Phylogenetic reconstruction based on morphometric data is hampered by homoplasies. For example, many similarities in cranial form between primate taxa more strongly reflect ecological similarities rather than phylogenetic relatedness. However, the way in which the different cranial bones constitute cranial form is, if at all, of less functional relevance and thus largely hidden from selection. We propose that these “constructional details” are better indicators of phylogenetic history than any large-scale shape feature or raw form variable. Within a geometric morphometric context, we show how to analyze the relative extent of bones independently of differences in overall shape. We also show how to decompose total shape variation into small-scale and large-scale shape variation. We apply both methods to the midsagittal cranial morphology of papionin monkeys, which are well known for the discrepancy between morphological similarities and phylogenetic relationships. We study phylogenetic signal and functional adaptation using a molecular phylogeny and contextual data on feeding ecology and locomotor behavior. As expected, total cranial shape, bone outline shape, and large-scale shape features were only weakly associated with phylogenetic distance. But the relative bone contributions and small-scale shape features were both highly correlated with phylogenetic distances. By contrast, the association with ecological and behavioral variables was strongest for the outline shape and large-scale shape features. Studies of morphological adaptation and phylogenetic history thus profit from a decomposition of shape variation into different spatial scales.

- Supplementary tables and figures for the main manuscript (Grunstra et al., Systematic Biology)

- 2D coordinates of 70 midsagittal landmarks, collected for 61 specimens from 16 different papionin species as well as 6 specimens from two non-papionin Old World monkey taxa, Cercopithecus mitis and Colobus guereza. Each line in the text file contains the coordinates of one specimen, ordered as (x1, y1, x2, y2, x3, y3, ...). These data were collected by N. Grunstra and S. Bartsch.

- Species labels for the 67 specimens

- R package "prWarps" (written by Anne Le Maítre) with functions to decompose total shape variation into outline and resudial variation and into small-scale and large-scale variation (see main manuscript).

- Worked example using the R package and the landmark data (written by Anne Le Maítre) .