From plant functional types to plant functional traits

Abstract

Dynamic global vegetation models (DGVMs) typically track the material and energy cycles in ecosystems with finite plant functional types (PFTs). Increasingly, the community ecology and modelling studies recognize that current PFT scheme is not sufficient for simulating ecological processes. Recent advances in the study of plant functional traits (FTs) in community ecology provide a novel and feasible approach for the improvement of PFT-based DGVMs. This paper reviews the development of current DGVMs over recent decades. After characterizing the advantages and disadvantages of the PFT-based scheme, it summarizes trait-based theories and discusses the possibility of incorporating FTs into DGVMs. More importantly, this paper summarizes three strategies for constructing next-generation DGVMs with FTs. Finally, the method’s limitations, current challenges and future research directions for FT theory are discussed for FT theory. We strongly recommend the inclusion of several FTs, namely specific leaf area (SLA), leaf nitrogen content (LNC), carbon isotope composition of leaves (Leaf δ¹³C), the ratio between leaf-internal and ambient mole fractions of CO₂ (Leaf C_i/C_a), seed mass and plant height. These are identified as the most important in constructing DGVMs based on FTs, which are also recognized as important ecological strategies for plants. The integration of FTs into dynamic vegetation models is a critical step towards improving the results of DGVM simulations; communication and cooperation among ecologists and modellers is equally important for the development of the next generation of DGVMs.

Keywords

DGVMs plant functional types (PFTs)plant functional traits (FTs)environmental filters ecosystem function climate change PFTs-traits hybrid model

I Introduction

Dynamic global vegetation models (DGVMs) are an essential component of earth system models (ESMs), which aim to understand the interactive processes between vegetation and the atmosphere (Prentice and Cowling, 2013; Prentice et al., 1992; Quillet et al., 2010; Van Bodegom et al., 2012). DGVMs are generally designed to track the structural and functional responses of vegetation to changes in climate and atmospheric CO₂ concentration and include coexistence of several plant functional types (PFTs) (Quillet et al., 2010; Reich et al., 2007). The PFT scheme not only performs well in simulating global vegetation dynamics, net primary productivity (NPP) and biomass under transient climate change scenarios (Peng, 2000; Quillet et al., 2010) but has also successfully reconstructed palaeo-vegetation patterns and palaeo-carbon storage (Peng, 2000; Prentice and Webb, 1998).

It has been assumed that these PFTs are fully capable of representing the dynamic processes of DGVMs in response to global change; however, there is greater variation within PFTs than among PFTs (Wright et al., 2005a), and the PFT scheme has also been criticized as inaccurate in simulating carbon cycle processes (Pavlick et al., 2013; Van Bodegom et al., 2012). Additionally, the plant attributes used to define PFTs are usually treated as fixed within each PFT, whereas they actually exhibit great variation (Kattge et al., 2011; Pavlick et al., 2013; Wright et al., 2004; Wright et al., 2005b). Growing evidence from current DGVM studies indicates that a plant functional traits (FTs) scheme is well suited for addressing these limitations (Ali et al., 2013; Pavlick et al., 2013; Quetier et al., 2007; Scheiter et al., 2013; Van Bodegom et al., 2012).

Traits-based theory originated from community ecology (McGill et al., 2006) and has become an effective method in ecological research. Traits-based theory focuses primarily on how environmental factors filter FTs (Díaz et al., 1998; Wright et al., 2005b) and on the relationships between traits and ecosystem functions (Díaz et al., 1999; Quetier et al., 2007; Savage et al., 2007). The FT scheme not only provides a widely applicable approach for forecasting ecosystem shifts and changes in ecosystem structure, but can also be linked with ecosystem functions under climate change (Quetier et al., 2007). Several scientists have addressed the inclusion of trait theory in DGVM development to improve the results of current vegetation models. Scheiter et al. (2013) have proposed that DGVM modelling could be improved by incorporating coexistence ecology and community assembly theory, and increasing evidence has shown that including FTs-based theory is necessary for constructing the next generation of DGVMs (Higgins et al., 2014; Scheiter et al., 2013; Van Bodegom et al., 2012; Verheijen et al., 2013).

The objectives of this study are to discuss the possibility of using a FT scheme instead of a PFT scheme for modelling DGVMs and to present viewpoints about constructing the next generation of DGVMs. We articulate our viewpoints by first describing recent advances in the development of DGVMs. Then, we present FTs theory and the possibility of assimilating FTs into current DGVMs. Next, we summarize three feasible approaches to developing the next generation of DGVMs based on FTs. Finally, we discuss the method’s limitations, challenges and future research directions.

II PFT scheme used in current DGVMs

1 Development of the PFT scheme

PFTs represent most of the world’s vegetation types and characteristics through their functional behaviours and attributes (Box, 1996). A PFT is a group of plants with similar functions based on their morphological, physiological, biochemical, reproductive and demographic characteristics (Box, 1981; Foley et al., 1996; Canadell et al., 2007; Prentice et al., 1992; Woodward and Cramer, 1996). Each PFT exhibits a similar response to climate change and performs similar ecosystem functions, which can effectively minimize the complexity of DGVMs (Díaz and Cabido, 1997; Woodward and Cramer, 1996). PFTs are commonly confined by empirical upper and lower limits for climate variables and play an important role in governing the structure and dynamics of models (Harrison et al., 2010).

The PFT scheme works well for predicting vegetation patterns and ecosystem functions. Taking the integrated biosphere simulator (IBIS; Figure 1) (Foley et al., 1996; Kucharik et al., 2000) as an example, we will show how the PFT scheme works in a typical DGVM. In IBIS, 12 PFTs are defined by several important ecological characteristics: basic physiognomy (i.e. trees and grass), leaf habit (evergreen and deciduous), photosynthetic pathway (C₃ and C₄) and leaf form (broadleaf and needle-leaf). A minimal set of climate constraints including winter cold tolerance limits, growing degree-day requirements and minimum chilling requirements are selected to determine the existence of PFTs in each grid, and consequently the vegetation type as a result of different combinations of coexisting PFTs. In IBIS, PFTs also determine the structure and ecosystem functions. Two vegetation layers are represented: woody plant functional types (trees) in the upper canopy and herbaceous PFTs in the lower canopy. PFTs in the upper vegetation layer are able to capture light first and shade the lower vegetation canopy; however, the lower-layer vegetation is able to take up soil moisture first. Different plant physiology parameters (i.e. photosynthesis and respiration parameters) and vegetation dynamics parameters (such as allocation of total photosynthetic production to leaf, root and stem, and the residence time of carbon in leaf, root and stem) are adopted for different PFTs to calculate the living biomass and leaf area index with a set of biochemical process equations. Different vegetation phenologies are also present in each PFT. Competition between PFTs is driven by differences in the annual carbon balance, which result from different ecological strategies.

Figure 1.

Schematic of the dynamic global vegetation model of Integrated BIosphere Simulator (IBIS) (Foley et al., 1996; Kucharik et al., 2000). T_min = absolute minimum temperature; T_w = temperature of the warmest month; GDD₅ = growing degree days calculated on 5°C.

The PFT scheme in DGVMs is a feasible and necessary approach for modelling terrestrial ecosystem structures and processes. A PFT scheme bridges the gap between plant physiology and ecosystem processes, making it possible to project the impacts of climate change on vegetation dynamics and carbon cycles (Díaz and Cabido, 1997). Nearly all current DGVMs rely on a PFT scheme to reduce effectively the complexity of DGVMs through functional classification. To summarize, the PFT scheme is the foundation of current DGVMs and makes it possible to model ecosystem processes based on multiple PFTs coexisting in each monitoring unit.

