Geographic Variations of the Wood Density and Fiber Dimensions of the Persian Oak Wood

: Persian oak ( Quercus brantii Lindl.) is a valuable native species in Iranian forests with very limited availability of data on its wood properties. The objective of the current study was to determine the inﬂuence of altitude and slope on physical properties and ﬁber dimensions of Persian oak wood. In addition, the relationship among wood properties, site conditions (temperature and rainfall) and growth traits of trees (tree height, DBH, basal area, age, crown diameter, crown basal area, volume and annual diameter increment) were studied by principal component analysis (PCA). Three altitude levels (1730, 1980 and 2250 m) and three slope classes ( < 30%, 30–45% and > 45%) were considered in the current study. It was determined that trees growing in the intermediate altitude (1980 m) showed the highest oven-dry density values, and those in the lowest altitude (1730 m) revealed the lowest ones. The results also indicate signiﬁcant statistical di ﬀ erences between altitude levels and slope classes on the ﬁber length, ﬁber diameter and volumetric swelling at the 99% conﬁdence interval while no signiﬁcant di ﬀ erences were found between average values of oven-dry density among di ﬀ erent altitudes and slopes. PCA analysis indicated that altitude and temperature are the most important factors a ﬀ ecting the wood properties. Knowledge of the relationship between wood properties and environmental factors are essential in terms of both forestry management and wood applications.


Study Area and Sampling
The research area is located between 49 • 59 43" N and 50 • 15 28" N and between 32 • 07 42" E and 32 • 22 25" E in the Bazoft region of Chaharmahal Va Bakhtiari, Iran ( Figure 1). Persian oak forests in this study area are spread between 1500 and 2300 m in altitude. The mean annual precipitation and temperature of the study area are 330 mm and 14 • C, respectively. This study was performed on trees collected from nine ecosites (three forest sample stands along an altitudinal gradient (1730,1980 and 2250 m a.s.l.)) × three terrain slope classes (less than 30%, 30-45% and more than 45%) in the Persian oak forests of Bazoft, creating three replicates in each ecosite (3 × 3 × 3 = 27 sampling plots, one sampled tree in each plot = 27 healthy dominant trees in total) ( Figure 1). Due to their remote location and the absence of evidence of human impact, it is assumed that all stands have been developed under the influence of natural impacts and disturbances. In each ecosite, three plots were implemented. In total, 27 healthy dominant trees (3 trees × 3 altitudes × 3 slopes) were selected. In each circular plot, all living trees of at least 7.5 cm diameter at breast height (DBH, 1.30 m above the ground) were identified, and their diameter at breast height, height, crown length and perpendicular diameters were recorded within 0.1 ha area. The caliper, vertex and diameter tape were used to measure tree diameters, height and crown diameter, respectively. Then, in each plot, one healthy dominant tree with the largest diameters at breast height (DBH) without any defects and reaction wood was sampled and one disk was taken from the tree trunk at DBH for the determination of wood properties. Then, the tree age at breast height (ABH) was obtained by counting the annual rings of the sampled disks. Finally, the mean annual diameter increment (MADI) was obtained by dividing DBH by the number of annual rings. Stand-level traits such as stand basal area, stand density and stand quadratic mean diameter (QMD) and tree-level traits such as stem basal area, stem volume, tree crown diameters and tree crown basal area in each ecosite were calculated based on 27 circular 0.1 ha sample plots data. The tree crown diameter was measured as the average crown spread is the average of the lengths of longest spread from edge to edge across the crown and the longest spread perpendicular to the first cross-section through the central mass of the crown. Tree crown basal area as tree crown area projection was calculated based on the crown area as a circle using average crown spread as its diameter. The annual records of total precipitation and mean annual temperature, the main climatic factors affecting the wood properties and growth of Brant oaks in the region, were obtained from the Koohrang Synoptic Meteorological Station (32.46° N, 50.13° E; 2365 m a.s.l.) [26].

