Extended Data Fig. 6: Sensitivity analysis at different points in the yield distribution (a-c) and IRR distribution (d-f). | Nature Food

Extended Data Fig. 6: Sensitivity analysis at different points in the yield distribution (a-c) and IRR distribution (d-f).

From: Profitability of climate-smart soil fertility investment varies widely across sub-Saharan Africa

Extended Data Fig. 6: Sensitivity analysis at different points in the yield distribution (a-c) and IRR distribution (d-f).The alternative text for this image may have been generated using AI.

Panels (a-c) show the predictor to which predicted yield response is most sensitive at the 5th percentile, 50th percentile, and 95th percentile of the yield distribution, respectively. Panels (d-f) show the input variable to which IRR is most sensitive at the 5th percentile, 50th percentile, and 95th percentile of the IRR distribution, respectively. Sensitivity to each input variable is defined using a local, one-at-a-time sensitivity analysis. The variable to which yield difference is most sensitive does not change significantly across its distribution, while the sensitivity of IRR to prices (specifically urea prices) increases as IRR increases. This, and the similarity in sensitivity of yield difference and IRR at low percentiles, suggest that yields are more limiting when profitability is low.

Back to article page