2 Limitations and challenges of the PFT scheme

Despite their effectiveness at simulating fluxes to and from terrestrial ecosystems, current DGVMs are insufficiently realistic and show limited room for improvement because they utilize PFTs with constant attributes (Van Bodegom et al., 2012). The PFT scheme provides a good way to use finite attributes to aggregate individual plants with similar behaviours, but the use of constant attributes to define PFTs can only explain between one-third and two-thirds of the variation in five important and quantitative leaf ecophysiological traits (Reich et al., 2007). At the same time, certain attributes defined for PFTs do not, in reality, differ much among PFTs (Cunningham and Read, 2002; Van Bodegom et al., 2012). PFTs in vegetation models describe the trait variation among coexisting species using a mean value for each trait, but there is a clear mismatch between allocating vegetation to discrete PFTs and calculating continuously varying fluxes through vegetation (Reich et al., 2007). Meanwhile, future climate changes are expected to result in different niches, so the future climate may have no analogue in present climate conditions, leading to the absence of corresponding PFTs for future climate scenarios (Van Bodegom et al., 2012; Williams et al., 2007).

DGVMs simulate ecosystem functions and structures using a limited number of PFTs (commonly fewer than 15) rather than using species, which can be useful for simulating biogeochemical processes and structures but might affect the precision of modelled vegetation distribution and biodiversity (McMahon et al., 2011). DGVMs represent the competition between PFTs, but competition occurring at a local scale and between individuals cannot be well simulated (Quillet et al., 2010). One possible way to overcome this limitation is to increase the number of PFTs. But this is a challenging task because the functional differences between PFTs become smaller and overlapping may occur when predicting the distribution of these PFTs. Although the PFT scheme is the basis of energy and material circulation, the limitations created by the PFT scheme are difficult to overcome.

A PFT classification based on bioclimatic response will need to be enhanced using information on FTs related to competition, successional dynamics and disturbance (Harrison et al., 2010). Current DGVMs rely on earlier classification such as that of Box (1981), which is a simple scheme and includes explicit bioclimatic limits, but PFTs are not fully characterized in terms of traits (Harrison et al., 2010). Therefore, vegetation modelling based on the PFT scheme as a combination of fixed attributes has limited explanatory power for forecasting the impacts of changing climate and disturbance.

DGVMs play a pivotal role in simulating atmosphere-land interactions, by quantifying the processes of global carbon, nitrogen and water cycles (Van Bodegom et al., 2012; Wullschleger et al., 2014). There is mounting evidence that uncertainty may arise from inadequate PFT parameters and incomplete PFT classification (Wullschleger et al., 2014). Recent study with remote sensing products and field data confirms that key parameters in PFTs exhibit overlapping ranges and have limits to their precision in representing carbon cycle processes (Alton, 2011). DGVMs poorly represent competition because of their assumption that competition occurs among PFTs rather than among individuals when applied to forest biodiversity research (Clark et al., 2011; Sato et al., 2007). Additionally, current PFT classifications often neglect the importance of root traits, which have shown great importance in the interaction of climate change with Arctic and boreal ecosystems (Hartley et al., 2012). Nitrogen and carbon cycles are tightly coupled in most DGVMs, and the C/N ratio is carried as a state variable in each biomass compartment. In researching of plant–soil interactions, Ostle et al. (2009) find that existing PFT categorizations are based on C function, growth and competition that may not necessarily reflect their N cycling characteristics, and C/N ratios of different plant compartments may propagate uncertainty owing to PFT definitions. Anderegg (2014) performs a meta-analysis and finds that PFTs are poorly suited to capturing key differences in hydraulic traits across species, indicating that a FTs-based approach works better than one based on PFTs. All of this means that it is imperative to identify a new scheme to replace the PFT scheme. One feasible approach is to incorporate a small number of dimensions of FTs into DGVMs, instead of the PFT scheme. This approach would enhance the predictive ability of DGVMs without increasing their complexity.

III From plant functional types (PFTs) to plant functional traits (FTs)

1 Plant functional traits

Trait-based theory is now widely applied at various biological levels from individuals to whole ecosystems (Ackerly and Cornwell, 2007; Ceulemans et al., 2011; Navas, 2012). Not every trait is a functional trait. Plant FTs are defined as morphological, physiological and phenological traits that indirectly impact individual fitness via their effects on the growth, reproduction and survival of the plant (Violle et al., 2007). Selected important FTs and their ecosystem functions are described in Table 1. FTs have been used recently to address a series of ecological problems including: (1) parameterizing DGVMs and constraining the parameters in a reasonable range (Díaz et al., 2001; McGill et al., 2006; Pavlick et al., 2013; Scheiter et al., 2013; Swenson and Weiser, 2010); (2) providing aids in the classification of PFTs (Chaturvedi et al., 2011; Lavorel et al., 1997; Witte et al., 2007); (3) predicting community assembly and community biodiversity (Andersen et al., 2012; Barnett et al., 2007; Gallagher et al., 2011; Reu et al., 2011; Suding et al., 2008); and (4) linking climate factors to ecosystem functions (Luck et al., 2012; Navas, 2012; Wallenstein and Hall, 2012).

Table 1.

Association of core plant functional traits with their ecosystem functions and responses to environmental change. Note that ‘+’ denotes that changes in climate or CO₂ greatly affect the trait or that the trait changes easily in response to disturbance. ‘?’ denotes that this relationship is not very clear from previous research.

