Diagnosing and assessing uncertainties of terrestrial ecosystem models in a multimodel ensemble experiment: 1. Primary production

We conducted an ensemble modeling exercise using the Terrestrial Observation and Prediction System (TOPS) to evaluate sources of uncertainty in carbon flux estimates resulting from structural differences among ecosystem models. The experiment ran public‐domain versions of biome‐bgc, lpj, casa , and tops‐bgc over North America at 8 km resolution and for the period of 1982–2006. We developed the Hierarchical Framework for Diagnosing Ecosystem Models (HFDEM) to separate the simulated biogeochemistry into a cascade of three functional tiers and sequentially examine their characteristics in climate (temperature–precipitation) and other spaces. Analysis of the simulated annual gross primary production (GPP) in the climate domain indicates a general agreement among the models, all showing optimal GPP in regions where the relationship between annual average temperature (T, °C) and annual total precipitation (P, mm) is defined by P=50T+500. However, differences in simulated GPP are identified in magnitudes and distribution patterns. For forests, the GPP gradient along P=50T+500 ranges from ～50 g C yr^?1 m^?2 °C^?1 (casa ) to ～125 g C yr^?1 m^?2 °C^?1 (biome‐bgc ) in cold/temperate regions; for nonforests, the diversity among GPP distributions is even larger. Positive linear relationships are found between annual GPP and annual mean leaf area index (LAI) in all models. For biome‐bgc and lpj , such relationships lead to a positive feedback from LAI growth to GPP enhancement. Different approaches to constrain this feedback lead to different sensitivity of the models to disturbances such as fire, which contribute significantly to the diversity in GPP stated above. The ratios between independently simulated NPP and GPP are close to 50% on average; however, their distribution patterns vary significantly between models, reflecting the difficulties in estimating autotrophic respiration across various climate regimes. Although these results are drawn from our experiments with the tested model versions, the developed methodology has potential for other model exercises.     

Evaluation of ecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJ dynamic global vegetation model

The Lund–Potsdam–Jena Dynamic Global Vegetation Model (LPJ) combines process‐based, large‐scale representations of terrestrial vegetation dynamics and land‐atmosphere carbon and water exchanges in a modular framework. Features include feedback through canopy conductance between photosynthesis and transpiration and interactive coupling between these ‘fast’ processes and other ecosystem processes including resource competition, tissue turnover, population dynamics, soil organic matter and litter dynamics and fire disturbance. Ten plants functional types (PFTs) are differentiated by physiological, morphological, phenological, bioclimatic and fire‐response attributes. Resource competition and differential responses to fire between PFTs influence their relative fractional cover from year to year. Photosynthesis, evapotranspiration and soil water dynamics are modelled on a daily time step, while vegetation structure and PFT population densities are updated annually. Simulations have been made over the industrial period both for specific sites where field measurements were available for model evaluation, and globally on a 0.5°° × 0.5°° grid. Modelled vegetation patterns are consistent with observations, including remotely sensed vegetation structure and phenology. Seasonal cycles of net ecosystem exchange and soil moisture compare well with local measurements. Global carbon exchange fields used as input to an atmospheric tracer transport model (TM2) provided a good fit to observed seasonal cycles of CO₂ concentration at all latitudes. Simulated inter‐annual variability of the global terrestrial carbon balance is in phase with and comparable in amplitude to observed variability in the growth rate of atmospheric CO₂. Global terrestrial carbon and water cycle parameters (pool sizes and fluxes) lie within their accepted ranges. The model is being used to study past, present and future terrestrial ecosystem dynamics, biochemical and biophysical interactions between ecosystems and the atmosphere, and as a component of coupled Earth system models.     

