Pest risk modelling for climate-smart vegetable production in East Africa

Abstract

Agricultural productivity in East Africa is increasingly threatened by climate-sensitive insect pests. This study investigates the occurrence of key vegetable pests: Brevicoryne brassicae, Plutella xylostella, Maruca vitrata, and Liriomyza mik. A comparative species distribution modelling evaluated pest risk projections under climate change. Four variable selection strategies were tested: case 1 (correlation and variance inflation factor [VIF] filtering across all predictors), case 2 (correlation and VIF filtering of bioclimatic variables merged with other predictors), case 3 (filtering based on environmental values at species occurrence point to approximate realized niches), and case 4 (dimensionality reduction using principal component analysis). These strategies were combined with 3 algorithms: generalized linear model, boosted regression tree, and maximum entropy. In total, 640 spatially occurrence records were analyzed with bioclimatic, elevation, normalized difference vegetation index, solar radiation, wind speed, and land use predictors. Future projections were generated under shared socioeconomic pathways, SSP2-4.5 and SSP5-8.5 (2041–2060), using 4 general circulation models. Model performance varied by species and strategy: case 3 performed best for B. brassicae and P. xylostella, case 1 was most effective for L. mik, and case 2 performed best for M. vitrata. Current suitability maps highlighted high risk zones in central and western Kenya, southwestern Uganda, and Rwanda’s northern province. Projections indicated consistent expansions for P. xylostella and L. mik, while B. brassicae and M. vitrata exhibited localized and uncertain shifts. These findings provide decision-ready pest risk maps that can inform climate-smart pest management, including targeted surveillance, resistant crop varieties, and biological control strategies across in East African vegetable systems.