably Hunt et al. (1991), who focused on semi-arid prairie, Van den Pol van Dasselaar and Lantinga (1995) who examined grasslands in the Netherlands, and a SCOPE project group, which used the CCGRASS and CENTURY models to quantify the effects of GCM predicted climates on grasslands worldwide (Breymeyer et al., 1996). These workers used models that are substantially less comprehensive than the Hurley Pasture Model (see below and the Appendix) and generally concentrated on differences between current and 2¬CO # climates. In this paper, we use the Hurley Pasture Model (Johnson and Thornley, 1985 ; Thornley and Verberne, 1989 ; Thornley and Johnson, 1990 ; Thornley 1996, 1998) to explore differences between transient and equilibrium climate responses and to quantify the separate and combined effects of increasing temperature and [CO ] on the net # primary productivity (NPP) and total carbon content (in the plant-soil system, Csys) of grasslands growing in two contrasting climates in Britain. bo970430 # 1997 Annals of Botany Company 206 Thornley and Cannell—Grassland Responses to Climate Change MODEL PROPERTIES AND CLIMATE SCENARIOS Hurley Pasture Model The version of the Hurley Pasture Model used here is fully described by Thornley (1998) including all equations and parameter values. A synopsis of the processes represented is given in the Appendix, including some source references, critical dependencies and parameters. The model is generic, deterministic, dynamic and largely mechanistic. It describes the fluxes of carbon (C), nitrogen (N) and water in a grazed soil-pasture-atmosphere system, coupling the C and N fluxes, modulating them by the plant and soil water status and by other features of the environment. Linked soil, plant and animal submodels represent the flows of C and N, a water submodel calculates the flow and status of water in the system, and an environment and management module provides the driving variables. Several other process-based grassland models have been reported [e.g. ELM (Innis, 1978), CENTURY (Parton et al., 1987, 1996), ERHYM-II (White, 1987), GEM (Hunt et al., 1991), CCGRASS (Verberne, 1992) and GEMT (Chen and Coughenour, 1994)]. The Hurley Pasture Model is distinctive in its comprehensive treatment of over 50 ecosystem processes, coupling of C, N and water fluxes in both plants and soil, explicit representation of assimilate transport and utilization, dynamic involvement of soil biomass in soil processes, modulation of all soil and plant biochemical processes by temperature and water status, and the inclusion of grazing animals. The model is complex because it includes many processes, none of which we regard as redundant to its purpose. To minimize complexity and achieve balance, all equations are as simple as possible— capturing only the essential responses and interactions— often more simply than in specialist models (e.g. of photosynthesis, stomatal conductance, light interception, soil hydrology and animal behaviour). Individual processes were modelled to respond to environmental and internal driving variables according to observations (e.g. photosynthetic responses). Processes were then assembled into submodels, and submodels into the whole model, at each stage ensuring that the simulations matched observed diurnal, seasonal and long term values and trends in the main measured variables in UK grasslands (LAI, NPP, root-shoot partitioning, soil N fluxes, C}N ratios, soil biomass, organic matter etc. ; Whitehead, 1995). Any discrepancies were remedied by re-examining the assumptions and altering the ways in which processes were represented and parameterized, until the dynamics of the system were stable and precisely determined by the underlying equations. During this process, the structure of the soil submodel was substantially modified from an earlier version (Thornley and Verberne, 1989) in order to simulate observed long term trends in soil organic matter at Rothamsted (Arah et al., 1997). However, the model was not constrained to fit any particular dataset, but was instead parameterized to give predictions that were qualitatively as observed and quantitatively within measured ranges, including responses to cutting, grazing, fertilizer application and changes in weather within a temperate climate (Whitehead, 1995 ; Thornley, 1998). It was assumed that the grass was responsive to elevated [CO ], with no downregulation, as observed for L. perenne # and some other grass species (Lutze and Gifford, 1995 ; T 1. Baseline climates and management conditions used for a lowland site in southern Britain (Hurley) and an upland site in northern Britain (Eskdalemuir) Latitude (°N) and [elevation (m)] N input (kg ha−" yr−") Windspeed at 2 m height (m s−") Relative humidity : Mean a.m. Annual variation in a.m. Mean p.m. Annual variation in p.m. PAR (MJ m−# d−") Mean Annual variation Air temperature (°C) Mean max. Annual variation in max. Mean min. Annual variation in min. Soil temperature (°C) Mean Annual variation Rainfall (m yr−") Mean Annual variation Grazing (sheep ha−"), continuous Hurley, England (southern, lowland) Eskdalemuir, Scotland (northern, upland) 51±54° [50] 50 1 55±27° [242] 20 2 0±80 0±1 0±7 0±1 0±84 0±07 0±75 0±13 4±8 4±1 3±8 3±1 14.0 7±0 6±5 5±5 11.0 7±0 4±0 5±0 10±0 6±0 8±5 6±0 0±66 0±22 10 1±53 0±51 5 207 Thornley and Cannell—Grassland Responses to Climate Change RESPONSES TO STEP CHANGES IN T E M P E R A T U R E A N D [C O ] C O M P A R E D # WITH EQUILIBRIUM RESPONSES Responses to step changes in climate were best revealed when simulating a grassland in the southern, lowland climate, without grazing. Qualitatively similar changes in underlying ecosystem properties were found in the northern, upland climate and with grazing, but with different effects on NPP (see Fig. 3, discussed below). Figure 1 shows the modelled responses of critical ecosystem properties to a step-wise increase of 5 °C in temperature and a doubling of [CO ] from 350 to # kg C m–2 kg C m–2 per year m2 m–2 Annual mean air and soil temperatures increased by 5°C Atmospheric CO2 increased from 350 to 700 µmol/mol Initial equil. Initial equil. 0 Long-term equilibrium A Total carbon (Csys) B Soil resp. 10 Long-term equilibrium H Csys I NPP 1.5 1.0 0.5 g N m–2 The model simulated grasslands continuously grazed by sheep in climates approximating the 30 year average meteorological conditions at Hurley in southern, lowland Britain and at Eskdalemuir in northern, upland Britain (Table 1). The soil was assumed to be a loam, with 10 % clay, having a soil water content of 40 % at field capacity. The magnitude of the soil pools at the start of each model run was that obtained at equilibrium in the climate. Daylength was calculated from latitude and time of year. Windspeed and daily N input from the atmosphere and were assumed to be constant. No fertilizers were applied. Other quantities were varied sinusoidally throughout the year : means and annual variations are given in Table 1. Photosynthetically active radiation (PAR) was maximal on 21 June, temperatures on 26 July and daily rainfall on 20 November. Daily variation in PAR was assumed to be sinusoidal between dawn and dusk, and in air temperature between dawn and 1500 h (France and Thornley, 1984). Soil temperatures were diurnally constant. Relative humidities varied seasonally and diurnally, with maxima on 30 December and at dawn. Climate treatments were of three kinds : (1) the model was run to equilibrium (when total annual C and N outputs equalled inputs) and was then subjected to a step change in temperature (5 °C—representing an extreme, in order to demonstrate the response) or [CO ] (350 to 700 µmol mol−") ; # (2) the model was run to equilibrium after changing the baseline climate in various ways ; and (3) the model was run to equilibrium in the baseline climate and then [CO ] and # temperatures were increased gradually to approximate the IPCC 1992 a climate change scenarios (Houghton et al., 1996). g N m–2 per year Baseline climates and treatments 20 8 Soil resp. NPP C LAI J LAI Soil mineral N K Soil mineral N L N leaching M S/R ratio N System C/N 4 0 6 D 4 2 0 E 8 N leaching 2 1 4 0 kg C kg–1 N kg kg–1 Casella, Soussana and Loiseau, 1996 ; Jones et al., 1996), unlike alpine and Mediterranean grasses according to Korner and Miglietta (1994) and Shappi and Korner (1996). The commonly observed responses of plants to elevated [CO ], such as decreased stomatal conductance, increased # photosynthetic temperature optimum, increased leaf thickness, increased tissue C}N ratio and increased partitioning to roots in N limiting conditions, are all simulated in the model (Baxter et al., 1994 ; Idso and Idso, 1994 ; Luo, Field and Mooney, 1994 ; Lutze and Gifford, 1995 ; Newton et al., 1995, 1996 ; Wolfenden and Diggle, 1995). F S/R ratio G System C/N 0 8 4 0 14 12 0 0 10 20 30 Years 0 10 20 30 Years F. 1. Modelled responses of an ungrazed grassland in southern, lowland Britain to step-wise changes in temperature and [CO ]. Left, # air and soil temperatures are increased by 5 °C. Right, [CO ] is doubled # − from 350 to 700 µmol mol ". The starting values are for the steady state (equilibrium) in the baseline climate (Table 1) and the long-term equilibrium values are shown in the changed climate. NPP is the instantaneous net primary productivity above and belowground accumulated over a year. The other variables are annual averages : total carbon in the soil-plant system (over 95 % in the soil) Csys ; soil respiration rate ; leaf area index (LAI) ; amount of soil mineral nitrogen ; amount of soil nitrogen leached ; shoot}root ratio ; total system carbon}nitrogen ratio (dominated by the soil). 