陆地生态系统碳源与碳汇及其影响机制研究进展

全球碳循环研究中发现,目前已知碳源与碳汇不能达到平衡。存在一个很大的碳失汇。大气、海洋和陆地生态系统是人工源CO2的3个可能的容纳汇,其中陆地生态系统最复杂、最具不确定性,因此陆地生态系统碳源与碳汇研究是全球碳循环研究的核心问题之一。大气成分监测、CO2通量测定、森林资源清查以及模型模拟等方面的研究都表明,CO2施肥效应、氮沉降增加、污染、全球气候变化以及土地利用变化,是影响陆地生态系统碳储量的主要生态机制,但不确定在过去的10～100年以及未来哪一种机制起最主要的作用。     

Global response of terrestrial ecosystem structure and function to CO2 and climate change: results from six dynamic global vegetation models

The possible responses of ecosystem processes to rising atmospheric CO₂ concentration and climate change are illustrated using six dynamic global vegetation models that explicitly represent the interactions of ecosystem carbon and water exchanges with vegetation dynamics. The models are driven by the IPCC IS92a scenario of rising CO₂ ( Wigley et al. 1991 ), and by climate changes resulting from effective CO₂ concentrations corresponding to IS92a, simulated by the coupled ocean atmosphere model HadCM2‐SUL. Simulations with changing CO₂ alone show a widely distributed terrestrial carbon sink of 1.4–3.8 Pg C y^?1 during the 1990s, rising to 3.7–8.6 Pg C y^?1 a century later. Simulations including climate change show a reduced sink both today (0.6–3.0 Pg C y^?1) and a century later (0.3–6.6 Pg C y^?1) as a result of the impacts of climate change on NEP of tropical and southern hemisphere ecosystems. In all models, the rate of increase of NEP begins to level off around 2030 as a consequence of the ‘diminishing return’ of physiological CO₂ effects at high CO₂ concentrations. Four out of the six models show a further, climate‐induced decline in NEP resulting from increased heterotrophic respiration and declining tropical NPP after 2050. Changes in vegetation structure influence the magnitude and spatial pattern of the carbon sink and, in combination with changing climate, also freshwater availability (runoff). It is shown that these changes, once set in motion, would continue to evolve for at least a century even if atmospheric CO₂ concentration and climate could be instantaneously stabilized. The results should be considered illustrative in the sense that the choice of CO₂ concentration scenario was arbitrary and only one climate model scenario was used. However, the results serve to indicate a range of possible biospheric responses to CO₂ and climate change. They reveal major uncertainties about the response of NEP to climate change resulting, primarily, from differences in the way that modelled global NPP responds to a changing climate. The simulations illustrate, however, that the magnitude of possible biospheric influences on the carbon balance requires that this factor is taken into account for future scenarios of atmospheric CO₂ and climate change.     

Predicting ecosystem dynamics at regional scales: an evaluation of a terrestrial biosphere model for the forests of northeastern North America

Terrestrial biosphere models are important tools for diagnosing both the current state of the terrestrial carbon cycle and forecasting terrestrial ecosystem responses to global change. While there are a number of ongoing assessments of the short-term predictive capabilities of terrestrial biosphere models using flux-tower measurements, to date there have been relatively few assessments of their ability to predict longer term, decadal-scale biomass dynamics. Here, we present the results of a regional-scale evaluation of the Ecosystem Demography version 2 (ED2)-structured terrestrial biosphere model, evaluating the model's predictions against forest inventory measurements for the northeast USA and Quebec from 1985 to 1995. Simulations were conducted using a default parametrization, which used parameter values from the literature, and a constrained model parametrization, which had been developed by constraining the model's predictions against 2 years of measurements from a single site, Harvard Forest (42.5° N, 72.1° W). The analysis shows that the constrained model parametrization offered marked improvements over the default model formulation, capturing large-scale variation in patterns of biomass dynamics despite marked differences in climate forcing, land-use history and species-composition across the region. These results imply that data-constrained parametrizations of structured biosphere models such as ED2 can be successfully used for regional-scale ecosystem prediction and forecasting. We also assess the model's ability to capture sub-grid scale heterogeneity in the dynamics of biomass growth and mortality of different sizes and types of trees, and then discuss the implications of these analyses for further reducing the remaining biases in the model's predictions.     

