Ecological and community-wide character displacement: the next generation


* E-mail:


Ecological character displacement, mostly seen as increased differences of size in sympatry between closely-related or similar species, is a focal hypothesis assuming that species too similar to one another could not coexist without diverging, owing to interspecific competition. Thus, ecological character displacement and community-wide character displacement (overdispersion in size of potential competitors within ecological guilds) were at the heart of the debate regarding the role of competition in structuring ecological communities. The debate has focused on the evidence presented in earlier studies and generated a new generation of rigorous, critical studies of communities. Character displacement research in the past two decades provides sound statistical support for the hypothesis in a wide variety of taxa, albeit with a phylogenetically skewed representation. A growing number of studies are strongly based in functional morphology, and some also demonstrate actual morphologically related resource partitioning. Phylogenetic models and experimental work have added to the scope and depth of earlier research, as have theoretical studies. However, many challenging ecological and evolutionary issues, regarding both selective forces (at the inter- and intraspecific level) and resultant patterns, remain to be addressed. Ecological character displacement and community-wide character displacement are here to stay as the focus of much exciting research.

Introduction – an early history

Almost half a century ago, Brown & Wilson (1956) first coined the term ‘ecological character displacement’, which was to become a controversial theme in ecology and evolutionary biology. This evolutionary phenomenon reflected the major role ascribed to competition in structuring ecological communities and the assumption that species that overlapped greatly ecologically could not coexist. Thus, while clearly presented as an evolutionary hypothesis to account for a speciation phenomenon, with roots as deep as Darwin's (1859) theory of evolution by natural selection, character displacement has gained a focal spot in community ecology.

Brown & Wilson (1956) suggested that ‘when two species of animals overlap geographically, their differences are accentuated in the zone of sympatry and weakened or lost entirely in parts of their ranges outside this zone’ (p. 63). Their hypothesis referred both to ‘reproductive character displacement’ that would evolve in order to avoid hybridization and to ‘ecological character displacement’, the evolution of size and shape differences that would reduce resource use overlap and thus interspecific competition. Brown & Wilson (1956) refer also to the historic/evolutionary scenario underlying their hypothesis: ‘Character displacement probably results most commonly from the first post-isolation contact of two newly evolved cognate species. Upon meeting, the two populations interact…in such a way as to diverge further from one another where they occur together’ (p. 63). Brown & Wilson (1956) also addressed the opposite phenomenon, later termed by Grant (1972)‘character release’: ‘two closely related species are distinct where they occur together, but where one member of the pair occurs alone it converges toward the second, even to the extent of being nearly identical with it in some characters’ (p. 49).

Hutchinson (1959), asking ‘why are there so many kinds of animals’, extended the discussion of competitively-driven morphological diversification. Assuming ecologically similar species that were morphologically too alike could not coexist, he suggested a minimum size difference between trophic apparati of potential competitors. These minimal ‘Hutchinsonian ratios’ were common in ecological literature of the 1960s and 1970s and provided quantitative expectations about what morphological shifts character displacement should generate.

The notion that species should differ in morphology to reduce resource overlap later produced another contention, that mean sizes of species belonging to an ecological guild (sensuRoot 1967) should be overdispersed, thereby producing size ratios more equal than would be expected by chance (Holmes & Pitelka 1968). This pattern was later termed ‘community-wide character displacement’ by Strong et al. (1979), a term that implies actual coevolution and is somewhat misleading as the same pattern may emerge through species assortment (Strong et al. 1979; Roughgarden 1983). Thus, while ecological character displacement was formulated as an evolutionary hypothesis, community-wide character displacement could arise through a mechanism of significance mainly to community assembly.

For years enthusiastically espoused by ecologists and evolutionists, the concept of ecological character displacement, first questioned by Grant (1972), played a major role in the debate of the late 1970s and early 1980s over tests for and interpretations of patterns in community ecology (Lewin 1983). Strong et al. (1979) used statistical tests of null models for size ratios of several island avifaunas and discovered that observed ratios were no greater than expected for random colonization. Simberloff & Boecklen (1981) tested apparent size-ratio equality between potential competitors and found that in the overwhelming majority of the published cases mean sizes were no more overdispersed than expected by chance. They further pointed out that species combinations or guilds studied were usually not chosen on clear and concrete grounds, that morphological characters analyzed differed between studies with weak justification or rationalization, and that ‘positive’ results were more likely to be published. Finally, Simberloff & Boecklen (1981) found no statistical support for the notion of Hutchinsonian ratios.

From an ‘acerbic and acrimonious’ debate (Lewin 1983) that focused on the role of competition in structuring ecological communities, community ecology emerged as a rigorous and challenging field (Simberloff 2004) where hypothesis-testing and null models are an integral part of the scientific tool kit. Here we review the paths character displacement has taken, with particular emphasis on recent research. Various other patterns of spacing or change that are perceived as competition-induced are occasionally cited, such as patterns of sonar frequency bands in bat communities (Heller & Von Helversen 1989) or regular temporal separation in pollen release (Stone et al. 1996). We limit our review to ecological character displacement and community-wide character displacement in morphological traits and to studies published in the past two decades (so we do not review some significant earlier cases). Our review is also restricted almost entirely to studies where character displacement was sought; there may be many others that describe manifestations of what may well be ecological character displacement but do not refer to this concept [see Robinson & Wilson (1994) for insightful discussion of this issue].

Ecological character displacement – post-1983

The debate surrounding the occurrence and evolutionary significance of ecological character displacement has addressed the statistical tools used, nature of the morphological evidence, choice of competing species (or guild membership), choice of morphological characters and their functional significance, and actual evidence for resource partitioning. Schluter & McPhail (1992) formalized criteria that must be met in order to ascertain that ecological character displacement occurs (cf. Losos 2000):

  • 1The pattern could not occur by chance.
  • 2Phenotypic differences should have a genetic basis.
  • 3Enhanced differences should result from actual evolutionary shifts.
  • 4Morphological differences should reflect differences in resource use.
  • 5Sites of sympatry and allopatry should not differ greatly in food, climate, or other environmental features affecting the phenotype.
  • 6There must be independent evidence for competition.

Some of these criteria are relevant only to character displacement between two species, not necessarily to community-wide character displacement. Criterion 2 is often difficult to prove although a pre-requisite for evolutionary change. Furthermore, meeting all six criteria in a research framework is usually not feasible (Robinson & Wilson 1994). Nevertheless, these criteria provide a coherent framework for research. Studies of ecological character displacement often entail a complex combination of community ecology, functional morphology, adaptation, quantitative genetics, and phylogenetic studies. Moreover, many theoretical models have explored this phenomenon.

Statistical tests are key to proper pattern analysis. First proposed by Simberloff & Boecklen (1981), statistical testing of ratios was debated (reviewed by Dayan & Simberloff 1998). Use of a test by Barton & David (1956) for size ratio equality was advocated by Simberloff & Boecklen (1981); Hopf & Brown (1986) and Pleasants (1994) subsequently suggested alternatives, and the very philosophy underlying tests was debated (Schoener 1984; Tonkyn & Cole 1986; cf. Ranta et al. 1994). The important point is that it is no longer acceptable to publish untested morphological patterns, and this change has produced a more rigorous approach to the study of ecological character displacement.

