Physical and mechanical properties of wood and their geographic variations in Larix sibirica trees naturally grown in Mongolia

We examined the physical and mechanical properties of wood in Siberian larch (Larix sibirica) trees that grow naturally in five Mongolian provenances (Khentii, Arkhangai, Zavkhan, Khuvsgul, and Selenge) and the geographic variations between them. Five trees with stem diameters of 20 to 30 cm at 1.3 m above ground were collected from each provenance. The mean values of the modulus of elasticity (MOE), modulus of rupture (MOR), compressive strength parallel to grain (CS), and shearing strength (SS) ranged from 7.03 to 9.51 GPa, 79.8 to 103.9 MPa, 46.3 to 51.1 MPa, and 10.4 to 13.0 MPa, respectively. Significant differences were found in radial and tangential shrinkage, MOE, MOR, and SS in wood among the five provenances. In addition, juvenile wood had inferior physical and mechanical properties in comparison to mature wood within and among provenances. Furthermore, there were significant differences in all examined properties, except for CS, in mature wood among the five provenances. Higher correlation coefficients were also obtained in mature wood among all mechanical properties, except for SS.

. Geographic and climatic information regarding the sampling stands and growth characteristics in L. sibirica trees used in the present study 33 . ASL above sea level, n number of sample trees, NAR number of annual rings at 1.3 m above ground level, D stem diameter at 1.3 m above ground level, TH tree height, SD standard deviation. Data on annual temperature and precipitation were provided by the Information and Research Institute of Meteorology, Hydrology, and Environment, Mongolia.

Provenance
Latitude Longitude ASL (m) www.nature.com/scientificreports/ Radial and tangential dimensions in the specimens were measured with a screw meter (MDC-25M, Mitutoyo) in air-dry and oven-dry conditions. The shrinkage in radial and tangential directions per 1% moisture content change was calculated.
Mechanical properties of wood. Mechanical properties of wood, such as bending properties, CS, and SS, were determined according to the JIS Z 2101:2009 37 .
The static bending tests were conducted using a universal testing machine (MSC 5/500-2, Tokyo Testing Machine) with a span of 280 mm. A load was applied to the center of the tangential surface of the specimen at a load rate of 5 mm/min. The load and deflection were recorded with a personal computer to calculate the MOE and MOR.
The compressive tests were conducted using a universal testing machine (RTF-2350, A&D) with a load rate of 0.5 mm/min. The CS was calculated by dividing the maximum load by the cross-sectional area.
The SS tests were conducted using a universal testing machine (MSC 5/500-2, Tokyo Testing Machine) with a load rate of 0.5 mm/min. The SS was determined by dividing the maximum load by the plane area.
Before testing, the dimensions and weight of each specimen were measured to calculate the density at testing. In addition, the moisture content of bending and compressive test specimens and SS test specimens was measured after testing by the oven-dry method 37 . The mean moisture content was 7.5% for bending and compressive test specimens and 9.8% for SS test specimens, respectively. Thus, obtained data of MOE, MOR, CS, and SS were adjusted to those values at 12% moisture content by the changing ratio due to the 1% moisture content change (4, 2, 6, and 3% for MOE, MOR, CS, and SS, respectively) described by Ishimaru et al. 38 . In the present study, trees with different ages were used. Thus, MOE and MOR values for the 10th and 30th annual rings from the pith were estimated from the logarithmic formula fitted to the radial variation of MOE and MOR in each tree. The annual ring numbers of the specimens were estimated from the radial variations of annual ring width 33 according to the method described in our previous reports 39-41 . Determination of boundary between juvenile and mature wood. Latewood tracheid lengths were measured to determine the boundary between juvenile and mature wood. Pith-to-bark strips were obtained from the 2-cm-thick disk. Small sticks of latewood were collected from each strip at every 5th annual ring, from the pith to the 20th annual ring from the pith, and also at every 10th annual ring. For Zavkhan, sticks were also collected at the 120th and 170th annual rings. The small sticks were macerated with Schulze's solution. A total of 50 tracheids at each radial position were measured using a microprojector (V-12B, Nikon) and a digital caliper (CD-30C, Mitutoyo). The percentage of the annual increment of tracheid length was calculated using a logarithmic formula 42 . Then, the boundary between juvenile and mature wood was determined as the point of 1% annual growth of the latewood tracheid length 42 .
Statistical analysis. Statistical analyses were conducted using R software 43 . Mean values of each physical and mechanical property were calculated by averaging the values of each radial position of harvested trees within a provenance. Due to failure to collect data during the static bending test, MOE and MOR were missing for one tree in Khentii. Using the mean values for each tree, the physical and mechanical properties among provenances (five trees in each provenance, total 25 trees) were evaluated with a one-way analysis of variance (ANOVA). Variances of populations were assumed as equal in the ANOVA test. In addition, a t-test was also performed to evaluate the differences in physical and mechanical properties between juvenile and mature wood, between OD and EOD, and between values of MOE and MOR at the 10th and 30th annual rings for each provenance (five trees) or all trees (25 trees from five provenances). To clarify the relationships between the measured properties, Pearson correlation coefficients were determined. Then, the test of no correlation was also applied.

