POTENTIAL SAMPLING BIAS IN LONG-TERM FOREST GROWTH TRENDS RECONSTRUCTED FROM TREE-RINGS - A CASE-STUDY FROM THE ITALIAN ALPS

Tree-ring studies of long-term growth trends have often produced contr oversial results. In such studies, the largest-diameter trees in a sta nd are usually sampled. We assessed the influence of stand dynamics on long-term growth trends by examining the past diameters of all the tr ees living in two uneven-aged subalpine Norway spruce (Picea abies (L. ) Karst.) stands in the Italian eastern Alps, as reconstructed from ri ng widths. The trees were ordered according to diameter, and groups of 12 trees (the 12 largest, the 12 smallest, etc.) were formed. Differe nt diameter groups have different increment curves. In both stands, th e 12 largest trees in 1992 have not had consistently faster growth rat es than smaller trees. This indicates that changes in diameter ranking order have occurred in the past and may be expected in the future. Du ring stand development, changes occur in the relative position of indi vidual trees, as ordered by diameter. The largest-diameter trees, at a ny time, may not always have been the largest trees and may not contin ue to be so. In a given year, the largest trees on average grow slower than other trees, which will become the largest in the future. The me an chronologies of the trees that were among the largest, prior to the harvest, and which presently (in 1992) are no longer in the top 12, a nd the mean chronologies of the trees that have moved up into the top 12 show very different growth trends. If analysed out of context, they would be interpreted differently, leading to different conclusions on long-term growth trends. when only the 12 largest-diameter trees are sampled, a bias may be present, as the trees may not have been open gr own and free of competition in the past. Consequently, studies of long -term growth may be seriously affected by bias attributable to stand d ynamics and sampling strategies. In future studies, the growth pattern s of all diameter classes in a stand should be assessed, rather than r estricting the sampling to the largest diameters. (C) 1998 Elsevier Sc ience B.V.