New tools for timber provenancing

Resumen

Trees are affected by a set of spatially and/or temporarily varying biotic and abiotic factors. The effect of these factors is reflected in the width and structure of tree rings, which are widely used to study species ecology and their responses to climate. Moreover, tree-ring width time series are also used for dating and identification of the geographical origin of archaeological or historical timber – dendroprovenancing. The latter is usually done by statistical matching of tree-ring width sequence of a given timber to a network of regional and local chronologies from the same species or genera. However, despite multiple successful applications, this method also has clear pitfalls, such as lack of long reference tree-ring width chronologies from areas with intensive logging history; or possible strong teleconnections between reference chronologies over large distances leading to coarse-scale results. Therefore, to overcome these limitations and to improve dendroprovenancing results, other wood-anatomical features or chemical characteristics have to be used. The potential of additional variables retrieved from the wood has been recently tested and showed promising results for timber tracking, e.g. its chemical composition or DNA. However, until now time series of xylem-anatomical features have been largely ignored in timber-provenancing studies. This is most likely due to the time-consuming and labour-intensive data acquisition and challenging preparation of archaeological timber for precise wood-anatomical measurements. But with recent improvements in wood- preparation techniques and image-analysis software, it became possible to acquire a sufficient amount of data derived from various xylem-anatomical features in an efficient way. This thesis aims to evaluate the value of xylem-anatomical features as well as of archaeological DNA (aDNA) in addition to TRW for timber-provenancing studies. In the context of the ForSEAdiscovery project, the potential of vessel size of ring-porous oak, blue intensity variables derived from pine trees, and aDNA extracted from historical oak timber to enhance dendroprovenancing precision was assessed. Moreover, a conceptually new method, i.e. based on individual time-series rather than on average chronologies, was applied. This was done by means of Principal Component Gradient Analyses (PCGA). A description is provided for each of the chapters on their research and findings below. To assess the potential of oak earlywood vessels for timber provenancing, samples from nine oak stands in Northern Spain (Cantabria and the Basque country) were collected (Chapter 2). From this set of samples, vessel size and latewood-width time series were created. Based on variation in latewood-width time series, it was possible to differentiate between Cantabrian and Basque forest stands, i.e. in East-West direction. The difference in response to the summer temperature was found to be the main factor leading to such a differentiation. At the same time, variation in vessels size enabled the grouping of trees from the Basque country according to the continentality gradient, i.e. North-South direction, based on a difference in response of vessel size to winter and spring temperatures. These results suggest that the approach of combining latewood width with vessel size leads to a higher precision dendroprovenancing and pinpointing the origin of oak timber on a finer scale. Leave-one-out analyses confirmed this conclusion. In Chapter 3, the potential of wood-density related Blue Intensity (BI) variables in addition to TRW for dendroprovenancing pine timber in drought limited areas was evaluated. The network of the BI and TRW time series was created from six pine forests from Central Spain and Southern Spain. PCGA of the derived time series revealed a grouping of trees according to their elevation category based on the BI time series. However, it failed to group trees according to their geographical provenance both for the TRW and BI series. Trees were correctly assigned to their origin based on TRW, but only within the predefined by the BI elevation groups. Based on these results, it is possible to conclude that a multi-variable approach comprising of BI and TRW assists in enhancing pine-timber dendroprovenancing in dry areas. In Chapter 4, the potential of extracting and using aDNA for oak-timber provenancing on different scales was evaluated. Thirty samples from historical buildings from Spain, Latvia and Denmark were analysed. Two different extraction protocols in two genetic laboratories were tested. Furthermore, two different haplotype identification methods were applied. From 60% of the samples, at least one marker showed the presence of the aDNA with a varying percentage per extraction protocol. An existing haplotype distribution map was used to identify the potential source area of material from two study cases. The results suggest that genetic analyses have a strong potential for pinpointing timber origin, though until now only in combination with the TRW based method. Improvements in DNA extraction from degraded wood and amplification protocols are essential for future applications in dendroprovenancing studies. In Chapter 5, the main outcomes of each of the core chapters are discussed and contextualized in a broader perspective of dendroprovenancing. The potential of oak earlywood vessels in solving actual timber provenancing challenges in other regions in Europe is discussed. The added value of using the individual tree approach was tested and its strong potential in improving timber provenancing is confirmed. Based on the results derived from the core chapters, and from other studies conducted in the framework of the ForSEAdiscovery project on the same samples, a decision tree was created in order to facilitate oak and pine timber provenancing in the region. In conclusion, the multivariable approach has demonstrated a strong potential to enhance the precision of timber provenancing. However, for a successful application, specific climatic gradients were required for both genera (Chapters 2 & 3). Therefore, the decision on a suitable approach should be based on the specific provenancing conditions and should be comprised of multiple variables.