Quantitative classification of a historic northern Wisconsin (USA) landscape: mapping forests at regional scales

We developed a quantitative and replicable classification system to improve understanding of historical composition and structure within northern Wisconsin's forests. The classification system was based on statistical cluster analysis and two forest metrics, relative dominance (% basal area) and relative importance (mean of relative dominance and relative density), as computed from the original U.S. Public Land Survey (PLS) bearing-tree data. Broad forest patterns are consistent between the two metrics; yet, detailed inspection highlights different aspects of historical structure. Maps produced characterize vegetation at regional scales and reveal patterns that can be interpreted in the context of environmental constraints. Our classifications have a fairly coarse spatial grain (2.6 km(2)) and fine-scale, patchily distributed ecosystems types are not represented. This resolution, however, is consistent with that of the PLS bearing-tree data, and maintaining it allowed retention of other beneficial map qualities, including quantitative representation of the data, replicability, flexibility, and an assessment of robustness and confidence. Our classifications are broadly applicable for regional-scale scientific and forest-management uses, including (i) assessing natural variability, (ii) determining the potential distribution of species, (iii) setting goals for ecological restoration, and (iv) calculating landscape change.