Data from: Machine learning can accurately assign fossil and extant species to crown toxicoferan (Reptilia: Squamata) groups using inner ear shape data.

This is one of the two sets of data files for: Meghan R Forcellati, James G Napoli, Dalton Meyer, Akinobu Watanabe, Roger Benson, Christopher J Raxworthy. "Machine learning can accurately assign fossil and extant species to crown toxicoferan (Reptilia: Squamata) groups using inner ear shape data." Zoological Journal of the Linnean Society, Volume 206, Issue 3, March 2026, zlaf188, https://doi.org/10.1093/zoolinnean/zlaf188

Please see the accompanying Data Dryad file, which contains the code: https://doi.org/10.5061/dryad.2fqz612zp

We used 3D geometric morphometric landmarks to capture data on the shape of the inner ears of squamates. As described in the main text, we used the 3D geometric morphometric data to perform both inferential analyses and statistical learning to understand whether inner ear shape data is associated with habitat usage in extant squamates, which can be predicted in fossil squamates, and also whether it contains reliable phylogenetic signal, which can be used to classify fossil squamates. All Landmark Data was collected in LandmarkEditor (Wiley, D. F., Amenta, N., Alcantara, D. A., Ghosh, D., Kil, Y. J., Delson, E., ... & Hamann, B. (2005, October). Evolutionary morphing. In VIS 05. IEEE Visualization, 2005. (pp. 431-438). IEEE). Thus, in summary, the main variables used in this analysis are: ecology/habitat, a measurement of whether species are arboreal, aquatic, fossorial, or 'other'/generalist (with conservative and semiconservative classifier variables being used to investigate whether being stricter in what we consider 'aquatic' effects our analyses); centroid size, a measurement calculated during generalized procrustes alignment of the centroids of the inner ear, which we used to measure allometry, and limb status, whether species are limb-reduced or fully-limbed.

All data, which are .pts files and data analysis files used in our paper, is published here under a CC BY-NC-SA-4.0 license. However, the individual specimens we collected this data from are from MorphoSource, and we include detailed descriptions of the licensing information for each one. Please reference the .csv file with licensing information ("SupplementaryDataS1_AcknowledgmentsReferencesSpecimens") for a precise understanding of exactly how to appropriately attribute each specimen in this dataset, and what terms you are allowed to use the data for. 

Thank you to MorphoSource contributors who shared their data. Please also see Yi, H., & Norell, M. A. (2015). The burrowing origin of modern snakes. Science advances, 1(10), e1500743; Zheng, Y., & Wiens, J. J. (2016). Combining phylogenomic and supermatrix approaches, and a time-calibrated phylogeny for squamate reptiles (lizards and snakes) based on 52 genes and 4162 species. Molecular phylogenetics and evolution, 94, 537-547.

Funding - Laidlaw Fellowship, Columbia Work Exemption Program Grants (2020, 2021, 2022, 2023), Richard Gilder Graduate School Fellowship, Yale Biospheric Studies Early Grant, American Museum of Natural History Vertebrate Paleontology Travel Fund, Supporting Tomorrow’s Researchers and Today’s Advocates Program (National Science Foundation Grant Division of Earth Sciences (EAR) 2331991); and to scan AMNH FARB 1645 and AMNH FARB 8138, the National Science Foundation Division of Earth Sciences (EAR) 0959384. All additional funding from the data used by this study is listed in the "SupplementaryDataS1_AcknowledgmentsReferencesSpecimens.csv" file.