We used discriminant analysis to assess taxonomic status of three closely related, rare taxa within the cactus genus Escobaria (E. guadalupensis, E. sneedii var. sneedii, and E. sneedii var. leei). These three taxa consist of nine isolated populations; six of these populations consist of morphologically typical specimens (i.e., similar to the type specimen), while three populations are of questionable taxonomic status. We used the six morphologically typical populations (referred to as assigned populations) to derive predictive discriminant functions for each taxon. These discriminant functions were then applied to assigned populations to determine the status and taxonomic boundaries of each taxon, and to the three questionable populations (referred to as unassigned populations) to interpret their taxonomic status. Our model included four predictor variables: number of immature stems, mature stem diameter, number of radial spines, and length of the innermost central spines. Discriminant analysis correctly classified >93% of 186 individuals from the six assigned populations, suggesting a distinct status for each taxon. Applying the predictive discriminant functions to unassigned populations resulted in classifying a large percentage of individuals from each unassigned population into each of two taxa. This suggests that the unassigned populations consist of morphological intermediates, possibly as the result of hybridization. Meiotic chromosome numbers of n = 11 were determined for E. guadalupensis, E. sneedii var. leei, and the three unassigned populations.

Communicating Editor: Richard Jensen