Conclusion: Where To From Here?
Conclusion: Where To From Here?
Abstract and Keywords
This book has explored the major methods and concepts in the field of ecophylogenetics. It has considered many of the common statistics and metrics used by ecologists when testing ecophylogenetic hypotheses at both small and large scales. The power of this approach is predicated on the assumption that phylogeny provides information on the evolutionary history of traits that cannot be inferred simply from community data. This concluding chapter reviews some of the advances that have been made in terms of predicting ecology from evolutionary patterns, combining trait and phylogenetic information, and developing a more predictive science of climate change and the biology of species invasions. It also discusses several possible trajectories for ecophylogenetic analyses in the future that will be important in moving the field forward. Finally, it looks at the benefits of a synthesis between ecology and evolution.
We can learn from history on condition that we understand it is history.
Throughout the chapters in this book we have explored the major methods and concepts in the field of ecophylogenetics. We have reviewed many of the common statistics and metrics adopted by ecologists when testing ecophylogenetic hypotheses at both small and large scales. The power of this approach is predicated on the assumption that phylogenies capture meaningful information about evolutionary influences on modern species’ ecological similarities and differences (chapter 1). In this book, we have mostly used phylogenetic relationships as a given. While we discuss approaches to reconstructing phylogenetic relationships (chapter 2), this certainly deserves more critical thought: How do methodological choices about phylogenetic estimation and tree building influence our ecophylogenetic tests? For example, should we use the “best” tree in our analysis, or ought we to use, say, the 1,000 most likely trees, or sample from the Bayesian posterior distribution of trees?
Further, we haven’t fully considered what these trees actually mean for our analyses. Although in chapter 5 we provide a brief overview of various evolutionary models that might describe the evolutionary trajectories of particular traits, we do not critically assess whether and how phylogenetic distances correspond to ecological differences. This link has come under increasing scrutiny in the recent literature, and ecologists need to embrace a more critical and sophisticated approach to ecophylogenetic analyses.
Beyond urging caution and critical thinking when employing phylogenetic methods, we note that there is also an underlying tension with trait-based analyses. Perhaps unsurprisingly, different researchers have differing views on interpreting what phylogeny and traits can tell us about ecological patterns. Some authors see the two as complementary, with both providing potentially meaningful information (Swenson 2014), while others believe that phylogenies simply provide “stand-ins” until the pertinent traits are uncovered. A number of analyses simply compare the explanatory power of traits versus phylogenies (e.g., Cadotte et al. 2009, Liu et al. 2015); it is then possible to select whichever explains more variation in some ecological pattern. However, given the complex relationship that may exist between phylogeny, traits, and the ecological patterns that link to them, it is not clear whether it is valid to simply compare the statistical power of the phylogeny and traits. In addition, phylogeny provides information on the evolutionary history of traits that cannot be inferred simply from community data. Such information might reveal the signature of how ecology has shaped evolution, whether through evolutionary character displacement, adaptive radiations, or niche conservatism, as (p.210) some examples. Some ways forward are starting to emerge, and we revisit some of these advances in the following sections.
10.1. Predicting Ecology from Evolutionary Patterns
Fundamental to testing many ecophylogenetic hypotheses is the translation of phylogenetic distances into expected niche differences (Letten and Cornwell 2014). To date, the community phylogenetic literature has largely focused on ecological hypotheses (e.g., competitive exclusion vs. environmental filtering) and has largely overlooked evolutionary hypotheses about trait and niche evolution, despite rapid advances in both fields. Most ecophylogenetic studies assume a positive relationship between ecological differences and phylogenetic distances (fig. 10.1A), given the premise that greater evolutionary distance has allowed species to accumulate more ecological differences. While this assumption may be common, it assumes a very particular evolutionary model, in which traits continuously diverge over time. This pattern of evolution is not commonly seen in comparative phylogenetic studies (Kelly et al. 2014), and a wide range of alternative models are available. Yet, despite extensive discussion in the comparative methods literature on models underlying trait evolution (O’Meara 2012), they are only rarely considered within ecophylogenetic studies (Peres-Neto et al. 2012, Letten and Cornwell 2014).
