The ecological transition in speciation


  • Donald A. Levin

    Corresponding author
    1. Section of Integrative Biology, University of Texas, Austin, TX 78731, USA
      Author for correspondence: Donald A. Levin Tel: +1 512 471 4685 Fax: +1 512 232 3402 Email:
    Search for more papers by this author

Author for correspondence: Donald A. Levin Tel: +1 512 471 4685 Fax: +1 512 232 3402 Email:


Ecological transitions are at the core of different modes of speciation. These transitions face both genetic and demographic hurdles. This paper focuses on how these hurdles are overcome, allowing ecological speciation and speciation via hybridization and/or polyploidy. Niche shifting is a two-step process. First there is the establishment of ill-adapted populations where ecological opportunity allows. This is followed by the genetic refinement of populations, which allows them to be integrated into novel communities and habitats. These steps are more readily accomplished in unsaturated floras, where competition is less intense. Ecological transitions in saturated floras may be facilitated by disturbance. Invasive species serve as heuristic model systems for understanding the early stages of speciation where niche shifts are involved.


Changes in the ecological attributes of populations, especially involving habitat and resource utilization, are an important component of speciation in many if not most flowering plant lineages (Levin, 2000). Indeed, such changes alone may result in the origin of species. Schluter (2000) deems the evolution of new species as a result of divergent natural selection mandated by abiotic or biotic environmental forces as ecological speciation. At some point in time, the divergence of genomes is accompanied by the emergence of post-pollination barriers.

Whereas phylogenetic analyses show the direction of niche shifts, we do not really appreciate how a population may enter a new habitat or exploit new resources, persist for many generations in the face of abiotic and biotic pressures to which it is not well adapted, and undergo genetic and phenotypic changes that enhances fitness and correlatively renders it a new species (neospecies). The purpose of this paper is to provide some insights into the process of local ecological speciation, taking advantage of our knowledge of invasive species. Invasive species serve as good surrogates for understanding processes of colonization and niche shifting. Recombinational speciation and speciation involving polyploidy also include ecological change. Accordingly, these modes of speciation will also be discussed. Before dealing with evolutionary dynamics, it is important to recognize the geographical and ecological contexts in which ecological speciation occurs.

Templates for ecological speciation

Ecological speciation, as envisaged here is a local phenomenon. That is, it most likely to occur within one or a few populations or a metapopulation. Ecological speciation is not likely to involve the transformation of an existing geographic race through the spread of novel genes from one local population to another. This is because genes conferring fitness in a new habitat would be detrimental to the race in its present habitat.

Ecological shifts are best known on oceanic islands, where species in several genera have radiated from the ancestral habitat into numerous others (e.g. Hawaiian silverswords, Tetramolopium and Cyanea; Wagner & Funk, 1995). Substantial ecological radiations also have occurred in continental genera, especially in areas recently vacated by glaciers (e.g. Espeletia; Monasterio & Sarmiento, 1991).

The manifest and rapid ecological diversification on oceanic islands and areas recently liberated by glacial retreat presumably occurred because of the ability of lineages to take advantage of ecological opportunities afforded in unsaturated floras (Levin, 2003). Open habitats and unsaturated resource spaces could be exploited because the flora was sparse and competition was low. The idea that there was greater ecological opportunity on islands than on continents is supported by the observation that alien species usually constitute a much larger percentage of the flora on distant islands than they do on continents (Heywood, 1989).

If ecological opportunity has been greater on islands for extended periods of time, then speciation rates should be higher on islands than on continents. Schluter (2000) described a speciation rate differential consistent with this argument. Upon comparing the species richness of plant lineages that radiated in isolated archipelagoes with the richness of their mainland sister lineages, he found that diversity usually was greater in the island lineages.

Ecological opportunity may also vary among islands in the same archipelago. It is likely to have been higher on younger islands, especially when there were substantial differences among islands in plant and animal diversity per unit area. This being the case, speciation rates should be higher on younger islands than on older ones. This indeed tends to be the case, as seen in the Hawaiian archipelago (Levin, 2000). The estimated speciation rate is a negative function of island age varying from 0.20 species per lineage per million years (Myr) on Kauai (5.7 Myr old) to 2.1 species per lineage per million years on Hawaii (0.5 Myr old). The same pattern also is seen in the Juan Fernandez group. The estimated speciation rate per lineage on Mastierra (4 Myr old) is 0.33 species per lineage per Myr vs 0.96 species per lineage per Myr on Masafuera (1–2 Myr old; Levin, 2000).

