About 35 wood species of the genus Carpinus of the Betulaceae are widely distributed in Europe, eastern Asia, and North and Central America. In addition, the highest density of species diversity has been reported in China ( Kiaei, 2011 ; Kiaei, 2012 ). Hornbeam is one of a native diffuse-porous hardwood species in Caspian forests. Besides, it grows in a mixture of oak and beech, as well as in some areas with Parrotia persica ( Abdi et al. , 2009 ). It is classified as a medium-density hardwood, hence, it is considered as a wood of semi-hard to hard, high volumetric shrinkage, and with low cleavage strength properties ( Moosavi et al. , 2017 ; Kiaei, 2012 ). Hornbeam, as one of the most important wood species in the paper-making industry, has a longer fiber length than other wooden species ( Kiaei, 2011 ).

Physiographic factors (altitude, aspect, slope, and side slope), edaphic factors (soil characteristics), climatic factors (light intensity, temperature, air humidity, precipitation, and wind), and biotic factors (humans, animals, plants, and microorganisms) can affect different wood properties ( Usta et al. , 2014 ; Kaygin et al. , 2016 ; Topaloglu et al. , 2016 ). There are various reports about the effects of ecological factors on wood properties mostly related to the relationship between altitude and wood properties. Many researchers, such as Noshiro et al. (1994) for Alnus nepalensis , Noshiro et al. (1995) for Rhododendron, and Yılmaz et al. (2008) for Quercus pontica , have reported that there are significant relationships between wood anatomical properties and ecological factors. On the other hand, there are reports indicating no significant relationships between the wood anatomical properties and altitude, including Dodonae aviscosa ( Liu & Noshiro, 2003 ), the genus Castanopsis ( Pande et al. , 2005 ), and Buddleja cordata ( Aguilar-Rodriguez et al. , 2006 ).

Many studies have focused on understanding the relationship between wood density and altitude. Govorcin et al. (2003) stated that wood density of Fagus sylvatica L. decreased with increasing altitude. Barij et al. , (2007) for Quercus pubescens and Kiaei (2011 , 2012 ) for Carpinus betulus reported that wood density increased by increasing the altitude. Hernandez & Restrepo (1995) stated that there was no change in the wood density of Alnus acuminata by increasing the altitude. Topaloglu et al. , (2016) reported that trees growing at altitudes of 400-600 m and 0-200 m had the highest and lowest density values, respectively. Besides, there are very few studies about the impact of altitude on the wood mechanical properties. Barij et al. (2007) found that the compression strength of Quercus pubescens increased by increasing the altitude. Kahveci (2012) reported that bending strength, modulus of elasticity, compressive strength parallel to grain, and dynamic bending strength of Alnus glutinosa decreased by increasing the altitude.

The main objectives of this study were (a) to in­vestigate variations of wood dry density, fiber dimen­sions and morphological properties including the coefficients of Runkel, flexibility, and slenderness of hornbeam ( Carpinus betulus ) at six different altitudes, (b) to examine the relationship between wood properties and site condition.

Eighteen mature and healthy trees (without reaction wood, decay, and insect damage or fungal infection) were clear-selected and harvested at six altitudes above sea levels (300, 500, 700, 900, 1100, and 1300m) in the northern forests of Iran (Nowshahr city, Mazandaran province). The main dimensions of the tree and site characteristics are listed in Table 1.

Disks were cut from the stem diameter at breast height (DBH = 1.30 m aboveground). Sampling was then performed after harvesting and removing pith and juvenile wood. The age demarcation point between juvenile and mature wood was estimated at about 19 years for all studied altitude levels ( Kiaei, 2006 ). The physical and morphological properties were measured using the specified procedure of ISO 3131 standard (wood-determination of density for physical and mechanical tests-international standard, 1975) and Franklin method (1945) , respectively. Site conditions at higher altitudes are very harsh for the growth of hornbeam trees due to the suppressed species. Therefore, the height and diameter of the trees are lower at higher than other altitudes.

