New ways to address an old idea: rethinking the regional species pool

Like many concepts in ecology (metacommunity, community), the idea of a regional species pool is useful, makes conceptual sense, and is incredibly difficult to apply to real data. Originally, the idea of a species pool came from the theory of island biogeography (MacArthur and Wilson, 1967), where it referred to all the species that could disperse to an island. Today, the regional species pool appears frequently, across null models, studies of community assembly both empirical and theoretical, and metacommunity theory. 

Understanding how particular processes shape community membership—whether the environmental, competition, or dispersal limitation—depends on knowing the identity of all the species that could have potentially assembled there. The species pool as defined by the research provides the frame of reference against which to consider a community's composition. Most null models of community assembly rely on correctly identifying this set of species, and worse, tend to be very sensitive to bias in how the regional pool is defined. If you include all species physically present in a region, in your species pool, environmental filtering may appear to be particularly important simply because many of those species can’t actually survive in your community (the narcissus effect). Given the importance of null models to community ecology, defining the species pool appropriately is an ongoing concern.

There are many decisions that can be made when asking 'which species could potentially be members of a community'? You could include all species that can physically arrive at a site (so only dispersal or geographic distance limits membership), or only include those species that can both arrive and establish (both dispersal and environmental conditions limit membership). Further, the availability of data is key: if you use observational data used to determine the environmental limitations, you may also incorporate the outcome of biotic interactions indirectly. If some species are rare and have low observation likelihoods, they will be under-represented. Abundances may be useful but inaccurate depending on how they are measured. Finally, it is common to define species as either present or not present for a species pool; this binary approach may conceal ecologically important information.

The 'filtering' heuristic for understanding community membership. Species groups 1-3 could each be defined as a regional species pool, depending on the definition applied.

A number of recent papers provide alternative approaches to constructing species pools, meant to avoid these pitfalls. Researchers can define multiple contrasting species pools, each pool representing an ecological process (or perhaps multiple processes) of interest. Each species pool can be modified further to reflect the strength of a particular process in constraining membership. The regional pool is not seen as a single entity but as a number of possible configurations whose utility is in comparison.

Lessard et al. (2016) illustrates how to produce this kind of process-based species pool with various constraints (figure below). Their three-step approach is to:

1. Define absolutely all possible members of regional pool. This is determined by identifying all assemblages in the region containing at least one species also found in the focal community (creating a 'dispersion field') (figure below, section A). This delineates a large region and identifies all species within it.
2. Calculate the probability of resampling a species from the focal community elsewhere in the dispersion field. This is done in the context of the process of interest. For example, the probability of observing a species in the focal community and another community might be determined based on the geographical or environmental distance between those sites. Every site in the dispersion field would now have a probability (or distance really) associated with it, representing some similarity with the focal site.
3. Finally, apply constraints to the calculated probabilities. You might choose to consider only the species within communities that are at least 50% similar to the focal community, for example. Such constraints would reflect the strength or importance of filtering by the process of interest.

Another recent paper (Karger et al., 2016) takes an approach with a number of commonalities to the Lessard et al. method. However, rather than resampling to produce potential pools of species (with species being defined as present or absent), they advocate a probabilistic approach to species pools. They suggest that species pools should be thought of as a set of probabilities of membership, which may be more reflective of ecological reality. In some ways, this is a simply a formalization of probabilistic sampling from Lessard, but instead of applying constraints, the researcher acknowledges that probabilities vary for different species. “Hence, a species pool can simply be defined as a function of probabilities of a species’ occurrence in the focal unit given the unit’s environmental and biotic conditions, geographical location and the time frame of interest”.

Both comparative and probabilistic approaches to defining species are logical advances, and one way of dealing with the untidy concept of the species pool. If this topic is of interest, a few other papers, albeit slightly less recent, are definitely worth reading: Pigot and Etienne 2015; Lessard et al. 2012, Carsten et al., 2013.

From Lessard et al., 2016. The three steps to build a species pool.

Coexistence theory: community assembly's next great hope?