The notion that younger islands are better arenas for speciation is supported by phylogenetic analyses in the Hawaiian silverswords. Here we see that the bulk of the speciation events probably occurred early in the history of the group when the flora was not saturated (Baldwin et al., 1998).

Although islands get a lot of attention, we know that ecological speciation also has occurred in saturated, continental floras. This is most apparent in regions with large numbers of endemics, such as the Cape Floristic Province of South Africa. Of the 90 000 plants species in this flora, 69% are endemic (Goldblatt & Manning, 2000). This diversity is concentrated in rather few genera that have radiated profusely by finely subdividing the abiotic and biotic niche space. The majority of endemics are edaphic and/or pollinator specialists. Thirteen genera, each with more than 100 species, account for 25% of all species in the flora (Cowling & Pressey, 2001). This diversification has occurred since the Pliocene, and thus has been quite rapid. The recent genesis of species clusters results in poor phylogenetic resolution in many clades. Rapid ecological radiation in the Cape flora is thought to occur in peripheral isolates in response fire and other factors that cause environmental perturbation.

Lapeirousa is one genus in the Cape Province and beyond that has undergone manifest ecological speciation (Goldblatt & Manning, 1996). The genus has diversified most profoundly in the edaphic niche and in the pollinator niche. Edaphic change has been from coarse and sandy substrates to clay soils. In some lineages bee and generalist pollinators have supplanted long-tongued flies and moths. Both ecological ‘migrations’ have occurred in some lineages.

The adaptive divergences that ultimately culminate in the formation of ecological species are apt to precede the development of strong post-pollination barriers between related lineages. This is indicated by the permeable barriers to gene exchange in many island assemblages. In the silverswords, for example, species cross readily and pollen fertility in hybrids of congeneric species varies from 30% to close to 100% (Carr, 1998). Even intergeneric hybrids have pollen fertilities from 10% to 30%. Most hybrids are robust. The fertility of F1 hybrids is positively correlated with the genetic identity of taxa, as estimated from allozyme data (Weller et al., 2001).

Post-pollination barriers between neospecies and their progenitors apparently arise after ecological differentiation as byproducts of divergence in nuclear genomes, cytoplasmic genomes, and chromosomal complements (Levin, 2000). Evidence that these barriers may arise rather rapidly comes from studies on wild plants and their domesticated derivatives. Cultivated barley, rice, amaranths, lettuce, soybean, and pea are among a long list of domesticates with partial barriers to gene exchange with their wild relatives (Smartt & Simmonds, 1995). It is not clear how long it took for barriers to emerge in these and other crops, but it probably took less than a few thousand generations. In Phlox drummondii, domesticates only 150-yr-old are partly cross-incompatible with their wild ancestors (Levin, 1976). Many of the changes implemented in artificial selection programs involved traits associated with plant life history, habitat preferences, morphology, and reproductive biology. These are the same kinds of traits whose divergence forms the basis for ecological speciation in wild plants.

Domesticated–wild taxon pairs provide more information on the incidental development of hybrid inviability, weakness and sterility than do wild progenitor-derivative pairs or phylogenetic trees. The former requires less guesswork about the path of evolution and the timing of lineage splitting.

The process of ecological speciation

Ecological speciation is envisioned as a two-stage process. The first stage involves the colonization of a new habitat or the invasion of a new community. Even if populations persist for several generations, they are unlikely to be well-adapted to the substrate or to their new competitors, pollinators or pests. The second stage of ecological speciation involves genetic and phenotypic refinement that brings the new lineage into accord with its physical and biological environments, and that in turn elevates it to the level of species.

The successful achievement of both stages of ecological speciation is opposed by genetic obstacles. Colonization requires the requisite genetic variation, and this is likely to be lacking in the source populations. Seeds are dispersing into novel habitats all time, but almost invariably they fail to germinate or the plants die before reproduction (Bradshaw, 1991). Even if a very rare genotype became established in a new environment, it probably would fail to mate with others of like tolerance because establishment is likely to be very sporadic in space and time.