Carbon dioxide balance of a fen ecosystem in northern Finland under elevated UV-B radiation

The effect of elevated UV‐B radiation on CO₂ exchange of a natural flark fen was studied in open‐field conditions during 2003–2005. The experimental site was located in Sodankylä in northern Finland (67°22′N, 26°38′E, 179 m a.s.l.). Altogether 30 study plots, each 120 cm × 120 cm in size, were randomly distributed between three treatments (n=10): ambient control, UV‐A control and UV‐B treatment. The UV‐B‐treated plots were exposed to elevated UV‐B radiation level for three growing seasons. The instantaneous net ecosystem CO₂ exchange (NEE) and dark respiration (R_TOT) were measured during the growing season using a closed chamber method. The wintertime CO₂ emissions were estimated using a gradient technique by analyzing the CO₂ concentration in the snow pack. In addition to the instantaneous CO₂ exchange, the seasonal CO₂ balances during the growing seasons were modeled using environmental data measured at the site. In general, the instantaneous NEE at light saturation was slightly higher in the UV‐B treatment compared with the ambient control, but the gross photosynthesis was unaffected by the exposure. The R_TOT was significantly lower under elevated UV‐B in the third study year. The modeled seasonal (June–September) CO₂ balance varied between the years depending on the ground water level and temperature conditions. During the driest year, the seasonal CO₂ balance was negative (net release of CO₂) in the ambient control and the UV‐B treatment was CO₂ neutral. During the third year, the seasonal CO₂ uptake was 43±36 g CO₂‐C m⁻² in the ambient control and 79±45 g CO₂‐C m⁻² in the UV‐B treatment. The results suggest that the long‐term exposure to high UV‐B radiation levels may slightly increase the CO₂ accumulation to fens resulting from a decrease in microbial activity in peat. However, it is unlikely that the predicted development of the level of UV‐B radiation would significantly affect the CO₂ balance of fen ecosystems in future.     

Severe drought effects on ecosystem CO2 and H2O fluxes at three Mediterranean evergreen sites: revision of current hypotheses?

Eddy covariance and sapflow data from three Mediterranean ecosystems were analysed via top‐down approaches in conjunction with a mechanistic ecosystem gas‐exchange model to test current assumptions about drought effects on ecosystem respiration and canopy CO₂/H₂O exchange. The three sites include two nearly monospecific Quercus ilex L. forests – one on karstic limestone (Puéchabon), the other on fluvial sand with access to ground water (Castelporziano) – and a typical mixed macchia on limestone (Arca di Noè). Estimates of ecosystem respiration were derived from light response curves of net ecosystem CO₂ exchange. Subsequently, values of ecosystem gross carbon uptake were computed from eddy covariance CO₂ fluxes and estimates of ecosystem respiration as a function of soil temperature and moisture. Bulk canopy conductance was calculated by inversion of the Penman‐Monteith equation. In a top‐down analysis, it was shown that all three sites exhibit similar behaviour in terms of their overall response to drought. In contrast to common assumptions, at all sites ecosystem respiration revealed a decreasing temperature sensitivity ( Q ₁₀) in response to drought. Soil temperature and soil water content explained 70–80% of the seasonal variability of ecosystem respiration. During the drought, light‐saturated ecosystem gross carbon uptake and day‐time averaged canopy conductance declined by up to 90%. These changes were closely related to soil water content. Ecosystem water‐use efficiency of gross carbon uptake decreased during the drought, regardless whether evapotranspiration from eddy covariance or transpiration from sapflow had been used for the calculation. We evidence that this clearly contrasts current models of canopy function which predict increasing ecosystem water‐use efficiency (WUE) during the drought. Four potential explanations to those results were identified (patchy stomatal closure, changes in physiological capacities of photosynthesis, decreases in mesophyll conductance for CO₂, and photoinhibition), which will be tested in a forthcoming paper. It is suggested to incorporate the new findings into current biogeochemical models after further testing as this will improve estimates of climate change effects on (semi)arid ecosystems' carbon balances.     

Reduction of ecosystem productivity and respiration during the European summer 2003 climate anomaly: a joint flux tower, remote sensing and modelling analysis

