Process-based forestry models are seen by the South African forestry industry as potentially useful tools for improving predictions of growth and water use in forest plantations. The 3-PG process-based forest model was evaluated in South Africa using limited Eucalyptus and Pinus trial data, with encouraging results. Some uncertainty exists, however, over whether 3-PG will prove to be a practical forest management tool on forestry estates that typically comprise scores of compartments (planted even-aged stands) covering a wide range of site conditions, species/clone and tree age. Two important questions require answers: (i) is 3-PG capable of modelling growth over a wide range of site growth potential to useful accuracy and (ii) are there practical methods for estimating or measuring all of the required parameter values for the model? To answer these questions, 12 stands of Eucalyptus grandis× camaldulensis hybrid clones, representing early, mid and late rotation age, and covering a wide range of site growth potential in the Zululand region (KwaZulu-Natal Province), were intensively studied over a period of 12 months. Measurements of initial and final biomass, leaf area index, biomass allometric ratios, litterfall, specific leaf area, sap-flow rates, pre-dawn xylem pressure potential and weather conditions were made throughout the year. Two influential site parameter values (maximum available soil-water capacity and a soil fertility rating) required indirect estimation, since relevant information was lacking and difficult to obtain from the field. A pragmatic approach to estimating these parameters is described. Annual tree growth predicted by the model, and daily ranges of sap flow, were compared to field measurements. Predictions of annual growth increment (dry mass of stems and branches) were acceptable for 11 of the 12 stands (Y=0.85X+1.7, r2=0.84). The twelfth stand, which had an extremely high annual growth increment, high leaf area and year-long access by the trees to shallow groundwater, was substantially underpredicted. This is attributed to likely physiological adaptions to the long-term absence of water stress at this site. We conclude that 3-PG can realistically simulate growth and water use over a wide range of rotation age and growth conditions. Preliminary sets of model parameter values for a range of diverse stands may be estimated over a relatively short period using simple field equipment and techniques. Further testing of these parameter values at physiologically significant times over the remainder of the rotation is recommended to test model output and permit fine-tuning of parameters. © 2004 Elsevier B.V. All rights reserved.