Even if colonization were achieved, the probability of long-term persistence is quite poor, unless conditions there were quite benign. Environmental and demographic stochasticity may cause the demise of the lineage in the not too distant future, if population size was not large or population growth rate was not considerable (Gomulkiewicz & Holt, 1995). Population growth is also required for the rapid genetic refinement of the population, because directional selection might be expected to cause populations to shrink (Reznick & Ghalambor, 2001).

Genetic refinement, the second stage of ecological speciation, may be difficult to achieve in ecological outliers because they have experienced genetic bottlenecks and thus are apt to have depauperate gene pools. Genetic studies on alien species support this supposition (Barrett & Husband, 1990). Even populations with large effective sizes may be constrained in undergoing genetic modification simply because they lack the appropriate variation. The common failure of artificial and natural selection to alter the habitat preferences of populations is ostensibly due this absence (Bradshaw, 1991).

A niche shift yielding a level of adaptation required to become integrated into a new community may take many generations, with the number depending on the genetic resources and intensity of selection. The gradual evolution of local adaptations has been observed in many species that recently invaded sites with high concentrations of heavy metals (Davies, 1993). The progressive ability of populations to tolerate salinity and herbicides has also been documented.

The probability of making a successful ecological transition declines the longer small, marginally fit populations persist in a new habitat without achieving increased adaptedness, because these populations are subject to extinction. Small populations with low growth rates are especially vulnerable to extinction via demographic and environmental stochasticity (Gomulkiewicz & Holt, 1995). Because strong directional selection might cause them to shrink, population growth is a prerequisite for the long-term survival of populations and for their rapid evolution (Lande, 1998).

Genetic refinement of ecological outliers is also opposed by immigration. Immigration frustrates local adaptation (Kirkpatrick & Barton, 1997). The closer the progenitor is to the ecological outlier, the greater will be the level of immigration. This differential would be magnified if outlier populations were small, because the immigration rates in small outlier populations would be proportionally higher than in larger outlier populations in the face of a constant pollen and seed rain from the progenitor (Levin, 2000). Given that gene flow opposes adaptation, it follows that ecological speciation is more likely to be completed in habitats distant from source populations than in neighboring habitats.

Genetic correlations (via pleiotropy or linkage) between traits related to fitness also may constrain genetic refinement in would-be invaders. Such correlations reduce the response to selection on a trait or trait combination (Antonovics, 1976). Complex correlations have been described in several plants (Waitt & Levin, 1998). The response to selection may also be constrained by developmental and functional correlations.

The rather static ecological posture of species in their response to long-term environmental change during the Quaternary indicates the presence of strong constraints on niche shifting. Species have two responses to environmental change, migration or evolution. The former has been the response of choice in most, if not all tree species (Huntley, 1991).

Niche differentiation associated with hybridization and polyploidy

Not all avenues of speciation that involve niche shifts are included within the realm of ecological speciation. Where speciation is associated with hybridization and/or polyploidy, ecological divergence is coincident with the formation of post-pollination barriers. These avenues of speciation will be considered here, because niche shifts are likely prerequisites for the establishment of the new entities.

Species arising from the stabilization of hybrid derivatives (recombinational species) have ecological amplitudes that differ from those of their parental species (Rieseberg, 1997). This result is expected because hybrid species combine the genomes of two ecologically distinctive entities. However, the habitats occupied by the hybrid entities often were novel or extreme relative to those of the parental taxa. Accordingly, having the genomes of two species does not necessarily mean that recombinational species will have intermediate ecological tolerances.

The three products of hybridization between Helianthus annuus and Helianthus petiolaris have novel ecological preferences (Rieseberg, 1997). The latter is more xerophytic, occurring on sandy soils rather than the heavier, wetter soils preferred by H. annuus. Two hybrid species, H. anomalus and H. deserticola, are more xerophytic than H. petiolaris, while H. paradoxus grows in saline and brackish marshes.

The acquisition of unusual ecological preferences in recombinational species is a subject of much interest. Rieseberg et al. (1999) showed that hybrids often have attributes that transcend those of both parental species, which they attribute to transgressive segregation. Examples of transgression include tolerance to cold, heat, drought, and heavy metals. Once generated, natural selection can stabilize the ‘extreme’ ecological, morphological, physiological features, and combinations thereof.

