Does signal-free detrending increase chronology coherence in large tree-ring networks?

Abstract

Over the past decade, dendrochronologists have increasingly adopted the signal-free detrending (SFD) method to remove age-size trends in tree-ring measurement series, amplify the common stand-wide signal in composite chronologies, and recover medium- to low-frequency patterns that may be inadvertently removed by other detrending approaches. However, since its introduction in 2008, no systematic evaluation of the effects of SFD on tree-ring chronologies has been performed. Here we conduct the first review of SFD in dendrochronology and assess its effects when applied to large tree-ring networks. We analyzed the PAGES North America 2 K database of nearly 300 temperature-sensitive chronologies and the Missouri River database of over 350 chronologies curated for the purpose of reconstructing Missouri River streamflow. Both databases contain multiple versions of each chronology generated by different detrending methods, including those produced with (and without) the signal-free procedure applied. We evaluated (i) whether SFD increases chronology coherence at the site level by boosting the between-tree agreement, (ii) whether SFD increases coherence on a regional basis by making neighboring chronologies more similar to each other, and (iii) whether signal-free chronologies retained more medium- to low-frequency variability than their traditional counterparts. We find that, while SFD increased the strength of common signals in many instances, the effect was not universal and some sites even show a decrease in signal coherence. At regional scales, SFD increases chronology coherence in temperature-sensitive records but had no detectable effect on moisture-sensitive records. Our results demonstrate the importance of evaluating the effects of SFD prior to deploying this method for chronology development and paleoclimate reconstruction.