A number of recent reviews have stressed the importance of including explicit evolutionary models in ecophylogenetic analysis (Mouquet et al. 2012, Srivastava et al. 2012, Letten and Cornwell 2014). The simplest and most commonly employed evolutionary model is Brownian motion (BM) (Felsenstein 1985), which we reviewed in chapter 5. Under BM, trait change occurs as a random walk, with deviations sampled from a normal distribution (Felsenstein 1985, Harvey and Pagel 1991), and resulting in greater trait divergence among lineages, on average, with increasing time. Thus, the net result of BM evolution for a single trait (e.g., body size) is an increase in variance over time, but the expectation of the mean trait value does not change (fig. 10.1B).
If we assume this type of evolution, then we would expect the mean niche or trait value to remain unchanged, but that the average difference between species increases over time. Further, because we should also expect an increase in the variance of pairwise differences in species’ trait values through time, BM evolution would actually result in a triangular relationship between ecological difference and phylogenetic distance (fig. 10.1C), and not a linear one as is often implicitly assumed.
Many traits reject a BM model of evolution, which gives rise to two important questions. First, does the type of evolutionary model change our expectations and how we articulate our hypotheses? Second, are there other models that might better describe patterns of trait evolution, and could such models help us to conduct more powerful tests? The answer to the first part of the second question is easy, because it is a simple yes. A detailed examination of the various evolutionary models is outside of the scope of this book; we touched on some in chapter 5, but more comprehensive discussions can be found elsewhere (Butler and King 2004, O’Meara 2012). The answer to the first question is more complicated and depends on the hypothesis of interest. If we wish to examine the relationship between phylogenetic distance and coexistence via niche differences (e.g., Slingsby and Verboom 2006, Connolly et al. 2011, Violle et al. 2011, Narwani et al. 2013, Godoy et al. 2014), then absolutely our choice of evolutionary model influences our results. However, ecophylogenetic (p.211)
tests that sum phylogenetic distances or test assembly using mean distances might be relatively insensitive to the particular evolutionary model assumed. Summing distances will always correspond to greater total niche differences regardless of whether the underlying trait distribution reflects increasing variance or linear divergence in absolute value. In general, we might therefore predict that assembly tests would be less sensitive to the particular model of evolution when exploring community structure in multispecies communities. For almost all evolutionary models of trait change (with the possible exception of strict punctualism), sister taxa will inherit the ancestral trait value such that very closely related species will always tend to be more ecologically similar then more distantly related species (Kelly et al. 2014), irrespective of the subsequent model of trait divergence following speciation. Therefore, we should still be able to detect evidence of limiting similarity from competitive (p.212) exclusion as overdispersion, because those species that are different will always be distantly related, while closely related species will be filtered out of the assemblage (along with similar but distantly related species).
10.2. Combining Trait and Phylogenetic Information
Traits and phylogeny may be related to one another in complex ways (Ackerly 2009, Evans et al. 2009), with trait divergence, convergence, and stasis all potentially represented on a phylogeny. If traits explain more variation than phylogeny, it is not that evolution is unimportant for ecological patterns—the traits are evolutionary products. It is instead that our measure of phylogenetic relationships does not adequately capture current species differences. Additionally, different traits will undoubtedly have evolved in different ways and at different rates. Some trait differences may be well captured by a particular phylogeny, but others may not be. Perhaps surprisingly, this is also true if traits evolved following a similar underlying model, such as BM; some traits will covary closely with phylogeny, but others might show little or no phylogenetic signal at all. Whether and how to combine multiple traits into a single multivariate distance is another major question, although one we do not expand on here (but see Swenson 2014); phylogeny simply provides us with one approach. Regardless, it is always useful to ask if traits and phylogeny explain the variation in the data, if one is better than another, or if they provide complementary information. In addition, a mismatch between traits and phylogenetic expectations might itself be interesting (see chapter 6).
We suggest that in many cases phylogenies represent a more flexible tool to understand the evolution of current ecological differences than measuring a handful of traits, because the phylogenies provide not only information on current patterns, but also capture the evolutionary history that has shaped them. We now have multiple models to link evolutionary time to ecological divergence, and we can translate edge lengths into potential differences in, say, niche overlap. With traits, we often lack this insight. For example, does a 10% difference along one niche axis, such as leaf size, correspond to a 10% reduction in overall niche overlap? Because niches have many axes, the answer is likely no. But then how do we combine multiple traits into a single measure? Analyses of phylogenetic trees also have limitations. There is no direct link between phylogeny and ecological mechanism; if we are interested in herbivory resistance, for example, we may do better to measure secondary chemicals. However, it would be even better to understand the evolutionary history of these chemicals, as this might give as additional insight into coevolutionary dynamics. Further, even if traits appear to explain more variation than phylogenetic distances, phylogeny may still explain additional variation that represents unmeasured traits (Bässler et al. 2014). In addition, if we observe a correlation between a trait and some process, we would want to make sure that we have identified the correct trait by correcting for phylogenetic nonindependence in the data.
