A fool's mission? A test of three common assumptions in dental metric analyses

Abstract

Three aspects of metric variation in the permanent dentition of humans are often simply accepted as true. The first is that formation of the permanent dentition occurs within morphogenetic fields broadly associated with tooth type and jaw. The second is that dental development of among females is characterized by a higher degree of ontogenetic buffering relative to males. The third is that expression of sex dimorphism in permanent tooth size is expressed uniformly among well-nourished human populations. This study tests these assumptions through an examination of mesiodistal and buccolingual dimensions of all non-canine permanent teeth, except third molars, among 2,709 living individuals of 15 ethnic groups from South Asia. With sexes pooled, only one in four contrasts of variance among key versus distal teeth within dental fields are significantly heterogeneous, while one in four contrasts yield higher levels of variance among key teeth relative to their distal counterparts within a dental field. Such results weaken considerably orthodox applications of Butler's dental field theory. When samples are the unit of analysis, male samples are marked by fewer dental fields with significantly heterogeneous levels of variance between key and distal members, while males and females are affected equally by significantly heterogeneous variation between key and distal members when dental fields are the unit of analysis. Such results suggest males and females are equally buffered against environmental perturbations that affect odontometric variation. One-way ANOVA indicates that a tooth's position within a dental field ac-counts for 15.5% to 23.1% of the observed varia-tion in tooth size, while two-way ANOVA reveals that when sex is added as a second factor, the percentage of variance in tooth size explained increases from 16.7% to 30.8%, an improvement of 27.2%. Such results indicate sex dimorphism in tooth size varies in both patterning and in magnitude among these samples, thereby explaining why discriminant functions developed for one population often perform more poorly when applied to other populations.