Climate-Responsive Forecasting of Hymenia recurvalis Outbreaks in Nigerian Agroecological Zones

The erratic nature of the current climate increase has resulted in the distribution of Hymenia recurvalis , a significant pest that affects the leafy vegetable cultivation of Nigeria across the agroecological zones. Nevertheless, the local models for early outbreak detection continue to be inadequate even as the pest becomes more of a threat. This problem is addressed in this study by developing a climate-based forecasting system that allows the improvement of integrated pest management strategies. The pest attack data were gathered from ten vegetable-production sites in the zones Agroecological—Sahel, Sudan Savanna, Northern Guinea Savanna, Southern Guinea Savanna, and Forest, over five years (2019–2023). During the research, the areas of Katsina, Kano, Zaria, Abuja, and Ibadan were geographically located for precise analysis. NiMet was the source of meteorological data, namely temperature, rainfall, and humidity. The record containing 600 geo-referenced records was analysed based on the algorithms of Random Forest and Logistic Regression. Random Forest was eventually found to be much more efficient in the predictions than Logistic Regression, with a strong 91% accuracy (AUC = 0.94) and high levels of other indicators as well. The regression analysis showed that both rainfall and the temperature minimum are the main factors responsible for the outbreak (p < 0.01). The maximal risk occurred in the post-rainy season, mainly in the southern part of the Guinea and Forest regions. The lead time of the high-risk situation of recorded cases proved that the model was correctly fitted to the historical observations of the prevalence. The practicality of a machine-tailored approach to the local scale of agroecology is proven in the sense that it would enable efficient countermeasures to be set in place in good time, and the farmer's efforts would be able to withstand the projected climate change. Work could be done along the lines of data and vegetation indices, as well as on-the-go mobile advisory outlets for the eruption alerts in real-time to perfect the predictive power of these tools. Extension frameworks of such tools will be instrumental in promoting sustainable pest management practices and, therefore, accelerating food security and agricultural productivity.