700 µmol mol−". The model began with the system at equilibrium in the baseline climate (Table 1) and was run to equilibrium in the changed climate (Fig. 1). Table 2 summarizes the transient and persistent (equilibrium) changes in ecosystem properties and Fig. 2 shows, diagrammatically, the sign of equilibrium changes in the main processes which altered NPP and Csys. Temperature responses An increase in temperature affected almost all processes in the model, but its most powerful effects were (1) to 208 Thornley and Cannell—Grassland Responses to Climate Change T 2. Contrasting short and long-term responses of grassland ecosystem properties to increases in temperature and atmospheric CO # Immediate response to step-wise increase in temperature Increase Long-term equilibrium response N leaching System C}N ratio Decrease Soil mineral N Shoot}root ratio Soil respiration Shoot}root ratio LAI Total system C NPP Temperature ↑ Leaf longevity ↓ LAI ↓ Drainage ↑ N leaching ↑ (Volatilized N ↑) Photosynthesis, respiration Soil organic matter ↓ and growth ↑ NPP (see Fig.3) System carbon ↓ Soil mineral N ↓ System C/N ratio ↓ CO2 ↑ Stomatal conductance ↓ Drainage ↓ N leaching ↓ LAI ↑ Photosynthesis ↑ Evapotranspiration ↑ or unchanged Soil mineral N concentration unchanged NPP ↑ Plant biomass and litter C/N ↑ Soil organic matter ↑ (Nitrification N loss ↑) System carbon ↑ # Decrease Increase No change Immediate response to step-wise increase in CO System C/N ↑ F. 2. Directional changes, at equilibrium, in grassland ecosystem properties in response to increases in temperature and [CO ]. These # changes occurred in both baseline climates, with and without grazing (the southern climate without grazing is shown in Fig. 1), but changes in net annual primary productivity (NPP) in response to temperature depended on the temperature range, climate and grazing, as shown in Fig. 3. increase the rate of leaf senescence and hence decrease leaf longevity, and (2) to increase soil respiration} decomposition}mineralization rates. In the model, these two primary effects had a number of interrelated short and long-term consequences (Figs 1 and 2) : (a) At equilibrium, Increase Decrease NPP, LAI, System C}N Soil respiration Total system C — — — Soil mineral N Shoot}root ratio N leaching NPP is the sum of the above and belowground litter inputs to the soil (plus animal intake in grazed systems) and is exactly balanced by soil respiration. In the model, a step increase in temperature always increased soil respiration more than NPP, even at high LAI. The result was a net loss of soil carbon, which occurred rapidly for a few years and then progressed slowly until a new lower equilibrium total system carbon level (Csys) was reached, when soil respiration was once again in balance with NPP (Fig. 1 A and B). The difference in temperature response of soil respiration and NPP occurred although the rates of all plant and soil biochemical processes (other than quantum efficiency of photosynthesis) were modified by the same cubic temperature function (see Appendix). This difference is critical in determining the magnitude of the carbon sink (Rotmans and den Elzen, 1993 ; Kirschbaum, 1995). (b) Decreased leaf longevity caused an immediate and sustained decrease in LAI (Fig. 1 C). When LAI fell below about 5 (which would be the equilibrium state at warmer or drier sites, or of grazed swards) both NPP and evapotranspiration were decreased. (c) In the short term, faster mineralization produced large amounts of soil mineral N, substantial amounts of which were leached, partly because the drop in LAI resulted in decreased evapotranspiration and increased drainage (Fig. 1 D and E). (d ) In the longer term (at equilibrium), higher temperatures decreased the amount of soil mineral N. Although mineralization occurred faster at higher temperatures, the total soil organic matter pool was smaller, more N was continually being lost by leaching, and higher temperatures increased N losses by ammonium volatilization (Figs 1 D and 2). (e) In the short term, high soil mineral N levels resulted in increased assimilate partitioning to shoots and an increase in shoot}root ratio, but in the long term this ratio decreased or changed relatively little (Fig. 1). ( f ) In the short term, increased mineralization caused N enrichment and a fall in the system C}N ratio, but in the long term the depletion of N by leaching and volatilization resulted in an increase in the system (mainly soil) C}N ratio (Fig. 1 G). CO responses # In the model, the only effects of increasing [CO ] were to # (1) increase photosynthesis, and (2) decrease stomatal conductance. However, as for temperature, there were Thornley and Cannell—Grassland Responses to Climate Change profound consequent changes in N relations and other ecosystem properties (Fig. 2). Figure 1 shows that most changes were the opposite of those caused by an increase in temperature : (a) A step-wise increase in photosynthesis immediately increased NPP and litter input to the soil, followed by an increase in soil respiration (Fig. 1 I). This lag between increased NPP and soil respiration resulted in an increase in soil carbon (including soil biomass) and hence in Csys towards a new equilibrium value (Fig. 1 H). This lag makes it inevitable that continually increasing [CO ] will # always be accompanied by an increase in ecosystem carbon storage, that is, in a carbon sink (Lloyd and Farquhar, 1996). (b) In the model, increased [CO ] and photosynthesis # also increased LAI (Fig. 1 J). (In grazed swards, elevated CO can decrease LAI in the short term, especially in N# poor conditions, owing to increased allocation to roots, but at equilibrium LAI is generally increased.) The increased LAI further increased NPP in swards with LAIs less than about 5 (see further discussion below). (c) Although stomatal conductances decreased in elevated [CO ], evapotrans# piration increased because the increase in LAI caused an increase in total canopy conductance. (d ) In the short term, an increase in carbon substrate in the plants resulted in increased allocation to roots (Fig. 1 M), greater N uptake and depletion of the soil mineral N pools (Fig. 1 K). Less N was leached (Fig. 1 L), partly because of this depletion and partly because greater evapotranspiration resulted in less drainage (the opposite of what occurred in response to increased temperature). (e) In the longer term, both soil mineral N levels (total amount of mineral N m−#) and shoot}root ratios recovered towards a level similar to their original equilibrium values (Fig. 1 K and M). That is, a balance was reestablished between processes which increased mineral N levels (e.g. more soil organic matter for mineralization and less drainage loss) and those which decreased them (e.g. greater plant uptake and greater nitrification losses because of greater soil biomass). ( f ) Elevated CO increased plant and litter C}N ratios (which # contributed to the lag in soil respiration) and increased the system C}N ratio to a new equilibrium value (Fig. 1 N). Interim conclusions Three points should be noted at this stage. First, the immediate changes in ecosystem properties following stepwise increases in temperature and CO were not only # different in magnitude to equilibrium changes, but were also sometimes different in sign. This was particularly true for temperature and for variables such as soil mineral N and shoot}root ratio (see Table 2). Secondly, the model predicted that a rise in temperature increased total soil respiration more than NPP, producing a net carbon source, whereas a rise in [CO ] increased NPP more than soil respiration, # producing a net carbon sink. Thirdly, effects of temperature and [CO ] on leaf area index were potentially critical in # determining grassland responses to elevated temperature and [CO ] (see Fig. 2). Decreasing LAI in response to # temperature had the potential to lower NPP, because less light was intercepted and more N was leached (owing to lowered evapotranspiration, increased drainage and faster 209 mineralization). By contrast, increasing LAI in response to elevated [CO ] had the potential to increase NPP, because # more light was intercepted and less N was leached— remembering that canopy evapotranspiration could be increased despite lowered stomatal conductance. The importance of changes in LAI indicated the potential importance of grazing in determining responses to climate change. EQUILIBRIUM RESPONSES OF NPP AND C A R B O N S T O R A G E (Csys) T O T E M P E R A T U R E A N D [CO ] : S I T E # DIFFERENCES AND EFFECTS OF GRAZING Clearly, the magnitude of the long-term equilibrium values to which grasslands tend, in response to slowly changing climate, depend on the magnitude of the increases in temperature and [CO ]. Figure 3 A shows the equilibrium # values of NPP, LAI and Csys predicted in response to temperature, varying 5 °C above and below the baseline annual mean air temperatures (shown with arrows ; with equivalent changes in soil temperatures) at both sites, with and without grazing. Figure 3 B shows the equilibrium values in response to [CO ] varying from 350 to # 800 µmol mol−". Ungrazed swards In the absence of grazing, in the baseline climates, the model predicted equilibrium LAIs of 6 in the southern climate and over 10 in the northern climate. (The model did not simulate flowering, and so allowed ungrazed swards to reach higher LAIs than would occur in nature.) LAIs