Modelling CO2 Impacts on Forest Productivity
Process-based models are important tools for understanding the dynamics and functioning of forest ecosystems. In nearly all models, forest productivity is largely driven by carbon (C) input via photosynthesis, which reacts strongly positive to elevated CO2 (eCO2). The simulated increases in forest productivity in terms of wood production, however, have been much higher than experimental and observational evidence suggests, at least in those models that did not account for nutrient limitations. Experimental results show that in most investigated systems, eCO2 leads to increasing allocation of C below ground, without sustained enhancements of wood production. This translocation of C below ground, which also enhances nutrient uptake, appears to be larger than simulated in our current models. However, because of a lack of experimental data from real ecosystems, in particular from the tropics, the big question remains unresolved whether increasing levels of CO2 have and will lead to substantially increased forest productivity in terms of stand-level wood increment. To reduce the uncertainties, it will be necessary to develop new theory, implement that in models, and rigorously compare model results with observations. They range from physiological measurements, results from “free air CO2 enrichment” experiments to forest inventories, tree rings and 13C analyses (as a proxy for changes in water use efficiency). Key issues that should be addressed in the models include the coupling of C and nutrient cycles, limitations to the C sink capacity of trees, dynamic changes in allocation, and potential feedbacks from tree population dynamics on stand productivity.
KeywordsForest productivity CO2 fertilization Net primary productivity (NPP) Forest modelling Allocation
CO2 is a limiting resource for photosynthesis [1, 2, 3]. CO2 also affects leaf stomatal density  and stomatal conductance . The direct effect on photosynthesis of increasing CO2 is often referred to as “CO2 fertilization,” which is more pronounced in plants with C3 photosynthesis  because in plants with C4 or CAM photosynthetic pathways, the CO2 concentration around the enzyme catalyzing the carboxylation reaction is elevated by the plant and decoupled from the atmospheric concentration. All trees, however, are C3 plants. Under elevated CO2 (eCO2), stomatal conductance generally decreases, which also leads to lower transpiration and water usage, whereby the decrease in transpiration also depends on how well leaves or needles are coupled to the atmosphere .
As photosynthesis is the primary entry point for all carbon (C) used by autotrophic life to synthesize carbohydrates and ultimately build biomass, understanding the effects of increasing levels of atmospheric CO2 on photosynthesis is of utmost importance for predicting forest growth. The basic biogeochemistry of photosynthesis, including the direct CO2 response under optimal conditions, is relatively well understood. However, it has been heavily debated how this response is modulated if growth conditions and the supply of other resources are limiting (e.g., [12, 13, 14]), and whether increased photosynthesis leads to enhanced forest productivity under those conditions [13, 15].
Other factors that might limit the extent to which plants can take advantage of increasing atmospheric CO2 are environmentally imposed stress effects on the plant, e.g., as a result of very high temperatures, and other limiting resources, in particular nutrients and water. As these other factors might exert substantial limitations on photosynthesis and growth, it has been suggested that at current CO2 levels, plants are generally no longer CO2-limited .
Enhanced photosynthesis does not necessarily lead to increased forest productivity in terms of wood biomass production, primarily because plants might allocate the extra C to other functions than biomass growth, such as increased fine root production and turnover, root exudates and C transfer to mycorrhiza. The two latter processes each may utilize up to 30 % of assimilated C . In a mature broad-leaved forest in Switzerland exposed to elevated CO2 with free air CO2 enrichment (FACE) technology, there was no significant enhancement of stem growth and leaf litter production, even though photosynthesis was persistently increased under eCO2, suggesting increased C allocation below ground [15, 17]. Also at a highly nutrient-limited site in northern Sweden, eCO2 in open-top chambers led to strong photosynthesis enhancements, but growth was only enhanced with optimum nutrient supply .
Forest models generally adopt one of two very different approaches for modelling productivity. (i) Empirically calibrated forest models commonly simulate a parameterized maximum productivity, expressed as biomass or stem diameter increment per year, which is modulated by environmental factors (soil conditions and climate, e.g., ). Such models cannot mechanistically represent the potential effects of changing CO2 levels and are therefore not discussed any further in the following. (ii) “Process-based” forest models aim to represent the physiological processes and, in some cases, population dynamics that ultimately affect forest productivity. Forest productivity in such models thus emerges as the result of complex interactions between a variety of processes, such as photosynthesis, autotrophic respiration, and allocation of C to different tree compartments.
In this paper, we focus on the approaches for modelling CO2 impacts on forest productivity in process-based forest and dynamic vegetation models, the results from these models up to date, and how well the model results correspond with available empirical evidence. We outline important shortcomings in current models and how the models could be improved.
Process-Based Models for Simulating Forest Productivity
Photosynthesis and Transpiration
Most process-based forest models adopt a modified version of the Farquhar photosynthesis model  for simulating the effects of light and CO2 on photosynthesis. The Farquhar model simulates the co-limitation of photosynthesis by carboxylation and electron transport capacities. In such models, photosynthesis increases strongly with increasing CO2 if other resources are not limiting, as shown in Fig. 1. The specificity for CO2 of the primary enzyme that catalyzes both the carboxylation and oxygenation (photorespiration) of the acceptor molecule of C3 photosynthesis (ribulose bisphosphate carboxylase/oxygenase [Rubisco]) decreases with temperature, implying stronger potential CO2 fertilization at higher temperatures . Acclimation to higher CO2 occurs in the models via increased leaf C:N ratios (and associated carboxylation capacity) [29••], which also has been observed in experiments (e.g., ), but to a lower extent than predicted by models ([29••], see also section on FACE experiments below).
