Expert systems in interactive species identification: a case of 2,000 caterpillars

The growing application of computer vision in species identification has significantly accelerated biodiversity data collection globally, and the ecologically and morphologically diverse group of exophytophagous caterpillars is no exception. However, many identification tools fail to incorporate critical aspects of a species' biology and ecological role within complex biotic and abiotic systems. This omission often compromises traditional knowledge preservation, contextual understanding, and the transfer of expertise inherent in classical identification keys. To bridge this gap, this study developed an interactive web application designed to enhance species identification and help link taxa to their ecological functions. Central to this work were three standardized data matrices encompassing morphological, ecological, and behavioral polyphenisms, constructed for 2,000 European exophytophagous caterpillar species, addressing 41 descriptive traits and several hundred character states. The matrices were developed through a combination of manual and semi-automated approaches, ensuring robust variability management and data consistency, while both crisp and fuzzy coding schemes were employed to accurately capture and replicate the expert identification process in a systematic format. This application features an intuitive filtering system that enables users to identify species efficiently while preserving the ecological context of each taxon. Beyond its practical utility, this tool supports biodiversity research and educational initiatives by integrating traditional taxonomic methods in contemporary data management, which can facilitate expert-to-user knowledge transfer and improve biodiversity literacy in an era of rapid environmental change.