Last week, we discussed a paper by Fitzpatrick et al. (2008) with the title ‘What, if anything, is sympatric speciation?’ Most of us felt that the points put forward were a good contribution to the ‘speciation debate’ and agreed that a general shift from the pattern orientated way of classifying speciation (allopatry vs sympatry) towards a more process based system would be fruitful.
Today, I would like to add some thoughts about the four criteria that were proposed by Coyne and Orr (2004), as we did not have time to discuss this during the meeting. I will do this by adding a little marine perspective to it. This is because a) roughly ¾ of the earth is covered by water, and b) approximately half of the vertebrate species are made up of fish (including jawless, cartilaginous and bony fish). I obviously had to deal with these criteria a lot during my PhD, so here comes what I think:
The four criteria that need to be met to corroborate that a species has evolved under sympatry:
1) a sympatric distribution of the most closely related sister-species;
2) genetic evidence for reproductive isolation;
3) lineage monophyly; and
4) an ecological setting in which historical allopatric differentiation in a biogeographical context is very unlikely.
Recent work suggests that the first criterion, 'the sympatric distribution of the most closely related sister-species', may not always be the case. For example, Doebeli and Dieckmann (2003) have shown that when ecological interactions actively drive diversification, then species that diverge in sympatry may subsequently become allopatric (i.e. to reduce competition). In light of this, the sympatric distribution of sister-species may not always be a reliable indicator of historical sympatry (or vice versa).
The fourth criterion is also problematic as it states that there must be an ecological setting in which allopatric differentiation is very unlikely e.g. remote oceanic islands, hosts for parasites or small crater lakes (Coyne and Orr 2004). These are all very geographically constrained and homogenous habitats and do not apply to most species, and in particular the vast majority of marine species. Why should sympatric speciation be any less likely in heterogeneous systems? Recent theoretical research has suggested that it is not (for details see Doebeli and Dieckmann 2004 and references therein). Recent theoretical models of sympatric speciation have shown that speciation may crucially depend on spatial structure. In a single and homogenous population, stochastic fluctuations will only be sufficiently large when the population size is small. Yet, with spatial structure, fluctuations can be considerable, even in a large population. Therefore, a small spatial component may greatly enlarge the potential for speciation, as local adaptation along an environmental gradient has the potential to increase the strength of frequency-dependent selection (Doebeli and Dieckmann 2003; Doebeli and Dieckmann 2004). Furthermore, inferring the historical patterns of speciation events (sensu Barraclough and Vogler 2000) is problematic because geographic distributions change over time (Losos and Glor 2003). Thus, it becomes clear that the current distribution of a species is not necessarily a reliable guide to its historical geographical range, therefore, it is crucial to acknowledge that geographical ranges shift and that geographical signal decays over time.
Another general criticism of the Coyne and Orr (2004) criteria is that there is an underlying burden of proof that requires sympatric models to exclude historical geographic barriers to gene flow. Conversely, no studies of allopatric speciation to date have excluded the possibility of historical sympatric distributions (Berlocher 1998). Researchers seem to be expected to treat allopatric speciation as the null model, even though several empirical studies have convincingly demonstrated sympatric speciation. Dieckmann et al. (2004) proposed that the time has come to do away with the notion of allopatric speciation as the null model, a notion that prevails partly because of the deceptive simplicity of allopatric scenarios. Once the bias towards detecting allopatric speciation in empirical data is removed, the data may actually suggest speciation without complete barriers to gene flow is the more likely explanation of many speciation events. This dichotomy can partly explain why evidence for sympatric speciation in heterogeneous environments is sparse. Given that most species inhabit heterogeneous environments, case studies of speciation events on small islands or lakes tell us little about the processes that act on species in spatially variable environments.
Ruling out that ancestral species may have once been allopatric is particularly problematic in marine fish species, as it is usually impossible to exclude the possibility that sympatric species were historically allopatric (Sponer and Roy 2002). Furthermore, speciation in the face of gene flow has been largely discounted in marine fishes because most species have a planktonic larval phase with the potential to disperse widely and generate genetic homogeneity over large spatial scales (Knowlton 1993; Palumbi 1994). However, there is increasing evidence that larvae do not always disperse long distances and some return to their natal reefs (Jones et al. 1999; Swearer et al. 1999; Swearer et al. 2002). Furthermore, habitat selection at settlement (reviewed in: Montgomery et al. 2001) and assortative mating (e.g. McMillan et al. 1999) can produce reproductive isolation at very fine spatial scales. This evidence suggests that a pelagic larval phase need not preclude the formation of fine-scale genetic structure because behaviour can override the potential for genetic mixing (Taylor and Hellberg 2003).
Before this post gets too long I would like to point to the excellent paper by Bolnick and Fitzpatrick (2007 ) in which they address in more detail the points 2 and 3.
I am sure some of you have some thoughts on the criteria proposed by Coyne and Orr as well, so please comment on this post.