Intra-ring and inter-ring variations of tracheid length in fast-grown versus slow-grown Norway spruces (Picea abies)

Our study was conducted on 40 Norway spruces [Picea abies (L.) Karst.] from a stand located in the Belgian Ardennes. Twenty trees were randomly sampled from a slow-growth category, and twenty others from a fast-growth category. The hypothesis under testing is fourfold: increased tree growth rate may affect 1) the intra-ring weighted frequency distribution of tracheid length, 2) the inter-ring variation (from pith to bark) of the parameters describing this frequency distribution, 3) the interring variation of the mean tracheid length, and 4) the correlation between yearly mean tracheid length and yearly ring width. Prior to hypothesis testing, a Gaussian non-linear model with two parameters, mu (central tendency) and sigma(2) (dispersion), was found to fit adequately the intra-ring weighted frequency distribution of tracheid length. As the estimates of mu and sigma(2) increased from pith to bark (with an exponential relationship between sigma(2) and mu), so did the yearly mean tracheid length. Differences between growth categories in the three variables (i.e., mu- and sigma(2)-estimates, mean tracheid length) were not significant, although the difference in mean tracheid length was constant in magnitude and sign over rings. The initially strong and negative correlation between mean tracheid length and ring width disappeared when the effects of autocorrelation and heterogeneity of variance were removed, except for trees whose average growth rate in circumference was maintained above 2.2 cm/year by thinnings.