Global terrestrial productivity in the Mesozoic era

Abstract This paper describes a global-scale modelling approach for investigating the structure and productivity of terrestrial vegetation in the Mesozoic era. General circulation model (GCM) palaeoclimate simulations (by the University of Reading) for Kimmeridgian and Cenomanian time are described as driving datasets for the University of Sheffield Dynamic Global Vegetation Model (SDGVM). The validity of these GCM palaeoclimates is reviewed. Global patterns of terrestrial net primary productivity (NPP) and leaf area index (LAI) have been simulated with the SDGVM, which models plant physiological processes and the biogeochemical cycling of carbon and nitrogen in soils and vegetation. An important feature of this modelling approach is that it requires no underlying map of soils or vegetation type. Global NPP in Kimmeridgian and Cenomanian time was high (117.6 and 106.8 Gt Ca −1 , respectively) compared with the present-day level (57.0 Gt Ca −1 ). The high concentrations of atmospheric CO 2 at each interval significantly influenced the NPP and LAI of Mesozoic vegetation, to an extent dependent on climate. CO 2 effects on the structure of vegetation in Kimmeridgian and Cenomanian time were sufficient to markedly influence key land surface variables required by GCMs for climate predictions, and point to the need to include direct CO 2 -vegetation interactions in climate models. Two approaches to testing the modelled NPP patterns were devised, one using palaeobotanical information on tree growth, and the other based on a comparison of measurements and predictions of the stable carbon isotope ratios of fossil plants. Both approaches provided some support for the global-scale simulations, indicating the feasibility of modelling vegetation activity in ancient climates from a knowledge of present-day processes.