Climatic information from tree-rings of Pinus sylvestris L. and a reconstruction of summer temperatures back to AD 1500 in Femundsmarka,eastern Norway,using partial least squares regression (PLS) analysis

Chronologies of tree-ring width (RW), earlywood width (EW) and latewood width (LW) were costructed back to AD 1500 (AD 1360 with small sample size) using Scots pine, Pinus sylvestris L. in the Femundsmarka area, eastern Norway. The influence of 17 monthly temperature, 17 monthly precipitation and three prior growth variables on RW, EW and LW indices were studied for different periods since AD 1872 using partial least squares regression (PLS). The most important variable for explaining radial growth was the current July temperature with regard to RW, EW and LW. Other important variables were the temperatures in the preceding December and the current August and June, and the tree-ring index of the previous year. No common and consistent pattern was found for monthly precipitation and previous summer temperatures. The climate-growth responses from the different standard 30-year periods varied considerably, indicating that the radial growth is very sensitive to even minor climatic changes. Estimates of July and July-August temperatures were made back to AD 1500 using PLS with RW, EW and LW chronologies as predictors. Best reconstructions were achieved with a six- or eight-variable model in which July and July-August temperatures in year t are estimated as functions of RW, EW and LW from two, three or four succeeding years (Q 1, t, t + 1, t + 2). The best reconstruction equations of July and July-August temperature accounted for 61%, and 56% of the temperature variance over a 61-year calibration period and 53%, and 34%/ of the variance over an independent 61-year verification period. The reconstructions indicate that, since AD 1500, warm summers clustered around AD 1510, 1770 and 1940, and the cool summers around AD 1600, 1710 and 1800. Summer temperatures have increased since AD 1860 even though there was a short cold period centred around AD 1910. It is concluded that the PLS method is well suited to the analysis of climate-growth relationships and creating models for reconstructing past climate variables.