Phylogenetic and Historical Biogeographical Relationships of the Genera in the Family Agamidae (Reptilia: Lacertilia).

Abstract

The morphology, taxonomy, fossil history and biogeography of the Agamidae are reviewed and a phylogenetic history of the 53 genera and 317 species is hypothesized. The morphological review includes the cranial and postcranial skeleton, abdominal and pectoral girdle muscles, forelimb and hindlimb nerve pathways, epidermal scalation, skin gl and s, dentition, features of the ear, eyeball, nose and hyoid apparatus, and karyology. The phenotypic variation is encoded as 143 binary state characters and reflects variation between supraspecific taxonomic categories. Principal component analysis and Prim and Wagner networks demonstrate structuring within the data matrix and hypothesize six major groups of agamids: I, Uromastyx and Leiolepis; II, Hydrosaurus and Physignathus; III, the Australian radiation of l4 genera; IV, Arua and Hypsilurus; V, primarily arboreal radiation of India and Southeast Asia of 27 genera; and VI, primarily terrestrial radiation of Africa and Asia of seven genera. Application of the Wagner tree algorithm predicts that half of the 143 characters are robust, i.e., the derived state is either unique or only one or two extra state changes occur on the phylogenetic tree. This algorithm estimates both the cladistic and patristic parameter and hypothesizes a most parsimonious directed tree. The Wagner tree algorithm is applied to both unweighted and weighted sets of characters. Individual character consistency ratios are calculated from the unweighted Wagner tree and are used as the weighted characters. The unweighted characters are analyzed by character compatibility analysis. This method is a strong weighting procedure in that subsets of only compatible (homologous) characters are used for predicting the cladistic events of a phylogenetic history. It does not allow reversal of character states. Nor does this method allow objective resolution of all taxonomic relationships. All three phylogenetic methods predict the six major agamid groups. The weighted Wagner tree and the character compatibility tree are the most similar in predicting the major dichotomies in agamid phylogenetic history. All three methods predict the six major agamid groups. Group I is hypothesized the sister taxon of the remaining agamid groups and is recognized as a separate family. The genera of Groups II and IV are sequentially ranked. Groups III, V and VI are hypothesized to be monophyletic radiations; the latter two are sister taxa. The relationships of the genera within these three groups are illustrated but a more stable or reliable hypothesis requires additional characters specific to these taxa. The plate tectonics of Gondwanal and , Southeast Asia and Indonesia is reviewed and related to vicariance biogeography. The question of Gondwanan or Laurasian affinity of Southeast Asia and Indonesia is treated in detail. Biogeographic analysis of the distribution of agamids demonstrates strong endemism of genera within well-defined tectonic regions. The Australian-New Guinean and Southeast Asian-Indonesian tectonic plates are both hypothesized to be the center of origin of the Agamidae. The boundary between these two regions corresponds with Wallace's Line and also delimits phylogenetically Groups III-IV from V-VI. This vicariance would corroborate the alternative tectonic model suggesting a Gondwanan connection for Southeast Asia and Indonesia. The Upper Cretaceous agamid fossil Mimeosaurus from Mongolia is considered strong biological evidence for this connection. The occurence of agamids in Africa, Europe and southwestern Asia and the presence of the Eocene agamid Tinosaurus in western North America are best explained by dispersal models.