Climate change poses an escalating threat to agricultural productivity and food security across Africa, where more than 60% of livelihoods depend on rain-fed farming. This study develops a predictive-modelling framework for assessing climate-driven impacts on agriculturally relevant rainfall patterns, using Nigeria as an in-depth case study representative of Sub-Saharan Africa's rain-fed agricultural systems. Monthly rainfall records from 41 meteorological stations across Nigeria's six ecological zones (2010–2021) were analysed using temporal decomposition, a Random Forest ensemble model, and the Standardised Precipitation Index to characterise rainfall variability, drought frequency, and their downstream implications for crop productivity. The Random Forest model predicted station-level rainfall with strong accuracy, enabling identification of zones most vulnerable to rainfall-driven productivity shocks. Results reveal a clear ecological gradient: the Forest zone receives the highest and most reliable rainfall, while the Sahel is most drought-prone, yet greater rainfall does not translate linearly into higher productivity, since each zone exhibits a threshold beyond which excess moisture reduces yields through waterlogging and related stress. The Guinea Savanna, despite only moderate rainfall variability, emerges as a disproportionate contributor to estimated productivity losses, reflecting its high exposure as a rain-fed cultivation hub. These findings suggest rainfall variability, more than mean rainfall decline alone, is a key pathway linking climate change to food-security risk in African agriculture. We discuss implications for climate-smart agricultural planning, early-warning systems, and adaptation financing, while noting the need for longer climatological records and multi-factor yield models to strengthen generalisability across the continent.
