Genetic differentiation among hybridizing populations reflects a complex mixture of factors, including migration dynamics, mating preferences, mutation, ancestral variation, and natural selection. The prospect for understanding the relative roles of these processes in the wild helps explain the preoccupation of evolutionary biologists with hybrid zones.
Although examining gene flow one marker at a time is a useful approach, measuring differentiation at multiple loci in the same populations provides a much richer portrait of evolutionary history. This is because different evolutionary processes can impart distinct signatures on patterns of gene flow. In particular, we expect demographic forces, such as migration, to affect loci across the genome in similar ways. Alternatively, natural selection targets particular genes, generally leaving locus-specific footprints. By studying introgression at multiple loci, these processes can be separated.
This idea is an exciting one for speciation research, since it raises the possibility of distinguishing genomic regions that contribute to reproductive isolation from those that do not. Specifically, genes that confer conspecific mating preferences or reduce fitness in hybrids are expected to show comparatively high differentiation, a prediction that has been confirmed for species with (Rieseberg et al. 1999) and without (Noor et al. 2001a; Machado et al. 2002) stable hybrid zones. Given evidence of genomic admixture from other loci, regions that fail to introgress between species could constitute the barriers that maintain species integrity.
Geraldes et al. (2008) used this rationale to dissect genetic differentiation in the European rabbit, Oryctolagus cuniculus. The two recognized subspecies, Oryctolagus cuniculus algirus and Oryctolagus cuniculus cuniculus, form a zone of secondary contact across the Iberian Peninsula. The authors genotyped restriction fragment length polymorphisms (RFLPs) in the SRY gene region of the Y chromosome in rabbits from 30 populations. Geraldes et al. (2008) also re-sequenced introns from three unlinked, autosomal genes in a subset of individuals from each subspecies for comparison.
Y-linked diversity was primarily partitioned between subspecies, with high differentiation (FST = 0.93) between the two geographical regions on either side of the contact zone. In contrast, only one of the three sequenced autosomal loci showed a high level of structure, with FST values at the other two loci lying near zero. By combining these data with previous surveys of additional X-linked and autosomal markers, Geraldes et al. (2008) demonstrated that this contrast between levels of differentiation extends to other parts of the rabbit genome. From a total panel of 26 loci, five were highly structured among subspecies (including the Y chromosome, one autosomal locus, two X-linked loci, and the mitochondrial cytb gene), whereas 21 loci showed clear evidence of gene flow. Such a ‘bimodal’ distribution of introgression is striking and has only rarely been observed in hybridizing taxa (Turner et al. 2005).
Comparisons of introgression between the Y chromosome, the X chromosome, the autosomes, and mitochondrial DNA are motivated by knowledge of the genetics of speciation. In particular, the established role of the sex chromosomes in reproductive isolation (Coyne & Orr 2004; Presgraves 2008) predicts that markers found on these chromosomes will often show restricted gene flow relative to other loci. Geraldes et al. (2008) add rabbits to a growing list of hybridizing taxa (Tucker et al. 1992; Saetre et al. 2003; Carling & Brumfield 2008) for which this is the case. Importantly, their results suggest that the Y chromosome contributes to reproductive barriers between rabbit subspecies, which implies that isolation is at least partly mediated by effects in males. This hypothesis can now be tested in experimental crosses.
An alternative explanation for variation in population structure among genomic compartments must be confronted. Because of differences in effective population size, relative levels of differentiation might simply reflect the rate of ancestral lineage sorting, which is expected to be fastest for the Y and the mtDNA, followed by the X chromosome, and then the autosomes. A superior fit of coalescent models with recent gene flow over models of isolated populations argues against the idea that most shared polymorphism has been retained from a common ancestral population in rabbits. Using similar approaches, it might also be possible to quantitatively predict differences in structure among the four locus types that would arise from drift alone (e.g. Y-linked polymorphism should sort four times faster than autosomal polymorphism, on average) and attribute observed deviations to differential gene flow and/or skews in the breeding sex ratio.
Although the barrier to gene flow on the rabbit Y chromosome is intriguing, an apparent similarity between the five low-introgression loci might have even broader implications for the examination of hybrid zones. The Y-linked and mitochondrial loci do not recombine, and the X-linked and autosomal factors are located near centromeres, suggesting that they experience reduced recombination as well. If such an association between introgression and recombination turns out to be widespread, it would bear on studies of gene flow in two important ways. First, information about recombination could be used to choose the most appropriate markers for a particular task. Migration rates, for example, should be characterized using genomic regions that recombine regularly to minimize effects of linked selection. Second, the association between recombination and gene flow might reflect a causal connection between crossing-over and speciation. Multiple models predict that when nascent species pass through a parapatric phase, incompatibilities will accumulate in chromosomal rearrangements (Rieseberg 2001; Noor et al. 2001b). Interestingly, reduced recombination is the salient characteristic of these models, suggesting that a more general association between introgression and recombination might be expected, even in the absence of rearrangements (Butlin 2005). Although examining gene flow in species with detailed recombination estimates will be most revealing, applying the logic of Geraldes et al. (e.g. comparing centromeric markers to those located elsewhere in the genome) should be feasible in many organisms.
Multilocus surveys of population differentiation have a bright future in speciation research. A time in which entire genomes can be scanned for introgression in hybridizing species is right around the corner (Turner et al. 2005). Importantly, this approach will still be challenged by the same question that plagues other genomic scans for selection: what does an outlier locus look like (Thornton et al. 2007)? The answer will depend on the evolutionary model, and new model-based analyses that efficiently separate reductions in gene flow caused by selection from neutral variance in introgression and ancestral lineage sorting are needed. If the results from available multilocus surveys are any indication, the genomic tapestry of differentiation among nascent species will be a fascinating and complex one indeed.