The Tree of Life in ecosystems: evolution of plant effects on carbon and nutrient cycling


  • Johannes H. C. Cornelissen,

    Corresponding author
    1. Systems Ecology, Department of Ecological Sciences, VU University, Amsterdam, The Netherlands
    Search for more papers by this author
  • William K. Cornwell

    1. Systems Ecology, Department of Ecological Sciences, VU University, Amsterdam, The Netherlands
    2. Evolution & Ecology Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW, Australia
    Search for more papers by this author


  1. The time is now ripe for ecologists and evolutionary biologists together to tackle a huge and important new challenge: mapping plant species effects on ecosystem functions onto the Tree of Life. This new research agenda is now tractable because of major recent advances in (i) screening plant species world-wide for the traits that support these functions, as well as in our understanding of how these traits support these functions; (ii) genetic screening and bioinformatics to build huge molecular plant phylogenies; and (iii) the comparative methods tools to analyse traits and phylogenies together.

  2. Understanding the evolutionary dynamics of traits related to biogeochemical cycling requires concerted research activity at a range of different phylogenetic scales, from analyses that consider data from all land plants to micro-evolution and other sources of intraspecific variation. This effort also requires study of the different plant organs involved in carbon and nutrient uptake, processing and transport, including the coordination between them.

  3. This overall effort should reveal (i) which lineages are associated with fast and which with slow turnover of carbon and nutrients (and water); (ii) how they are associated with fast versus slow turnover in terms of the traits involved; (iii) when in evolutionary history key transitions between fast and slow occurred; and (iv) what were and are the consequences of such variation in plant carbon and nutrient traits and turnover rates for past, present and future ecosystem functioning and services.

  4. Here, we introduce seven papers that together help set up this research agenda, each at a different phylogenetic level and resolution, and for different plant organs. Recent studies have begun to reveal evolutionary patterns in traits that matter to carbon and nutrient cycling for other groups of organisms, such as decomposing fungi and macrodetritivores and herbivorous animals. Ultimately such studies will all help towards the goal of mapping effects of organisms on biogeochemical cycling onto the entire Tree of Life.


The study of plant effects on carbon and nutrient cycling has traditionally been the domain of plant ecologists who focused on the ecology of present-day species in their communities, ecosystems and biomes. This literature has on the one hand covered (mostly interspecific) variation in plant traits related to carbon and nutrient gain to support vegetation productivity. Examples of such traits include inherent relative growth rate, specific leaf area, foliar nutrient concentrations, root dry matter content, wood density and anatomy (Lambers & Poorter 1992; Grime et al. 1997; Reich, Walters & Ellsworth 1997; Poorter & Bongers 2006; Castro-Diez et al. 1998; Westoby et al. 2002; Díaz et al. 2004; Garnier et al. 2004; Wright et al. 2004; Freschet et al. 2010). An interesting trait in this context is the type (and intensity) of mycorrhizal association of a plant species (Cornelissen et al. 2001), which itself may depend on the functional traits of the fungal taxa involved and the plant's ability to regulate the symbiosis (Van der Heijden & Scheublin 2007; Kiers et al. 2011). On the other hand, a complementary literature has covered interspecific variation in traits related to carbon and nutrient losses from plant material, for example through fire (Scarff & Westoby 2006; Schwilk & Caprio 2011) and decomposition (Hobbie 1996; Aerts & Chapin 2000; Cornwell et al. 2008; Fortunel et al. 2009; Freschet et al. 2013), and in (partly overlapping) traits related to soil carbon sequestration (De Deyn, Cornelissen & Bardgett 2008).

