Abstract
Bootstrap confidence intervals are developed for mean ring-width chronologies of red spruce (Picearubens Sarg.). These confidence intervals are robust, easy to compute, and useful for determining the "significance of recent trends in radial growth that may be related to forest decline. By comparing the mean chronologies of different age-classes of red spruce trees in two different collections, it is shown that a serious age-related bias may result if the series are blindly averaged without considering the differences in growth rate related to tree age and stand dynamics. However, even when age differences are taken into account, there is evidence for a common increase in radial growth rate of red spruce in the 1950s followed by rapid and sustained decrease after 1960. The cause of this widespread growth increase and decrease has not yet been established, and some current hypotheses are discussed.