Discussion
Thinning increased the structural complexity of individual Scots pine trees confirming our hypothesis H1. Crown projection area showed a positive relationship (mean correlation coefficient >0.6) with structural complexity whereas tree height did not, leading to partially accepting the hypothesis about associations between structural complexity and horizontal and vertical measures of Scots pine trees (H2). Finally, the hypothesis H3 is also partly accepted as there was practically no relationship between structural complexity and benefit-to-cost ratio, tree growth, or the light availability but crown and growth attributes were, however, significant when estimating structural complexity. Forest management, and thinning especially, affected structural complexity of individual Scot pine trees. Stem, crown, and growth variables were found significant predictors for it indicating them as drivers for structural complexity.
Crown dimensions affected structural complexity more than stem attributes (i.e. DBH, tree height, and volume) which is similar findings to Seidel et al. (2019b) who studied the relationship between structural complexity and stem and crown attributes of four deciduous species (i.e.Fagus sylvatica L., Fraxinus excelsior L., Acer pseudoplatanus L., and Carpinus betulus L.). Thinning affected stem or growth attributes, benefit-to-cost ratio, and light availability of Scots pine trees, as either moderate or intensive thinning resulted in differing values for these attributes compared to Scots pine trees without a thinning treatment (i.e. control plots).
There was no relationship (R2<0.2 and correlation coefficient <0.1, not significant) between structural complexity and benefit-to-cost ratio indicating that it does not affect structural complexity of Scots pine trees. This is contradictory to the findings by Seidel et al. (2019a) who found R2 of at least 0.25 between structural complexity and cost-to-benefit ratio for 76 deciduous trees (i.e. 46 Fagus sylvatica L., 25 Fraxinus excelsior L., and 5 Acer pseudoplatanus L.). Their study site was a mixed, unmanaged deciduous forest and there were ~150 trees/ha whereas in our study the tree density per ha was at least twice of that. Thus, it can be assumed that there was more space for the deciduous trees to grow. Additionally, tree form in general is different between conifers and deciduous trees due to the difference in their shoot growth, in other words the degree of apical dominance differs being strong for conifers and weaker for deciduous trees (Kozlowski 1964). This can produce differences in structural complexity and its relationship to benefit-to-cost ratio.
Light availability (measured through competition) did not affect structural complexity of Scots pine trees and there was no relationship between them. However, larger variation in light availability was found in plots without thinning treatments, which is expected as it has shown in previous studies that thinning decreases competition (Mäkinen & Isomäki 2004, Juchheim et al. 2017, Jacobs et al. 2020). Seidel et al. (2019a) reported decreasing structural complexity when competition increased but their study only included 93 read oak (Quercus rubra L.) trees from ten different sites compared to our ~740 Scot pine trees. Similar to the results of benefit-to-cost ratio discussed above, deciduous trees have a different branching pattern compared to conifers. Future research should focus on the response of structural complexity to light availability of different species with enough trees to clearly understand their relationship.
Ehbrecht et al. (2017) studied stand structural complexity in mixed stands with even-aged and uneven-aged forest management. They discovered that stand structural complexity differed between tree species and even-aged coniferous stands had less complex structure compared toFagus sylvatica (L.) stands. We only studied one tree species, namely Scots pine, but found that structural complexity varied between trees with different management history, although all the plots were even-aged and single layer that could have decreased their structural complexity.
Forest structural complexity has been widely studied (Camarretta et al. 2020, McElhinny et al. 2005, Ishii et al. 2004) but there is less research on tree structural complexity, which was the focus of this study. TLS can provide traditional tree attributes (i.e. DBH, volume, crown dimensions) for characterizing tree structure but its potential can also be expanded for new attributes. Here, we only utilized box dimension to measure structural complexity, which can be considered as a weakness of the study as we cannot compare these results to other measures. However, the box dimension was chosen because it was easy to calculate, it has shown its potential in characterizing individual tree structure (Seidel et al. 2019a, b) , and because no other measures for individual tree structural complexity was found in the literature. As it was shown, there was a relationship between box dimension and stem, crown, and growth attributes indicating that these attributes can explain structural complexity if individual trees. None of them, however, considers the entire 3D tree structure to which the box dimension brings added value. The strength of the study is that the study design allowed us to compare forest management practices without confounding factors that could have been a challenge in uneven-aged or mixed-species stands.
Forest management in general, and thinning in particular, controls the between-tree competition especially by removing part of shadowing canopy mass to enhance the growth of remaining trees (White 1980). Although there is a strong economic incentive in forest management for wood production (Puettman et al. 2015), there is an increasing understanding how forest management can also be applied for supporting diversity of forest structure (Bergeron et al. 2002, Kuuluvainen 2009), biodiversity (Fedrowitz et al. 2014), resilience (Messier et al. 2013), and carbon uptake (Hardiman et al. 2011). Additionally, there is an increasing respect towards recreational opportunities and landscape amenities (Butler & Leatherberry 2004, Hugosson & Ingemarson 2004, Urquhart & Cortney 2011). Therefore, structural diversity has been identified as a silvicultural principle that could be addressed with non-conventional forest management practices (Puettmann et al. 2015). Furthermore, forest structural variety can be used as a measure for biodiversity and structurally complex forests enhance carbon uptake (Gough et al. 2019), which are important ecosystem services provided by forests. Thus, it can be expected that structural diversity can also be beneficial for functional and species diversity of forests.