Traits	Description	Ecosystem function	Climate response	CO₂ response	Response to disturbance
Whole-plant traits
Growth form	Tree, shrub, liana/vine, herb, forb, etc.	Deciding the basic structure, biodiversity and behaviour of communities. Vegetation succession will occur under different climate scenarios represented by different growth forms	+	+	+
Life form	Phanerophytes, chamaephytes, hemicryptophyte, cryptophytes, therophytes, aerophytes, epiphytes	Reflecting plant adaptation to survive unfavourable conditions (C Raunkiaer proposed in 1904). Phanerophytes usually dominate in warm and humid environments, whereas chamaephytes, hemicryptophytes, and cryptophytes are better adapted to dry and cold environments	+	+	+
Plant height	Quantitative data	Species with different heights use individual resources, interact with neighbours and disperse in time and space (Garnier and Navas, 2012). Plant height is also related to critical ecosystem variables such as carbon storage capacity (Moles et al., 2009)	+	+	+
Leaf traits
Specific leaf area(SLA) / leaf mass per area (LMA)	Quantitative data. SLA is defined as the ratio of leaf area to dry mass	SLA reflects the capacity for resource capture. High-SLA leaves are usually more productive, but they are also short-lived and susceptible to herbivores (Coley et al., 1985; Wilson et al., 1999). SLA has exhibited a positive effect on decomposition rates in previous research (Allison and Vitousek, 2004; Bakker et al., 2011). LMA is simply the inverse of SLA	+	+	+
Leaf dry matter content	Quantitative data	Leaf dry matter content is a better predictor of resource capture (Wilson et al., 1999)	+	?	+
Leaf N concentration	Quantitative data	Leaf nitrogen concentration is integral to the production of the proteins involved in the photosynthetic machinery, especially Rubisco (Wright et al., 2004)	+	+	+
Leaf P concentration	Quantitative data	Leaf phosphorus is a constituent element of nucleic acids, lipid membranes and bioenergetic molecules such as ATP (Wright et al., 2004)	+	+	+
Photosynthetic pathway	C₃, C₄, Crassulacean Acid Metabolism (CAM)	C₄ and CAM photosynthesis are evolutionarily evolved from C₃ photosynthesis. C₃ plants have significant increases in photorespiration at low CO₂ concentrations. CAM increased water-use efficiency in dry environments (Ehleringer and Monson, 1993)	+	+	+
Leaf δ¹³C	Carbon isotope(δ¹³C) composition of leaves	The leaf carbon isotope ratio (δ¹³C) of C₃ plants is inversely related to the drawdown of CO₂ during photosynthesis (Farquhar et al., 1982; Prentice et al., 2011). Leaf δ¹³C has a close relationship with water use efficiency (WUE) (Verlinden et al., 2014)	+	+	+
Leaf C_i/C_a	The ratio between leaf-internal(C_i) and ambient(C_a) mole fractions of CO₂	C_i/C_a regulates the balance between carbon gain and water loss, which is lower at dry or cold conditions than wet or hot sites (Prentice et al., 2014; Wright et al., 2003)	+	+	+
V_cmax25	Maximum carboxylation rate at 25ºC	V_cmax25 is the key parameter to calculate the photosynthesis for PFTs following Farquhar et al. (1980) and also determines the reference dark respiration at 25ºC (Verheijen et al., 2013)	+	+	+
Leaf lifespan	Quantitative data. Leaf life span describes the average duration of the revenue stream from each leaf constructed (Wright et al., 2004)	Leaf life span is highly correlated with leaf structural variables and stomata conductance to water vapour, and both mass and area-based net photosynthesis decrease with increasing leaf life span (Reich et al., 1991)	+	+	+
Leaf venation traits	Quantitative data. Leaf venation traits include the dimensions of vascular cells and of whole veins, vein hierarchy, vein length per unit area (VLA) and major vein density	Greater phloem conduit numbers or size can contribute to faster phloem transport rates. High VLA enables greater stomatal density, stomatal conductance and higher rates of gas exchange per unit leaf area (Sack and Scoffoni, 2013). These traits play an important role in mass transport, mechanical support and herbivore defence	+	+	+
Stem traits
Wood density	Wood density is the oven-dry mass divided by green volume (Chave et al., 2009)	Wood density is correlated with mechanical support, water transport and the storage capacity of woody tissues. Leaf size and stem growth rate decrease with increasing wood density (Chave et al., 2009)	+	+	+
Root traits
Root longevity	Quantitative data	Root turnover is important to the global carbon budget and to the success of individual plants. Root longevity is positively correlated with mycorrhizal colonization and decreases with increasing nitrogen concentration, root maintenance respiration and specific root length (Eissenstat et al., 2000)	+	+	?
Specific root length (SRL)	Ratio between root length and root biomass	SRL can successfully be used as an indicator of environmental change when nutrient availability is manipulated. SRL responds negatively to reduced light and to elevated temperature and CO₂ (Ostonen et al., 2007)	+	+	+
Regenerative traits
Seed mass	Quantitative data	Seed mass links the ecology of reproduction and seedling establishment (Leishman et al., 2000)	+	?	+
Seed size	Quantitative data	Small seeds disperse easily and are more widely distributed, whereas larger seeds are relatively less abundant but produce seedlings with greater competitive ability (Khurana et al., 2006)	+	?	+

Among all FTs that can be measured on an individual plant, those that are closely related to ecosystem functions and exhibit a consistent relationship with climate or environmental gradients are suitable for global syntheses and modelling. Obviously, certain FTs, such as leaf type (i.e. broadleaves, needle-leaves or scaled-leaves), specific leaf area(SLA), seed mass, leaf nitrogen content(LNC) and plant height, can be directly observed or easily measured (Drenovsky et al., 2012; Weiher et al., 1999), whereas measuring other FTs might be difficult or time-consuming, e.g. dispersal distance, relative growth rate, competitive effect and response. However, those FTs usually directly influence plant physiological processes and ecosystem functions. Weiher et al. (1999) and Cornelissen et al. (2003) referred to the former group as ‘soft’ traits and to the latter as ‘hard’ traits. Because of the time investment in measurements and the ecological importance of ‘hard’ traits, we need to develop more easily measured or estimated analogues for them.

Several global FT databases have been initiated by organisations and research networks. For example, GLOPNET (http://bio.mq.edu.au/∼iwright/glopian.htm) is a multi-investigator group studying global plant traits; their dataset quantifies a worldwide ‘leaf economic spectrum’ consisting of key chemical, structural and physiological leaf attributes for 2548 species spanning 175 sites(Reich et al., 2007; Wright et al., 2004). GLOPNET is available to scientists and open for use. TRY, created in 2007 (http://www.try-db.org/TryWeb/Home.php), is a network of vegetation scientists headed by DIVERSITAS, IGBP and the Max Planck Institute for Biogeochemistry. The TRY database integrates 93 plant trait databases at a worldwide scale and represents more than 1000 traits for 70,000 species (Kattge et al., 2011). The TRY database contains both public and non-public data. The public data are freely available for downloading. However, the majority of the data are non-public and cannot be used without the approval of the appropriate administrators. The usefulness of TRY is limited to a certain extent by the cumbersome procedure required to access most of its contents. We anticipate that this obstacle will be removed in the future.

2 Relationships between FTs and climate factors

Relationships between FTs and climate have been emphasized for at least a century (Wright et al., 2004). For example, leaf nitrogen and phosphorus decrease and their ratio (N:P) increases with increasing environmental temperature (Reich and Oleksyn, 2004). Additionally, a global quantification of leaf traits reveals strong correlations between soil nutrient fertility, SLA and leaf nitrogen concentration (LNC) (Ordoñez et al., 2009). Using the GLOPNET dataset, multiple regression equations describing traits–climate relationships are presented in Figure 2. Usually, ecologists have endeavoured to quantify relationships between plant traits and climate factors at a global level. However, it is difficult to construct equations for relationships at a global scale on account of insufficient data, different criteria and complicated interactions between plants and the environment. The TRY and GLOPNET datasets each combine data from many case studies of FTs and climate, which makes it possible to build global regression equations.

Figure 2.

Relationships between leaf functional traits and climate factors and among leaf functional traits. The data were averaged by sites and derived from GLOPNET (Wright et al., 2004). LMA = leaf mass per area; N_mass = mass-based leaf nitrogen; N_area = area-based leaf nitrogen; MAT = mean annual temperature; MAP = mean annual precipitation.

Predicting the responses of community composition and ecosystem function to a rapidly changing climate is a major research challenge in ecology. Webb et al. (2010) have proposed a conceptual foundation for the environmental filtering of trait distributions. Three elements were summarized: trait distributions, performance filters and environmental gradients. The distribution of traits is decided by the pool of possible traits of individual plants, the performance filters are an expression of fitness in a given environment, and the environmental gradients can be used to filter the distribution of traits at different points in space and time (Webb et al., 2010). In fact, the filtering of FTs by the environment is a result of natural selection or ecological sorting, which makes traits-relative theories close to the theory of evolution by natural selection.

3 FTs and ecosystem functions in relation to global change

Representing plant species as a set of FTs instead of a fixed PFT provides possibilities for analysing the ecosystem functions (Lavorel and Garnier, 2002). FTs show close association with growth, establishment, dispersal, competition and response to disturbance (Poorter and Bongers, 2006; Weiher et al., 1999). Environmental change or disturbance will alter the pool of FTs and correspondingly influence ecosystem functions.