The stabilization of novel character packages is unlikely to occur in habitats to which the parental species are well-adapted. The hybrid population would either be out-competed or it would be swamped by one or both species. Rather, stabilization most likely depends on the penetration of new communities, or by the alteration of the parental communities, both of which would provide some degree of isolation from the parental species. Both forms of invasion would be facilitated by disturbance.

Turning to polyploids, the ecological amplitudes of diploids and autopolyploids may be variously disparate (Levin, 2002). Cytotypes may differ in tolerances to high or low temperature, drought, salinity, shade and herbivory. Cytotypes also may exploit different pollinators. Alternative visitation practices are related to differences in floral structure and rewards.

Why do related cytotypes exploit different niches? The most obvious answer is that autopolyploids may be preadapted for habitats and resources off limits to their progenitors. Just because chromosome doubling may change the ecological tolerances of populations does not mean that populations will be finely tuned to a novel habitat. However, it does mean that polyploids might get a toehold in one such habitat. This would be followed by a period of genetic refinement through natural selection.

The arguments made with reference to niche shifts associated with recombinational speciation and autopolyploidy may be combined and applied to allopolyploids. Allopolyploids often grow in habitats that are different, but not simply intermediate, to those exploited by their progenitors (Levin, 2002).

The raw materials for the ecological divergence of polyploids from their parents are generated by recombination in diploid hybrids and/or in polyploids and from the effects of chromosome doubling. Newly generated allopolyploids may be somewhat preadapted to exploit novel habitats. Once transported to them, selection will hone them into ecologically proficient entities. Disturbance would facilitate the persistence of hybrids across generations, the establishment of neoallopolyploids, and their entrance into new habitats.

The role of disturbance in ecological speciation

Although ecological opportunity seems to be substantial in unsaturated island floras and in floras developed in areas recently vacated by glaciers, there is less opportunity in saturated continental floras. There are more species and more individuals contesting for limited resources. Even if populations have the ability to enter a new habitat, they are unlikely to do so because of the demographic and genetic hurdles they must surmount. This bodes poorly for ecological speciation in saturated floras. Yet, speciation based on changes in niche exploitation has been a prominent feature of angiosperm evolution. How is this possible when the invasibility of communities declines as communities mature? (Rejmánek, 1989). How is it possible for ecological opportunities to be enhanced in relatively closed communities?

I propose that ecological speciation in relatively saturated communities may be facilitated by disturbance, because it relaxes the competitive pressures on invading populations. Disturbance allows marginally adapted immigrants an opportunity to become established and form reproducing populations. These populations may then evolve greater levels of adaptedness, thereby placing them on a speciation trajectory. This argument assumes that speciation occurs locally, in one or a few populations or in a metapopulation.

Disturbance makes communities more invasible. It does so primarily by disrupting strong species interactions, thereby reducing competitive pressures between ecological associates (Crawley, 1987). Disturbance may also negatively impact the viability of native populations. By altering the composition and density of the local flora, invaders may be released from the prevailing levels of herbivory and disease.

There are numerous niche axes (abiotic and biotic) along which ecological speciation may occur in saturated communities. Two axes, the substrate and the pollinator fauna, stand out above the rest.

It is relatively easy to envision how an edaphic niche shift may proceed. Disturbance in a community built on one substrate may allow invasion by species adapted to a different substrate. Eventually, the ability of the invader to exploit the new substrate increases to the point where the species can hold its own in the new community without disturbance.

Ecological diversification can include exploitation of different assemblages of pollinators. Changes in pollinator exploitation require alterations in the pollination environment that would impose new selective pressures on floral traits. I propose that disturbed areas would provide marginal conditions for the prevailing pollinators. In these areas, the level of competition between genera for pollinators is likely to be much less than before disturbance, because floral resources for them would be less abundant. Pollinators that might not usually visit a plant might do so if its principal nectar source was scarce or absent. Pollination niche transfer does occur when patch size declines and interpatch distance becomes greater (Murcia, 1996). Pollinator responses to fragmentation are relevant here, because both disturbance and fragmentation generate discontinuous habitat patches that are distinct from the landscape in which they are embedded.