Wood dry density (WDD)

ANOVA revealed a highly significant effect of al­titudes on the WDD (P ˂ 0.05, Table 2). The lowest and highest WDD values were measured at altitudes of 300 m and 1300 m, respectively. Average values of WDD (698 kg/m ³ ) and CV (3.10%) were obtained for hornbeam wood (Table 3). The lowest variability (CV = 2.5%) and the highest variability (CV = 3.47%) of the WDD were measured at altitudes of 1100 m and 700 m, respectively.

Fiber length (FL)

ANOVA showed that in addition to FL regular variations at various altitude levels, the effect of altitude levels on FL was highly significant (P ˂ 0.05, Table 2). In addition, the value of FL decreased with increasing altitudes (Table 3). The average values of FL and CV were 1.42 mm and 11.67%, respectively.

Fiber diameter (FD)

ANOVA results indicated significant differences in fiber diameter among the altitude levels (P ˂ 0.05, Table 2). The average FD decreased with increasing altitude at a CV of 10.84% (Table 3). The mean of FD was 25.58 μm in the six altitudes.

Cell wall thickness (CWT)

According to ANOVA results, the altitude sig­nificantly affected the CWT (P ˂ 0.05, Table 2), with the highest and lowest values measured at 1300 m and 300 m, respectively, and an average of 5.72 μm and a CV of 15.23% (Table 3).

Morphological coefficients

Slenderness coefficient (SC)

As shown by ANOVA results, significant effects of altitude levels were observed on the SC with an average of 55.55 and a CV of 13.14%. Additionally, the highest and lowest SC was measured at 500 m and 1300 m, respectively (Fig. 1).

Runkel coefficient (RC)

RC represents the paper's resistance to rupture, which is obtained from the thickness of two cell walls divided by the fiber diameter. ANOVA results illustrated sig­nificant effects of altitude levels on the RC with an average of 0.93, and a CV of 5.37%. The lowest and highest RC values were found at 300 m and 1300 m, respectively. Moreover, the values of RC increased with increasing altitudes (Fig. 2).

Flexibility coefficient (FC)

The results of ANOVA revealed significant effects of altitude levels on the FC, which is the ratio of the lumen diameter divided by the fiber diameter. FC averaged 54.04% with a CV of 12.43%. Furthermore, the value of FC decreased with increasing altitudes. The highest and lowest FC values were measured at 300m and 1300 m, respectively (Fig. 3).

Relationship among studied variables

The relationship between the studied wood pro­perties (WDD, FL, FD, CWT, FC, RC, and SC) with site conditions (such as temperature, precipitation, crown canopy, and understory herb layer) and tree main dimensions (such as tree height, tree diameter at breast height, and tree age) were investigated by the Pearson matrix correlation (Table 4) and forward stepwise re­gression (Table 5).

Results of Pearson matrix correlation indicated that the temperature, crown canopy, tree height and DBH (diameter at breast height) had significant relationships with the studied wood properties (except SC). A correlation coefficient of above 80% was measured between the environmental variables and the studied wood properties (except SC).

A correlation coefficient of less than 72% was obtained between the understory herb layer and the studied wood properties (WDD, FL, FD, CWT, RC, FC, and SC), while a correlation of over 72% was found between temperature, crown canopy, and tree diameter and height with the studied wood properties. Therefore, it can be concluded that the relationship between the understory herb layer with the studied wood properties was lower than the other environmental variables.

The results illustrated that there was a significant relationship between fiber length and fiber diameter with precipitation rate. The precipitation effect on other wood properties was not significant. Furthermore, correlation coefficients between the fiber length and fiber diameter with precipitation rate were measured more than 87%, while the relationship between pre­cipitation and the other wood properties was less than 79%.

There were significant relationships between tree's age and WDD, CWT, and RC. Besides, a correlation coefficient of more than 82% was measured between tree's age and the mentioned wood properties. The tree's age effect on the other wood properties was not significant.

The results indicated no significant differences between all site conditions/tree main dimensions with the SC, with a correlation coefficient of less than 79.6% between all environmental conditions and FC. As a result, in all of relationships, the lowest and highest of correlation coefficients were measured between the understory herb layer and CWT (less than 11%), as well as the temperature and Runkel coefficients (about 98.8%), respectively.

