Sexing bison metapodials using principal component analysis

Bison remains are a common and important component of many North American archaeological and paleontological sites. Interpretations of bison remains, however, often are hampered by the inability to determine sex reliably in this dimorphic taxon. Metapodials are among the most common bison element recovered in archaeological assemblages, but have proven difficult to sex. The methods currently in use to estimate sex range from those using only bivariate plots and ratios of various metapodial measurements to those using discriminant function analysis. Each method has advantages and disadvantages, with no one method producing certain, unambiguous results. The designation of sex in borderline specimens remains uncertain in all current methods and must be determined subjectively. By reducing data through the use of ratios, the current bivariate methods fail to make full use of the size variation present between male and female bison. The requirements of discriminant function analysis likewise limit this method's utility for sexing metapodials, particularly for assemblages containing small sample sizes. The use of principal component analysis utilizing several of the most dimorphic, commonly measured variables produces a more confident assessment of sex for both complete and partial specimens of modern bison metapodials.