Assessing the application of landmark-free morphometrics to macroevolutionary analyses

Area of Science:

• Evolutionary biology
• Morphometrics
• Computational anatomy

Background:

• 3D geometric morphometrics revolutionized phenotypic evolution studies.
• Manual landmarking is time-consuming, operator-dependent, and limits cross-taxa comparisons.
• Automated, landmark-free methods offer potential but are underexplored for macroevolutionary analyses.

Purpose of the Study:

• To evaluate automated, landmark-free approaches for macroevolutionary analyses.
• To compare Deterministic Atlas Analysis (DAA) with high-density geometric morphometrics.
• To assess the utility of these methods across diverse mammalian taxa.

Main Methods:

• Applied Large Deformation Diffeomorphic Metric Mapping (LDDMM) via DAA.
• Utilized a dataset of 322 mammals from 180 families.
• Standardized mixed-modality scan data (CT and surface scans) using Poisson surface reconstruction.

Main Results:

• Standardization improved DAA-manual landmarking correspondence, though differences persisted for specific taxa (Primates, Cetacea).
• Both DAA and manual methods yielded comparable, yet varying, estimates for phylogenetic signal, morphological disparity, and evolutionary rates.
• DAA demonstrated potential for large-scale, cross-taxa studies due to enhanced efficiency.

Conclusions:

• Landmark-free methods like DAA offer efficiency gains for macroevolutionary morphometrics.
• Challenges remain, particularly regarding accuracy across morphologically disparate taxa.
• Further research and methodological improvements are needed for widespread adoption of landmark-free analyses.