Michael Osunga^1,, Felix Mutua¹, Robinson Mugo²

Maize lethal necrosis (MLN) is a disease that attacks maize crops with significant impacts on both food security and nutrition security on smallholder farmers in Kenya. The study used spatial regression analysis to model MLN severity on sampled farm fields in Bomet County, Kenya. The modelling analysis integrated spatial information based on derived crop mask, on-site derived MLN disease severity index at an optimal maize growing season and phenological stage. Relevant ecological variables derived spatially including temperature, rainfall, soil moisture and slope were identified and fed into a spatial regression model. Significant ecological variables were weighted and used as basis for generating spatially explicit MLN severity index map. MLN affected farms have spatial dependence with MLN severity becoming less correlated the further away from each MLN affected farm field. The ecological variables have negative influence on MLN severity except for temperature. Soil moisture, rainfall and slope are the most significant determinants of MLN severity index in Bomet (all<p 0.05), with high MLN severity areas identified in Chebunyo, Sigor and Kipreres. This study would help in MLN epidemiological surveillance and in developing site-specific control measures and interventions. The spatial model used in this study could be replicated and up-scaled to other MLN prone areas in Kenya and in Africa coupled with other statistically significant spatiotemporal ecological variables to fully understand and ascertain MLN disease outbreak.

maize lethal necrosis, disease severity index, spatial regression