Importance of sampling design to increase climate signal detection in shrub ring chronologies

Shrub dendrochronology is gaining increasing momentum in temperate high mountain regions to decipher climatic controls on current shrub expansion. Yet, a lack of consensus still persists in terms of sampling protocols, thus hampering comparability of results from different studies. For instance, serial sectioning, i.e. the sampling of multiple sections along the same shrub stem is recommended as it increases the detection of partial and missing rings, but has only been employed in few studies as it is time-consuming. Similarly, as a result of serial sectioning, chronologies frequently combine sections sampled at different positions along the stem and at the root collar which hinders the detection of climatic signals. Here, we used cross-sections sampled on 21 Rhododendron ferrugineum shrubs from the French Pyrenees to define a parsimonious protocol enabling detection of partial and missing rings while increasing the strength of the climate signal in the shrub ring chronology. We demonstrate that partial and missing rings are almost evenly distributed along Rhododendron ferrugineum stems and that they can be detected optimally using two sections on which growth rings are measured along three radii. Our results also evidence that chronologies which include only ring-width series from basal sections more strongly integrate summer temperature fluctuations than stem-based or mixed chronologies. Noteworthy, the snowpack signal is stronger in chronologies with individuals from the upper stem sections. Overall, our results confirm that sampling design - serial sectioning and caution in ring-width series aggregation - is key to ensure robustness of dendroecological studies on dwarf shrubs in alpine environments.