Multiple forward stepwise correlation analysis sho­wed that the wood properties could be explained by the model: Y= a + b*β ₁ + c*β ₂ , which presented a high correlation coefficient (R > 0.96) as well. The wood properties were affected by one environmental variable only (site conditions/tree main dimensions), except the FD that was corrected to two variables (Table 5). In addition, more than 96% of the variations in wood dry density, CWT, and RC depended on the ambient temperature. About 92% of fiber length variation was related to the tree DBH. The FD variations were corrected with tree diameter (84%) and tree age (13%). About 97% of FC variations were related to the crown canopy. As a result, there were no relationships between the site conditions/tree main dimensions with SC.

Wood density is one of the most important wood qualities in softwood and hardwood species. It is affected by cell wall thickness, cell diameter, ratio of earlywood to latewood, and chemical content of the wood ( Zobel & van Buijtenen, 1989 ). In the present study, we showed that the effect of altitude levels was significant on wood dry density of hornbeam wood. Also, the average of WDD was increased by altitude increasing. Similar results were previously reported by Barij et al. , (2007) for Quercus pubescens , Topaloglu et al. , (2016) for oriental beech, & Kiaei (2011 , 2012 ) for C. betulus .

The average WDD of hornbeam wood at the studied altitudes (698 kg/m ³ ) was lower than that in Turkey ( Gunduz et al. , 2009 ; 794 kg/m ³ ) and in the Guilan (732 kg/m ³ , Golbabaei et al. , 2004 ) and Golestan (717 kg/m ³ , Golbabaei et al. , 2004 ) sites (Iran), but it was higher than that of Mazandaran site (Iran) (688 kg/m ³ , Golbabaei et al. , 2004 ).

The influences of wood fiber dimensions and their derived values (slenderness ratio, flexibility coefficient, and Runkel ratio) are well described on pulp and paper mechanical properties. Wood fibers are one of the most important factors in the production of pulp and paper, or the fiberboards. Kellogg & Thykeson (1975) and Matolcsy (1975) also reported the importance of fiber dimensions in predicting wood pulp mechanical properties. It has been reported that the morphological features of fiber are important because they determine the suitability of lignocellulosic materials prior to production ( Ververis et al. , 2004 ; Ona et al. , 2001 ). The altitude levels affected significantly the FL, FD, and CWT of C. betulus wood. The average of FL and FD decreased while CWT increased with increasing altitude. Similar results were reported by many re­se­archers for Alnus nepalensis ( Noshiro et al. , 1994 ), Rhododendron ( Noshiro et al. , 1995 ), and Quercus pontica ( Yılmaz et al. , 2008 ).

The fibers were classified into three groups. The first group was considered as short fibers, such as some hardwood species (length > 0.9 mm). The se­­cond group had an average length of 0.9-1.9 mm. The third group consisted of fibers with a length of more than 1.9 mm ( Salehi, 2001 ). Furthermore, the results showed that the average fiber length of the hornbeam wood was equal to 1.2-1.8 mm. therefore it is concluded that the fibers of current study are classified in the second group. Also, the average fiber length (1.42mm) at the studied altitudes was similar to that of Turkish hornbeam wood (1.49 mm; Tank, 1978 ).

The average fiber diameter of hornbeam wood in studied altitudes was about 25.58 μm which is located in the normal range compared with hardwood fibers (20-40 μm: Atchison, 1987 ; San et al. , 2016 ), and is also higher than that of Turkish hornbeam wood (21.93 μm: Tank, 1978 ).