There are multiple ways that we can combine traits and phylogeny in ecophylogenetic analyses. The simplest is to generate separate distance matrices for each data type and put both distances into singular statistical models, such as multiple regressions or path analysis, to compare explanatory power and statistical interactions. However, this approach is limited because covariance between traits and phylogeny may influence analyses. Alternative approaches allow us to integrate traits and phylogeny directly. For example, we can (p.213) scale edge lengths by trait values (e.g., Blomberg et al. 2003). This approach assumes the phylogenetic topology as given but allows edge lengths to reflect differences in the rates of trait evolution. Such an approach would effectively free us from the assumption that trait evolution follows a Brownian motion model, but of course requires the specification of an alternative model of how traits evolve. As we saw in chapter 5, there are many alternative models to choose from. The adoption of methods that allow us to better choose between them could enhance the ecological relevance of phylogenetic trees, although these methods have not been widely adopted in the ecophylogenetics literature.
A different approach allows us to combine phylogenetic distances (PDist) and trait or functional distances (FDist) into a single functional-phylogenetic distance (FPDist). Values for FPDist weight (a) the independent contributions of FDist and PDist and allow us to combine them nonlinearly (p), as in the following (Cadotte et al. 2013):
Functional distance (FPDist) is essentially phylogenetic distances that include trait convergence and divergence; or conversely, functional distance accounting for information from unmeasured traits that have a phylogenetic signal, and thus should better estimate species functional differences. The calculation of FPDist is possible in the pez package using the funct.phylo.dist() function. This approach can be used to test assembly mechanisms or biodiversity-ecosystem relationships, or can be included in any model that employs species’ distances (Cadotte et al. 2013, Bässler et al. 2014).
10.3. Phylogenetic Insights into a Changing World
E.O. Wilson’s (1965) chapter in the seminal volume Genetics of Colonizing Species (edited by H. G. Baker and G. L. Stebbins) starts with a sober assessment of the current state of ecological and evolutionary theory:
It is commonly stated that the true test of a theory is its predictive power. By this criterion we must admit to the lack of any theory worthy of the name that treats the subject of the ecology of colonization. The failure is dramatized in the case of biological control in economic entomology, a field where correct prediction is all-important to the success of costly projects.
This passage was written over half a century ago. The challenge of using theory to understand basic ecological processes and the potential impacts from invasive species and pests is as great or greater today. In addition, we now face new challenges, including climate change and extinction. We have briefly touched upon extinctions in the preceding chapters, in the following few sections we discuss how phylogeny has also helped us to develop a more predictive science of invasion biology and climate change.
10.3.1. Species Invasions
One major trend that has directly altered the composition of communities is the human-facilitated movement of species around the world. The introduction of nonindigenous species (NIS) brings together species that have been evolving separately for millions of years. When introduced, NIS bring with them a suite of traits shaped by past environments that might (p.214) not match to the past of native species. As we have seen throughout this book, species’ similarities and differences are fundamental in shaping ecological interactions. While taxa with shared past interactions may have diverged in key traits or behaviors allowing for their coexistence (e.g., by partitioning niche space), NIS will have had no such history with the native pool. Thus the ecological dynamics of NIS are highly unpredictable. Many NIS fail to thrive at all, while others go on to spread invasively with dramatic impacts on the ecosystems that they invade (Cadotte and Jin 2014).
Darwin (1859) first articulated the alternative expectations for the phylogenetic patterns of non-native invasions. On the one hand, he supposed that NIS distantly related to the native residents should be more successful. This is often referred to as “Darwin’s naturalization hypothesis” (Daehler 2001). Elsewhere in the Origin of the Species, Darwin suggested that NIS should be more successful where their close relatives are found because they share the necessary adaptations for success in new environments. This is referred to as the “pre-adaptation hypothesis” (Ricciardi and Mottiar 2006). This conflict is often referred to as Darwin’s “naturalization conundrum.”