Forest models generally differ in the way limitations by environmental factors or other resources are implemented. In the widely used Lund-Potsdam-Jena (LPJ) Dynamic Global Vegetation Model (DGVM) , for example, an empirical temperature inhibition function is used to constrain photosynthesis at low and high temperatures. Water limitation is represented as a decrease in the stomatal conductance (limiting CO2 uptake into the leaf) when water supply cannot meet the transpirational water demand associated with optimum photosynthesis without water limitation. Water supply is modelled as a function of root-distribution-weighted soil water content and a maximum transpiration rate . This approach takes account of the close coupling between CO2 uptake and water loss through stomata . In most models, photosynthesis and transpiration are coupled and stomatal or canopy conductance decreases with eCO2, but the details of this coupling differ between models . Water use efficiency (WUE, i.e., C uptake per unit water transpired) is generally predicted to increase under eCO2, which has also been observed in FACE experiments . An increase in WUE over the last 150 years as a result of increasing CO2 has also been inferred from tree ring analyses and isotopic data (13C). However, additional analyses of ring widths indicated that the increase in WUE was not translated into increased biomass growth [34, 35, 36]. Note, however, that there is only weak, if any, evidence for conifers to reduce stomatal conductance under eCO2 [6, 37]. Forest models generally adopt generalized schemes for the coupling of photosynthesis and transpiration, which simulate reductions in stomatal conductance for broad-leaved trees and conifers. Simulated substantial increases in runoff because of reduced transpiration under eCO2 (e.g., ) might, therefore, not be realistic for the large areas covered by boreal conifers .
Part of the assimilated C is then used by respiratory processes, and the remainder is the net primary productivity (NPP), which is available for plant growth, the replacement of tissues, e.g., replacing leaf fall and fine root turnover, root exudates (not explicitly represented in most models), etc. Plant respiration is either modelled as a fixed fraction of gross primary productivity (GPP, approximately 50 %) or split into leaf respiration (scaled to maximum carboxylation capacity), growth respiration (commonly 25 % of GPP or new growth), and temperature- and tissue C:N-dependent maintenance respiration of wood and fine roots (e.g., ).
In the latter approach, maintenance respiration increases with temperature either adopting a Q10 relationship (where respiration rates increase with a fixed relative rate per given temperature increase, resulting in an exponential increase) or an Arrhenius-type response, which assumes that the Q10 factor decreases with temperature . Models that only increase photosynthesis and simulate plant maintenance respiration independently of photosynthesis can simulate higher CO2 fertilization (in terms of NPP) than models that scale total autotrophic respiration to photosynthesis. However, under strong temperature increases, models that assume a fixed fraction of plant respiration avoid carbon starvation that can potentially arise from increasing respiration when photosynthesis is, e.g., strongly limited by water. They can, therefore, also yield more positive NPP enhancement under eCO2. If and how eCO2 affects plants respiratory processes is poorly understood .
NPP is allocated to different compartments of the tree, such as fine roots, stems, and leaves, and to replace tissue that is turned over (especially fine roots and leaves). This allocation is crucial for plant growth. In many models, NPP allocation is simulated by allocating a fixed fraction of assimilated C to each plant compartment. The limitation of this approach is that allocation is not responding to environmental changes such as increasing atmospheric CO2 concentrations (e.g., ). Thus, a given NPP enhancement directly leads to a certain increase in biomass production. Instead, there is good empirical evidence that carbon allocation patterns can change strongly under eCO2 conditions [42••]. Results from FACE experiments have shown that, under eCO2, more C is allocated below ground, including fine roots, and most probably also root exudates and C transfer to mycorrhiza fungi, which can increase the N uptake by plants ([42••, 43], see also section on FACE below). Most models do not explicitly represent root exudates and C transfer to mycorrhiza. In models that simulate increased allocation to fine roots, if nutrients become more limiting under eCO2, this allocation shift also increases the nutrient uptake capacity [42••].
Early theoretical modelling work demonstrated that the NPP response to increasing atmospheric CO2 should become progressively limited by N over time, a concept known as progressive nitrogen limitation (PNL). In a simple ecosystem model (the Generic Decomposition and Yield [GDAY] model), PNL occurred after a step increase of atmospheric CO2 as N was sequestered in woody biomass and soil organic matter, leaving less soil N to be mineralized, and leaf C:N ratios increased at eCO2 . In this study, soil C–N cycling was simulated based on the CENTURY model , which has since been adopted by many forest models (e.g., LPJ-GUESS  and O–CN ). These results are consistent with experimental evidence that NPP is strongly limited by N availability in many ecosystems across the world , in particular in boreal forests .
In the widely adopted CENTURY approach, the soil N cycle is basically attached to a soil C cycle via C:N ratios of different soil organic matter and microbial pools, which can change within given bounds . Thus, additional C input under eCO2 locks up N if soil organic matter and/or microbial C:N ratios do not increase. It follows that carbon storage in N-limited regions can only increase if one of the following occurs: (i) total N increases, (ii) C:N ratios of a given pool increase, (iii) pools with wide C:N ratios (such as wood) increase relative to pools with narrow C:N ratios, or (iv) N losses decrease (see also ).
In tropical regions, P is assumed to be a main limiting factor for forest productivity (e.g., [50-53]). As in the case of N, theoretical consideration and modelling studies suggest that higher C uptake under eCO2 leads to more immobilization of P in organic matter (assuming that soil organic matter C:P ratios do not substantially change), leaving less P available for plants (e.g., ). However, results from modelling studies also show that, under eCO2, plants may actively make additional P available by increasing carbon allocation to fine roots, mycorrhiza fungi, and soil microbes and thus maintain increased productivity [51, 53], and increasing eCO2 can also lead to reduced nutrient losses from the ecosystem . However, we are not aware of any forest stand-scale experimental data to test these results. On the other hand, limitations from other resources, e.g., calcium or potassium, may become more limiting under eCO2 (e.g., ) and are rarely considered in forest models. Goll et al.  included a simple N and P cycle in the land surface model JSBACH and found a reduced land C uptake by 25 % compared to simulations without nutrient cycling.
In spite of the clear importance of nutrient limitations, to date, they have been ignored in many forest or dynamic global vegetation modelling studies. Not surprisingly, the model results suggested strongly increasing NPP in a future world with higher atmospheric CO2, at least in those areas where other resources were not strongly limiting (e.g., [55, 56]). The models commonly simulated a rather “green world” in the future (e.g., [56, 57••]), which has been criticized as overly optimistic, for example, because it would be unrealistic to expect the stoichiometric N demand for the simulated increase in biomass to be met . Progress has been hampered by the fact that most DGVM studies focused on the overall carbon cycle, i.e., the C sink or source strength of the global terrestrial biosphere, instead of further investigating the component processes. The results from the first global modelling studies that included N feedbacks on C cycling indeed suggested that the future C uptake by terrestrial ecosystems will be much smaller than indicated by the results of C-only models [59, 60, 61, 62].
