Hop, step and gape: do the social displays of the Pelecaniformes reflect phylogeny?

Abstract

Several authors have recently revived the old ethological claim that behavioural characters could be used to provide accurate estimates of phylogeny. To test this contention we constructed a behavioural data set for the pelecaniforms based upon van Tets’ (1965, Ornithol. Monogr., 2,1–88) classic comparative study of their social behaviour. Parsimony analysis of the resulting 20 taxa, 37-character data set produced 12 shortest trees. These trees fitted the behavioural data well. A permutation tail probability test found that the behavioural data contained more structure than would be expected by chance alone. The behavioural trees were compared with previously published morphological and genetic estimates of pelecaniform phylogeny to test if they were more congruent than would be expected by chance. In all cases the behavioural trees were far more similar to the morphological and genetic trees than chance alone would predict. By mapping the distribution of the behavioural characters onto a best-estimate evolutionary tree constructed from the independent trees, it was possible to investigate the homology of the behavioural characters. All but three of the characters appeared to be homologous. The behavioural characters were, therefore, no more homoplasious than other types of characters. The best-estimate tree was used to predict the social displays of the pied cormorant, Phalacrocorax varius, and to evaluate hypotheses about the derivation of male advertising displays.