Don't forget the details! Trait ecology and generality

The search for generality is perhaps the greatest driver of modern ecology and probably also the greatest source of ecological angst. Though ecological trends frequently reflect the newest, brightest hope for generality, the search for generality (perhaps by definition) encourages us to ignored details and complexities. Maybe this means that some areas of study won't develop fully until they've fallen out of fashion. And maybe this means that the most interesting science happens when the pressure to 'save community ecology' is gone. A great example of the kind of post-hype, thoughtful approach for trait-based ecology comes from Reynolds et al. (2017) in Tree Physiology. They do a really nice job of highlighting some of the details that must inform trait-based ecology. Here, Reynolds et al. take a broad comparative approach across species, but incorporate important details that have at times been overlooked - especially the role of the environment, recognizing and measuring both constitutive and plastic traits, captures that there are multiple paths (or trait combinations) that can result in similar functioning.

The authors look at four conspecific tree species (Brachychiton spp.) with different average positions along an observed moisture gradient (CMD or climate moisture deficit). Two species occupied drier areas of Australia ('xeric species'), while the other two were found in more moderate areas ('mesic species'). The authors assumed that the different distributions of these species reflect different hydraulic niches. Were species' hydraulic niches associate meaningfully with their traits, specifically those trait associated with drought stress responses. Though these species are closely related--and so huge divergences in form and function might not be expected--the costs and benefits of drought resistance should differ among the species. In dry environments, drought resistance strategies should be more important, and may select for particular traits or sets of traits. Trait states associated with drought conditions include "reduced leaf area, enhanced stomatal control, safer or more efficient xylem, increased tissue water capacitance...and/or deeper root systems " may all be selected for. On the other hand, investment in these traits when water is not limited is often costly, reducing growth and competition. This suggests a meaningful selective regime associated with the CMD gradient and trait values might exist.

One important, but oft-overlooked aspect of trait ecology is that trait values depend on both genes and the environment. Reynold et al. incorporate this fact this by manipulating water availability between drought and control treatments. They measured both constitutive components of trait values – those driven by genetics and expressed regardless of environments – and the plastic or environment-dependent components. For instance, in the presence of prolonged drought, trees might increase root production or change leaf characteristics. In addition to manipulating water availability between treatments, the authors measured nine traits related to morphology and allocation.

From Reynolds et al. 

Given the expectation that trait values reflect the complex interaction of genetics and the environment in different species, is it possible to even make simple predictions about trait-environment relationships? The authors find that "These complex relationships illustrate that assuming that individual traits (often measured on individuals under a single set of environmental conditions) reflect drought resistance is likely to be overly simplistic and may be erroneous for many species. However, our results do suggest that generalization may be possible, provided multiple traits are measured to explore specific integrated drought strategies."

Indeed, some results are relatively predictable relationships: under well-watered control conditions, the allocation of biomass matched the expectation: xeric species had higher investment in below-ground biomass and in transport tissues than the mesic species (both characteristic of a water-conserving species).

On the other hand, leaf traits such as SLA did not show any trend related to species' assumed drought tolerance, either for constitutive or plastic trait components. Sometimes traits associated with the leaf economic spectrum such as SLA are assumed to indicate stress tolerance, but this was not the case.

By far the most interesting result was the observation that the xeric species had the highest assimilation and stomatal conductance rate and the lowest water use efficiency under well-watered conditions. Only by also examining these species under drought conditions was it possible to observe that they are highly plastic with regards to water use efficiency. In fact, they show a feast or famine approach to water usage - "where high photosynthetic rates per unit leaf area and high investment in root and stem tissue even in well-watered conditions are achieved through profligate water use during rare periods of water availability, in order to establish a root system and stem storage tissues necessary to survive long periods of water stress." Under drought conditions, these species show reduced root tissue investment; in contrast, mesic species follow expected patterns and plastically increase root tissue investment.

This paper is a reminder that the details are also fascinating and informative. As humans, we may have a simplistic understanding of the realized environment sometimes. To us perhaps all water stress is similar, but for each species in this study, the long term selective pressures may be meaningfully different - in timing, duration, and life stage. This creates the potential for complex differences between species which may best be reflected via life history strategies involving multiple traits. That may still imply some degree of generality is possible, but it is multi-dimensional.

Works cited:
Victoria A Reynolds, Leander D L Anderegg, Xingwen Loy, Janneke HilleRisLambers, Margaret M Mayfield; Unexpected drought resistance strategies in seedlings of four Brachychiton species, Tree Physiology, https://doi.org/10.1093/treephys/tpx143

Gained in translation: translational ecology for the Anthropocene

A recent evaluation of the state of science around the world run by 3M found that 86% of the 14,000 people surveyed believed that they knew 'little to nothing' about science. 1/3 of all respondents also said they were skeptical of science and 20% went farther, saying that they mistrust scientists and their claims.

Those attitudes wouldn't surprise anyone following US politics these days. But they're still troubling statistics for ecologists. Perhaps more than most scientific disciplines, ecologists feel that their work needs to be communicated, shared, and acted on. That's because modern ecology can't help but explicitly or implicitly include humans – we are keystone species and powerful ecosystem engineers. And in a time where the effects of global warming are more impactful than ever, and where habitat loss and degradation underlie an age of human-caused extinction, ecology is more relevant than ever.

The difficulties in converting primary ecological literature into applications are often construed as being caused (at least in part) by the poor communication abilities of professional scientists. Typically, there is a call for ecologists to provide better science education and improve their communication skills. But perhaps this is an 'eco-centric' viewpoint – one that defaults to the assumption that ecologists have all the knowledge and just need to communicate it better. A more holistic approach must recognize that the gap between science and policy can only be bridged by meaningful two-way communication between scientists and stakeholders, and this communication must be iterative and focused on relevance for end-users.

William H. Schlessinger first proposed this practice - called Translational Ecology (TE) - nearly 8 years ago. More recently an entire special issue in Frontiers in Ecology and the Environment was devoted to the topic of translational ecology in 2017. [The introduction by F. Stuart Chapin is well worth a read, and I'm jealous of the brilliant use of Dickens in the epigraph: “It was the best of times, it was the worst of times, it was the age of wisdom, it was the age of foolishness, it was the epoch of belief, it was the epoch of incredulity.”]

Although applied ecology also is focused on producing and applying ecological knowledge for human problems, translational ecology can be distinguished by its necessary involvement of the end user and policy. Enquist et al. (2017, TE special issue) note: "Ecologists who specialize in translational ecology (TE) seek to link ecological knowledge to decision making by integrating ecological science with the full complement of social dimensions that underlie today's complex environmental issues."