Certain FTs directly or indirectly influence the growth of the plant. LNC is integral to the production of the proteins involved in the photosynthetic machinery, especially Rubisco, which is responsible for the drawdown of CO₂ within leaves (Wright et al., 2004). SLA is the area of one side of a single leaf divided by its dry mass and affects the efficiency of leaves at capturing light and carbon dioxide (Vendramini et al., 2002). The combination of SLA and LNC can be related to leaf span and to predict the maximum photosynthetic rate, which has a significant impact on the primary productivity and nutrient cycling of an ecosystem (Reich et al., 1997). Leaf dry matter content (LDMC) affects the investment of plant production in persistent leaf structure and nutrient retention and is suggested to correlate negatively with potential relative growth rate (Cornelissen et al., 2003). Overall, these few FTs characterize a spectrum of strategy ranges from fast-growing species with high biomass turnover to slow-growing species with permanent leaf structures (Suter and Edwards, 2013; Wright et al., 2004).

After seeds arrive at a new location, they must become established. Previous research has identified a set of FTs that have close relationships with the dispersal, establishment and survival of plants. Seed size and mass are key FTs in determining species distributions and play an important role in reproduction. Small and light seeds have greater dispersal ability and are more widely distributed compared with heavier and larger seeds, but heavier and larger seeds have more establishment opportunities, increasing their chance of survival in stressful environments (Khurana et al., 2006). Apart from seed size and mass, seed shape also affects the dispersal of a plant species. When seeds germinate, a rapid seedling relative growth rate (RGR) generally increases a seedling’s success, but a slower RGR is associated with better seedling survival under adverse environmental conditions (Leishman, 1999).

Plant species frequently compete with each other for various limited resources, such as nutrients and light, and there are also many FTs that reflect this mechanism. It has been suggested that, under nutrient-rich conditions, a rapid growth rate is crucial for high competitive ability (Grime, 1977); therefore, SLA, which measures the ratio of light-capturing area to dry biomass, directly reflects competitive ability (Weiher et al., 1999). Plant height is an important part of a plant’s ecological strategy and is strongly correlated with life span, seed mass and time to maturity, which is a major determinant factor of light capture (Moles et al., 2009). To a certain extent, seed mass and size can also enhance the competitive ability of a seedling (Turnbull et al., 1999). Rooting depth is reported as an additional key FT in terms of competition for water in water-limited environments (Violle et al., 2009).

A plant’s resistance or resilience to disturbance depends on local environmental conditions, further modifying the pool of species and FTs (Bernhardt-Römermann et al., 2011). Fire, hurricanes, logging, drought or frost events, grazing, erosion and herbivores are examples of major disturbances. After the destruction of above-ground biomass, resprouting capacity directly influences the re-establishment of a plant (Perez-Harguindeguy et al., 2013). RGR is also a prominent indicator of plant strategy with respect to productivity after disturbances (Perez-Harguindeguy et al., 2013). Díaz et al. (2001) proposed the combination of plant height, life history and leaf mass as the best prediction of species’ grazing response. In a case study, SLA was reported as the best predictor of response to land-use change (Meers et al., 2008).

FTs constitute a highly useful concept for forecasting changes in plant communities, and their associated ecosystem services, in response to climate change (Soudzilovskaia et al., 2013). FTs are closely related to climate regulation, carbon storage, water regulation, soil stability and tolerance of disturbance (de Bello et al., 2010; Lavorel and Grigulis, 2012). FTs provide a new perspective with which to evaluate ecosystem services. Environmental factors influence ecosystem services primarily through underlying ecosystem processes, and there has been little quantitative study of FT/ecosystem service relationships in field experiments, so it will be effective and meaningful to qualify these relationships with the help of DGVMs.

4 Key FTs for constructing the next generation of DGVMs

Selection of the core FTs that strongly influence ecosystem structure and function is vital for modelling ecosystem processes. A synthesis of empirical and theoretical studies has proposed that at least four dimensions of FTs should be considered (Westoby et al., 2002). The first dimension is leaf mass per area and leaf lifespan (LMA-LL), which reflects the turnover of plant leaves, nutrient residence times and response to growth conditions. The second dimension is seed mass and seed output(SM-SO), which is correlated with dispersal and establishment opportunities. The third is leaf size and twig size (LS-TS), which has a close relationship with the texture of canopies. The fourth dimension is potential plant height, which is a proxy for competitive ability. These four dimensions vary across ecological zones and reflect the adaptation of species to climate changes.

For integrating FTs into the next generation of DGVMs, we highly recommend several FTs including SLA, LNC, carbon isotope composition of leaves (Leaf δ¹³C), the ratio between leaf-internal and ambient mole fractions of CO₂ (Leaf C_i/C_a), seed mass and plant height. SLA and LNC are closely related to plant growth. SLA reflects a plant’s ability to capture light and CO₂. High-SLA leaves are usually associated with high productivity, whereas lower-SLA leaves are more efficient in resource-limited environments (Wilson et al., 1999). Because of its clearer and more important ecological interpretation and wide application to different floras, SLA appears to be the best candidate in large databases (Vendramini et al., 2002; Wilson et al., 1999). LNC is closely related to photosynthesis (Wright et al., 2004). Seed mass is related primarily to seedling establishment and competitive ability and reflects the adaptability of a plant to disturbance (Turnbull et al., 1999). The leaf carbon isotope ratio (δ¹³C) of C₃ plant is inversely related to the drawdown of CO₂ during photosynthesis (Prentice et al., 2011) and Leaf δ¹³C has a close relationship with water use efficiency (WUE) (Verlinden et al., 2014). The ratio between leaf-internal and ambient mole fractions of CO₂ (C_i/C_a) regulates the balance between carbon gain and water loss, which is lower at dry or cold conditions than at wet or hot sites (Prentice et al., 2014). Plant height is closely related to competition for light, animal diversity and carbon storage capacity (Moles et al., 2009). These FTs are related to the important role of photosynthesis and reflect the most important functions driving plant establishment, growth, dispersal and competition, which constitute the basic and indispensable structure and function parameters in DGVMs. Thus, these FTs were selected as suitable and feasible first choices to integrate into a new generation of DGVMs.

5 FTs and vegetation classification

Deriving vegetation maps from climate-vegetation or trait-vegetation relationships is one of the most important parts of ecological modelling. Vegetation classification in DGVMs would greatly enhance our ability to predict both pulsed and gradual environmental changes (Barnett et al., 2007). The classic BIOME model predicted the distribution of PFTs based on threshold values of five bioclimatic variables (Prentice et al., 1992). Witte et al. (2007) extracted three indicator values for moisture, nutrients and acidity and fitted these indicator values into a Gaussian Mixture Model to yield the occurrence probabilities of specific vegetation types. A Gaussian mixture density procedure that predicted PFTs from trait combinations was also employed in a conceptual and traits-based vegetation model (Van Bodegom et al., 2012). This method can be employed in modelling vegetation classification when the number of vegetation types is uncertain. The pattern of predicted FTs and biome types shown in Figure 3 was generated based on simple linear regression equations relating FTs and climate factors.

Figure 3.

The boundaries of global biome type (a) in relation to climate factors in Whittaker (1970) and (b) in relation to predicted plant functional traits based on the regression equation between FTs and climate factors (Wright et al., 2004; Reich and Oleksyn, 2004). LMA = leaf mass per area; N_mass = mass-based leaf nitrogen; MAT = mean annual temperature; MAP = mean annual precipitation.

