Inferring locomotor behaviours in Miocene New World monkeys using finite element analysis, geometric morphometrics and machine-learning classification techniques applied to talar morphology

Abstract

The talus is one of the most commonly preserved post-cranial elements in the platyrrhine fossil record. Talar morphology can provide information about postural adaptations because it is the anatomical structure responsible for transmitting body mass forces from the leg to the foot. The aim of this study is to test whether the locomotor behaviour of fossil Miocene platyrrhines could be inferred from their talus morphology. The extant sample was classified into three different locomotor categories and then talar strength was compared using finite-element analysis. Geometric morphometrics were used to quantify talar shape and to assess its association with biomechanical strength. Finally, several machine-learning (ML) algorithms were trained using both the biomechanical and morphometric data from the extant taxa to infer the possible locomotor behaviour of the Miocene fossil sample. The obtained results show that the different locomotor categories are distinguishable using either biomechanical or morphometric data. The ML algorithms categorized most of the fossil sample as arboreal quadrupeds. This study has shown that a combined approach can contribute to the understanding of platyrrhine talar morphology and its relationship with locomotion. This approach is likely to be beneficial for determining the locomotor habits in other fossil taxa.

Keywords: Platyrrhini; finite-element modelling; morphometrics; positional behaviour; statistical learning; talus.

Figure 1.

Broad platyrrhine ecophyletic groups. Colours represent different main locomotion modes. (Online version in colour.)

Figure 2.

(a) Loading scenario tested in the FEA; (b) the 30 landmarks used in the GM analyses. (Online version in colour.)

Figure 3.

von Mises stress distribution for all the analysed specimens. (Online version in colour.)

Figure 4.

Boxplots of von Mises stress distributions for all the analysed specimens. (Online version in colour.)

Figure 5.

Phylomorphospace of the first two morphometric PCs and stress values (MWM) as vertical z-axis. One of the models closest to the mean shape was warped to match the multivariate mean using the thin-plate spline method, and then the obtained average model was warped to represent the variation along the PC axes. In addition, the von Mises stress maps of two extreme models are displayed to facilitate the understanding of the z-axis. (Online version in colour.)

Figure 6.

(a) Standard PLS and (b) the phylogenetic PLS analysis of the shape variables and stress percentile values. One of the models closest to the mean shape was warped to match the multivariate mean using the thin-plate spline method and then the obtained average model was warped to represent the covariation between the two blocks of data for PLS1. (Online version in colour.)

Figure 7.

Dot-plot comparing the accuracy and Cohen’s Kappa values of the different classification models applied to biomechanical (blue squares) and morphometric (red dots) data. The dots represent the average accuracy and Cohen’s Kappa values after performing the ‘leave-group-out’ cross-validation (200 repeats), while the whiskers display their respective 0.95 confidence level. Model acronyms: RF, random forest; LDA, linear discriminant analysis; KNN, k-nearest neighbours; SVM, support vector machine; NB, Naive Bayes; CART, classification and regression trees. (Online version in colour.)

Figure 8.

Decision boundary plots for (a) biomechanical and (b) morphometric data. In (a), only the seventh and 10th intervals are displayed because they contribute the most to class separation, while in (b) only the first two PCs are shown. The space is coloured depending on what locomotor category the (a) SVM or the (b) RF algorithm predict that region belongs to, whereas the lines between coloured areas represent the decision boundaries. Colour intensity indicates the certainty of the prediction in a particular graph area (i.e. darker colours imply a higher probability of belonging to a particular class). Symbols surrounded by a white rim represent misclassified specimens. In (b), one of the models closest to the mean shape was warped to match the multivariate mean using the thin-plate spline method, and then the obtained average model was warped to represent the variation along the two PC axes. (Online version in colour).