Rethinking Community Assembly through the Lens of Coexistence Theory
J. HilleRisLambers, P.B. Adler, W.S. Harpole, J.M. Levine, and M.M. Mayfield

The big (literally, at 24 pages) paper to read this year is a review by a number of well-known community ecologists that aims to package years of often contradictory and confusing results from community assembly research (Weiher & Keddy 2001) into a manageable package using coexistence theory. Coexistence theory arose particularly out of Peter Chesson’s work (particularly his own annual review paper (Chesson 2000)), and rests in the idea that coexistence between species is the result of a balance of stabilizing forces (i.e. niche differences) and equalizing forces (i.e. fitness similarity) between those species. Coexistence is stable when stabilizing forces dominate, so a species competes more strongly with itself than with other, more dissimilar, species. The most successful adaptations of this framework to “real world” experiments have come from Jonathan Levine’s lab (in collaboration with many of the coauthors on this work). Indeed, there are probably few people more qualified to attempt to re-explain the often complicated findings in community assembly research using coexistence theory.

The classic heuristic model for community assembly involves a regional species pool that is consecutively filtered through environmental and then biotic filters, selecting only for those species adapted to the local environment. While logically appealing, this model may have constrained thinking about assembly: after all, our definition of a niche recognizes that species are impacted by and impact their environments (Chase & Leibold 2003), and unlike a expectations for a biotic "filter", arrival order can alter the outcome of biotic interactions. But does coexistence theory do a better job of capturing these dynamics? 

The important message to take from coexistence theory, the authors suggest, is that stabilizing niche differences facilitate coexistence, whereas relative fitness differences drive competitive exclusion. And although this yields predictions about how similar or different coexisting species should be, coexistence theory diverges in a number of ways from trait-based or phylogenetic approaches to community assembly. “First, competitive exclusion can either preferentially eliminate taxa that are too functionally similar when trait differences function as stabilizing niche differences or preferentially eliminate all taxa that do not possess the near optimal trait when such trait differences translate into fitness differences. Second, both stabilizing niche differences and relative fitness differences are influenced by abiotic and biotic factors. For both reasons, patterns of trait dissimilarity or similarity cannot easily be used to infer the relative importance of environmental versus biotic (competitive) filters, which is an important goal of community assembly studies.”

There are a number of ways in which pre-existing research might provide evidence for the predictions of coexistence theory. You can look at studies which modify fitness differences between species (for example, through nutrient addition experiments), those which modify niche differences (for example, manipulating colonization differences between species), and those which manipulate the types of species competing to establish. You can take advantage of trait or phylogenetic information about communities (and traits are valuable because they provide a mechanistic linkage), although Mayfield and Levine (2010) have already shown there are clear limitations to such approaches. A particularly useful approach may be to look at demographic rates, particularly looking for frequency-dependent growth rates, an indicator of niche differences between species – when niche differences are large, species should have higher growth rates at low density (lower intraspecific competition) than at high density. And indeed, there is some evidence for the effect of fitness differences or niche differences on community composition.

Ultimately reanalyzing old research has its limitations: is it possible that nutrient additions leading to changes in community structure are evidence of fitness differences? Yes. Are there other possible explanations? Yes. Convincing evidence will take new studies, and the authors make some excellent  suggestions to this end: that we need to combine demographic and trait-based approaches so that assembly studies results suggest at mechanisms, not patterns. The focus would be on correlating niche and fitness differences with traits, rather than correlating traits with species’ presence or absence in the community. 

Given the muddle that is community assembly research, a review that offers a new approach is always timely, and this one is very comprehensive and sure to be well cited. Strangely, for me this paper perhaps lacked the moment of insight I felt when I read about coexistence theory being applied to invasive species (MacDougall et al 2009) or phylogenetic analyses of communities (Mayfield and Levine, 2010). There are a few reasons why that might be: one is that there are difficulties that are not well explored, particularly that traits may not realistically be able to be categorized in an either-niche-or-fitness fashion, and that abiotic and biotic factors can interact with traits. The predictions this framework makes for community assembly are less clear: even the tidiness of coexistence theory can't escape the complications of community assembly. But perhaps that is a pessimistic take on community assembly. Regardless, the paper has a lot to offer researchers and will hopefully encourage new work exploring the role of niche and fitness differences in community assembly.

