The complex mix of economic and ecological objectives facing today's forest managers necessitates the development of growth models with a capacity for simulating a wide range of forest conditions while producing outputs useful for economic analyses. We calibrated the gap model ZELIG to simulate stand-level forest development in the Oregon Coast Range as part of a landscape-scale assessment of different forest management strategies. Our goal was to incorporate the predictive ability of an empirical model with the flexibility of a forest succession model. We emphasized the development of commercial-aged stands of Douglas-fir, the dominant tree species in the study area and primary source of timber. In addition, we judged that the ecological approach of ZELIG would be robust to the variety of other forest conditions and practices encountered in the Coast Range, including mixed-species stands, small-scale gap formation, innovative silvicultural methods, and reserve areas where forests grow unmanaged for long periods of time. We parameterized the model to distinguish forest development among two ecoregions, three forest types and two site productivity classes using three data sources: chronosequences of forest inventory data, long-term research data, and simulations from an empirical growth-and-yield model. The calibrated model was tested with independent, long-term measurements from 11 Douglas-fir plots (6 unthinned, 5 thinned), 3 spruce-hemlock plots, and 1 red alder plot. ZELIG closely approximated developmental trajectories of basal area and large trees in the Douglas-fir plots. Differences between simulated and observed conifer basal area for these plots ranged from -2.6 to 2.4m²/ha; differences in the number of trees/ha >=50cmdbh ranged from -8.8 to 7.3tph. Achieving these results required the use of a diameter-growth multiplier, suggesting some underlying constraints on tree growth such as the temperature response function. ZELIG also tended to overestimate regeneration of shade-tolerant trees and underestimate total tree density (i.e., higher rates of tree mortality). However, comparisons with the chronosequences of forest inventory data indicated that the simulated data are within the range of variability observed in the Coast Range. Further exploration and improvement of ZELIG is warranted in three key areas: (1) modeling rapid rates of conifer tree growth without the need for a diameter-growth multiplier; (2) understanding and remedying rates of tree mortality that were higher than those observed in the independent data; and (3) improving the tree regeneration module to account for competition with understory vegetation.
