Site classification and growth models for Sitka spruce plantations in Ireland

Site classification has been identified as an important tool in forestry for the quantification of a forest site’s productivity potential. It is useful in forest management and silvicultural decision making for example when deciding on when to start thinning and or when to clearfell a forest of a particular species, and for estimating stands development patterns across sites. This paper presents top height-age site classification and growth models for Sitka spruce (Picea sitchensis (Bong.) Carr.) forest plantations in Ireland using nonlinear quantile regression (NLQR) methodology.Repeated top height – age measurements were recorded from over 700 Coillte Teoranta (Irish Forestry Board) silvicultural permanent sample plots (PSPs) in seven Irish forest regions over the past six decades.The conditional quantiles (0 ⩽ τ ⩽ 1) of the top height distribution at a given age were used as a sorrugate to classify Irish Sitka spruce plantations into five polymorphic site classes of I, II, III, IV and V in order of decreasing productivity, and for fitting the site class growth models using the Chapman-Richard nonlinear growth function.The NLQR methodology performed better than the traditional guide curve methodology for constructing top height-age site index curves when compared to the NLQR median (τ = 0.5) predictive growth model. The fit statistics deviance of 10639.6 for the NLQR median (τ = 0.5) predictive growth model was considerably smaller than the fit statistics deviance of 71736.3 for the comparable Nonlinear least square (NLS) mean top height-age growth model.The model predicted top heights of 27.6 m ± 2.4%, 23.3 m ± 2.9%, 20.4 m ± 2.8%, 17.4 m ± 2.9% and 16.0 m ± 2.1% are to be expected in Sitka spruce stands at age 30 years on site classes I, II, III, IV and V respectively.

► A nonlinear quantile regression method for site classification and growth modelling.
► Site classes’ model parameter estimates were independently and statistically derived.
► Nonlinear quantile regression method performed better the site index methodology.
► The site growth curves were all sigmoid and varied in a polymorphic way with site.
► Top height maximum mean annual increment occurred earlier on the best site.