From Hallet et al. (2017, TE special issue)

The essential component of translational ecology is a reliance on people or groups known as boundary spanners, which are the key to (effectively) bridging the chasm between research and application. These people or organizations have particular expertise and skill sets to straddle the divide between "information producers and users". Boundary spanners are accountable to the science and the user, and generally enable communication between those two groups.

Boundary spanners likely have interdisciplinary backgrounds, and integrate knowledge and skills from ecology and biology, as well as disciplines such as anthropology, human geography, sociology, law, or politics. The key issue in that boundary spanners can overcome is the lack of trust between information users and producers. Translational ecology – through communication, translation, and mediation – is especially focused on developing relationships with stakeholders and boundary spanners are meant to be particularly skilled at this. 

For example, academics publish papers, and then the transmission of information to potential users is usually allowed to occur passively. At best, this can be slow and inefficient. At worst, potential end users lack access, time, and awareness of the work. Boundary spanners (including academics) can ensure this knowledge is accessibly by producing synthetic articles, policy briefs and white papers, by creating web-based decision-support tools, or by communicating directly with end users in other ways. A great example of existing boundary spanners are Coop extension offices hosted at US land grant universities. Coops are extensions of government offices (e.g. USDA) whose mission is to span the knowledge produced by research and to bring it to users through informal education and communication. 

For working academics, it may feel difficult to jump into translational ecology. There can be strong institutional or time constraints, and for those without tenure, fear that translational activities will interfere with other requirements. Institutions interested in working with ecologists also often face limitations of time and funding, and variable funding cycles can mean that boundary-spanning activities lack continuity.

But what's hopeful about the discussion of translational ecology in this issue is that it doesn't have an individualistic viewpoint: translational ecology requires teams and communities to be successful, and everyone can contribute. I think there is sometimes a very simplistic expectation that individual scientists can and must be exceptional generalists able to do excellent research, write and give talks for peers, teach and lecture, mentor, and also communicate effectively with the general public (in addition to taking care of administration, human resources, ordering and receiving, and laboratory management). We can all contribute, especially by training boundary spanners in our departments and labs. As F.S. Chapin says, "The key role of context in translational ecology also means that there are roles that fit the interests, passions, and skills of almost any ecologist, from theoreticians and disciplinarians to people more focused on spanning boundaries between disciplines or between theory and practice. We don't need to choose between translational ecology and other scientific approaches; we just need to provide space, respect, and rigorous training for those who decide to make translational ecology a component of their science." 

From Enquist et al. (2017, TE special issue).

References:
Special Issue: Translational ecology. Volume 15, Issue 10. December 2017. Frontiers in Ecology and the Environment

Moving on up to the regional scale

Like the blind men and the elephant, perspective drives understanding in ecology. The temporal and spatial scale of analysis (let alone the system and species you focus on) has major implications for your conclusions. Most ecologists recognize this fact, but consider only particular systems, scales or contexts due to practical limitations (funding, reasonable experimental time frames, studentship lengths). 

Ecologists have long known that regional processes affect local communities and that local processes affect regional patterns. Entire subfields like landscape ecology, metapopulations, metacommunities, and biogeography (species area relationships) highlight these spatial dependencies. But high-profile ecological research into biodiversity and ecosystem functioning ('BEF') primarily considers only local communities. Recently though, the literature has started to fill this gap and asking what BEF relationships look like at larger spatial scales, and how well local BEF relationships predict those at larger spatial scales.

'Traditional' BEF experiments were done at relatively small spatial scales (often only a few meters^2). Positive BEF relationships were commonly observed, but often were quite saturating – that is, only a few species were necessary to optimize the function of interest. If the impact of biodiversity saturates with only a few species, it would seem that surprisingly few species are necessary to maintain functioning. True, studies that considered multiple ecosystem functions are more likely to conclude that additional diversity is required for optimal functioning (e.g. Zavaleta et al. 2010). But a simplistic evaluation of the facts that a) ecosystem functioning rapidly saturates with diversity, and b) locally, diversity may not be generally decreasing (Vellend et al. 2017), could lead to overly confident conclusions about the dangers of biodiversity loss. Research on BEF relationships, as they transition from local to larger spatial scales, is increasingly suggesting that our understanding is incomplete, and that BEF relationships can grow stronger at large spatial scales.

A number of recent papers have explored this question, and in considering the essential role of spatial scale. Predictions about how BEF relationships will change with spatial scale vary. On one hand, in most systems there are only a few dominant species and these species may disproportionately contribute to ecosystem functions, regardless of the spatial scale. On the other hand, species-area relationships tend to increase rapidly at small scales, as community composition turns over. If that is the case, then different species may make important contributions in different places. Winifree et al. (2018) contrasted these predictions for three crop species that rely on natural bee pollinators (cranberries, blueberries, and watermelons). They censused pollinators at 48 sites, over a total extent of ~3700 km^2. Though at local scales very few bee species were required to reach pollination goals, the same goals at larger spatial scales required nearly an order of magnitude more bee species. These results in particular appeared to be driven by species turnover among sites--perhaps due to underlying environmental heterogeneity.

From Winifree et al. "Cumulative number of bee species required to maintain thresholds of 25% (orange), 50% (black), and 75% (purple) of the mean observed level of pollination, at each of n sites (16). Horizontal dashed lines indicate the total number of bee species observed in each study. Error bars represent 1 SD over all possible starting sites for expanding the spatial extent. For all three crops combined, each x-axis increment represents the addition of one site per crop".

Another mechanism for increased BEF at larger scales is insurance effects. The presence of greater diversity can interact with spatial and temporal environmental variation to increase or stabilize ecosystem functioning. Greater diversity should maximize the differential responses of species to changing conditions, and so buffer variation in ecosystem functioning. Such effects, when they occur through time are temporal insurance, and when they occur via dispersal among sites, spatial insurance. Wilcox et al. (2018) considered the role of synchrony and asynchrony among populations, communities, and metacommunities to ask whether local asynchrony affected stability (see Figure below for a nice conceptual explanation). Across hundreds of plant data sets, they found that asynchrony of populations did enhance stability. However, the degree to which it affected stability varied from very weak to very important (e.g. by 1% to 300%). Maximizing species or population differences at local scales apparently can have implications for dynamics, and so potentially stability of functioning, at much larger scales.

