Divergence in latewood density response of Norway spruce to temperature is not resolved by enlarged sets of climatic predictors and their non-linearities

In softwoods, latewood density (LWD) is strongly correlated with temperature, and serves as a basis for climatic reconstructions. However, an unexplained decline in the strength of this relationship has been reported from 1970 to 2000, which has been termed the 'divergence' problem. We inquired whether 'divergence' may be resolved by (i) the relationships between LWD and an enlarged set of climatic factors unaccounted for in traditional approaches or (ii) non-linearities in the relationships between climatic variables and LWD. We developed multi-factorial climatic models of LWD historical variations in tree-ring series of Norway spruce in north-eastern France, where divergence has been observed. The relationships between these LWD variations and 38 climatic variables were analysed. Climatic variables were obtained from homogenised climatic series and included monthly maximum and minimum temperature, potential evapotranspiration and water balance. Two climatic models of LWD were built, using projection to latent structure (PLS) regression analysis: (i) a model including linear responses to climatic variables (M1) and (ii) a model including linear and quadratic responses to climatic variables (M2). Models M1 and M2 accounted for 68.9% and 70.4% of LWD variations, respectively. Model predictions still revealed a divergence from 1965 to 1980, a period where LWD strongly decreased. When using both linear and quadratic forms of climatic variables (model M2), model predictions revealed a less acute divergence phenomenon. We concluded that our hypotheses were infirmed and that the divergence phenomenon neither fully resulted from multi-factorial, nor non-linear, climatic control of LWD, and that it may be caused by non-climatic factors, of which atmopsheric pollution that remains untested.