The selection of taxonomic characters for morphometric analysis: a case study based on southern African Aethomys (Mammalia: Rodentia: Muridae)

Abstract

Prior to a systematic revision of southern African rock rats of the genus Aethomys Thomas, statistical procedures were used to select morphometric characters for recording and subsequent use in a revision of the group. The procedure advocated could have wider application in morphometric studies. An initial set of 66 cranial characters was reduced to 51 after the data set was subjected to routine assumptions tests. The remaining 51 characters, considered to be statistically problem-free, were reduced to a final set of 11 characters after subjecting them to cluster and ordination procedures to summarize patterns of correlations between characters, develop criteria for the selection of representative measurements within cluster analysis-generated subclusters, and the subsequent removal of redundant variables. The procedure attempts to economize while at the same time adequately represent the phenotype, an approach consistent with the concept of morphological integration. Four additional cranial and four standard external characters are also included in the final data set, but for descriptive and comparative purposes only.