Distinguishing local stresses affecting trees using internal statistics from tree ring chronologies

Traditional approaches to climate reconstruction from tree rings average over many cores in order to suppress noise, which implicitly invokes a model in which trees record signal plus noise. An alternative model, the Local Liebig Stress (LLS) model, suggests that within-site noise is dominated by differential stresses within the site who se expression is modulated by Liebig's Law. This alternate model implies that climate signals are not evenly distributed across trees in a stand at any given time, and that averaging cores across sites is not the optimal way to extract climate signals. These two models, the signal plus noise and the LLS model, make distinct predictions for the variability of internal statistics, including the mean interseries correlation (R bar ) and higher moments . In this study we use synthetic tree growth series to investigate the implications of each of these two models on internal site statistics. We simulate how these statistics vary under different climates and environments and identify new diagnostics for quantifying the relative importance of local stresses at the tree and site level , and how this stress regime changes over time in response to environmental change. Results predicts that Rbar should covary with climate if within-site noise is modulated by Liebig's Law. W e analyze a global network of tree-ring observations and find that Rbar increases during the Little Ice Age, consistent with a stronger temperature stress resulting in more frequent expression of common, as opposed to local, stresses in tree ring growth.