Detecting Change in Forest Spatial Patterns From Reference Conditions

Timber harvest, fire suppression, road construction, and domestic livestock grazing have transformed spatial patterns of Interior Northwest forests. As a consequence, parameters of current disturbance regimes differ radically from historical regimes; present-day wildlife habitat distributions differ from historical distributions; and long-term survival of some native terrestrial species is uncertain. Public land managers are under increasing scientific and social pressure to mold existing forest spatial patterns to reflect those resulting from natural disturbance regimes and patterns of biophysical environments. However, knowledge of the characteristics of natural spatial patterns is unavailable. Using a dichotomized ordination procedure, we grouped the 343 forested subwatersheds (mean area, 8000 ha) on the eastern slope of the Cascade Mountains in Washington State into ecological subregions by similarity of area in potential vegetation and climate attributes. We built spatially continuous “historical” (1938–1956) and “current” (1985–1993) vegetation maps for 48 randomly selected subwatersheds from aerial photo interpretations. From remotely sensed attributes, we classified cover types, structural classes, and potential vegetation types and attributed them to individual patches. We then estimated a reference variation (RV) in spatial patterns of patch types (cover type and structural class), by subwatersheds and five forested ecological subregions, using the 48 historical vegetation maps stratified by subregion and a spatial pattern analysis program. Finally, we compared the current pattern of an example subwatershed (MET_11) with the RV estimates of its corresponding subregion to illustrate how reference conditions can be used to evaluate the importance of spatial pattern change. By evaluating pattern changes in light of RV estimates (nominally, the sample median 80% range of a metric) and the full range of class and landscape metrics, we could identify both current and historical conditions of MET_11 that fell outside the RV. This approach gives land managers a tool to compare characteristics of present-day managed landscapes with reference conditions to reveal significant pattern departures, as well as to identify specific pattern characteristics that might be modified through management. It also provides a means to identify “outlier” conditions, relative to subregion RV estimates, that may occasionally be the object of pattern restoration activities.