SENSITIVITY ANALYSIS OF A FOREST SUCCESSION MODEL

Gap models can simulate forest structure, dynamics and succession. A sensitivity analysis is presented for a gap model, FORSKA. FORSKA was developed to simulate boreal forests and uses an explicit vertical canopy structure that allows self-pruning in response to shading of the lower crowns. A simulation of 190 years' succession in a old-growth stand dominated by pine (Pinus sylvestris L.) and spruce (Picea abies (L.) Karst.) was used as a basis for the analysis. These species differ in their shade-tolerance characteristics, leaf-area sapwood-area ratio and diameter/height relation. Response curves of the sensitivity of each species to all model parameters were determined by varying the standard values up and down by 10% and 50%. Sensitivity indices were estimated by regression. The most critical parameters proved to be those directly affecting growth rates, light availability within the canopy and the species' response to shading through deterministic algorithms. Other parameters, including many that are difficult to estimate precisely, had relatively little effect. Although the stochastic processes of recruitment and mortality are qualitatively important to the outcome of the simulation, the low sensitivity to the parameters of these processes suggest that the system is strongly buffered against natural variability in these parameters.