of 5 are sufficient to intercept most solar radiation. LAIs fell below this value in ungrazed swards only when the southern climate was warmed to mean air temperatures above 12 °C (Fig. 3 A). Thus, with the latter exception, changes to NPP in ungrazed swards in response to temperature and [CO ] # were not very sensitive to changes in LAI. They occurred primarily in response to temperature—through effects on photosynthesis, respiration, growth, soil N and water status (the southern climate was markedly drier)—and in response to elevated [CO ] through effects on photosynthesis and # stomatal conductance. In both climates, the NPP of ungrazed swards increased in response to warming to maxima at 12–13 °C. Over the range 5–12 °C, NPPs were higher in the upland climate, and increased more in response to temperature over this range (38 % compared with 20 %, Fig. 3 A), because stomatal closure occurred less in the upland climate as a result of higher rainfall and humidity. The decrease in Csys with temperature, over the range 5–12 °C, was correspondingly greater in the lowland climate (2±0 compared with 1±5 kg C m−# loss per °C). Without grazing, NPP increased less in response to a doubling of [CO ] in the upland climate than in the lowland # climate (34 % compared with 45 %, Fig. 3 B) because of the greater response of photosynthesis to elevated [CO ] at # higher temperatures. Csys increased correspondingly less in 210 Thornley and Cannell—Grassland Responses to Climate Change Without grazing Northern, upland 2.0 30 NPP 20 Without grazing 20 1.0 1.0 10 Csys >10 2.0 10 >10 LAI With grazing With grazing 1.0 20 20 NPP 10 1.0 0.5 10 Csys 0.5 LAI 0 5 10 15 20 Temperature (°C) 0 5 10 15 Temperature (°C) Southern, lowland Leaf area index (LAI); System carbon (Csys) (kg m–2) Leaf area index (LAI); System carbon (Csys) (kg m–2) Southern, lowland 30 30 Without grazing NPP Northern, upland 2.0 30 Without grazing 1.5 20 1.5 20 Csys 1.0 10 LAI 1.0 10 LAI > 10 0.5 With grazing 20 1.5 NPP 0.5 With grazing 20 1.0 10 2.0 Csys 1.5 1.0 10 0.5 0.5 LAI 0 400 600 800 [CO2] µmol mol–1 0.0 0 400 600 800 –1 [CO2] µmol mol 0.0 Net annual primary productivity (NPP) (kg C m–2 per year) B Net annual primary productivity (NPP) (kg C m–2 per year) A F. 3. Modelled equilibrium responses of British grasslands to changes in temperature (A) and [CO ] (B), with and without grazing. # Temperatures were varied equally for the soil and air ; those shown are annual mean air temperatures and arrows mark the baseline values (Table 1). The quantities shown are annual net primary productivity (NPP), total system carbon (Csys ) and leaf area index (LAI). Grazing Part of plant N returned to soil in urine and faeces LAI ↓ Part of NPP eaten and respired Evapotranspiration ↓ NPP ↓ System carbon ↓ Drainage ↑ N leaching ↑ N volatilization ↑ System C/N ratio ↓ F. 4. Directional changes in grassland ecosystem properties in response to grazing. the upland climate (21 % compared with 42 % in the lowland climate). In short, the NPP of ungrazed grasslands in the cool, wet, northern climate responded most to temperature (reducing the loss of carbon relative to that at the southern site) and least to elevated [CO ] (increasing the gain in carbon relative # to that at the southern site). Grazed swards As expected, grazing per se (in the absence of climate change) had major effects on the ecosystem, principally by (1) decreasing LAI (grazing was a function of LAI, independent of temperature) ; (2) consuming part of the plant NPP and losing it in animal respiration rather than transferring it to the soil ; and (3) returning most of the consumed N as urine and faeces (see Fig. 4). The consequent effects on the ecosystem were to lower NPP substantially and to increase N losses by leaching (because of greater drainage) and volatilization (from animal excreta) so that, at equilibrium, Csys and system C}N ratios were lower in grazed than in ungrazed swards (Fig. 4). Figure 3 A shows that, with grazing, NPPs were generally much lower than without grazing, but increased at low temperatures up to 5–7 °C, because of increased rates of photosynthesis and growth (as in ungrazed swards). But above 5–7 °C, the NPP of grazed swards decreased because the decline in LAI (resulting from shorter leaf longevity) outweighed the positive effects of temperature on photosynthesis and growth. Consequently, NPP was 25 % lower in the warm, southern baseline climate than in the northern baseline climate (marked with arrows in Fig. 3), and, in contrast to the ungrazed swards, any increase in temperature above the baseline values decreased NPP in both climates (by about 0±06 kg C m−# per year per °C). As expected, (1) the grazed swards contained markedly less carbon than the ungrazed ones (Csys) ; (2) the upland grazed grassland contained much more carbon than the lowland one ; and (3) both suffered a dramatic decrease in Csys with increase in mean annual air temperature above 5 °C (over 1 kg C m−# decrease per °C in the upland climate). Figure 3 B shows that, in grazed swards, increases in NPP, LAI and Csys in response to elevated [CO ] were greater at # the upland than the lowland site. (Whereas, without grazing, NPP, LAI and Csys increased most at the lowland site.) This occurred because the increase in LAI in response to elevated [CO ] was greater at cool temperatures, and increases in # LAI above the low values in grazed swards increased light interception, NPP and Csys. (By contrast, without grazing, increases in LAI had little effect on NPP and Csys.) Thornley and Cannell—Grassland Responses to Climate Change EQUILIBRIUM RESPONSES TO A RANGE OF CLIMATE AND MANAGEMENT VARIABLES WITH GRAZING Figure 5 gives the equilibrium values of NPP, LAI, Csys and annual animal intake (Ci,an) in the baseline climates (row 6) and shows the effects on equilibrium values of changes in a number of climate and management variables. The rows are ordered according to equilibrium Csys in the southern climate. In all conditions, the northern, upland grassland was more productive in NPP (though not in Ci,an) owing mainly to larger LAIs at lower temperatures (and with only five sheep ha−" compared with ten sheep ha−" at the southern site), despite lower N inputs and solar radiation levels (Table 1). Increasing rainfall and windspeed (affecting the water balance) had little effect on the grasslands—a 30 % increase in summer rainfall in the southern climate did not prevent water stress. But it is interesting that decreasing rainfall had a deleterious effect at the southern site (because of increased water stress) but a beneficial effect at the northern site (because of decreased N leaching). Increasing water stress by decreasing relative humidity by 10 % decreased productivity at both sites, and disabling the model to prevent N leaching increased productivity at both sites. The treatments which decreased Csys most were increases in temperature, because of decreased LAI, NPP and accelerated soil organic matter decomposition, and animal grazing, because of severely reduced LAI and NPP. The treatments which increased Csys most were increases in CO , # owing to greatly increased NPP (without a proportionate increase in LAI) and N inputs, especially in the growing season, owing to large increases in LAI and NPP. Southern, lowland PRODUCTIVITY AND CARBON STORAGE IN RESPONSE TO GRADUALLY I N C R E A S I N G [C O ] A N D T E M P E R A T U R E # The upper half of Fig. 6 shows the simple linear increases in [CO ] and mean annual air temperature which were imposed # on grazed swards in the lowland and upland climates for 50 years, after bringing the ecosystems to equilibrium in the baseline climates at time zero. As mentioned, at equilibrium, the NPP of the upland grassland was greater than that of the lowland grassland (Fig. 3 A), so the two ecosystems had different starting values in Fig. 6. A linear increase in [CO ] increased NPP at # both sites, but, as explained above (Fig. 3 B) the upland grassland responded most. Also, at the upland site, a linear increase in temperature (above the annual mean of 7±5 °C) slightly increased NPP, whereas at the lowland site an increase in temperature (above the annual mean of 10±25 °C) slightly decreased NPP (Fig. 6). (Note that, in the long term, at equilibrium, an increase in temperature decreased NPP at the upland site.) The net effect was that, at the upland site, increasing [CO ] and temperature together increased NPP more than # increasing [CO ] alone, and the absolute rate of increase in # NPP in response to climate change was greater at the upland site. Also, decreasing the rate of temperature rise from 0±03 to 0±02 °C yr−" decreased the overall response of the upland grassland but had a small beneficial effect on the lowland grassland (Fig. 6). The effect of gradual climate change on changes in total carbon storage (Csys) is shown in Fig. 7. As expected, rising temperature caused a loss of carbon, at an ever-increasing rate with increase in temperature, whereas elevating [CO ] # Northern, upland Row Csys LAI NPP Ci,an Csys LAI NPP 1 3.14 0.77 0.46 0.127 6.24 1.38 0.67 0.120 Temperature, +2°C 2 3.26 0.60 0.38 0.118 5.99 0.92 0.47 0.117 Animal stocking, +50% 3 4.20 0.88 0.52 0.153 8.25 2.33 0.76 0.155 Relative humidity, –0.1 4 4.41 0.91 0.55 0.159 8.46 2.51 0.79 0.160 Rainfall –30% 5 4.48 0.93 0.57 0.164 8.22 2.28 0.75 0.152 Rainfall +30% 6 4.51 0.94 0.57 0.165 8.36 2.42 0.77 0.157 Standard conditions 7 4.63 0.95 0.59 0.168 8.27 2.31 0.76 0.154 Windspeed × 2 8 4.97 1.08 0.65 0.187 9.04 3.10 0.87 0.178 No leaching 9 5.93 1.24 0.80 0.203 9.92 3.43 1.00 0.184 Radiation (PAR), +20% 10 7.94 2.79 1.10 0.298 13.58 9.95 1.35 0.196 +300 kg N ha–1 year–1, continuous 11 8.14 3.00 1.13 0.306 13.69 10.18 1.36 0.196 3 * 100 kg N ha–1, 2 Mar, 1 May, 30 Jun 12 9.14 1.79 1.28 0.241 14.33 CO2, 700 µmol mol–1 0 5 10 15 Sequestered C (kg C m–2) 211 Ci,an 8.15 1.53 0.199 Environment and management 0 5 10 15 Sequestered C (kg C m–2) F. 5. Modelled equilibrium changes in the amount of carbon sequestered in the system (Csys, the histograms) and some other properties of grazed British grasslands in response to changes in environmental and management variables relative to the baseline conditions in Table 1. The other properties are leaf area index (LAI), net annual primary production (NPP, kg C m−# per year), and the annual carbon input to grazing animals (Ci,an, kg C m−# per year). 