Results
Physical and mechanical properties of wood. The AD values in the five provenances ranged from 0.62 to 0.68 g/cm 3 , while the OD values ranged from 0.58 to 0.65 g/cm 3 ( Table 2). The highest values were found in Khentii and the lowest values were found in Khuvsgul for both AD and OD. Cold-water extractive content ranged from 7.3 to 16.1% in the five provenances (Table 3). Radial variations of AD and OD were similar for the five sites: slightly lower values were found near the pith compared with those at other radial positions (Fig. 1).
The mean values of radial shrinkage per 1% moisture content change ranged from 0.16 to 0.19%, and tangential shrinkage per 1% moisture content change ranged from 0.30 to 0.36% among the five sites ( Table 2). The shrinkage in radial direction was almost constant from the pith to the bark, whereas the shrinkage in the tangential direction increased up to 4 cm from the pith and then remained constant around 0.3 to 0.4% (Fig. 1). Significant differences among provenances were found in shrinkage and cold-water extractive content, whereas no differences were found in AD, OD, and EOD (Tables 2 and 3). Table 4 shows the mean values of the mechanical properties in each provenance. The mean values of MOE, MOR, CS, and SS ranged from 7.03 to 9.51 GPa, 79.8 to 103.  (Fig. 2). Significant differences among provenances were found in all mechanical properties except for CS (Table 4). Estimated MOE and MOR values at the 10th and 30th annual rings from the pith are shown in Table 5. The mean values of MOE and MOR were 6.89 GPa and 78.5 MPa at the 10th annual ring from the pith and 9.60 GPa and 108.5 MPa at the 30th annual ring from the pith, respectively. Significant differences among provenances were found in MOE and MOR at the 30th annual ring from pith.  Table 2). The latewood tracheid length in all provenances increased up to about the 20th annual ring from the pith and then slightly increased toward the bark (Fig. 3). The boundary was determined as the point of 1% annual growth of the latewood tracheid length, according to the method described by Shiokura 42 . As shown in Table 6, the boundaries ranged from 17 to 24 annual rings from the pith among the provenances. Significant differences were found in tracheid length between the provenances ( Table 2). Mean values of the physical and mechanical properties in juvenile and mature wood are listed in Table 7. Significant differences were found in the mean values of the physical properties, tracheid length, and mechanical properties between juvenile and mature wood, except for SS. Significant among-provenance differences were found for all properties except for AD and OD in juvenile wood and CS in both juvenile and mature wood. Figure 4 shows the correlation coefficients for the physical and mechanical properties of three different wood types (juvenile wood, mature wood, and both types of wood). The mechanical properties had a strong correlation with each other for all wood types, except for SS. No significant correlation coefficients were found between RS or bending properties (MOE and MOR) and wood density (AD and OD), whereas EOD was significantly correlated with RS, MOE, or MOR in both mature wood and all wood types. However, CS and SS were significantly correlated with AD and OD in all wood types, except for SS and OD in juvenile wood. Table 2. Mean values and standard deviations of physical properties and tracheid length in each provenance. n number of trees, AD air-dry density, OD oven-dry density, RS shrinkage in radial direction per 1% moisture content change, TS shrinkage in tangential direction per 1% moisture content change, TL latewood tracheid length, SD standard deviation. Each value of physical properties and TL in each provenance was calculated by averaging the mean values of five trees. The mean value in each tree was calculated by averaging the values obtained from different radial positions. F-values and p-values were obtained with a one-way ANOVA among provenances.  Table 3. Mean values and standard deviations of oven-dry density after cold water extraction in each provenance. n number of trees, EOD oven-dry density after cold water extraction, SD standard deviation, Sig p-values obtained using a t-test between the OD listed in Table 2 and the EOD in each provenance or all sample trees. Each value of EOD and cold-water extractive content in each provenance was calculated by averaging the mean values of five trees. The mean value in each tree was calculated by averaging the values obtained from different radial positions. F-values and p-values were obtained with a one-way ANOVA among provenances. www.nature.com/scientificreports/

