Omland, K. E. 1997. Examining two standard assumptions of ancestral reconstructions: the history of plumage dimorphism in dabbling ducks (Anatini). Evolution 51:1636-1646.

Abstract. -- The use of phylogenies to reconstruct ancestral states is a valuable technique for studying the evolution of characters that leave no fossil record. Few researchers have attempted to assess the reliability of ancestral reconstructions. Here I test for congruence of a phylogenetic reconstruction with a widely accepted scenario based on independent lines of evidence. I used Livezey's (1991) phylogeny to infer the history of plumage dimorphism in dabbling ducks (Anatini). This procedure reconstructs monomorphic ancestors for the genus Anas as well as each of its six main clades. This reconstruction differs strongly from the widely accepted scenario of speciation and plumage evolution in the group (e.g., Delacour and Mayr 1945; Sibley 1957). This incongruence may occur because two standard assumptions of character reconstruction are probably not met in this case. Violating either of these two assumptions would be source of error sufficient to create misleading reconstructions. The first assumption that probably does not apply to ducks is that terminal taxa (in this case species) are monophyletic. Many of the widespread dimorphic species of ducks may be paraphyletic and ancestral to isolated monomorphic species. Three lines of evidence support this scenario: population genetic data, biogeography, and vestigial plumage patterns. A second assumption that probably does not apply to duck plumage color is that gains and losses of character states are equally likely. Three other lines of evidence suggest that dimorphic plumage might be lost more easily than gained: weak female preferences for bright male plumage, biases toward the loss of sexually dimorphic characters, and biases toward the loss of complex characters. In order to recover the accepted scenario using Livezey's phylogeny, losses of dimorphism must be five times more likely than gains. These results caution against the uncritical use of unordered parsimony at the sole criterion for inferring ancestral states. Detailed population-based sampling is needed and altered transformation weighting may be warranted in ducks and in many other groups and character types with similar attributes.

Key words. -- Character reconstruction, ancestral reconstruction, paraphyletic species, sexual color dimorphism, comparative method, Anas