How Often Does Sympatry Affect Sexual Isolation inDrosophila?

Abstract

In a seminal article and its follow-up, Coyne and Orr (1989, 1997) found greater prezygotic (primarily sexual) reproductive isolation in sympatric Drosophila species pairs than in allopatric pairs of similar genetic divergence. Several processes can explain this observed pattern. For example, natural selection could have strengthened the existing mating discrimination between sympatric species to reduce either maladaptive hybridization (Dobzhansky 1940) or interference of mate recognition signals (Howard 1993). Similarly, achieving stable sympatry may require a minimum level of prezygotic isolation, hence selecting for groups that are more sexually isolated (Templeton 1981). Ecological character displacement in sympatric populations can also produce prezygotic isolation as a correlated response (e.g., Dodd 1989; Rice and Salt 1990). For simplicity, I will refer to all of these processes jointly as ‘‘reinforcement.’’ My use of this term refers solely to the pattern of greater prezygotic isolation among sympatric species and not to any particular process. Although the Coyne and Orr (1989, 1997) data sets clearly show that sympatry influences prezygotic isolation, the frequency with which interactions among sympatric taxa cause increases in prezygotic isolation is not clear. Some of the species included in their studies were used in multiple comparisons with other sympatric species. This practice could overestimate the number of independent reinforcement events for the following reason. Reinforcement, in the form of direct selection for mating discrimination, could occur by reducing the range of phenotypes acceptable as mates, perhaps by increasing the level of female discrimination (Ratcliffe and Grant 1983; Noor 1995a; Kelly and Noor 1996). This mode of reinforcement would fortuitously increase prezygotic isolation from all other species also. Hence, a single sympatric interaction might cause the appearance of numerous cases of reinforcement in studies that invoke multiple comparisons, such as those of Coyne and Orr. The comparative method may be used to estimate more accurately the fre-