Phylogenetic analysis of geometric morphometric data: A study case in Didelphidae

• Published in: Zoologica scripta Vol. 54; no. 4; pp. 437 - 453
• Main Authors: Saguir, Sergio O., Haidr, Nadia S., Flores, David A., Catalano, Santiago A.
• Format: Journal Article
• Language: English
• Published: Oslo Wiley Subscription Services, Inc 01.07.2025
• Subjects: Datasets; Didelphidae; Empirical analysis; landmarks; morphology; Morphometry; Performance evaluation; Phylogenetics; Phylogeny; Sampling; semi‐landmarks; skeleton; superimposition
• ISSN: 0300-3256; 1463-6409
• DOI: 10.1111/zsc.12722

The analysis of Geometric Morphometric Data (GMD) in a phylogenetic context is a common practice in current evolutionary analyses. However, its use as evidence to recover phylogenetic relationships remains controversial. While simulation analyses have indicated theoretical limits for phylogenetic inference based on GMD, empirical analyses have shown mixed results and emphasized the importance of proper character sampling. In this study, we evaluated the phylogenetic performance of GMD in phylogenetic reconstruction using a newly generated dataset on the Didelphidae family. This dataset comprises the largest character sampling generated to date, including information from 10 different skeletal structures represented by 14 landmark configurations. Specifically, our objectives were: (i) to evaluate how different superimposition procedures and the inclusion of semilandmarks affect phylogenetic inferences, (ii) to compare the phylogenetic performance of GMD and traditional characters, and (iii) to compare the phylogenetic information of cranial and postcranial data. We found that trees obtained from GMD and traditional discrete characters exhibited similar levels of correspondence with a reference molecular tree. The inclusion of semilandmarks led to worsened results, regardless of the methodology used to place them, and we found no clear evidence for the superiority of any particular landmark superimposition approach. Our results align with previous analyses demonstrating that the inclusion of a higher number of skeletal structures improves results. We discuss these findings in the context of the ongoing debate about the utility of GMD to infer phylogenetic relationships.