We investigated male sexual behaviour and the cost of sexual harassment, as measured by the reduction of female feeding time in the presence of a male, in a cavedwelling population of Poecilia mexicana, in which sexual harassment does not occur naturally. We asked whether the lack of sexual harassment in this population is due to low sexual activity of the males, or low feeding motivation of the females. We experimentally increased the sexual activity of males or the females' feeding motivation, or we used a combination of both treatments. Female feeding time was not lower in the presence of a male than in the presence of a female after sexual deprivation of the males or food deprivation of the females. Only in the combined experiment was female feeding time lower in the presence of a large male than in the presence of a small male, indicating a weak effect of sexual harassment by large males. Virgin females did not suffer a cost of sexual harassment, indicating that sexual experience does not cause the lack of sexual harassment in cave mollies. Males from a surface population, where sexual harassment occurs, significantly reduced the feeding time of cavedwelling females even though these males exhibited surprisingly little sexual behaviour. The sexual activity of cave mollies did not correlate with male body size in any experiment, indicating that even after sexual deprivation, small cave molly males do not switch to the "alternative" mating behaviour known in surface-dwelling P. mexicana, where sexual activity is correlated negatively with male body size.

The Atlantic molly Poecilia mexicana inhabits a variety of habitats. In previous studies, small males from clear-water populations showed considerably more sexual behaviours than large males. Males from a sulphur creek, the El Azufre, or from a sulphur cave (cave molly) showed comparatively low sexual activity, and typical size-dependent mating behaviour was either absent or large males showed even more sexual behaviour than smaller ones. Sexually deprived cave molly males increased their sexual activity, but small males still did not show more sexual behaviour than large ones, suggesting that the genetic basis for typical size-dependent mating behaviour has been lost. Likely, the extreme abiotic conditions (hypoxia and toxic hydrogen sulphide) have selected against energetically costly behaviours. In this paper, we examined how El Azufre males would respond to sexual deprivation. We found these males to also increase their sexual activity after sexual deprivation but, unlike cave mollies, small males now showed more sexual behaviour than large ones, suggesting that the genetic basis for this behaviour is not lost in this population. Differences between populations may be due to the less harsh characteristics of El Azufre (e.g. less H 2 S), resulting in weaker selection against energy-consuming behaviours. Furthermore, we tested for potential costs of male sexual harassment for females in terms of reduced feeding efficiency in the presence of a male, as was shown for P. mexicana from a clear-water population, but was not found in cave mollies or the El Azufre fish. Test females were either starved for 1 week to increase hunger levels or male sexual deprivation and female starvation were combined; however, we found no evidence for male sexual harassment even after these treatments. Our results parallel previous findings in the cave molly and suggest that El Azufre males have lost specific behavioural traits that cause sexual harassment.