On the difficulty of detecting density?dependent selection on polymorphic females of the damselflyIschnura graellsii: Failure to reject the null

Abstract

Female colour polymorphism is a prevalent, but perplexing characteristic of many species of coenagrionid damselflies. The colour of the thorax and abdomen of an 'andromorphic' female is similar to that of the male, whereas the colour of a 'heteromorphic' female differs from that of the male. Robertson (1985) proposed that andromorphs were 'male mimics' and, thus, more likely than heteromorphs to be ignored by males as potential mates. He suggested that the benefit of avoiding time-consuming supernumerary matings balanced the cost of relatively greater predation on the purportedly more conspicuous andromorphs. Hinnekint (1987) later used 'male mimicry' as the mechanism underlying his density-dependent selection model. Hinnekint's hypothesis stated that at high male density, andromorphs were better able to avoid unnecessary matings, but at low male density, they incurred a greater risk of not mating at all. Both Robertson (1985) and Hinnekint (1987) predict that at any male density, andromorphs mate less frequently than heteromorphs. To test that prediction, Cordero (1992) quantified lifetime mating frequency and survivorship of three colour morphs of the damselfly Ischnura graellsii in two populations in Spain. Within both populations, neither the mean number of matings per female nor their mean life-span differed among the andromorphic (A) females and the two types of heteromorphic (I and O) females. Nevertheless, Cordero (1992) concluded that his data supported Hinnekint's (1987) densitydependent hypothesis. Although Codero did not quantify population density, in the population he designated as 'low density', he found that andromorphs were less likely than heteromorphs to be seen to mate at least once. In the 'high-density' population, Cordero found no difference in mating failure among morphs. But Cordero's data failed to reject the null hypothesis that female coloration makes no difference to a female's fitness. In what follows, I show that the null hypothesis appears to be the best.