The response of crop growth and yield to CO2 and ozone is known to depend on climatic conditions and is difficult to quantify due to the complexity of the processes involved. Two modified mechanistic crop simulation models (AFRCWHEAT2-O3 and LINTULCC), which differ in the levels of mechanistic detail, were used to simulate the effects of CO2 (ambient, ambient x2) and ozone (ambient, ambient x1.5) on growth and developmental processes of spring wheat in response to climatic conditions. Simulations were analysed using data from the ESPACE-wheat project in which spring wheat cv. Minaret was grown in open-top chambers at nine sites throughout Europe and for up to 3 years at each site. Both models closely predicted phenological development and the average measured biomass at maturity. However, intermediate growth variables such as biomass and leaf area index (LAI) at anthesis, seasonal accumulated photosynthetically active radiation intercepted by the crop (Sigma IPAR), the average seasonal light use efficiency (LUE) and the light saturated rate of flag leaf photosynthesis (A(sat)) were predicted differently and less accurately by the two models. The effect of CO2 on the final biomass was underestimated by AFRCWHEAT2-O3 due to its poor simulation of the effect of CO2 on tillering, and LAI.LINTULCC overestimated the response of biomass production to changes in CO2 level due to an overprediction of the effect of CO2 on LUE. The measured effect of ozone exposure on final biomass was predicted closely by the two models. The models also simulated the observed interactive effect of CO2 and ozone on biomass production. However, the effects of ozone on LAI, Sigma IPAR and A(sat) were simulated differently by the models and less accurately with LINTULCC for the ozone effects on LAI and Sigma IPAR. Predictions of the variation between sites and years of growth and development parameters and of their responses to CO2 and ozone were poor for both AFRCWHEAT2-O3 and LINTULCC. It was concluded that other factors than those considered in the models such as chamber design and soil properties may have affected the growth and development of cv. Minaret. An analysis of the relationships between growth parameters calculated from the simulations supported this conclusion. In order to apply models for global change impact assessment studies, the difficulties in simulating biomass production in response to CO2 need to be considered. We suggest that the simulation of leaf area dynamics deserves particular attention in this regard. (C) 1999 Elsevier Science B.V. All rights reserved.