A number of studies representing different taxa and ecosystems have examined patterns of NIS relatedness to the communities they invade (Rejmanek and Richardson 1996, Duncan and Williams 2002, Ricciardi and Atkinson 2004, Lambdon and Hulme 2006, Strauss et al. 2006, Diez et al. 2008, Diez et al. 2009, Carboni et al. 2013, Park and Potter 2013). Some studies have found evidence that NIS are more closely related to native residents than expected by chance (e.g., Daehler 2001, Duncan and Williams 2002, Diez et al. 2008, Park and Potter 2013), while others show the opposite (e.g., Ricciardi and Atkinson 2004, Strauss et al. 2006, Strecker and Olden 2014). These mixed results hint at the need to more carefully consider the context of species invasion. Opposing findings may have examined assemblages at different spatial scales, studied different stages of invasion, used different null expectations, and considered different introduction histories (Proches et al. 2008, Thuiller et al. 2010, Cadotte 2014). For strong inference we require data on both successes and failures (introduced species that have not gone on to become invasive), as well as information across invasion stages (see e.g., Strauss et al. 2006, Bezeng et al. 2015, Li et al. 2015). Fortunately, such data are becoming more common.
10.3.2. Climate Change
Climate change represents a great unknown, and scientists are being called on to make predictions about how such change will affect global biodiversity patterns (Sala et al. 2000). There has been much work recently trying to predict how species and communities will respond to projected climate shifts (Gruter et al. 2006, Benton et al. 2007, Norf et al. 2007, Altermatt et al. 2008, Davis et al. 2010, Alexander et al. 2011, Sorte et al. 2012). The term “climate change” is a catchall that encapsulates a number of important environmental changes (e.g., temperature, precipitation, seasonality, snowpack, etc.), each of which can affect an organism’s performance, phenology, and interactions with others, either in isolation or in combination (Wolkovich et al. 2012). Shifts in climate might influence the strength of the environmental filters within which community assembly mechanisms operate. Such alterations to environmental filters could result in the selection of predictable assemblages (Alexander et al. 2011), assuming that species fit closely to their climate niche. However, this ignores the importance of biotic interactions; climate change may have more pervasive and subtle impacts than just on where a species can live. Shifts in climate will also (p.215) alter the timing of species activities, with inevitable effects on mutualistic, antagonistic, and competitive interactions among species (Rafferty and Ives 2012, Willmer 2012, Wolkovich et al. 2012).
Phylogenetic relationships provide information that can be used to aid in the understanding and predicting of species’ range shifts (Lawing and Polly 2011) and of ultimate changes in large-scale biodiversity patterns (Thuiller et al. 2011). For example, while we may still not know exactly how thermal tolerances have evolved across a phylogeny, there is evidence that they are relatively conserved (Martinez-Meyer and Peterson 2006, Donoghue 2008, Kozak and Wiens 2010, Lawing and Polly 2011). We might, therefore, expect closely related species to respond similarly to shifts in climate. Moreover, the phylogenetic lineages present in a given region can provide information on physiological processes resulting from altered composition and environmental ecophysiological drivers. For example, information on phylogeny might help predict the presence of different photosynthetic pathways important for estimating rates of carbon sequestration at ecosystem scales (Edwards et al. 2007).
However, key questions remain. For example, as species face changing climates, what will be the relative balance between conserved traits driving range shifts and in situ evolution that allows populations to persist in suboptimal environments (Parmesan 2006)? Another emerging question is how the historical evolution of flowering time will dictate community flowering patterns with earlier springs and altered growing seasons. Harsh environments with short growing seasons appear to constrain flowering time, but in less constrained environments there is a strong phylogenetic signal to flowering time (Lessard-Therrien et al. 2014). Thus, it will be necessary to predict flowering time shifts under different environments, and phylogeny may help with this prediction (Davies et al. 2013, Mazer et al. 2013).
Finally, as community composition changes along with the timing of species interactions, the traditional assembly mechanisms structuring communities also changes. The phylogenetic methods employed throughout this book offer some insights into how we can measure these changes, and perhaps provide predictions on how these changing environments might restructure communities.