The European CARBOEUROPE/FLUXNET monitoring sites, spatial remote sensing observations via the EOS‐MODIS sensor and ecosystem modelling provide independent and complementary views on the effect of the 2003 heatwave on the European biosphere's productivity and carbon balance. In our analysis, these data streams consistently demonstrate a strong negative anomaly of the primary productivity during the summer of 2003. FLUXNET eddy‐covariance data indicate that the drop in productivity was not primarily caused by high temperatures (‘heat stress’) but rather by limitation of water (drought stress) and that, contrary to the classical expectation about a heat wave, not only gross primary productivity but also ecosystem respiration declined by up to more than to 80 gC m⁻² month⁻¹. Anomalies of carbon and water fluxes were strongly correlated. While there are large between‐site differences in water‐use efficiency (WUE, 1–6 kg C kg⁻¹ H₂O) here defined as gross carbon uptake divided by evapotranspiration (WUE=GPP/ET), the year‐to‐year changes in WUE were small (<1 g kg⁻¹) and quite similar for most sites (i.e. WUE decreased during the year of the heatwave). Remote sensing data from MODIS and AVHRR both indicate a strong negative anomaly of the fraction of absorbed photosynthetically active radiation in summer 2003, at more than five standard deviations of the previous years. The spatial differentiation of this anomaly follows climatic and land‐use patterns: Largest anomalies occur in the centre of the meteorological anomaly (central Western Europe) and in areas dominated by crops or grassland. A preliminary model intercomparison along a gradient from data‐oriented models to process‐oriented models indicates that all approaches are similarly describing the spatial pattern of ecosystem sensitivity to the climatic 2003 event with major exceptions in the Alps and parts of Eastern Europe, but differed with respect to their interannual variability.     

Modeling analysis of primary controls on net ecosystem productivity of seven boreal and temperate coniferous forests across a continental transect

Process‐based models are effective tools to synthesize and/or extrapolate measured carbon (C) exchanges from individual sites to large scales. In this study, we used a C‐ and nitrogen (N)‐cycle coupled ecosystem model named CN‐CLASS (Carbon Nitrogen‐Canadian Land Surface Scheme) to study the role of primary climatic controls and site‐specific C stocks on the net ecosystem productivity (NEP) of seven intermediate‐aged to mature coniferous forest sites across an east–west continental transect in Canada. The model was parameterized using a common set of parameters, except for two used in empirical canopy conductance–assimilation, and leaf area–sapwood relationships, and then validated using observed eddy covariance flux data. Leaf Rubisco‐N dynamics that are associated with soil–plant N cycling, and depend on canopy temperature, enabled the model to simulate site‐specific gross ecosystem productivity (GEP) reasonably well for all seven sites. Overall GEP simulations had relatively smaller differences compared with observations vs. ecosystem respiration (RE), which was the sum of many plant and soil components with larger variability and/or uncertainty associated with them. Both observed and simulated data showed that, on an annual basis, boreal forest sites were either carbon‐neutral or a weak C sink, ranging from 30 to 180 g C m^?2 yr^?1; while temperate forests were either a medium or strong C sink, ranging from 150 to 500 g C m^?2 yr^?1, depending on forest age and climatic regime. Model sensitivity tests illustrated that air temperature, among climate variables, and aboveground biomass, among major C stocks, were dominant factors impacting annual NEP. Vegetation biomass effects on annual GEP, RE and NEP showed similar patterns of variability at four boreal and three temperate forests. Air temperature showed different impacts on GEP and RE, and the response varied considerably from site to site. Higher solar radiation enhanced GEP, while precipitation differences had a minor effect. Magnitude of forest litter content and soil organic matter (SOM) affected RE. SOM also affected GEP, but only at low levels of SOM, because of low N mineralization that limited soil nutrient (N) availability. The results of this study will help to evaluate the impact of future climatic changes and/or forest C stock variations on C uptake and loss in forest ecosystems growing in diverse environments.     

Soil CO2 efflux in contrasting boreal deciduous and coniferous stands and its contribution to the ecosystem carbon balance

Similar nonsteady‐state automated chamber systems were used to measure and partition soil CO₂ efflux in contrasting deciduous (trembling aspen) and coniferous (black spruce and jack pine) stands located within 100 km of each other near the southern edge of the Boreal forest in Canada. The stands were exposed to similar climate forcing in 2003, including marked seasonal variations in soil water availability, which provided a unique opportunity to investigate the influence of climate and stand characteristics on soil CO₂ efflux and to quantify its contribution to the net ecosystem CO₂ exchange (NEE) as measured with the eddy‐covariance technique. Partitioning of soil CO₂ efflux between soil respiration (including forest‐floor vegetation) and forest‐floor photosynthesis showed that short‐ and long‐term temporal variations of soil CO₂ efflux were related to the influence of (1) soil temperature and water content on soil respiration and (2) below‐canopy light availability, plant water status and forest‐floor plant species composition on forest‐floor photosynthesis. Overall, the three stands were weak to moderate sinks for CO₂ in 2003 (NEE of ?103, ?80 and ?28 g C m^?2 yr^?1 for aspen, black spruce and jack pine, respectively). Forest‐floor respiration accounted for 86%, 73% and 75% of annual ecosystem respiration, in the three respective stands, while forest‐floor photosynthesis contributed to 11% and 14% of annual gross ecosystem photosynthesis in the black spruce and jack pine stands, respectively. The results emphasize the need to perform concomitant measurements of NEE and soil CO₂ efflux at longer time scales in different ecosystems in order to better understand the impacts of future interannual climate variability and vegetation dynamics associated with climate change on each component of the carbon balance.     

