A negative correlation between trait and mate preference in Xiphophorus pygmaeus

Abstract. Variation in female mating preferences was compared with variation in male traits across populations of Xiphophorus pygmaeus. Females of this species prefer the large courting males of X. nigrensis to their own small, non-courting males. It was predicted that if large, courting X. pygmaeus males were to occur, they would be preferred. The discovery, distribution and subsequent spread, of large X. pygmaeus males provided an unusual opportunity to analyse how sexual selection might act on a trait not currently fixed in a species. The present study showed variation between populations in female preference for large male body size that was negatively correlated with the presence of large males. This variation in preference persisted over a 5-year period. Female preference for larger body size and courtship are evolutionarily decoupled from the male traits of large size and courtship in X. pygmaeus. This pattern is not consistent with indirect selection models (e.g. runaway selection), which predict that female mating preferences and male traits evolve in concert due to a genetic correlation between preference and trait.? 1996 The Association for the Study of Animal Behaviour Several models have been proposed to explain how female mating preferences evolve (Kirkpatrick & Ryan 1991; Maynard Smith 1991). Good genes and runaway models predict that female prefer-ences coevolve with preferred male traits because of a genetic correlation between the preference and the trait that arises due to non-random mating (Pomiankowski 1988). These are models of indirect selection (Kirkpatrick & Ryan 1991), because preferences evolve due to selection on male traits rather than selection directly on female preferences. In contrast, some direct selection models predict that female mating preferences and male traits are evolutionarily decoupled: prefer-ences can arise before the traits they favour, as in sensory exploitation or sensory bias models (Basolo 1990; Ryan et al. 1990a), and preferences for existing traits can be lost because of costs associated with the preference (Kirkpatrick 1987