Simulation study to determine necessary sample sizes for image analysis-based quantitative wood anatomy of vessels of beech (Fagus sylvatica)

Quantitative analysis of wood anatomy is a powerful dendroecological tool to provide new insights into environmental signals encoded in tree-rings. Nevertheless, studies with long time series or large sample sizes are not very common for diffuse-porous tree species due to the laborious sample preparation and image analysis. Finding a compromise between sample size and accuracy of sample estimators is therefore of crucial importance. In this simulation study, we present a subsampling method, which can make the quantitative analysis of wood anatomy of European beech more efficient. Based on our study material we demonstrate that for the parameters vessel size, vessel density, total vessel area and accumulated hydraulic conductivity the environmental signals are captured with high trueness (deviation < 5%) when analysing only a 1 mm wide radial stripe along the increment core. Nevertheless, when information about vessel grouping indices is required, for high trueness of the results a wider radial stripe of ∼2.2 mm needs to be analysed.