Improving models of wood density by including genetic effects: A case study in Douglas-fir

Many models have been published for relating wood characteristics, such as wood density, to growth traits. At a tree population level, ring density is known to be significantly correlated with cambial age and ring width. Howe ver, at the individual tree level, the predictive value of models based on this relationship is usually poor, as there is an important, so-called "tre e effect" in the residuals of such models. We hypothesise that this effect arises from within population genetic variability, and have tested this hyp othesis by adjusting linear models for Douglas-fir populations with differe nt levels of genetic variability, ranging from provenances to clones. The a ddition of a genetic effect significantly increased the predictive value of the model and decreased the residuals. At the clone level, for example, in clusion of the genetic effect increased the explained variance (adjusted R- 2 value) from 20% to 54%. It is suggested that most of the observed variabi lity in the wood density/growth relationship of Douglas-fir populations has a genetic origin.