From Wilcox et al. "Conceptual figure showing how stability and synchrony at various spatial scales within a metacommunity combine to determine the stability of ecosystem function (here, productivity). In (a), high synchrony of species within and among local communities results in low stability at the scale of the metacommunity. In (b), species remain synchronised within local communities, but the two communities exhibit asynchronous dynamics due to low population synchrony among local patches. This results in relatively high gamma stability. Lastly, in (c), species exhibit asynchronous dynamics within local communities through time, and species-level dynamics are similar across communities (i.e. high population synchrony). This results in relatively high gamma stability. Blue boxes on the right outline stability components and mechanisms, and the hierarchical level at which they operate. Adapted from Mellin et al. (2014)."

Finally, Isbell et al. (2018) describe ways in which ecosystem functioning and other contributions of nature to humanity are scale-dependent, laying out the most useful paths for future work (see figure below).

From Isbell et al. 2018.

These papers make nearly identical points worth reiterating here: 1) we have done far too little work beyond the smallest spatial scales (~3 m^2) and so lack necessary knowledge of the impacts of losing of biodiversity, and 2) policy decisions and conservation activities are occurring at much larger scales – at the scale of the park, the state, or the nation. Bridging this gap is essential if we are to make any reasonable arguments as to why ecosystem function figure into  large-scale conservation activities.

References:
Sustaining multiple ecosystem functions in grassland communities requires higher biodiversity. Erika S. Zavaleta, Jae R. Pasari, Kristin B. Hulvey, G. David Tilman. Proceedings of the National Academy of Sciences Jan 2010, 107 (4) 1443-1446; DOI: 10.1073/pnas.0906829107. 

Plant biodiversity change across scales during the Anthropocene. Vellend, Mark, et al. Annual review of plant biology 68 (2017): 563-586.

Species turnover promotes the importance of bee diversity for crop pollination at regional scales. RACHAEL WINFREE, JAMES R. REILLY, IGNASI BARTOMEUS, DANIEL P. CARIVEAU, NEAL M. WILLIAMS, JASON GIBBS. SCIENCE16 FEB 2018 : 791-793

Asynchrony among local communities stabilises ecosystem function of metacommunities. Kevin R. Wilcox, et al. Ecology Letters. Volume 20, Issue 12, Pages 1534–1545.

Isbell, Forest, et al. "Linking the influence and dependence of people on biodiversity across scales." Nature 546.7656 (2017): 65.

A general expectation for the paradox of coexistence

There are several popular approaches to the goal of finding generalities in ecology. One is essentially top down, searching for generalities across ecological patterns in multiple places and at multiple scales and then attempting to understand the underlying mechanisms (e.g. metabolic scaling theory and allometric approaches). Alternatively, the approach can be bottom up. It may consider multiple models or multiple individual mechanisms and find generalities in the patterns or relationships they predict. 

A great example of generalities from multiple models is in a recent paper published in PNAS (from Sakavara et al. 2018). It relies on, links together, and adds to, our understanding of community assembly and the effects of competition on the distribution of niches in communities. In particular, it adds additional support to the assertion that both combinations of either highly similar or highly divergent species can coexist, across a wide variety of models.

Work published in 2006 by Scheffer and van Nes played an important early role towards a reconciliation of neutral theory and niche-based approaches. They used a Lotka-Volterra model to highlight that communities could assemble with clusters of coexisting, similar species evenly spaced along a niche axis (Figure 1). Neutrality, or at least near-neutrality, could result even when dynamics were determined by niche differences. [Scheffer, van Nes, and Remi Vergnon also provide a nice commentary on the Sakavara et al. paper found here].

Fig. 1: From Scheffer and van Nes, emergent 'lumpiness' in communities.

One possibility is that Scheffer and van Nes's results might be due to the specifics of the L-V model rather than representing a general and biologically realistic expectation. Sakavara et al. address this issue using a mechanistic consumer-resource model in "Lumpy species coexistence arises robustly in fluctuating resource environments". Under this model, originally from Tilman's classic work with algae, coexistence is limited by the number of resources that limit a species' growth. For 2 species, for example, 2 resources must be present that limit species growth, and further the species must experience a tradeoff in their competitive abilities for the 2 resources. Coexistence can occur when each species is limited more by the resource on which it is most competitive (Figure 2). Such a model– in which resources limit coexistence—leads to an expectation that communities will assemble to maximize the dissimilarity of species.

Fig 2. From Sakavara et al. (2017).

Such a result occurs when resources are provided constantly, but in reality the rates of resource supply may well be cyclical or unpredictable. Will community assembly be similar (resulting in patterns of limiting similarity) when resources are variable in their supply? Or will clumps of similar species be able to coexist? Sakavara et al. considered this question using consumer-resource models of competition, where there are two fluctuating limiting resources. They simulated the dynamics of 300 competing species, which were assigned different trait values along a trait gradient. Here the traits were the half-saturation coefficients for the 2 limiting resources: these were related via a tradeoff between the half saturation constants for each resource.

What they found is strikingly similar to the results from Scheffer and van Nes and dissimilar to the the results that emerge when resources are constant. Clumps of coexisting species emerged along the trait axis. When resource fluctuations occurred rapidly, only fairly specialized species survived in these clumps (R* values that were high for either resource 1, or resource 2, rather than intermediate). But when fluctuations were less frequent, clusters of species also survived at intermediate points along the trait axis. However, in all cases the community organized into clumps composed of very similar species that were coexisting (see Figure 3). It appears that this occurs because the fluctuating resources result in the system having non-stationary conditions. That is, similar sets of species can coexist because the system varies between those species' requirements for persistence and growth. 

Fig. 3. "Lumpy species coexistence". The y-axis shows the trait value (here, the R*) of species present under 360 day periodicity of resource supply.  

Using many of the dominant models of competition in ecology, it is clearly possible to explain the coexistence of both similar or dissimilar species. This is true across approaches from the Lotka-Volterra results of Scheffer and van Nes, to Tilman's R* resource competition, to Chessonian coexistence (2000). It provides a unifying expectation upon which further research can build. Perhaps the paradox of the planktons is not really a paradox anymore?

-->

Reference:
Athanasia Sakavara, George Tsirtsis, Daniel L. Roelke, Rebecca Mancy, and Sofie Spatharis. Lumpy species coexistence arises robustly in fluctuating resource environments. PNAS 2017; published ahead of print December 20, 2017, doi:10.1073/pnas.1705944115

More authors, more joy?

