SHEETS, H.D.; SWIDERSKI, D.L.; Canisius College, Buffalo, NY.; Museum of Zoology, Univ. of Michigan, Ann Arbor: Methods of combining landmarks and outline data, with analyses of scapula shape in selected Neotropical bats

Landmark based geometric morphometrics has developed as a powerful set of statistical and visual tools for the study of organismal shape. The approach is limited in the kinds of shape information accessible to it, however, by the need to employ discrete homologous landmarks as the basis for comparison. In particular, curves and complex outlines are difficult to study by existing methods. Data of this type may be incorporated into the study of shape by the use of semi-landmark methods, which allow information about curved surfaces to be incorporated into the framework of landmark based geometric morphometrics. We present software and statistical approaches needed to carry out combined landmark and semi-landmark analysis. We demonstrate the method and compare it to standard landmark based methods in a variety of analyses, including Principal Components Analysis, pair-wise tests for shape differences between taxa, and analyses of disparity within taxa. These analyses are performed on scapulae of selected Neotropical bats of the Phyllostomidae. In these bats, 12 landmarks can be consistently recognized; most are near the joint with the humerus. Analyses using only these landmarks are successful in discriminating among taxa with markedly different flight and foraging patterns. Analyses including semi-landmarks digitized on margin of the anterodorsal (supraspinous) fossa discriminate among the same taxa and also supply additional detail about changes in the shape of that fossa. This comparison demonstrates that semi-landmark methods can be useful for systems where landmarks are sparse.