We developed a network of tree-ring width chronologies of larch (Larix sibirica Led.) from upper tree-lines of the southeast Altai Mountains, South Siberia. Annual tree-ring variability of chronologies since A.D. 1710 was compared using factor analysis. The factor analysis clustered eight tree-ring chronologies into two groups that were used for compositing chronologies. One resulting composite chronology (A.D. 1582-1994) averaged sites from upper tree-lines in glacier-free areas and another chronology (A.D. 1090-1999) captured the sites at upper tree-lines in valleys of the Korumdu, Aktru, Yan-Karasu and Kizil-Tash Glaciers (North-Chuya Range). There is no significant difference in the estimated strength of temperature signals (June and July) of the composite chronologies. However, we observed a remarkable contrast in the decadal variability of larch growth between upper tree-lines of glacier-free areas and glacier valleys. The tree-ring growth of larch was coherent among the chronologies for the period A.D. 1582-1725. Suddenly, low-frequency similarity declined around A.D. 1730. The magnitude of differences became more pronounced after A.D. 1775 indicating three periods with opposite growth tendency (1775-1850, 1900-1915 and 1960-1994) that alternated with short periods of coherent growth. We assume that the low-frequency signal in the glacier valley larch chronology accommodates oscillations of both summer temperature and glacier dynamics. The periods of low-frequency departures are consistent with the 19th Century advance and tremendous 20th Century retreat of the glaciers. We argue that expanded glaciers enhance harmful impacts of katabatic wind on larch growth. It appears that employing tree rings from upper tree-lines of glaciated areas for estimation of decadal and centennial variability climatic proxies should be selected with great caution.

Considerable research has occurred in recent years to build Agathis australis (D. Don) Lindley (kauri) tree-ring chronologies for paleoclimate applications and to identify statistical relationships between kauri tree rings and climate. This paper reports on a multi-year study of the seasonal growth of kauri, designed to assist in the interpretation of identified statistical relationships, and to determine if kauri’s seasonal growth characteristics are dependent on tree size. To achieve this, 43 kauri (0.09-2.00 m diameter) at Huapai Scientific Reserve were fitted with vernier bands to measure circumference change over 3-4 growing seasons. Absolute (mm) and relative (proportion of total ring) monthly growth rates were calculated for each tree and statistics characterizing the timing of growth were calculated (e.g. date corresponding to 50% of growth). Tree size-related differences were assessed by splitting the data into three subsets based on size, then comparing the monthly growth rates and growth timing statistics for the subsets. The growth timing statistics were also correlated with tree diameter. A key finding is the strong dominance of spring growth, with October and November alone accounting for 38-50% of the total ring width. This result is consistent across age cohorts, although the largest trees tended to peak in November, rather than October. This indicates that kauri tree rings are likely to have value in terms of reconstructing spring conditions; consistent with reported statistical relationships between kauri tree rings and the El Niño-Southern Oscillation phenomenon. High inter-tree variance in growth rates characterized the results, but little of this variance was accounted for by tree size. Although relationships between tree size and growth characteristics were generally weak and inconsistent, they are considered sufficient to warrant a precautionary approach in the development of tree-ring chronologies for climate reconstruction purposes.