Impact of the Asian monsoon climate on ecosystem carbon and water exchanges: a wavelet analysis and its ecosystem modeling implications

The monsoon system is an important natural driver of ecosystem carbon and water exchanges in Asia and is being altered by anthropogenic forcings. This system is accompanied by heavy rainfall and typhoons in the main growing season, thus causing alterations of environmental conditions such as rainfall, wind, and temperature; therefore, it acts as a natural disturbance to forests in Asia. Therefore, degradation of ecosystem service by monsoon activity reinforced by anthropogenic factors in a changing climate is of great concern. In this study, we presented observational evidences for the interplay of terrestrial carbon and water dynamics with the Asian monsoon and their implication in ecosystem modeling. We analyzed 3‐year eddy‐covariance data at a temperate deciduous forest in Korea. We used wavelet power and coherence spectra to investigate the Asian monsoon system and to determine its impact on the ecosystem. During the study period, our analysis showed strong coupling between ecosystem functioning and temporal variations of monsoon climate. Further scrutiny on the model outputs showed that the model did not accurately reproduce the observed plant phenology and thus ecosystem carbon and water exchanges disturbed by monsoon activities. Our findings suggest that under projected climate scenarios, terrestrial carbon sinks in monsoon Asia will decline if the monsoon disturbance will exceed its natural range of variation and if there is no enhancement in the robustness of the ecosystem in this region.     

Fishery-induced changes in a marine ecosystem: insight from models of the Gulf of Thailand

Two mass-balance trophic models are constructed to describe the Gulf of Thailand ecosystem (10–50 m depth): one model pertains to the initial phase of fisheries development, and the other to when the resources were severely depleted. The two phases are compared, and changes brought about by fishing discussed. A dynamic simulation model, Ecosim, is then used successfully to reproduce the 1980 state of the fishery based on the 1963 model and the development in catches. In addition the 1980 model is used to predict how the ecosystem groups may bounce back following marked reduction in fishing pressure. Finally, the 1963 model is used to study alternative scenarios for how the fisheries development could take place, notably the effect of exploiting only the resources of larger species. The study validates that the Ecosim model can be used to predict ecosystem level changes following changes in fishing pressure, therefore fishing induced changes can to a large extent explain the changes in ecosystem pools and fluxes observed over time.     

Importance of recent shifts in soil thermal dynamics on growing season length, productivity, and carbon sequestration in terrestrial high-latitude ecosystems

In terrestrial high‐latitude regions, observations indicate recent changes in snow cover, permafrost, and soil freeze–thaw transitions due to climate change. These modifications may result in temporal shifts in the growing season and the associated rates of terrestrial productivity. Changes in productivity will influence the ability of these ecosystems to sequester atmospheric CO₂. We use the terrestrial ecosystem model (TEM), which simulates the soil thermal regime, in addition to terrestrial carbon (C), nitrogen and water dynamics, to explore these issues over the years 1960–2100 in extratropical regions (30–90°N). Our model simulations show decreases in snow cover and permafrost stability from 1960 to 2100. Decreases in snow cover agree well with National Oceanic and Atmospheric Administration satellite observations collected between the years 1972 and 2000, with Pearson rank correlation coefficients between 0.58 and 0.65. Model analyses also indicate a trend towards an earlier thaw date of frozen soils and the onset of the growing season in the spring by approximately 2–4 days from 1988 to 2000. Between 1988 and 2000, satellite records yield a slightly stronger trend in thaw and the onset of the growing season, averaging between 5 and 8 days earlier. In both, the TEM simulations and satellite records, trends in day of freeze in the autumn are weaker, such that overall increases in growing season length are due primarily to earlier thaw. Although regions with the longest snow cover duration displayed the greatest increase in growing season length, these regions maintained smaller increases in productivity and heterotrophic respiration than those regions with shorter duration of snow cover and less of an increase in growing season length. Concurrent with increases in growing season length, we found a reduction in soil C and increases in vegetation C, with greatest losses of soil C occurring in those areas with more vegetation, but simulations also suggest that this trend could reverse in the future. Our results reveal noteworthy changes in snow, permafrost, growing season length, productivity, and net C uptake, indicating that prediction of terrestrial C dynamics from one decade to the next will require that large‐scale models adequately take into account the corresponding changes in soil thermal regimes.     