10.3.3. Phylogenetic Repercussions of Biotic Homogenization
As human activities alter habitats and abiotic conditions globally, the compositions of regional biotas are becoming more similar (McKinney and Lockwood 1999, Smith 2006, Dormann et al. 2007, McKinney and La Sorte 2007), due in part to the human-facilitated movement of species; but this increasing similarity is also caused by the range expansion of generalist species and species adapted to disturbed environments. A global analysis of long-term observations has shown that while local-scale a diversity has not systematically declined over time, the composition of the assemblages is changing in predictable ways (Dornelas et al. 2014), and this is the engine for biotic homogenization. This homogenization can be measured across a number of different metrics (Olden and Rooney 2006, Winter et al. 2009), and reveals how the multiple facets of biodiversity are converging.
As the phylogenies of regional biotas become more similar, there is a need to consider the reasons and the consequences. There may be phylogenetic reasons to expect that communities become more homogenous; for example, it could be due to the greater filtering that appears to occur in disturbed or modified environments and that selects for certain lineages (see chapter 3; Helmus et al. 2010). Nonindigenous species (NIS) are a cause of (p.216) greater homogenization at larger scales because closely related NIS tend to fill out similar ranges and occur together in anthropogenic habitats (McKinney 2004, 2006; Winter et al. 2009; Cadotte, Borer, et al. 2010). The consequence of such changes is as yet unclear.
Much of this book is predicated on the premise that evolutionary history and phylogenetic relationships influence where species occur and their ecological interactions, and that this has consequences for ecosystem function and services. There is great potential in the application of these methods to address central questions on the dynamics and repercussions of biotic homogenization. What does it mean for the conservation of phylogenetic diversity? Does phylogenetic homogenization lead to homogenization of ecosystem functions and services? To date, we have barely scratched the surface in our efforts to address these questions.
10.4. Where to go from Here?
Ecophylogenetics is a rapidly evolving area of research, and in five years methods might look very different than how they do today. Various core assumptions currently need to be tested and refined, and phylogenetic analyses of communities must better account for ecological, biogeographical, and evolutionary processes. Here we identify several possible trajectories for ecophylogenetic analyses in the future that we think will be important in moving the field forward.
10.4.1. Linking Coevolved Networks
Ecophylogenetic analyses most often use a single phylogeny representing a group of species from a single trophic level (e.g., a plant community). This framework assumes implicitly that trophic interactions are less important in structuring communities. However, organisms evolve in a complex network of interactions that span across trophic levels, and include mutualisms, exploitation, parasitism, and predation; and species from different trophic groups have intimately intertwined evolutionary histories. We have developed theoretical models to make predictions on the effect of one trophic level, a herbivore, on another trophic level, a plant community (Cavender-Bares et al. 2009).
In addition, empirical data suggest herbivores can significantly alter plant phylogenetic community structure (Yessoufou et al. 2013), and that the phylogenetic relatedness of plants can influence patterns of herbivory (Pearse and Hipp 2014). However, we still lack a dynamic phylogenetic framework that allows for feedback between trophic levels, and that considers community structure simultaneously across trophic components.
Advances in the modeling of interaction networks hold much promise. For example, recent models have explored the complex interconnectedness between two groups of organisms (Bascompte et al. 2003, Bastolla et al. 2009, Ings et al. 2009). A classic example of such a network would be a plant-pollinator mutualistic network (e.g., fig. 10.2). Such a network can reveal which flowers are visited by the most pollinator species and which pollinator visits the greatest number of flowers. These networks can also provide additional information that is not simple to extract from the phylogeny for a single trophic group; an example is the nestedness of interactions and the degree of specialization of these interactions (Bascompte et al. 2003). Mutualistic networks are also valuable for predicting the potential for cascading coextinctions resulting from the disappearance of mutualistic partners (Koh et al. 2004). (p.217)
These types of interaction networks have often been analyzed at the species level, but the underlying phylogenies can provide a deeper understanding of species interactions and the evolution of these interactions (Rezende, Jordano, et al. 2007; Rezende, Lavabre, et al. 2007). Phylogenetic interaction networks are not in themselves new, and they have been used to provide insights into coevolutionary dynamics (Legendre et al. 2002); however, our ability to draw ecological inferences from them are developing (Rafferty and Ives 2013). Viewing a phylogenetic interaction network (e.g., host-pathogen; fig. 10.3) can reveal examples of phylogenetic concordance, where a pathogen or group of pathogens only infects a clade of closely related hosts, evolutionary generalists or specialists, or alternatively can identify cases of host switching (e.g., see the impressive work by Streicker et al.  on rabies in bats). While these tools have become available, to date very little work has been done to apply them to community dynamics and assembly.
