Empirical models of tree top breakage and scattered wind-throw in a Pinus radiata and Pseudot.suga menziesii plantation forest in New Zealand. (2020)
Type of ContentElectronic Thesis or Dissertation
Degree NameMaster of Forestry Science
PublisherUniversity of Canterbury
AuthorsWhiteley, Yanninashow all
In this thesis, attritional damage to trees attributed to wind and snow was studied in Pinus radiata (radiata pine) and Pseudotsuga menziesii (Douglas-fir) at Geraldine Forest, a 5,500 hectare forest in the South Island of New Zealand, based on tree damage descriptions from forest sample plot data. This damage largely comprises the breakage of live stems, and also a smaller number of windthrown trees. The mean levels of damage were compared by species, and it was established that the damage levels are significantly different, with Douglas-fir exhibiting higher broken heights, lower proportion of trees damaged, and lower proportions of trees alive than radiata pine.
With these results established, damage was modelled for each species separately, using both mixed- effects linear (or generalised linear) regression and random forests to create empirical models. The three response variables were the mean height of broken trees per plot (P_tree_ht_mean_BRKN); the proportion of trees with damaged tops per plot (Tops_prpn_DAM); and the proportion of live trees per plot (Prpn_LIVE). None of the models created were suitable for re-use with new data, due to bias in the model results and the reliance of the models on mixed-effects.
Three models had sufficient explanatory power to demonstrate that some particular tree and topographic variables correlate with damage levels. These models, all at the plot level, were 1) radiata pine P_tree_ht_mean_BRKN by linear regression with mixed-effects, 2) radiata pine P_tree_ht_mean_BRKN by random forest, and 3) radiata pine Tops_prpn_DAM by logistic regression with mixed-effects, using only data where the top of every tree had been assessed.
For the model of radiata pine P_tree_ht_mean_BRKN by linear regression with mixed-effects, the age of the trees, the proportion of trees pruned, and the aspect correlated with the height at which trees break, with the stand identity as a mixed effect. For the model of radiata P_tree_ht_mean_BRKN by random forests, the year of establishment, the per-hectare equivalent basal area, the age of the trees, the height of the unbroken trees, the pruned proportion, the mean pruned height, and a group of weather variables correlated with the height at which trees break. For the model of radiata pine Tops_prpn_DAM by generalised linear model with mixed-effects, the per-hectare equivalent stocking, the mean diameter of unbroken trees, and the proportion of live trees correlated with the proportion of trees damaged, with the stand identity and the plot number as mixed effects.
From these results, and by comparison with previous research into empirical models of damage to trees by wind and snow, some management recommendations have been made to reduce future damage by wind and snow at Geraldine Forest. The first is that if low levels of damage are highly desired, then Douglas-fir is the better species to plant. To reduce the levels of damage in radiata pine, any or all of the following measures apply. The first is to avoid growing radiata pine on slopes with north-east and or south-east aspects, and/or in areas of low topographic shelter, both of which positively correlate with higher proportion damaged. The second is to choose a low stocking for radiata pine, as high stocking is correlated with higher proportion damaged, but without implementing very heavy or very late thinning. The third recommendation is short rotations for radiata pine, as the age of trees is a strong predictor of damage levels. The fourth recommendation is to plant radiata pine at low elevations; height growth is faster at lower elevations and so trees will attain a desirable size in a shorter rotation; also, taller trees have higher broken heights, leaving a longer salvageable portion of stem below any breaks. The fifth recommendation (which runs somewhat counter to the third) is to prune the radiata pine crop, because pruned radiata pine breaks at higher heights, again leaving more salvageable stem.