A Morphological Analysis of the Structure of Communities of Lizards in Desert Habitats

Using nine measurements of the head, body, and appendages we characterize the morphological relationships among lizards in deserts of western Australia, southern Africa, and western North America. Each species was placed in a morphological space whose dimensions were the logarithms of the original measurements. We calculated the dispersion of the species in each region along principal components to assess the number of dimensions and volume of morphological space occupied. Euclidean distances between all pairs of species were calculated to assess the density of species packing within the morphological space. We also generated random subsets of the species pool within each region and calculated matrices of Euclidean distance for them. In the combined sample of 83 species, the first three principal components accounted for 86, 8 and 3%, respectively, of the total variance in the dispersion of species in morphological space. Pairwise discriminant analysis revealed significant differences between the morphological distributions of the lizards in the three regions. When analyzed separately, the 54 Australian lizards occupied the largest volume, followed by the 11 North American species, and, lastly, the 18 southern African species. We used the average distance to the nearest neighbor (NND) as a measure of species packing. NND's in Australian localities (18—36 species) and in North American localities (4—10 species) were considerably greater than those in southern African localities (10—15 species). When assemblages were adjusted for number of species, the Australian lizards were more widely spaced morphologically than those in either the North American or African localities. Attributes of randomly generated assemblages did not differ morphologically from the subsamples of species found in each locality. Furthermore, the standard deviations of NND's, a measure of the regularity of species packing, were similar in natural and randomly generated communities. Hence the present analysis provides little indication of interaction between species according to morphological attributes. Our morphological analysis confirmed some patterns revealed by studies of ecological relationships in the same localities.