The cell wall thickness (CWT) of the fiber affects the strength of individual fibers. It is known that the paper made from a pulp prepared by very thin cell wall fibers has a very low tear resistance. The fibers with very thick cell wall cause low resistance and high-volume papers because they do not properly flatten at the time of sheet formation ( Seth & Page, 1988 ). Multiple reports indicated that there were positive relationships between wood density and CWT in many species. Our results showed that the lowest WDD and the thinnest CWT of the fibers, as well as the highest WDD and the thickest CWT of the fibers, were measured at 300 m and 1300 m, respectively. The decreased density at low altitudes may be related to both reduced CWT of the fibers and the cellulose content ( Topaloglu et al. , 2016 ). Denne & Hale (1999) observed that trees with lower mean density had thinner fiber walls and wider vessel lumens than those with higher mean density. The avera­ge CWT at the studied altitudes was 5.72 μm, which is close to that of Turkish hornbeam wood (CWT: 5.85 μm).

Slenderness coefficient (SC) shows the quality of the paper, which is obtained from the fiber length divided by the fiber diameter. High value of this ratio provides better forming and a well-bonded paper ( Ashori & Nourbakhsh, 2009 ). According to the physical pro­perties of paper test, this feature as an important factor has a significant effect on strength, tear, burst, breaking off, and double folding resistance. In addition, it is usually between 70-90 for softwoods and 40-60 for hardwoods for papermaking ( Akgul & Tozluoglu, 2009 ). The mean of SC had placed in hardwood range for all of studied altitudes. The acceptable value for a SC of the paper is more than 33 ( Xu et al. , 2006 ; Kiaei et al. , 2014 ; Enayati et al. , 2009 ), as also measured for the Iranian horn­beam wood in the studied altitudes.

Runkel coefficient (RC) is usually used to determine the suitability of a fibrous material for the pulp and paper production. Wood species with a RC of more than 1 have a stiff fiber, less flexibility, and poor bonding ability. In addition, the fibers with ratios less than 1 produce good quality pulp and paper ( Xu et al. , 2006 ; Jang & Seth, 1998 ). Therefore, according to the results of the RC, the fibers at altitudes of 300-700 m are suitable for paper production. Flexibility coefficient (FC) represents the paper resistance against rupture and burst, and also there is a direct relationship between the FC and the paper resistance ( Enayati et al. , 2009 ; Ashori & Nourbakhsh, 2009 ). There are four groups of FC, namely highly elastic (˃ 75), elastic (50-75), rigid (35-50), and highly rigid (< 30) ( Bektas et al. , 1999 ; Kiaei et al. , 2014 ). Hence, the measured characteristics of wood fibers at 300 m, 500 m, and 700 m belong to the elastic group being suitable for paper production. The fibers at altitudes of 900-1300 m are classified in the rigid group, which is not suitable for paper production due to inefficient elasticity. Therefore, they are mainly used in fiber plates, rigid cardboards, and cardboard production.

According to the forward matrix regression results, temperature, the DBH, crown canopy, and age of the trees were important factors in the wood variations compared to other factors. The temperature contribu­ted significantly to the variations of WDD, cell wall thickness, and Runkel coefficients. This result of mul­tiple stepwise correlation analysis are supported by those of Bakhshi et al. (2011) for maple wood. They found that fiber diameter dimensions were related to the precipitations in November (52.9%), August (19.6%), January's minimum temperature (5.6%), and November's maximum temperature (4.4%).

In this study, the wood dry density, fiber length, fiber diameter, cell wall thickness, and morphological properties of hornbeam wood were studied at six altitudes above sea level in Nowshahr site. The follo­wing is a summary of the most important findings:

· There are significant differences in WDD, FL, FD, CWT, SC, RC, and FC of hornbeam wood at various altitude levels.

· The average values of FL, FD, SC, and FC decreased by increasing the altitude to 51%, 40%, 7.7%, and 87%, respectively. However, the average values of WDD, CWT, and RC increased by increasing the altitude levels to 6%, 32%, and 74%, respectively.

· The temperature and tree DBH played an impor­tant role in the wood properties as shown by forward step­wise regression. The variations of WDD, CWT, and RC were related to the temperature. FL and FD were affected by the tree DBH.

· Hornbeam wood at low altitudes is suitable for paper and pulp production due to longer fiber length, favorable flexibility coefficients, lower Runkel coef­ficients, and more desirable slenderness coefficients.