Evaluation of a vegetation simulator for roe deer habitat predictions

Abstract

In this manuscript the prediction of the primary roe deer habitat variables canopy cover, stand structural classes, proportion of evergreen trees and understory vegetation cover (classified according to palatability for roe deer and vegetation height into barren ground, graminoids, Vaccinium spp., herbs, conifers, Rubus spp. ferns deciduous and shrubs) with the growth model PROGNAUS was evaluated for a 10-year period. Canopy cover was calculated as a function of crown radius with a correction for crown overlap and also estimated using multiple linear regression. Using these two methods observed canopy cover was underestimated by 14% and 2%, respectively. Coefficients of correlation between observed and predicted canopy cover were 0.47 and 0.51. Stand structural classes were predicted within 10% of the actual proportion and 68% of all sample plots were assigned to the correct structural class. The proportion of canopy cover of evergreen species was predicted within 1% and was highly correlated ( r = 0.6) to the basal area proportion of the evergreen trees. All understory vegetation classes were predicted within 5% of the actual proportion. 4 were correctly classified into five cover classes (0%, 1–25%, 25–50%, 50–75% and 75–100%) and the proportion of each of these covers classes for each vegetation type was well predicted. A visual comparison showed that observed and predicted spatial patterns were similar. Predictions of both forest growth and understory vegetation models are reasonably accurate for roe deer habitat predictions.