Carbon balance of young birch trees grown in ambient and elevated atmospheric CO2 concentrations

We constructed a carbon budget for young birch trees grown in ambient and elevated CO₂ concentrations over their fourth year of growth. The annual total of net leaf photosynthesis was 110% more in elevated CO₂ than in ambient CO₂. However, the trees in elevated CO₂ grew only 59% more biomass than the trees in ambient CO₂ over the year. Modelling studies showed that larger loss of carbon from fine-root production and growth of the root-associated mycorrhiza by the trees in elevated CO₂ probably accounted for all the remaining difference in net photosynthesis between the two treatments. Our modelling also showed that the fraction of net photosynthate consumed by respiration of nonleaf tissue was similar in the two CO₂ treatments, and was 26% and 24% for trees in ambient and elevated CO₂, respectively. Trees in elevated CO₂ had 43% more leaves, and produced 110% more net photosynthate than trees in ambient CO₂, even though the maximum rate of carboxylation per unit leaf nitrogen decreased by 21%. Sensitivity studies showed that down-regulation reduced the annual net photosynthetic production of the trees in elevated CO₂ by only 6%. Direct effects of higher CO₂ on photosynthesis and greater leaf area of the trees in elevated CO₂ increased the net photosynthesis of the trees by 68% and 60%, respectively; and together accounted for most of the difference in net photosynthesis between the two treatments.     

Carbon storage in forest soil of Finland. 2. Size and regional pattern

For confidently estimating the amount of carbon stored in boreal forestsoil, better knowledge of smaller regions is needed. In order to estimatethe amount of soil C in forests on mineral soil in Finland, i.e. excludingpeatland forests, and illustrate the regional patterns of the storage,statistical models were first made for the C densities of the organic and0–1 m mineral soil layers. A forest type, which indicated siteproductivity, and the effective temperature sum were used asexplanatory variables of the models. In addition, a constant C densitywas applied for the soil layer below the depth of 1 m on sortedsediments. Using these models the C densities were calculated for atotal of 46673 sites of the National Forest Inventory (NFI). The amountof the soil C was then calculated in two ways: 1) weighting the Cdensities of the NFI sites by the land area represented by these sites and2) interpolating the C densities of the NFI sites for 4 ha blocks to coverthe whole land area of Finland and summing up the blocks on forestedmineral soil. The soil C storage totalled 1109 Tg and 1315 Tg, whencalculated by the areal weighting and the interpolated blocks,respectively. Of that storage, 28% was in the organic layer, 68% inthe 0–1 m mineral soil layer and 4% in the layer below 1 m. The totalsoil C equals more than two times the amount of C in tree biomass and20% of the amount of C in peat in Finland. Soil C maps made usingthe interpolated blocks indicated that the largest soil C reserves arelocated in central parts of southern Finland. The C storage of theorganic layer was assessed to be overestimated at largest by 13% andthat of the 0–1 m mineral soil layer by 29%. The largest error in theorganic layer estimate is associated with the effects of forest harvestingand in the mineral soil estimate with the stone content of the soil.     

