Preliminary evaluation of anatomical characteristics of four common Mongolian softwoods

Abstract To effectively and sustainably utilize wood resources from boreal forests in Mongolia, anatomical characteristics, tracheid morphology, cell proportion, annual ring width, and latewood percentage were preliminary determined in Pinus sylvestris, Pinus sibirica, Picea obovata, and Larix sibirica trees naturally growing in Mongolia. Based on the observation, the anatomical characteristics of four common Mongolian softwoods were the same as those previously observed in the same species or the same genus species. Based on the parameters of the Gompertz functions for annual ring width, silvicultural management, such as thinning timing and harvesting age, should be considered depending on the species when the plantation is established. The results of the model selection for relationships between latewood percentage and basic density indicated that the increase ratio of basic density corresponded to an increase of latewood percentage is almost the same irrespective of species, although there are species-specific values of basic density corresponding to specific latewood percentages. The results obtained in the present study contribute effective and sustainable utilization of wood resources from Mongolian forestry.


Introduction
Natural forest of Mongolia is mainly classified into two types: boreal forests and saxaul forests (FRDC 2020). Boreal forests are distributed in northern Mongolia and occupy 84.9% (9,911,327 ha) of the total forest area (11,674,438 ha) (FRDC 2020). Saxaul forests are distributed in the deserts and steppes in southern Mongolia and occupy 15.1% (1,763,111 ha) of the total forest area (FRDC 2020). A total of 11 coniferous species, including Abies, Picea, Pinus, Larix, and Juniperus, are naturally distributed in Mongolia (Damdinsuren 2014). Among the 11 species, 4 species-Pinus sylvestris L., Pinus sibirica Du Tour., Picea obovata Ledeb., and Larix sibirica Ledeb.-are known as important forestry species in the country (Tsogtbaatar 2004;Altrell 2019;Sukhbaatar et al. 2020). Stock volume of the four species reaches around 93.2% of the country's total stock volume (FRDC 2020). In 1990, Mongolia made a dramatical change in its political and economical systems, the basic concept of which is the transition from a single-party political system to a democratic form of society (Jamsran 2004). As the results, the annual logging volume in Mongolia decreased from 2.2 million m 3 in 1980s, 0.86 million m 3 in 1992, to 0.5 million m 3 in 2000 (Nyamjav et al. 2007). However, in recent decades, the production of industrial roundwood has gradually increased from 49,000 m 3 in 2010 to 162,000 m 3 in 2019 (FAO 2014(FAO , 2019. Thus, there is some possibility of increasing demand for wood resources in Mongolia. To promote the sustainable forestry using natural resources, efficient utilization of wood resources is needed for Mongolian forestry. In softwoods, anatomical characteristics, such as tracheid diameter, tracheid wall thickness, and latewood percentage, are closely related to the physical and mechanical properties of wood (Bendtsen and Senft 1986;Takata et al. 1992;Kang 1993;Koizumi et al. 2005;Peltola et al. 2009;Gryc et al. 2011). For example, Koizumi et al. (2005) reported that in Larix kaempferi, the effect of latewood percentage on average wood density was greatest in outer wood (mature wood). In addition, wood anatomical characteristics are also related to liquid or gas permeabilities which affect processibility in wood drying and wood preservation (Milita et al. 1995;Fujii et al. 1997;Matsumura et al. 1998;Usta and Hale 2006). Milita et al. (1995) reported that factors such as the presence of juvenile or mature wood and between-tree variability have a greater effect on permeability than does wood density in plantation grown Pinus taeda. For efficient wood utilization, anatomical characteristics concerning the physical and mechanical properties of wood should be clarified.
A sigmoid or S-type curve resembles trends in the life cycle of trees (Salas-Eljatib et al. 2021). Current annual increment (CAI) and mean annual increment (MAI) can be determined by the sigmoid models, such as the logistic model, the Gompertz model, and others (Salas-Eljatib et al. 2021). These annual increment values are useful for determining the suitable rotation age with respect to volume yield (Guzm an et al. 2012;Resende et al. 2021). Guzm an et al. (2012) reported that the rotation length that maximized wood production (maximum sustainable yield) was 27.5 years for the best site and 33.5 years for the poorest site of Pinus radiata plantations based on the MAI value. Thus, the optimum age to apply thinning and harvesting with respect to volume yield can be estimated based on these annual increment values in radial growth.
We have investigated wood properties in Mongolian softwoods Tumenjargal et al. 2018Tumenjargal et al. , 2020Sarkhad et al. 2020Sarkhad et al. , 2022. In a previous paper, we clarified differences in wood properties between juvenile wood and mature wood using linear or nonlinear mixed-effects modeling for radial variations of wood properties in four common Mongolian softwood, P. sylvestris, P. sibirica, P. obovata, and L. sibirica (Sarkhad et al. 2022). However, basic information on anatomical characteristics and the effect of these characteristics on physical properties were still unclear for these Mongolian softwoods. Wood anatomical characteristics in relation to dendrochronology and wood properties have been researched on P. sylvestris and L. sibirica (Schweingruber 1990;Koizumi et al. 2003;Benkova and Schweingruber 2004;Pritzkow et al. 2014;Fonti and Babushkina 2016;Khansaritoreh et al. 2018) because these species are naturally distributed in Europe and were utilized as plantation species (Farjon 2017). However, information about wood properties and anatomical characteristics in the other two species, P. sibirica and P. obovata are not so much, as these species are naturally distributed in limited countries, such as Russia, Mongolia, and others (Farjon 2017). Thus, additional information on anatomical characteristics and their impacts on other wood properties should be investigated for the Mongolian softwoods harvested from natural forests.
To effectively and sustainably utilize wood resources in Mongolia, anatomical characteristics (tracheid morphology, cell proportion, annual ring width, and latewood percentage) were preliminary investigated for four common Mongolia softwoods in this study, although number of sample trees were limited.

