Abstract

Introduction:  Carollia is characterized by the difficulty in identifying individuals from different species. In Panama, no study has addressed the accurate classification and identification of this genus; however, molecular and phylogenetic studies conducted in other regions of the Americas highlight the challenges of morphological differentiation. Taxonomic keys for identifying this genus tend to vary, complicating species identification in Panamanian localities.

Objective:  To evaluate the external morphometric and morphological characteristics of Carollia specimens to facilitate species identification through multivariate statistical techniques.

Methods:  We used existing data matrices, which were updated in the field from October 2022 to January 2023 using mist nets. External morphometric measurements (tail, forearm, hand wing, tibia, calcaneus, tragus, hair color, and body size) and individual characteristics were recorded. A total of 263 specimens representing the four species reported in Panama were documented. We used univariate statistics to compare each characteristic across species. Subsequently, we applied multivariate analyses, including principal component analysis (PCA), linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), and partial least squares discriminant analysis (PLS-DA), to identify the species based on external morphological and morphometric characteristics. Decision trees were also used for species classification.

Results:  Linear discriminant analysis (LDA) and decision trees proved to be the best methods for species classification, achieving up to 99% accuracy. The most relevant characteristics for classification were tail and forearm lengths.

Conclusion:  Morphometric characteristics alone do not provide sufficient discrimination among species. However, when parameters are analyzed using multivariate models, the discriminatory accuracy is significantly improved.

Key words: principal component analysis; discriminant analysis; decision trees