However, as the N cycle became incorporated into more models, it became clear that the story might not be that simple. Wårlind et al. [63•] summarized the results from a number of DGVMs run as C-only or C–N models in terms of the simulated C sequestration between the years 1850 and 2100 (using a climate and CO2 change scenario for the future). All models except LPJ-GUESS simulated lower increases (over time) in C sequestration in the C–N version than in the C-only version (8 to 27 % lower), whereas LPJ-GUESS simulated 17 % higher C sequestration in the C–N version (even though biomass pools were generally much lower in the C–N version, in particular in boreal forests, where the simulation results appeared to be more realistic than with the C-only version). CO2 fertilization effects were indeed substantially smaller in the C–N version than in the C-only version, but higher temperatures lead to higher N mineralization, especially in the mid to high latitudes. This climatically driven “N fertilization” led to strong increases in woody biomass growth under climate warming, which overrode soil C losses as a result of warming [63•]. Note also that additional N demand (to sustain increased biomass) under a business-as-usual climate and CO2 scenario in the LPJ-GUESS C–N version was between the low and high N supply estimate by Hungate et al.  and, thus, fully consistent with their analysis . The lower additional N demand in LPJ-GUESS compared to earlier DGVMs (also with the C-only version) might be rooted in the more detailed and presumably more realistic treatment of vegetation dynamics in this tree-individual-based model. Detailed representations of vegetation dynamics can result in longer lag times in the vegetation response and more realistic representations of ecosystem-scale changes in allometry (and associated C:N ratios), which emerge from the growth of individual trees. In LPJ-GUESS, additional carbon storage in mid to high latitudes under eCO2 and climate change were partly driven by increasing woody biomass with wide C:N ratios [46, 63•].
C Sink Limitation
As outlined above, allocation of NPP to different plant compartments and plant growth is in most forest models treated in a rather simplistic manner and strongly driven by photosynthesis. This may result from the fact that our current understanding of allocation processes is limited and that few data are available to calibrate or test models [42••]. Most forest models assume that plant growth is primarily limited by the availability of carbon (“carbon limitation”) rather than by the potential to use the available carbon (“sink limitation,” [16, 64, 65]). Sink activity may be limited under environmental conditions that are not beneficial to growth, such as low temperatures, which limit any metabolic activity, drought (e.g., as growth may also be limited by cell turgor), or low nutrient availability [66•]. These processes are not explicitly represented in most forest models. In this context, non-structural carbohydrates (NSCs) are assumed to play an important regulating role in plant metabolism (e.g., [64, 67, 68]). Such mobile NSC stores in the plant can be seen as indicators for the plants’ status regarding sink limitation . Elevated atmospheric CO2 concentrations will clearly lead to increased source activity (i.e., photosynthesis) and increased C supply but different constraints will work on the sink activity such as temperature and water limitations for tissue formation . Fatichi et al. [66•] suggested that current DGVMs should be improved by treating the amount of C allocated to growth independently from NPP. Leuzinger et al.  developed a first attempt for simulating sink limitation to plant growth along tree lines as a result of cold growing seasons, which resulted in lower and more realistic biomass estimates by LPJ-DGVM. However, in the closely related LPJ-GUESS model, biomass at the tree line also decreases substantially when nitrogen limitation is accounted for , and one should keep in mind that forest models, even without explicit representation of sink limitations, include factors to account for it implicitly, such as the empirically calibrated temperature inhibition function for photosynthesis (which also constrains plant growth), or the coupling between water demand and supply in the LPJ family of models mentioned above. Nevertheless, we agree with the general criticism that current forest models are too much source-driven and that sink limitation should be addressed more explicitly in forest models.
Dynamic vegetation models, in particular, also simulate the establishment of tree saplings and tree mortality, which influence the vegetation structure and, via changes in leaf area index (LAI), forest productivity. Establishment of saplings commonly decreases with canopy closure, and tree mortality increases as the growth efficiency, expressed as, e.g., NPP per leaf area, decreases and as trees approach their maximum age or size. These processes are strongly affected by competition for resources. As trees can reach their maximum age faster under eCO2, increases in CO2 could lead to higher mortality. Bugmann and Bigler [71••] used observations on the relationship between growth and longevity and the ForClim model to investigate the potential effects of feedbacks from changes in mortality on stand-level CO2 fertilization. (Note that CO2 fertilization effects in that study were implemented as a simple multiplier on tree growth, not mechanistically). The authors concluded that the reduction of tree longevity as a result of eCO2 tends to offset growth-related biomass increases in the model. Even if the growth-induced reduction in longevity was not accounted for in the model, eCO2 only led to very minor biomass increments at the stand level because of more intense competition and increases in tree mortality. These results confirm the idea that eCO2, at least in forests with dense canopies and intense competition, only accelerates C turnover in ecosystems and has no pronounced effects on carbon storage . Disturbances, such as forest pests, wind, and fire, are also important drivers of population dynamics. Forest models adopt a wide variety of approaches to represent disturbances, but summarizing these would be beyond the scope of this article.
Simulation Results Concerning CO2 Effects on Forest Productivity
We start with summarizing model results because research on the physiological effects of CO2 on forest productivity has been very much driven by model results. Early on, it has become clear that CO2 effects in the models are very substantial, and that it is hardly possible to project forest productivity and carbon storage into the future without understanding physiological CO2 effects (e.g., ). Then, the modelling results were increasingly confronted with data.
One common approach to capture the uncertainty associated with potential physiological CO2 effects is to run the models with and without CO2 effects (the latter by keeping CO2 constant, thereby switching off both the effects on photosynthesis and stomatal conductance, but maintaining the climatic effects of CO2). Cramer et al.  found that even the sign of the CO2 transfer between the atmosphere and terrestrial biosphere projected by six DGVMs strongly depended on CO2 effects.
