Determining the best morphological characters for taxonomic identification of Inga species in the Colombian Andes using correlation, entropy and a discriminant index

The genus <i>Inga</i> (Leguminosae, Caesalpinioideae) is a key component of tropical American forests, yet its high species diversity and morphological complexity hinder accurate identification, especially in montane environments. In this study, we evaluated 49 morphological characters (32 categorical and 17 continuous) from 73 <i>Inga</i> species occurring above 1000 m in Colombia. We applied a multi-criteria framework combining Spearman correlation, Shannon entropy and a discriminant index to prioritise diagnostic characters, based on redundancy, variability and discriminatory power. The results indicate that categorical characters, particularly stipule shape, indumentum type on both lamina surfaces and the petiole, interfoliar nectaries shape, leaflet apex and base shapes are the most informative for species identification and therefore received the highest scores. In contrast, continuous characters showed limited variation among species and were less useful and received lower scores. A usability test demonstrated higher accuracy and faster identifications when using the highest-scoring characters. This framework not only improves identification efficiency in field and herbarium settings but also offers a scalable model for intelligent, user-friendly taxonomic tools. As a complementary output, we present an open-access, interactive identification key, designed to improve the quality and reliability of <i>Inga</i> identifications across the Colombian Andes.