It seems that ecologists have been complaining that no one writes single author manuscripts anymore since at least the 1960s. de Solla Price predicted in 1963:

"…the proportion of multi-author papers has accelerated steadily and powerfully, and it is now so large that if it continues at the present rate, by 1980 the single-author paper will be extinct”

Fortunately, an interesting new editorial in the Journal of Applied Ecology has the data (from their archives of published and submitted papers) to evaluate to ask whether this disastrous outcome has actually occurred.

It turns out that Price was wrong about single-author extinction, although he hadn't misread the trends. Since the 1970s, the proportion of single-authored papers at the journal have declined to less than 4% and the mean number of authors has risen to more than 5 (Figure 1).

Fig 1. 

It's also notable that single-authored papers are cited significantly less often and are 2.5x less likely to be accepted (!). (If that statistic doesn't make you want to gather some coauthors, nothing will). These trends agree with others reported in the literature.

The authors hypothesize that a number of factors drive this result. Ecology has gotten 'bigger' in many ways - analyses are less likely to focus on single populations or species and more likely to be replicated through space and or time. This increased breadth requires more students or assistants to aid with experimental or field work, or collaborations with other labs to bring such data together. Similarly, ecological data collection and analyses often require multiple types of specialized knowledge, whether statistical, mathematical, technological, or systems-based. And by relying on multiple researchers to play specialized roles, the overall quality of a manuscript might be higher (as compared to a jack-of-all trades). The authors also suggest that factors including the growing number of ecologists, the more international scope of many research activities, and more democratic approaches to authorship have increased the mean number of co-authors.

What makes these results particularly interesting is that I think there is still something of a cachet for the sole-authored paper. The conceit is that writing a sole authored paper means that you have a fully realized research plan, and you're accomplished enough to bring it to fruition by yourself. But these stats at least seem to suggest that you're better off with a few friends :)


Barlow, Jos, Stephens, Philip A., Bode, Michael, Cadotte, Marc W., Lucas, Kirsty, Newton, Erika, Nuñez, Martin A., Pettorelli, Nathalie. On the extinction of the single-authored paper: The causes and consequences of increasingly collaborative applied ecological research. J Appl Ecol. 55(1): 1365-2664. doi.org/10.1111/1365-2664.13040

Decomposing diversity effects within species

The relationship between biodiversity and ecosystem functioning is so frequently discussed in the ecological literature that it has its own ubiquitous acronym (BEF). The literature has moved from early discussions and disagreements about mechanism, experimental design, and species richness to ask how different components of biodiversity might contribute differentially to functioning. The search is for mechanisms which hopefully will lend predictability to biodiversity-function relationships. One approach is to independently manipulate different facets of biodiversity – whether species, phylogenetic, trait-based, or genetic diversity – to help disentangle the relative contribution of each.

A new paper extends this question by considering how within-species diversity – including genotypic richness, genetic differences, and trait differences – contribute to functioning. Abbott et al. (2017, Ecology) use a field-based eelgrass system to explore how independent manipulations of genotypic richness and genetic relatedness affected biomass production and invertebrate community richness. They collected 41 unique genotypes of eelgrass (Zostera marina), and used 11 species-relevant loci to determine the relatedness of each genotype pair. The authors also measured 17 traits relevant to performance including "growth rate, nutrient uptake, photosynthetic efficiency, phenolic content, susceptibility to herbivores, and detrital production ".

Eelgrass meadow.
From
http://www.centralcoastbiodiversity.org/
eelgrass-bull-zostera-marina.html

Each of these of these measures are inter-related, but not necessarily in clear, predictable fashions. Genotypes likely differ functionally, but some traits and some genotypes will vary more than others. Genetic distances or relatedness between species similarly may be proxies for trait differences, but this depends on the underlying evolutionary processes. The relationship between any of these measures and functions such as biomass production are no doubt varied and dependent on the mechanism.

The authors established plots with two levels of genotypic richness, either 2 genotypes or 6 genotypes, where genotypes varied among the 41 available. Fully crossed with the genotypic richness treatment was a genetic relatedness treatment: genotypes were either more closely related than a random selection, less closely related, or as closely related as random. At the end of the experiment, above and belowground biomass were collected, and epifaunal invertebrates were collected, and modelled as a component of the biodiversity components.

Because of early die-offs in many plots, planted genotype richness differed from final richness greatly (very few plots had 6 genotypes remaining, for example). For that reason, final diversity measures were used in the models. The relationship between aboveground biomass or belowground biomass and biodiversity were similar: both genotypic richness and genotypic evenness were positively related to total final biomass, but genetic relatedness was negatively correlated. That is, plots with more related genotypes were less productive. Other variables such as trait diversity was not as important, and in fact they did not find any relationship between trait differences and degree of genetic relatedness between genotypes. Since relatedness seemed unrelated to functional similarities, between genotypes, the authors suggested that possibly that reduced biomass among related genotypes is due to self-recognition mechanisms. Most interestingly, the best predictors of invertebrate grazer diversity were opposite -  – the best predictor was trait diversity, not genotypic richness or genetic relatedness.

Even in this case, where Abbott et al. were able to separate different diversity components experimentally, it's clear that simplistic predictions as to how they contribute to functioning are insufficient. The contributions of genotypic versus trait diversity were not strongly related. Further, trait diversity performed best on the function for which genotypic diversity performed worst. Understanding what this means is difficult - are the traits relevant for understanding intraspecific interactions (resource usage, etc) so incredibly different from those relevant for interspecific interactions with herbivores? Are the 17 traits too few to capture all differences, or too many irrelevant traits? Do we expect different biodiversity facets have unique independent effects on ecosystem functions, or does the need to consider multiple facets simply mean we have an imperfect understanding of how different facets are related? 

Positive cost-benefit analysis for conservation spending

In a time when most news about human impacts on the Earth's biodiversity seems to be negative, a new paper in Nature provides a glint of good news about our ability to change the current trend of loss. Encouraging new conservation efforts and funding may be contingent on providing evidence that such efforts will actually be effective.

The new report from Waldron et al. (2017) provides evidence for a predictable relationship between conservation spending and reduction of biodiversity loss. They focused on signatory countries of the Earth Summit's Convention on Biological Diversity and Sustainable Development Goals, and developed a pressures-and-conservation-impact’ (PACI) model to predict how biodiversity loss changed in these countries between 1996-2008. Improvements were driven by conservation spending (relativized to reflect differences in buying power between nations) and were counteracted by GDP growth and agricultural expansion. 