212 Thornley and Cannell—Grassland Responses to Climate Change r C pe 0.03° year 0.02°C per 450 11 10 CO2 400 CO2 C 0.03° 9 ear per y 8 ear per y 0.02°C 350 10 20 30 40 50 0 10 20 30 40 50 7 1.1 1.1 CO2 and temp 1.0 1.0 0.9 0.9 CO2 0.8 0.8 CO2 Temp 0.7 CO2 and temp 0.7 0.6 Temp 0.6 0.5 0 10 20 30 Years 40 50 0 10 20 30 Years 40 50 0.5 Mean annual air temperature (°C) 12 year 0 Net annual primary productivity (kg C m–2 per year) Northern, upland Net annual primary productivity (kg C m–2 per year) Atmospheric [CO2] µmol mol–1 Southern, lowland 500 F. 6. Modelled changes in annual net primary productivity of grazed grasslands in lowland and upland Britain in response to increasing temperature (0±03 and 0±02 °C yr−") and CO (3 µmol mol−" yr-"). The systems were at equilibrium at year zero in the baseline climates defined # in Table 1. System carbon gain/loss (g C m–2 per year) Southern, lowland Sink 20 CO2 Northern, upland 20 CO2 10 CO2 and 0.02°C per year CO2 and 0.02°C per year 10 CO2 and 0.03°C per year 0 CO2 and 0.03°C per year 0 0.02°C per year 0.02°C per year –10 Source 0.03°C per year 0 10 20 30 40 50 Years –10 0.03°C per year 0 10 20 30 40 50 Years F. 7. Modelled annual changes in the amount of carbon in lowland and upland British grasslands (plants and soil) in response to increasing temperature (0±03 and 0±02 °C yr−") and CO (3 µmol mol−" yr−"). The # systems were at equilibrium at year zero in the baseline climates defined in Table 1. caused an accumulation of carbon towards an asymptote at high [CO ]. The net effect of increasing [CO ] and # # temperature together was to produce a carbon sink of around 10 g C m−# yr−", which was greater with slow rather than rapid warming, and which decreased with time as the response to [CO ] levelled off. Differences in the responses # of the two sites can be explained in the same way as the equilibrium responses discussed above. CONCLUSIONS AND DISCUSSION This analysis advances discussion on two aspects of the impacts of climate change on grasslands, namely on the interpretation of experiments at the ecosystem level, especially on temperature effects, and the role of temperate grasslands as carbon sinks in response to increasing temperature and [CO ]. # Interpretation of experiments Although experiments on elevated [CO ] and temperature # effects on photosynthesis and other individual processes may give interpretable results, it is more difficult to determine responses at the ecosystem level by experimentation because : (1) short-term experiments may observe misleading transient responses ; (2) it is difficult to simulate animal grazing ; and (3) effects on NPP can be highly specific to the conditions of the experiment. These three points are elaborated below. Initial ecosystem responses to step changes in temperature and [CO ] can be ery different in magnitude, and een be # different in sign, to equilibrium responses or responses to gradually changing climate (Fig. 1 and Table 2). Our model, and all others which couple vegetation and soil processes (e.g. Hunt et al., 1991 ; Friend et al., 1996 ; McGuire, Kicklighter and Melillo, 1996 ; Parton et al., 1996 ; Thornley and Cannell, 1996) show that ecosystem responses to most Thornley and Cannell—Grassland Responses to Climate Change aspects of climate change are highly dependent upon soil processes which control the supply of mineral nutrients. These processes take decades to reach near-equilibrium. Consequently, experiments which impose sudden changes in temperature or [CO ] on an ecosystem in the field or on # monoliths brought into the laboratory, and which last only a few years, will rarely produce quantitative changes in NPP, Csys or other ecosystem properties which indicate likely long-term trends in response to gradual climate change (see also Rastetter and Shaver, 1992). In our view, this fact makes it practically impossible to quantify realworld responses of NPP and Csys to climate change other than by modelling. Grazed swards react ery differently from ungrazed swards to changes in temperature and [CO ], because the productiity # of grazed swards responds to changes in LAI as well as direct effects on other processes (Figs 3 and 4). Some experiments on grasslands simulate mowing (Newton et al., 1995, 1996 ; Jones et al., 1996) but few experiments or models simulate grazing (Ojima et al., 1996). Almost all grasslands in the world are subject to grazing, which has profound effects on ecosystem properties—lowering LAI, NPP, Csys and system C}N ratio (Fig. 4). Most importantly, light interception is often incomplete, so that grazed swards are responsive to anything that changes LAI—NPP being roughly proportional to intercepted photosynthetically active radiation at low LAIs (Monteith, 1977). If it is recognized that leaf longevity is decreased by warming, and assumed that grazing is independent of temperature, then increasing temperatures above about 5 °C can lower the LAI of temperate, grazed swards to such an extent that NPP is actually decreased—despite accelerated photosynthesis, nutrient release by mineralization and growth (Fig. 3 A). The response of ungrazed swards to temperature is quite different because of their large LAIs (see Fig. 3 A). For the same reason, ungrazed swards respond less to nitrogen addition (Fig. 5 ; Van den Pol van Dasselaar and Lantinga, 1995). Effects of climate change on grasslands are likely to be highly specific to particular climatic conditions, intensity of grazing and nutrient supply. The SCOPE study, reported by Parton et al. (1996), predicted very different responses of grassland to increasing [CO ] and changes in temperature # and rainfall at the biome level. Elevated [CO ] generally # increased NPP and soil organic matter levels (approx. Csys)—but to different extents in different biomes—whereas changes in temperature and rainfall had dramatically different effects depending on the current climate and feedbacks involving, primarily, nutrient mineralization and water use. This analysis goes a step further in showing that responses of grasslands within a biome are likely to be very variable, depending on the current climate and the intensity of grazing. Again, the main sources of variation were responses to temperature and rainfall, rather than to [CO ]. Rising # temperatures could increase or decrease NPP, depending on the current climate (cf the two sites in Fig. 6) and the level of grazing (Fig. 3). The optimum temperature for NPP was 12–13 °C without grazing, but only 5–7 °C with the levels of grazing chosen in this study (Fig. 3). Increasing rainfall 213 increased NPP where it relieved summer droughts, but decreased NPP where it increased N leaching (Fig. 5). Similarly, amounts of N applied to grasslands can alter climate change responses (Thornley, 1998). Clearly, differing responses of NPP to climate change mean different amounts of litter input to the soil and different effects on Csys and the operation of grasslands as carbon sinks. It is not surprising that experiments give a variety of responses, some of which may be correctly attributed to species, but others are likely to be explained by differences in environmental and management conditions (e.g. Owensby et al., 1993 ; Lutze and Gifford, 1995 ; Newton et al., 1995 ; Casella et al., 1996 ; Jones et al., 1996 ; Soussana, Casella and Loiseau, 1996). Our overall conclusion from the three points made above is that research should be focused on the development of models which represent the essential feedback processes that are altered by climate. Experiments should be directed towards lessening uncertainty about the operation and parameterization of those processes, and rather than towards answering questions at the ecosystem level directly (see also Rastetter, 1996). Temperate grasslands as carbon sinks in response to increasing temperature and [CO ] # One of the most important and difficult problems in climate change research is to quantify the net effect on carbon sequestration of the opposing effects of increasing [CO ] and temperature. It is a difficult problem, because the # net flux is a small difference between two very large numbers. Also, it is not always realized that the hypothesized terrestrial carbon sink of 0±5–2±0 Gt C yr−" due to CO and # nitrogen fertilization (Mellilo et al., 1996) represents a sink of only 5–20 g C m−# y−" if spread evenly over the vegetated surface of the earth (8–9¬10* ha). This is a small number, well within the error of direct estimates of CO fluxes and # detectable changes in carbon pools. Our analysis indicates the rising [CO ] alone produces a sink, rising temperatures # alone produce a source, and the net effect is likely to be a sink for