A good null model is hard to find

de Bello, F. The quest for trait convergence and divergence in community assembly:are null-models the magic wand? Global Ecology and Biogeography, (Global Ecol.Biogeogr.) (2012) 21, 312–317

Ecologists have always found the question of how communities assemble to be of great interest. However, studies of community assembly are often thwarted by the large temporal and spatial scales over which processes occur, making experimental tests of assembly theory difficult. As a result, researchers are often forced to rely on observational data and make inferences about the mechanisms at play from patterns alone. While historical assembly research focused on inferring evidence of competition or environmental filtering from patterns of species co-occurrence, more recent research often looks at patterns of phylogenetic or trait similarity in a community to answer these questions. 

Not surprisingly, when methods rely heavily on observational data they are open to criticism: one of the most important outcomes of early community assembly literature was the recognition that patterns that appeared to support a hypothesis about competition or environmental filtering could in fact result by random chance. This ultimately lead to the widespread incorporation of null models, which are meant to simulate patterns that might be observed by random chance (or other processes not under study), against which the observed data can be compared. Patterns of functional and phylogenetic information in communities can also be compared against null expectations to ensure that patterns of phylogenetic or functional over- or under-dispersion can't arise due to chance alone. However, while null models are an important tool in assembly research, they are sometimes as the foolproof solution to all of its problems.

In a new paper by Francesco de Bello, the author states frankly “whilst reading null-model methods applied in the literature (indeed including my work), one may have the impression of reading a book of magic spells”. While null models are increasingly sophisticated, allowing researchers to determine which processes to control for and which to leave out, de Bello suggests that the decision to include or omit particular factors from a null model can be unclear, making it difficult to interpret results or compare results across studies. Further, results from null models may not mean what researchers expect them to mean.

Using the example of functional diversity (FD; variation in trait values among species in a community), de Bellow illustrates how null models may have different meanings than expected. Ideally, a null model for FD should produce random values of FD, against which the observed values of FD can be compared. Interpreting the difference between the observed and random results can be done using the standardized effect size (SES, the standardized difference between the observed and randomly generated FD values); SES values >0 show that traits are more divergent than expected by chance, suggesting competition structures communities. If SES<0, traits are more convergent than expected by chance, suggesting environmental conditions structure communities. Finally, if SES ~0, then trait values aren’t different from random. However, de Bello shows that the SES is driven by the observed FD values, because the ‘random’ FD values are dependent on the pool of observations sampled. This means that the values the null model produces are ultimately dependent on those observed values, despite the fact you plan to make inferences by comparing the null and observed values as though they are independent. For example, consider the situation where you are building a null model of community structure for plant communities found along two vegetation belts. If the null model is constructed using all the plant communities, regardless of the habitat they are found in, the resulting null FD value will be higher, since species that are dissimilar and found in different vegetation belts are being randomly selected as occurring in a community. If null models are constructed separately for both vegetation belts, the null FD value is lower, since species are more similar. The magnitude of the difference between the null model and the observed values, and further, the biological conclusions one would take from this study, would therefore depend on which null model was constructed.

from de Bello 2012, illustrating how combining species pools (right) can lead to entirely different decisions about whether communities are convergent or divergent in terms of traits than when they are considered separately (left, centre).

De Bello’s findings make important points about the limitations of null models, particularly for functional diversity, but likely for other types of response variable. The type of null model he explores is relatively simplistic (reshuffling of species among sites), and the suggestion that the species pool affects the null model is not unique (Shipley & Weiher, 1995). However, even sophisticated and complex null models need to be biologically relevant and interpretable, and null models are still frequently used incorrectly. Although only mentioned briefly, De Bello also notes another problem with studies of community assembly, which is that popular indices like FD, PD, and others may not always be able to distinguish correctly between different assembly mechanisms (Mouchet et al. 2010, Mayfield & Levine, 2010), something that null model do not control for. 

