Exploring potential drivers of divergence in tree-ring based temperature reconstructions of NW North America

Non-stationary growth responses have been identified in tree-ring width (TRW) and maximum latewood density (MXD) chronologies of north-west North America. Here, we present MXD and latewood blue intensity (LWBI) data from two areas of the Yukon Territory (YT) to explore divergent climate-growth relationships until 2021 CE and evaluate the underlying reasons considering different detrending methods and instrumental datasets. We examine divergent long-term trends and changing inter-annual signals using well-replicated chronologies integrating a mixture of young and mature trees. Both tree-ring parameters correlate significantly (p < 0.05) with May–August temperatures, but the MXD results are stronger and show less divergence in trend. Variability among differently detrended MXD chronologies is smaller and a signal-free version of age-dependent spline detrending appears to be optimal for both YT sites. Comparison of instrumental data products reveals that the highest and most stable correlations are achieved using the Berkeley Earth dataset. Additionally, using different sub-diurnal temperatures affects both trend and correlation divergence with maximum temperature consistently showing the strongest and minimum temperature the weakest results. We conclude that regional divergence in the YT is characterized by trend rather than high-frequency issues and is larger in LWBI than MXD data. Altering detrending methods and diurnal temperatures is of greater importance than varying instrumental data products. Most stationary responses are recorded when applying signal-free age-dependent spline detrending to tree-ring data and targeting Berkeley Earth maximum temperatures. Disregarding these methodological choices may amplify divergence in YT MXD and LWBI calibration models.