Probabilistic assessment of adaptation options from an ensemble of crop models: a case study in the Mediterranean

Uncertainty about future climate change impacts increases the complexity of assessing adaptations and risks at regional level. In modelling studies, uncertainty may arise from climate projections, field data and crop models. Approaches are required for effectively quantifying climate impacts and the effect of adaptations, managing inherent uncertainties and communicating the results. Here, we focus on assessing adaptation of wheat in a Mediterranean environment in 2030 and 2050 under the A1B scenario. A probabilistic framework for evaluating the effect of feasible adaptation strategies for winter wheat in northern Spain was applied with an ensemble of 17 crop models. First, adaptations response surfaces (ARSs) were created. These are bi-dimensional surfaces in which the effect of an adaptation option (e.g. changes in crop yield compared to the standard management) is plotted against two explanatory variables (e.g. changes in temperature and precipitation). Then, the likelihood of the effect of adaptations was calculated using ARSs and probabilistic projections (PP) of climate change. The latter are joint probabilities of changes in the same explanatory variables used for drawing the ARSs. Therefore, ARSs were constructed and climate PP superimposed. Based on ARSs, the most effective adaptations were mainly based on spring wheat, current and shorter cycle duration and early sowing date. Based on PP, these options increase yield with respect to no adaptations on average by 11% in 2030 and 15% in 2050. Also, the likelihood of maintaining current yields (i.e. standard management under current climate) is extremely likely (>95%). Other combinations of sowing dates and cycle duration were only promising when a single supplementary irrigation was applied.