Researcher spotlight: Tadashi Fukami

Increasingly, ecological explanations for extant community patterns are relying on dynamics operating across multiple spatial and temporal scales, linking small and large scales, and the here and now with evolutionary history. The traditional boundaries of sub-disciplines are blurring. I think that few other young scientists straddle these boundaries as successfully as Tad Fukami, and new assistant professor in the Department of Biology at Stanford University. Tad uses a broad array of theoretical and experimental approaches to understand how ecological communities are put together. From laboratory microcosms to rat-infested islands, and from the computer to remote locations, he is able to pull together disparate pieces of information into a central narrative about the assembly of communities.

I asked him why the question of community assembly interested him so much, and he gives much credit to his advisors, Jim Drake (also my PhD advisor) and Dan Simberloff both in the Department of Ecology and Evolutionary Biology at the University of Tennessee. But more than this, he says that:

“you need to look into the historical background of species interactions to understand the apparently inexplicable variation in the way species interact and the way communities are structured by the interactions.”

and certain aspects of this research obviously excite him. He goes on to say:

“One particularly intriguing thing is the great effect that small chance events that cause variation in early immigration history can have on long-term community development.”

Most ecologists gain their expertise by coming to understand and appreciate the details and intricacy of particular organisms or ecosystems. But Tad is especially noted for his use of an amazingly broad assemblage of systems and methods. I asked him why he used so many different systems, and how he chose those to test his ideas. He said that his work has benefitted from many exciting collaborations and that he has:

“been very lucky to meet many great people who have expertise on specific organisms and systems that a person with diffuse interests like me doesn't have.”

But I think that there may be something deeper and more reassuring. That is, the fact that one could study a multitude of systems, testing the basic dynamics of community assembly, means that there are regularities in how communities are assembled. That you can study stochastic historical events in bacterial microcosms and inform your understanding of plant succession means that while we individually take on these, at times, daunting research projects, our collective understanding of ecological processes are threaded together in a great fabric. And no one is a microcosm of this more than Tad Fukami.

Key recent papers

Fukami, T., Beaumont, H. J. E., Zhang, X.-X. & Rainey, P. B. (2007) Immigration history controls diversification in experimental adaptive radiation. Nature 446: 436-439.

Fukami, T., Wardle, D. A., Bellingham, P. J., Mulder, C. P. H., Towns, D. R., Yeates, G. W., Bonner, K. I., Durrett, M. S., Grant-Hoffman, M. N. & Williamson, W. M. (2006) Above- and below-ground impacts of introduced predators in seabird-dominated island ecosystems. Ecology Letters 9: 1299-1307.

Fukami, T., Bezemer, T. M., Mortimer, S. R. & Van der Putten, W. H. (2005) Species divergence and trait convergence in experimental plant community assembly. Ecology Letters 8: 1283-1290.

Whence diversity?

It is a truism to say that ecological communities are diverse. They often contain dozens or hundreds or thousands of species that represent many of the deep origins in the tree of life. A recent paper by Prinzing and colleagues published in Ecology Letters tested the hypothesis that communities of plants that include more of the ancient divergences from the evolutionary tree of plants should also contain a greater diversity of physical traits. They examined over 9000 plant communities and found that those that contain fewer evolutionary lineages actually had greater trait diversity than those randomly assembled from more lineages. This result reveals that when communities are assembled from a few lineages (likely due to strong environmental selection -e.g., drought tolerance) those members tended to have evolved large differences. That is, while species may be constrained to certain habitat types due to their evolutionary heritage, successful coexistence depends on maximizing differences.
Andreas Prinzing, Reineke Reiffers, Wim G. Braakhekke, Stephan M. Hennekens, Oliver Tackenberg, Wim A. Ozinga, Joop H. J. Schamine, Jan M. van Groenendael (2008). Less lineages more trait variation: phylogenetically clustered plant communities are functionally more diverse Ecology Letters, 11 (8), 809-819 DOI: 10.1111/j.1461-0248.2008.01189.x