Preparation of Samples
For determination of physical properties and fiber dimensions, 5-cm-thick disks were taken at 1.30 m in height. Tree sampling and specimen preparation for physical properties and fiber dimensions are shown in Figure 2. In each circular plot, all living trees of at least 7.5 cm diameter at breast height (DBH, 1.30 m above the ground) were identified, and their diameter at breast height, height, crown length and perpendicular diameters were recorded within 0.1 ha area. The caliper, vertex and diameter tape were used to measure tree diameters, height and crown diameter, respectively. Then, in each plot, one healthy dominant tree with the largest diameters at breast height (DBH) without any defects and reaction wood was sampled and one disk was taken from the tree trunk at DBH for the determination of wood properties. Then, the tree age at breast height (ABH) was obtained by counting the annual rings of the sampled disks. Finally, the mean annual diameter increment (MADI) was obtained by dividing DBH by the number of annual rings. Stand-level traits such as stand basal area, stand density and stand quadratic mean diameter (QMD) and tree-level traits such as stem basal area, stem volume, tree crown diameters and tree crown basal area in each ecosite were calculated based on 27 circular 0.1 ha sample plots data. The tree crown diameter was measured as the average crown spread is the average of the lengths of longest spread from edge to edge across the crown and the longest spread perpendicular to the first cross-section through the central mass of the crown. Tree crown basal area as tree crown area projection was calculated based on the crown area as a circle using average crown spread as its diameter. The annual records of total precipitation and mean annual temperature, the main climatic factors affecting the wood properties and growth of Brant oaks in the region, were obtained from the Koohrang Synoptic Meteorological Station (32.46 • N, 50.13 • E; 2365 m a.s.l.) [26].

Preparation of Samples
For determination of physical properties and fiber dimensions, 5-cm-thick disks were taken at 1.30 m in height. Tree sampling and specimen preparation for physical properties and fiber dimensions are shown in Figure 2.

Physical Properties
After the preparation process, specimens with dimensions of 3 cm × 2 cm × 2 cm were prepared in accordance with ISO 13061-14 [27] for the investigation of oven-dry density and volumetric swelling. In total, 270 specimens from different parts of disks (10 specimens per each disk) were prepared.
Sample dimensions were measured in green (saturated) and oven-dry condition with a slide caliper; oven-dry mass was determined with an electric balance to an accuracy of 0.01 g. Swelling was calculated using the dimensional change from the green to oven-dry condition. The physical properties were calculated according to the following equations: where: D0 is the oven dry density (g . cm −3 ), αv is the volumetric swelling (%), vs. is the volume in state of saturate (cm 3 ), V0 is the volume in state of oven-dry (cm 3 ) and P0 is the mass in state of oven dry (g).

Wood Fiber Properties
Separation of individual wood fiber was performed using Franklin method [28]. From each disk, wood specimens with the dimension of 15 mm × 10 mm × 2 mm were saturated in a mixture (1:1) of acetic acid and oxygenized water in test tubes. Afterwards, the specimens were kept in the oven at 65 ± 3 °C for 48 h. After maceration, the specimens were washed (2-3 times) in distilled water and then they were immersed with distilled water, shacked and the fiber dimensions (fiber length, fiber diameter and cell wall thickness) were measured with the use of microscope. From each disk, at least 50 fibers were used for the measurements.

Statistical Analysis
Normal distribution of data is one of the preconditions of the multivariate analysis of variance (MANOVA). Therefore, datasets were tested for normal distribution using the Shapiro-Wilk test of

Physical Properties
After the preparation process, specimens with dimensions of 3 cm × 2 cm × 2 cm were prepared in accordance with ISO 13061-14 [27] for the investigation of oven-dry density and volumetric swelling. In total, 270 specimens from different parts of disks (10 specimens per each disk) were prepared.
Sample dimensions were measured in green (saturated) and oven-dry condition with a slide caliper; oven-dry mass was determined with an electric balance to an accuracy of 0.01 g. Swelling was calculated using the dimensional change from the green to oven-dry condition. The physical properties were calculated according to the following equations: where: D 0 is the oven dry density (g·cm −3 ), α v is the volumetric swelling (%), vs. is the volume in state of saturate (cm 3 ), V 0 is the volume in state of oven-dry (cm 3 ) and P 0 is the mass in state of oven dry (g).

