With nature "red in tooth and claw" (1), how do so many species coexist? Why do some species not out-compete the others to extinction? This is a question that has long fascinated ecologists and evolutionary biologists (2), especially given the incredible diversity of seemingly similar species in tropical forests or on coral reefs around the world. Most explanations for coexistence posit either that a given pair of species are so dissimilar that they compete very little with each other (the

more »

... explanation), or that two species are so similar that they coexist for a long but not infinite period of time with neither able to gain an upper hand (the neutral explanation). Both explanations essentially focus on ecological processes like competition and predation, but in this issue of PNAS, Leibold et al. propose a new mechanism for coexistence that couples niche and neutral processes together with rapid evolution (3). This research builds from the increasing recognition that evolution is not just a slow process that plays out over geological time, but also a rapid process that affects interactions among species from year to year and generation to generation. Daphnia ("water fleas") in Lake Constance, Germany, for example, evolved tolerance to a diet with otherwise harmful cyanobacteria after cyanobacteria became abundant in the 1960s and 1970s (4). This evolution may have helped control toxic algal blooms. Guppies in Trinidadian pools with abundant predators evolved earlier maturity and other life history differences, including altered dietary preferences, that altered ecosystem structure and productivity (5). In part because the impacts of rapid evolution on ecological processes remain challenging to detect, they are likely more important than is widely recognized. Theory is helpful in exploring this relatively blank slate. The coexistence model from Leibold and colleagues is based around two species that occupy a set of habitat patches across the landscape, which could be forest patches, individual flowers, pools of water, coral reefs, or many other habitat types. The species compete strongly, so that only one species can occupy a patch at a time, though a storm or other disturbance occasionally opens a patch to re-colonization by removing the resident species. Without evolution, one species occupies primarily one type of patch and the other species occupies primarily the other type of patch ( Fig. 1a for a cartoon with colors for patch types and shapes for species identity). Each species can occasionally occur on the other patch type if it colonizes after a disturbance, but the species gets eventually outcompeted. This is classic niche-based coexistence: each species outcompetes the other species in its own preferred habitat type. With evolution, the picture changes substantially. In this model, each species can (with some low probability) evolve traits that match the other patch type. This is a process of local adaptation, and it allows some members of each species to compete for the other habitat type. In the visual symbols of Fig. 1 , this is represented by the blue triangle evolving to a green triangle and being able to persist on a green habitat patch, or by a green circle evolving to a blue circle and persisting on a blue habitat patch. The authors start their simulations with each species only having one trait set, so the model with evolution closely resembles the model without evolution at first (Fig. 1b) . It is only once each species has an opportunity to locally adapt and evolve the other trait set that the results diverge. Once both species have individuals with both traits, the state of a given habitat patch can change from a mismatch to a match either through evolution or by competition from individuals in the same species or in the other species (e.g., blue circle on green habitat evolving to a green circle, being outcompeted by green circle, or being outcompeted by green triangle, Fig. 1b ). 3 The outcome is an interesting hybrid of niche and neutral theory: the species are differentiated at the scale of each patch, with one species outcompeting the other deterministically in most encounters (niche theory) (Fig. 1b) . However, the species become equivalent at the scale of the landscape, with both species capable of inhabiting both habitat types (neutral theory). Each species has a narrow niche at a particular time at the scale of a habitat patch, but a broad niche that completely overlaps with the other species at the scale of the landscape. Such scale-dependence fits with a broader theme that understanding how processes operate across spatial and organizational scales is key to understanding ecological, evolutionary, and complex system dynamics (6, 7). One consequence of this equivalency is that coexistence on a finite number of patches over infinite time horizons is no longer possible, as in neutral theory. The number of patches occupied by each species will drift up and down, and eventually one species will (by chance) hit zero occupancy. However, such transient dynamics may take a very long time; so long in fact, that the final state may not be all that relevant (8). Leibold and colleagues explore a finite world and other potential complications with an individual-based model, though the results generally