Approaches to studying endocranial morphology in Euarchontoglires: Assessing sources of error for a novel and biologically informative set of landmarks

Abstract

Despite its importance for making biologically meaningful conclusions in geometric morphometric analyses, landmark error is inconsistently assessed. We evaluate a set of 30 landmarks designed to capture shape variation among six major brain regions on endocasts of Euarchontoglires (Primates, Scandentia, Dermoptera, Lagomorpha, Rodentia). Seven trials were performed by three observers on virtual endocasts of three species: Alouatta palliata (Primates), Ochotona pallasi (Lagomorpha), and Octodon degus (Rodentia). Standard deviation for all landmarks was evaluated relative to mean inter-landmark distance, centroid size, and centroid radius. A Procrustes analysis of variance (ANOVA) was performed to assess inter- and intraobserver error. Results show that standard deviations in landmark placement across trials, species, and observers are low (≤ 1.5%). The Procrustes ANOVA found significant differences between observers, accounting for only a small portion of the variation (R² = 0.002, p ≤ 0.0001); most of the variation is attributed to species (R² = 0.993, p ≤ 0.0001). In a separate analysis, a landmark sampling evaluation curve (LaSEC) which included two additional curve semilandmark sets along the sagittal aspect of the neocortex and cerebellum was applied to 40 landmarked species spanning all five orders of Euarchontoglires. The LaSEC indicated that 99% of the variation is captured at by 30 landmarks. Morphological patterns previously thought to characterize these groups are replicated in a principal component analysis of landmark data for 40 species. Specifically, most variation relates to the relative scale of the neocortex and olfactory bulbs and the flexion of the basicranium. Overall, these landmarks are highly replicable and able to represent morphological patterns within a diverse group such as Euarchontoglires.