4.1. Effects of Stand Density on Wood Quality Characteristics
The obtained results demonstrate how the effect of stand density on wood quality can be identified by measurement of the main wood properties—wood density (WD), global modulus of elasticity (MOE), and bending strength (MOR). The stand density influences the growth of trees, the productivity of stands, and the quality of the produced wood [41
The findings of this study showed that different stand densities caused various responses of wood quality characteristics of Scots pine and Norway spruce trees. Specifically, the mean values of WD increased with increasing stand density, and the lowest values were obtained at the sites with the lowest stand density. Other studies similarly concluded that thinning (decreasing the number of remaining trees per hectare) decreased WD of Scots pine [11
] and Norway spruce [14
]. Most authors indicated that these changes were a consequence of a higher growth rate. However, studies on numerous species reported no significant effect or little effect of stand density on WD [23
] or even an increase [42
]. This discrepancy may be attributed to differences in species and geographical location.
There were statistically significant differences in the characteristics of the wood between various stand densities. A significant effect of stand density on MOE and MOR of Scots pine and Norway spruce was found, i.e., the lowest MOE and MOR were observed at the lowest stand density. Few studies are available on the impact of thinning or stand density on MOE or MOR. However, in line with our results, ambivalent results were observed for MOE and MOR of other species. These wood mechanical properties significantly decreased in Sitka spruce influenced by early thinning [44
] and in Douglas fir and Norway spruce with thinning [42
]. However, in black spruce (Picea mariana
), MOE slightly increased, but no changes of MOR values after thinning were reported [43
]. No changes in both the MOE and MOR of loblolly pine [45
] and no decrease in the MOE of Douglas fir [46
] with varying silvicultural intensity were reported. However, both MOE and MOR increased in Sitka spruce [42
]. According to Stöd et al. [28
], the first thinnings provided saw timber with the lowest MOR and MOE, whereas the material from the second thinnings provided the higher values.
For timber researchers, it is important to understand the key principles and limitations of the wood strength-grading system. Based on the theoretic strength-class distribution, Norway spruce wood corresponded to the strength class of C16 at the sites with the highest stand density. However, Scots pine wood did not reach the requirements of any strength class. Compared with other studies, about 12% of the samples were rejected for the strength class of C14 and 20% for the strength class of C16, when testing spruce wood [47
]. Similarly, Norway spruce wood met the requirements of the strength classes C18 to C22 in both thinned and unthinned stands, while Sitka spruce wood was classified in the C16 to C18 classes [42
4.3. Modeling Wood Quality Parameters in Relation to Tree Characteristics
It is known that the use of MOE and MOR models enables the industry to assess the quality of wood products and to predict the bending stiffness and strength values based on tree characteristics. MOE and MOR are considered to be essential wood properties. For the predictions of MOR, MOE was selected by the stepwise procedure. In earlier studies, a close relationship of MOE and MOR was also recorded for many tree species [22
], and MOR can be best estimated from MOE and tree characteristics [50
]. Using linear regression models, attempts were made to determine MOE and MOR from site and tree indicators. In the study of Lei et al. [48
], a stepwise method was applied to identify the best variables for predicting MOE and MOR based on the stand and tree characteristics in black spruce. The mentioned study indicated that for the prediction of MOE, stem taper was the best explanatory variable (R2
= 0.56), followed by tree crown length, DBH, stand density, and tree crown width. With the exception of stem taper and DBH, the variables positively influenced MOE. For the prediction of MOR, the MOE was the best explanatory variable, followed by tree DBH and tree crown length (R2
= 0.79). Further studies on black spruce reported that the best MOE model (R2
= 0.65) consisted of three reliable indicators: tree DBH, crown length, and WD [51
]. These authors found that the MOE model was best explained by WD. The MOR model was best described by WD and DBH (R2
Scots pine MOE and MOR were modeled in Finland and France according to three indicators: WD, ring width, and wood age [22
]. The better MOE (R2
= 0.72) and MOR (R2
= 0.84) models were determined for Finnish Scots pine than for French Scots pine (R2
= 0.52 and R2
= 0.42, respectively). Including only the MOE index in the model, the best MOR models were found (R2
= 0.79–0.95). A study in Finland found that WD had the greatest influence on the modeling of MOE and MOR in Scots pine. Another important indicator that negatively affected these parameters was the branch thickness [28
]. As indicated by Castéra et al., 1996 [50
], the effect of knots on wood strength is great, which may partly explain the relatively low R2
value of MOR.
The variations of WD, MOE, and MOR have been analysed using linear mixed models [18
]. For the best model for MOE (R2
= 0.80), the authors included four fixed indicators: the indicator property, calculated from resonance frequencies, and board length; WD at 12% moisture content; ratio of DBH of the sample trees to the mean DBH of the stand; and relative longitudinal position, calculated as the proportion of the longitudinal board position to tree height. Similarly, the best model for MOR consisted of the same fixed effects (R2
For Norway spruce and Scots pine, general linear models were applied to determine the MOE and MOR based on tree and sample indices [20
]. Only three random variables were used for MOE models: WD, mean ring width, and knot area ratio. To determine the MOR, the MOE was included in the models. The studies showed higher coefficients of determination for pine than for spruce.
In general, the models presented in this paper did not perform very well in describing the relationship of wood properties with stand and tree characteristics. The relationships found here can be expected to change with increases in stand age, the accumulation of mature wood or including specific crown parameters, site, and climate indices. In any case, these models could be an alternative tool to predict the wood strength from MOE and some stand and tree characteristics, since MOE can be obtained by various non-destructive testing methods. Further research should therefore be undertaken to examine the applicability of these findings to more fertile sites and mixed-species stands.
Differences in wood properties occur due to the different genotypes and environments of the trees, i.e., the soil and climatic conditions, in which the trees grow. When we explain the impact of forest management on wood properties, one of the explanatory statements is that any changes in tree growth conditions affect the wood properties [41
]. The forest management recommendations assume that thinning in commercial forests will take place during certain rotation periods. The first thinning primarily is aimed at improving forest growth and yield in the future. When the trees are removed during thinning, the wood quality of the removed trees may be lower than that of mature forests. It can be assumed that this was a limitation of this study because the wood samples were taken from the trees that were removed during the thinning operations. The findings, presented in this study, should be also interpreted with caution because the proportion of juvenile wood was not evaluated. In practice, the proportion of juvenile wood should be minimized due to the specific anatomical properties (short wood cells, high amount of lignin, low WD, etc.). As noted by Yang and Hazenberg in 1994, the properties of juvenile and mature wood are also affected differently by different stand densities [52
]. These authors found that the growth rate of juvenile wood was significantly different when different stand densities were compared.
Considerably more work will need to be done to determine the wood quality in the later stand development stages, sampling the wood at final felling, and testing the wood sampled from trees of different Kraft’s classes (social class that corresponds to different positions in the stand structure and crown development). Furthermore, there is a need for a cost-efficient and end-user oriented study on wood quality properties in the Baltic region.