Inadequate pollinator service also may promote a shift toward autogamy. Autogamy contributes to the reproductive isolation of species by reducing the potential for interspecific pollen exchange.

Neospecies as invaders

Invasive species can serve as model systems for gaining a new perspective on ecological speciation and on the conditions that promote or thwart it. Both successful speciation and invasion involve the introduction of a population into a new habitat, the establishment of reproductive populations, and the expansion of the entity.

All ecological neospecies, by definition, are ecologically divergent from their progenitors. All ecological neospecies must invade niches off limits to their progenitors. The occupation of new habitats does not occur in the absence of other plants, animals, fungi, and microbes. Rather, neospecies must thrive among a broad array of ecological associates with which the progenitor had little interaction, as well as cope with a new physical environment. Here is the parallel with invasive species, which also must become established in new habitats and expand in the face of ecological associates to which they were not previously exposed.

Ecological opportunity is a requirement for would-be neospecies and would-be invaders. Disturbance provides this opportunity in saturated floras (Hobbs, 1989). It follows, and indeed is observed, that the number of invasive plant species is the highest in the very early stages of a successional sequence and then declines exponentially as the vegetation changes (Rejmánek, 1989).

Disturbance typically is required for the colonization and subsequent expansion of aliens (e.g. Bromus tectorum and Poa pratensis; Mack, 1989). For some invaders (e.g. pines of Australia and South Africa) habitat disruption seems only necessary for colonization; and subsequent spread into a nonperturbed version of the same community can occur without it (Richardson & Higgins, 1998). It is notable that species may invade habitats quite distinct from those in which they became established. For example, in Australia the exotic shrub Acacia nilotica initially invaded riparian areas disturbed by cattle, but then moved into adjacent upland areas (Brown & Carter, 1998).

The fact that alien species requiring disturbance for establishment may invade more mature communities is important for the speciation model I am proposing. It means that incipient ecological neospecies are not necessarily restricted to disturbed habitats. They may be able to better function in more closed communities as they undergo genetic refinement.

Neospecies conventionally are viewed as monophytic entities that differentiate little during the early course of their evolution. Information on invasive plants suggests that we re-examine this notion. The matter of monophyly will be considered after some background on invasive species.

Many, if not most, exotic species have arrived in a given region from multiple source populations. These independent lineages may then interbreed, thereby exchanging genes and enriching local gene pools. The spectacular expansions of Lythrum salicaria in North America and Echium plantagineum in Australia seem to have been fostered by amalgamation of populations originating in different parts of Europe (Barrett, 2000).

We are used to thinking that ecological alterations leading to a new species have to be singular events. Why does this have to be so, if alien species can undergo multiple colonization of sites somewhat divergent from sites at home? Ecological speciation could involve recurrent shifts from the parental habitat to a new one. There is an example of this in the evolution of the sand dune endemic Helianthus anomalus, which is a product of hybrid speciation. Lineages generated in hybrid swarms involving H. annuus and H. petiolaris thrice made the transition to the more xeric sites that H. anomalus prefers (Schwartzbach & Rieseberg, 2002).

There is a tacit assumption that ecological differentiation occurs only in more mature species. We do not see mention of ecotype or race formation in species purported to be of recent origin. The idea that neospecies may not undergo ecological diversification is challenged by the observation that alien species often undergo rapid and conspicuous genetic differentiation. For example, multitrait latitudinal clines have evolved in Solidago gigantea and Solidago altissima, which were introduced to Europe from North America about 250 yr ago (Weber & Schmid, 1998). Clines were observed in garden populations for shoot height, leaf and inflorescence sizes, and phenology.

In closing, our understanding of speciation has advanced considerably in recent decades as a result of mapping ecologically relevant traits onto molecular phylogenetic trees and of studies on the genetic architecture of traits. In some groups we have insights into the direction of niche shifts and on the number of genes or quantitative trait loci involved in single character changes. However, little attention has been given to the population biology of ecological transitions associated with many modes of speciation. Knowing the direction of change and the number of genes controlling some character changes does not fully illuminate the evolutionary process. We must continue to ask what is involved at the population level in transforming one species into another.