Using this model, the authors could predict how the conservation investments made in these nations had affected their loss of biodiversity, as compared to the scenario in which no investment had been made. Amazingly, the median loss of biodiversity per nation was 29% lower than would otherwise have been expected. Over 1996-2008, seven countries even had net biodiversity improvements: Mauritius, Seychelles, Fiji, Samoa, Tonga, Poland and Ukraine.

Fig 1. Map of biodiversity decline scores (BDS) for signatory nations.
"Colours show percentage of all global declines (total BDS) associated with each country. Pie charts show the predicted reduction in decline (in black) if spending had been I$5 million higher (for selected countries); pie size represents the square root of the BDS. Inset shows predicted versus observed BDS (log-transformed) for the continuous model".

They discuss a number of interactions among model terms that capture greater socio-economic complexity - for example, the impacts of GDP growth on biodiversity loss are lower when a country's base GDP is very low. Such large scale studies naturally face data limitations - here, they use mammal and bird Red List status changes to develop a quantitative measure of biodiversity loss. Other taxa presumably show similar trends, but we lack the data to incorporate them at this moment.

Hopefully by demonstrating this cost-benefit analysis for conservation actions, Waldron et al. (2017) encourage future 'investors' as to the payoff of spending on conservation. 

Reductions in global biodiversity loss predicted from conservation spending. Anthony Waldron, Daniel C. Miller, Dave Redding, Arne Mooers, Tyler S. Kuhn,  Nate Nibbelink, J. Timmons Roberts, Joseph A. Tobias & John L. Gittleman. Nature (2017). doi:10.1038/nature24295. 

Blogging about science for yourself

In case you missed it, a new paper in Royal Society Open Science from seven popular ecology blogs discusses the highlights and values of science community blogging. It provides some insights into the motivations behind posting and the reach and impacts that result. It's a must-read if you've considered or already have a blog about science.

It was nice to see how universal the 'pros' of blogging seem to be – the things I most appreciate about contributing to a blog are pretty similar to the things the authors here reported on too. According to the archives, I've been posting here since 2010, when I was a pretty naïve PhD student interacting with the ecological literature for the first time. I had a degree of enthusiasm and wonder upon interacting with ideas for the first time that I miss, actually. I just started a faculty job this fall, and I think that the blog allowed me to explore and experiment with ideas as I figured out where I was going as a scientist (which is still an ongoing process).

As Saunders et al. note, one of the other major upsides to blogging is the extent to which it produces networking and connections with colleagues. In a pretty crowded job market, I think it probably helped me, although only as a complement to the usual suspects (publications, 'fit', research plans, interviewing skills). Saunders et al. also mentioned blogging as relevant to NSF's Broader Impacts section, which I actually hadn't considered. Beyond that, the greatest benefit by far for me is that forcing oneself to post regularly and publicly is amazing practice for writing about science.

Despite these positives, I don't necessarily think a science blog is for everyone and there are definitely things to consider before jumping in to it. It can be hard to justify posting on a blog when your to-do list overflows, and not everyone will –understandably- think that's a good use of their time. There is a time commitment and degree of prioritisation required that is difficult. This is one reason that having co-bloggers can be a lifesaver. It is also true that while writing a blog is great practice, it probably selects for people able to write quickly (and perhaps without perfectionistic tendencies).

When students ask me about blogging, they often hint at concerns in sharing their ideas and writing. It can be really difficult to put your ideas and writing out there (why invite more judgement and criticism?) and this is can feedback with imposter syndrome (speaking from my own experience). For a long time, minorities, women, students have been under-represented in ecology blogs, and I think this may be a contributor to that. It's nice to see more women blogging about these days, and hopefully there is a positive feedback from increasing the visibility of under-represented groups.

In any case, this paper was especially timely for me, because I've been re-evaluating over the past few months about whether to keep blogging or not, and this provided a reminder of the positive impacts that are easy to overlook.

When do descriptive methods exceed the sum of their points?

The last post here mused on the connection between (but also, distinctness of) the scientific goals of "understanding" and "prediction". An additional goal of science is "description", the attempt to define and classify phenomenon. Much as understanding and prediction are distinct but interconnected, it can be difficult to separate research activities between description and understanding. Descriptive research is frequently considered preliminary or incomplete on its own, meant to be an initial step prior to further analysis. (On the other hand, the decline of more descriptive approaches such as natural history is often bemoaned). With that in mind, it was interesting to see several recent papers in high-impact journals that rely primarily on descriptive methods (especially ordinations) to provide generalizations. It's fairly uncommon to see ordination plots as the key figure in journals like Nature or The American Naturalist, and it opens up the question of 'when do descriptive methods exceed description and provide new insights & understanding?'

For example, Diaz et al.'s 2016 Nature paper took advantage of a massive database of trait data (from ~46000 species) to explore the inter-relationships between 6 ecologically relevant plant traits. The resulting PCA plot (figure below) illustrates, across many species, that well-known tradeoffs between a) organ size and scaling and b) the tissue economic spectrum appear fairly universal. Variation in plant form and function may be huge, but the Diaz et al. ordination highlights that it still is relatively constrained, and that many strategies (trait combinations) are apparently untenable.

From Diaz et al. 2016.

Similarly, a new paper in The American Naturalist relies on ordination methods  to try to identify 'a periodic table of niches' of lizards (Winemiller et al. 2015 first presented this idea) – i.e. a classification framework capturing the minimal, clarifying set of universal positions taken by a set of taxa. Using the data and expert knowledge on lizard species collected over a lifetime of research by E. Pianka and L. Vitt, Pianka et al. (2017) first determine the most important life history axes -- habitat, diet, life history, metabolism, and defense attributes. They use PCoA to calculate the position of each of 134 species in terms of each of the 5 life history axes, and then combined the separate axes into a single ordination (see figure below). This ordination highlights that niche convergence (distant relatives occupy very similar niche space) and niche conservation (close relatives occupy very similar niche space) are both common outcomes of evolution. (For more discussion, this piece from Jonathon Losos is a great). Their results are less clarifying than those in Diaz et al. (2016): a key reason may simply be the smaller size of Pianka et al.'s data set and its greater reliance on descriptive (rather than quantitative) traits.

