Forest processes from stands to landscapes: exploring model forecast uncertainties using cross-scale model comparison

Forest management practices conducted primarily at the stand scale result in simplified forests with regeneration problems and low structural and biological diversity. Landscape models have been used to help design management strategies to address these problems. However, there remains a great deal of uncertainty that the actual management practices result in the desired sustainable landscape structure. To investigate our ability to meet sustainable forest management goals across scales, we assessed how two models of forest dynamics, a scaled-up individual-tree model and a landscape model, simulate forest dynamics under three types of harvesting regimes: clearcut, gap, and uniform thinning. Althougth 50-100 year forecasts predicted average successional patterns that differed by less than 20% between models, understory dynamics of the landscape model were simplified relative to the scaled-up tree model, whereas successional patterns of the scaled-up tree model deviated from empirical studies on the driest and wettest landtypes. The scale dependencies of both models revealed important weaknesses when the models were used alone; however, when used together, they could provide a heuristic method that could improve our ability to design sustainable forest management practices.