Detecting global trends of cereal yield stability by adjusting the coefficient of variation

High stability of crop yields is a key objective in crop production and breeding, especially under the conditions of a changing climate. Reliable indices are therefore needed for quantifying yield stability. Recently it was shown that some frequently used indices of yield stability, such as the coefficient of variation (CV) may be wrongly interpreted if there is a systematic dependence of the variance σ² from the mean yield μ following Taylor's power law. Here we propose a method to adjust the standard CV to account for the systematic dependence of σ² from μ. This adjusted CV can be used as a stability index that is expressed in units that are equivalent to the standard CV, as a percentage of the mean, and can therefore be used in agronomic studies that aim to provide guidance for farmers and advisors. Applying this adjusted CV (called aCV) to FAO cereal yield data, we show that the temporal yield stability of both wheat and rye has weakly but significantly decreased over the last five decades and this trend was not picked up with the standard CV in wheat, and was more marked with the aCV than with the standard CV in rye. For the intensifying research on yield stability in agronomy, the suggested method is a novel alternative to estimate yield stability more conclusively, allowing straight-forward interpretation and providing the basis for developing cropping systems with higher yield stability in the future.