ROXAS – an efficient and accurate tool to detect vessels in diffuse-porous species

Abstract

Wood-anatomical parameters form a valuable archive to study past limitations on tree growth and act as a link between dendrochronology and ecophysiology. Yet, analysing these parameters is a time-consuming procedure and only few long chronologies exist. To increase measurement efficiency of wood-anatomical parameters, novel tools like the automated image-analysis system ROXAS were developed. So far, ROXAS has only been applied to measure large earlywood vessels in ring-porous species.In this study, we evaluate if ROXAS is also suitable for efficient and accurate detection and measurement of vessels in diffuse-porous European beech. To do so, we compared the outcome of ROXAS with that of the established measurement programme Image-Pro Plus in terms of efficiency and accuracy.The two methods differed substantially in efficiency with automatic measurements using ROXAS being 19 times faster than with Image-Pro Plus. Although the procedures led to similar patterns in annual variation of mean vessel area and vessel density, the absolute values differed. Image-Pro Plus measured systematically lower mean vessel areas and higher vessel densities than ROXAS. This was attributed to the species-specific technical settings in ROXAS, leading to more realistic results than those obtained using the default settings in Image-Pro Plus. A shortcoming of ROXAS was, however, that small vessels (<100 μm²) could not be detected with sufficient accuracy. Yet, based on thin sections of European beech, it is generally difficult to distinguish such small vessels from parenchyma cells. Moreover, these small vessels do not contribute substantially to conductive efficiency. Therefore, we do not foresee any problems for most studies if the lower vessel area threshold to be measured is set to 100 μm². Overall, ROXAS proved to be useful for measuring vessel parameters in diffuse-porous tree species, allowing accurate and efficient analyses of large numbers of samples.