Current issue: 53(2)
Under compilation: 53(3)
Much of forestry data is characterized by a longitudinal or repeated measures structure where multiple observations taken on some units of interest are correlated. Such dependencies are often ignored in favour of an apparently simpler analysis at the cost of invalid inferences. The last decade has brought to light many new statistical techniques that enable one to successfully deal with dependent observations. Although apparently distinct at first, the theory of Estimating Functions provides a natural extension of classical estimation that encompasses many of these new approaches. This contribution introduces Estimating Function Theory as a principle with potential for unification and presents examples covering a variety of modelling issues to demonstrate its applicability.
A spatial growth model is presented for Scots pine (Pinus sylvestris L.) on a dwarf-shrub pine mire drained 14 years earlier. The growth model accounts for the variation in tree diameter growth owing to the competition between trees, the distance between tree and ditch, and the time passed since drainage. The model was used to study the effect of tree arrangement on the post-drainage growth of a pine stand. Clustering of trees decreased the volume growth by 9–20% as compared to a regular spatial distribution. Stand volume growth, for a given number of stems, was at its maximum and variation in diameter growth at its minimum when the stand density on the ditch border was 1.5–5 higher than midway between two adjacent ditches.