Constructing a multiple‐part morphospace using a multiblock method

Abstract

Popular current methods for quantifying variation in biological shape are well‐suited to analyses of isolated parts (e.g. the same bone from the skeletons of many individuals). An analytical challenge exists for quantifying variation between the shapes of multiple‐part objects where each part has a different position, rotation or scale (e.g. partial or whole articulated skeletons). We investigated regularised consensus principal component analysis (RCPCA) as a multiblock method for quantifying variation in the shape of multiple‐part objects. Multiblock methods are routinely used in other big data research fields such as bioinformatics/medicine, marketing and food research, but have not been widely embraced for evolutionary biology research. We have created the new package morphoBlocks for the r programming language to make RCPCA more accessible for shape evolution research. morphoBlocks provides a complete workflow for formatting, analysing and visualising the variation between multiple‐part objects by integrating functions from a diverse range of other packages. In particular, global components produced by RCPCA provide a consensus space that we present here as a morphospace for multiple‐part objects. morphoBlocks is demonstrated with a case study of manually placed landmarks and automatically placed pseudolandmarks from the partial wing skeletons of 15 extant penguin species and five fossil penguin species. Our case study provides quantitative support for a historical hypothesis about the magnitude and mode of morphological change across the evolutionary history of penguins. RCPCA can be used to analyse two‐ or three‐dimensional datasets with 10s of landmarks, or 100s to 1,000s of semilandmarks or pseudolandmarks, from 10s to 100s of specimens comprised of two or more parts. We use morphoBlocks on a small three‐bone case study and provide a framework for applying this method to much larger studies investigating the ecological or evolutionary significance of multiple‐part objects.

Small‐ to medium‐sized mammals show greater morphological disparity in cervical than lumbar vertebrae across different terrestrial modes of locomotion

“…All analyses were done in R Statistical Environment version 4.2.3 (R Core Team, 2023 ). The analytical libraries and functions used (herein noted as library::function ) were geomorph v.4.0.6 (Adams et al., 2023 ; Baken et al., 2021 ), morphoBlocks v.0.1.0 (Thomas & Harmer, 2022 ), RRPP v.1.4.0 (Collyer et al., 2018 ; Collyer & Adams, 2023 ), and vegan v.2.6–4 (Oksanen et al., 2022 ).…”

“…This is because geometric morphometric methods are not suitable for multi‐part, articulated, and moveable structures (but see Vidal‐Garcia et al., 2018 and Rhoda et al., 2021 for different solutions). The recently available technique of multi‐block method for morphometric purposes (Thomas et al., 2023 ) serves as a suitable method for evaluating patterns of shape variation among multiple elements simultaneously, such as different vertebrae from a vertebral column. This approach has the benefit that the shape of all vertebrae can be examined together, irrespective of the shape of the articulated column from which they come.…”

Realising the promise of large data and complex models

“…when analysing skeletons rather than individual bones). With this focus, Thomas et al (2023) propose a method based on regularised consensus principal components analysis to be able to summarise and compare shape variation in multipart morphospaces. Importantly, they also provide an accompanying R package, to permit wider usage and impact within the large scientific community.…”

“…when analysing skeletons rather than individual bones). With this focus, Thomas et al (2023) propose a method based on regularised consensus principal components analysis to be able to summarise and compare shape variation in multipart morphospaces.…”

Evolutionary ecomorphology for the twenty-first century: examples from mammalian carnivores

“…However, new approaches have recently been proposed for considering multiple elements [ 144 ]. For example, ‘Morphoblocks’ [ 145 ] examines variance–covariance matrices within and between multiple independent ‘blocks’ of landmarks to identify both common and unique morphological variation. Techniques such as these could be very powerful for testing ecomorphological signal across different combinations of elements, to facilitate extracting the maximum biological information from rare and precious fossil material.…”