Wood Fiber Properties
Separation of individual wood fiber was performed using Franklin method [28]. From each disk, wood specimens with the dimension of 15 mm × 10 mm × 2 mm were saturated in a mixture (1:1) of acetic acid and oxygenized water in test tubes. Afterwards, the specimens were kept in the oven at 65 ± 3 • C for 48 h. After maceration, the specimens were washed (2-3 times) in distilled water and then they were immersed with distilled water, shacked and the fiber dimensions (fiber length, fiber diameter and cell wall thickness) were measured with the use of microscope. From each disk, at least 50 fibers were used for the measurements.

Statistical Analysis
Normal distribution of data is one of the preconditions of the multivariate analysis of variance (MANOVA). Therefore, datasets were tested for normal distribution using the Shapiro-Wilk test of normality, and then data transformation was applied for those that were not normally distributed. Levene's test was employed to examine the homogeneity of variances. MANOVA was conducted to evaluate significant differences between mean values of studied wood properties at altitude levels and slope classes using SPPS 23 (IBM, Armonk, NY, USA). This method generated a multivariate dataset that was interpreted by using principal component analysis (PCA). The patterns of variation of the measured traits were selected and used for performing PCA based on the method of alternating least squares (PRINQUAL procedure in SAS). A Kruskal secondary least-squares monotonic transformation was applied to all variables, with the restriction that ties were preserved. The results are presented by a PCA biplot that shows the transformed variables (e.g., the measured traits) projected onto the two-dimensional plane of the analysis described by the two principal components. PCA allowed the multivariate dataset to be reduced by minimizing multicollinearity and associating the correlated traits in two principal axes [13]. This approach improved data exploration and simplified the interpretation of the results.

Results
Descriptive statistics for the Persian oak sample trees, studied traits and forest stands in the study area are presented in Table 1.
Data normality was checked by the Shapiro-Wilk test (the data was normal; sig > 0.05), while the homogeneity of variances was tested using the Levene's test (the variances were homogeneous; sig > 0.05).

Oven-Dry Density
The average values of oven-dry density determined for Persian oak wood in three different altitude levels and slope classes are presented in Table 2. Multivariate analysis of variance (MANOVA) results revealed no significant differences between mean values of oven-dry density at altitude levels and slope classes ( Table 3). The highest (0.83 g/cm 3 ) and lowest (0.73 g/cm 3 ) values of oven-dry density were identified in the intermediate altitude step on the slope of 30-45% and in the lowest altitude on the slope of <30%, respectively. Overall, a mean value of oven-dry density of 0.78 g/cm 3 and a coefficient of variation of 7.84% was achieved for Persian oak wood.   Table 2 shows the mean values of volumetric swelling for Persian oak. From the MANOVA, it can be derived that the effects of altitude levels and slope classes on volumetric swelling are significant at the 0.01 significance level ( Table 3). The highest (21.74%) and lowest (10.92%) values of volumetric swelling were found in the intermediate altitude level on the slope of 30-45% and in the lowest altitude step on the slope of >45%, respectively. Overall, a mean value of volumetric swelling of 17.34% and a coefficient of variation of 23.80% were determined.

Fiber Length
The highest (0.96 mm) and lowest (0.70 mm) values of fiber length were found in the lowest altitude step on the slope of 30-45% and in the high altitude on the slope of >45%, respectively (Table 4). Average values of fiber length (0.87 mm) and coefficient of variation (2.05%) were achieved in total. According to the results of MANOVA, altitude and slope have a significant statistical effect on the fiber length at the 99% confidence interval (Table 3).

Fiber Diameter
A maximum value of 23.30 µm and a minimum value of 19.70 µm for fiber diameter were determined in the lowest altitude level on the slope of 30-45% and in the highest altitude level on the slope of <30%, respectively (Table 4). Overall, average values of fiber diameter of 20.49 µm and coefficient of variation of 9.42% were determined. MANOVA test results indicate significant differences between average values of fiber diameter between three altitude and slope groups at the 99% confidence interval (Table 3).