Materials
Four common Mongolian softwoods-P. sylvestris, P. sibirica, P. obovata, and L. sibirica-were collected from Mandal, Selenge, Mongolia (48 49'N,106 53'E and 48 41'N,106 38'E; Figure 1). Figure 1 also shows the mean monthly temperature and precipitation at the nearest meteorological station located in Mandal, Selenge, Mongolia. Pinus sylvestris trees were harvested from a natural mixed forest with Betula platyphylla. Remained other three species were collected from another mixed forest with B. platyphylla. The altitude of both stands was about 1120 m above sea level. Unfortunately, detailed information, such as stand density, soil conditions, and others was unknown. In each natural forest, we selected sample trees having 20 to 30 cm in stem diameter at 1.3 m above the ground, because the size is commercial harvesting size in Mongolia. Five trees in each species were harvested, and 3 cm in thickness disks were collected at 1.3 m above the ground. Table 1 shows the growth characteristics, annual ring number, basic density, and boundary cambial age between juvenile and mature wood of the sample trees (Sarkhad et al. 2020(Sarkhad et al. , 2022. After collecting the disks, bark-to-bark radial strips with a pith (5 cm in width) were prepared for the following experiments. To avoid the compression wood, the radial strips were collected from positions that did not include eccentric growth and dark-colored earlywood (typical appearance of compression wood in softwoods).

Anatomical characteristics
Generally, wood properties in softwoods can be divided into two parts: juvenile and mature wood (Bendtsen and Senft 1986;Zobel and van Buijtenen 1989;Moore and Cown 2017;Sarkhad et al. 2022). The boundary between juvenile and mature wood normally was around the 15th to 20th annual ring from the pith (Shiokura 1982;Bendtsen and Senft 1986;Zobel and van Buijtenen 1989;Moore and Cown 2017;Sarkhad et al. 2022). In fact, the boundary between juvenile and mature wood determined by radial variation modeling of tracheid length ranged from 20th to 24th annual ring from the pith in the same sample trees for all four tree species of this study (Table 1, Sarkhad et al. 2022). Thus, in this study, cell morphology and proportion were determined at the 10th and 40th annual rings from the pith as the representatives of juvenile and mature wood, respectively. Small wood blocks were collected from the 10th and 40th annual rings from the pith of the radial strips collected at 1.3 m above the ground. Transverse sections (20 mm in thickness) were prepared using the sliding microtome (REM-710, Yamatokohki, Saitama, Japan). These transverse sections were used for the following experiments.
The diameter and wall thickness of tracheids in earlywood and latewood were determined as tracheid morphology. The transverse sections without dehydration were put on grass slides and then mounted with 25% glycerin and cover slips. Transverse images were captured using a digital camera (DS-2210, Sato Shouji Inc., Kawasaki, Japan) and a microscope (BX51, Olympus, Tokyo, Japan). Diameters in radial and tangential directions and tangential wall thickness were measured for 20 tracheids at each annual ring position and wood type (earlywood and latewood) by ImageJ (National Institutes of Health, MD, USA). Tracheid diameter was also calculated by averaging diameters in the radial and tangential directions. The boundary between earlywood and latewood was determined according to Mork's definition (Denne 1989).
Cell proportions were determined using the point counting method (Denne and Hale 1999). Transverse sections were stained with safranin, and dehydrated by graded ethanol. Then, the sections were then dipped into xylene and mounted by biolite (Ohken, Tokyo, Japan). Transverse images with current annual rings were taken using DS-2210 and BX51. Mesh images (100 mm interval) were overlapped to the transverse images by ImageJ, and each point was counted and classified into the following cell types: tracheid, ray parenchyma, and axial intercellular (resin) canal in each earlywood and latewood. About 400 to 1000 points in a photomicrograph of each species were measured in an annual ring of each species. Cell percentage was calculated by dividing the number of grids in each cell type by the total number of grids.