From Winemiller et al. 2017

Finally, a new TREE paper from Daru et al. (In press) attempts to identify some of the processes underlying the formation of regional assemblages (what they call phylogenetic regionalization, e.g. distinct phylogenetically delimited biogeographic units). They similarly rely on ordinations to take measurements of phylogenetic turnover and then identify clusters of phylogenetically similar sites. Daru et al.'s paper is slightly different, in that rather than presenting insights from descriptive methods, it provides a descriptive method that they feel will lead to such insights.

Part of this blip of descriptive results and methods may be related to a general return to the concept of multidimensional or hypervolume niche (e.g. 1, 2). Models are much more difficult in this context and so description is a reasonable starting point. In addition, the most useful descriptive approaches are like those seen here - where new data or a lot of data (or new techniques that can transform existing data) - are available. In these cases, they provide a route to identifying generalization. (This also leads to an interesting question – are these kind of analyses simply brute force solutions to generalization? Or do descriptive results sometimes exceed the sum of their individual data points?)

References:

Díaz S, Kattge J, Cornelissen JH, Wright IJ, Lavorel S, Dray S, Reu B, Kleyer M, Wirth C, Prentice IC, Garnier E. (2016). The global spectrum of plant form and function. Nature. 529(7585):167.

Eric R. Pianka, Laurie J. Vitt, Nicolás Pelegrin, Daniel B. Fitzgerald, and Kirk O.Winemiller. (2017). Toward a Periodic Table of Niches, or Exploring the Lizard Niche Hypervolume. The American Naturalist. https://doi.org/10.1086/693781

Barnabas H. Daru, Tammy L. Elliott,  Daniel S. Park, T. Jonathan Davies. (In press). Understanding the Processes Underpinning Patterns of Phylogenetic Regionalization. TREE. DOI: http://dx.doi.org/10.1016/j.tree.2017.08.013

Why is prediction not a priority in ecology?

When we learn about the scientific method, the focus is usually on hypothesis testing and deductive reasoning. Less time is spent on considering the various the outcomes of scientific research, specifically: description, understanding, and prediction. Description involves parsimoniously capturing data structure, and may use statistical methods such as PCA to reduce data complexity and identify important axes of variation. Understanding involves the explanation of phenomenon by identifying causal relationships (such as via parameter estimation in models). Finally, prediction involves estimating the values of new or future observations. Naturally, some approaches in ecology orient more closely toward one of these outcomes than others and some areas of research historically have valued one outcome over others. For example, applied approaches such as fisheries population models emphasize predictive accuracy (but even there, there are worries about limits on prediction). On the other hand, studies of biotic interactions or trophic structure typically emphasize identifying causal relationships. The focus in different subdisciplines no doubt owes something to culture and historical priority effects.

In various ways these outcomes feedback on each other – description can inform explanatory models, and explanatory models can be evaluated based on their predictions. In a recent paper in Oikos, Houlahan et al. discuss the tendency of many ecological fields to under-emphasize predictive approaches and instead focus on explanatory statistical models. They note that prediction is rarely at the centre of ecological research and that this may be limiting ecological progress. There are lots of interesting questions that ecologists should be asking, including what are the predictive horizons (spatial and temporal scales) over which predictive accuracy decays? Currently, we don't even know what a typical upper limit on model predictive ability is in ecology.

Although the authors argue for the primacy of prediction ["Prediction is the only way to demonstrate scientific understanding", and "any potentially useful model must make predictions about some unknown state of the natural world"], I think there is some nuance to be gained by recognizing that understanding and prediction are separate outcomes and that their relationship is not always straightforward (for a thorough discussion see Shmueli 2010). Ideally, a mutually informative feedback between explanation and prediction should exist, but it is also true that prediction can be useful and worthy for reasons that are not dependent on explanation and vice versa. Further, to understand why and where prediction is limited or difficult, and what is required to correct this, it is useful to consider it separately from explanation.

Understanding/explanation can be valuable and inspire further research, even if prediction is impossible. The goal of explanatory models is to have the model [e.g., f(x)] match as closely as possible the actual mechanism [F(x)]. A divergence between understanding and prediction can naturally occur when there is a difference between concepts or theoretical constructs and our ability to measure them. In physics, theories explaining phenomenon may arise many years before they can actually be tested (e.g. gravitational waves). Even if useful causal models are available, limitations on prediction can be present: in particle physics, the Heisenberg uncertainty principle identifies limits on the precision at which you can know both the position of a particle and its momentum. In ecology, a major limitation to prediction may simply be data availability. In a similar field (meteorology) in which many processes are important and nonlinearities common, predictions require massive data inputs (frequently collected over near continuous time) and models that can be evaluated only via supercomputers. We rarely collect biotic data at those scales in ecology. We can still gain understanding if predictions are impossible, and hopefully eventually the desire to make predictions will motivate the development of new methods or data collection. In many ecological fields, it might be worth thinking about what can be done in the future to enable predictions, even if they aren't really possible right now.

Approaches that emphasize prediction frequently improve understanding, but this is not necessarily true either. Statistically, understanding can come at the cost of predictive ability. Further, a predictive model may provide accurate predictions, but do so using collinear or synthetic variables that are hard to interpret. For example, a macroecological relationship between temperature and diversity may effectively predict diversity in a new habitat, and yet do little on its own to identify specific mechanisms. Prediction does not require interpretability or explanatory ability, as is clear from papers such as "Model-free forecasting outperforms the correct mechanistic model for simulated and experimental data". So it's worth being wary of the idea that a predictive model is necessarily 'better'.

With this difference between prediction and understanding in mind, it is perhaps easier to understand why ecologists have lagged in prediction. For a long time, statistical approaches used in ecology were biased toward those meant to improve understanding, such as regression models, where parameters estimate the strength and direction of a relationship. This is partially responsible for our obsession with p-values and R^2 terms. What Houlahan et al. do a great job of emphasizing is that by ignoring prediction as a goal, researchers are often limiting their ability confirm their understanding. Predictions that are derived from explanatory models Some approaches in ecology have already moved naturally towards emphasizing prediction, especially SDMs/ecological niche models. They recognized that it was not enough to describe species-environment relationships; testing predictions allowed them to determine how universal and mechanistic these relationships actually were. A number of macroecological models fit nicely with predictive statistical approaches, and could adopt Houlahan’s suggestions quite readily (e.g. reporting measures of predictive ability and testing models on withheld data). But for some approaches, the search for mechanism is so deeply integrated into how they approach science that it will take longer and be more difficult (but not impossible)*. Even for these areas, prediction is a worthy goal, just not necessarily an easy one. 