The potentially crucial role of eCO2 on vegetation productivity, biomass stocks, and the global carbon cycle has since been confirmed by many modelling studies. Ollinger et al. , for example, used the PnET-CN model to simulate forest growth in the northeastern USA for the period of 1700–2000. They found substantial enhancements by increasing CO2, although N deposition, ozone exposure, and former land use were also found to be important drivers. Also based on results from the PnET-CN model, Pan et al.  concluded that changes in atmospheric chemistry (N deposition, ozone, and CO2) might have caused an NPP increase of 29 % in Mid-Atlantic temperate forests in North America during the twentieth century, of which 14 % was caused by increasing CO2 according to the model.
Over the last decades, there have indeed been many reports of increasing forest productivity from many forest regions across the world (e.g., [74, 75, 76, 77]). It is often assumed that CO2 fertilization is one important driver of these increases, but this assumption is mainly based on model results (e.g., ). However, any increase in forest productivity over the last decades is consistent with results from forest models that respond very positively to steadily increasing CO2 without necessarily proving causality. Successional changes, for example, could also play an important role, and the actual stand age and structure have commonly not been implemented in the models. Thus, we think that the common assumption that CO2 effects played an important role for these increases (e.g., ) is not based on very solid evidence.
Regarding future projections, Pan et al.  used three biogeochemistry models (the Terrestrial Ecosystem Model [TEM], the BioGeochmical Cycles [BIOME-BGC] model, and CENTURY) to simulate the potential NPP response for the conterminous USA to a doubling in atmospheric CO2. The models simulated increases between 5 and 11 %. Kirschbaum et al.  projected wood productivity responses of radiata pine (Pinus radiata) forests in New Zealand into the future using the CenW model with climate change projections from several global climate models (GCMs) and CO2 emission scenarios from the Special Report on Emissions Scenarios (SRES). With physiological CO2 effects enabled, the model projected an average increase of 37 % by 2090, while projected growth responses were only minor when keeping CO2 constant. Keenan et al.  used the GOTILWA+ model to project the productivity of three common tree species in Spain into the future, using scenarios of climate and CO2 change. When only including climatic effects, the model predicted average future decreases in NPP of about 30 % by the end of the century. Simulations with climatic and CO2 effects combined gave increases of about 20 % for two species and only minor changes for the third species. Scheiter and Higgins , Higgins and Scheiter , and Scheiter et al.  simulated substantial future increases in forest productivity and biomass, as well as savannah to forest transitions, with the adaptive DGVM (aDGVM) across Africa and savannah regions in northern Australia. For Australia, the authors also conducted factorial experiments to separate the effects of increasing CO2 and climate change. Under the SRES A1B emission scenario, average above-ground biomass increased from 14.2 to 32.8 t ha−1 (assuming no change in rainfall seasonality). The factorial analyses revealed that only CO2 had a statistically significant effect on the biomass increase . These simulations also highlighted the role of complex interactions between resource competition (in particular between C3 and C4 plants), tree population dynamics, and fire, such as a higher chance for tree saplings to escape the fire horizon under eCO2 [81, 83].
The Amazon Forest
Experimental Results and Model-Data Intercomparisons
Early experiments under laboratory conditions or with open-top chambers showed strong effects of eCO2 (commonly around 700 ppm) on photosynthesis and plant growth , i.e., strong increases in photosynthesis and growth under optimal nutrient supply and lower NPP enhancements with low nutrient supplies . However, the relevance of such experiments to real ecosystems has been questioned . Therefore, so-called free air CO2 enrichment (FACE, [15, 93]) experiments have been established.
FACE Model-Data Intercomparisons
In an early synthesis of the results from four such FACE experiments, Norby et al.  concluded that CO2 enhancement to about 550 ppm leads to an NPP enhancement of about 23 %, which was highly conserved across a broad range of productivities. Hickler et al.  parameterized the LPJ-GUESS vegetation model to represent the trees and stand structure at the sites referred to by Norby et al.  and applied the model with weather data recorded at those sites to test whether LPJ-GUESS could reproduce the observed NPP enhancement. The model indeed predicted an average response for the sites that was similar to the observations, but the results also indicated that the FACE results from temperate forests are not representative for other climate zones. In LPJ-GUESS, the simulated NPP enhancement is larger in warmer climates because Rubisco specificity for CO2 decreases with temperature (see above). In other DGVMs, water limitation has a stronger effect on the modelled NPP enhancement, with higher enhancement under dry conditions (compare Fig. S4 in  and Fig. A1 in ). As all these models have adopted similar representations of photosynthesis, differences in the implementation of water limitations and the coupling between C uptake and transpiration are most probably the reason for this discrepancy. A FACE result synthesis by Nowak et al. (, Fig. 7), however, suggests that the NPP enhancement is smaller if trees are strongly water limited, possibly reflecting sink limitation (see above).
As mentioned above, NPP enhancements do not necessarily lead to increased wood production. Increased wood production has been reported for the Duke forest (a loblolly pine [Pinus taeda] stand in North Carolina) for more than 10 years of CO2 enhancement, but at the Oak Ridge National Laboratory site (ORNL, a sweet gum [Liquidambar styraciflua] stand in Tennessee), most of the extra C has been allocated to compartments with high turnover rates, in particular fine roots [42••, 97]. In addition, results from Duke clearly show that the NPP enhancement has been larger in FACE rings with higher nutrient availability , and the initial strong NPP enhancement at ORNL has not been maintained due to PNL [99•].
In 2008, the FACE model-data synthesis (FACE-MDS, ) project was established. In this project, four ecosystem models, five land surface models (which are also embedded in climate models), and two dynamic vegetation models were compared to experimental results from the two of the FACE experiments that most closely resemble forests (Duke and ORNL) using a novel assumption-centred model intercomparison approach, in which individual component processes of the models were compared to experimental data [100••]. With this approach, it is possible to evaluate if individual processes in the models correspond well with experimental results and to identify shortcomings in the current models that should be addressed by further model development.
