On applications of geometric morphometrics to studies of ontogeny and phylogeny.

In press − − do not cite 2 The field of geometric morphometrics is relatively new (see Bookstein, 1991, Rohlf and Marcus, 1993) and has shown very rapid progress over the last few years. As might be expected during a period of rapid development, there can be technical problems in some of the pioneering studies as biologists attempt to apply the new tools. It is only now becoming clear how the new techniques should be combined in order to carry out comprehensive analyses of real datasets. Bookstein (1996b) gave a list of recommendations for such applications and Bookstein (1996a) gave several comprehensive examples of morphometric analyses. However, these accounts do not address some of the types of applications that are of particular interest in systematic biology. Bookstein (1994) pointed out problems with using geometric morphometric methods in the usual character-based cladistic studies. He emphasized that morphometrics cannot supply homologous shape characters. The purpose of this note is to comment some recent applications of morphometric methods in systematic biology. This note is particularly concerned with the morphometric methods used by Zelditch et al. These methods were also used by Burke et al. (1996). For convenience, these studies will be referred to using the acronym Z&F. In these studies the authors investigated the use of partial warps (Bookstein, 1991) as variables in ontogenetic and taxonomic studies. They were impressed by their observation that differences in partial warps scores (Rohlf, 1993b) corresponded to shape differences that could be localized on the bodies of the organisms. They also found differences in these variables between developmental stages and between species. They concluded that partial warps could be interpreted and used as traditional taxonomic characters and would be useful in evolutionary studies. Lynch et al. (1996) tried to interpret the partial warps they obtained in their study but they were cautious about using them in the ways advocated by Z&F. They suggested that extensive simulation studies needed to be done to validate the Z&F approach. Naylor (1996) investigated their approach using data based on a simulated phylogeny. Even though his simulated phylogeny was based on a sequence of simple morphological changes and had no homoplasy, the results showed high levels of homoplasy and did not recover the morphological changes used to create the phylogeny. The present paper is concerned with theoretical problems rather than empirical problems requiring validation through simulations. The fundamental problem is that Z&F …