Recent research published in Local environmental changes In August 2025, we utilized a machine learning-driven approach to estimate future changes in livestock capacity across East Africa and to diagnose underlying climate drivers.
This paper addresses the sustained knowledge gaps of how climate change affects grazing land productivity and livestock capacity at a local scale. Traditional methods involving localized investigations and process-based models struggle with spatial variability and inadequate calibration in the data scarce domain.
Instead, researchers led by Duku and colleagues develop a machine learning framework that integrates satellite-based biomass data (via remote sensing), integrates climate forecasts to model net primary productivity (NPP), and translates it into livestock transport capabilities (measured in tropical livestock units).
The expected outcomes will vary widely depending on the country and production system.
- in Ethiopiahistorically, the Mixed Crop Livestock Rainfate Temperature System (MRT), which supports about half of the country's cattle and sheep, represents a disastrous decline. Projections reduce capacity by up to 37%, especially towards the end of the century, especially under the emissions SSP5-8.5 scenario. Already, current livestock numbers are near the limits of the MRT system, so declining capacity risks serious livestock sustainability issues.
- in Kenyatrends vary by system:
- In idyllic/dry LGA systems, capacity may increase by 11-26%.
- Conversely, mixing temperature (MRT) and livestock-only temperate (LGT) systems are expected to decrease by 12-24% and 10-31%, respectively.
- Tanzania Shows a relatively minor shift: a slight increase (approximately 12% may decrease) in most systems (up to 6%) except for LGT.
- in Ugandaoverall profits are predicted, but MRT systems could decrease by 15-32%, while mixed stormwater wetting systems (MRHs) could increase by 7-12%.
What drives these shifts?
This study uses explanability analysis to dig into the driver. On Ethiopia's MRT, the worst quarterly increase in precipitation, reduced temperature seasonality, and the driest quarterly increase in temperature are the major contributors to the decline in NPP and thus decrease in the carrier capacity.
In the Kenya LGA system, the increase is primarily due to increased precipitation over cold and wet months, but rising dry season temperatures can mitigate profits.
CGIAR Communications reflects these findingshighlighting particularly sharply predicted declines. This amounts to 37% in Ethiopia's mixed crop livestock system and 24% in Kenya. They emphasize the urgency of customized adaptive responses.
The impact on policy is profound:
- Ethiopia Surveillance systems must be urgently strengthened, invested in local adaptation strategies (including presumably breed shifts or resilient feed) and burdened with the safe livelihood of farm keepers.
- Kenya and Uganda New opportunities should be seized by promoting sustainable livestock systems that maximize productivity without compromising environmental health.
This new machine learning approach not only provides a robust and scalable projection of livestock capacity, but also illuminates the complex climate drivers behind them.