The main findings from FACE-MDS [101••] include that (i) even under nutrient-poor conditions, NPP can be enhanced under eCO2 [29••]; (ii) simulated NPP enhancements correspond reasonably well with the experimental results, but often for the wrong reasons ([29••]; see the next two points); (iii) in the models, substantial decreases in foliar N concentration make persistent NPP enhancements possible, but the models strongly overestimate the extent of changes in C:N ratios [29••]; (iv) the models do not reproduce the magnitude of observed increases of plant N uptake under eCO2, which, in reality, is probably caused by increased allocation of carbon below ground (fine roots, root exudates and transfer of C to mycorrhiza) [29••, 42••, 43]; (v) the results from models that use allometric relationships that vary in response to changing resource supplies correspond better with the observations than models with fixed allometric coefficients [42••]; (vi) PNL can be relaxed in models in the longer term if plant N uptake is not only a function of N supply but also of N demand, because in these models, eCO2 leads to lower N losses , which is consistent with observations at ORNL ; (vii) the observations show that changes in stomatal conductance and photosynthesis are closely coupled, as it has been implemented in most models ; and (viii) even though both sites are situated in a similar climate and on similar soils, the observed responses at the sites differed markedly: PNL was very strong at ORNL [99•] but not generally across FACE rings at Duke [29••]; C allocation shifted strongly below ground at ORNL but not at Duke [42••]; priming of soil N release was stronger at Duke [29••]; leaf mass area, which has often been reported to increase under eCO2 , increased substantially only at ORNL, and the pattern was not captured by any model. These large differences between the sites make it clear that further experiments will be necessary to derive generalizations that can be used in models to predict the future. Site and even species-specific responses have also been reported elsewhere. Dawes et al. , for example, reported that in a 9-year FACE experiment at the Swiss tree line, European larch (Larix decidua) showed increased tree ring increments but mountain pine (Pinus mugo ssp. uncinata) did not. Unfortunately, both the Duke and the ORNL sites have now been shut down. As a follow-up, model-data intercomparisons using the assumption-centred approach are now being carried out with the same set of models at four additional FACE sites located in the USA covering diverse ecosystems (scrub oak in Florida; mixed-grass prairie in Wyoming; desert shrub community in the Mojave Desert, Nevada; aspen clones in Wisconsin). To fill our knowledge gap concerning eCO2 effects in the tropics, a FACE experiment has recently been set up in a mature almost-original Eucalyptus forest close to Sydney (http://www.uws.edu.au/hie/facilities/EucFACE) and one is currently being established in the Amazon rainforest .
Understanding the physiological effects of increasing CO2 is of utmost importance for modelling past and future forest productivity. Substantial advances have recently been made in model development, in particular concerning the coupling of N and C, and through model-data intercomparisons. Nevertheless, large uncertainties remain. These include, e.g., the response of allocation patterns under changing levels of CO2 (in particular below-ground processes), the role of P limitation in the tropics, the functioning of C sink limitations, and feedbacks from tree population dynamics on stand-scale productivity. Increasing levels of CO2 strongly increase photosynthesis and, in most cases, also benefit tree growth through reduced water losses. But the extent to which forest productivity in terms of wood production has and will increase as a result of eCO2 is still highly uncertain. It seems likely that in most model simulations to date, CO2 effects on forest wood production have been overestimated. Reducing the uncertainties will require the development of new theory, model algorithms, and rigorous comparisons between models and data. Various types of data, ranging from physiological variables, such as NSC, to results from FACE experiments and forest inventories to isotopically inferred estimates of WUE will be necessary to improve our models.
The Oak Ridge and Duke FACE sites were supported by the US Department of Energy (DOE) Office of Science’s Biological and Environmental Research (BER) and the FACE model-data synthesis (FACE-MDS) activity by DOE and the US National Centre for Ecological Analysis and Synthesis (NCEAS, grant EF-0553768). Special thanks also to Richard J. Norby, Ram Oren, Belinda E. Medlyn, Soenke Zaehle, Anthony P. Walker, and Martin G. De Kauwe for very much driving FACE-MDS. A. Rammig acknowledges funding from the EU-FP7 project AMAZALERT (grant agreement no. 282664) and from the Helmholtz Alliance "Remote Sensing and Earth System Dynamics". We also acknowledge the World Climate Research Programme’s Working Group on Coupled Modelling, which is responsible for the Coupled Model Intercomparison Project (CMIP), and we thank the climate modelling groups for producing and making available their model output. For CMIP, the US Department of Energy’s Program for Climate Model Diagnosis and Intercomparison provides coordinating support and led development of software infrastructure in partnership with the Global Organization for Earth System Science Portals. Finally, we would like to thank the Section Editor Miko Kirschbaum for the extremely insightful comments on an earlier draft and for providing the photosynthesis model used to produce Fig. 1.
Compliance with Ethics Guidelines
Conflict of Interest
Dr. Hickler reports grants from US NCEAS during the conduct of the study.
Dr. Rammig and Dr. Werner both declare no conflict of interest.
Human and Animal Rights and Informed Consent
This article contains no studies with human or animal subjects performed by the author.