Cell Wall Thickness
The maximum value of the parameter cell wall thickness (6.02 µm) as well as the minimum (5.55 µm) were determined for the selected oak trees in the lowest altitude level on the slope of <30% and in the lowest altitude level on the slope of 30-45%, respectively (Table 4). In total, average values of cell wall thickness of 5.77 µm and coefficient of variation of 3.95% were achieved. MANOVA results in Table 3 reveal that both altitude and slope have no effect on the parameter cell wall thickness.

Principal Component Analysis
The results of Kaiser-Meyer-Olkin (KMO = 0.57) and Bartlett's test of sphericity (p < 0.01) confirmed the adequacy of dataset for PCA. The transformed parameters of the measured traits were projected onto two dimensional planes generated by the first two canonical axes ( Figure 3). Graph of study parameter and biplot of sample plots as well as study parameter based on first and second components are presented in Figures 3 and 4 Canonical Axis 1 represented 33.62% of the variability, while the variance accounted for by Canonical Axes 2 and 3 represented 23.52% and 12.33% of the variability, respectively (Figure 3 and Table 5). The analysis of the PCA shows that three components explain 69.47% of the total variance (Table 5). Similar directions with respect to the origin revealed high correlations between the parameter of the measured traits. This is related to the volumetric swelling, fiber diameter and fiber length. Total variance explained by five principal components is 85.68% (Table 5).  The first component explains 33.62% of the total variance and is most highly correlated with the following study variables: tree volume, DBH, tree basal area, tree age and tree height. These traits are not particularly correlated with the other principal components. The second component describes 23.52% of the variance mainly influenced by the volumetric swelling, altitude, temperature, fiber diameter and fiber length. The third component explains 12.33% of the total variance relating to the tree crown diameter and tree crown basal area. The fourth component describes 9.64% of the variance influenced by the tree annual diameter increment, slope and precipitation. The fifth component explains 6.57% of the variance influenced by the settings cell wall thickness and oven-dry wood density (Tables 5 and 6).