Annual ring width and latewood percentage
Annual ring and latewood widths from the pith to bark were measured in all annual rings of both directions of the strips. The image data (1200 dpi) of the transverse sections of the strips were captured on a personal computer equipped with a scanner (GT-9300UF, EPSON, Nagano, Japan). Using the scanned images, the annual ring and latewood widths from pith to bark were measured at each annual ring using ImageJ. Annual ring and latewood widths of each tree were expressed as mean values of two directions. Table 1. Growth characteristics, annual ring number, basic density, and boundary cambial age between juvenile and mature wood of the sample trees (Sarkhad et al. 2020(Sarkhad et al. , 2022 Note: D: stem diameter at 1.3 m above the ground; TH: tree height; ARN: annual ring number at 1.3 m above the ground; BD: basic density; Min.: minimum; Max.: maximum. Mean and standard deviation (SD) for each species were calculated from the data of five trees in each species. Boundary cambial age between juvenile and mature wood was determined by the results of nonlinear mixed-effects model for radial variations of latewood tracheid length (Sarkhad et al. 2022).
Boundary between earlywood and latewood was determined by the color differences observed by naked eyes.

Data analysis
Statistical analysis was conducted using R software (R Core Team 2020). A paired t-test was used for detecting the differences in cell morphology and cell proportions between the 10th and 40th annual rings from pith or earlywood and latewood. Also, correlations between data up to the 20th annual ring from the pith and data of the 21st to 40th annual ring from the pith were evaluated in annual ring width and latewood percentage using Pearson's correlation coefficients. Radial variations of cumulative annual ring width and latewood percentage were evaluated by developing linear and nonlinear mixed-effect models using the lmer function in lme4 packages (Bates et al. 2015) and the nlme function in nlme package (Pinheiro and Bates 2000).
To evaluate the radial variation of cumulative annual ring width, the model (1) was developed based on the Gompertz function ( Table 2). The developed models were selected by the value of the Akaike Information Criterion (AIC, Akaike 1998). Using the estimated parameters in the selected models, MAI and CAI were calculated.
Radial variations in latewood percentage were evaluated using the following models (2 and 3): where LWP ij is the latewood percentage of the ith annual ring number from the pith of the jth individual tree, ARN ij is the ith annual ring number from the pith of the jth individual tree, b 0 , b 1 , c 0 , and c 1 are the fixed-effect parameters, Tree j is the random-effect parameter of the jth individual tree, and e ij is the residuals. The developed models were selected by the value of the AIC.
To evaluate the differences in the relationships between basic density and latewood percentage amongtree species, the following mixed-effects models (4) were developed: where BD ij is the basic density of the ith individual tree of the jth species, Species j is random-effect parameters of the jth species, LWP ij is the latewood percentage of the ith individual tree within the jth species, d 0 and d 1 are fixed-effect parameters, and e ij is residuals. The basic density was measured at 10 years intervals from the pith (Sarkhad et al. 2022). Thus, latewood percentage values were averaged at each 10 years intervals from the pith.