Theoretical models

While the hard evidence for ecological character displacement was hotly debated, the circumstances under which the phenomenon would arise were also questioned. Models addressing this issue almost all combine a genetic model for control and evolution of the character in question with an ecological model describing interactions between species (primarily limited to competition) that generate natural selection. The starting point for recent theoretical explanations has usually been a two-species model by Slatkin (1980), termed a ‘null model for character displacement’ by Doebeli (1996) because it is very general and has few assumption; Taper & Case (1992) called this model a QGR because it rests on a quantitative genetic recursion model. The underlying genetic model is for a quantitative trait determined additively by several linked loci, while the ecological model is a standard Lotka-Volterra competition model with (1) an individual's ability to use a resource dependent only on the value of a trophic character, and (2) the extent of competition between individuals (conspecific or heterospecific) solely dependent on relative values of the character in each. An important aspect of Slatkin's model is that within-phenotype niche width is fixed.

Slatkin's essential result was that, although ecological character displacement was possible, for most of parameter space it would not arise, and he concluded that the key conditions for its evolution were either differences in resource spectra of the two species or constraints on trait variance. Many theoreticians (e.g. Milligan 1985; Taper & Case 1985, 1992; Doebeli 1996; Drossel & McKane 1999) have subsequently sophisticated this QGR model or relaxed various assumptions to attempt to show that circumstances for ecological character displacement to arise are not so stringent. For instance, Milligan (1985) used a somewhat different genetic model and allowed the two competitors to differ more in resource use, finding character displacement to be a much more likely outcome. Doebeli (1996) explicitly modelled many additive loci without Slatkin's assumption of a normal distribution of the trait in each generation, and he found substantial character displacement even without differences in resource spectra or constraints on trait variance. Drossel & McKane (1999) relaxed assumptions about shapes of the carrying capacity and the competition function, also finding substantial character displacement to be quite likely.

Models in a second category, termed coevolutionarily stable community (CSC) models by Taper & Case (1992), assume that evolution will act to maximize conditional population densities. Developed particularly by Rummel & Roughgarden (1985) and Roughgarden & Pacala (1989), this model is at the heart of an invasion-structured faunal build-up scenario for island Anolis lizard communities, described below.

A third class of models, termed evolutionarily stable strategy (ESS) by Taper & Case (1992), has been championed by Abrams and his colleagues (Abrams 1986, 1987, 1989, 1990; Abrams & Matsuda 1994; Abrams & Chen 2002) and rests on maximization of individual fitness under frequency-dependent selection. These authors have modified the basic model in various ways to make it more realistic. Abrams & Matsuda (1994), for example, include population dynamic consequences of increased competitive trait values, while Abrams & Chen (2002) add an indirect interaction between competitors via shared predators. A persistent theme in these models is that, under many circumstances, sizes of two competitors will evolve in parallel or converge, rather than diverge.

Taper & Case (1992), comparing the three types of model (QGR, CSC, and ESS) by simulation for a limited number of parameter sets, found similar predictions until competition was made asymmetric, at which point CSC predictions differ greatly from those of QGR and ESS models. They recommend the latter two classes, or a melded version that they feel achieves adequate realism while being computationally tractable.

There has been much less modelling of species-sorting and of community-wide character displacement than of ecological character displacement. Leibold (1998) used a ZNGI (zero net growth isocline) model that has no evolution and accounts for species’ requirements for environmental factors and their impacts on environmental factors, including resources. He found that, for a wide array of circumstances, the opposite pattern to community-wide character displacement obtained: the species retained in a community are more similar than would be a random subset from the regional biota.

For ecological character displacement, the key result of intensive modelling as described above is that the phenomenon is expected under a greater range of circumstances than Slatkin (1980) suggested. However, problems remain with respect to how theory relates to empirical results. As Abrams has long lamented (see, e.g. Abrams 1996), both his ESS models and others predict that character convergence or parallelism will occur, probably for a greater range of parameters than character displacement, yet little attention has been paid to this fact in the literature, and few empirical examples are known (but see Sidorovich et al. 1999) – a striking contrast to the many adduced for divergent character displacement. Abrams (1996) suggests this anomaly may result from an asymmetry in the nature of empirical evidence that would implicate convergence or parallelism. Most empirical evidence for competitive character displacement derives from ‘natural experiments’, as we show in this review. However, ‘natural experiments’ could not so easily implicate convergence or parallelism because one might expect either of these responses even absent competition.

Another aspect of modelling to date that seems not to match empirical results is that many models (e.g. those of Taper & Case 1985) predict the pace of evolution to be glacially slow, at least during part of the divergence, yet, as has been shown by some empirical evidence (e.g. Yom-Tov et al. 1999), divergence at least sometimes is quite rapid. Also, the parameterization that would make these models predictive (e.g. foretell what would happen when a species is added or deleted) or explanatory for particular cases is probably impossible with available data.

Models of community-wide character displacement (as opposed to statistical tests for its presence) are far less common than models of ecological character displacement between pairs of species. Yet, in nature, species are usually embedded in larger communities, and understanding patterns (as well as predicting divergence or lack thereof for species additions and deletions) will require a commensurate modelling scale.

Finally, Slatkin (1984) studied the conditions under which sexual size dimorphism would be selected for by intraspecific competition between sexes. His model was analogous to that of Slatkin (1980) for ecological character displacement, and he found a similarly restrictive set of circumstances under which the phenomenon (in this case, dimorphism) would evolve. Bolnick & Doebeli (2003) model evolution of sexual dimorphism with a more sophisticated genetic model and find the phenomenon more likely. In light of extensive empirical examples (cited by Dayan & Simberloff 1998) in which community-wide character displacement is strongly manifested, but only if sexes are treated as ‘morphospecies’, theoretical treatment of these phenomena acting jointly is desirable, but the literature has largely been silent on the matter.

Patterns in nature

The actual evidence for occurrence of ecological character displacement, methods for testing it, and conclusions to be drawn were at the heart of the debate on ecological character displacement. In the past two decades, many studies have sought morphological patterns of ecological character displacement or community-wide character displacement in a diverse array of extant and even extinct taxa – mammals, birds, reptiles, amphibians, insects, other invertebrates, and even plants. We first review briefly the nature of evidence for morphological patterns in these taxa. We then ask whether functional morphology was studied or properly argued in these studies, whether species choices or guild memberships were clearly established, and whether supportive studies of actual resource partitioning were carried out. Finally, we ask whether actual coevolution has been described and whether species-sorting could underlie published patterns.

Among mammals, carnivores have played a major role in the character displacement literature owing, no doubt, to the many species with large geographic ranges and with significant morphological variation (Ralls & Harvey 1985; Dayan et al. 1989a,b, 1990, 1991, 1992; Reig 1992; Dayan & Simberloff 1994a; VanValkenburgh & Wayne 1994; Kieser 1995; Sidorovich et al. 1999; Simberloff et al. 2000). Moreover, it is intuitively easy to grasp how different-sized carnivores would be able to take prey of different sizes (e.g. Gittleman 1985) and thus partition resources. The study of Australian dasyurid carnivores seems also a natural choice (Jones 1997). Most other studies have dealt with insectivores (Malmquist 1985; Loy & Capanna 1998; Racz & Demeter 1998; Ochocinska & Taylor 2003), bats (Heller & Von Helversen 1989; Yom-Tov 1993; Arita 1997; Kingston et al. 2000), and rodents (Alcantara 1991; Yom-Tov 1991; Dayan & Simberloff 1994b; Parra et al. 1999).

The classical bird studies are of Galapagos finches (Schluter et al. 1985; Diamond 1987; Schluter 1988), but honeyeaters (Diamond et al. 1989), grebes (Fjeldsa 1983), fairy-wrens (Tiedmann & Schodde 1989), vultures (Hertel 1994), crossbills (Benkman 2003), and owls (Gehlbach 2003) were studied as well.

