A Machine Learning Approach to Study Demographic Alterations in Honeybee Colonies Using SDS–PAGE Fingerprinting

SIMPLE SUMMARY: Honeybees are vital pollinators for the human food chain. Colony depopulation is a serious threat to Apis mellifera populations and unfortunately it is also one of the most elusive and difficult to study. This research deals with the problem at its foundation: population imbalances. The proposed method allows to discriminate, with remarkably good performances, precocious foragers from proper aged ones using SDS-PAGE patterns of haemolymph proteins. Implications and future perspectives are discussed. ABSTRACT: Honeybees, as social insects, live in highly organised colonies where tasks reflect the age of individuals. As is widely known, in this context, emergent properties arise from interactions between them. The accelerated maturation of nurses into foragers, stimulated by many negative factors, may disrupt this complex equilibrium. This complexity needs a paradigm shift: from the study of a single stressor to the study of the effects exerted by multiple stressors on colony homeostasis. The aim of this research is, therefore, to study colony population disturbances by discriminating overaged nurses from proper aged nurses and precocious foragers from proper aged foragers using SDS-PAGE patterns of haemolymph proteins and a machine-learning algorithm. The KNN (K Nearest Neighbours) model fitted on the forager dataset showed remarkably good performances (accuracy 0.93, sensitivity 0.88, specificity 1.00) in discriminating precocious foragers from proper aged ones. The main strength of this innovative approach lies in the possibility of it being deployed as a preventive tool. Depopulation is an elusive syndrome in bee pathology and early detection with the method described could shed more light on the phenomenon. In addition, it enables countermeasures to revert this vicious circle.