This literature did not address how such variation in traits relating to carbon and nutrient gains, losses and storage arose evolutionarily. Neither did it focus on whether considering the evolution of such traits might help to better understand the functioning of both extant and past plant communities and ecosystems. Paleo-ecologists have focused on past ecosystem functions using reconstructions based on fossil plant remains of deep-time ancestors rather than on the traits of their extant descendants. At the same time, phylogeneticists have been concerned with increasingly accurate estimates of evolutionary relatedness among species, that is, the topology and branch lengths of the true Tree of Life. They used variation in traits (or ‘characters’) also, but often as a tool to help them with this mapping per se, or in some cases to reconstruct the evolution of organismal functioning, without much regard for the ecosystem consequences of the adaptations of different species, now or in the past. Recently, ecologists and evolutionary biologists have started to join forces to apply multi-species evolutionary information to help understand the functioning of communities and ecosystems in the past, present and future (e.g. Schwilk & Ackerly 2001; Ackerly, Schwilk & Webb 2006; Prinzing et al. 2008; Berendse & Scheffer 2009; Donovan et al. 2010; Mouquet et al. 2012; Cavender-Bares et al. 2009; Smith, Beaulieu & Donoghue 2009; Stock & Verboom 2012; Diaz et al. 2013; Swenson 2013).

Now, for the first time, we are in a good position to exploit this momentum of linking ecological and evolutionary information. This now enables us to tackle a huge and exciting new challenge, which is to build the Tree of Life of plant species effects on carbon and nutrient cycling – with possible extensions to water cycling. For instance, we will soon know (i) which (large or tiny) clades are associated with fast and which with slow turnover of carbon and nutrients (and water) in ecosystems, (ii) how these rates can be explained from their traits, (iii) when in evolutionary history they evolved and prevailed and (iv) with what consequences for past, present and future communities and ecosystems. We can do this by mapping the key plant traits of extant species that drive aspects of these cycles onto the phylogeny, or we may even directly map out the phylogeny of ‘ecosystem effect functions’ such as species’ litter decomposability or flammability, which are underpinned by several ‘effect traits’ simultaneously (Diaz et al. 2013). Achieving this goal will generate a powerful new line of inference (in addition to fossil data) about rates of cycling of matter through geological time. In addition, new understanding about the evolution of traits that drive biogeochemical cycling will produce a deeper understanding of the dynamics of carbon and nutrient dynamics in ecosystems through time. Several key carbon and nutrient cycling traits, and drought tolerance, have a clear relationship with phylogeny (Brodribb & Field 2010) and profound effects on ecosystem services (Diaz et al. 2006), implying that there is much scope for optimism about the task ahead.

To firmly position this new research agenda, we organized a symposium at the INTECOL-BES conference in London in August 2013, as well as this associated Special Feature in Journal of Ecology. The two major goals of this symposium, with the same title as this paper, were to (i) integrate information from plant trait data bases, molecular phylogeny and micro-evolutionary theory, to capture the most complete picture of the evolution of plant traits to date; (ii) integrate this information with empirically-derived relationships between interspecific variation in plant traits and plant effects on carbon and nutrient cycling functions and processes. The seven very different contributions presented here all highlight the state-of-the-art possibilities but also problems and caveats related to this overall agenda (Fig. 1). They highlight the fact that the success of tackling this eco-evolutionary challenge depends on simultaneous progress in different complementary research fields at different scales in terms of phylogenetic resolution. The success may also depend on whether we manage to take a whole-plant view and piece together data for the different plant organs and their interactions. It is clear that all these elements have their particular research methodologies, some already more advanced than others.

Figure 1.

Linking the Tree of Life and carbon and nutrient cycling through plant and mycorrhizal traits at different scales. The circular arrow (top left) represents carbon and nutrient cycling. The numbers relate to the seven contributions in this Special Feature: (1) Reich (2014); (2) Donovan et al. (2014); (3) Behm & Kiers (2014); (4) Edwards et al. (2014); (5) Liu et al. (2014); (6) Cornwell et al. (2014); (7) Barot et al. (2014). The numbers on the left side of the panel relate to the phylogenetic scale covered by each paper, from within and between species (2,3), via genus level (2,4) down to larger branches of the phylogeny (1,5,6). Eco-evolutionary feedback to nutrient cycling is covered by contribution 7. The numbers on the right relate to the scale of the organs for which traits are covered by each paper, from leaves (2,5,7), foliated branches (4), roots (2) including their arbuscular mycorrhizal fungi (3), above-ground plant (6) to whole plant (1). Contributions 2,3,4 and 6 cover traits related to carbon and nutrient gain, contribution 5 covers decomposition (and thereby nutrient turnover) while contributions 1 and 7 cover both carbon and nutrient gain and release processes. Figure derived with permission from original artwork by Eliza Jewitt.