Results and discussion
Anatomical characteristics Table 3 shows the tracheid morphologies at the 10th and 40th annual ring from pith. Earlywood and latewood tracheid diameters in both radial and tangential directions at the 40th annual ring from the pith exceeded those at the 10th annual ring from the pith in all species, except for latewood of L. sibirica. Tracheid wall thickness in earlywood and latewood at the 10th annual ring from the pith was almost the same as that at the 40th annual ring from the pith, except for P. obovata (Figure 2). When compared to earlywood, the tracheid diameter in both radial positions decreased in latewood, and tracheid wall thickness increased in latewood. However, no significant differences in tracheid wall thickness were found between earlywood and latewood in P. sibirica. Transition from earlywood to latewood was gradual in P. sibirica (Figure 2). These tracheid morphological features in P. sibirica are very similar to those of other five-needle pine, such as Pinus pentaphylla grown in Japan and Pinus koraiensis grown in Korea (Wood Technological Association of Japan 1984; Kang 1993; Lee et al. 1997).
Out of the four tested species, tracheid morphologies were characterized by larger diameter and thicker wall in L. sibirica, and smaller diameter and thinner wall in P. sibirica. Also, mean values of tracheid morphologies in four Mongolian softwoods were in the range of those of the same species or the same genus species reported by other researchers (Wood Technological Association of Japan 1984; Kang 1993;Ilvessalo-Pf€ affli 1995).
In the four species, cell proportions in tracheids, rays, and intercellular canals ranged from 83.6 to 90.6%, 8.8 to 15.9%, and 0.3 to 0.8%, respectively (Table 4). There were no differences between radial Note: Equation (1-i), fixed-effects model; Equations (1-ii) to (1-iv), mixed-effects model; CRW i , cumulative annual ring number at the ith annual ring number from pith; CRW ij , cumulative annual ring number at the ith annual ring number from pith in the jth individual tree; ARN i , the ith annual ring number from pith; ARN ij , the ith annual ring number from pith in the jth individual tree; a 0 , a 1 , and a 2 , fixed-effect parameters; Tree 0j , Tree 1j , and Tree 2j , random-effect parameters of a 0 , a 1 , and a 2 at the individual tree level; e i and e ij , residuals. Bold value indicates minimum AIC among four developed models in each species.
positions (10th and 40th annual rings from the pith), except for ray proportions in P. sibirica. Lower tracheid proportions and higher ray proportions were obtained in this study compared to those of previous research (Wood Technological Association of Japan 1984; Ilvessalo-Pf€ affli 1995).
Annual ring width Figure 3 shows the radial variations of the annual ring width. Mean annual ring widths were 1.44, 1.85, 2.06, and 2.04 mm in P. sylvestris, P. sibirica, P. obovata, and L. sibirica, respectively (Table 5). Among the four species, P. obovata and L. sibirica showed larger mean annual ring widths. Based on the model selection for radial variation of cumulative annual ring width, selected models included random effects of individual trees in slope (Equation (1-iv)) for P. sylvestris, in the start position of the curve (Equation (1-iii)) for P. sibirica, and in asymptotic value (Equation (1-ii)) for P. obovata and L. sibirica (Table 6). The annual ring number from the pith showing maximum MAI and CAI were determined using the fixed-effect parameters in the selected models in each species (Figure 4). As a result, the smallest annual ring number showing maximum CAI and MAI was found in L. sibirica (about the 20th annual ring), followed by P. obovata, P. sibirica, and P. sylvestris (about 60 to 70th annual ring) (Table 7). Thus, appropriate silvicultural management, such as thinning timing, harvesting age, and others, should be considered depending on the species when the plantation is established. Figure 5 shows the relationships between the mean values of the annual ring width from pith to 20th annual ring (position I in Figure 5) and the mean values from 21st to 40th (position II in Figure 5). A significant correlation coefficient was not found, suggesting that trees with faster radial growth in the relatively early stage of growth did not always show a faster radial growth rate after 20 years of growth. Takata et al. (1992) examined the annual ring width of L. kaempferi from 25 seed provenances, and they reported that a significant negative correlation (r ¼ À0.18) was found between core wood (juvenile wood) and outer wood (mature wood) of annual ring width. In L. sibirica naturally growing in Mongolia, Tumenjargal et al. (2018) reported that a significant positive correlation was found between mean values of annual ring width from 1st to 10th annual rings and those from 11th to 20th annual rings, but mean values from 1st to 10th annual rings were insignificantly correlated with those from 21st to 30th, and from 31st to  40th. This study's results parallel those reported by Tumenjargal et al. (2018).

