Effect of sample selection on the environmental signal derived from tree-ring series

Possibilities to improve the expressed population signal (EPS) derived from tree-ring chronology were studied on 19 trees sampled from a permanent Scots pine ( Pinus sylvestris L.) thinning experiment in southern Finland. Trees of varying relative size from thinned and unthinned control plots, as well as radii in different compass directions and different positions along the stem, were compared. The EPS values increased with increasing number of chronologies sampled. The effect of increasing tree number was more pronounced compared to increasing number of radii per tree. EPS also increased when the social status of a tree improved, but no clear differences were found between unthinned and thinned plots. Furthermore, no clear trend existed between different heights along the stem. The highest EPS values occurred in the directions west, southwest, and south; correspondingly, the lowest value was on the north side of the bole, followed by the northeast and east. It was concluded that some reduction in the noise present in tree-ring chronologies can be achieved by paying more attention to sample selection within and between stems. The sunny side of the bole especially, as well as dominant trees, should be preferred. However, the most important factor in noise reduction is adequate replication.