Constraints and advances in methodology and data availability

One reason why we could not start to build the Tree of Life of plant-associated biogeochemistry until now is that there were severe limitations of methodology and (standardized) data availability. But now the two parallel fields in plant ecology and systematics that matter most to tackling this challenge have both seen recent, massive progress. The first is the global synthesis of plant trait data. On the productive side of carbon and nutrient cycling, large global data sets of traits representing the trade-off between photosynthetic rates and growth rates versus organ life span and nutrient conservation have now been standardized and assembled (Wright et al. 2004; Kleyer et al. 2008; Chave et al. 2009; Kattge et al. 2011; Pérez-Harguindeguy et al. 2013). On the carbon and nutrient release side, multi-species data bases of traits and specific effect functions representing ‘afterlife effects’ of plant organs on organic matter turnover are growing fast. These include litter decomposability and flammability as well as their various underlying traits (Scarff & Westoby 2006; Cornwell et al. 2008; Schwilk & Caprio 2011; Freschet et al. 2013). The effects of interspecific variation in these plant traits on major aspects of the global carbon and nutrient cycles, including productivity, decomposition and fire regimes, are now beginning to be understood better (Brovkin et al. 2012; Verheijen et al. 2013).

However, these plant traits did not arise de novo. Instead, they are the result of a long process of evolution. This brings us to the second advance, namely the rapid growth of GenBank and bioinformatic tools, which now allow us to build molecular phylogenies that cover thousands of plant species (e.g. Zanne et al. 2013); together with Bayesian approaches, it is now possible to obtain a better time-calibrated picture of the evolutionary history of plants than ever before. It is worth noting that these advances may be applied with great precision at small phylogenetic scales or with less precision at the largest scale of the Tree of Life. Both large-scale and small-scale analyses are crucial: to interpret and generalize findings from larger branches of the plant phylogeny, it is important to know the tempo and mode of evolution of the relevant traits at the tips of these branches over more recent history and over shorter time-scales that matter to current climatic changes; and we need to quantify sources of intraspecific trait variation (e.g. genotypic variation between and within populations, phenotypic plasticity) in different taxa and different environmental contexts.

Whereas important and promising synthesis of phylogenetic and functional trait data sets is beginning to be achieved (e.g. Smith & Donoghue 2008; Jones et al. 2014), to date, the phylogenies and trait data sets have been piecemeal in nature, examining one trait at a time (e.g. Moles et al. 2005). The success or failure of a species (or a clade) depends on a suite of plant traits in relation to environmental conditions and co-occurring lineages at each point in time. As such, more progress can be made by examining the correlated divergences across many traits. For example, early divergence in vascular transport pathways may be crucial in this regard (Brodribb & Field 2010). Some plant traits have been linked to shifts in rates of molecular evolution within small- to medium-sized clades (e.g. growth form; Smith & Donoghue 2008). However, despite the rapid recent growth of both trait and genetic data bases, the full matrix of information on a global scale still has many gaps and the complete multivariate picture of plant evolution has yet to emerge. The seven papers in this Special Feature, briefly summarized below, together indicate both the promise and the challenges ahead for this research agenda.

