What use are ontogenetic data anyway? Challenges in multivariate modelling of primate tooth formation.

Background: The development of the permanent dentition provides a reliable substrate to infer ontogenetic patterns within and among species. Multivariate methods offer a promising approach to compare taxon-specific patterns.

Aim: This study used multivariate statistical approaches to compare ontogenetic patterns by more comprehensively quantifying variation in crypt and tooth formation scores for the permanent dentition in five catarrhine primate taxa, Homo sapiens, Pan paniscus, Pan troglodytes, Hylobates lar, and Papio anubis.

Subjects and methods: Tooth formation was scored according to published standards for each specimen. Multivariate relationships between teeth were modelled according to a Bayesian multivariate cumulative probit model. Relationships among and between teeth were summarised with correlation matrices, variable loadings plots, and the Frobenius norm. Univariate boxplots were used to contextualise and check the biological salience of the multivariate results.

Results: H. sapiens results corroborate previous research and show a degree of modularity that separates early forming and later-forming teeth. All four other species may show broad correlative patterns, but clear biological patterns are masked due to small sample sizes and/or sample composition.

Conclusion: Even with careful application of statistical procedures, ontogenetic inferences are only as good as the data are comprehensive.