COMPUTATIONAL FRAMEWORK OF CLIMATE-INDUCED FOOD INSECURITY VIA FIREFLY-GA HYBRID MODEL IN NIGERIA’S CEREAL PRODUCTION

Abstract

Climate variability continues to threaten agricultural productivity across sub-Saharan Africa, with noticeable effects on yield stability in climate-sensitive production systems. This study develops a cascade–parallel Firefly–Genetic Algorithm (FA–GA) hybrid framework for crop yield prediction and agronomic optimization using twenty years (2006–2024) of Nigerian climate and crop data. The proposed architecture integrates the global search capability of the Genetic Algorithm with localized refinement via the Firefly Algorithm to enable balanced exploration and exploitation within a high-dimensional decision space. Model performance was evaluated using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Coefficient of Determination (R²), and normalized MAE-derived accuracy. The FA–GA hybrid achieved an accuracy of 96.4%, an R² of 0.96, an RMSE of 0.042 t/ha, and an MAE of 0.031 t/ha, outperforming the standalone GA, standalone FA, and Random Forest benchmarks under identical experimental conditions. Convergence analysis showed approximately 35% fewer iterations compared to standalone GA, indicating improved computational efficiency. Stress simulations further demonstrated adaptive capacity. Optimized planting windows under a 15-day dry spell retained approximately 15% of maize yield relative to non-optimized scenarios, while targeted potassium adjustment mitigated projected heat-induced rice yield losses by 8%. The results indicate that the proposed hybrid framework delivers both high predictive precision and actionable agronomic recommendations under climate perturbations. This dual capability positions the FA–GA model as a viable decision-support tool for climate-resilient agriculture and national food security planning.