Papers of particular interest, published recently, have been highlighted as: • Of importance •• Of major importance
- 5.Medlyn BE, Barton CVM, Broadmeadow MSJ et al. Stomatal conductance of forest species after long-term exposure to elevated CO2 concentration: a synthesis. New Phytol. 2001;149(2):247–64.Google Scholar
- 6.Körner C, Morgan JA, Norby R. CO2 fertilisation: when, where, how much? In: Canadell SG, Pataki DE, Pitelka LF, editors. Terrestrial ecosystems in a changing world. Berlin: Springer; 2007. p. 9-21.Google Scholar
- 7.Long SP, Ainsworth EA, Rogers A et al. Rising atmospheric carbon dioxide: plants FACE the future. Annu Rev Plant Biol. 2004;55(1):591–628.Google Scholar
- 11.Chapin III RS, Matson PA, Mooney HA. Principles of terrestrial ecosystem ecology. New York: Springer; 2002.Google Scholar
- 12.Kirschbaum MUF, Medlyn BE, King DA et al. Modelling forest-growth response to increasing CO2 concentration in relation to various factors affecting nutrient supply. Glob Chang Biol. 1998;4(1):23–41.Google Scholar
- 14.Oren R, Ellsworth DS, Johnsen KH et al. Soil fertility limits carbon sequestration by forest ecosystems in a CO2-enriched atmosphere. Nature. 2001;411:469–72.Google Scholar
- 15.Körner C, Asshoff R, Bignucolo O et al. Carbon flux and growth in mature deciduous forest trees exposed to elevated CO2. Science. 2005;309(5739):1360–2.Google Scholar
- 20.Lasch P, Badeck FW, Suckow F et al. Model-based analysis of management alternatives at stand and regional level in Brandenburg (Germany). For Ecol Manag. 2005;207:59–74.Google Scholar
- 24.Keenan T, Davidson E, Moffat AM et al. Predicting the future of forests in the Mediterranean under climate change, with niche- and process-based models: CO2 matters! Glob Chang Biol. 2011;17(1):565–79.Google Scholar
- 25.Peters EB, Wythers KR, Zhang S et al. Potential climate change impacts on temperate forest ecosystem processes. Can J For Res. 2013;43:939–50.Google Scholar
- 26.Hickler T, Vohland K, Feehan J et al. Projecting the future distribution of European potential natural vegetation zones with a generalized, tree species-based dynamic vegetation model. Glob Ecol Biogeogr. 2012;21(1):50–63.Google Scholar
- 28.Zaehle S, Friend AD. Carbon and nitrogen cycle dynamics in the O-CN land surface model: 1. Model description, site-scale evaluation, and sensitivity to parameter estimates. Glob Biogeochem Cycles. 2010;24(1), GB1005.Google Scholar
- 29.••Zaehle S, Medlyn BE, De Kauwe MG et al. Evaluation of 11 terrestrial carbon-nitrogen cycle models against observations from two temperate Free-Air CO2 Enrichment studies. New Phytol. 2014;202(3):803–22. The authors compared individual processes in 11 ecosystem models to results from FACE experiments, with a special focus on processes that control NPP.Google Scholar
- 30.Crous KY, Walters MB, Ellsworth DS. Elevated CO2 concentration affects leaf photosynthesis - nitrogen relationships in Pinus taeda over nine years in FACE. Tree Physiol. 2008;28(4):607–14.Google Scholar
- 31.Sitch S, Smith B, Prentice IC et al. Evaluation of ecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJ Dynamic Global Vegetation Model. Glob Chang Biol. 2003;9:161–85.Google Scholar
- 32.Ball JT, Woodrow IE, Berry JA. A model predicting stomatal conductance and its contribution to the control of photosynthesis under different environmental conditions. Progress in Photosynthesis Research. Vol. 9; 1987.Google Scholar
- 33.De Kauwe MG, BE Medlyn, S. Zaehle et al. Forest water use and water use efficiency at elevated CO2: a model-data intercomparison at two contrasting temperate forest FACE sites. Glob Chang Biol. 2013;19(6):1759-79.Google Scholar
- 36.van der Sleen P, Groenendijk P, Vlam M et al. No growth stimulation of tropical trees by 150 years of CO2 fertilization but water-use efficiency increased. Nat Geosci. 2015;8(1):24–8.Google Scholar
- 38.Betts RA, Cox PM, Collins M et al. The role of ecosystem-atmosphere interactions in simulated Amazonian precipitation decrease and forest dieback under global climate. Theor Appl Climatol. 2004;78(1–3):157–75.Google Scholar
- 40.Leakey ADB, Ainsworth EA, Bernacchi CJ et al. Elevated CO2 effects on plant carbon, nitrogen, and water relations: six important lessons from FACE. J Exp Bot. 2009;60(10):2859–76.Google Scholar
- 41.Franklin O, Johansson J, Dewar RC et al. Modeling carbon allocation in trees: a search for principles. Tree Physiol. 2012;32(6):648–66.Google Scholar
- 42.••De Kauwe MG, Medlyn BE, Zaehle S, et al. Where does all the carbon go? A model-data intercomparison of vegetation carbon allocation and turnover processes at two temperate forest free-air CO2 enrichment sites. New Phytol. 2014;203(3):883-99. This paper demonstrated the crucial role of allocation changes for simulating the response of forest ecosystems to eCO 2.Google Scholar
- 43.Finzi AC, Norby RJ, Calfapietra C et al. Increases in nitrogen uptake rather than nitrogen-use efficiency support higher rates of temperate forest productivity under elevated CO2. Proc Natl Acad Sci. 2007;104(35):14014–9.Google Scholar
- 45.Parton WJ, Schimel D, Cole CV et al. Analysis of factors controlling soil organic matter levels in Great Plains grasslands. Soil Sci Soc Am J. 1987;51:1173–9.Google Scholar
- 46.Smith B, Wårlind D, Arneth A et al. Implications of incorporating N cycling and N limitations on primary production in an individual-based dynamic vegetation model. Biogeosciences. 2014;11(7):2027–54.Google Scholar
- 48.Kostiainen K, Kaakinen S, Saranpää P et al. Stem wood properties of mature Norway spruce after 3 years of continuous exposure to elevated [CO2] and temperature. Glob Chang Biol. 2009;15(2):368–79.Google Scholar