Island Anolis lizards are the best-studied reptile taxon for character displacement (Rummel & Roughgarden 1985; Roughgarden & Pacala 1989; Losos 1990, 1992, 1994; Case & Bolger 1991; Losos et al. 1993; Losos & Irschick 1994; Roughgarden 1995; Miles & Dunham 1996; Butler & Losos 1997, 2002; Schneider et al. 2001; Rogowitz 2003). A few other lizard taxa have been examined (Galagher et al. 1986; Radtkey 1996; Radtkey et al. 1997; Carranza et al. 2000; Vitt et al. 2000; Melville 2002); most are also island-dwellers. Map turtles have also been studied (Lindeman 2000). Research on ecological character displacement in amphibians has focused on salamanders (Irschick & Schaeffer 1997; Maret & Collins 1997; Adams & Rohlf 2000; Jaeger et al. 2002) and spadefoot toad tadpoles (Pfennig & Murphy 2000, 2002, 2003).

Robinson & Wilson (1994) argued that the prevalence of ecological character displacement among fishes has largely been neglected for a variety of reasons including the fact that, while some of the most spectacular adaptive radiations have occurred in fishes, many researchers tend to discount the significance of competition in generating them. They point out that only five references to teleost fishes appear in the conceptually oriented literature on character displacement although competition is often a diversifying force, creating differences among fishes that evolve in predictable ways (Robinson & Wilson 1994). They argue that character displacement examples appear in fisheries journals not routinely cited by ecologists and, more significantly, that criteria for demonstrating the phenomenon are too stringent (Robinson & Wilson 1994). We feel fishes have come a long way since then, and while threespine sticklebacks account for a large part of the published studies (Schluter & McPhail 1992; Schluter 1993, 1994, 1995, 1996b, 2003; Caldecutt & Adams 1998; Taylor & McPhail 1999; Rundle et al. 2000, 2003; Scott & Foster 2000; Pritchard & Schluter 2001; Gray & Robinson 2002; Vamosi 2002, 2003; Vamosi & Schluter 2002, 2004; Hart 2003; Bolnick 2004), other fish taxa have had a share: East African cichlids (Ruber & Adams 2001; Allender et al. 2003), sunfishes (Robinson et al. 1993; Wainwright 1996; Jastrebski & Robinson 2004), stream fishes (Knouft 2003), salmonids (Alexeev et al. 1986; Pigeon et al. 1997; Dynes et al. 1999; Wood et al. 1999; Jonsson & Skulason 2000; Forseth et al. 2003), paragalaxias in Tasmanian lakes (McDowall 1998), perch (Hjelm et al. 2000; Svanback & Eklov 2003), lake-dwelling fishes of the southern Andes and Patagonia (Cussac et al. 1998; Ruzzante et al. 1998, 2003; Logan et al. 2000), and others (Poeser 1998; Lu & Bernatchez 1999; Matthiessen et al. 2003; additional references in Robinson & Wilson 1994).

Although Hutchinson's (1959) seminal address followed his observations of two species of Corixidae, not many studies deal with morphological patterns among insects. The few publications treat ants (Foitzik & Heinze 1999; Gotelli & Ellison 2002), beetles [Barr & Crowley 1981; Kawano 1995, 2003; Sota et al. 2000; Satoh et al. 2003 (some cases may actually reflect reproductive rather than ecological character displacement)], and aphids (Sorensen 1994). In other invertebrate studies, snails play a major role (Cherrill & James 1987; Grahame & Mill 1989; Cowie 1992; Camargo 1993; Chiba 1993, 1996, 2004; Saloniemi 1993; Emberton 1994, 1995; Gorbushin 1996; Barker & Mayhill 1999; Grudemo & Johannesson 1999; De Francesco & Isla 2003), with a few studies of brachiopods (Leighton 1998), crustaceans (Yamaguchi 2003; Marchinko et al. 2004), millipedes (Bond & Sierwald 2002), and earthworms (Fragoso & Rojas 1997).

The hypothesis of ecological character displacement was developed specifically for animal species. However, several plant studies sought this phenomenon; the majority deal with characters related to plant–pollinator interactions (Murray et al. 1987; Armbruster et al. 1994; Hansen et al. 2000; Miyake & Inoue 2003), which could generate both ecological and reproductive character displacement (e.g. Grant 1994); pollinators are a shared resource for flowering plants, and morphological spacing of flower structures partitions this resource between them. However, the character displacement that evolves reinforces reproductive isolation.

Veech et al. (2000) studied patterns in seed sizes of coexisting conifer species, assuming that seed sizes reflect an evolutionary trade-off between seed number and seed mass and so can affect the outcome of seedling competition. While they found no regular patterns between conifer seed sizes, non-random patterns obtained between seeds in pine-only assemblages, suggesting that this pattern may promote coexistence (Veech et al. 2000) – an unusual claim of ecological character displacement in plants.

In sum, studies on a wide array of taxa have explored ecological character displacement and community-wide character displacement, but particular taxa dominate the literature: mammalian carnivores, Galapagos finches, island Anolis lizards, threespine stickleback fish from western Canada, and snails. Vertebrate studies far outnumber those of invertebrates, and most studies report positive results, that is, suggest the phenomenon is operating.

Character displacement, functional morphology and resource partitioning

While Brown & Wilson (1956) referred in general to morphological displacement, their examples include Lack's studies of bird beaks, which dealt with both size and shape, and Hutchinson (1959) specifically pointed to differences in metric characters related to the trophic apparatus. Almost all subsequent research has dealt with size. Morphological traits must be functionally significant for different character states to evolve in response to competition and to resource partitioning. Simberloff & Boecklen (1981) noted that earlier ecological character displacement studies used different and often poorly rationalized characters. The association of character displacement with resource partitioning is such that in some studies (e.g. Milton 1991; Alterio & Moller 1997; Brantingham 1998) the term character displacement was used loosely as a synonym for resource partitioning or dietary specialization. However, we focus on research that actually aims to show a relationship between the morphology and the diet.

While some researchers still use body mass or length of potentially competing species (e.g. Cavallini 1995; Jimenez et al. 1995; Gotelli & Ellison 2002), probably often because such data are published, many have begun to study characters more directly related to resource use. Malmquist (1985) judged jaw measures to be adapted to food niche in shrews, Racz & Demeter (1998) studied character displacement in mandible shape and size in two species of water shrews, Kieser (1995) studied mandibulo-dental size patterns of foxes and compared those with size patterns of guild members, and Loy & Capanna (1998) found enhanced differences in areas related to the zygomatic region in two sympatric voles and suggested that character displacement mainly relates to efficiency of food gathering and processing. For gerbillids Yom-Tov (1991) studied upper toothrow length, a character he thought related to seed size and to amount of vegetative materials consumed. Similarly, Satoh et al. (2003) studied mandible size of tiger beetles, because mandible size is correlated with prey size in adults of this taxon.

