A hierarchical approach for simulating northern forest dynamics
Complexity in ecological systems has challenged forest simulation modelers for years, resulting in a number of approaches with varying degrees of success. Arguments in favor of hierarchical modeling are made, especially for considering a complex environmental issue like widespread eastern hemlock regeneration failure. We present the philosophy and basic framework for the NORTHern Woodland Dynamics Simulator (NORTHWDS) Integrated Hierarchical Model System (NIHMS). NIHMS has an individual tree component (the NORTHWDS Individual Response Model (NIRM)), a mesoscale stand simulator (NORTHWDS), and a landscape model (NORTHWDS Landscape Model (NLM, presented in another paper)). NIRM predicts the behavior of a tree given the physical and biotic environment that constrains its performance, using process-response functions at a scale larger than the individual plant. The NORTHWDS model integrates both the structure of the individual tree model (including tree growth and mortality functions) with a series of ecosystem processes (e.g. competition, site biogeochemistry, small-scale disturbance, deer browsing) and even larger scale events (e.g. catastrophic windthrow) to predict long-term stand dynamics on a 5-year simulation cycle. The boundaries in time and space between the NIRM, NORTHWDS, and NLM models are not discrete, but overlap due to the multiscale expression of ecological and physiological processes. For example, the NORTHWDS model represents both the intersection between the NIRM and NLM models with additional unique mesoscale processes (e.g. intertree competition). At the highest level of NIHMS, NLM provides the environmental context for NORTHWDS, with all levels operating in an internally consistent and parsimonious manner.
Three case study scenarios are used to illustrate some of the potential applications of NIRM. Scenario 1: simulation of northern red oak survivorship, crown dynamics, diameter increment, and cumulative propagule production under different local stand densities; Scenario 2: the response of white ash along an available nitrogen gradient with respect to mortality, crown surface area, diameter growth, tree biomass, and propagule production; and Scenario 3: the survivorship and total propagule production of black spruce along a soil moisture gradient. Under Scenario 1, crown size decreased appreciably as local stand density increased from open-grown (<1m2 /ha) to closed canopy conditions 15-30 m2/ha, triggering reduced annual increment, lower established propagule production, and increased small tree mortality. Similarly, nitrogen (for white ash) and moisture (for black spruce) gradients significantly affected crown size and growth potential, with many of the same consequences as noted for increased competition in northern red oak. These predictions were consistent with ecological expectations for all scenarios except the black spruce moisture gradient response, which arose because of scale-related issues and the complexity of gradients.
To evaluate NORTHWDS model behavior, a 36 ha synthetic stand resembling a hardwood-dominated forest from northern Wisconsin was simulated for 300 years using four windthrow disturbance scenarios (no windthrow, acute windthrow only, chronic windthrow only, and both types of windthrow). Acute (particularly severe but locally discrete and infrequent events) windthrow patterns were designed to generate many small, low intensity events, while chronic windthrow (pervasive yet low intensitycyclic loss of trees) losses depended on species, stature, rooting depth, and tree exposure. These scenarios were compared by examining differences in structural (i.e. biomass, tree richness) and compositional attributes for a number of key species. NORTHWDS predicted the maturation of a pole-sized aspen stand and its eventual conversion to a predominantly sugar maple forest. Quantitatively, aboveground biomass levels compa