Does Choice of Response Function Matter in Setting Maximum Allowable N-Application Rates in Danish Agriculture?

Abstract

Response research during the last couple of decades tends to support agronomic knowledge pointing to decreasing marginal yield response up to a point where additional nitrogen does not result in further yield increase but rather a yield plateau. In contrast to a non-linear response with plateau model (NLRP) a cubic polynomial specification is generally used by those responsible for recommending/setting nitrogen norms in compliance with Danish environmental policy. The flexibility of the cubic polynomial ensures a relatively good fit to observed nitrogen/yield data and, surprisingly, often renders a close emulation of the NLRP model. Unfortunately, the cubic also often produces unrealistic optimal nitrogen application estimates. In those cases, Danish authorities use a simple quadratic specification as a fallback. We argue that an NLRP specification would be a better first choice model than the cubic – among other things, avoiding the need for a fallback model. Further progress in response research and choice of proper model specification could possibly be achieved by accounting for total nitrogen supply by including nitrogen provided in the form of mineralized nitrogen from soil sources.