Other studies focused on characters that appear to have even more direct functional significance for resource partitioning. In birds, the classic trait is beak size, related to food size, dating back to Lack's (1947) study of Darwin's finches (Fjeldsa 1983; Hertel 1994; Gehlbach 2003). Dayan et al. (1989a,b, 1990, 1992) and Dayan & Simberloff (1994a) studied morphological patterns in carnivore specialized dentitions (canines and carnassials); mustelids, viverrids, and felids kill by inserting an upper canine between two cervical vertebrae, so Dayan et al. (1989a, 1990) and Dayan & Simberloff (1994a) studied maximum diameter of the upper canines, suggesting it would best reflect preferred prey size. For canids, which kill differently, Dayan et al. (1989b, 1992) studied carnassial lengths, as did VanValkenburgh & Wayne (1994) for East African jackals and Werdelin (1996) for late Miocene and early Pliocene hyaenids in Eurasia and Africa. Dayan & Simberloff (1994b) studied width of the upper incisors of heteromyid rodents, used to husk seeds, as did Ben-Moshe et al. (2001) for gerbillids. Parra et al. (1999); Millien-Parra (2000); Millien-Parra & Loreau (2000), and Millien & Jaeger (2001) also studied rodent incisors. In fish, morphological differences (such as in numbers and lengths of gill rakers) of two threespine stickleback species coexisting in post-glacial lakes in western Canada have been the subject of studies affirming the adaptive role of the different morphs (Schluter & McPhail 1992).

Indeed, an increasing number of researchers have studied the relationship between morphological characters and actual resource partitioning. Some patterns of resource partitioning (but still a minority) provide strong support for the hypothesis of competition-induced character displacement. For example, Fjeldsa (1983) found differences in prey of grebes between sympatric and allopatric conspecific populations, and between coexisting species, that are related to their bill sizes, with larger-billed populations taking more and larger fish and smaller-billed grebes taking small invertebrates. Jones (1997) perceived canine strength as significant for prey partitioning among dasyurid carnivores, and her field work on the food habits of different species indeed revealed resource partitioning between them. In salamanders character displacement was demonstrated in trophic anatomical differences related to functional and biomechanical differences in jaw closure; one species developed a faster closing jaw and the other a slower but more forceful jaw closure (Adams & Rohlf 2000). These differences were associated with differences in prey consumption. Ben-Moshe et al. (2001) discovered regular size patterns in gerbillid incisors, used for husking seeds, and found in cafeteria experiments that larger species did indeed prefer larger seeds.

In a particularly elegant series of studies, Schluter (1993, 1995) examined habitat use efficiencies, and growth rate as a surrogate of fitness, for two morphologically distinct threespine sticklebacks (Gasterosteus spp.), one of several coexisting fish pairs that formed in coastal British Columbia lakes after the last glacial retreat. Sticklebacks occurring on their own appear intermediate in morphology and habitat use, so this system has become a major focus of character displacement research. Schluter (1993, 1995) found that in the benthic habitat the larger, deeper bodied species (with wide mouth and few, short gill-rakers) was a better forager than the smaller species (with narrow mouth and many, long gill-rakers) and consequently exhibited higher growth rates, while the reverse was true in open waters. Hybrids were intermediate in success, suggesting the evolution of fitness tradeoffs in this habitat gradient (see also Robinson & Wilson 1994); that is, that adaptation to one habitat has occurred at the expense of feeding rate in the other, while hybrids have low fitness (Schluter 1993, 1995). Similarly, trade-offs in foraging efficiency leading to divergent natural selection have been demonstrated in polymorphic perch populations, where pelagic and littoral perch morphs exhibit divergent morphologies functionally related to differences in search and attack velocities of prey in different habitats (Svanback & Eklov 2003). The functional morphology and feeding biology of sunfishes have been thoroughly studied as well and show how adaptations to one resource are at the expense of adaptations to others (Wainwright 1996).

There has been little study of the degree of heritable inter-individual difference in resource use and the relationship of such differences to morphological variation, and these must become research foci to support the general hypothesis that character displacement is induced by presence of a similar species. Bolnick et al. (2003) find much evidence for individual resource specialization in a recent review.

Resource partitioning studies do not always unequivocally support morphological patterns; Dayan & Simberloff (1994a) found regular inter- and intraspecific spacing in mustelid canines in the British Isles, while McDonald (2002), analysing studies of diets of the same species, felt his results supported resource partitioning between sexes but not species along a prey size axis. Likewise, Wood et al. (1999), studying anadromous and non-anadromous morphs of sockeye salmon (Oncorhynchus nerka), discovered that, despite a major difference in gill-raker number between morphs, taken in other species to reflect competition-induced selection for differences in trophic apparati (e.g. Schluter & McPhail 1992), no differences were found in diets of the two populations when they coexist during their first year.

Notably, studies of character displacement use a growing number of morphological measures (e.g. Mikulova & Frynta 2001; Lee & Mill 2004), reflecting both increased awareness of the significance of understanding functional morphology of the species studied and widespread access to suitable computational software. For example, following several studies of ecological character displacement in mustelids that focused on univariate analysis of one or few characters (Ralls & Harvey 1985; Dayan et al. 1989a; Dayan & Simberloff 1994a), Lee & Mill (2004) used canonical variate analysis on 19 morphological measurements to study cranial variation in British mustelids. They found that width of the zygomatic arch and height of the sagittal crest are the key traits differentiating eight species and that these traits related to the manner of killing prey.

The use of multivariate studies facilitates analysis of shape as well as size. Changes in size can be associated with changes in shape and can have functional significance. For example, Adams & Rohlf (2000) found shifts in sympatry in two salamander species not only in size, but also in relative lengths of the squamosal and dentary, which are functionally associated with jaw strength and gape closure speed. These differences accord well with shifts in their prey size in sympatry. Likewise, Robinson & Wilson (1994), reviewing character displacement in fishes, conclude that competition is a strong diversifying force among them and that differentiation often takes place along predictable pathways and almost always includes morphologically distinct pelagic and benthic forms. Stickleback shape differences have recently attracted much attention (Schluter & McPhail 1992; Schluter 2000).

While the hypothesis of ecological character displacement and ensuing studies have focused on size shifts and patterns, it is a tenable hypothesis that shape changes could produce divergence in resource use, irrespective of size changes. VanValkenburgh & Wayne (1994) studied three jackal species in East Africa, where they overlap geographically, finding size convergence of the three species but divergence in shape, with black-backed jackals (Canis mesomelas) adapted for increased carnivory.

In sum, while earlier studies of ecological character displacement often focused on size alone, with little heed to the functional significance of the characters, the past two decades see a sharp increase in studies using sophisticated approaches to functional morphology, and increasingly also investigating actual resource partitioning.

Species and guilds

Ecological character displacement was assumed to occur between closely related newly evolved species that come into contact (Brown & Wilson 1956). Community-wide character displacement, by contrast, was expected to occur within ecological guilds (sensuRoot 1967). Guilds comprise species that use the same class of resources in a similar way: ‘the term groups together species without regard to taxonomic position, that overlap significantly in their niche requirements’ (Root 1967). Guild assignment could have a critical influence on the outcome of morphological analyses of size ratio equality (Simberloff & Boecklen 1981), but guild membership is not easily determined (Simberloff & Dayan 1991). Simberloff & Dayan (1991) pointed out that studies of guilds differed particularly in the significance attached to similarity in resource use (see Hawkins & MacMahon 1989) and that, whenever guilds are used, it should be with a clear rationale for which species are included and which excluded.

Not many studies actually address character displacement between ‘cognate species’ or sister-species. This is probably because neither phylogenetic nor biogeographic histories are usually part of the study, except for island lizards (see below). Often phylogenies are not fully resolved, and we doubt that most ecologists studying competition are equally concerned with resolving phylogenetic scenarios. Studies of day-geckos (Phelsuma) in the Seychelles Archipelago (Radtkey 1996) and scincid lizards (Melville 2002) probably do relate to sister-species.