*I was asked for examples of 'unpredictable' areas of ecology. This may be pessimistic, but I think that something like accurately predicting the composition (both species' abundance and identity) of diverse communities at small spatial scales might always be difficult, especially given the temporal dynamics. But I could be wrong! 

...if the Simpsons could predict Trump, I suppose there's hope for ecologists too...

**This has been edited to correctly spell the author's name.

Novel habitat, predictable responses: niche breadth evolution in geckos

At a time of immense ecological change (such as the Anthropocene), organisms have a few options. They can move, tolerate, adapt, or, in failing to do so, face extinction. One or most of those options may not be available to most species. For example, the question of whether most species can adapt rapidly enough to maintain populations in degrading habitats, rising temperatures and increasing environmental variability has (at least in part) motivated the study of rapid or contemporary evolution. Studying the probability of successful selection and adaptation over ecological timescales may be very important for understanding the options available to species.

de Amorim et al. (2017, PNAS) describe one such example, where the result of novel environmental change provides a unique opportunity to observe rapid evolution. Beginning in 1996, a reservoir in Central Brazil was created by flooding a huge area, creating nearly 300 islands and massively affecting local wildlife. Gymnodactylus amarali was the most common lizard (a termite-specialized gecko), and the authors sought to determine the impacts of rapid isolation on the species.

Isolation on islands created an new set of biotic conditions – other termite eating lizards went extinct on islands, increasing the available diet breadth, particularly increasing the availability of larger termites. Larger termites require geckos have the physical ability to catch and processes them. One possibility is that to take advantage of this new resource, G. amarali on islands would need larger heads. Because larger heads and bodies come with increased energy requirements, the authors predicted that the island geckos would have larger heads, but no change in overall body size.

Termite size increased on average on islands; for the same body size, head length tended to be larger on islands. 

Indeed, island geckos had higher diet breadths, driven by the availability of larger termites and an increased ability to catch them via larger head lengths. Increased diet breadth was accompanied by increased head size, but not body size.

Notably, this change in diet and associated characters occurred independently across multiple reservoir islands, beginning once they were isolated from the mainland. This is an interesting example of rapid evolution precisely because evolution took the same path in every case, and because it occurred so rapidly (less than 15 years). This is not always the expectation - in many cases, human activities (e.g. fragmentation) will increase decrease population sizes and genetic diversity, thereby increasing drift and decreasing the predictability (and speed) and adaptation. Contrasts between successful and unsuccessful adaptive responses will help us understand better how and when fragmentation threatens populations.

Mariana Eloy de Amorim, Thomas W. Schoener, Guilherme Ramalho Chagas Cataldi Santoro, Anna Carolina Ramalho Lins, Jonah Piovia-Scott, and Reuber Albuquerque Brandão. 2017. Lizards on newly created islands independently and rapidly adapt in morphology and diet. PNAS. 114 (33) 8812-8816.

Experimental macroevolution at microscales

Sometimes I find myself defending the value of microcosms and model organisms for ecological research. Research systems do not always have to involve a perfect mimicry of nature to provide useful information. A new paper in Evolution is a great example of how microcosms provide information that may not be accessible in any other system, making them a valuable tool in ecological research.

For example, macroevolutionary hypotheses are generally only testable using observational data. They suffer from the obvious problem that they generally relate to processes of speciation and extinction that occurred millions of years ago. The exception is the case of short generation, fast evolving microcosms, in which experimental macroevolution is actually possible. Which makes them really cool :-) In a new paper, Jiaqui Tan, Xian Yang and Lin Jiang showing that “Species ecological similarity modulates the importance of colonization history for adaptive radiation”. The question of how ecological factors such as competition and predation impact evolutionary processes such as the rapid diversification of a lineage (adaptive radiation) is an important one, but generally difficult to address (Nuismer & Harmon, 2015; Gillespie, 2004). Species that arrive to a new site will experience particular abiotic and biotic conditions that in turn may alter the likelihood that adaptive radiation will occur. Potentially, arriving early—before competitors are present—could maximize opportunities for usage of niche space and so allow adaptive radiation. Arriving later, once competitors are established, might suppress adaptive radiation.

More realistically, arrival order will interact with resident composition, and so the effects of arriving earlier or later are modified by the identities of the other species present in a site. After all, competitors may use similar resources, and compete less, or have greater resource usage and so compete more. Although hypotheses regarding adaptive radiation are often phrased in terms of a vague ‘niche space’, they might better be phrased in terms of niche differences and fitness differences. Under such a framework, simply having species present or not present at a site does not provide information about the amount of niche overlap. Using coexistence theory, Tan et al. produced a set of hypotheses predicting when adaptive radiation should be expected, given the biotic composition of the site (Figure below). In particular, they predicted that colonization history (order of arrival) would be less important in cases where species present interacted very little. Equally, when species had large fitness differences, they predicted that one species would suppress the other, and the order in which they arrived would be immaterial. ­

From Tan et al. 2017

The authors tested this using a bacterial microcosm with 6 bacterial competitors and a focal species – Pseudomonas fluorescens SBW25. SBW25 is known for its rapid evolution, which can produce genetically distinct phenotypes. Microcosm patches contained 2 species, SBW25 and one competitor species, and their order of arrival was varied. After 12 days, the phenotypic richness of SBW25 was measured in all replicates.

From Tan et al. 2017. Competitor order of arrival in general altered the final phenotypic richness of SBW25.

Both order of arrival and the identity of the competitor did indeed matter as predictors of final phenotypic richness (i.e. adaptive radiation) of SBW25. Further, these two variables interacted to significantly. Arrival order was most important when the 2 species were strong competitors (similar niche and fitness differences), in which case late arrival of SBW25 suppressed its radiation. On the other hand, when species interact weakly, arrival order had little affect on radiation. The effect of different interactions were not entirely simple, but particularly interesting to me was that fitness differences, rather than niche differences, often had important effects (see Figure below). The move away from considering the adaptive radiation hypothesis in terms of niche space, and restating it more precisely, here allowed important insights into the underlying mechanisms. Especially as researchers are developing more complex models of macroevolution, which incorporate factors such as evolution, having this kind of data available to inform them is really important.

Interaction between final phenotype richness and arrival order for B) niche differences and D) fitness differences. S-C refers to arrival of SWB25 first, C-S refers to its later arrival. 

