New insights into the phenotypic covariance structure of the anthropoid cranium

In complex organisms, suites of non-random, highly intercorrelated phenotypic traits, organized according to their developmental history and forming semi-autonomous units (i.e. modules), have the potential to impose constraints on morphological diversification or to improve evolvability. Because of its structural, developmental and functional complexity, the cranium is arguably one of the best models for studying the interplay between developmental history and the need for various parts of a structure to specialize in different functions. This study evaluated the significance of two specific types of developmental imprints in the adult anthropoid cranium, those imposed by ossification pattern (i.e. ossification with and without a pre-existing cartilaginous phase) and those imposed by tissue origin (i.e. tissues derived principally from neural-crest vs. those derived from paraxial mesoderm). Specifically, this study tests the hypothesis that the face and the basicranium form two distinct modules with higher within-unit trait integration magnitudes compared with the cranium as a whole. Data on 12 anthropoid primate species were collected in the form of 20 three-dimensional landmarks digitized on cranial surface models that sample the basicranium as well as regions of functional importance during feeding [corrected]. The presence of a significant modularity imprint in the adult cranium was assessed using a between-region within-species comparison of multivariate correlations (RV coefficients) obtained with partial least-squares, using within-module within-species eigenvalue variance (EV), and using cluster analyses and non-metric multidimensional scaling. In addition to addressing the validity of the cranial modularity hypothesis in anthropoids, this study addressed methodological aspects of the interspecific comparison of morphological integration, namely the effect of sample size and the effect of landmark number on integration magnitudes. Two methodological findings that are of significance to research in morphological integration are that: (i) a smaller sample size increases integration magnitude, but preserves the pattern of variation of integration magnitudes from block to block within species; and that (ii) the number of landmarks per cranial block does not significantly impact block integration magnitude measured as EV. Results from the analyses testing for cranial modularity imprints in the adult anthropoid cranium show that some facial landmarks covary more strongly with basicranial landmarks than with other facial landmarks. Cluster methods, non-metric multidimensional scaling and, to an extent, RV results show that the rostral and the zygomatic landmarks covary more strongly with the basicranial landmarks than they do with the molar landmarks. However, the rostral-zygomatic-basicranial block, the molar block, the facial block, the basicranial block and the other analyzed cranial and facial blocks are not more integrated than the cranium. Thus, the morphological variation in the adult anthropoid cranium is not significantly constrained by at least two of the potential developmental sources of its covariance structure.