Biases in RCS tree ring chronologies due to sampling heights of trees

International audienceThe Regional Curve Standardization (RCS) is one of the most employed standardization methods to remove biological signals in long tree ring chronologies. The approach assumes that an overall age-related growth trend typify all tree ring series to be included in a standardized tree ring chronology. Although several potential problems of the method have been examined, the influence of varying the sampling height along tree stems has not been evaluated. Considering that age-related growth trends may vary with stem height, biases may arise when combining samples from unknown or variable sampling heights, a frequent situation with subfossil logs. In this study we perform a detailed stem analysis of 15 lakeshore black spruce (Picea mariana Mill. B.S.P.) trees in the taiga of eastern Canada to describe how the age-related growth trend varies with stem height and evaluate associated biases in RCS chronologies built from living and subfossil trees. Results show that the age-related growth trends vary markedly and systematically along stems, potentially generating large methodological biases in RCS chronologies, especially near the recent chronology end. These biases may lead to erroneous reconstructions of recent climatic trends and cause false divergence between tree ring and climate series. We have developed a correction procedure that appears efficient in removing these biases from chronologies built with the lakeshore trees and associated subfossil logs