several decades. These three points are elaborated below. As long as rising [CO ] continues to increase gross # photosynthesis and grassland NPP, then total system carbon (Csys, mainly in soil organic matter) will increase, representing a carbon sink, because of the time lag between increasing NPP and increasing soil respiration. This point was elaborated by Taylor and Lloyd (1992) and Lloyd and Farquhar (1996). The argument is that there are time lags between the instantaneous rise in rates of photosynthesis with increase in [CO ], consequent increases in NPP, litterfall # and soil organic matter pools. In general, the total litter and soil respiration CO fluxes are proportional to the sizes of # their decomposing carbon pools, so that the increase in loss of carbon from litter and soils is not in equilibrium with the increase in NPP. This lag between NPP and litter}soil respiration produces a transient carbon sink. The Hurley Pasture Model, and most other process-based ecosystem models, simulate this effect (Fig. 1 H and I) and make it inevitable that rising [CO ] results in rising Csys—that # 214 Thornley and Cannell—Grassland Responses to Climate Change is, in a carbon sink (Fig. 7, CO effect)—provided # photosynthesis does not totally downregulate, or become CO saturated, and that NPP responses are not severely # limited by nutrient and water supplies (i.e. are not lessened by greater nutrient and water capture and}or use efficiencies ; Idso and Idso, 1994). Any effect of elevated [CO ] on # decreasing maintenance respiration rates (not represented in our model ; Gonzales-Meler, Drake and Azcon-Bieto, 1996) or soil respiration rates, by changing litter quality (which is represented) will exaggerate the effect. By contrast, increasing temperature generally decreases Csys, although NPP may be increased or decreased (Figs 1–3). As mentioned, ecosystem NPP can increase or decrease in response to increasing temperature depending on the starting temperature (the current climate), level of grazing, nutrient input and the extent of positive and negative feedbacks on nutrient supplies and water stress (McGuire et al., 1996 ; Thornley and Cannell, 1996). However, even when NPP increases, the net result is generally a decrease in Csys—that is, a carbon source (Figs 1 A, B, 7). This means that soil respiration rates generally increase more rapidly than rates of NPP with increase in temperature—a fact supported by their different Q values "! (Kirschbaum, 1995) and by the observation that Csys tends to decrease from the tundra to the equator (Post et al., 1982). In this analysis, both NPP and Csys were greater at the cool, upland site than at the warmer, lowland site (Fig. 5). The important point is that, in the Hurley Pasture Model, the expected result of decreasing Csys with increase in temperature was obtained although the same temperature response function was used for all soil and plant biochemical processes (other than quantum efficiency of photosynthesis, and with additional modifiers for photosynthesis ; see Appendix). This temperature function was similar to that used for soil decomposition in the CCGRASS and CENTURY models (Van den Pol van Dasselaar and Lantinga, 1995 ; Parton et al., 1996) and, when applied to maintenance respiration and other plant processes, it predicted a relatively small change in the fraction of gross photosynthate lost by respiration over a large range of temperatures (see Thornley, 1998) as found experimentally by Gifford (1994). We recognize that temperature response functions will differ for plants and microbes adapted to different environments, but we have less reason to believe that the response functions of biochemical processes in plants and soils differ greatly within any one environment. The net effect of current and projected increases in [CO ] # and temperature (oer the next few decades) is likely to be an increase in NPP and Csys in humid, temperate grassland, with a carbon sink of 5–15 g C m−# yr−". This sink will decrease to zero and become negatie after seeral decades (Figs 6 and 7). Most researchers have addressed the problem of determining the magnitude of the carbon sink}source produced by grasslands in response to climate change by running models in a 2¬CO climate either to equilibrium (Thornley, Fowler # and Cannell, 1991) or for 50–100 years—using much simpler vegetation models than the Hurley Pasture Model (e.g. grassland models : Hunt et al., 1991 ; Ojima et al., 1993 ; Parton, Ojima and Schimel, 1994 ; Hall et al., 1995 ; Van den Pol van Dasselaar and Lantinga, 1995 ; McGuire et al., 1996 ; Parton et al., 1996). Some models predict a net source for most grasslands (Ojima et al., 1993 ; Hall et al., 1995 ; Van den Pol van Dasselaar and Lantinga, 1995) others a small sink, depending mostly on the magnitude of increased nutrient availability in response to warming and the respiration response (McGuire et al., 1996). The Hurley Pasture Model predicts that the combined rates of increase in [CO ] and temperature that are occurring # now, and are likely to continue for a few decades, will produce a sink of about 5–15 g C m−# yr−" in temperate grasslands for several decades (Fig. 7). This is the right order of magnitude, if it occurs globally (see above). However, it is inevitable that the positive effects of increasing [CO ] will decrease as the photosynthetic system saturates, # whereas the negative effects of temperature will continue to increase in non-tropical regions. Clearly, the timescale over which the global terrestrial sink weakens is critical for current negotiations on the control of greenhouse gas emissions and requires further modelling research on realworld transient responses. A C K N O W L E D G E M E N TS The work has been supported by NERC through its TIGER (Terrestrial Initiative in Global Environmental Research) programme, the Department of the Environment Contract No EPG1}1}3 on carbon sequestration, the European Community EPOCH (EPOC-CT90–0022) and Espace (EV5V-CT93–0292) grassland programmes, and the BBSRC Institute of Grassland and Environmental Research at Northwyke, Devon, UK. LITERATURE CITED Arah JRM, Thornley JHM, Poulton PR, Richter DD. 1997. Soil organic matter in the ITE Edinburgh Forest and Hurley Pasture ecosystem models : comparison with long-term datasets. Geoderma (in press). Baxter R, Ashenden TW, Sparks TH, Farrar JF. 1994. Effects of elevated carbon dioxide on three montane grass species. I. Growth and dry matter partitioning. Journal of Experimental Botany 45 : 305–315. Breymeyer AI, Hall DO, Melillo JM, Agren GI. 1996. Global effects on coniferous forests and grasslands. Chichester : John Wiley and Sons. Brookes PC, Landman A, Pruden G, Jenkinson DS. 1985. Chloroform fumigation and the release of soil nitrogen : a rapid direct extraction method to measure microbial biomass nitrogen in soil. Soil Biology and Biochemistry 17 : 837–842. Campbell GS. 1974. A simple method for determining unsaturated conductivity from moisture retention data. Soil Science 117 : 311–315. Cannell MGR, Dewar RC. 1994. Carbon allocation in trees : a review of concepts for modelling. Adances in Ecological Research 25 : 59–104. Casella E, Soussana JF, Loiseau P. 1996. Long-term effects of CO # enrichment and temperature increase on a temperate grass sward. Plant and Soil 182 : 83–99. Chen D-X, Coughenour MB. 1994. GEMT – a general model for energy and mass transfer of land surfaces and its application at the Fife sites. Agricultural and Forest Meteorology 68 : 145–171. Clarkson DT, Jones LHP, Purves JV. 1992. Absorption of nitrate and ammonium ions by Lolium perenne from flowing solution cultures at low root temperatures. Plant, Cell and Enironment 15 : 99–106. Eason WR, Newman EI. 1990. Rapid cycling of nitrogen and Thornley and Cannell—Grassland Responses to Climate Change phosphorus from dying roots of Lolium perenne. Oecologia 82 : 432–436. Evans JR, Farquhar GD. 1991. Modeling canopy photosynthesis from biochemistry of the C3 chloroplast. In : Boote KJ, Loomis RS, eds. Modeling crop photosynthesis – from biochemistry to canopy. CSSA Special Publication Number 19. Madison, Wisconsin : American Society of Agronomy, 1–16. Evans JR, Terashima I. 1988. Photosynthetic characteristics of spinach leaves grown with different nitrogen treatments. Plant and Cell Physiology 29 : 157–165. France J, Thornley JHM. 1984. Mathematical modelling in agriculture. London : Butterworths. Friend AD, Stevens AK, Knox RG, Cannell MGR. 1996. A processbased, biogeochemical, terrestrial biosphere model of ecosystem dynamics (Hybrd v3.0). Ecological Modelling (in press). Gifford RM. 1994. Implications of CO effects on vegetation for the # global carbon budget. In : Heimann M, ed. The global carbon cycle. NATO ASI Series Vol I, 15. Berlin : Springer-Verlag, 159–199. Gonzalez-Meler M, Drake BG, Azcon-Bieto J. 1996. Rising atmospheric carbon dioxide and plant respiration. In : Breymeyer AI, Hall DO, Melillo JM, Agren GI, eds. Global change : effects on coniferous forests and grasslands. (SCOPE 56). Chichester : John Wiley and Sons. Gregson K, Hector DJ, McGowan M. 1987. A one-parameter model for the soil water characteristic. Journal of Soil Science 38 : 483–486. Hall DO, Ojima DS, Parton WJ, Scurlock JMO. 1995. Response of temperate and tropical grasslands to CO and climate change. # Journal of Biogeography 22 : 537–547. Hirose T, Werger M JA. 1987. Nitrogen use efficiency in instantaneous and daily photosynthesis of leaves in the canopy of a Solidago altissima stand. Physiologia Plantarum 70 : 