Sensitivity of wheat yield and soil nitrogen (N) losses to stepwise changes in means and variances of climatic variables were determined using the FASSET model. The LARS-WG was used to generate climate scenarios using observed climate data (1961–90) from two sites in Denmark, which differed in climate and soil conditions. Scenarios involved changes to (i) mean temperature alone, (ii) mean and variability of temperature, (iii) winter and summer precipitation amounts and (iv) duration of dry and wet series.

The model predicted lower grain yield and N uptake in response to increases in mean temperatures, caused by early maturity, with little change in variability. This, however, increased soil mineral N causing increased N losses. On sandy loam, larger temperature variability lowered grain yields and increased N losses coupled with higher variability at all the mean temperature ranges. On coarse sand, grain yields either remained unaltered or were slightly reduced when larger temperature variability was introduced to increase in mean temperatures of up to +2°C above baseline. However, introducing variability to further increase in mean temperatures lowered yields without any change in variability. Larger temperature variability did not affect soil mineral N and N2O emissions, but increased N leaching on coarse sand.

Large response in grain yield, N uptake and soil N cycling, and in their variability was predicted when summer precipitation was varied, whereas only N leaching responded to changes in winter precipitation. Doubling the duration of dry series lowered grain yield and N removed by grain, but increased N leaching, whereas doubling the duration of wet series showed opposite effect. Predicted responses to changes in precipitation patterns were larger on coarse sand than on sandy loam. This study illustrates the importance of considering effects of changes to mean climatic factors, climatic variability and soil types on both crop yield and soil N losses.