BAI BAI bias – An evaluation of uncertainties in calculating basal area increments from cores

Abstract

Basal area increment (BAI) is increasingly used in tree-ring based studies as it provides a direct measure of wood production and thus allows for the interpretation of growth trends. BAIs from increment cores are generally calculated while assuming circular growth patterns. However, observation of stem discs shows that many ring shapes are characterized by some form of eccentricity, not only characterized by deviations from circular ring shapes but also by piths not being centrally located. This observation poses to the question with what accuracy BAIs are calculated from increment cores. To quantify the estimation bias in BAI, we have developed a method that mimics eccentric tree growth by simulation. Various aspects of eccentricity are incorporated to created ‘stem discs’ with realistic appearances. Since BAI time series for our simulated discs are known, we can evaluate the accuracy of BAI calculation methods from cores. The ‘coring’ is simulated by taking cores at the thickest and thinnest sides of the simulated discs, whereby the number of cores is varied from one to four. In our simulations, we choose two calculation methods, namely the traditional circular approach and one that is based on the assumption of elliptical growth shapes. We find that bias in calculated BAI values is highly influenced by the number of cores taken, with a dramatic decrease from one to four cores. Furthermore, trend patterns in BAI series might be misleading in case of highly eccentric growth patterns. Based on these findings, we discuss the consequences for the interpretation of existing literature, where BAI analyses are based on one or two cores (along with the assumption of circular ring shapes). Such consequences are, however, difficult to quantify since we have no eccentricity statistics of tree growth within a forest stand. Therefore, we do not know the randomness of eccentricity within a stand, and thus to what extent chronology building (i.e. averaging BAI estimates over multiple trees) may reduce estimation bias. To lower BAI bias, we recommend to base BAI calculations on as many cores as possible. For individual trees with high levels of eccentricity, taking four increment cores seems necessary to reasonably estimate their basal area increments.