Detection and evaluation of an early divergence problem in northern Fennoscandian tree-ring data

• Published in: Oikos
• Main Authors: Schneider, L., Esper, J., Timonen, M., Büntgen, U. (Ulf)
• Format: Article in Journal/Newspaper
• Language: English
• Published: 2014
• Subjects: growth; 20th-century; temperature; sensitiviti; trends; Alps; Fennoscandia; Fennoscandian
• Online Access: https://doi.org/10.1111/j.1600-0706.2013.00836.x; http://hdl.handle.net/11104/0237429

Although not yet fully understood, reduced sensitivity of tree growth to temperature at high northern latitudes during the last similar to 40 years is often linked to concurrent anthropogenic changes of atmospheric composition and global warming. The idea that a temporal localization of the problem could improve its understanding initiated a search for erratic growth-patterns in earlier periods of high quality dendrochronological archives. An extensive network of maximum latewood density (MXD) measurements from northern Fennoscandia likely represents one of the most reliable regional summer-temperature reconstructions. The strong coherence between proxy and instrumental data is, however, interrupted by a short, but significant correlation decrease from similar to 1900 till 1925, a period of distinct summer-temperature warming. Here we analyze this early 20th century divergence period (EDP). We therefore use long instrumental station records and tree-ring density chronologies including 878 Pinus sylvestris and 126 Picea abies samples. Our results indicate that EDP was accompanied by a simultaneous decline of inter-site and inter-station correlations. This could be ascribed to substantially reduced inter-annual summer temperature variability from 1905-1919. Stable correlations of the MXD network with high-pass filtered sea level pressure and precipitation records imply tree-growth to be additionally controlled by other factors, e.g. light conditions, in periods of low summer temperature variability. Within the scope of this study, the causes for EDP could be confined to a limited area and a short period. Calibration of proxy data and reconstruction skills thus remain unaffected in this case of divergence.