Hybridization may rarely promote speciation

Abstract

Abbott et al. (2013) point out many of the multitude of ways in which hybridization and speciation interact. The classical view of the speciation process is that it occurs most easily in strict allopatry and that it is inhibited by gene flow. Recently, however, the focus of research has shifted into examining ways in which hybridization can actually promote the speciation process. Empirical evidence for the facilitation of speciation by hybridization is often mixed or incomplete, and some of the mechanisms proposed in the review by Abbott et al. (2013) are theoretically poorly understood or controversial. Below, we raise some cautionary notes on these mechanisms. We conclude that the conditions under which hybridization can promote speciation may be more restrictive than the review by Abbott et al. (2013) implies. The possibility that hybridization can, counter to initial impressions, actually promote speciation is certainly intriguing. Homoploid and allopolyploid hybrid speciation, which Abbott et al. (2013) write extensively about, are two processes by which hybridization can enable potentially rapid speciation. Abbott et al. (2013) also, however, discuss several ways in which hybridization may contribute to a more gradual build-up of reproductive isolating mechanisms; these are more controversial. The most classic of these is perhaps the process of reinforcement, whereby the evolution of reproductive isolation is promoted by selection against interspecific matings, including by low hybrid fitness (Dobzhansky, 1940; Servedio & Noor, 2003). Reinforcement has long been recognized as a way in which speciation may proceed, at least to some degree, when hybridization is occurring. A process analogous to reinforcement is ‘adaptive speciation’, in which speciation begins in sympatry and continues as hybridization (in the sense of continued gene flow) between the incipient species is occurring. Mathematical models of this process are in many ways identical to reinforcement models, albeit with different starting conditions; they can thus be discussed together. Many theoretical models have found favourable conditions for reinforcement and adaptive speciation, and empirical evidence also suggests that reinforcement, at least, may not be uncommon (see Coyne & Orr, 2004). One important question that relates to whether hybridization promotes speciation is whether reinforcement merely allows speciation to proceed despite gene flow or whether it actually speeds up the process of divergence. There is some evidence, though taxonomically restricted, that suggests the latter (e.g. Coyne & Orr, 1989, 2004). However, there are a number of theoretical reasons, increasingly well understood in recent years, to be cautious about the generality of this finding. The spread of alleles leading to increased reproductive isolation has been found to be very slow in several mathematical models of speciation with gene flow (e.g. Bolnick, 2004; Proulx & Servedio, 2009), and other models suggest that the degree of trait divergence or reproductive isolation that evolves may be limited (e.g. Kirkpatrick & Servedio, 1999; Matessi et al., 2001; Kirkpatrick, 2000; Pennings et al., 2008; Servedio, 2011; but see Bank et al., 2012a). These limitations are often due to the fact that the evolution of premating isolation during reinforcement and adaptive speciation relies on linkage disequilibrium between the genes that cause reproductive isolation and the genes under selection (those, for example, that cause low hybrid fitness); this genetic association is often weak, whereas gene flow itself is relatively strong (Kirkpatrick, 2000; note that, in reinforcement models, weaker gene flow reduces the occurrence of selection against hybrids as well, so the qualitative relationship between the forces does not change). Additional obstacles to speciation with gene flow are created by positive frequency-dependent sexual selection undermining species coexistence, stabilizing sexual selection acting against the divergence of mating strategies, the loss of genetic variation in traits underlying assortment and the effects of search costs (e.g. Matessi et al., 2001; van Doorn et al., 2004; Kirkpatrick & Nuismer, 2004; Burger & Schneider, 2006; Kopp & Hermisson, 2008; Otto et al., 2008; Pennings et al., 2008). There are several mitigating factors that may increase the likelihood or extent of reinforcement and adaptive speciation, including very specific types of selection (e.g. van Doorn et al., 2009), certain mutational step sizes of assortment combined with specific geographic conditions (e.g. Rettelbach et al., 2011; Servedio, 2011; Bank et al., 2012a) or the presence of ‘magic’ traits that are both under divergent selection and are targets of assortment (Gavrilets, 2004). It is not known how often any of these mitigating conditions will occur. Abbott et al. (2013) also suggest that hybridization may contribute to speciation by acting as a source of Correspondence: Maria R. Servedio, Department of Biology, University of North Carolina, Chapel Hill, NC 27599, USA. Tel.: +1 919 843 2692; fax: +1 919 962 1625; e-mail: servedio@email.unc.edu