Problems with over-generalizing the dynamics of communities

Community ecologists talk about communities as experiencing particular processes in a rather general way. We fall into rather Clementsian language, asking whether environmental filtering dominates a community or if biotic interactions are disproportionately strong. This is in contrast to the typical theoretical focus on pairwise interactions, as it acts as though all species in a community are responding similarly to similar processes.

Some approaches to community ecology have eschewed this generality, particularly those that focus on ecological ‘strategies’ differentiating between species. For example, Grimes argued that species in a community represented a tradeoff between three potential strategies - competitive, stress-tolerant, and ruderal (CRS). Other related work describes rarity as the outcome of very strong density-dependence. The core-transient approach to understanding communities differentiates between core species, which have deterministic dynamics tied to the mean local environment, in contrast to transient species which are decoupled from local environmental conditions and have dynamics are driven by stochastic events (immigration, environmental fluctuations, source-sink dynamics). Assuming environmental stationarity, core species will have predictable and consistent abundances through time, in comparison to transient species.

If species do respond differently to different processes, then attempting to analyse all members of a community in the same way and in relation to the same processes will be less informative. Tests for environment-trait relationships to understand community composition will be weaker, since the species present in a community do not equally reflect the environmental conditions. In “A core-transient framework for trait-based community ecology: an example from a tropical tree seedling community”, Umana et al (2017) ask whether differentiating between core and transient species can improve trait-based analyses. They analyse tropical forest communities in Yunnan, China, predicting that core species "will have strong trait–environment relationships that increase the growth rates and probability of survival that will lead to greater reproductive success, population persistence and abundance".

The data for this test came from 218 1 m2 seedling plots, which differed in soil and light availability. The authors estimated the performance of individual seedlings in terms of relative growth rate (RGR). They also gathered eight traits related to biomass accumulation, and stem, root and leaf organ characteristics. They were particularly interested in how the RGR of any individual seedling differed from the mean expectation for their species. Did this RGR deviation relate to environmental differences between sites?  If a species’ presence is strongly influenced by the environment, then RGR deviation should vary predictably based on environmental conditions.

They then modelled RGR deviation as a function of the traits or environmental conditions (PCA axes). They considered various approaches for binning species based on commonness vs. rarity, but the general result was that bins containing rarer species had fewer PCA axes significantly associated with their RGR deviation and/or those relationships were weaker (e.g. see Figure below).

They conclude  that “the main results of our study show that the strength of demography-environment/trait and trait-environment relationships is not consistent across species in a community and the strength of these effects is related to abundance”. Note that other studies similarly find variation in the apparent mechanism of coexistence in communities. For example, Kraft et al. 2015  found that local fitness and niche differences only predict coexistence for a fraction of species co-occurring in their sites.

Umana et al.'s result is a reminder that work looking for general processes at the community level may be misleading. It isn't clear that there is a good reason to divide species into only two categories (e.g. core versus transients): like unhappy families, transient species may each be transient in their own way.

Progress on biodiversity-ecosystem function requires looking back

Williams, L. J., et al. 2017. Spatial complementarity in tree crowns explains overyielding in species mixtures. - Nature Ecology & Evolution 1: 0063.

It seems at times that the focus on whether biodiversity has a positive relationship with ecosystem functioning has been a bit limiting. Questions about the BEF relationships are important, of course, since they support arguments for protecting biodiversity and suggests a cost of failing to do so. But as a hypothesis ('higher diversity is associated with higher functioning'), they can be rather one-dimensional. It's easy to think of situations in which other types of BEF relationships (neutral, negative) exist. So is it enough to ask if positive BEF relationships exist?

It’s nice then that there is increasingly a focus on identifying mechanisms behind BEF relationships, using both theory and empirical research. A new paper along these lines is “Spatial complementarity in tree crowns explains overyielding in species mixtures” from Laura Williams et al. (2017). "Overyielding" is the phenomenon in which greater total biomass is produced in a mixture of species compared to the expectation based on their biomass production in monoculture. Overyielding would suggest a benefit in maintaining polycultures, rather than having monocultures, and is a common response variable in BEF studies.

This study focused on the production of stem biomass in monocultures vs. polycultures of forest trees. Experimental communities of young tree species were planted with orthogonal gradients of species richness and functional richness, allowing the effects of species number and trait diversity to be disentangled. Complementarity in tree canopy structure in these communities may be an important predictor of overyielding in stem biomass. Complementarity among tree crowns (that is, the extent to which they fit together spatially without overlapping, see Fig below) should reflect the ability of a set of species to maximize the efficiency of light usage as it hits the canopy. Such variation in crown canopy shapes among species could lead to a positive effect of having multiple species present in a community. 

Example of crown complementarity.
From Williams et al. 2017.

To test this, the authors estimated crown architecture for each species using traits that reflect crown shape and size. These measures were used to predict the spatial complementarity expected with different combinations of tree species. In addition, a single integrative trait – maximum growth rate – was measured for each species. The authors hypothesized that the variation in growth rate of species in a community would be associated with variation in crown heights and so also a good predictor of overyielding.

They found that crown complementarity occurred in nearly all of the experimental polycultures and on average was 29% greater in mixtures than monocultures. Controlling for the number of species, communities with greater variation in growth rate did in fact have greater crown complementarity, as predicted. Further, higher levels of crown complementarity were strongly associated (R2~0.6) with stem biomass overyielding.

Fig 2&3 from Williams et al (2017). For experimental communities:
a) the relationship between crown complementarity and variation in growth rate.
b) the relationship between crown complementarity and stem biomass overyielding.

These results provide a clear potential mechanism for a positive effect of biodiversity (particularly trait-based variation) in similar forests. (As they state, "We posit that crown complementarity is an important mechanism that may contribute to diversity-enhanced productivity in forests"). Given the importance of the sun as a limiting resource in forests, the finding that mixing species that combining shade intolerant and shade tolerant strategies are more productive (the authors note that "growth rate aligns with shade tolerance and traits indicative of a tree’s resource strategy") is not necessarily surprising. It fits within existing forestry models and practices for mixed stands. This is a reminder that we already understand many of the basic components of positive (and neutral and negative) diversity-functioning relationships. The good news is that ecology has accumulated a large body of literature on the components of overyielding (limiting resources, niche partitioning, evolution of alternate adaptive strategies, constraints on these, the strength of competition, etc). From the literature, we can identify the strongest mechanisms of niche partitioning and identify the contexts in which these are likely to be relevant. For example, sun in forests and canopy complementarity, or water limitation in grasslands and so root complementarity might be a good focal trait.