Many researchers simply assume that closely related species are more likely to compete than are distantly related ones. Many studies therefore deal with character displacement among congeners (e.g. Malmquist 1985; Reig 1992; Jimenez et al. 1995; Loy & Capanna 1998; Sidorovich et al. 1999; Woodman 2000; Mikulova & Frynta 2001; Aunapuu & Oksanen 2003), often without actually testing for occurrence of competition.

However, some authors have grouped distantly related species within the same guilds (reviewed by Simberloff & Dayan 1991; Dayan & Simberloff 1998). For example, Dayan et al. (1989a) studied community-wide character displacement in a carnivore guild comprising members of two families. Community-wide character displacement was often found to result from actual coevolution (Dayan & Simberloff 1998), so the study of this phenomenon in distantly related taxa implies that Brown & Wilson's (1956) definition may be overly restrictive. An unusual case of character displacement resulting from interactions between distantly related species is that of finches and bees in the Galapagos Islands. Schluter (1986) reports that Geospiza fuliginosa and G. difficilis take significantly more nectar on islands from which the larger carpenter bee (Xylocopa darwini) is absent and that their body sizes are significantly smaller than those of populations on islands with bees. Individual finches that exploit flower nectar are smaller than conspecifics that do not (Schluter 1986).

Mechanisms – coevolution, species assortment, taxon-cycles, and textural discontinuity

Character displacement explicitly refers to competition-induced morphological change in sister-taxa or other closely related taxa that overlap post-divergence (Brown & Wilson 1956). In fact, to ascertain that character displacement has occurred, morphological patterns that do not result from actual coevolution must be factored out. Community-wide character displacement, on the other hand, can result from either coevolution between competitors or species assortment; that is, only species sufficiently dissimilar from one another would be able to enter an ecological community.

Some studies of non-random morphological patterns do not explore the mechanism that has generated these patterns. However, many studies provide evidence for coevolution by examining variation in morphology of the food-gathering apparati, and sometimes in food use, among congeners in allopatry and sympatry (e.g. Dayan et al. 1989a,b, 1990; Dayan & Simberloff 1994a; Adams & Rohlf 2000). On the other hand, few studies present compelling evidence for species assortment. Bowers & Brown (1982) provided an interesting case, patterns in body sizes of granivorous rodents of the North American Southwest. They found that species of similar size coexist less frequently in local communities and overlap less in geographic distributions than expected on the basis of chance and concluded that this pattern reflects interspecific competition. The scale used for this study is that of local coexistence; only species of the regional species pool that ‘obeyed’ a size difference rule could actually co-occur. Similarly, Gotelli & Ellison (2002) studied New England ant assemblages in bog and forest habitats and found that at the scale of local coexistence in bogs there was weak segregation by body size. At the wider scale of geographic ranges, Letcher et al. (1994) found for Palearctic mammals and those of the British Isles that species that overlap less than expected by chance are more similar in body size than those that overlap more, and they concluded that competition may cause species with similar body sizes to reduce geographic range overlap.

The occurrence of both character displacement and species assortment in island lizards has long been suggested (reviewed by Case & Bolger 1991), and the development of sophisticated phylogenetic tools opens exciting prospects for gaining insight into the history of colonization as well as of size patterns; do observed sizes reflect morphological evolution or can they be explained more parsimoniously by sizes of the ancestors?

Radtkey (1996) studied day-active geckos (Phelsuma) in the Seychelles Archipelago, where a monophyletic lineage diverged post-colonization, and found evidence for evolutionary change in body size in response to the presence of a congener. Subsequently, Radtkey et al. (1997) explored a previously suggested case of ecological character displacement of whiptail lizards (Cnemidophorus) on the Baja peninsula and on oceanic islands in the Sea of Cortez and found evidence of character release on islands. However, another study of gekkonid lizards (Tarentola) that invaded North Atlantic islands, using molecular phylogenies, revealed size patterns that were best explained simply by ancestral sizes (Carranza et al. 2000).

One of the most intensively studied model systems consists of Anolis lizards on Caribbean islands, where c. 140 of the nearly 400 species are found (Losos 1994; Roughgarden 1995; Jackman et al. 1999). Schoener (1970) described non-random patterns in the sizes of these species on different islands and variation in size with number of coexisting species, and various historical scenarios have been deduced starting with Williams (1972). Several recent conflicting studies have attempted to gain insight into the evolution of size patterns in this system. Rummel & Roughgarden (1985) developed the theory of faunal buildup, which distinguishes between invasion-structured and coevolution-structured ecological communities. According to this theory, Anolis lizards on two-species islands of the Lesser Antilles consist of an original species (which as a single species expresses the optimal body size for the island) and a larger invader. Competition prevents a species with the same size as the resident from invading; body size differences on the two-species islands express the difference needed for a second species to invade an island successfully (Roughgarden & Pacala 1989). Invoking the taxon cycle (Wilson 1961), they suggest that the resident species evolved reduced body size so as to reduce competition with the invader, a scenario not unlike ecological character displacement. However, the resident also evolves to become smaller so as to use the empty space that opened up along the centre of the resource axis (Roughgarden & Pacala 1989); the resident may also be driven to extinction.

Losos (1990, 1992) studied Anolis lizards of the Lesser Antilles and concluded that a relative rarity of size evolution suggests that size assortment must cause non-random patterns. Additionally, he suggested that similar-sized species may not coexist because they interbreed and coalesce into one gene pool, a possibility that Giannasi et al. (2000) found unconvincing for want of actual evidence of hybridization. Miles & Dunham's (1996) analyses based on ancestor reconstruction methods supported the taxon-cycle model but failed to reject the character displacement model. Finally, Giannasi et al. (2000) carried out a phylogenetic analysis of body size evolution in Caribbean Anolis and concluded that observed patterns of body size variation resulted from a combination of size assortment and ecological character displacement. Their analysis demonstrated that size increase occurs more often than size decrease. However, this seems not to be the end of the story, with new evidence demonstrating that after severe storms islands may be recolonized by overwater dispersal from neighbouring islands (Schoener et al. 2001), in routes that are congruent with the direction of ocean currents and in ways that may limit adaptive divergence (Calsbeek & Smith 2003).

Holling (1992), studying birds and mammals of boreal forests and short-grass prairie, suggested that gaps between body sizes of members of regional communities are far greater than chance would predict – a version of community-wide character displacement. He ascribed this pattern to a sorting generated by discontinuous landscape ‘texture’, the latter defined in terms of time- and space scales that are believed to divide landscapes into a small number of discrete ‘habitat quanta’. Both assortment and evolution would then match residents by body mass with particular quanta, and Allen et al. (1999) argue that this matching process determines which species can invade particular landscapes and which will go extinct. Although widely cited in the landscape ecological literature, this hypothesis has been cogently criticized by Siemann & Brown (1999) on the grounds that, by statistical test, gaps in body sizes are not significantly larger than random.

That so few published cases actually deal with or support a pattern of species assortment implies that, while ecologists have demonstrated the evolutionary role of competition and the occurrence of coevolution, we are still far from proving a significant role for size-based species assortment. Note that, because the occurrence of Hutchinsonian ratios is not supported by statistical analysis (Simberloff & Boecklen 1981; Dayan & Simberloff 1998), a minimal size ratio that limits species membership within a community is at present not known. Few cases show clearly that competition affects species assembly in ecological communities on the basis of size assortment. Guilds within ecological communities may be size-structured, but support for size-based rules of their assembly, of great significance to community ecologists, is scant. A scenario of species entering a community only if sufficient size differences occur between invaders and residents has been suggested for patterns of invasion of non-native birds in the Hawaiian islands (Moulton 1985; Moulton & Lockwood 1992) and Anolis lizards in the Caribbean (Losos et al. 1993). However, if species evolve depending on sizes of other species present, and species assortment is the exception rather than the rule, then size restrictions tell us little about community invasibility.

