Stand-level sampling designs for bark stripping caused by red deer (Cervus elaphus L.): simulation studies based on nine fully censused stands

Abstract

Precise assessment of bark stripping damage is of high economic importance, since bark stripping makes wood unusable for saw timber and it is important for compensation payments for game damage. Bark stripping is clustered and decreases with increasing tree diameter, so that common forest inventories, optimized for assessing timber production variables such as standing timber volume, do not provide adequately precise estimates of bark stripping damage. In this study we analysed different sampling designs (random sampling, systematic sampling), tree selection methods (fixed radius plot, angle count sampling) and number of plots and plot sizes (plot radius: 2–20 m; basal area factor: 1–6m2/ha) for bark stripping assessment. The analysis is based on simulation studies in 9 fully censused stands (9026 trees). Simulations were done for actually assessed damage and randomly distributed damage and each scenario was repeated 100 times with different random points or different random grid locations. Systematic sampling was considerably more precise than random sampling in both scenarios. Sampling intensities to attain a standard error of 10% ranged between 12 and 18% dependent on the plot size. For a given sampling intensity, precision increased with decreasing plot size or increasing basal area factor. This implies, however, a large number of plots to be measured, which is expensive, when travel costs are high. Differences between tree selection by fixed radius plots or angle count sampling were minor. For bark stripping damage, we recommend sampling with fixed radius plots with a radius of 4–6 m and the measurement of approximately 230 or 150 plots, respectively.