10.4.2. Linking Biogeography to Phylogenetic Analyses
As we discussed in chapter 4, the species pool used for ecophylogenetic analysis can influence the power to detect nonrandom patterns, even changing their direction and magnitude. However, there is a deeper, subtler issue: the biogeographic history of the species pool. The species pool provides the source material for local assemblages, but it is itself a product of large-scale historical and stochastic processes. For example, if we analyze local plant communities in Sweden, the presence in the species pool of a few gymnosperms, which are distantly related to all nongymnosperms, will increase the probability of detecting local phylogenetic clustering (e.g., Cadotte 2014). If we compare this to an analysis of semi-arid plant communities in western Madagascar with relatively young and small-ranged taxa, we would be more likely to detect overdispersion locally. These biogeographic processes can influence the detection of local phylogenetic patterns independent of the actual local (p.218)
processes influencing community assembly. This can be both a strength and a weakness for ecophylogenetic analyses. The impact of the biogeographical pool should not be overlooked (Lessard, Borregaard, et al. 2012, Carstensen et al. 2013).
There is a further need to link predictions from evolutionary models to the biogeographical processes that have shaped the species pool. For example, an analysis of ecophylogenetic patterns among cichlid fish should explicitly consider an evolutionary model that allows for rapid recent divergence in ecological characters (e.g., Seehausen 2006). If species stemming from a recent adaptive radiation are included in an analysis with older taxa that have evolved relatively slowly, then the models used to generate hypotheses should be flexible and allow for different evolutionary rates within the phylogeny (Purvis et al. 1995).
One glaring ecological example where biogeographic context is especially important is in the analysis of native and nonindigenous communities. As we discuss above, native communities are comprised of species that have been coevolving together for much of their recent evolutionary history, and this sympatry may be responsible for driving ecological divergence, especially among closely related species (Schluter 1994). However, closely related species that are allopatric have not been subject to the same directional evolutionary pressures. Thus, when NIS are introduced into a species pool that contains close relatives, we predict ecological interactions could be either very strong, with both competing for the same resource, or very weak, with the native and non-native species having specialized on a different local resources in allopatry. In addition, it seems likely that we should model ecological differences separately for native and nonindigenous species. If NIS come from different species pools, we might assume that that they have experienced different selection pressures and competitive loads. As NIS have evolved independently from natives, we may expect NIS to appear as if they have evolved according to Brownian motion relative to the natives, which are likely to have evolved coexistence strategies with other sympatric species (Cadotte and Jin 2014). In combination, (p.219) native and nonindigenous species might, therefore, be expected to show different phylogenetic community assembly patterns.
10.4.3. A Call for Carefully Designed Experiments
As we have mentioned repeatedly, the various analyses of ecophylogenetic patterns covered in this book rely on some key assumptions about the relationship between phylogenetic distance and ecological or niche differences. We have argued that more sophisticated evolutionary models need to be used. In addition, we suggest that any ecological predictions stemming from phylogenetic patterns should be tested with experiments. Ecophylogenetic experiments (Liu et al. 2011, Violle et al. 2011, Narwani et al. 2013, Godoy et al. 2014) provide perhaps the only approach for verifying whether inferences drawn from phylogenetic relationships match to ecological mechanisms. There are several potential pitfalls of which experimenters need to be careful when designing such studies. Here we outline three major issues that need to be carefully considered in any experimental tests.
The first issue involves phylogenetic pseudoreplication. Any good experiment requires the replication of statistically independent entities for proper inference (Hurlbert 1984). When we hear this, we naturally think of the physical design, such as randomizing the spatial arrangement of pots or aquaria. Of course treatments also need to be independent, meaning that our pots are assigned to a particular treatment independently from other pot treatments. However, an ecophylogenetic experiment that manipulates the amount of phylogenetic diversity in an assemblage might use different combinations of the same limited number of species to represent different phylogenetic diversity levels. As a result of this design, the same phylogenetic edges are present in multiple diversity treatments. The extreme example of this design is the use of a single species tested against multiple competitors of varying phylogenetic distances (fig. 10.4). In such a study, half of the smallest phylogenetic distance is completely nested within all other distance treatments. If evolution of the relevant trait(s) follows Brownian motion, then we have simply selected a single trajectory, as in figure 10.1B, and we lack an ability to estimate variance among lineages. Regardless
of the actual pattern observed, our inference can only be about the single evolutionary trajectory leading to our test species, and is thus not a reliable test of broader phylogenetic diversity. While the experiment in figure 10.4 is an extreme case assuming a completely pectinate phylogeny, experimenters should be wary of resampling internal edges in competition trials.