- 49.Rastetter EB, McKane RB, Shaver GR et al. Changes in C storage by terrestrial ecosystems: how can C-N interactions restrict responses to CO2 and temperature. Water Air Soil Pollut. 1992;64(1–2):327–44.Google Scholar
- 50.Quesada CA, Phillips OL, Schwarz M et al. Basin-wide variations in Amazon forest structure and function are mediated by both soils and climate. Biogeosciences. 2012;9(6):2203–46.Google Scholar
- 51.Yang X, Thornton PE, Ricciuto DM et al. The role of phosphorus dynamics in tropical forests—a modeling study using CLM-CNP. Biogeosciences. 2014;11:1667–81.Google Scholar
- 52.Vitousek PM, Porder S, Houlton BZ et al. Terrestrial phosphorus limitation: mechanisms, implications, and nitrogen-phosphorus interactions. Ecol Appl. 2010;20(1):5–15.Google Scholar
- 53.Cernusak LA, Winter K, Dalling JW, et al. Tropical forest responses to increasing atmospheric CO2: current knowledge and opportunities for future research. Funct Plant Biol. 2013;40(6):531-551.Google Scholar
- 54.Goll DS, Brovkin V, Parida BR et al. Nutrient limitation reduces land carbon uptake in simulations with a model of combined carbon, nitrogen and phosphorus cycling. Biogeosciences. 2012;9(9):3547–69.Google Scholar
- 55.Cramer W, Bondeau A, Woodward FI et al. Global response of terrestrial ecosystem structure and function to CO2 and climate change: results from six dynamic global vegetation models. Glob Chang Biol. 2001;7:357–73.Google Scholar
- 56.Sitch S, Huntingford C, Gedney N et al. Evaluation of the terrestrial carbon cycle, future plant geography and climate-carbon cycle feedbacks using five Dynamic Global Vegetation Models (DGVMs). Glob Chang Biol. 2008;14:1–25.Google Scholar
- 57.••Huntingford C, Zelazowski P, Galbraith D et al. Simulated resilience of tropical rainforests to CO2-induced climate change. Nat Geosci. 2013;6(4):268–73. The authors disentangled simulated future climatic and CO 2 effects in a DGVM and showed that, across the tropics, very strong positive CO 2 effects on vegetation biomass overrode negative effects of increasing temperatures in the model.Google Scholar
- 58.Hungate BA, Dukes JS, Shaw MR et al. Nitrogen and climate change. Science. 2003;302(5650):1512–3.Google Scholar
- 59.Bonan G. Carbon cycle: fertilizing change. Nat Geosci. 2008;1(10):645–6.Google Scholar
- 60.Sokolov AP, Kicklighter DW, Melillo JM et al. Consequences of considering carbon–nitrogen interactions on the feedbacks between climate and the terrestrial carbon cycle. J Clim. 2008;21(15):3776–96.Google Scholar
- 61.Thornton P, Lamarque J-F, Rosenbloom NA et al. Influence of carbon-nitrogen cycle coupling on land model response to CO2 fertilization and climate variability. Glob Biogeochem Cycles. 2007;21, GB4018.Google Scholar
- 62.Zaehle S. Terrestrial nitrogen—carbon cycle interactions at the global scale. Philos Trans R Soc B Biol Sci. 2013;368(1621):20130125.Google Scholar
- 63.•Wårlind D, Smith B, Hickler T et al. Nitrogen feedbacks increase future terrestrial ecosystem carbon uptake in an individual-based dynamic vegetation model. Biogeosciences. 2014;11:6131–46. The paper summarized the effects of including N limitations in global dynamic vegetation and biogeochemical models when simulating past and future responses to climate and CO 2 change.Google Scholar
- 64.Sala A, Woodruff DR, Meinzer FC. Carbon dynamics in trees: feast or famine? Tree Physiol. 2012;32(6):764–75.Google Scholar
- 65.Wiley E, Helliker B. A re-evaluation of carbon storage in trees lends greater support for carbon limitation to growth. New Phytol. 2012;195(2):285–9.Google Scholar
- 66.•Fatichi S, Leuzinger S, Körner C. Moving beyond photosynthesis: from carbon source to sink-driven vegetation modeling. New Phytol. 2013;201(4):1086–95. The authors outlined how C sink limitation might be included in vegetation models.Google Scholar
- 67.Carbone MS, Czimczik CI, Keenan T et al. Age, allocation and availability of nonstructural carbon in mature red maple trees. New Phytol. 2013;200:1145–55.Google Scholar
- 68.Dietze MC, Sala A, Carbone MS et al. Nonstructural carbon in woody plants. Annu Rev Plant Biol. 2014;65:667–87.Google Scholar
- 69.Körner C. Growth controls photosynthesis—mostly. Nova Acta Leopold. 2013;391:273–83.Google Scholar
- 70.Leuzinger S, Manusch C, Bugmann H et al. A sink-limited growth model improves biomass estimation along boreal and alpine tree lines. Glob Ecol Biogeogr. 2013;22(8):924–32.Google Scholar
- 71.••Bugmann H, Bigler C. Will the CO2 fertilization effect in forests be offset by reduced tree longevity? Oecologia. 2011;165(2):533–44. The authors showed that CO 2 effects on tree population dynamics can offset growth enhancements of individual trees at the stand scale.Google Scholar
- 72.Ollinger SV, Aber JD, Reich PB et al. Interactive effects of nitrogen deposition, tropospheric ozone, elevated CO2 and land use history on the carbon dynamics of northern hardwood forests. Glob Chang Biol. 2002;8(6):545–62.Google Scholar
- 73.Pan Y, Birdsey R, Hom J et al. Separating effects of changes in atmospheric composition, climate and land-use on carbon sequestration of U.S. Mid-Atlantic temperate forests. For Ecol Manag. 2009;259(2):151–64.Google Scholar
- 74.Ciais P, Schelhaas MJ, Zaehle S et al. Carbon accumulation in European forests. Nat Geosci. 2008;1(7):425–9.Google Scholar
- 75.Pan Y, Birdsey RA, Fang J et al. A large and persistent carbon sink in the world’s forests. Science. 2011;333(6045):988–93.Google Scholar
- 76.Pretzsch H, Biber P, Schuetze G et al. Forest stand growth dynamics in Central Europe have accelerated since 1870. Nat Commun. 2014;5:4967.Google Scholar
- 77.McMahon SM, Parker GG, Miller DR. Evidence for a recent increase in forest growth. Proc Natl Acad Sci. 2010;107(8):3611–5.Google Scholar
