Adaptive radiation (i.e. the rapid proliferation of an ecologically and morphologically differentiated species assemblage from one ancestral species as a consequence of the adaptation to various ecological niches) is thought to have played a prominent role in organismal diversification (Schluter 2000). The spectacular diversity of forms that has emerged in adaptive radiations and the explosive mode of species formation have fascinated empiricists and theoreticians in likewise manner. This common interest has facilitated the integration of empirical data into large-scale, individual-based, and spatially explicit models. Applying a stochastic model, Gavrilets & Vose (2005) provided theoretical support for common patterns of adaptive radiations including the ‘overshooting effect’, which describes the phenomenon of greater species-richness early in the radiation, or the ‘area effect’ referring to higher diversification rates in larger habitats. The same authors have observed in their simulations that the differentiation at neutral loci between populations of different species is similar to that between populations of the same species. In early phases of adaptive radiations, the spatial structure within a species might in fact be more pronounced than the spatial genetic structure overall. This is due to local gene flow, which removes differentiation in neutral markers. In this issue of Molecular Ecology, Puebla et al. (2008) test some of the predictions and observations of Gavrilets & Vose (2005) in a sympatric species pair of hamlets.
Hamlets are small, synchronously hermaphroditic seabasses in the genus Hypoplectrus (Perciformes: Serranidae) that inhabit coral reefs of the Caribbean Sea. Although up to 12 species have been recognized, hamlets show little morphological differentiation and are primarily distinguished on the basis of body colouration, which is why some authors prefer to list the different types as colour morphs instead of species. At the same time, the diverse colouration has invoked speculations on the speciation process in this group. It has long been hypothesized that hamlet species/morphs have diversified on the basis of a fitness advantage resulting from their ‘aggressive mimicry’ behaviour, whereby predatory hamlets mimic nonpredatory fishes to outsmart their prey (Tresher 1978). In another recent study, Puebla et al. (2007) provided observational evidence for aggressive mimicry. Since the hamlet species under study were genetically distinct and mated assortatively, the authors argued that disruptive selection on colouration — mediated through aggressive mimicry — has been driving hamlet speciation.
In their present work, Puebla et al. (2008) compare the intra- and interspecific genetic structure in two Hypoplectrus species, the barred hamlet (H. puella) and the black hamlet (H. nigricans), from three sites in Barbados, Belize, and Panama (Fig. 1). In agreement with the predictions of Gavrilets & Vose (2005), the spatial genetic structure within H. nigricans is greater than that calculated over all samples and, also, that between H. puella and H. nigricans over all locations. Local gene flow between the sympatric hamlet species is a likely explanation for this observation (Fig. 1). This fits well with the idea that genomes can be rather ‘porous’ with respect to gene flow in neutral markers (Wu 2001). Similarly, in the simulations by Gavrilets & Vose (2005), the different virtual species maintained their divergence in selected loci despite of substantial gene flow at neutral loci. It would now be highly useful to have some of the selected loci in hand (e.g. genes underlying colouration in hamlets) to test whether genetic differentiation is more pronounced there.
Interestingly, the genetic structure of the second species, H. puella, is not in line with theoretical predictions. There are still significant levels of differentiation between the three populations, demonstrating that H. puella is not panmictic in spite of the short, 3-week planktonic larval phase common to hamlets. Yet, these differences in FST based on 10 microsatellite markers are much smaller than those in H. nigricans, and smaller than those between the two species over all locations. One possible explanation for this difference lies in demography. In an extensive field survey, the authors determined that H. puella is about twice as abundant as H. nigricans. Applying an individual-based simulation model, they demonstrate that differences in abundance may indeed influence the spatial genetic structure of the barred hamlet and the black hamlet.
A charming alternative hypothesis is that H. nigricans-type hamlets have evolved in situ and repeatedly from H. puella with aggressive mimicry as the underlying mechanism. There are indeed some reasons to believe that H. puella is the ancestral form (e.g. it does not show aggressive mimicry). The observed pairwise FST values between populations would be compatible with such a scenario. Further support comes from a population assignment test with the computer software structure (Pritchard et al. 2000), which groups the populations of H. nigricans and H. puella from Belize into a single cluster suggesting that there is no genetic structure at all at this locality. Unfortunately, so far there is no phylogenetic support that would corroborate the sympatric speciation scenario. Using mitochondrial DNA, Ramon et al. (2003) uncovered a rather close relationship and substantial haplotype sharing between nine hamlet species including H. puella and H. nigricans. Since the most frequent mitochondrial haplotypes were found in both species, and since haplotypic diversity was comparable, mitochondrial demographic and phylogeographical patterns (see e.g. Barluenga et al. 2006) do not provide enough information to back up sympatric speciation. A future analysis based on nuclear DNA and including all hamlet species might do so, though.
On the basis of their impressive amount of population-genetic, observational, demographic and modelling data, Puebla et al. (2008) point to the importance of local processes in the radiation of hamlets. Colouration, likely under selection due to aggressive mimicry, is a very plausible candidate for an adaptively relevant character: could the colour/mimicry system even be a key innovation, creating the opportunity to radiate? The question is now whether divergent selection on colouration alone is sufficient for ecological speciation to proceed or whether other factors, such as mate recognition, habitat preference and/or specialization to certain types of food, are important, too (see e.g. Gavrilets et al. 2007). Notably, hamlet species do mate assortatively with respect to colouration (Barreto & McCartney 2008; Puebla et al. 2007), and there is at least some indication for trophic specialization despite the generally large dietary overlap between the different taxa (Whiteman et al. 2007).
The findings of Puebla et al. (2007, 2008) make the hamlet assemblage one of the best-documented examples of adaptive radiation in marine fishes. The lack of well-described marine fish radiations is surprising given the multitude of examples in freshwater fishes and the fact that, as with the species flocks of cichlids in East Africa, some of the most spectacular adaptive radiations have occurred in fishes (Kocher 2004; Salzburger & Meyer 2004; Seehausen 2006). One reason for this discrepancy might be that adaptive radiations are simply more apparent in geographically well-defined areas. All of the famous examples — Galapagos finches, Anolis lizards of the Greater Antilles, Hawaiian Drosophila and silverswords, East African cichlids — have occurred on islands (note that, from the fishes point of view, a lake is very much like an island). Interestingly, the marine examples also typically involve geographically confined areas (e.g. Caribbean hamlets, Antarctic icefishes). Perhaps geographical circumscription is a prerequisite for a marine adaptive radiation to occur.