Notes towards an optimal sampling strategy in dendroclimatology

Though the extraction of increment cores is common practice in tree-ring research, there is no standard for the number of samples per tree, or trees per site needed to accurately describe the common growth pattern of a discrete population of trees over space and time. Tree-ring chronologies composed of living, subfossil and archaeological material often combine an uneven distribution of increment cores and disc samples. The effects of taking one or two cores per tree, or even the inclusion of multiple radii measurements from entire discs, on chronology development and quality remain unreported. Here, we present four new larch (Larix cajanderi Mayr) ring width chronologies from the same 20 trees in northeastern Siberia that have been independently developed using different combinations of core and disc samples. Our experiment reveals: i) sawing is much faster than coring, with the latter not always hitting the pith; ii) the disc-based chronology contains fewer locally absent rings, extends further back in time and exhibits more growth coherency; iii) although the sampling design has little impact on the overall chronology behaviour, lower frequency information is more robustly obtained from the disc measurements that also tend to reflect a slightly stronger temperature signal. In quantifying the influence of sampling strategy on the quality of tree-ring width chronologies, and their suitability for climate reconstructions, this study provides useful insights for optimizing fieldwork campaigns, as well as for developing composite chronologies from different wood sources.