Figure 5. Terrestrial laser scanning point clouds from example trees with varying structural complexity (i.e. box dimension).
When assessing drivers of structural complexity, all stem, crown, and growth attributes were found significant (p<0.05) in the nested two-level linear mixed-effects models (Table 3). Benefit-to-cost ratio and light availability, on the other hand, were not. Intensive thinning increased tree height, benefit-to-cost ratio, height growth, and light availability significantly (p<0.05), whereas benefit-to-costa ratio, height growth, light availability was significantly (p<0.05) smaller in the linear mixed-effects models (Table 3). However, coefficient of determination was <0.2 for all other architectural attributes (Figures S1-S4) except for crown dimensions where it was 0.5 between box dimension and crown projection area and crown volume (Figure S5).
Thinning treatment was statistically significant (p<0.05) in mixed-effects models where height and volume growth, benefit-to-cost ratio, light availability, and each stem attribute was included as predictor variable at a time. Tukey’s honest significance test revealed that there was a statistical difference (p<0.05) between moderate and intensive thinning in all stem attributes, benefit-to-cost ratio, height and volume growth, and light availability. Thinning, either moderate or intensive, resulted in significant difference (p<0.05) in all stem attributes, crown width, benefit-to-cost ratio, all growth attributes, and light availability when compared to trees without a thinning treatment.
Table 3. Results of the nested two-level linear mixed-effects models with box dimension as dependent variable and thinning treatment together with stem and crown attributes, benefit-to-cost ratio, growth attributes, and light availability as independent variables. * denotes statistical significance of p-value<0.05.