Multi-pronged abundance prediction of bee pests’ spatial proliferation in Kenya

• DOI: 10.1016/j.jag.2024.103738
• Publication Date: 2024-02-29T10:50:36Z
• Authors: David Masereti Ma...; Elfatih M. Abdel-...; John Odindi; Onisimo Mutanga; Tobias Landmann; Henri E.Z. Tonnang
• Keywords: Physical geography; Food security; 04 agricultural and veterinary sciences; Beehealth; GB3-5030; Environmental sciences; Machine learning; Climate change; Human footprint; 0401 agriculture, forestry, and fisheries; GE1-350; Bee pest abundance

Bee farming and beehealth are threatened by climate change, agricultural and agrochemicals intensification, and bee pests and diseases. Among these threats, bee pests have particularly been identified as a major obstacle to beehealth. Although previous studies have endeavoured to establish bee pests’ spatial distribution, their seasonal abundance in the landscape remains poorly understood. Hence, this study sought to determine factors that influence the abundance and spatial proliferation of bee pests in Kenya. Abundance data on Varroa destructor, Oplostomus haroldi, Galleria mellonella and Aethina tumida were collected from apiaries in Kenya during the wet and dry seasons. The abundance data were fitted to non-conflating human footprint datasets, satellite derived vegetation phenological, topographical and bioclimatic variables. The results indicated a significant (p ≤ 0.05) seasonal influence on bee pests’ abundance, while precipitation was the most relevant on most bee pests’ abundance prediction models. Topographic and vegetation phenological influence varied across the landscapes while anthropogenic influence was comparatively low. High seasonality in bioclimatic variables influenced the projected (year 2055) spatial and abundance risk levels of bee pests across the study area. The V. destructor and A. tumida prediction models for current and future epochs ranked excellent in their performance, while O. haroldi and G. mellonella were ranked good and fair, respectively. Due to their precision, this study concluded that these models could reliably be used to establish bee pests’ high-risk areas for management and mitigation purposes.