Discussion
Denser woods are characterized by higher mechanical properties and are more resistant to (mechanical) failures. Wood density is generally a complex variable and is related to many factors such as anatomical characteristics, e.g., vessel and fiber morphology, ecological site, moisture content and chemical constitutes [29,30]. In the present study, the density of Q. brantii is higher than that of Q. robur [31], Q. cerris [32] and Q. rubra L. [33]. However, oven-dry density values of Q. brantii evaluated for the Lordegan site in Iran (1.01 g/cm 3 , [25]) are significantly higher as compared to values of the Bazoft region in this study. This difference could be related to the site effects and stand age. Regarding the impact of site factors, there are no significant differences for average values of oven-dry density between altitudes and slope groups. This agrees well with findings by Guilley et al. [34] for Q. petraea. Overall, the highest average value of oven-dry density was found in the intermediate altitude. These findings are in line with previous findings of Sopushynskyy et al. [35] for F. sylvatica L and Kaygin et al. [17] for P. sylvestris L. However, Berges et al. [24] indicated that wood density decreased with increasing altitude for sessile oak.
Furthermore, the volumetric swelling is mainly affected by wood density (Guler et al. [36]). In the current study, the high altitude corresponds to the higher volumetric swelling, while the lowest altitude level had the lowest volumetric swelling. As is well known, the relationship between wood density and volumetric swelling is positive [29]. A similar finding was reported by Kiaei [37] for Carpinus betulus.
Fiber morphology and orientation are the primary elements responsible for the strength of wood and play an important role in determining the qualitative and quantitative wood properties and specific utilization of lignocellulosic materials [38,39]. The statistical evaluation of the wood fiber dimensions show that the altitude affects fiber length and fiber diameter of Persian oak but it is not influencing cell wall thickness. The average value of fiber length and fiber diameter declined while cell wall thickness increased with increasing altitude. Similar observations were previously reported by Kieai et al. [11] for Carpinus betulus, Noshiro et al. [18] for Alnus nepalensis and Yılmaz et al. [19] for Quercus pontica.
According to Wheeler et al. [40], fibers are classified into three groups: (1) short fibers with a length less than 0.90 mm; (2) fibers of medium length between 0.90 and 1.90 mm, including Persian oak with an average fiber length of 0.95 mm; and (3) fibers longer than 1.90 mm.
The measured average fiber length of Persian oak is lower than that reported, e.g., for Cork oak [41] and most hardwoods [42].
Fiber diameter and cell wall thickness depend on exogenous growth conditions and annual ring width [30]. The average value of fiber diameter of Persian oak wood at different attitude and slopes amounts to 20.49 µm, which is in agreement with those values reported for other hardwood fibers [43]. Plomion et al. [44] reported that the variations in the fiber diameter depend on molecular and physiological changes occurring in the vascular cambium as well as in the wood cell walls throughout the tree growth.
Cell wall thickness plays a key role in wood quality and its strength properties. Basically, this important anatomical parameter is variable among species and sites, between and within trees, and it is as well highly correlated with wood density. A positive correlation between wood density and cell wall thickness is described by many researchers (e.g., [10,45]). According to the results of the present study, the highest average values of wood density and cell wall thickness were obtained at the altitude of 1800-2000 m. The increased density at the representative altitudes can be explained with higher fiber cell wall thickness and content of cellulose, lignin and extractives [11]. For comparable oak species of Q. acutissima, Q. dealbata, Q. fenestrate, Q. lanceofolia and Q. semiserrata, cell wall thicknesses of fibers in the range of 7.49-18.26, 5.30-19.86, 6.49-17.05, 7.52-27.42 and 6.37-14.92 µm were described, respectively, by Sharma et al. [46]. The values of these oak species are slightly higher as compared to the results of our own measurements for Q. brantii, which can be explained by individual genetic, physiological and silvicultural impacts.
By using PCA, not only the number of comparisons between treatment means is reduced, but also the meaningfulness of these comparisons is enhanced. In the present study, multivariate analysis was used to identify obvious differences in wood properties and the selected traits. The results show that the first axis has the highest correlation with tree-level traits and the second axis has the highest correlation with the traits wood properties, altitude and temperature. Comparable results were reported by Kiaei et al. [11], Noshiro et al. [18], Yılmaz et al. [19] and Kaygin et al. [23].
As mentioned above, the location of the traits wood properties, tree characteristics and forest stand in different regions of PCA axes is based on correlation coefficients between the traits. Therefore, the location of the traits in the diagram (Figure 3) is significant and important. For example, the traits forest stands and wood properties that occupied a specific region and are close to each other have several similar characteristics between each other and consequently the correlation coefficient between those characteristics occupying opposites places in the diagram ( Figure 3); for example, opposite directions of an axis show distinct variations. According to the PCA results, altitude and temperature are the most important traits affecting the wood properties. Rossi et al. (2015) [13] and Van Der Maaten-Theunissen et al. (2013) [47] pointed that variations in the wood properties are anticipated, especially in terms of latitude and altitude, which are inherently related to temperature and rainfall. Summarizing, PCA shows noticeable variations of wood properties in the study area across altitude and slope gradient. Since this method is of high accuracy and has different abilities, it could be used for a habitat analysis and determination of effective ecological factors. Analyzing ecological data using ordination methods provides a simpler understanding of the complex relationship among growth traits, wood properties and environmental gradients. In addition, this method is impaired by the existence of ineffective factors and the data complexity of ecological methods. Further work is needed to investigate the anatomical properties (fibers proportions, rays, vessels and parenchyma), ring width and cambial age changes with geographic variations.

Conclusions
In the present study, physical and wood fiber properties of Persian oak wood were investigated at different altitudes above the sea level and slopes. The results indicate that there is a significant statistical effect of altitude and slope on the fiber length, fiber diameter and volumetric swelling of Persian oak wood. In contrast, no significant differences were found between average values of oven-dry density. The results also show that altitude and temperature are the most important factors affecting the wood properties, as shown by PCA analysis. Respective study can be used for predication of wood quality due to climate change. The properties of the wood from higher altitude will become similar to the wood from lower altitudes.
Author Contributions: N.N., data curation; M.B., designing and performing experiments and writing the original paper; S.K., analyzing data and editing; M.H., review and editing; and G.K., review and editing. All authors have read and agreed to the published version of the manuscript Funding: This research received no external funding.