Climate change threatens smallholder agriculture and food security in the Global South. While cropland expansion is often used to counter adverse climate effects despite ecological trade-offs, the benefits for diets and nutrition remain unclear. This study quantitatively examines relationships between climate anomalies, forest loss from cropland expansion, and dietary outcomes in Nigeria, Africa’s most populous country. Combining high-resolution data on forest cover and climate variables within random forest and panel regression models, we find that 25 to 31% of annual forest loss is linked to climate variability. Using georeferenced household survey data, we then find that changes in forest cover have a significant positive association with changes in child diet diversity—a key proxy of nutritional adequacy—while cropland expansion does not, suggesting that such forest conversions may be an ineffective climate adaptation strategy for improving nutrition. Our findings highlight the potential of nutrition-sensitive climate adaptation to enhance yields, promote nutritious cropping choices, and protect remaining forests.
10 April 2026 in Environmental Research: Food Systems
Smallholder systems contribute to nearly three-quarters of food production in sub-Saharan Africa and employ half of the continent’s population. At the same time, these primarily rainfed systems are sensitive to climate variability and have limited capital for obtaining buffering inputs. Both their significance and vulnerability require improved agricultural monitoring. However, the small field sizes of smallholder farms (<2 ha), their complex multi- and intercropped systems, and persisting growing season cloud cover in tropical and sub-tropical regions where many smallholder systems occur mean that the fine-scale cropping patterns of smallholder landscapes remain poorly understood. Focusing on the Oyo state in southwest Nigeria, the country with the largest smallholder population in Africa, we seek to overcome these challenges by mapping the extent of maize and cassava cultivation, two of the country’s most widely cultivated staple crops. Using geolocated GoPro images collected as ground-truth data, we trained machine/deep-learning models (i.e. random forest and convolutional neural network) to classify high-resolution (10 m) multispectral satellite imagery. In doing so, we estimated that 175 604 ha of cassava and 150 538 ha of maize would be cultivated across the state in 2022, in close agreement with official statistics and coarser global crop-type products. We then evaluated a suite of dry-season vegetation indices to assess the prevalence of maize–cassava intercropping (in which maize is harvested at the end of the rainy season and cassava persists), finding indicative evidence of the practice in more than a third (45.2%) of maize locations. Thus, our findings challenge conventional remote sensing assumptions (i.e. monocultured cultivation) and provide important advances for scalable and transferable fine-scale crop type mapping in complex smallholder systems.
Irrigation plays a critical role in global food production and climate adaptation and exercises profound influence over humanity's water use. Yet despite its critical importance, there is a persistent lack of understanding of fine-scale irrigation patterns across the planet, knowledge which is essential for informing global food security and sustainability targets. Utilizing either statistical downscaling or remote sensing approaches, existing global irrigation datasets are constrained by coarse spatial resolutions, a lack of timeliness, or varying robustness and reliability. To address this gap, here we integrate multi-source Earth observation and environmental datasets and use machine learning to develop a medium-resolution (30 m) global irrigated area dataset for the 2023/24 growing season. Within existing cropland extent, we leverage a newly compiled set of georeferenced irrigated (N=230,683) and non-irrigated (N=153,194) ground-truth points and integrate seasonal vegetation metrics derived from Landsat 8/9 imagery with agroecological-zone information and hydroclimatic and topographic variables. We subsequently develop and evaluate two machine-learning frameworks, a continental Agro-Ecological Zone (AEZ) tile-based framework and a continental-scale framework, and apply the best-performing approach for each continent. Evaluation using held-out test samples yielded a global accuracy of 80.5 ± 2.1%. The resulting maps were also validated against independent global and national irrigation datasets and statistics, demonstrating broad agreement in the spatial distribution of irrigated areas. This approach is robust and reliable because it is built on a harmonized global ground-truth database, incorporates multiple predictors, and is rigorously validated using independent datasets. All code, ground-truth, and data products are freely and publicly available and can serve as a robust, scale-neutral, and fully reproducible framework for fine-resolution irrigation mapping. These advances provide the critical and long-needed foundation for near-real-time monitoring and early warning systems, and fine-scale land and water resource management.