View article

REGEN is an expert system that predicts future species composition at the onset of stem exclusion using preharvest stand conditions. To extend coverage into hardwood stands of the Central Appalachians, we developed REGEN knowledge bases for four site qualities (xeric, subxeric, submesic, mesic) based on relevant literature and expert opinion. Data were collected from 48 paired stands to calibrate our preliminary REGEN knowledge bases. Data from 17 additional paired stands were used to independently validate the final performance of the model. Predictions for species groups were within 4 percentage points of measured values on average, and model error was typically no more than approximately 20 percentage points for any species group. These independent validation results confirmed the potential of REGEN to predict species composition of stands regenerated using the clearcut method in the Appalachians of Virginia and West Virginia.