COMMENTARY Hybridization and the build-up of genomic divergence during speciation J. L. FEDER*, S. M. FLAXMAN † , S. P. EGAN* & P. NOSIL ‡ *Department of Biological Sciences, University of Notre Dame, South Bend, IN, USA †Department of Ecology and Evolutionary Biology, University of Colorado, Boulder, CO, USA ‡Department of Animal and Plant Sciences, University of Sheffield, Sheffield, UK Abbott et al. (2013) extensively review our understand- ing of the roles of hybridization in speciation. The topic is discussed in terms of both the homogenizing effects hybridization may have impeding population diver- gence (e.g. Hendry et al., 2001; Taylor et al., 2006; Seehausen et al., 2008) and its creative potential for generating new biodiversity via the production of novel genotypes/phenotypes (e.g. Rieseberg et al., 2003; Seehausen, 2004). It is the latter topic that is the focus of the review and is an aspect of hybridization that is garnering increased attention and acceptance in recent years, not just in plants, but also in animals. As the creative aspects of hybridization are admirably covered in Abbott et al.’s (2013) study, we instead con- centrate on the conditions under which the homoge- nizing effects of hybridization can be overcome to result in population divergence. This is a fundamental question regardless of the source of variation underly- ing divergence or the geographical mode of differentia- tion (initial differentiation in allopatry, parapatry or sympatry). The term ‘speciation-with-gene-flow’ cir- cumscribes the issue and focuses attention on whether at a given level of hybridization – usually defined in terms of the gross migration rate (m) – it is possible for increased reproductive isolation to evolve. From this perspective, speciation-with-gene-flow reduces emphasis on how populations initially evolved to reach a level of reduced m and instead focuses on whether they will diverge further, remain at an equilibrium state, or reverse course and fuse. Selective processes promoting genomic divergence There are three selective processes that generally act to further population divergence in the face of gene flow (reviewed in Feder et al., 2012b). The first is selection acting directly on a gene region. When the selection coefficient (s) favouring an allele in one pop- ulation vs. another is ~m/2 or greater, there is a non- negligible chance that direct selection on the locus alone will act to establish the variant and increase population divergence, and when s ≫ m, there is a high probability (Feder et al., 2012b). Thus, if the sup- ply of new mutations with large selective effects is not strongly limited, then there is little impediment to spe- ciation-with-gene-flow. However, when this is not the case, two forms of genetic hitchhiking can help facili- tate divergence. The first form, termed ‘divergence hitchhiking’ (DH), is due to local reductions in effective gene flow (m e ) for sites physically linked to genes subject to divergent selection. This occurs because divergent selection can reduce gene flow at sites linked to the direct targets of selection before alleles at these sites have a chance to recombine away and introgress into the other popula- tion. This reduction in m e means that a linked site does not require as large an s value to overcome the homog- enizing effects of gene flow. Divergence can also be facilitated by ‘genome hitch- hiking’ (GH). In this case, the combined effect of all genes causing reproductive isolation reduces average m e across the genome. Although the degree of reduced m e will vary across the genome according to the distribution of selected sites, if average m e is reduced enough globally, then divergence is not limited to just strongly selected genes and those tightly physically linked. Rather, diver- gence can accumulate genome-wide. DH and GH are not mutually exclusive, and both are consequences of diver- gent selection. The key difference is the focus on physical linkage to selected sites vs. more widespread divergence arising from global reductions in gene flow. The three processes enhancing population divergence have been recently integrated into a four-phase model for speciation-with-gene-flow (Feder et al., 2012a). In this framework, populations successively pass through ‘phases’ where direct selection on individual genes, DH and GH sequentially assume greater relative importance for generating increased genomic differentiation (Fig. 1). We stress that these different ‘phases’ are not discrete and do not have strong boundaries. Rather, they describe different parts of a continuum of diver- gence where the relative importance of DS, DH and GH changes for facilitating differentiation. A key point that we make here not covered by Feder et al. (2012a) is that these phases can be associated with different species concepts (Fig. 1). Moreover, although the model was developed for de novo speciation in sympatry, it is also applicable for divergence initiated in parapatry or allopatry. Indeed, Wu (2001) described a complementary model involving secondary contact based on different patterns of genomic divergence rather than on the evolutionary processes generating them. A challenge is therefore to equate patterns observed in Correspondence: Jeffrey L. Feder, Department of Biological Sciences, University of Notre Dame, South Bend, IN 46556, USA. Tel.:+1 574 631 4159; fax: +1 574 631 7413; e-mail: feder.2@nd.edu ª 2013 THE AUTHORS. J. EVOL. BIOL. 26 (2013) 261–266 261 JOURNAL OF EVOLUTIONARY BIOLOGY ª 2013 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY doi: 10.1111/jeb.12009