Remote sensing images are, and have been, widely used in vegetation mapping, utilizing spectral bands from the visible to microwave portions of the spectrum (Andrew et al., 2014; Berry and Roderick, 2002; de Bello et al., 2010; Poulter et al., 2011). Poulter et al. (2011) developed four PFT datasets based on land-cover information extracted from EOD-MODIS, SPOT4 VEGETATION and ENVISAT-MERIS. Satellite-derived data could be used to improve the quality of model evaluations and to reduce uncertainties (Murray et al., 2012). More recently, Andrew et al. (2014) have discussed and proposed to use remote sensing images, especially utilizing the current and rapidly developing generation of hyperspectral and LiDAR instruments, to map quantitative plant traits, including (1) chemical traits: pigment composition, water content and nitrogen content; (2) structural traits: height, biomass, LAI, life form, crown morphology, canopy cover and canopy roughness. Many products of remote sensing, such as MODIS LAI data and tree height data extracted from RADAR images, not only make global validation of model outputs possible, but also provide estimates of several plant traits at a global scale. Above all, satellite-derived data make it possible to reveal the global pattern of FTs.

IV New strategies in modelling DGVMs with FTs

1 Varying traits within PFTs

Current vegetation models simulate vegetation structure and functions with fixed PFTs under restrictions of climate and other environmental factors. To date, the quantification of plant species response to climate change has been described at the level of fixed PFTs, an approach limited by its inflexibility as, in fact, there is much interspecific functional variation (Soudzilovskaia et al., 2013). PFTs monitored in every grid are permitted to coexist, and different PFTs have fixed attributes to describe. PFT-based models have reached their limits, and fixed attributes need to be more flexible to incorporate into models (Gerber et al., 2010; Van Bodegom et al., 2012). Verheijen et al. (2013) allowed three traits to vary within PFTs – specific leaf area, maximum carboxylation rate at 25°C and maximum electron transport rate at 25°C. This method permitted more variation in vegetation responses in DGVMs and resolved the issue that traditional PFT-based DGVMs may not include PFTs that will be present in future climate scenarios. However, it has not been validated with observational data and still cannot resolve the other limitations posed by the traditional PFT scheme.

2 Traits-based methods for constructing next-generation DGVMs

Along with the development of FT-related theories, several plant scientists have proposed a series of conceptual frameworks of the next generation of DGVMs based on FTs. Advancing the understanding of the functional basis of plant traits will improve our understanding of DGVMs (Canadell et al., 2007). From simplification of the filter theory of Keddy (1992) and Woodward and Diament (1991) to the summarized theory of Lavorel and Garnier (2002), FT/environment and FT/ecosystem function relationships have laid the foundation for the development of traits-based DGVMs. The CATS model relies on FTs to predict species relative abundances and transforms ‘trait-based’ habitat filtering into a quantitative framework (Frenette-Dussault et al., 2013; Shipley, 2010); Van Bodegom et al. (2012) have provided a conceptual DGVM framework based on traits by quantifying community assembly concepts. Continuous and process-based trait-environment relationships were central to the content of their conceptual model, which was a typical conceptual model for the next generation of DGVMs.

In the next generation of DGVMs, the PFT scheme will most likely be replaced by a cluster of FTs. Several scientists have already attempted to apply FT methods in DGVMs. For example, a dynamic plant carbon–nitrogen model has been used to simulate the long-term response of plants to increasing atmospheric CO₂. In this model, a plant species was represented by a vector of trait values, and physiological processes were also presented as traits (Ali et al., 2013). Scheiter et al. (2013) proposed a conceptual DGVM based in community ecology and coexistence theory. Their proposed DGVM adopted an individual-based approach to simulate many individual plants with a set of trait values. These traits of individuals described the rates of resource assimilation, growth, carbon allocation and respiration. Mutation and crossover were included in a community seed bank to generate new trait combinations and update the community trait pool (Scheiter et al., 2013). This method sufficiently considered the role of individuals and integrated FTs into the processes of establishment, growth, dispersal and reproduction. JeDi-DGVM (Pavlick et al., 2013) generates a large number of random plant growth strategies, each of which is defined by 15 FTs that characterize the plant functions of carbon allocation, phenology and ecophysiology. The survival of plant strategies (where a strategy is represented by a group of FTs) in each grid is filtered by the environment in this model. Although the results of JeDi-DGVM have been criticized because its traits are ‘not measurable’(Higgins et al., 2014), these approaches provide different perspectives on constructing FT-based DGVMs and have also furnished the foundation for our future study.

3 A transitional framework for a PFT-traits hybrid DGVM

To explore an effective and quick way to create a traits-based DGVM, we propose a PFT-traits hybrid and transitional DGVM framework (Figure 4). In this conceptual model, PFTs are also engaged in the ecological structure and function, but rather than being directly linked to environmental factors. PFTs are linked to environmental factors through FT/climate relationships and the Gaussian Mixture Model (GMM) classifier. GMM has been applied successfully in predicting global vegetation distribution through the FT-based approach (Douma et al., 2012; Van Bodegom et al., 2014). Five steps are involved in describing this conceptual model:

Mathematical relationships are constructed between selected traits and environmental variables (in addition to climate variables, these should include nutrient and water availability, disturbance and acidity).

A group of traits and their corresponding plant functional types are used to train a GMM; meanwhile, the traits space is divided into different sub-spaces that belong to different PFTs.

The predicted traits are classified into different PFTs according to the location of traits vectors in N-dimensional space.

In every monitoring cell, several PFTs may exist, so the monitoring cell is assigned a vegetation type according to the probability of presence of each PFT, which can be obtained from the GMM classifier.

The predicted distribution of vegetation is validated with the natural vegetation map.

Figure 4.

A conceptual framework for improving DGVMs modified from Douma et al. (2012), Van Bodegom et al. (2012) and Webb et al. (2010). GMM = Gaussian Mixture Model; PFTs = plant functional types.

The predicted traits can be then used to predict the key status values of ecosystem processes or to represent the species composition and abundance. This type of conceptual DGVM not only reduces the complexity of traits-based DGVMs but also builds a bridge between the PFTs and FTs schemes, as needed to transition from the PFT scheme to FTs-based DGVMs.

V Future challenges and conclusions

1 Future challenges and research directions

The creation of a new generation of DGVMs is long-term work and requires a new traits-based framework, so collecting global FTs furnishes the foundation to construct the new DGVMs. TRY, the GLOPNET database and other trait-related databases have already been constructed and provide many traits that could be used in constructing a new generation of DGVMs. However, no existing database is ideal for this purpose; for example, the GLOPNET database includes few leaf traits from central and northern Africa, Russia, China and Canada (Wright et al., 2005b), so collecting trait data in these areas would help us to understand better trait–climate relationships. Moreover, previous research has focused primarily on above-ground traits, and fewer have endeavoured to include underground traits (Canadell et al., 2007). Identifying the traits correlated with root functions and testing the relationships between root, stem and leaf traits at a global scale is necessary (Canadell et al., 2007). It is important to identify traits that are easy to measure, but there are also many traits that are functionally important but impossible to measure owing to a lack of an easy analogue and logistical problems (Weiher et al., 1999), so developing new methods to measure these traits is an additional challenge for ecologists.

Major advances have been made in describing how climate factors filter traits and the relationships among traits in a region, but uncertainty increases when scaling up to the landscape and global levels (Tang and Bartlein, 2008; Zheng et al., 2010). A general decline in diversity or predictable functional shifts need to be considered if the scale changes (Loreau et al., 2001). Ackerly and Cornwell (2007) also proposed that traits with strong correlations at the regional and global scales might be decoupled at a local scale. Typically, DGVM modellers attempt to adjust model parameters via observation. Unfortunately, sometimes the parameters in the model do not match the observations because of limitations on temporal and spatial scales (Fisher et al., 2010). Improving the quality of model input data and validation data and promoting the coupling of different DGVMs are equally vital to calibrating model results and will help us greatly in understanding the mechanisms by which plants respond to climate change (Tang and Bartlein, 2008).

