Standard descriptive matrices in the identification of exophytophagous caterpillars

Abstract

Identification of exophytophagous lepidopteran larvae is a necessity for researchers in biological disciplines ranging from biodiversity inventorying to research in parasitoid evolution and species monitoring. The lack of expertise in the field jeopardizes the outcomes of further investigations and recording of the multilevel plasticity of juvenile Lepidoptera. This paper offers an improvement to the existing haphazard approach by developing 41 simplified characters that include 150 morphological, behavioral and autecological states and their delineation, visual validation, and a descriptive matrix for 83 heterogeneous species. By combining the states into all possible identification scenarios, the matrix revealed 582 morphological, habitat and resource polyphenisms for the mentioned species. The categorical nature of the data implied the use of categorical principal component analysis to visualize the discriminative capacity without character relationship assumptions. The object-point biplot was used to derive the K value for K-mode clustering, while the cluster membership was introduced as a labeling variable to further inspect the grouping pattern. The results of this descriptive analytic research indicate that descriptive matrices will allow continuous expansion and fine examination of many different species assemblages. From interactive identification keys to machine learning training, the presented framework can make data storage and interpretation significantly more attainable.