Application of sensitivity analysis to a model of Eucalyptus globulus plantation productivity

A simple forest growth model, ProMod, was developed to assess productivity by plantation grown Eucalyptus globulus. It is based on a sound understanding of the basic physiology of tree growth and predicts leaf area index (LAI) following canopy closure, annual net photosynthetic production and water use, soil water balance, and stemwood production, all in response to climatic and site factors. Minimum inputs are site latitude, monthly means of daily maximum and minimum temperatures, solar irradiance, rainfall and open-pan evaporation, mean monthly rain-days, and a simple classification of soil depth, texture, stoniness, drainage and fertility. Output from ProMod is affected by processes included as part of the model structure, physiological parameters characterising these processes, and variables characterising site climate and soil. A sensitivity analysis based on predicted net total annual production (Ga) and annual water use efficiency (ωα) was applied to the structure of ProMod and to its parameter values. The structural sensitivity analysis shows that, with the possible exception of the temperature dependence of respiration, simplification of any of the processes in ProMod results in a loss of generality. The parameter sensitivity analysis shows that Ga and ωα are highly sensitive to a small subset of parameters: light saturated photosynthetic rate, shape of the light response curve and its dependence on temperature, low temperature response of LAI, and parameters in the relationship between water use efficiency and vapour pressure deficit. There is little or no sensitivity to some parameters which are difficult to measure. In many cases sensitivity varied significantly from site to site. This information is important when ProMod is to be parameterised for different species, and impacts on traits to be selected for in a tree-breeding program.