New progress and challenges in mapping plant biogeochemistry traits onto the Tree of Life

In this Special Feature, Reich (2014) first sets the scene with a monumental literature-based synthesis of plant trait variation related to biogeochemical cycling as it has been measured across extant species, subject to physiological, biochemical and structural trade-offs and coordination of plant design. He examines scales from the plant organ level (1 in Fig. 1), that is, from the ‘leaf economics spectrum’ (Wright et al. 2004) to the ‘plant economics spectrum’, integrating the major plant organs involved in carbon and nutrient dynamics (Freschet et al. 2010; Fortunel, Fine & Baraloto 2012). This review frames a number of current research questions including whether adaptations to low or high availability of different resources, that is nutrients, carbon (through light regime) and water, are all subject to the same trade-offs. In other words, he asks whether plants are designed to have either generally fast internal turnover of all these different resources or generally slow turnover. He provides multiple empirical examples to conclude with a positive answer to this big question, a benchmark for subsequent researchers to further test and apply in different environmental contexts. In the spirit of this Special Feature, Reich (2014) also illustrates and discusses how this cross-species, cross-organ, cross-resource economics spectrum can help us to understand and integrate both ecosystem-scale cycling of carbon, nitrogen and phosphorus and that of water.

Now, against this background of present-day variation in plant traits related to carbon and nutrient cycling, we move to the actual mapping of such traits onto the Tree of Life, following a sequence from fine to coarse phylogenetic resolution. Starting at the tips of the Tree's branches (2, 3 in Fig. 1), Donovan et al. (2014) demonstrate how key traits in the context of carbon and nutrient cycling (leaf and root [N], leaf mass per area (LMA), specific root length) vary strongly at the tips of the phylogeny, in this case among and within closely related Helianthus species. They find large trait variation among very closely related species and populations. For instance, Helianthus species show substantial intraspecific variation due to genetic differences, environmental responses and ontogeny. The intraspecific variation for leaf [N] and LMA is even large enough to overlap with a significant section of the leaf economics spectrum across species world-wide (cf. Wright et al. 2004). An important message from this paper is that one has to be aware of the risks associated with taking a species’ mean trait value (e.g. from a trait data base) for phylogenetic trait mapping purposes and that the level of this risk may depend also on the phylogenetic scale and position of a study.

It can get even more complex than this. As mentioned above, mycorrhizal association type can be seen as a plant trait to some extent. However, Behm and Kiers (2014) argue that within a type, in their case arbuscular mycorrhizal fungi (AMF), major trait variation may exist both between and within fungal species. Intraspecific variation may be due to both phenotypic plasticity and to genotypic variation. As for the latter, the fusing of hyphae leads to cytoplasms with more than one nucleus, which not only inflates genotypic variation but also makes it difficult to delimit and distinguish individual species. Moreover, some but not all AMF genotypes are loyal to particular host plants, and their effects on the performance of these host plants may also vary (Helgason et al. 2002), which could give mismatches of phylogenetic scale and position of traits of host plant versus those of AMF. At this stage, we cannot quantify what these potentially complex relations could mean for plant carbon and nutrient dynamics in the Tree of Life and this certainly deserves in-depth study. Behm and Kiers (2014) make a very useful contribution towards this by laying out a framework for screening AMF for functional traits relevant to carbon and nutrient dynamics and for quantifying the genetic versus environment-driven contributions to AMF trait variation.

A little deeper down into the Tree of Life (4 in Fig. 1), at genus level, Edwards et al. (2014) examine the evolution of leaf economic traits within the deciduous species of the genus Viburnum. Among these species, there is a very different trait correlation pattern for LMA, leaf [N] and photosynthetic rate compared to the pattern across all species, that is, compared to Wright et al. (2004). Instead of following the overall pattern, in a novel finding, they show that within deciduous Viburnum, whole-plant architecture, which is relatively conserved, has a crucial connection to leaf traits. This shows that while there may be one pattern across all species, completely different trait coordination may occur within specific clades.

At a much broader phylogenetic scale, Liu et al. (2014) directly measure variation in an effect trait or specific effect function – decomposability. One untested hypothesis in the literature was that the rise to dominance of angiosperms in general represented a shift towards faster nutrient cycling (Berendse & Scheffer 2009). Some data support this idea – the leaves of angiosperms decompose faster than the gymnosperms on average (Cornwell et al. 2008). However, there was little phylogenetic resolution to these data, and very little was known about the pattern within the basal clades of the angiosperms. With a new data set from southern China, based on a large litter incubation experiment, Liu et al. (2014) show that within the magnoliid clade, leaf decomposability is slow, with only a few exceptions. This suggests that the real shift to fast potential decomposition rates for litter from tree species occurred later, once the eudicots became the dominants of tropical forests.