Although much work is available on traits-related theory, incorporating these results into DGVMs, improving the accuracy of results and reducing the uncertainty of model simulations are still large challenges. Certainly, we need not only large amounts of traits-based data but also proper methods with which to address these data. Meta-analysis (Curtis and Wang, 1998; Hedges et al., 1999) has been widely used in combining regional observation datasets into a global scale, so it will play an important role in constructing a global traits database. Model-data fusion (MDF) using inverse modelling and data assimilation provides a new quantitative method for constraining model prediction to a reasonable range (Peng et al., 2011). MDF will be an effective method for developing the new generation of DGVMs given the complexity of ecosystem processes and functions.

The priority in a new framework for DGVMs is to propose a traits-based scheme to replace the PFT scheme. Although much research has focused on traits–climate relationships, the simulation of vegetation establishment, growth and mortality without PFTs is also a challenge for ecologists. At the community level, understanding these mechanisms through FTs – as driven by climate factors and biotic interactions and the relationships between FTs and ecological functions – will require more theoretical, experimental and modelling approaches (Garnier and Navas, 2012).

2 Summary and conclusions

One of the main challenges in Earth system science is the integration of our growing knowledge from observations and experimental results into a coherent modelling framework that leads to the development of DGVMs. There have been several efforts to integrate the traits-based theory and scheme into PFT-based DGVMs. Although it is in its early stages, this approach appears promising. Several ecologists have proposed insightful conceptual frameworks for the development of future DGVMs. A number of these frameworks have been applied successfully at the regional scale and for parts of ecosystem processes, but there is still a long way to go and this research direction will face many unexpected problems when it is extrapolated to the global scale. Here, we proposed a transitional framework for a PFT-traits hybrid DGVM. In the PFT-traits hybrid DGVM, the PFT still exists, but will be replaced gradually by FTs in future work. Obviously, developing a new generation of DGVMs will undoubtedly prove challenging.

Footnotes

Acknowledgements

We thank IC Prentice for his constructive and valuable comments and suggestions on the initial version of this manuscript. We are grateful to the editor and two anonymous referees for their constructive comments and suggestions for improving the paper. Dr Martin Parkes is thanked for his great help with our writing.

Funding

This study was financially supported by the National Basic Research Program of China (2013CB956602), the National Natural Science Foundation of China (41201079), the Programme of New Century Excellent Talents in University of China (NCET) and Engineering Research Council of Canada (NSERC) Discover Grant.

References

Ackerly

Cornwell

(2007) A trait-based approach to community assembly: Partitioning of species trait values into within- and among-community components. Ecology Letters 10: 135–145.

Ali

Medlyn

Crous

. (2013) A trait-based ecosystem model suggests that long-term responsiveness to rising atmospheric CO₂ concentration is greater in slow-growing than fast-growing plants. Functional Ecology 27: 1011–1022.

Allison

Vitousek

(2004) Rapid nutrient cycling in leaf litter from invasive plants in Hawaii. Oecologia 141: 612–619.

Alton

(2011) How useful are plant functional types in global simulations of the carbon, water and energy cycles? Journal of Geophysical Research 116: 1–13.

Anderegg

WRL

(2014) Spatial and temporal variation in plant hydraulic traits and their relevance for climate change impacts on vegetation. New Phytologist 205:1008–1014.

Andersen

Endara

Turner

. (2012) Trait-based community assembly of understory palms along a soil nutrient gradient in a lower montane tropical forest. Oecologia 168: 519–531.

Andrew

Wulder

Nelson

(2014) Potential contributions of remote sensing to ecosystem service assessments. Progress in Physical Geography 38: 328–353.

Bakker

Carreño-Rocabado

Poorter

(2011) Leaf economics traits predict litter decomposition of tropical plants and differ among land use types. Functional Ecology 25: 473–483.

Barnett

Finlay

Beisner

(2007) Functional diversity of crustacean zooplankton communities: Towards a trait-based classification. Freshwater Biology 52: 796–813.

10.

Bernhardt-Römermann

Gray

Vanbergen

. (2011) Functional traits and local environment predict vegetation responses to disturbance: A pan-European multi-site experiment. Journal of Ecology 99: 777–787.

11.

Berry

Roderick

(2002) Estimating mixtures of leaf functional types using continental-scale satellite and climatic data. Global Ecology and Biogeography 11: 23–39.

12.

Box

(1981) Macroclimate and Plant Forms: An Introduction to Predictive Modeling in Phytogeography. Hague: Dr W Junk Publishers.

13.

Box

(1996) Plant functional types and climate at the global scale. Journal of Vegetation Science 7: 309–320.

14.

Canadell

Pataki

Pitelka

(eds) (2007) Terrestrial Ecosystems in a Changing World. Berlin Heidelberg: Springer-Verlag.

15.

Ceulemans

Merckx

Hens

. (2011) A trait-based analysis of the role of phosphorus vs nitrogen enrichment in plant species loss across north-west European grasslands. Journal of Applied Ecology 48: 1155–1163.

16.

Chaturvedi

Raghubanshi

Singh

(2011) Plant functional traits with particular reference to tropical deciduous forests: A review. Journal of Biosciences 36: 963–981.

17.

Chave

Coomes

Jansen

. (2009) Towards a worldwide wood economics spectrum. Ecology Letters 12: 351–366.

18.

Clark

Bell

Hersh

. (2011) Climate change vulnerability of forest biodiversity: Climate and competition tracking of demographic rates. Global Change Biology 17: 1834–1849.

19.

Coley

Bryant

Chapin

III (1985) Resource availability and plant antiherbivore defense. Science (Washington) 230: 895–899.

20.

Cornelissen

Lavorel

Garnier

. (2003) A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Australian Journal of Botany 51: 335–380.

21.

Cunningham

Read

(2002) Comparison of temperate and tropical rainforest tree species: Photosynthetic responses to growth temperature. Oecologia 133: 112–119.

22.

Curtis

Wang

(1998) A meta-analysis of elevated CO₂ effects on woody plant mass, form and physiology. Oecologia 113: 299–313.

23.

de Bello

Lavorel

Díaz

. (2010) Towards an assessment of multiple ecosystem processes and services via functional traits. Biodiversity and Conservation 19: 2873–2893.

24.

Díaz

Cabido

(1997) Plant functional types and ecosystem function in relation to global change. Journal of Vegetation Science 8: 463–474.

25.

Díaz

Cabido

Casanoves

(1998) Plant functional traits and environmental filters at a regional scale. Journal of Vegetation Science 9: 113–122.

26.

Díaz

Cabido

Zak

. (1999) Plant functional traits, ecosystem structure and land use history along a climatic gradient in central western Argentina. Journal of Vegetation Science 10: 651–660.

27.

Díaz

Noy-Meir

Cabido

(2001) Can grazing response of herbaceous plants be predicted from simple vegetative traits? Journal of Applied Ecology 38: 497–508.

28.

Douma

Witte

JPM

Aerts

. (2012) Towards a functional basis for predicting vegetation patterns: Incorporating plant traits in habitat distribution models. Ecography 35: 294–305.

29.

Drenovsky

Grewell

D’Antonio

. (2012) A functional trait perspective on plant invasion. Annals of Botany 110: 141–153.