Discussion
Physical and mechanical properties of wood. The AD (0.62 to 0.68 g/cm 3 ) and OD (0.58 to 0.65 g/ cm 3 ) values obtained this study were similar with respect to basic density 33 in the same sample trees (Table 2), whereas our results for mean AD values were relatively higher than those for L. sibirica reported by Ishiguri et al. 27 and lower than those reported by Koizumi et al. 5 . Radial variations of AD and OD showed similar patterns to those reported by other researchers of L. sibirica 5 and L. kaempferi 16 . Meanwhile, Cáceres et al. 3 reported  www.nature.com/scientificreports/ an influence of extractives on density in L. kaempferi. They found that the hot-water extractive content of L. kaempferi varied between 2.9 to 6.9% among 20 provenances, suggesting that actual wood density might be about 5% lower than AD. As shown in Table 3, cold-water extractive content ranged from 7.3 to 16.1%, and the mean values of EOD (0.54 g/cm 3 , Table 2) were about 10% lower values compared to OD (0.61 g/cm 3 , Table 2). These results indicated that the effect of cold-or hot-water extractives on wood density might be greater in L. sibirica compared to other Larix species.  Table 6. Boundary between juvenile and mature wood based on radial variation of latewood tracheid length. www.nature.com/scientificreports/ Ishiguri et al. 27 reported that radial shrinkage at 1% moisture content change showed almost constant values from pith to bark, whereas tangential shrinkage increased up to 4 cm from pith and then became constant at around 0.3%. The mean values and radial variation patterns examined in this study for shrinkage in both the radial and tangential directions in L. sibirica were similar to those of L. sibirica examined by Ishiguri et al. 27 .
Although the tree ages varied, the mean values of MOE, MOR, and CS of the L. sibirica trees in the present study (Table 4) were similar to those found in a previous study for L. sibirica that grow naturally in Mongolia 27 but lower than those for L. sibirica that grow naturally in Russia 5 and higher than those for L. kaempferi planted in Japan 22,24 . The mean SS was higher than that of L. sibirica planted in Finland 9 and L. kaempferi planted in Japan 24 . In the radial variation, similar radial trends were found in L. sibirica that grow naturally in Mongolia 27 and in L. kaempferi planted in Japan 22 .
Based on the obtained results, the mean values of the physical and mechanical properties of L. sibirica collected from five different provenances in Mongolia are similar to those of L. sibirica and other Larix species found in other countries. Thus, wood resources from L. sibirica harvested in Mongolia can be used for similar purposes to other Larix species, such as construction materials.
Juvenile and mature wood. The boundary between juvenile and mature wood ranged from the 17th to 24th annual rings from the pith ( Table 6). The results were similar to those reported for L. kaempferi trees 17,22,42 . However, Ishiguri et al. 27 showed that juvenile wood might exist within 4 cm from the pith in L. sibirica. In the present study, the boundary was within 2 to 5 cm from the pith among the provenances, suggesting that juvenile  www.nature.com/scientificreports/ wood formation in L. sibirica trees that grow naturally in Mongolia is not only affected by tree age but also by growing conditions. We previously reported that mean values of annual ring width were 1.55, 2.47, 0.49, 1.86, and 1.74 for Khentii, Arkhangai, Zavkhan, Khuvsgul, and Selenge, respectively 33 . This result indicates that the radial growth rate was extremely slow in Zavkhan compared to other four provenances. Shiokura and Watanabe 28 reported that suppressed radial growth in the initial stage of tree growth resulted in prolonging the juvenile wood formation period in Picea jezoensis and Abies sachalinensis. Although significant differences among provenances were also found in annual ring number from the pith in the boundary between juvenile and mature wood (Table 6), the difference in the earliest (17th) and the latest (24th) annual ring number from the pith in the boundary was only 7 years. Thus, the radial growth rate in L. sibirica does not have a strong effect on the cambial age at which the www.nature.com/scientificreports/ production of mature wood cells begins. However, further research is needed to clarify the relationship between the radial growth rate and annual ring number from the pith in the boundary between juvenile and mature wood in this species. As shown in Table 7, significant differences between juvenile and mature wood were found in the mean values of physical properties, tracheid length, and mechanical properties, except for SS: the values of the physical and mechanical properties of juvenile wood were lower than those of mature wood. These lower values can be explained by shorter tracheid length and lower wood density. Similar results were obtained by several researchers of softwood species 17,22,24,28,29 . For example, Koizumi et al. 24 found that, in L. kaempferi, the mean MOE, MOR, CS, and SS values were 8.2 GPa, 93.3 MPa, 54.0 MPa, and 11.5 MPa in juvenile wood and 9.5 GPa, 97.2 MPa, 55.1 MPa, and 11.4 MPa in mature wood, respectively. Bao et al. 25 reported that the mechanical properties of juvenile wood were significantly lower than those of mature wood in Larix olgenis and L. kaempferi. We also found lower mechanical properties, basic density, and shorter latewood tracheid length of juvenile wood in 67-year-old L. kaempferi 22 . Thus, the presence of juvenile wood should be considered when utilizing wood resources of this species as construction materials requiring higher strength properties.
