Analysis of the primate vestibular apparatus: a comparison of 3D geometric morphometric and diffeomorphism approaches
Résumé
Inferring the positional behavior of extinct taxa is crucial for understanding the evolution of the ape and human clade (Homi noidea), with implications for reconstructing the locomotion of the last common ancestor between humans and chimps preceding human bipedalism. Besides a morphofunctional approach applied to the postcranium , several authors have relied on the morphology of the vestibular apparatus to infer the locomotion of extinct primates [ l -2]. The vestibular apparatus, housed in the bony labyrinth, is composed of three semicircular canals, the utricle, and the saccule. These structures detect angular accelerations and provide sensory input for stabilizing the vision during rotary movements of the head, thereby being functionally related with maintaining balance during locomotion [ l -4 ). Different approaches based on inner ear morphology have been applied to gain insight on the locomotor repertoire of fossil primates [1-2). Nevertheless, there is no current consensus about the most efficient methodology for assessing the locomotor repertoire of extinct species. Here we compare the results of two three-dimensional geometric morphometrics (3D GM) methods applied to the vestibular apparatus of extant primates, with emphasis on rheir potential for predicting the positional behavior of the investigated taxa: landmark-based and deformation-based 3DGM . The former relies on generalized Procrustes superimposition, whereas the latter directly compare the shape of different surfaces via continuous and invertible deformation [5]. While the preparation of the landmark set is more rime consuming, deformation analysis needs higher computational power. We used a set of 9 landmarks and 128 semilandmarks distributed along each canal and the common crus. The deformation was calculated on the surfaces of the vestibular apparatus, cut at the connection with the cochlea. The studied sample includes 14 different catarrhine (hominoid and cercopithecoid ) genera, representing 21 species. Principal components analysis (PC A) was performed for each method to assess morphological variation among genera. Univariate and multivariate Blomberg's K and Pagel'sÀ were used to evaluate phylogenetic signal in landmark-based method with in the main axes of shape variation. Both analyses yield similar results and discriminate well between humans, great apes and the remaining cararrhines. Monkeys cluster together in the center of the morphospace (hylobatids overlap wirh cercopithecoids in the deformation results), whereas great apes are distinguished (especially in PC l ) for having more eccentric and compressed semicircular canals, and Homo occupies a very distinct position in both PC l and PC 2, due to laceral canal morphology. Focusing on hominoids alone, discrimination among genera increases, although there is still some overlap between gorillas, chimpanzees and orangutans. The analysis therefore discriminates well between the bipedal humans, the agile ricochetal brachiators (gibbons), and the larger-bodied great apes, although among the latter there is some overlap between the knuckle-walking African apes and the more arboreal orangutans. Univariate estimation of the phylogenetic signal presents significant results for PC l (K=2.7;À=0 .99 ) and PC 3 (K=0.47), indicating that PC l reflects shape changes closely approaching phylogenetic divergence. However, multivariate analysis indicates higher variance than that expected under Brownian motion between close relatives(K=0.4; À=0.53), possibly implying some degree of homoplasy in overall vestibular morphology. In conclusion , both 3DGM analyses of primate vestibular apparat us give comparable results and appear to be reliable for discriminating among the studied primates genera on the grounds of positional behavior. Although phylogeny has an effect on the major axis of shape variation, and will be analyzed in future studies, overall vestibular appararus shape is driven by other factors of functional relevance.