Sexual Size Dimorphism in Seabirds: A Reply to Croxall

Abstract

Croxall (1995) has attacked our analysis of sexual size dimorphism in southern-Hemisphere seabirds (Fairbairn and Shine 1993) on several grounds, and suggested that our conclusions are in error because of deficiencies in the ways that we compiled and analysed our data sets. We believe that he is wrong, and we explain our reasons in this rebuttal. Croxall's major arguments are as follows: (1) our estimates of mean body sizes for males and females of several taxa are unreliable, because they contain numerous errors, omit important species, are based on samples that vary considerably in size and reliability, and do not take into account short-term fluctuations in body mass in breeding seabirds. (2) the direction and degree of sexual size dimorphism (SSD) in seabirds show strong phylogenetic conservatism, so that the associations we documented between SSD and two other variables (absolute body size, and oceanic productivity) may be artifacts of phylogenetic conservatism in these variables also. First, we consider Croxall's criticisms of our data set. He identified only one mathematical error (a transposition of two digits for mean mass of males in one population of rockhopper penguins) among our data on 99 populations (66 species) of seabirds. However, Croxall also identified numerous other problems: his Appendix lists 17 populations that we omitted despite the availability of published data, three data sets that combined information gathered in different ways (one sample of Diomedea exulans was gathered during the breeding season whereas another was not; one sample of Fregetta tropical was based on measurements of live animals whereas another was not; the data for Oceanites nereis combined data from different localities) and four data sets that were based on inadequate information (small sample sizes for Pterodroma solandi, P. cooki, and Puffinus gavia; reliance on beachcast specimens for Pterodroma brevirostris). He also criticised the use of a temporally unstable measure such as body mass as an index of size dimorphism. Below, we consider each of these issues in turn. First, we consider the omissions. As our paper indicated, we relied on a single encyclopedic review (Marchant and Higgins 1990). Two of the omissions cited by Croxall concern taxa (Diomedea m. melanophrys and D. chrysostoma) that were not listed in this review, and hence were not incorporated in our study for this reason. Another supposed omission (of a very large data set for Eudyptula minor) is an error on Croxall's part: the sample is incorporated in our Table and in our analysis. However, Croxall also identified 15 additional taxa that were listed in Marchant and Higgins (1990), but which we did not include. Ironically, in view of Croxall's concerns about small sample sizes, we had omitted 9 of these samples because they were based on small numbers (? 10 specimens of one sex). The other six missing populations were omitted inadvertently, or because we had the same kinds of reservations as Croxall about using samples based only on skins rather than live birds. We have repeated our analyses with the full data set (i.e., incorporating all taxa listed in Marchant and Higgin's review, including the 15 identified by Croxall and three others also omitted from the original analysis). Our conclusions are unchanged. In analyses that treat each population as an independent unit, SSD correlates significantly both with mean absolute adult body mass (n = 118, r = -0.44, P <0.0001) and with oceanic productivity (n = 100, r=-0.55, P<0.0001). The same patterns remain with a phylogenetically based analysis, where the magnitudes of phylogenetic changes in one variable are compared to concurrent changes in another variable (SSD vs size n=54, r=-0.29, P<0.035; SSD vs productivity n=57, r=-0.36, P<0.007). Thus, incorporating these additional taxa has no influence on our conclusions. Croxall also raises a more general issue of omission of taxa, by suggesting that we should have included additional species from other parts of the world. It is un-