Testing the utility of the 3-PG model for growth of Eucalyptus grandis X urophylla with natural and manipulated supplies of water and nutrients

The productivity of fast-growing tropical plantations depends, in part, on the ability of trees to obtain and utilize site resources, and the allocation of fixed carbon (C) to wood production. Simulation models can represent these processes and interactions, but the value of these models depends on their ability to improve predictions of stand growth relative to simpler empirical approaches. We evaluated the 3-PG process-based model for simulating the response of Eucalyptus grandis x urophylla to changes in soil fertility and climate. This was done by calibrating the model with a complete C budget from an irrigated plantation, and then validating the model using independent data based on 2 years of growth from 40 pairs of fertilized and unfertilized stands. The 3-PG predictions were tested against actual production, and against a classic, empirical approach to estimating stand yield. The 3-PG parameter for site fertility was based on an objective fertilization response from the paired-plots. The 3-PG model responded well to the range of soil and climatic conditions during calibration, and was particularly sensitive to estimates of leaf area index. Actual wood production for the 40 validation stands ranged from 2 to 51 Mg ha(-1) per year, compared with model estimates of 10-42 Mg ha(-1) per year (r(2) = 0.78). Both 3-PG and the empirical model provided good estimates of wood production for average conditions, but 3-PG successfully represented the wet years and dry years that were not differentiated in the empirical model. This sensitivity of 3-PG to climate may be very useful for the prediction of wood production during short rotations, where a few years of unusual weather may strongly influence yield. Process-based models can play an important role in improving the management of these almost-agricultural forests, especially in regions with high rainfall variability. (C) 2004 Elsevier B.V. All rights reserved.