The impact of 3-dimensionality on 2D and 3D comparative geometric morphometric studies: a case study on the skulls of Southeastern European moles

Abstract The potential of shape variation to distinguish between species can be examined using quantitative approaches such as 2D or 3D geometric morphometrics (GM). As these methods differ in requirements and effort, their ability to resolve between different species is crucial. Thus, we tested two hypotheses: (1) approaches perform better with increasing complexity, i.e., 3D GM > 2D GM and (2) 2D approach performs comparatively better in flat and rather 2-dimensional structures than in complex 3-dimensional structures. These hypotheses were tested using crania and mandibles of four mole species of the genus Talpa ( T. stankovici , T. europaea , T. caeca , T. martinorum ) from the Balkan and Italian peninsula. Our results based on 3D CT scans and 2D images rejected both hypotheses: (1) 2D GM performs equally well in the highly 3-dimensional cranium. This becomes evident only when the 3D dorsal and ventral landmarks are analysed separately from each other (2) on the other hand 3D GM performs better in the flat mandible in relation to 2D. A potential explanation lies in the complexity of the structures, which was measured by calculating the Dirichlet Normal Energy (DNE). Our results showed that DNE is higher in the 3-dimensional cranium than in the flat mandible. However, the landmark-based mesh of the mandible had a higher complexity than the one of the cranium. This implies that the landmark selection and number have a stronger impact on performance than the method itself. Nevertheless, a broader comparison between different structures and organisms is required to confirm these patterns.