ABSTRACT

Stable carbon isotope ratios (δ¹³C) in tree rings have been increasingly used as indicators of climatic and ecological processes. However, compared to inter-annual δ¹³C variability, the factors related to inter-tree and -site variability (i.e., dendroclimatic noise) have attracted less attention. Here we present a statistical study quantifying the inter-tree and -site variability in a tree-ring δ¹³C dataset in the context of biogeographical, climatic, and forestry variables. This dataset originates from sixty-seven Scots pine (Pinus sylvestris L.) trees growing in seven sites in the north boreal forests and near the subarctic timberline in Finnish Lapland. Using estimates from a linear mixed model, we identified a set of three constraints, possibly behind the inter-tree and -site δ¹³C variations. First, a latitudinal constraint was found with more negative δ¹³C values in colder (timberline) conditions, in comparison to trees in south of timberline. Second, a longitudinal constraint appeared to result in more negative δ¹³C levels in the eastern sites. Third, an altitudinal constraint led to less negative δ¹³C levels at higher altitudes. Additional source of inter-tree and -site δ¹³C variations attributable to tree size variables resulted in less negative δ¹³C values in taller trees with larger crowns. As the tree size and age are closely connected in our data, these results supported earlier findings that Pinus δ¹³C data may exhibit age-related changes. These results were consistent with mechanistic links between environmental drivers and carbon isotope fractionation. The extent of inter-tree and -site variability affected the length of confidence intervals around the mean δ¹³C value. The use of statistical power analyses was highlighted for assessing the required sample size to estimate the mean value.