Enhanced dendroprovenancing through high-resolution soil- and climate data

Instruments aiming to avoid illegal logging such as certification chains require data-driven solutions to verify timber origin. One approach to timber tracing is dendroprovenancing, which uses the spatial and temporal consistency of tree ring width patterns to match unknown samples to reference samples from known locations. Best matching reference samples indicate the potential source location of the unknown sample. Gaps in temporal and spatial coverage of reference chronologies however currently limit applicability of dendroprovenancing, with additional data acquisition being both time-consuming and expensive. This study presents a novel general dendroprovenancing framework, aiming to overcome this shortcoming. It relies on modelling and spatially exhaustive prediction of reference chronologies using a regression model and gridded high-resolution soil- and climate data with global coverage. The presented framework is explored through a case study on Quercus robur using 107 tree-ring chronologies from western and central Europe. We tested three scenarios using leave one out cross-validation: 1) the dating of the chronology is unknown, 2) the source location of the chronology is unknown, and 3) both the dating and source location of the chronology are unknown, with the latter most closely resembling a real-world scenario. We found that tracing accuracy was high, even in the scenario in which both the dating and source location of the chronology were unknown. 82.2% of the chronologies were traced to within a radius of 250 kilometres from the ground truth and correctly dated. The findings highlight newfound potential of dendroprovenancing for timber tracing.