A traditional view of the nature of species holds that populations within species maintain genetic cohesion through substantial intraspecific gene flow and that the whole genome of a species is protected from interspecific gene flow by strong reproductive isolation. Both aspects of this view have been challenged. Reproductive barriers between recently evolved species can be incomplete and permeable to gene flow, and in geographically structured environments, intraspecific gene flow may be limited. Whether species that combine these features can evolve as cohesive evolutionary units remains an open question. In this issue of Molecular Ecology, Palma-Silva et al. (2011) investigate this issue in an ideal system. They characterize gene flow within and between four sympatric species of Pitcairnia bromeliads on isolated rock outcrops, called inselbergs, in Brazil (Fig. 1). They show that despite very little intraspecific gene flow between inselbergs and substantial introgression, each species manages to maintain genetic integrity. Furthermore, certain regions of the genome appear to introgress more easily than others. This, taken together with previous studies of premating isolation in these species (Wendt et al. 2001, 2002), suggests that their reproductive barriers are strong but permeable. These data reinforce recent work suggesting that speciation must be thought of not as a whole-genome phenomenon but rather on a locus-by-locus basis, with neutral loci readily exchanged between species, but genes that contribute to reproductive isolation, so-called speciation genes, relatively unlikely to introgress (Wu 2001; Feder & Nosil 2010; Rieseberg & Blackman 2010; Nosil & Schluter 2011).
Palma-Silva et al. (2011) explored genetic patterns across species and inselbergs using both plastid and nuclear markers. They found a strong phylogeographic signal in the plastid DNA, with only 1 of 25 haplotypes shared between coastal and inland inselbergs, yet 7 haplotypes shared among species. Interspecific gene flow, ancestral polymorphism and homoplasy could all explain this pattern; however, the strong geographical signal, the existence of wild hybrids and the fact that both rare and common haplotypes were shared suggest a major role for introgression. On the other hand, an analysis of nuclear microsatellites, characterized for the two species in which putative wild hybrids have been described (P. staminea and P. albiflos), indicated strong between-species differentiation. The pattern of more frequent introgression of plastid haplotypes than nuclear alleles found here is common in many plants and parallels patterns of mitochondrial introgression in animals (Chan & Levin 2005).
A negative relationship between inter- and intraspecific gene flow has been predicted, because in a large, panmictic population heterospecific alleles entering the gene pool via rare hybridization events are easily swamped (Petit & Excoffier 2009). Estimates of the effective migration rate confirmed that these bromeliads experience extremely limited gene flow (Nem < 1) between populations of the same species isolated on different inselbergs. With such a low Nem, gene flow among populations alone cannot homogenize a species’ gene pool; hybridization with sympatric congeners could easily breach species boundaries at a local scale, and consequently, species cohesion becomes difficult. Yet the four species remain largely intact, at least with respect to their nuclear genomes. This suggests a role for selection in species cohesion because alleles under selection require lower levels of gene flow to spread among populations (Slatkin 1976; Morjan & Rieseberg 2004; Barbaráet al. 2007). In this case, it is the interaction between selection and gene flow that reduces population divergence.
This potential role of selection in maintaining the inselberg Pitcairnia species bears on recent ideas about the genetic architecture of speciation with gene flow and indeed the nature of species (Wu 2001; Via & West 2008; Nosil et al. 2009; Feder & Nosil 2010; Hausdorf 2011). While selection may ensure that certain diverging loci are not shared between hybridizing species, recombination and recurrent hybridization could allow the introgression of neutral alleles. This view of a ‘porous genome’, although not new, has opened up interesting avenues in speciation research when combined with modern genetic tools. Understanding which parts of the genome are most easily shared between species and, conversely, which are retained and why is the next major step for understanding speciation in this system.
Palma-Silva et al. (2011) take a first step towards answering these questions in the absence of genomic resources for these species by looking for asymmetric introgression of individual loci or genomic compartments. Here, the lack of significant cytonuclear associations in P. staminea × P. albiflos hybrids indicates that hybridization occurs in both directions. However, introgression of plastid haplotypes is asymmetric, with P. staminea haplotypes found in P. albiflos nuclear genetic backgrounds but not the reverse. Additionally, two nuclear markers showed biased introgression from P. staminea to P. albiflos. Further investigation of these particular loci is one approach to understanding hybridization dynamics.
A complementary approach is to measure the strengths of barriers to interspecific gene flow. Quantifying when such barriers fail may help explain the directionality and frequency of introgression. It is not clear what reproductive isolating mechanisms are most important in this system, but both pre- and postmating barriers appear to exist (differences in floral morphology and mating systems are summarized in Fig. 2). Focussing again on P. staminea and P. albiflos, the authors note that their flowering times (though with some overlap), pollinators and mating systems differ. Furthermore, hybrids begin flowering earlier, resulting in more F2s than backcrosses, and their flowering period overlaps more with P. albiflos, in keeping with the biased introgression into this species. Evidence for either extrinsic or intrinsic postzygotic isolation, apart from lower germination success in hybrids (Wendt et al. 2001), is weaker, and in fact, the presence of F2 and backcross individuals suggests more limited postzygotic barriers.
This study focused on hybridization between two of the four species, but future work on reproductive isolating mechanisms between other members of the group will be fruitful. Because P. albiflos flowers nocturnally and uses different pollinators than the other three species, it may be expected to have the strongest isolating barriers. However, differences in flowering phenology and P. corcovadensis’s high selfing rate could provide further premating barriers. Quantifying multiple isolating mechanisms in this system would be helpful in understanding the directionality of introgression and identifying the traits and later the loci that are most important for species cohesion.
This system could also allow further tests of recent ideas on the genetic architecture of speciation. The porous genome concept predicts that the genes responsible for divergence between the species should reside within the areas of the genome that are most resistant to introgression, because in fact selection on these genes created the resistance. “Divergence hitchhiking” around these genes can lead to the formation of “genomic islands of speciation” containing not only the speciation genes but other linked loci (Via & West 2008; Feder & Nosil 2010), although the size of these islands of divergence remains contentious (Scascitelli et al. 2010). Particularly interesting would be evidence that the genes in question reside in areas of low recombination, which have received much attention for their ability to keep species-specific alleles together (Ortiz-Barrientos et al. 2002).
Another potential consequence of leaky species boundaries is the sharing of adaptive alleles between species. Alleles that are beneficial in both species should introgress more readily than neutral alleles, and furthermore, introgressed neutral alleles could become adaptive following environmental change (Barrett & Schluter 2008; De Carvalho et al. 2010). Such a mechanism has been proposed to be an important diversifying force in adaptive radiation (Seehausen 2004). The search for such genes in inselberg Pitcairnia is likely to be challenging, but in the light of the similar habitats occupied by all four species is potentially enlightening.
The roles of inter- and intraspecific gene flow during speciation have been extensively modelled, and Palma-Silva et al. (2011) have tested some of these ideas in a fascinating and highly apt natural system. Their findings of extensive plastid haplotype sharing between species, extremely low intraspecific migration between inselbergs and strong but porous barriers to hybridization bear on our understanding of how species persist in the face not only of hybridization but also of low migration between populations. Work on populations in highly subdivided environments such as the Brazilian inselbergs could offer many insights on the genetics of speciation with gene flow.