mimic the simpler analytical model that they also develop. This theory fits with a growing convergence of niche and neutral theory, even though niche and neutral theory can otherwise appear to be opposites. It would seem that either species are different or species are equivalent. However, we now have examples where the distinctions are much blurrier. High species diversity can create community patterns that are effectively neutral, despite strong niche differentiation among the species (9). A set of species may also be effectively equivalent compared to each other and therefore abundances can drift neutrally up and down, even though that set may be strongly differentiated from other sets (10). The Leibold model adds to this literature by demonstrating that neutrality can also appear with only two species (rather than many) and that it can be a scale-dependent phenomenon. The process of local adaptation matches species traits with their environment, which can at least somewhat decouple species identity from the ecological role that a species plays in the community. The Leibold model is an extreme version of this thinking because one species can be identical to the other species. However, the recognition that local adaptation can be widespread also suggests that traits will often be more useful than species identity for understanding ecological dynamics. Traits like leaf area, leaf nitrogen content, wood density, seed mass, and tree height, for example, help explain which trees coexist in a forest (11). Plant traits also influence primary production, decomposition, and carbon storage, a linkage that has important consequences for ecosystem processes now that traits are changing rapidly with climate warming (12). Trait-based approaches are also being used productively to understand population and community dynamics for fishes, fungi, microbes, and other taxa, though efforts to identify the most relevant and easily measured traits remain a challenge. The potential for local adaptation highlights the importance of measuring traits at the population rather than species level. Much as new perspectives on species coexistence are interesting in and of themselves, a challenge for the model from Leibold and colleagues will be to identify if and where it applies in the real world rather than just as a theoretical novelty in silico. An ecological system that matches their theory would presumably occupy discrete habitat patches of two or more types that are occasionally disturbed and would have relatively low rates of dispersal between habitat 4 patches so that local adaptation could occur. It also requires that two species converge on effectively equivalent local adaptations to different habitat types, which could be a bar too high to cross for the real world in which many factors influence fitness and shape natural selection at once (as opposed to a model world constrained to two phenotypes). Certain amphipod crustaceans (Niphargus spp.) that live in freshwater springs on the Istrian Peninsula in Croatia and Slovenia might fit the description, since they appear to compete strongly at a local level but to fill similar ecological roles at a regional scale (13). Microbial communities across distinct local habitats, like host animals, could represent another system where coexistence matches the Leibold model. However, both systems have not been studied extensively enough from this perspective to know if they really fit, or just appear to fit superficially, and other coexistence mechanisms have also been proposed (14) . More complete tests of the theory would likely require experimentation, and testing for regional-scale neutrality will be especially difficult. Salmon (Oncorhynchus spp.) or sticklebacks (Gasterosteidae) are also intriguing, since they are locally adapted to stream conditions and occupy discrete but often-disturbed habitats (15, 16). However, species differences appear large enough to allow coexistence and hence avoid the region-scale neutrality envisioned by the Leibold model. A system matching the Leibold model also raises intriguing questions. If the species share a recent common ancestor, will the boundaries between species be porous to gene flow or will pre-zygotic barriers evolve? If evolution happens fast, will the species experience an evolutionary arms race to compete for the available habitat? Will evolution drive the species to differentiate and no longer compete, so that the dynamics envisioned by the Leibold model might themselves be transient? How similar do two species need to be in order to exhibit effectively neutral dynamics at a regional scale? Even if the model turns out to apply rarely or not at all in the real world, the Leibold paper makes clear how important local adaptation and rapid evolution can be in questions of species coexistence. Overall, the research highlights the need for a more integrated field that spans both ecology and evolution, including eco-evolutionary dynamics and the feedbacks between the two processes. The two fields remain too often separated by tradition, training, research techniques, and journals, and yet we have ample evidence that both ecological and evolutionary dynamics act on the same timescales. Further discoveries await at the intersection of the two. Acknowledgments I thank