Refine
Year of publication
- 2024 (2)
Document Type
- Article (2)
Language
- English (2)
Has Fulltext
- yes (2)
Is part of the Bibliography
- no (2)
Keywords
- group differences (2)
- ANOVA (1)
- North American marmots (1)
- P values (1)
- allometry (1)
- discriminant analysis (1)
- founder effect (1)
- interspecific variation (1)
- island population (1)
- mandible (1)
Taxonomy lays the foundations for the study of biodiversity and its conservation. Procrustean geometric morphometrics (GMM) is a most common technique for the taxonomic assessment of phenotypic population differences. To measure biological variation and detect evolutionarily significant units, GMM is often used on its own, although it is much more powerful with an integrative approach, in combination with molecular, ecological and behavioural data, as well as with meristic morphological traits. GMM is particularly effective in taxonomic research, when applied to 2D images, which are fast and low cost to obtain. Yet, taxonomists who may want to explore the usefulness of GMM are rarely experts in multivariate statistical analyses of size and shape differences. In these twin papers, I aim to provide a detailed step-by-step guideline to taxonomic analysis employing Procrustean GMM in user-friendly software (with tips for R users). In the first part (A) of the study, I will focus on preliminary analyses (mainly, measurement error, outliers and statistical power), which are fundamental for accuracy, but often neglected. I will also use this first paper, and its appendix (Appendix A), to informally introduce, and discuss, general topics in GMM and statistics, that are relevant to taxonomic applications. In the second part (B) of the work, I will move on to the main taxonomic analyses. Thus, I will show how to compare size and shape among groups, but I will also explore allometry and briefly examine differences in variance, as a potential clue to population bottlenecks in peripheral isolates. A large sample of North American marmot mandibles provides the example data (available online, for readers to replicate the study and practice with analyses). However, as this sample is larger than in previous studies and mostly unpublished, it also offers a chance to further explore the patterns of interspecific morphological variation in a group, that has been prominent in mammalian sociobiology, and whose evolutionary divergence is complex and only partially understood.
In this second part of the study, using a ‘clean’ dataset without very low precision landmarks and outliers, I describe how to compare mandibular size and shape using Procrustes methods in adult North American marmots. After demonstrating that sex differences are negligible, females and males are pooled together with specimens of unknown sex and species are compared using a battery of tests, that estimate both statistical significance and effect size. The importance of allometric variation and its potential effect on shape differences is also explored. Finally, to provide potential clues on founder effects, I compare the magnitude of variance in mandibular size and shape between the Vancouver Island marmot (VAN) and the hoary marmot, its sister species on the mainland. In almost all main analyses, I explore the sensitivity of results to heterogeneous sample size and small samples using subsamples and randomized selection experiments. For both size and shape, I find a degree of overlap among species variation but, with very few exceptions, mean interspecific differences are well supported in all analyses. Shape, in particular, is an accurate predictor of taxonomic affiliation. Allometry in adults, however, explains a modest amount of within-species shape change. Yet, there is a degree of divergence in allometric trajectories that seems consistent with subgeneric separation. VAN is the most distinctive species for mandibular shape and mandibular morphology suggests a long history of reduced variation in this insular population. Geometric morphometrics (GMM) is a powerful tool to aid taxonomic research. Regardless of the effectiveness of this family of methods and the apparent robustness of results obtained with GMM, however, large samples and careful measurements remain essential for accuracy. Even with excellent data, morphometrics is important, but its findings must be corroborated with an integrative approach that combines multiple lines of evidence to taxonomic assessment. The analytical protocol I suggest is described in detail, with a summary checklist, in the Appendix, not to miss important steps. All the analyses can be replicated using the entire dataset, which is freely available online. Beginners may follow all the steps, whereas more experienced researchers can focus on one specific aspect and read only the relevant chapter. There are limitations, but the protocol is flexible and easy to improve or implement using a programming language such as R.