3D geometric morphometric analyses of the human torso morphology reveal new understandings about the torso integration in primates

Classic studies in Palaeoanthropology suggest that in hominins and other primates thoraces and pelves are anatomically integrated. However, this torso integration hypothesis has been only tested in isolated bones so far, but not in anatomically connected torsos. Here we aim at testing the torso integration hypothesis in two modern human populations. We considered and excluded the effect of sexual dimorphism, which is an important factor influencing the human torso shape. We segmented 50 anatomically connected torsos (CT-scans) from two different populations (25 Mediterraneans, 25 Sub-Saharan Africans). We compared the maximum medio-lateral width at 7th-9th rib levels with pelvic bi-iliac breadth in males and females within both populations. From a 3D geometric morphometrics approach, we measured 1030 (semi)landmarks on each torso model, and we quantified torso shape variation, mean size and shape differences, thoraco-pelvic covariation and allometric effects. Females show narrow thoraces and wide pelves and males show the opposite trend in both populations. Indeed, we found equal thoracic and pelvic widths and curvatures only in absence of sexual dimorphism, with the lowest rib levels and the ilia showing the highest covariation. In addition, larger torsos were narrower than smaller torsos. Our study reveals therefore that sexual dimorphism leads to a complex thoraco-pelvic integration model in H. sapiens previously unrevealed by measuring isolated bones. These findings question some assumptions derived from the classical torso integration hypothesis and emphasize the role of sexual dimorphism when reconstructing fossil hominin body shape