Effect of ring width, cambial age, and climatic variables on the within-ring wood density profile of Norway spruce Picea abies (L.) Karst.

Studying the effects of dendrometric and climatic variables on within-ring density variations needs flexible and interpretable models. We described the within-ring density profile using a piecewise linear regression and studied its dependence on (1) dendrometric variables such as cambial age (CA) and ring width (RW), and (2) climatic variables. Based on X-ray analysis of 5,191 Norway spruce rings, a six-parameter three-segmented model was fitted on each within-ring density profile. Each model parameter was related to dendrometric and climatic variables using multiple linear regressions. Then, these models were assembled in two models relating the within-ring density profile to (1) RW and CA (model M1), and (2) climatic variables (model M2). M1 showed an R (2) of 83.4 % and a residual standard error of 68.5 kg m(-3). Larger rings were associated with a decrease of latewood proportion and mean ring density. Rings with high CA were characterised by high maximum ring density. M2 showed an R (2) of 60.9 % and a residual standard error of 94.9 kg m(-3). Warm summers increased the maximum ring density. Years with favourable water status decreased mean ring density. The piecewise linear models allowed the classification of within-ring density profiles in three types. Considering CA and RW led to the most explicative model since RW described many processes such as silviculture or climate. Earlywood density was impacted by water status while latewood density was conditioned by both temperatures and water status. Our approach may be used for the identification of within-ring density fluctuations or to assess the effects of silviculture or global change on the within-ring density profile.