Predicting pasture yield response to nitrogenous fertiliser in Australia using a meta-analysis-derived model, with field validation

Abstract

An improved ability to predict pasture dry matter (DM) yield response to applied nitrogen (N) is a crucial step in determining the production and economic benefits of N fertiliser inputs with associated environmental benefits from reducing inefficient N fertiliser use. Pasture DM yield responses to applied N fertiliser from 920 independent field trial sites were used from a database repository of Australian fertiliser experiments. These data were analysed and a quantitative non-linear mixed-effects model based on the Mitscherlich function was developed. The fitted model provided a good fit to a large body of data (R2 = 0.92), using readily interpretable coefficients, including fixed effects for state by season, phosphorus status and harvest type (initial or residual), and nested random effects for location and trial or subtrial. The model was limited by patchiness of metadata, uneven representation of regions and few very high rates of applied N in the data. Nonetheless, model predictions were comparable with independent spring pasture DM responses to applied N fertiliser from subsequent field studies on three contrasting pastures on commercial dairy farms in Victoria. The final derived model can be used to predict pasture yield response to applied N fertiliser as a proportion of obtainable yield and can be scaled to absolute response using the fitted model estimates of maximal yield or, more usefully, a specified maximal yield by the user. Importantly, the response function exhibits diminishing returns, enabling marginal economic analysis and determination of optimum N fertiliser application rate to a specified pasture.