Alternative selective forces

Brown & Wilson (1956) viewed ecological character displacement as a competition-driven speciation phenomenon, usually arising when two closely related newly evolved species meet. However, resource competition is but one of various ecological interactions that may select for evolution of animal size or shape. In the past few years different studies have invoked interference competition, predation, and even facilitation as alternative selective forces that can produce morphological diversification and character displacement. Moreover, some authors have noted that results of competition-driven displacement can affect the outcome of other ecological interactions of competing species. This is an arena that has just begun to attract researchers, and while we provide several examples, much more research is required to assess the significance of these interactions.

Body size is commonly associated with an animal's abilities in interference competition, with larger individuals usually having the upper hand (see Discussion in Yom-Tov & Dayan 1996). Therefore, increased body size in sympatry with a competitor might also evolve in response to ‘alpha selection,’ through interference competition (Jaeger et al. 2002). This alternative hypothesis was posed and tested for salamanders (following Adams & Rohlf 2000), because shifts in jaw morphology can relate to shifts in prey size, but also to shifts in bite strength and speed. While experiments found no support for its occurrence in one species pair (Jaeger et al. 2002), the significance of changes in body size among competitors remains to be studied.

On the other hand, Melville (2002) described ecological character displacement in two species of alpine scincid lizards (Niveoscincus) in Tasmania; the larger species dominated interference competition. She suggested that, while the larger species is selected for interference abilities, aggression between similar-sized individuals may select for decreased body size of the smaller species.

In carnivores, Whitehead & Walde (1993) argued that even spacing of the logarithmic diameter of canine teeth in mustelids and felids shown by Dayan et al. (1989a, 1990) may more likely result from behaviourally mediated competition for space than from exploitative competition for prey. Modelling the ecological and evolutionary divergence of a character associated with territory defence such as canine size, they show that an assumption of size-dependent territorial aggressiveness can lead both to character divergence and to sexual size dimorphism. McDonald (2002) supports the same hypothesis. Likewise, citing aggressive interactions documented by observations using radio-tracking, Sidorovich et al. (1999) interpret the size increase of male and female European mink and female polecats in response to invasive American mink as a response to direct aggression from American mink rather than to resource competition.

In recent years, the significance of predation in promoting divergence has been suggested either through ‘apparent competition’ or ‘competition for enemy-free space’ or in reinforcing divergence, once evolved, because different morphs face different predation pressures (see Vamosi & Schluter 2002). This is a new line of investigation, the outcome of which is still uncertain.

Vamosi (2003) experimented with predator avoidance tactics of ‘limnetic’ and ‘benthic’ threespine sticklebacks; both species survived better in preferred habitat. He concluded that, while competition may cause diversification across the two habitats, habitat-specific predation pressures may ‘sharpen’ adaptive peaks and contribute to observed divergence (Vamosi 2002). Experiments with native trout (O. clarki) as a predator for both sticklebacks in ponds revealed that trout affected survival of the limnetic morph but not the benthic morph, while hybrid survival was low, with or without trout. Vamosi & Schluter (2002) suggest that different predation pressures between the habitats can promote divergence owing to between-habitat differences in predation risk but could also simply reduce the value of open waters, thus limiting divergence.

Day & Young (2004) suggested on the basis of theoretical considerations and experimental studies with microbes that facilitation may also select for sympatric speciation and diversification, and they propose an experimental method to distinguish between competition and facilitation as selective forces. They suggest that, as facilitation increases the number of resources, hence number of niches, evolutionary branching resulting from facilitation is a likely phenomenon (Day & Young 2004). While this hypothesis remains to be tested, it could bear on the evolution of ecological character displacement.

In sum, in theory size divergence may result from other evolutionary forces that can either select for it or reinforce it, but more research is required to investigate this possibility.

Intraspecific patterns – sexual size dimorphism, trophic polymorphisms, and disruptive selection

While the hypothesis of ecological character displacement was conceived to account for an interspecific phenomenon, intraspecific size patterns have also been incorporated in this framework. Individuals within populations and species exhibit some degree of variation in the trophic apparatus (Bolnick et al. 2003) that can be continuous but also discontinuous (Smith & Skulason 1996). The effect of competition on patterns of continuous variation has been the focus of the niche variation hypothesis (Van Valen 1965), recently reviewed and tested elsewhere (Meiri et al. 2005). Another form of intraspecific variation is ontogenetic change; it has been suggested that, in many insect herbivores, substantial ontogenetic character displacement occurs in mandibular morphology, which is correlated with changes in feeding strategy and behaviour. This phenomenon is termed ontogenetic character displacement (Hochuli 2001). Whether this pattern merits the term ecological character displacement is questionable, but the phenomenon itself must be significant in slow-growing taxa where an animal may spend much of its life, possibly the more critical parts of it, in a wide array of sub-adult sizes, and the severity of selection for a given adult size may be influenced accordingly.

Here we focus on relevant patterns of discontinuous variation. Particularly important are sexual size dimorphism, common in many taxa, and the less common phenomenon of trophic polymorphism.

Sexual size dimorphism, a form of intraspecific variation, is associated with natural and/or sexual selection. Natural selection is perceived to promote size differences between the sexes that will reduce resource overlap between them. Therefore, in recent years researchers have sought non-random morphological patterns that involve both inter- and intraspecific character displacement. While genetic links between males and females are obviously strong, research shows that different selective pressures can and do affect males and females, resulting in different sizes and morphologies.

A large scientific literature on ecological causes of sexual size dimorphism is reviewed by Shine (1989), Fairbairn (1997), and Meiri et al. (2005), and we will not review it here. We point only to research in which each sex was considered a ‘morphospecies’ and all morphospecies were analysed at once in search of non-random size patterns. Dayan et al. (1989a, 1990) and Dayan & Simberloff (1994a) studied canine size in sexually dimorphic carnivore species and found community-wide character displacement between the different morphospecies, with males and females sometimes interspersed in size (Dayan et al. 1990). Similarly, Jones (1997) studied patterns of canine strength in dasyurid carnivores. However, the interpretation of sexual size dimorphism, particularly in characters related to inter- and intraspecific aggression, remains controversial (Whitehead & Walde 1993; Gittleman & VanValkenburgh 1997; Johnson & MacDonald 2001); the trophic apparatus of many species is used also for defence or aggression. Moreover, males and females can diverge in size, shape, or both (Butler & Losos 2002), and clearly much remains to be studied here. Nevertheless, it is quite amazing to note that, for many dimorphic species, only a single sex (almost always males) was studied for many years, including two taxa that figure heavily in the character displacement literature, carnivores and Anolis lizards, where patterns were sought while ignoring a full 50% of the population!

Same sex adults of different species may also exhibit morphological variation within the same or overlapping populations. Vaisanen & Heliovaara (1989), who studied two parapatric alternate-year cohorts of a pine bark bug species (Aradus cinnamomeus), found that the closer the allochronic bug populations were to each other spatially, the greater the morphological difference. They suggested redefining ecological character displacement to include intraspecific evolutionary unit changes also.

