Evaluating SORTIE simulations

Evaluating SORTIE

The emergence of the large-scale dynamics as a consequence of individual interactions is a vital but poorly understood aspect of complex ecosystem dynamics. We have presented a series of simulations that predict forests with and without disturbance, with increasing mortality, and after different clear-cut treatments. Several consistent features of the dynamic development of the forest emerge from these analyses. Estimates of total biomass are consistent with observed values from old-growth forest in the northeastern United States, even though no data from old-growth stands were used in model parameterization (Pacala et al. 1993; Pacala et al 1996). In the absence of disturbance, the young forest is initially dominated by fast-growing, light-demanding trees (often black cherry, red oak, and white pine); forest maturation is characterized by a shift away from these species to shade-tolerant species capable of growing in low light (hemlock and particularly beech). Eventually, only beech and hemlock can recruit under the closed canopy, eliminating all other species. This prediction is very sensitive to the patterns of disturbance on the landscape.

Base line
no disturbance

Base line
with disturbance

Partial
clear-cut

High
mortality

Views of the model under diverse conditions.

Simulations with frequent multitree disturbances favor yellow birch because it can quickly colonize newly created gaps. Yellow birch has the longest dispersal and fastest growth in moderate light and eventually outcompetes beech and hemlock under these conditions. Increasing individual-tree mortality is not equivalent to the multitree disturbance. Slight increases in mortality shift the dominance from beech to hemlock, a shade-tolerant species with slightly higher growth rate. Higher mortality essentially traps the forest in a perpetual pioneer state dominated by the fastest growing, light-demanding trees: black cherry, red oak, and white pine.

In order to understand range of prediction and evaluate the model, we introduce a series of model experiments. First, the uncertainty associated with parameter estimation is propagated through the model. These models repeatedly produce plausible forests with early mixed-species assemblages giving rise to a community dominated by one or more shade-tolerant species. The identities of these competitive dominant species are not resolvable given the uncertainty in the parameters. The clear-cut experiments show that the several key species are dispersal-limited. Survival of seedlings (advanced regeneration) has pronounced effect on the model, leading to a strong founder effect. Increasing mortality of individual trees is not the same as increased disturbance. As mortality increases, the community shifts from shade-tolerant species to fast-growing pioneer species.

These prediction are biologically plausible and consistent with our broad understanding of forest dynamics. In developing this model, there was no attempt to match the large-scale dynamics with similarly large-scale forest data. Instead, SORTIE was conceptualized at the individual scale and parameterized from field data. This model, assembled at this detailed level, is truly predictive of the large-scale forest behavior under a diverse set of conditions. Building and testing data-defined, detailed models represents a powerful tool for investigating complex ecological phenomena (Huston et al. 1988; DeAngelis and Gross 1992; Uchmanski and Grimm 1996; Bolker et al. 1997; Levin et al. 1997).