At the broadest phylogenetic scale considered within these issues – all vascular plants – Cornwell et al. (2014) present an analysis of data from 48 324 species, that is >10% of known global diversity. At this broad phylogenetic scale, the diversity and heterogeneity of evolutionary processes are considerable, and rather than try to model all of those processes, Cornwell et al. (2014) present a quantitative method for finding and describing the most distinctive lineages for each trait. In many cases, these trait lineages confirm existing natural history anecdotes, but other trait lineage combinations are completely novel. Furthermore, the identification of important lineage–trait combinations raises new questions about the processes that created these patterns. For example, Proteaceae with their tough leaves are global outliers with respect to LMA; is that due to an evolutionary jump or a gradual transition to distinctive trait values?

Finally, Barot et al. (2014) add a very different dimension to this theme by modelling trait-based eco-evo feedbacks with regard to nutrient cycling. They do not stop at the way in which plant trait evolution affects soil nutrient dynamics; they go full circle by incorporating a long-debated hypothesis (e.g. Berendse 1994) that species’ promotion of nutrient mineralization rates can be seen as an adaptive effect trait or specific effect function. They build a spatially explicit, individual-based simulation model in which plants are allowed to show evolutionary change (through mutations within given species or species replacements in the community) in their nutrient losses and effect on local nutrient mineralization rate. They show how the outcome of the feedback through plant mineralization strategy for plant biomass and soil carbon pools depends on spatial scale, local competition, seed dispersal and soil nutrient availability and distribution. The authors hypothesize, based on their model outcome, that this eco-evo feedback could lead to different effects on, for instance, soil carbon storage than ecological feedback alone, something of great interest and in need of real-life validation in the context of global carbon budgets and climate.


Here, we have introduced seven wide-ranging examples that are all complementary pieces of the big puzzle to map plant effects on biogeochemical cycling onto the Tree of Life and to apply this information to learn about the rates of matter cycling in past, present and future. Together, these contributions stress the important challenge of building an understanding of trait evolution across different phylogenetic scales. For instance, the confidence with which we can take species means for studying evolutionary trait patterns, and the ecosystem functions they represent, depends on the phylogenetic level and resolution of the analysis. Intraspecific variation may affect the outcome of such analysis significantly in many cases, but more strongly so towards the tips of the phylogeny. Such effort need not be limited to vascular plants. Several other organismal types and taxa have also been studied recently for evolutionary pattern in traits that matter to biogeochemistry. For instance, there have been recent advances in our understanding of phylogenetic pattern in fungal traits involved in the degradation of recalcitrant cell wall compounds (Eastwood et al. 2011; Floudas et al. 2012). Lang et al. (2009) revealed strong phylogenetic relations in the litter decomposability and underlying traits of non-vascular cryptogams including bryophytes and lichens. For some animal taxa, relatively large numbers of species have already been tested for traits involved in herbivory (e.g. in grasshoppers: Van der Plas, Anderson & Olff 2012; Moretti et al. 2013) and in organic matter decomposition (e.g. macrodetritivores: Heemsbergen et al. 2004; termites: Lo, Watanabe & Sugimura 2003; Freymann et al. 2008). Such studies beyond vascular plants all make important contributions towards a comprehensive Tree of Life of carbon and nutrient cycling. Thus, the synthesis of the connections between species and ecosystem biogeochemistry, molecular phylogenies and evolutionary theory using the approaches described above, and presented in this Special Feature, will become a benchmark to both ecologists and evolutionary biologists in the coming decades.


We thank Amy Austin, Lauren Sandhu and Amy Everard for helping to organize the INTECOL-BES symposium and this Special Feature; and the British Ecological Society for providing funds for the speakers at this symposium. JHCC benefitted from Grant CEP-12CDP007 by the Royal Netherlands Academy of Arts and Sciences (KNAW) and WKC from Grant 142.16.3032 of the Darwin Center for Biogeosciences.