215–222. Houghton JT, Meira Filho LG, Callander BA, Harris N, Kattenberg A, Maskell K. 1996. Climate change 1995. The science of climate change. Contribution of working group I to the second assessment report of the IPCC. Cambridge : Cambridge University Press. Hunt WF. 1983. Nitrogen cycling through senescent leaves and litter in swards of Ruanui and Nui ryegrass with high and low nitrogen inputs. New Zealand Journal of Agricultural Science, Cambridge 102 : 703–709. Hunt HW, Trlica MJ, Redente EF, Moore JC, Detling JK, Kittel TGF, Walter DE, Fowler MC, Klein DA, Elliot ET. 1991. Simulation model for the effects of climate change on temperate grassland ecosystems. Ecological Modelling 53 : 205–246. Idso KB, Idso SB. 1994. Plant responses to elevated CO enrichment in # the face of environmental constraints : a review of the past 10 years research. Agricultural and Forest Meteorology 69 : 153–203. Innis GS. 1978. Grassland simulation model. Ecological Studies 26. New York : Springer-Verlag. Jenkinson DS. 1990. The turnover of organic carbon and nitrogen in soil. Philosophical Transactions of the Royal Society, London B 329 : 361–368. Johnson IR, Thornley JHM. 1985. Dynamic model of the response of a vegetative grass crop to light, temperature and nitrogen. Plant, Cell and Enironment 8 : 485–499. Jones HG. 1992. Plants and microclimate. Cambridge : Cambridge University Press. Jones MB, Jongen M, Doyle T. 1996. Effects of elevated carbon dioxide concentrations on agricultural grassland production. Agricultural and Forest Meteorology 79 : 243–252. Kirschbaum MUF. 1995. The temperature dependence of soil organic matter decomposition and the effect of global warming on soil organic carbon storage. Soil Biology and Biochemistry 27 : 753–760. Korner C, Miglietta F. 1994. Long term effects of naturally elevated CO on Mediterranean grassland and forest trees. Oecologia 99 : # 343–351. Lloyd J, Farquhar GD. 1996. The CO dependence of photosynthesis, # plant growth and response to elevated CO concentrations and # their interaction with soil nutrient status. I. General principles and forest ecosystems. Functional Ecology 10 : 4–32. Long SP. 1991. Modification of the response of photosynthetic productivity to rising temperature by atmospheric CO con- # 215 centration : has its importance been underestimated ? Plant, Cell and Enironment 14 : 729–739. Lundgren B, Soderstrom B. 1983. Bacterial numbers in a pine forest soil in relation to environmental factors. Soil Biology and Biochemistry 15 : 625–630. Luo Y, Field CB, Mooney HA. 1994. Predicting responses of photosynthesis and root fraction to elevated CO : interactions # among carbon, N and growth. Plant, Cell and Enironment 17 : 1195–1204. Lutze JL, Gifford RM. 1995. Carbon storage and productivity of a carbon dioxide enriched nitrogen limited grass sward after one year’s growth. Journal of Biogeography 22 : 227–233. McGuire AD, Kicklighter DW, Melillo JM. 1996. Global climate change and carbon cycling in grasslands and conifer forests. In : Breymeyer AI, Hall DO, Melillo JM, Agren GI, eds. Global change : effects on coniferous forests and grasslands. (SCOPE 56). Chichester : John Wiley and Sons, 389–411. Maron SH, Prutton CF. 1965. Principles of physical chemistry. New York : MacMillan. Melillo JM, Prentice IC, Farquhar GD, Schultze E-D, Sala OE. 1996. Terrestrial biotic responses to environmental change and feedbacks to climate. In : Houghton JH, Meira Filho LG, Callander BA, Harris N, Kattenberg A, Maskell K. Climate change 1995. The science of climate change. Contribution of working party I to the second assessment report of the IPCC. Cambridge : Cambridge University Press, 444–481. Monteith JL. 1977. Climate and the efficiency of crop production in Britain. Philosophical Transactions of the Royal Society (London), Series B 281 : 277–294. Morison JIL. 1985. Intercellular CO concentration and stomatal # responses to CO . In : Zeiger E, Cowan I, Farquhar GD, eds. # Stomatal function. Stanford, USA : University Press, 229–251. Nambiar ESK, Fife DN. 1991. Nutrient retranslocation in temperate conifers. Tree Physiology 9 : 185–207. Newton PCD, Clark H, Bell CC, Glasgow EM. 1996. Interaction of soil moisture and elevated CO on the above-ground growth rate, root # length density and gas exchange of turves from temperate pasture. Journal of Experimental Botany 47 : 771–779. Newton PCD, Clark H, Bell CC, Glasgow EM, Tate KR, Ross DJ, Yeates GW. 1995. Plant growth and soil processes in temperate grassland communities at elevated CO . Journal of Biogeography # 22 : 235–240. Ojima DS, Parton WJ, Coughenour MB, Scurlock JMO, Kirchner TB, Kittel TGF, Hall DO, Schimel DS, Garcia Moya E, Gilmanov TG, Seastedt TR, Kamnalrut A, Kinyamario JI, Long SP, Menaut J-C, Sala OE, Scoles RJ, van Veen JA. 1996. Impact of climate and atmospheric carbon dioxide on grasslands of the world. In : Breymeyer AI, Hall DO, Melillo JM, Agren GI, eds. Global change : effects on coniferous forests and grasslands. (SCOPE 56). Chichester : John Wiley and Sons, 271–309. Ojima DS, Parton WJ, Schimel DS, Scurlock JMO, Kittel TGF. 1993. Modelling the effects of climatic and CO changes on grassland # storage of soil C. Water, Air and Soil Pollution 70 : 643–657. Owensby CE, Coyne PI, Ham JM, Auen LM, Knapp AK. 1993. Biomass production in a tallgrass prairie ecosystem exposed to ambient and elevated CO . Ecological Applications 3 : 644–653. # Pachevsky LB, Haskett JD, Acock B. 1996. An adequate model of photosynthesis I. Parameterization, validation and comparison of models. Agricultural Systems 50 : 209–225. Parton WJ, Coughenour MB, Scurlock JMO, Ojima DS, Gilmanov TG, Scholes RJ, Schimel DS, Kirchner TB, Menaut J-C, Seastedt TR, Garcia Moya E, Kamnalrut A, Kinyamario JI, Hall DO. 1996. Global grassland ecosystem modelling : development and test of ecosystem models for grassland systems. In : Breymeyer AI, Hall DO, Melillo JM, Agren GI, eds. Global change : effects on coniferous forests and grasslands. (SCOPE 56). Chichester : John Wiley and Sons, 229–267. Parton WJ, Ojima DS, Schimel DS. 1994. Environmental change in grasslands : assessment using models. Climatic Change 28 : 111–141. Parton WJ, Schimel DS, Cole, CV, Ojima DS. 1987. Analysis of factors controlling soil organic matter levels in Great Plains Grasslands. Soil Society of America Journal 51 : 1173–1179. 216 Thornley and Cannell—Grassland Responses to Climate Change Parton WJ, Scurlock JMO, Ojima DS, Gilmanov TG, Scoles RJ, Schimel DS, Kirchner T, Menaut J-C, Seastedt T, Garcia Moya E, Apinan Kamnalrut, Kinyamario, JI. 1993. Observations and modeling of biomass and soil organic matter dynamica for the grassland biome worldwide. Global Biogeochemical Cycles 7 : 785–809. Penning PD, Parsons AJ, Orr RJ, Treacher TT. 1991. Intake and behaviour responses by sheep to changes in sward characteristics under continuous stocking. Grass and Forage Science 46 : 15–28. Penning de Vries FWT, Brunsting AHM, van Laar HH. 1974. Products, requirements and efficiency of biosynthesis : a quantitative approach. Journal of Theoretical Biology 45 : 339–377. Post WM, Emanuel WR, Zinke PJ, Stangenberger AG. 1982. Soil carbon pools and world life zones. Nature 298 : 156–159. Rastetter EB. 1996. Validating models of ecosystem response to global change. BioScience 46 : 190–198. Rastetter EB, Shaver GR. 1992. A model of multiple-element limitation for acclimating vegetation. Ecology 73 : 1157–1174. Robson MJ, Ryle GJA, Woledge J. 1988. The grass plant–its form and function. In : Jones MB, Lazenby A, eds. The grass crop. London : Chapman and Hall, 25–83. Rotmans J, den Elzen MGJ. 1993. Modelling feedback mechanisms in the carbon cycle : balancing the carbon budget. Tellus 45B : 301–320. Rowell DL. 1994. Soil science : methods and applications. Harlow, Essex : Longman. Ryan MG. 1995. Foliar maintenance respiration of subalpine and boreal trees and shrubs in relation to nitrogen content. Plant, Cell and Enironment 18 : 765–772. Schappi B, Korner C. 1996. Growth responses of an alpine grassland to elevated CO . Oecologia 105 : 43–52. # Schenk U, Manderscheid R, Hugen J, Weigel H-J. 1995. Effects of CO # enrichment and intraspecific competition on biomass partitioning, nitrogen content and microbial biomass carbon in soil of perennial ryegrass and white clover. Journal of Experimental Botany 46 : 987–993. Soussana JF, Casella E, Loiseau P. 1996. Long-term effects of CO # enrichment and temperature increase on a temperate grass sward. Plant and Soil 182 : 101–114. Sutton MA, Pitcairn CER, Fowler D. 1993. The exchange of ammonia between the atmosphere and plant communities. Adances in Ecological Research 24 : 302–394. Taylor JA, Lloyd J. 1992. Sources and sinks of atmospheric CO . # Australian Journal of Botany 40 : 407–418. Thornley JHM. 1972. A model to describe the partitioning of photosynthate during vegetative plant growth. Annals of Botany 36 : 419–430. Thornley JHM. 1996. Modelling water in crops and plant ecosystems. Annals of Botany 77 : 261–275. Thornley JHM. 1998. Grassland dynamics. An ecosystem simulation model. Wallingford, Oxon : CAB International. (in press.) Thornley JHM, Cannell MGR. 1992. Nitrogen relations in a forest plantation-soil organic matter ecosystem model. Annals of Botany 70 : 137–151. Thornley JHM, Cannell MGR. 1996. Temperate forest responses to carbon dioxide, temperature and nitrogen : a model analysis. Plant, Cell and Enironment 19 : 1331–1348. Thornley JHM, Fowler D, Cannell MGR. 1991. Terrestrial carbon storage resulting from CO and nitrogen fertilization in temperate # grasslands. Plant, Cell and Enironment 14 : 1007–1011. Thornley JHM, Johnson IR. 1990. Plant and crop modelling. Oxford : Oxford University Press. Thornley JHM, Verberne ELJ. 1989. A model of nitrogen flows in grassland. Plant, Cell and Enironment 12 : 863–886. Van den Pol van Dasselaar A, Lantinga EA. 1995. Modelling the carbon cycle of grassland in the Netherlands under various management strategies and environmental conditions. Netherlands Journal of Agricultural Science 43 : 183–194. Verberne ELJ. 1992. Simulation of the nitrogen and water balance in a system of grassland and soil. Nota 258. DLO-Institut voor Bodemvruchtbaarheid, Oosterweg 92, 9750 RA, Haren, Netherlands. Verberne ELJ, Hassink J, de Willigen P, Groot JJR, van Veen JA. 1990. Modelling organic matter dynamics in different soils. Netherlands Journal of Agricultural Science 38 : 221–238. White JR. 1987. ERHYM-II : model description and user guide for the BASIC ersion. US Department of Agriculture, Agricultural Research Service, ARS-59, Washington, D.C. Whitehead DC. 1995. Grassland nitrogen. Wallingford, UK : CAB International. Williams EJ, Hutchinson GC, Fehsenfeld FC. 1992. NOx and N O # emissions from soils. Global Biogeochemical Cycles 6 : 351–388. Wolfenden J, Diggle PJ. 1995. Canopy gas exchange and growth of upland pasture swards in elevated CO . New Phytologist 130 : # 369–380. APPENDIX A synopsis of processes and properties represented in the Hurley Pasture Model A full description of the Hurley Pasture Model is given by Thornley (1998) and the ACSL source program, pasture.csl, is available by ftp from buua.nbu.ac.uk in pub}tree}Pasture, or e-mail to johnt!buua.nbu.ac.uk. Temperature and water modifiers of biochemical processes Temperature. The rates of all biochemical processes in plants and soil are modified by the same cubic temperature function, which increases sigmoidally from zero at 0 °C to maximal activity at 30 °C, with the inflexion at 15 °C, and falling to zero activity at 45 °C. Exceptions are the photosynthetic processes. Water status. The rates of all biochemical processes in the plant and soil are made dependent on the ‘ chemical activity ’ of water, which is derived from the tissue or soil water potential (Maron and Prutton, 1965 ; see Thornley, 1996, 1998). Shoot water potential is used as a surrogate to derive surface litter water activity. Plant submodel (21 state ariables, 60 parameters) Species and oeriew. The grassland is assumed to be composed entirely of a C grass such as Lolium perenne. The $ grass plant model (Fig. A1) is driven by C input from photosynthesis and N input from soil mineral N pools (ammonium and nitrate). The soil submodel influences plant growth only by altering the supply of mineral N supply and water. Plant state ariables. These are leaves (lamina), leaf area, sheath plus stem, and roots ; all have four age categories. Carbon (C) and nitrogen (N) substrate pools in the shoot (leaves, sheaths and stem) and root are defined separately from the structure. Light interception. Calculated using the Beer-Lambert law, with light extinction coefficient 0±5 and leaf transmission coefficient 0±1. Photosynthesis. Uses a non-rectangular hyperbola dependence of gross photosynthesis on light, with three parameters. The sharpness of the knee of the hyperbola is constant at 0±95. The initial slope (α, the inverse of quantum requirement) is set to a value equivalent to 19 quanta per molecule of CO (Jones, 1992), with modifiers for [CO ] # # (representing photorespiration suppression), temperature 217 Thornley and Cannell—Grassland Responses to Climate Change Plant submodel Removal of shoot structure and substrates Harvesting Animal grazing Animal submodel SHOOT Lamina area Lamina Sheath + stem Four age categories: i = 1 → 2→ 3 → 4 Senescence and litter submodel Recycling of C and N Photosynthesis Soil Grazing/degradation by other fauna Shoot C substrate Shoot N substrate Growth Maintenance CO2 respiration Transport Transport CO2 Growth respiration Growth Root C substrate CO2 Respiration with N uptake Exudation Root N substrate Recycling of C and N i = 1 → 2→ 3 → 4: Four age categories of root mass Root density ROOT N uptake Soil and Senescence Litter Exudation sub- Grazing/degradation by other fauna model F. A1. Hurley Pasture Model : the plant submodel. (decreasing α by 1±5 % per °C above 15 °C) and shoot water status ; α is not dependent on N concentration (Evans and Terashima, 1988). The asymptote (Pmax) has a notional maximum at 350 µmol mol−" CO of 1¬10−' kg CO m−# s−", # # equivalent to 23 µmol CO m−# s−", typical of C grasses, $ # with cubic modifiers for temperature and water status as above, and modifiers for leaf N level (with minimum and maximum set at 1 and 2 g N m−#, Hirose and Werger, 1987), CO and stomatal conductance. The optimum temperature # for Pmax is [CO ] dependent (Long, 1991) although this is # unimportant in a temperate climate. The model gives responses similar to the biochemical model of Evans and Farquhar (1991, see Pachevsky, Haskett and Acock, 1996). Values of α and Pmax are not ‘ downregulated ’ with increase in [CO ], but leaf photosynthesis decreases as a result of # increased leaf thickness, C}N ratio and stomatal resistance. Canopy photosynthesis. Calculated using a single analytical expression where the contributions of all leaves are added by using an algebraic expression for the integral through the canopy. This method gave similar results to two alternative methods evaluated by Thornley (1998) who showed that canopy photosynthesis was not appreciably affected by decreasing foliage N concentration from the youngest to oldest age category. Therefore uniform canopy N concentrations were assumed (cf. Friend et al., 1996). Stomatal conductance. Stomata are open only in the day. Conductance varies linearly, between fully open and closed, with shoot relative water content (i.e. water content at prevailing water potential divided by water content when the potential is zero). Conductance is modified by [CO ], so # that doubling [CO ] from 350 to 700 µmol mol−" CO # # caused a 40 % reduction (Morison, 1985). There is no root message control of stomatal conductance. Allocation of assimilates. The transport}utilization method of Thornley (1972) is used. Allocation depends on the growth rates of the plant parts (see below) and the conductances to C and N substrate transport along opposing root-shoot concentration gradients within the plant. Transport conductances increase with temperature and water status and are scaled with plant mass. This mechanism simulates widely observed behaviour of shoot}root allocation related to plant C}N status, including responses of L. perenne to [CO ] (Cannell and Dewar, 1994 ; Shenck et # al., 1995). Shoot and root growth. Shoot and root growth are proportional to their structural mass times the product of local C and N substrate concentrations, modulated by temperature and water status. With neither limiting, the relative shoot growth rate is 0±2 d−" (Robson, Ryle and Woledge, 1988). Leaf expansion is proportional to shoot 218 Thornley and Cannell—Grassland Responses to Climate Change relative water content, as well as being affected by the dependence of shoot growth and shoot-root partitioning on water status. Specific leaf area. The maximum specific leaf area is 25 m# kg−". Actual incremental specific leaf area decreases linearly with increase in C substrate concentration (in high light and [CO ], at low temperatures) and with increasing # water stress. Respiration. Constant fractions of C substrate are used for growth respiration, using established conversion factors (Penning de Vries, Brunsting and van Laar, 1974). Maintenance respiration rates are proportional to structural mass, weighted by age category, and modified by temperature and water status. N dependence operates through the increase in C}N ratio with age category rather than as a direct N dependence (cf. Ryan, 1995). There is also no direct [CO ] dependence of maintenance respiration rate # coefficients (cf. Gonzalez-Meler et al., 1996). Root N uptake. N uptake rate from soil ammonium and nitrate pools has Michaelis-Menten dependence on the mineral pool sizes [weighted towards ammonium at low temperatures, Clarkson, Jones and Purves (1992)] and is modified by temperature, root water status, and soil relative water content (m$ water}m$ soil). Uptake is proportional to root mass (weighted by age), is increased by increasing root C substrate concentration and is decreased by increasing root N substrate concentration. Root exudation. Roots exude a maximum 2 % of their C substrate and 0±5 % of their N substrate per day, modulated by root structural mass, C and N substrate concentration, temperature and water status. Root density (kg m−$ soil). This modifies water uptake, applies to new root and is dependent on soil water content, being maximal at field capacity. Shoot and root aging. Transfer to older age categories is accelerated with increase in temperature and water stress. At 20 °C, with water non-limiting, leaves have a half-life of 24 d and roots 26 d (Eason and Newman, 1990). Material from the oldest categories is transferred to the three litter pools (see below). Retranslocation from senescing shoot and root. The lower the N substrate concentration, the more N is withdrawn and recycled before the tissue dies (Hunt, 1983). This is assumed to occur in both shoot and root (cf. Nambiar and Fife, 1991). The fraction of litter that is recyclable (proteinaceous material) has a C}N ratio of 2±7 kg kg−". Soil and