Trophic polymorphisms are perhaps the most intriguing type of intraspecific variation that has been explored from the perspective of ecological character displacement. Here, within a population, two or more discrete morphs exist that can be distinguished by coordinated changes involving many individual characters, which are usually related to feeding structures (reviewed by Hanken & Hall 1993). The magnitude of morphological differences in trophic structures suggests adaptation (Hanken & Hall 1993), and it is tempting to view these adaptations in the framework of competitively driven morphological divergence.

Maret & Collins (1997) argued that some alternative morphologies within a single species have a common phylogenetic history and are thus likely to be under selection by ecological conditions experienced in sympatry. Trophic polymorphisms can therefore be used as model systems for exploring ecological factors that promote morphological divergence, as well as to gain insight into ecological conditions that promote sympatric speciation (Maret & Collins 1997). They conducted field surveys and experiments with typical and cannibal morphs of tiger salamander (Ambystoma tigrinum nebulosum) larvae, the latter comprising the largest individuals in the population. Their results suggest that the cannibal morph is advantageous in dense populations with intense resource competition (as well as abundant conspecific prey), as would be theoretically expected (Maret & Collins 1997).

Hanken & Hall (1993) ask whether trophic polymorphism is an incipient stage in the evolution of interspecific morphological diversity. Brown & Wilson (1956) focused on newly evolved species that meet following allopatric speciation:

‘Divergence between two species where they occur together, coupled with convergence where they do not, is a pattern that strongly suggests some form of interaction in the evolutionary history of the pair. The usual case may be one in which the members of the pair are cognate (derived from the same immediate parental population) and have recently made secondary contact following the geographical isolation that has mediated their divergence to species level…. Of conceivably equal or greater importance is the process of ecological displacement. It seems clear from an a priori basis that any further ecological divergence lessening competition between the overlapping populations will be favoured by natural selection if it has a genetic basis’. (p. 59)

The conception of the model as limited to the evolution of morphological differences between species that evolved in allopatry is unsurprising if one considers the near hegemony of the model of geographic speciation for much of the last century [for reviews of ecological and sympatric speciation, see Orr & Smith (1998) and Via (2001)]. However, competition-induced selection for differing morphologies within a single sympatric population implies that competition may promote sympatric speciation (see, e.g., Rosenzweig 1978). In fact, simple intraspecific divergence in size could imply the same. Recent evidence of extensive sympatric speciation in several taxa by hybridization (see, e.g. Schliewen & Klee 2004; Seehausen 2004) does not rule out the possibility of subsequent competition-driven selection for morphological differentiation.

Even if trophic polymorphisms associated with specialization (or even intraspecific differences in size) are an evolutionary stable point (Hanken & Hall 1993), we remain with the question of which forces selected for these patterns, including, of course, intraspecific competition. This is in particular because food most commonly promotes the development of alternate trophic morphs (Hanken & Hall 1993). Here, too, a complex picture emerges since intraspecific and even intrasexual morphological differences may also arise from other selective forces.

Although it is supported by theoretical models, there is currently little evidence for the notion that intraspecific competition generates disruptive selection (Bolnick 2004). In a recent experimental study of threespine sticklebacks, using body size and gonad mass as surrogates for fitness, Bolnick (2004) showed disruptive selection in a trophic apparatus (gill-raker length) in a natural intermediate population in one lake and experimented with different levels of population density (hence selection), finding that increased density favoured extreme morphs.

Gaining insight into the role of intraspecific competition in generating intraspecific morphological divergence and consequently sympatric speciation still requires much empirical research. If ecological character displacement is indeed redefined to encompass competition-driven intraspecific morphological divergence without isolation, this opens new horizons for character displacement research.

Ecological speciation and adaptive radiation

Adaptive radiation is time-honoured in the evolutionary literature, but recent years have seen a resurgence of interest in this phenomenon (Schluter 1996a,b, 2000; Givnish & Sytsma 1997). Adaptive radiation, the rise of a diversity of ecological roles and speciation within a lineage [see Givnish (1997) for review of definitions] has been of interest to ecologists and evolutionary biologists alike, and to some it embodies the role competition and its absence play in large-scale evolutionary processes. The diversification of a lineage into species that exploit a variety of different resources (a type of ecological speciation; see Schluter 2001) requires that the different species differ in the morphological or physiological traits used to exploit those resources (Schluter 1996a).

Some would suggest that after colonization of ecologically available territory lineages should experience a burst of rapid evolutionary diversification (Jackman et al. 1999). For example, Schluter (1988) interprets the great divergence in beak size among congeneric granivorous finches in the Galapagos and honeycreepers in the Hawaii Islands as resulting from character displacement, with no competition from distantly related species. Morphological and ecological divergence is facilitated, which can lead to adaptive radiation. This view couches Lack's (1947) original observations in character displacement terms. Likewise, Roughgarden (1995) states that Caribbean anoles occupy a niche filled by birds in North America and Europe. The notion that niches that are empty of distantly related taxa can be taken up by others is shared by many.

In recent years a burgeoning scientific literature on ecological character displacement, ecological speciation, and adaptive radiation focuses on threespine sticklebacks as a model system. Threespine sticklebacks in Canadian post-glacial lakes have been used to test whether speciation during adaptive radiations involves resource-based natural selection and competition, with generally positive conclusions (Schluter 1996a,b, 2001; Rundle et al. 2000).

Meyer (1993) points out that the cichlids in the East African lakes have diversified dramatically and ascribes this phenomenon at least in part to a special morphological feature, a second set of jaws in the back of the buccal cavity, which allows cichlids to become extremely specialized on prey types and may give them a competitive edge. Genetic (mtDNA) variation among cichlid species studied in Lake Victoria, the youngest lake, is very low and supports a hypothesis of intra-lacustrine speciation (Meyer 1993); although nuclear markers reveal much variation, they do not contradict a hypothesis of recent cladogenesis within the lake (Seehausen et al. 2003). The different species in Lake Victoria could have evolved by allopatric speciation during periods of desiccation and by microallopatric speciation, owing to different microhabitat use. These hypotheses are currently less favoured than that of sympatric speciation (reviewed by Galis & Metz 1998), although Meyer (1993) points out that the lake is vast and the species within it mostly restricted in distribution, so intra-lacustine speciation does not refute a model of allopatric speciation. Incipient sympatric speciation has recently been suggested in the Midas cichlid species complex in Nicaraguan lakes (Barluenga & Meyer 2004).

While Liem (1990) asserts that aquatic vertebrate feeding systems are characterized by relaxed competition and the absence of character displacement, Robinson & Wilson (1994) have no doubt that competition acts as a diversifying evolutionary force in freshwater fish communities, intraspecifically when other species are absent and interspecifically when they are present. Robinson et al. (1993) have shown that pumpkinseed sunfishes (Lepomis gibbosus) from an Adirondack lake without bluegill sunfish (L. machrochirus) have differentiated into two morphological forms. We have previously mentioned additional examples, such as that of Bolnick (2004) on stickleback gill-raker length, and some are cited by Robinson & Wilson (1994). In fact, Robinson & Wilson (1994) point to generalizations regarding morphological divergence in fish: the phenomenon dominates in species-poor communities and most commonly involves benthic and pelagic forms ‘which suggests that most lakes offer a similar array of habitats and resources that can be regarded as ‘niches’ that exist apart from the species that inhabit them’ (Robinson & Wilson 1994; see also McDowall 1998; Dynes et al. 1999; Rundle et al. 2000; Gray & Robinson 2002, and Jastrebski & Robinson 2004, for replicated trophic polymorphisms).