Latewood percentage
Pinus sylvestris and L. sibirica showed relatively higher latewood percentages compared to the other two species (Table 4). Compared to the same species or the same genus species, the mean values obtained in this study showed similar values (Takata et al. 1992;Koizumi et al. 2005;Peltola et al. 2009;Gryc et al. 2011;Tanabe et al. 2016) or relatively lower values (Kang 1993;Karlman et al. 2005;Gryc et al. 2011). Figure 6 shows the radial variations of latewood percentage. While logarithmic formula (Equation (3)) was fitted on P. sylvestris and L. sibirica, linear formula (Equation (2)) was adapted to P. sibirica and P. obovata, suggesting that latewood percentage increased from pith to bark and then stabilized in P. sylvestris and L. sibirica, while it gradually increased from pith to bark side in P. sibirica and P. obovata. However, value of slope in the linear formula in P. sibirica showed almost zero (0.0048), indicating that latewood percentage was almost constant from pith to bark. Similar radial patterns in latewood percentages were recognized in the same genus species (Kang 1993;Karlman et al. 2005;Koizumi et al. 2005;Tanabe et al. 2016). Figure 5 shows the relationships between mean values of latewood percentages from pith to 20th annual    ring (position I in Figure 5) and mean values from 21st to 40th (position II in Figure 5). A significant positive correlation was found, suggesting that trees with a higher latewood percentage around the pith area have a higher latewood percentage wood after 20 years of growth. The results also indicate that the selection of trees with higher latewood percentages may be possible when tree breeding programs for wood quality are established for these four Mongolian softwood species. Effects of tree species on relationships between latewood percentage and basic density were evaluated by a mixed-effects model. The model with random slope and intercept (Table 8, Equation (4-i)) was not converged. The AIC value in the model with random intercept (Equation (4-iii)) was higher than that in the model with random slope (Equation (4-ii)), suggesting that the best model among the three developed models was the model with only random intercept (Equation (4-iii)). The results of the model selection indicated that the increase ratio of basic density corresponded to an increase of latewood percentage is almost the same irrespective of species, although there are speciesspecific values of basic density corresponding to certain latewood percentages (Figure 7). If the latewood percentage was the same in four species, L. sibirica showed the highest basic density, followed by P. sibirica, P. sylvestris, and P. obovata (Table 8). Comparison of two Pinus species, P. sylvestris has a lower speciesspecific value of basic density (Figure 7) but has a higher latewood percentage (Table 5) rather than that in P. sibirica. The results might be closely related to the physiological phenomena of trees. Further research is needed to clarify the differences in the relationships Table 6. Estimated values of the fixed-and random-effect parameters of the selected radial growth model.   (Tables 6 and 7). The curve with black color based on the only fixed-effect parameters (Table 6). Dashed lines in vertical direction in CAI and MAI figures indicate annual ring number from pith showing maximum CAI and MAI that showed after CAI reached maximum (Table 7) based on the only fixed-effect parameters (Table 6).
between latewood percentage and basic density at the species level.

Conclusions
In this study, anatomical characteristics were preliminary investigated in four common Mongolian softwoods-P. sylvestris, P. sibirica, P. obovata, and L. sibirica, although the number of samples were limited.
Microscopic features of the four species paralleled those obtained in the same species or the same genus species by the previous researchers. Tracheid morphology in examined four softwoods was also similar to that in the same species or the same genus species. These results may suggest that wood from these four Mongolian softwoods can be utilized for the same purposes in the same species or the same genus species found in the other countries. Based on the Gompertz formula for annual ring width, annual ring number from pith showing the maximum values of the CAI and MAI for P. sylvestris, P. sibirica, P. obovata, and L. sibirica ranged from 59 to 68, 32 to 48, 24 to 34, and 17 to 23 years, respectively. The results suggested that appropriate silvicultural management, such as thinning timing, harvesting age, and others, should be considered depending on the species when the plantation is established. No significant correlation coefficient of mean annual ring width was found between the pith area and 21st to 40th annual ring from the pith, suggesting that trees with faster radial growth in the relatively early stage of growth did not always show a faster radial growth rate after 20 years of growth. On the other hand, a positive correlation in latewood percentage was found between two different positions. Thus, trees with a higher latewood percentage around the pith area have a higher latewood percentage wood after 20 years of growth. The results of the model selection for explaining the relationships between latewood percentage and basic density indicated that the increase ratio of basic density corresponded to an increase of latewood percentage is almost the same irrespective of species, although there are species-specific values of basic density corresponding to specific latewood percentages. The results obtained in the present study can contribute to establishing sustainable forestry using  natural forests and plantations in boreal regions and Mongolia.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
Part of this research was financially supported by the Higher Engineering Education Development Project, implemented by the Ministry of Education, Culture, Science, and Sports, Mongolia. Nonlinear mixed-effect model (logarithmic formula; LWP ij ¼ c 0 ln (ARN ij ) þ c 1 þ Tree j þ e ij ) was adapted to Pinus sylvestris and Larix sibirica, whereas radial variations of Pinus sibirica and Picea obovata were explained by linear mixed-effect model (LWP ij ¼ b 0 ARN ij þ b 1 þ Tree j þ e ij ). Figure 7. Differences in the relationship between latewood percentage and basic density among-tree species. Regression lines are based on the selected model (Equation (4-iii)) with fixed-and random-effect parameters at the species level (Table 8).