On the use of avian mortality patterns to test sexual selection theory

-To test the hypothesis that reduced survival of large males limits the extent to which sexual selection can increase male size, we examined mortality patterns in Redwinged Blackbirds (Agelaius phoeniceus) subjected to a surfactant spray. Univariate comparisons of wing measurements did not support the predicted mortality patterns. Although principal component analysis revealed that mortality was nonrandom with respect to some morphological characters, there was again no evidence of the predicted patterns either of directional selection favoring smaller males or of stabilizing selection favoring intermediatesized females. This result led us to recognize that, like previous authors, we had made the incorrect assumption that large males should be more vulnerable to all mortality factors rather than the correct interpretation of the theory, that differential mortality should be seen as the net effect of all mortality factors. Only two published data sets are appropriate for testing the hypothesis, and neither shows differential mortality amongst males on the basis of size. Received 22 February 1983, accepted 14 September 1983. THE correlation between the degree of sexual size dimorphism in birds and the variance in mating success of males relative to females provides one of the main supports for sexual-selection theory (Darwin 1871; Selander 1972), although in some specific instances this correlation has not been found (e.g. Weatherhead 1980). The implication of the relationship is that larger size enhances an individual's ability to compete for mates. In considering what pressures constrain male size from getting even larger, Selander (1965) hypothesized that the enhanced mating ability of large individuals is balanced by their lower survival. Because females of most avian species are under weak sexual-selection pressure relative to males, their size should approximate the ecological optimum for individuals of that species, that is, the size that maximizes survival. In birds therefore, females can be considered a control group against which one can compare relative male survival to assess the extent to which sexual selection has moved males away from the ecological optimum for the species. 3 Present address: Oikos Ecological Research Associates Ltd., P.O. Box 8818, Saskatoon, Saskatchewan S7K 6S7, Canada. Searcy and Yasukawa (1981) have described two methods of testing Selander's (1965) hypothesis. The first is to compare male and female survival between species that occur along a sexual-dimorphism gradient. Using the family Icteridae, which includes species with varying degrees of sexual dimorphism, Searcy and Yasukawa (1981) found that the survival of males relative to females decreased as the degree of dimorphism, and thus sexual selection, became more pronounced. This result is in accordance with Selander's (1965) hypothesis. The second approach to testing the hypothesis is to compare mortality patterns within species. If male size is above the survival optimum and female size approximates the survival optimum, several predictions regarding mortality patterns should be met. In males one should find higher mortality amongst larger individuals, resulting in directional selection toward the female phenotype. If females are at the optimum, one should find higher mortality amongst females both larger and smaller than the mean, resulting in a pattern of stabilizing selection around the mean. Searcy and Yasukawa (1981) cite several studies that have provided data suitable for testing at least one of the above predictions. In some cases the predictions were supported and in others they were 134 The Auk 101: 134-139. January 1984 This content downloaded from 157.55.39.78 on Sat, 25 Jun 2016 06:57:09 UTC All use subject to http://about.jstor.org/terms January 1984] Testing Sexual Selection Theory 135 not. Our aim in this paper is to report the results of a study similar in design to one cited as supporting the hypothesis of higher mortality amongst larger males (Baker and Fox 1978). We use the results of our study to develop the argument that the data from a number of the studies cited by Searcy and Yasukawa (1981) are actually inappropriate for testing the predictions regarding mortality patterns in sexually dimorphic species for which the dimorphism is a product of sexual selection. To use these data for that purpose constitutes a misinterpretation of Selander's (1965) hypothesis. AN APPARENT TEST OF SELANDER'S