Phylogenetic signals in morphometric data

Although many of the goals and concepts of qualitative morphological analysis and morphometrics are similar, systematists have largely rejected the use of morphometric methods in phylogenetic analysis on a variety of grounds. This review finds that (1) the concepts of a cladistic character and a morphometric variable are essentially identical, (2) morphometric methods can be instrumental in discovering and documenting new morphological character and character states, (3) prior objections to the use of morphometric variables because of their continuous nature confuse the issues of variable type with those surrounding the distributions of sets of observations, (4) morphometrics offers the best method of determining whether morphological observations are discontinuous (= can be coded as discrete character states) or continuous (= cannot be coded as discrete character states), (5) constellations of landmark-based morphometric variables represent adequate summaries of putative structure-level homologues for use in phylogenetic analyses, (6) partial warp analyses do not perform well in either simulated or actual phylogenetic systematic analyses because of their inherent instability and lack of adequate spatial localization, and (7) a new method of subdivided relative warp analysis (described herein) performs very well at recognizing simulated morphological character states and recovering a simulated morphological phylogenetic hierarchy. Based on these results it is concluded that the potential of morphometric data analysis methods (especially relative warp-based methods) to contribute to phylogenetic-systematic investigations should be explored further.