- 78.Settele J, Scholes R, Betts R et al. Terrestrial and inland water systems. In: Field CB et al., editors. Climate change 2014: impacts, adaptation, and vulnerability. Part A: global and sectoral aspects. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel of Climate Change. Cambridge: Cambridge University Press; 2014. p. 271–359.Google Scholar
- 79.Pan Y, Mellilo JM, McGuire AD et al. Modeled responses of terrestrial ecosystems to elevated atmospheric CO2: a comparison of simulations by biogeochemistry models of the Vegetation/Ecosystem Modeling and Analysis Project (VEMAP). Oecologia. 1998;114:389–404.Google Scholar
- 80.Kirschbaum MUF, Watt MS, Tait A et al. Future wood productivity of Pinus radiata in New Zealand under expected climatic changes. Glob Chang Biol. 2012;18(4):1342–56.Google Scholar
- 81.Scheiter S, Higgins SI. Impacts of climate change on the vegetation of Africa: an adaptive dynamic vegetation modelling approach. Glob Chang Biol. 2009;15(9):2224–46.Google Scholar
- 82.Higgins SI, Scheiter S. Atmospheric CO2 forces abrupt vegetation shifts locally, but not globally. Nature. 2012;488(7410):209–12.Google Scholar
- 83.Scheiter S, Higgins SI, Beringer J et al. Climate change and long-term fire management impacts on Australian savannas. New Phytol. 2015;205(3):1211–26.Google Scholar
- 84.Cox PM, Betts RE, Jones CD et al. Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model. Nature. 2000;408:185–7.Google Scholar
- 85.Cox PM, Betts RA, Collins M et al. Amazonian forest dieback under climate-carbon cycle projections for the 21st century. Theor Appl Climatol. 2004;78(1–3):137–56.Google Scholar
- 86.Huntingford C, Harris PP, Gedney N et al. Using a GCM analogue model to investigate the potential for Amazonian forest dieback. Theor Appl Climatol. 2004;78(1–3):177–85.Google Scholar
- 87.Hickler T, Smith B, Prentice IC et al. CO2 fertilization in temperate FACE experiments not representative of boreal and tropical forests. Glob Chang Biol. 2008;14:1531–42.Google Scholar
- 88.•Rammig A, Jupp TE, Thonicke K et al. Estimating the risk of Amazonian forest dieback. New Phytol. 2010;187:694–706. This paper showed that understanding physiological CO 2 effects is crucial for projecting future changes in biomass across the Amazon.Google Scholar
- 89.Schaphoff S, Heyder U, Ostberg S et al. Contribution of permafrost soils to the global carbon budget. Environ Res Lett. 2013;8(1):014026.Google Scholar
- 90.Taylor KE, Stouffer RJ, Meehl GA. An overview of CMIP5 and the experiment design. Bull Am Meteorol Soc. 2012;93(4):485–98.Google Scholar
- 91.Galbraith D, Levy PE, Sitch S et al. Multiple mechanisms of Amazonian forest biomass losses in three dynamic global vegetation models under climate change. New Phytol. 2010;187(3):647–65.Google Scholar
- 92.Norby RJ, Wullschleger SD, Gunderson CA et al. Tree responses to rising CO2 in field experiments: implications for the future forest. Plant Cell Environ. 1999;22:683–714.Google Scholar
- 93.Hendrey GR, Ellsworth DS, Lewin KF et al. A free-air enrichment system for exposing tall forest vegetation to elevated atmospheric CO2. Glob Chang Biol. 1999;5:293–309.Google Scholar
- 94.Norby RJ, Lucia EHD, Gielen B et al. Forest response to elevated CO2 is conserved across a broad range of productivity. Proc Natl Acad Sci U S A. 2005;102(50):18052–6.Google Scholar
- 95.Randerson JT, Hoffman FM, Thornton PE et al. Systematic assessment of terrestrial biogeochemistry in coupled climate-carbon models. Glob Chang Biol. 2009;15(10):2462–84.Google Scholar
- 96.Nowak RS, Ellsworth DS, Smith SD. Functional responses of plants to elevated atmospheric CO2: do photosynthetic and productivity data from FACE experiments support early predictions? New Phytol. 2004;162(2):253–80.Google Scholar
- 97.Norby RJ, Hanson PJ, O'Neil EG et al. Net primary productivity of a CO2 enriched deciduous forest and the implications for carbon storage. Ecol Appl. 2002;12(5):1261–6.Google Scholar
- 98.McCarthy HR, Oren R, Johnsen KH et al. Re-assessment of plant carbon dynamics at the Duke free-air CO2 enrichment site: interactions of atmospheric [CO2] with nitrogen and water availability over stand development. New Phytol. 2010;185(2):514–28.Google Scholar
- 99.•Norby RJ, Warren JM, Iversen CM et al. CO2 enhancement of forest productivity constrained by limited nitrogen availability. Proc Natl Acad Sci. 2010;107(45):19368–73. This paper demonstrated the occurrence of progressive nitrogen limitation at the Oak Ridge FACE site.Google Scholar
- 100.Walker AP, Hanson PJ, De Kauwe MG et al. Comprehensive ecosystem model-data synthesis using multiple data sets at two temperate forest free-air CO2 enrichment experiments: model performance at ambient CO2 concentration. J Geophys Res Biogeosci. 2014;119(5):2013JG002553.Google Scholar
- 101.••Medlyn BE, Zaehle S, De Kauwe MG, et al. Using ecosystem experiments to improve vegetation models. Nat Clim Chang. 2015; in press. This paper summarized the approach and results of the FACE model-data synthesis, comparing 11 models with experimental results from two FACE sites (Duke and Oak Ridge). Google Scholar
- 102.Walker AP, Zaehle S, Medlyn BE et al. Predicting long-term carbon sequestration in response to CO2 enrichment: how and why do current ecosystem models differ? Glob Biogeochem Cycles. 2015. accepted for publication.Google Scholar
- 103.Johnson DW, Cheng W, Joslin JD et al. Effects of elevated CO2 on nutrient cycling in a sweetgum plantation. Biogeochemistry. 2004;69(3):379–403.Google Scholar
- 104.Dawes MA, Hättenschwiler S, Bebi P et al. Species-specific tree growth responses to 9 years of CO2 enrichment at the alpine treeline. J Ecol. 2011;99(2):383–94.Google Scholar
- 105.Tollefson J. Experiment aims to steep rainforest in carbon dioxide. Nature. 2013;496:405–6.Google Scholar