30.

Ehleringer

Monson

(1993) Evolutionary and ecological aspects of photosynthetic pathway variation. Annual Review of Ecology and Systematics 24: 411–439.

31.

Eissenstat

Wells

Yanai

. (2000) Building roots in a changing environment: Implications for root longevity. New Phytologist 147: 33–42.

32.

Farquhar

Caemmerer

Berry

(1980) A biochemical model of photosynthetic CO₂ assimilation in leaves of C₃ species. Planta 149: 78–90.

33.

Farquhar

O’Leary

Berry

(1982) On the relationship between carbon isotope discrimination and the intercellular carbon dioxide concentration in leaves. Functional Plant Biology 9: 121–137.

34.

Fisher

McDowell

Purves

. (2010) Assessing uncertainties in a second-generation dynamic vegetation model caused by ecological scale limitations. New Phytologist 187: 666–681.

35.

Foley

Prentice

Ramankutty

. (1996) An integrated biosphere model of land surface processes, terrestrial carbon balance and vegetation dynamics. Global Biogeochemical Cycles 10: 603–628.

36.

Frenette-Dussault

Shipley

Meziane

. (2013) Trait-based climate change predictions of plant community structure in arid steppes. Journal of Ecology 101: 484–492.

37.

Gallagher

Leishman

Moles

(2011) Traits and ecological strategies of Australian tropical and temperate climbing plants. Journal of Biogeography 38: 828–839.

38.

Garnier

Navas

(2012) A trait-based approach to comparative functional plant ecology: Concepts, methods and applications for agroecology. A review. Agronomy for Sustainable Development 32: 365–399.

39.

Gerber

Hedin

Oppenheimer

. (2010) Nitrogen cycling and feedbacks in a global dynamic land model. Global Biogeochemical Cycles 24: GB1001.

40.

Grime

(1977) Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. American Naturalist: 1169–1194.

41.

Harrison

Prentice

Barboni

. (2010) Ecophysiological and bioclimatic foundations for a global plant functional classification. Journal of Vegetation Science 21: 300–317.

42.

Hartley

Garnett

Sommerkorn

. (2012) A potential loss of carbon associated with greater plant growth in the European Arctic. Nature Climate Change 2: 875–879.

43.

Hedges

Gurevitch

Curtis

(1999) The meta-analysis of response ratios in experimental ecology. Ecology 80: 1150–1156.

44.

Higgins

Langan

Scheiter

(2014) Progress in DGVMs: A comment on ‘Impacts of trait variation through observed trait–climate relationships on performance of an Earth system model: A conceptual analysis’ by Verheijen et al. (2013). Biogeosciences 11: 4357–4360.

45.

Kattge

Díaz

Lavorel

. (2011) TRY: A global database of plant traits. Global Change Biology 17: 2905–2935.

46.

Keddy

(1992) Assembly and response rules: Two goals for predictive community ecology. Journal of Vegetation Science 3: 157–164.

47.

Khurana

Sagar

Singh

(2006) Seed size: A key trait determining species distribution and diversity of dry tropical forest in northern India. Acta Oecologia 29: 196–204.

48.

Kucharik

Foley

Delire

. (2000) Testing the performance of a dynamic global ecosystem model: Water balance, carbon balance and vegetation structure. Global Biogeochemical Cycles 14: 795–825.

49.

Lavorel

Garnier

(2002) Predicting changes in community composition and ecosystem functioning from plant traits: Revisiting the Holy Grail. Functional Ecology 16: 545–556.

50.

Lavorel

Grigulis

(2012) How fundamental plant functional trait relationships scale-up to trade-offs and synergies in ecosystem services. Journal of Ecology 100: 128–140.

51.

Lavorel

McIntyre

Landsberg

. (1997) Plant functional classifications: From general groups to specific groups based on response to disturbance. Trends in Ecology & Evolution 12: 474–478.

52.

Leishman

(1999) How well do plant traits correlate with establishment ability? Evidence from a study of 16 calcareous grassland species. New Phytologist 141: 487–496.

53.

Leishman

Wright

Moles

. (2000) The evolutionary ecology of seed size. Seeds: The ecology of Regeneration in Plant Communities 2: 31–57.

54.

Loreau

Naeem

Inchausti

. (2001) Biodiversity and ecosystem functioning: Current knowledge and future challenges. Science 294: 804–808.

55.

Luck

Lavorel

McIntyre

. (2012) Improving the application of vertebrate trait-based frameworks to the study of ecosystem services. Journal of Animal Ecology 81: 1065–1076.

56.

McGill

Enquist

Weiher

. (2006) Rebuilding community ecology from functional traits. Trends in Ecology & Evolution 21: 178–185.

57.

McMahon

Harrison

Armbruster

. (2011) Improving assessment and modelling of climate change impacts on global terrestrial biodiversity. Trends in Ecology & Evolution 26: 249–259.

58.

Meers

Bell

Enright

. (2008) Role of plant functional traits in determining vegetation composition of abandoned grazing land in north-eastern Victoria, Australia. Journal of Vegetation Science 19: 515–524.

59.

Moles

Warton

Warman

. (2009) Global patterns in plant height. Journal of Ecology 97: 923–932.

60.

Murray

Watson

Prentice

(2012) The use of dynamic global vegetation models for simulating hydrology and the potential integration of satellite observations. Progress in Physical Geography 37: 63–97.

61.

Navas

(2012) Trait-based approaches to unravelling the assembly of weed communities and their impact on agro-ecosystem functioning. Weed Research 52: 479–488.

62.

Ordoñez

Van Bodegom

Witte

JPM

. (2009) A global study of relationships between leaf traits, climate and soil measures of nutrient fertility. Global Ecology and Biogeography 18: 137–149.

63.

Ostle

Smith

Fisher

. (2009) Integrating plant–soil interactions into global carbon cycle models. Journal of Ecology 97: 851–863.

64.

Ostonen

Puttsepp

Biel

. (2007) Specific root length as an indicator of environmental change. Plant Biosystems 141: 426–442.

65.

Pavlick

Drewry

Bohn

. (2013) The Jena Diversity-Dynamic Global Vegetation Model (JeDi-DGVM): A diverse approach to representing terrestrial biogeography and biogeochemistry based on plant functional trade-offs. Biogeosciences 10: 4137–4177.

66.

Peng

(2000) From static biogeographical model to dynamic global vegetation model: A global perspective on modelling vegetation dynamics. Ecological Modelling 135: 33–54.

67.

Peng

Guiot

. (2011) Integrating models with data in ecology and palaeoecology: Advances towards a model-data fusion approach. Ecology Letters 14: 522–536.

68.

Perez-Harguindeguy

Diaz

Garnier

. (2013) New handbook for standardised measurement of plant functional traits worldwide. Australian Journal of Botany 61: 167–234.

69.

Poorter

Bongers

(2006) Leaf traits are good predictors of plant performance across 53 rain forest species. Ecology 87: 1733–1743.

70.

Poulter

Ciais

Hodson

. (2011) Plant functional type mapping for earth system models. Geoscientific Model Development 4: 993–1010.

71.

Prentice

Cowling

(2013) Dynamic global vegetation models. In: Levin

(ed) Encyclopedia of Biodiversity, 2nd edn. Waltham: Academic Press, 670–689.

72.

Prentice

Webb

(1998) BIOME 6000: Reconstructing global mid-Holocene vegetation patterns from palaeoecological records. Journal of Biogeography 25: 997–1005.

73.

Prentice

Cramer

Harrison

. (1992) A global biome model based on plant physiology and dominance, soil properties and climate. Journal of Biogeography 19: 117–134.