Experiments and ‘natural experiments’– on evolutionary processes and rates

Experiments and ‘natural experiments’ can offer compelling evidence for evolutionary processes, as well as insight regarding evolutionary rates. We refer here to experiments on the phenomenon of character divergence rather than with experiments that support the notion of competition, valuable as they may be.

Natural experiments can occur at a variety of temporal scales. For example, the two species of threespine sticklebacks (Gasterosteus aculeatus complex) inhabiting small lakes of coastal British Columbia evolved at the end of the Pleistocene into benthic and limnetic morphs that co-occur in some lakes and that are intermediate in body form where they occur on their own, exploiting both habitats (Schluter & McPhail 1992; Taylor & McPhail 2000). This ‘natural experiment’ therefore provides a clear time frame.

Diamond et al. (1989) explored another ‘natural experiment’, in which a major volcanic eruption occurred about three centuries ago on Long Island, off the coast of New Guinea. The eruption removed most vegetation and apparently did the same to two adjacent islands. These three islands are now the only ones inhabited by two congeneric honeyeaters (Myzomela pammelaena and M. sclateri); Diamond et al. (1989) suggest that the honeyeaters colonized these islands from allopatric populations on other islands and found that the larger species became significantly larger in sympatry, while the smaller became significantly smaller.

Introduced species provide excellent ‘natural experiments’ to study rates of size change in response to changes in competition. Yom-Tov et al. (1999) found evidence for rapid evolutionary change in murids on Pacific and New Zealand islands within 150 years of their entering new ecological communities; some of these changes can be ascribed to character displacement. Simberloff et al. (2000) found a pattern of character release in canine and skull sizes of the small Indian mongoose (Herpestes javanicus) within a century of its introduction to various islands.

At an even shorter time scale, Sidorovich et al. (1999) found in a 10-year study that, when American mink (Mustela vison) were introduced to a local carnivore community in north-eastern Belarus, European mink (M. lutreola) increased in size while the invader decreased in size. Thus size convergence occurred rapidly.

Planned experiments, of course, have the advantage of controls and replicates (cf. Underwood 1990). Experimenting with morphological evolution is a major challenge that has so far been met by few elegant studies that take advantage of specific amenable systems. The threespine stickleback system provided the background for experimental research on character displacement (Schluter 1994). The target experimental population comprised an equal number of hybrids of a solitary species of intermediate form with both limnetic and benthic species, as well as crosses between intermediate morphs, in order to produce wide morphological variation. Limnetic morphs were added to this population, and individuals most affected in growth by several months of increased competition were those most similar to the limnetic morph (Schluter 1994). This study was criticized for a variety of reasons (Bernardo et al. 1995; Murtaugh 1995), most significantly that it confounds introduction of a heterospecific with an increase in total fish density (Bernardo et al. 1995), but the key result of frequency-dependent selection during character displacement was confirmed by Schluter (2003). Bolnick (2004) recently used body size and gonad mass as surrogates for fitness and experimented with different levels of population density (hence selection) of intermediate morphs, finding that increased density favoured extreme morphs.

Gray & Robinson (2002) experimented on a pair of divergent brook sticklebacks (Culaea inconstans) and ninespine sticklebacks (Pungitius pungitius) from postglacial Ontario lakes. Brook trout are found in sympatry with ninespine sticklebacks but also in allopatric populations that differ in resource-related traits. Gray & Robinson (2002) measured the short-term fitness (growth) response of individuals from sympatric and allopatric populations while enclosed alone and with ninespine sticklebacks. They found that growth was greater in absence of ninespine sticklebacks and that sympatric forms did better in the presence of ninespine sticklebacks, suggesting ecological character displacement has indeed occurred.

Pritchard & Schluter (2001) later conducted a pond experiment with threespine sticklebacks to test whether competition between species declines as character divergence proceeds, yielding descendants whose present-day interaction is a ‘ghost’ of its former strength. They put marine sticklebacks together with intermediate sticklebacks (allopatric population in lakes, simulating ‘pre-character displacement’) and with benthic sticklebacks (from sympatric populations, simulating ‘post-character displacement’). Growth rate and niche specialization of marine sticklebacks were higher in the ‘post-displacement’ treatment than in the ‘pre-displacement’ treatment, suggesting a progressive decline in competition strength (Pritchard & Schluter 2001).

With fish, Schluter (1994) used closely related species that have only recently diverged and for which crosses are still possible. Alternative experimental conditions are where populations contain polyphonic (polymorphic) individuals (Pfennig & Murphy 2000, 2002, 2003). Tadpoles of two species of spadefoot toads (Spea bombifrons and S. multiplicata), each of which has an omnivore and a carnivore morph, provided such a system. Under experimental conditions, Pfennig & Murphy (2000, 2002) found that, when the new species were raised together, S. bombifrons tended to have mostly the carnivore morph while S. multiplicata mostly the omnivore morph. Kept separately, both species exhibited both morphs. Moreover, in natural ponds the carnivorous morph of sympatric S. bombifrons is significantly more carnivore-like than the same morph in allopatric S. bombifrons. Conversely, sympatric S. multiplicata are significantly more omnivore-like than allopatric individuals of the same morph, and in both species these differences were found only in trophic structures (Pfennig & Murphy 2003). Moreover, the genetic basis for character displacement in these species finds support in local genetic adaptation (Pfennig & Murphy 2002).


In the past 20 years, ecological character displacement and community-wide character displacement have retained a focal role in studies of community ecology and evolutionary biology. Within a few years after the early 1980s debates, research on ecological character displacement has retaken centre stage. Studies of character displacement have proliferated, the majority addressing criticisms and concerns levelled at earlier research. Several dozen well analysed cases now support the notion that close competitors coevolve to increase size differences between them, or that ecological guild members tend to be overdispersed in sizes of their trophic apparatus. Experimental studies and studies with a strong phylogenetic basis are exciting developments of the past two decades. Most significantly, the overwhelming majority of published studies are statistically tested. Moreover, most studies include detailed consideration of aspects of functional morphology, some also testing actual resource partitioning.

Studies of ecological character displacement and community-wide character displacement span a wide range of taxa including plants; the majority support the hypothesis. The studies, however, come in taxonomic clusters of well-studied test-cases, notably carnivorous mammals, Galapagos finches, Anolis lizards on islands, threespine sticklebacks, and snails. Vertebrates still take the lead, whereas evidence for character displacement in many species-rich invertebrate taxa is still lacking.

Many issues still require research. The frequency with which species assortment also produces regular patterns, a major concern of community ecologists, is still far from clear. Further studies are required to elucidate the significance of size in assembly of ecological communities and in community invasibility. There is a dearth of research on character displacement manifested by shape rather than size and on the relationship of heritable intraspecific differences in morphology to resource use. Alternative selective forces have been suggested that generate similar non-random morphological patterns and this possibility requires more research. Understanding patterns of intraspecific variation in the theoretical framework of ecological character displacement requires much more study. In particular, the role of competition-induced morphological divergence in promoting sympatric speciation is virtually an untapped field.

We predict that ecological character displacement will remain an important part of community ecology and evolutionary research; the combination of functional morphology, phylogenetic insight, ecological and behavioural field research, historical studies, and experimental research has added depth and evolutionary significance to this scientific endeavour. We foresee an exciting future for this interface between the fields of community ecology, functional morphology, and evolutionary biology.


This manuscript benefited from discussions with James Hanken and comments from two referees. T. Dayan is grateful to M. Tebo and to O. and D. Bar-Yosef for their warm hospitality.