litter submodel (15 state ariables, 68 parameters) Oeriew. There is only one soil horizon. The original soil submodel of Thornley and Verberne (1989) has been expanded to include more litter and soil organic matter (SOM) pools, giving more realistic SOM values (Arah et al., 1997) and making the model more akin to the CENTURY model (Parton et al., 1987). However, unlike soil models by Jenkinson (1990), Verberne et al. (1990) and Parton et al. (1993) the soil biomass is actively involved in dynamic C and N transfer processes, made possible with the representation of a small soluble C pool. The compartment flows are shown in Fig. A2 ; individual pool inputs and outputs are described in equations given by Thornley (1998). Soil type and clay content. The soil is a loam with 10 % clay. The clay content regulates the relative fractions of dying biomass carbon and decomposing lignin entering the unprotected and protected SOM pools. Also, the C}N ratio of the stabilized SOM pool depends on clay content. Litter pools. Shoot litter and faeces enter surface litter pools ; root litter enters soil litter pools. Both surface and soil litter pools are divided into metabolizable (met), cellulose (cel) and lignin (lig) subpools. The percentages of litter entering each subpool are 20, 65 and 15 % to each pool, respectively. The C}N ratio of the met. subpool is variable, depending on the litter ; the C}N ratios of the cel. and lig. subpools are 150 and 100, respectively. The rates of decomposition at 20 °C are : met. 0±2 d−", cel. 0±1 d−" and lig. 0±02 d−", modified by temperature, water status and (for the met. subpool) C}N ratio, as in the CENTURY model. The temperature of the surface litter pools is the mean between soil and air temperature. Outputs are shown in Fig. A2. Soil organic matter (SOM) pools. There are unprotected, protected (by clay particles) and stabilized SOM pools. All have varying C}N ratios because the C}N ratios of their inputs vary, depending on the size of the soil mineral N pools (ammonium and nitrate). Inputs and outputs are shown in Fig. A2. Decomposition rates depend on soil temperature and water status. At 20 °C, and with ample water, the half lives of the pools are : unprotected 1 year, protected 12 years and stabilized 100 years. Stabilization transfers occur from unprotected to protected to stabilized SOM at rates modified by soil temperature and water status. Mineral N pools. The ammonium mineral N pool receives N from atmospheric deposition, any N fertilizers applied, non-symbiotic N fixation, animal urine, root exudates and dying soil biomass, as well as from mineralizing litter and SOM (Fig. A2). Ammonium N can be taken up by plants, nitrified to nitrate, volatilized, immobilized or used in soil biomass growth. The nitrate N pool receives N from the ammonium pool, atmospheric deposition and fertilizers and loses N by plant uptake, immobilization, leaching and denitrification. Soil biomass. This consists of microbes, invertebrates and all other soil organisms. The biomass has a constant C}N ratio of 8. It grows, using substrates from the soluble C pool and two mineral N pools, at a semi-exponential rate, modified by soil temperature and water status, to a maximum biomass proportional to the size of the three SOM pools (Brookes et al., 1985). Predicted growth rates are similar to those observed (Lundgren and Soderstrom, 1983). Death rates are also dependent on soil temperature and water status. Soluble C pool. This small pool, which derives C from root exudates, dying soil biomass and mineralizing litter and SOM pools, provides the C substrate for soil biomass growth, and can be depleted by leaching and respiration. Mineralization}immobilization. The definitions of mineralization and immobilization are always somewhat arbitrary. Here, mineralization is defined as N release from the three decomposing SOM pools. Immobilization is the N 219 Thornley and Cannell—Grassland Responses to Climate Change Soil and litter submodel Shoot litter Surface litter Animal faeces metabolic cellulose C, N variable C/N = 150 lignin C/N = 100 CO2 Unprotected SOM C, N variable CO2 Root litter Surface litter metabolic cellulose C, N variable C/N = 150 C/N = 100 CO2 Protected SOM C, N variable CO2 Soluble C Leaching Volatilization CO2 CO2 Animal urine Death Soil biomass Ammonium N Growth Fertilizer, root exudate Atmospheric deposition Nitrogen fixation N lignin Root exudate CO2 CO2 C/N = 8 N Stabilized SOM C, N variable Nitrification Immobilization Fertilizer Nitrate N Atmospheric deposition Plant uptake CO2 Leaching Denitrification Mineralization CO2 F. A2. Hurley Pasture Model : the soil and litter submodel. required from the mineral N pools so that newly created SOM possesses a C}N ratio that depends on the concentration of mineral N. Thus, the immobilization flux depends on the C}N ratio of the material which is currently entering the SOM pools as well as the mineral N concentration. Nitrification and denitrification. Nitrification, converting ammonium to nitrate, occurs at a rate dependent on the ammonium concentration and soil biomass, modified by soil temperature and water status. At 20 °C, with 0±2 kg m−# soil biomass, the nitrification flux is about 300 kg N ha−" yr−", 2 % of which is lost as N gases. Denitrification occurs only when the soil is above field capacity (in anaerobic conditions) at a rate dependent on the nitrate concentration and soil biomass, modified by soil temperature. At 20 °C, with 0±2 kg m−# soil biomass, the flux is about 7 kg N ha−" yr−". Nitrification and denitrification fluxes are within measured ranges for N-rich grasslands (Williams, Hutchinson and Fehsenfeld, 1992 ; Whitehead, 1995). Volatilization. Volatilization to ammonia gas depends on the ammonium concentration and increases with tem- perature. At 20 °C, with 0±001 kg N m−# in the ammonium pool, the flux is 73 kg N ha−" yr−" (Sutton, Pitcairn and Fowler, 1993). N fixation by non-symbiotic soil microbes. Fixation occurs # at a rate proportional to soil biomass, modified by soil temperature and water status, inhibited by high level of mineral N, with Michaelis-Menten dependence on the soluble C pool. With 0±2 kg m−# soil biomass, the maximum fixation rate at 20 °C is 37 kg N ha−" yr−". Leaching. N and C are leached from the nitrate N and soluble C pools, respectively, depending on the concentrations of N and C in those pools and the amount of drainage. A N concentration of 0±001 kg N m−# is equivalent to about 10 mg l−" in leachate (Rowell, 1994). Animal submodel (2 state ariables, 9 parameters) Species. The pasture is assumed to be grazed by mature, non-pregnant, non-lactating sheep. Grazing. Shoot removal by grazing is a sigmoidal asymptotic function of LAI, with a maximum rate of 220 Thornley and Cannell—Grassland Responses to Climate Change Animal submodel Outputs to environment: Maintenance respiration Methane production Growth respiration C retained for animal growth C Inputs of C, N from pasture submodel by grazing shoot material N N retained for animal growth C N N Urine C Faeces Outputs to the soil and litter submodel F. A3. Hurley Pasture Model : the animal submodel. Water submodel Transpiration SHOOT Water content Relative water content Water potential Rain Rainfall interception by and evaporation from canopy ROOT Water content Relative water content Water potential Soil evaporation (assumed zero) SOIL Water content Relative water content Water potential Surface run-off (assumed zero) Drainage F. A4. Hurley Pasture Model : the water submodel. 2±5 kg DM per animal d−" and a half-maximum rate attained at LAI ¯ 1 (Penning et al., 1991). Animal growth. It is assumed that 1 % of N ingested is retained and that the C}N ratio of animal protein is 3±5. Gaseous losses. These occur by maintenance respiration (60 % of C), methane production (5 % of C) and growth respiration. Urine and faeces. This output is obtained by difference. The fraction of N excreted as urine increases with increase in C}N ratio of ingested material. Water submodel (3 state ariables, 40 parameters) Oeriew. This submodel is fully described by Thornley (1996). It applies to a closed canopy and so has no soil evaporation. It calculates the amount and status of water in Thornley and Cannell—Grassland Responses to Climate Change the soil (single compartment), roots and shoots. Water fluxes are driven by water potential gradients. Interception. The fraction of rainfall intercepted by the canopy depends on the LAI and canopy extinction coefficient. The fraction that evaporates each month is given a numerical value each month depending on the intensity and duration of rainfall. Soil water. Soil water potentials and hydraulic conductivities are calculated using the equations of Gregson, Hector and McGowan (1987) and Campbell (1974), respectively. Soil-root-shoot flux. The flux of water from the soil to the roots depends on root mass, root density and the resistance 221 in the bulk soil, which depends on soil hydraulic conductivity and soil relative water content. The flux from root to shoot depends on the water potential difference divided by a rootshoot resistance to water flow, which is scaled to root and shoot mass. Eapotranspiration. The Penman-Monteith equation is used. Plant water status. The separation of substrates from structure permits detailed treatment of tissue water relations. The components of water potential—osmotic potential and pressure potential—are calculated, as well as relative water content. Osmoregulation occurs in response to increased C substrate concentration.