Correlation among physical and mechanical properties of wood. Figure 4 shows the correlation coefficients of the physical and mechanical properties of three different wood types (all types of wood, juvenile wood, and mature wood). In general, wood density is positively related to shrinkage in the radial and tangential directions 44 . The results of this study showed significant correlations between radial shrinkage at 1% moisture www.nature.com/scientificreports/ content and EOD in mature wood and all wood, suggesting that EOD can predict shrinkage in the radial direction in this species. Wood density is also positively correlated with many types of mechanical properties of wood 45,46 . CS was positively correlated with all types of wood densities measured in this study. The MOE and MOR in mature wood and all wood only exhibited a significant positive correlation with EOD. These results indicate that MOE and MOR values were correlated with wood substances without extractives, and these values in juvenile wood might be related to other properties, such as microfibril angle. Luostarinen and Heräjärvi 10 reported that water-soluble arabinogalactan contents were weakly correlated with SS in L. sibirica. SS was significantly correlated with AD, but not with EOD, suggesting that cold water-soluble extractives, such as arabinogalactan, might be affected on the SS in this species. Based on these results, strength properties (e.g., bending properties and compressive strength) can be estimated with each other and predicted by EOD. In addition, SS might be influenced by the presence of cold watersoluble extractives, such as arabinogalactan.
Among-provenance variations. Cáceres et al. 3 reported that significant among-provenance differences were not found in basic and oven-dry densities, whereas hot-water extractive content was significantly affected by provenances in L. kaempferi. We also previously demonstrated that no significant differences among provenances were found in the basic density of L. sibirica naturally grown in Mongolia 33 . Although the cold-water extractive content significantly differed among provenances in this study, all examined densities, such as AD, OD, and EOD, showed no significant differences among the five provenances (Tables 2 and 3 www.nature.com/scientificreports/ wood density might not vary greatly among provenances. Thus, it can be concluded that genetic variations in relation to wood density might be small in L. sibirica trees naturally grown in Mongolia. In half-sib families of P. jezoensis, F-values obtained by an ANOVA test for AD, MOE, and MOR among families gradually decreased from juvenile to mature wood 47 . In addition, Kumar et al. 48 reported that estimates of narrow-sense heritability for MOE were generally higher in the corewood than in the outer wood in Pinus radiata. For Larix species, significant differences in wood density, CS, and SS but not in MOE and MOR were found in outer wood among 23 provenances for 31-year-old L. kaempferi 24 . Thus, genetic variations in the physical and mechanical properties of juvenile wood were higher than in mature wood in many softwood species. Significant differences were also found in most of the mechanical properties among provenances, except for CS (Table 4). In addition, significant differences were found in all examined physical and mechanical properties except for CS in mature wood among the five provenances, while no differences were found in juvenile wood for many properties (Table 7). Similar results were obtained in estimated MOE and MOR values at the 10th and 30th annual rings from the pith: no significant among-provenance variations were found in MOE and MOR at the 10th annual ring from the pith, but significant differences were found in the 30th annual ring from the pith ( Table 5). Although the environmental conditions in the five provenances were not the same, the genetic variations in physical and mechanical properties among provenances were large in mature wood compared to juvenile wood for L. sibirica grown naturally in Mongolia. Further research is needed to clarify the genetic factors of the physical and mechanical properties of wood in L. sibirica.
Based on the results, there are significant among-provenance differences in the physical and mechanical properties of wood, especially in mature wood, in L. sibirica grown naturally in Mongolia. The physical and mechanical properties of wood in this species, especially in mature wood, can be improved by establishing tree breeding programs: families or clones with higher mechanical properties can be produced to achieve sustainable forestry in Mongolia.

conclusions
This study examined physical and mechanical properties of wood and their geographic variations in L. sibirica trees that grow naturally in Mongolia. Significant differences were found in RS, TS, tracheid length, MOE, MOR, and SS among five provenances. Differences in wood densities, such as AD, OD, and EOD, among the provenances were not significant. However, a significant difference was found in cold-water extractives between the provenances. The results show that the physical and mechanical properties of wood in this species can be improved by establishing appropriate tree breeding programs. In addition, the physical and mechanical properties of wood varied significantly between juvenile and mature wood. Based on the results, it is suggested that the identification of juvenile and mature wood is important for utilizing the wood of this species. In addition, identifying the extracted wood density in L. sibirica is also important for the practical use of the wood of this species.

Data availability
The data sets used in the present study can be obtained upon request from the corresponding author.