Quantifying shell outline variability in extant and fossil Laqueus (Brachiopoda: Terebratulida): are outlines good proxies for long-looped brachidial morphology and can they help us characterize species?
Lopez Carranza, Natalia; Carlson, Sandra (2020), Quantifying shell outline variability in extant and fossil Laqueus (Brachiopoda: Terebratulida): are outlines good proxies for long-looped brachidial morphology and can they help us characterize species?, Dryad, Dataset, https://doi.org/10.5061/dryad.sqv9s4n23
Extant and extinct terebratulide brachiopod species have been defined primarily on the basis of morphology. What is the fidelity of morphological species to biological species? And how can we test this fidelity with fossils? Taxonomically and phylogenetically, the most informative internal feature in the brachiopod suborder Terebratellidina is the geometrically complex long-looped brachidium, which, given their fragile nature, are not commonly preserved in the fossil record. In their absence, it is essential to test other sources of morphological data when trying to recognize and identify species. We analyzed valve outlines and brachidia in the genus Laqueus to explore the utility of shell shape in discriminating extant and fossil species. Using geometric morphometric methods, we quantified valve outline variability using elliptical Fourier methods and tested whether long-looped brachidial morphology correlates with shell outline shape. We then built classification models based on machine learning algorithms using outlines as shape variables to predict fossil species’ identities. Our results demonstrate that valve outline shape is significantly correlated with long-looped brachidial shape and that even relatively simple outlines are sufficiently morphologically distinct to enable extant Laqueus species to be identified, validating current taxonomic assignments. These are encouraging results for the study and delimitation of fossil terebratulide species, and their recognition as biological species. In addition, machine learning algorithms can be successfully applied to help solve species recognition and delimitation problems in paleontology, especially when morphology can be characterized quantitatively and analyzed statistically.