Sensitivity of four wheat simulation models to climate change

Abstract

Crop simulation models may be valuable in anticipating crop production under a changed climate. We compared four computer simulation models of wheat, crop estimation through resources and environment synthesis (CERES), erosion productivity impact calculator (EPIC), Stewart and Sinclair, for evaluating the impact of climate change on dryland spring wheat yield for continuous rotation in southern Alberta. To a varying extent, the four models showed decreases in dryland spring wheat yields due to high temperature and low precipitation. All the models except Stewart had similar sensitivity to low precipitation; however, they showed differences to high-moisture conditions. Within the range considered, the Sinclair model was the most sensitive to temperature, followed by CERES and Stewart. Only EPIC indicated optimum temperature and precipitation levels, while CERES had the most pronounced precipitation optimum. Although the CERES, Stewart and Sinclair models have different phenology submodels, they predicted similar phenological response to a doubled CO 2 climate scenario generated from the Canadian Climate Center General Circulation Model for Lethbridge, AB. Growing seasons shortened by 19 d were predicted using CERES and 18 d by using the Sinclair and Stewart models. The CERES, Stewart and Sinclair models were modified to include the effect of CO 2 on radiation-use-efficiency. With current atmospheric CO 2 concentration in the future climate scenario, the EPIC and Stewart models predicted significant (25%) and non-significant (3%) yield increases for dryland wheat and Sinclair and CERES predicted yield losses. Higher CO 2 levels may compensate for the effect of global warming; doubling CO 2 from present levels in a warmer climate scenario resulted in yield increase predictions at different amplitudes using EPIC, Stewart and CERES and a slight yield decrease with Sinclair. Key words: Sensitivity, climate change, dryland spring wheat