Predicting plant available water holding capacity of soils from crop yield.

Aims: Quantification of variations in plant available water holding capacity (PAWC) of soils helps to improve yield forecast and inform spatially variable management practices in dryland agriculture systems. We developed and tested a general inverse approach to estimate PAWC from crop yield. Methods: The APSIM model was used to simulate wheat yield on synthetic soils with contrasting PAWC and climates. The simulated results were used to develop an empirical model to relate simulated yield to PAWC. The empirical model was inversely used to predict PAWC from observed crop yield. Potential prediction ability was quantified using independently simulated wheat yield on actual soils. The actual ability was assessed with measured wheat yields and PAWC. Results: The approach had higher accuracy for sites with high rainfall or dominant summer rainfall. It could potentially provide acceptable PAWC predictions across contrasting climate regions (prediction error < 37 mm, 33.5%). The prediction error using crop yield against measured PAWC was <25 mm (26.5%). Conclusions: Our results demonstrate that soil PAWC can be reliably predicted from crop yield. This approach provides an alternative way to predict PAWC rather than directly measuring it via soil sampling, with profound implications for reducing labour and costs. [ABSTRACT FROM AUTHOR]