Abstract:
Variation in 19 cranial and four external measurements was assessed for 65 populations of Microtus mexicanus from throughout its range. Discriminant function analysis resulted in the separation of two major groupings of populations. One group included all populations from Mexico, with the exception of the populations from San Antonio de las Alazanas, Coahuila and near Galena, Nuevo Leon. The other group included these exceptions, plus all populations from the united States. These two groups separated on canonical vector I, which was primarily a size factor with all characters, except mastoidal breadth, having negative character loadings. This separation suggested that M. mexicanus is represented by two species: however, principal component analysis failed to separate the two groups. Multivariate analyses of variance between neighboring populations resulted in the grouping of populations that were not significantly (E > 0.001) from each other. These non significant groupings were often concordant with the distributions of named subspecies. The United States showed relatively more non significant differences between neighboring populations than populations from Mexico. Populations from the united states were divided into five groups, each representing a subspecies, based upon the non significant groupings derived from the multivariate analyses of variance; this was also supported by a principal component analysis for those populations with small sample sizes from the Music Mountains, Hualapai Mountains, and south of Prescott, Arizona and neighboring populations. Populations from Mexico showed relatively more differentiation as compared to populations from the united states. The multivariate analyses of variance resulted in the separation of 13 groups in Mexico. The greater divergence of Mexican populations is explained by populations from Mexico being relatively old with the united states populations representing a more recent invasion north during the Illinoian glacial period.