The second issue involves unequal numbers of comparisons. Since a phylogeny is hierarchical, the most commonly observed pairwise distances would be from the root split (fig. 10.5A). The result is that we end up with relatively few small phylogenetic distances and many large phylogenetic distances (see chapter 4). A number of recent studies exhibit this pattern. In a detailed study of the phylogenetic patterns of co-occurrence in South African sedges, Slingsby and Verboom’s (2006) data exhibit this skew in phylogenetic distances (fig. 10.5B). The statistical issues associated with this are especially critical in laboratory experiments that may have limited numbers of species and thus very few short distances. Experiments should try to elevate the replication of short phylogenetic distances, at the expense of large distances if space and resources are limited (Cadotte 2013).
The third major issue involves whether niche opportunities match nature. Species evolve in response to many different selection pressures, and for the evolution of competitive coexistence, divergent selection could result in species utilizing different dimensions of spatial and temporal heterogeneity that reduce the strength of negative interactions. Laboratory and controlled field experiments necessarily try to remove unwanted variability in order to test specific hypotheses or mechanisms (Resetarits and Bernardo 1998). However, phylogenetic hypotheses are often ambiguous about the precise mechanisms allowing coexistence, instead preferring to frame hypotheses around a general concept of “niche differences.” Controlled experiments that look for correlations between niche overlap and phylogenetic distances may actually be biased against observing a strong relationship since much of the ecologically relevant niche axes have been removed (fig. 10.6).
Experiments in controlled environments will limit the opportunity for species to take advantage of the heterogeneity they have evolved to utilize. Instead, what we should see from these experiments is a limited ability to detect niche differences and clear inequality (p.221)
in species’ fitnesses, because a subset of species may be well suited to the environment represented by the experimental venue. Thus, to fully assess the evolution of niche differences, experiments need to be carefully matched to species’ resource and habitat requirements, and ecologically relevant heterogeneity needs to be included.
10.5. Heralding the Ecology-Evolution Synthesis?
It is well documented that the studies of ecology and evolution were estranged for much of the twentieth century (Kohler 2002). However, with the advancement of population genetics and the ability to analyze transcription products within organisms, population ecology has served as the vanguard for the (re)merging of ecology and evolution (Johnson and Stinchcombe 2007). This is because population growth and individual fitness are intimately tied to competition among genotypes resulting in changes in gene frequencies (Sakai 1965). These types of ecoevolutionary dynamics operate over short time scales and explain dynamics within a population or among interacting species (i.e., community genetics). The field of ecophylogenetics emphasizes how the evolutionary past informs where species can occur and how they interact with one another today. In this framework, it is evolutionary history that informs population fitness.
We are in the midst of a new synthesis of ecology and evolution, driven by advances in population genetics and ecophylogenetics. But there remains a risk that a new divide might emerge to reinforce a border between within-species population research and work done on communities and ecosystems. This new synthesis must therefore be able to embrace theories, concepts, and tools that allow us to bridge short-term dynamics with the patterns generated through deep evolutionary time. Much of this book discusses patterns at the species level and higher, but many methods can be applied equally across multiple levels of organization, from species to populations, colonies, individuals, and so forth. Along with some of the advances we have discussed in this book, we therefore suggest future work should also aim to merge within-species or population adaptive processes with among-species ecological divergence. There are several possibilities that could be explored. For example, different models can inform different parts of a phylogeny (e.g., population genetics coalescence models below the species model, and ancestral state reconstructions above). In (p.222) addition, reticulate networks that capture gene flow between populations and species could replace strictly hierarchical phylogenetic trees. Many (but not all) phylogenetic metrics can be applied to networks just as easily as they can be applied to phylogenetic trees. It would be possible to also incorporate different gene trees representing different evolutionary histories by, for example, using mitochondrial and nuclear DNA and coding and noncoding regions, or targeted genes with particular functions or under strong selection. As the phylogenetics revolution that emerged in the 1990s comes together with the current revolution in genomics, we suggest there is much potential for new insights and scope for growth.
The recent merger of evolution and ecology has led to exciting new hypotheses and methods, and has stimulated a lot of recent research. We hope that this excitement continues into the future and serves as a harbinger of a broader synthesis of ecology and evolution.