Learning in the context of sexual behaviour and danger in female and male Drosophila pseudoobscura

Mate preference for a phenotypically plastic trait is learned, and may facilitate preference-phenotype matching

Fixed, genetically determined, mate preferences for species whose adult phenotype varies with rearing environment may be maladaptive, as the phenotype that is most fit in the parental environment may be absent in the offspring environment. Mate preference in species with polyphenisms (environmentally dependent alternative phenotypes) should therefore either not focus on polyphenic traits, be polyphenic themselves, or learned each generation. Here, we test these alternative hypotheses by first describing a female-limited seasonal polyphenism in a sexually dimorphic trait in the butterfly Bicyclus anynana, dorsal hindwing spot number (DHSN), and then testing whether male and female mate preferences for this trait exist, and whether they are seasonally polyphenic, or learned. Neither naïve males nor naïve females in either seasonal form exhibited mating preferences for DHSN. However, males, but not females, noticed DHSN variation and learned mate preferences for DHSN. These results suggest that individuals may accommodate environmentally dependent variation in morphological traits via learned mate preferences in each generation, and that learned mate preference plasticity can be sexually dimorphic.

Effects on population divergence of within-generational learning about prospective mates

Although learned mate preferences are suspected to have important effects during speciation, theoretical models have largely neglected the effects on speciation and population divergence of within-generational learning, that is, learning based upon prior experience with potential mates. Here, we use population genetic models to address this deficit. Focusing on the situation of secondary contact between populations that still hybridize, we consider models of learning by females and by males under polygyny. We assess the effects of learning to prefer conspecifics from previous conspecific encounters, learning to avoid heterospecifics from previous heterospecific encounters, and learning to prefer familiar types. We examine the amount of population divergence that results from learning in these models. We also assess the effect of learning on the spread of an allele that strengthens assortative mating in both models. We find that learning can have counterintuitive, but logical and understandable effects that differ with the version of the model assessed. In general, population divergence is expected to increase most consistently when females learn to strengthen their preferences for conspecifics from previous encounters with conspecifics. Our results also suggest that within-generational learning will generally inhibit the spread of alleles strengthening assortative mating.