Investigation Into the Usability of Geometric Morphometric Analysis in Assessment of Sexual Dimorphism E. Pretorius,* M. Steyn, and Y. Scholtz Department of Anatomy, University of Pretoria, Pretoria 0001, Republic of South Africa KEY WORDS ramus flexure; orbit shape; greater sciatic notch; geometric morphometrics ABSTRACT Understanding sexual dimorphism is very important in studies of human evolution and ske- letal biology. Sexual dimorphic characteristics can be studied morphologically and metrically, although mor- phologic studies pose several problems such as difficul- ties with quantification and interobserver error. Geomet- ric morphometrics is a relatively new method that allows better assessment of morphologic characteristics. This paper aims to investigate the usability of this method by assessing three different morphologic characteristics in a sample of South African blacks: shape of the greater sci- atic notch, mandibular ramus flexure, and shape of the orbits. Relative warps, thin-plate splines, and canonical variates analysis (CVA) analyses were performed. As expected, the shape of the greater sciatic notch provided the best separation between the sexes. Surprisingly, however, the shape of the orbits performed better that ramus flexure. Several possible explanations for this re- sult are possible, which include the possibility that orbit shape is more sexually dimorphic than previously ex- pected, or that biological reality is not reflected by this technique. More research is, however, needed. Am J Phys Anthropol 129:64–70, 2006. V V C 2005 Wiley-Liss, Inc. Determination of sex is one of the vital aspects in any skeletal analysis. Sexually dimorphic characteristics can be assessed metrically and morphologically. The use of metric parameters is usually rather straightforward, as numerical results are obtained that are easy to assess and interpret (e.g., DiBennardo and Taylor, 1983; _ I ¸ scan and Miller-Shaivitz, 1984; Steyn and _ I ¸ scan, 1998, 1999; Asala, 2001). Many characteristics, however, can only be assessed morphologically, e.g., prominence of the gla- bella, shape of the pubic bone, or flexure of the mandibu- lar ramus (e.g., De Villiers, 1968; Rogers and Saunders, 1994; Loth and Henneberg, 1996; Patriquin et al., 2003). These morphological characteristics of the skeleton are often difficult to assess, due to a number of factors such as inter- and intraobserver errors, problems with classification of morphological characteristics (e.g., wide, narrow, or intermediate), and statistical analysis. The experience and interpretation of the observer also plays a role. Geometric morphometrics is a relatively new method that provides a mechanism to quantify morpho- logical characteristics (Kendall, 1981, 1984; Bookstein, 1989, 1991, 1996; Rohlf and Slice, 1990; Rohlf and Mar- cus, 1993; Slice, 1993; Rohlf, 1995, 2002). It allows for a detailed assessment of the aspects in which the morphol- ogy between various skeletons differs, and may provide some statistics that can be used to interpret findings. Although it has been used to quantify morphology since the late 1980s, it is a technique that just recently started to become popular in physical anthropology (e.g., Lynch et al., 1996; Wood and Lynch, 1996; Hennessy and Stringer, 2002; Rosas and Bastir, 2002). In two recent studies, Steyn et al. (2004) and Oettle ´ et al. (2005) used this method to assess differences in the shape of the greater sciatic notch and ramus flexure of the mandible, respectively. Through new developments in techniques of geometric morphometrics, it is now possible to obtain P-values to test for statistical significance when comparing various shapes. It is also possible to do classifications, through canonical variates analysis (CVA) analyses. From CVA analyses, the percentage of correctly assigned member- ship can be obtained. The aim of this study is to test the workability and validity of these methods, when applied to sexually dimorphic characteristics of the human skeleton. For this purpose, we decided to use three morphological characteristics: shape of the orbits, ramus flexure, and greater sciatic notch shape. The shape of the orbits is not a known, well-proved method of determining sex. Some (e.g., Krogman and _ I ¸ scan, 1986) suggested that the shape of the orbit is more round in females, and more squared in males. To our knowledge, however, this has not been tested before, and this method can therefore be expected to show no statistically significant difference between the sexes (high P-value), and a low classifica- tion accuracy. The flexure of the mandibular ramus is of ‘‘intermedi- ate’’ value when used in sex determination. Initial publi- cations indicated a very high accuracy of 99% (Loth and Hennebrg, 1996), showing that the ramus is flexed in males and straight in females. Subsequent studies could not obtain the same rates of accuracy, and accuracies Grant sponsor: National Research Foundation of South Africa. *Correspondence to: Prof. E. Pretorius, PO Box 2034, Pretoria 0001, Republic of South Africa. E-mail: resia.pretorius@up.ac.za Received 4 August 2004; accepted 3 December 2004. DOI 10.1002/ajpa.20251 Published online 21 October 2005 in Wiley InterScience (www.interscience.wiley.com). V V C 2005 WILEY-LISS, INC. AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 129:64–70 (2006)