74.

Prentice

Dong

Gleason

. (2014) Balancing the costs of carbon gain and water transport: Testing a new theoretical framework for plant functional ecology. Ecology Letters 17: 82–91.

75.

Prentice

Meng

Wang

. (2011) Evidence of a universal scaling relationship for leaf CO₂ drawdown along an aridity gradient. New Phytologist 190: 169–180.

76.

Quetier

Lavorel

Thuiller

. (2007) Plant-trait-based modeling assessment of ecosystem-service sensitivity to land-use change. Ecological Applications 17: 2377–2386.

77.

Quillet

Peng

Garneau

(2010) Toward dynamic global vegetation models for simulating vegetation-climate interactions and feedbacks: Recent developments, limitations and future challenges. Environmental Reviews 18: 333–353.

78.

Reich

Uhl

Walters

. (1991) Leaf lifespan as a determinant of leaf structure and function among 23 Amazonian tree species. Oecologia 86: 16–24.

79.

Reich

Oleksyn

(2004) Global patterns of plant leaf N and P in relation to temperature and latitude. Proceedings of the National Academy of Sciences 101: 11,001–11,006.

80.

Reich

Walters

Ellsworth

(1997) From tropics to tundra: Global convergence in plant functioning. Proceedings of the National Academy of Sciences 94: 13,730–13,734.

81.

Reich

Wright

Lusk

(2007) Predicting leaf physiology from simple plant and climate attributes: A global GLOPNET analysis. Ecological Applications 17: 1982–1988.

82.

Reu

Proulx

Bohn

. (2011) The role of climate and plant functional trade-offs in shaping global biome and biodiversity patterns. Global Ecology and Biogeography 20: 570–581.

83.

Sack

Scoffoni

(2013) Leaf venation: Structure, function, development, evolution, ecology and applications in the past, present and future. New Phytologist 198: 983–1000.

84.

Sato

Itoh

Kohyama

(2007) SEIB–DGVM: A new dynamic global vegetation model using a spatially explicit individual-based approach. Ecological Modelling 200: 279–307.

85.

Savage

Webb

Norberg

(2007) A general multi-trait-based framework for studying the effects of biodiversity on ecosystem functioning. Journal of Theoretical Biology 247: 213–229.

86.

Scheiter

Langan

Higgins

(2013) Next-generation dynamic global vegetation models: Learning from community ecology. New Phytologist 198: 957–969.

87.

Shipley

(2010) From Plant Traits to Vegetation Structure. Cambridge: Cambridge University Press.

88.

Soudzilovskaia

Elumeeva

Onipchenko

. (2013) Functional traits predict relationship between plant abundance dynamic and long-term climate warming. Proceedings of the National Academy of Sciences 110: 18,180–18,184.

89.

Suding

Lavorel

Chapin

. (2008) Scaling environmental change through the community-level: A trait-based response-and-effect framework for plants. Global Change Biology 14: 1125–1140.

90.

Suter

Edwards

(2013) Convergent succession of plant communities is linked to species’ functional traits. Perspectives in Plant Ecology, Evolution and Systematics 15: 217–225.

91.

Swenson

Weiser

(2010) Plant geography upon the basis of functional traits: An example from eastern North American trees. Ecology 91: 2234–2241.

92.

Tang

Bartlein

(2008) Simulating the climatic effects on vegetation: Approaches, issues and challenges. Progress in Physical Geography 32: 543–556.

93.

Turnbull

Rees

Crawley

(1999) Seed mass and the competition/colonization trade-off: A sowing experiment. Journal of Ecology 87: 899–912.

94.

Van Bodegom

Douma

Verheijen

(2014) A fully traits-based approach to modeling global vegetation distribution. Proceedings of the National Academy of Sciences 111: 13,733–13,738.

95.

Van Bodegom

Douma

Witte

JPM

. (2012) Going beyond limitations of plant functional types when predicting global ecosystem-atmosphere fluxes: Exploring the merits of traits-based approaches. Global Ecology and Biogeography 21: 625–636.

96.

Vendramini

Díaz

Gurvich

. (2002) Leaf traits as indicators of resource-use strategy in floras with succulent species. New Phytologist 154: 147–157.

97.

Verheijen

Brovkin

Aerts

. (2013) Impacts of trait variation through observed trait–climate relationships on performance of an Earth system model: A conceptual analysis. Biogeosciences 10: 5497–5515.

98.

Verlinden

Fichot

Broeckx

. (2015) Carbon isotope compositions (^δ13C) of leaf, wood and holocellulose differ among genotypes of poplar and between previous land uses in a short-rotation biomass plantation. Plant Cell Environment 38: 144–156.

99.

Violle

Garnier

Lecoeur

. (2009) Competition, traits and resource depletion in plant communities. Oecologia 160: 747–755.

100.

Violle

Navas

Vile

. (2007) Let the concept of trait be functional! Oikos 116: 882–892.

101.

Wallenstein

Hall

(2012) A trait-based framework for predicting when and where microbial adaptation to climate change will affect ecosystem functioning. Biogeochemistry 109: 35–47.

102.

Webb

Hoeting

Ames

. (2010) A structured and dynamic framework to advance traits-based theory and prediction in ecology. Ecology Letters 13: 267–283.

103.

Weiher

Werf

Thompson

. (1999) Challenging Theophrastus: A common core list of plant traits for functional ecology. Journal of Vegetation Science 10: 609–620.

104.

Westoby

Falster

Moles

. (2002) Plant ecological strategies: Some leading dimensions of variation between species. Annual Review of Ecology and Systematics 33: 125–159.

105.

Whittaker

(1970) Communities and Ecosystems. London: Macmillan Company, Collier-Macmillan Limited.

106.

Williams

Jackson

Kutzbach

(2007) Projected distributions of novel and disappearing climates by 2100 ad . Proceedings of the National Academy of Sciences 104: 5738–5742.

107.

Wilson

Thompson

Hodgson

(1999) Specific leaf area and leaf dry matter content as alternative predictors of plant strategies. New Phytologist 143: 155–162.

108.

Witte

JPM

Wójcik

Torfs

. (2007) Bayesian classification of vegetation types with Gaussian mixture density fitting to indicator values. Journal of Vegetation Science 18: 605–612.

109.

Woodward

Diament

(1991) Functional approaches to predicting the ecological effects of global change. Functional Ecology: 202–212.

110.

Woodward

Cramer

(1996) Plant functional types and climatic change: Introduction. Journal of Vegetation Science 7: 306–308.

111.

Wright

Reich

Cornelissen

. (2005a) Assessing the generality of global leaf trait relationships. New Phytologist 166: 485–496.

112.

Wright

Reich

Cornelissen

JHC

. (2005b) Modulation of leaf economic traits and trait relationships by climate. Global Ecology and Biogeography 14: 411–421.

113.

Wright

Reich

Westoby

(2003) Least-cost input mixtures of water and nitrogen for photosynthesis. The American Naturalist 161: 98–111.

114.

Wright

Reich

Westoby

. (2004) The worldwide leaf economics spectrum. Nature 428: 821–827.

115.

Wullschleger

Epstein

Box

. (2014) Plant functional types in Earth system models: Past experiences and future directions for application of dynamic vegetation models in high-latitude ecosystems. Annals of Botany 114: 1–16.

116.

Zheng

Ren

Lan

. (2010) Effects of grazing on leaf traits and ecosystem functioning in Inner Mongolia grasslands: Scaling from species to community. Biogeosciences 7: 1117–1132.