Building and Visualising Ordinations of Shape Data in R With Morphospace

Abstract

Ordination is a critical step of geometric morphometrics that allows simplification of high-dimensional shape spaces into lowdimensional representations summarising shape variation. While this is routinely used to visualise the main patterns of morphometric data, the lack of a unified approach curbs researchers' ability to make the most of available ordination and visualisation methods. Here we introduce morphospace, an R package providing a streamlined approach to shape ordination and visualisation, intended to enhance the biological relevance of the analysis. This package integrates a series of tools into a workflow including (1) filtering variation associated with nuisance factors out from shape data; (2) building ordinations emphasising different types of biological signal and mapping shape variation into them; (3) projecting elements onto this morphospace, including empirical or theoretical shapes, shape clusters, morphometric axes, phylogenetic trees or performance landscapes; and (4) combining shape variation with non-shape information to create alternative visualisations. We describe the basic mechanics and capabilities of morphospace, as well as its integration with other R packages, and briefly discuss the patterns it can help expose. Then, we showcase its applications by working three examples focused on phenomena commonly targeted in palaeontological and evolutionary research. The tools and workflow provided by morphospace facilitate the creation of publication-ready visualisations to accompany results of dedicated statistical tests and enhance heuristic exploration of the patterns present in shape data and the underlying processes that generated them.