Effects of Soil Texture on Belowground Carbon and Nutrient Storage in a Lowland Amazonian Forest Ecosystem

Soil texture plays a key role in belowground C storage in forest ecosystems and strongly influences nutrient availability and retention, particularly in highly weathered soils. We used field data and the Century ecosystem model to explore the role of soil texture in belowground C storage, nutrient pool sizes, and N fluxes in highly weathered soils in an Amazonian forest ecosystem. Our field results showed that sandy soils stored approximately 113 Mg C ha^-1 to a 1-m depth versus 101 Mg C ha^-1 in clay soils. Coarse root C represented a large and significant ecosystem C pool, amounting to 62% and 48% of the surface soil C pool on sands and clays, respectively, and 34% and 22% of the soil C pool on sands and clays to 1-m depth. The quantity of labile soil P, the soil C:N ratio, and live and dead fine root biomass in the 0–10-cm soil depth decreased along a gradient from sands to clays, whereas the opposite trend was observed for total P, mineral N, potential N mineralization, and denitrification enzyme activity. The Century model was able to predict the observed trends in surface soil C and N in loams and sands but underestimated C and N pools in the sands by approximately 45%. The model predicted that total belowground C (0–20 cm depth) in sands would be approximately half that of the clays, in contrast to the 89% we measured. This discrepancy is likely to be due to an underestimation of the role of belowground C allocation with low litter quality in sands, as well as an overestimation of the role of physical C protection by clays in this ecosystem. Changes in P and water availability had little effect on model outputs, whereas adding N greatly increased soil organic matter pools and productivity, illustrating the need for further integration of model structure and tropical forest biogeochemical cycling. Received 3 March 1999; accepted 27 August 1999.     

Seasonal hysteresis of net ecosystem exchange in response to temperature change: patterns and causes

Understanding how net ecosystem exchange (NEE) changes with temperature is central to the debate on climate change‐carbon cycle feedbacks, but still remains unclear. Here, we used eddy covariance measurements of NEE from 20 FLUXNET sites (203 site‐years of data) in mid‐ and high‐latitude forests to investigate the temperature response of NEE. Years were divided into two half thermal years (increasing temperature in spring and decreasing temperature in autumn) using the maximum daily mean temperature. We observed a parabolic‐like pattern of NEE in response to temperature change in both the spring and autumn half thermal years. However, at similar temperatures, NEE was considerably depressed during the decreasing temperature season as compared with the increasing temperature season, inducing a counter‐clockwise hysteresis pattern in the NEE–temperature relation at most sites. The magnitude of this hysteresis was attributable mostly (68%) to gross primary production (GPP) differences but little (8%) to ecosystem respiration (ER) differences between the two half thermal years. The main environmental factors contributing to the hysteresis responses of NEE and GPP were daily accumulated radiation. Soil water content (SWC) also contributed to the hysteresis response of GPP but only at some sites. Shorter day length, lower light intensity, lower SWC and reduced photosynthetic capacity may all have contributed to the depressed GPP and net carbon uptake during the decreasing temperature seasons. The resultant hysteresis loop is an important indicator of the existence of limiting factors. As such, the role of radiation, LAI and SWC should be considered when modeling the dynamics of carbon cycling in response to temperature change.     

Constraining rooting depths in tropical rainforests using satellite data and ecosystem modeling for accurate simulation of gross primary production seasonality

Accurate parameterization of rooting depth is difficult but important for capturing the spatio-temporal dynamics of carbon, water and energy cycles in tropical forests. In this study, we adopted a new approach to constrain rooting depth in terrestrial ecosystem models over the Amazon using satellite data [moderate resolution imaging spectroradiometer (MODIS) enhanced vegetation index (EVI)] and Biome-BGC terrestrial ecosystem model. We simulated seasonal variations in gross primary production (GPP) using different rooting depths (1, 3, 5, and 10 m) at point and spatial scales to investigate how rooting depth affects modeled seasonal GPP variations and to determine which rooting depth simulates GPP consistent with satellite-based observations. First, we confirmed that rooting depth strongly controls modeled GPP seasonal variations and that only deep rooting systems can successfully track flux-based GPP seasonality at the Tapajos km67 flux site. Second, spatial analysis showed that the model can reproduce the seasonal variations in satellite-based EVI seasonality, however, with required rooting depths strongly dependent on precipitation and the dry season length. For example, a shallow rooting depth (1–3 m) is sufficient in regions with a short dry season (e.g. 0–2 months), and deeper roots are required in regions with a longer dry season (e.g. 3–5 and 5–10 m for the regions with 3–4 and 5–6 months dry season, respectively). Our analysis suggests that setting of proper rooting depths is important to simulating GPP seasonality in tropical forests, and the use of satellite data can help to constrain the spatial variability of rooting depth.