morphometrics with machine learning to enhance taxonomic resolution

Understanding morphological diversity in bats is central to ecological classification, evolutionary interpretation, and conservation planning. This study analyzed key morphological, ecological, and genetic traits to distinguish microbat and megabat lineages using an integrated set of computational tools. Kernel Density Estimation (KDE) was first applied to determine dominant morphological signatures, revealing that elongated fingers with membrane wings, large wingspans, and patagium-supported forelimbs were the most consistently expressed traits across the dataset. Principal Component Analysis (PCA) further demonstrated clear multidimensional separation between species groups, with PC1 capturing variation associated with agile, echolocation-driven flight, while PC2 reflected robust body forms and craniofacial specialization. These component loadings confirmed strong evolutionary divergence between microbats and megabats. To deepen the analysis, Random Forest and Logistic Regression models were employed to quantify the predictive importance of ecological, physiological, and genetic factors. Results showed that habitat, DNA characteristics, shape, protection from external threats, and lifespan exerted substantial influence on species classification. These findings indicate that morphological differentiation is not driven by structural traits alone but emerges from an interplay of ecological pressure, life-history strategy, and genomic adaptation. Overall, the study demonstrates the power of integrating morphometrics with machine learning to enhance taxonomic resolution and interpret evolutionary patterns. By combining distribution-based, multivariate, and predictive modelling approaches, this research contributes a comprehensive and methodologically innovative framework for understanding species divergence and morphological variability within one of the most diverse mammalian orders