This study was conducted in the Caspian forest of Iran. These forests are managing by selection cutting silviculture as close to nature method. The geographical coordinates of the study area were 37° 38’ 34” to 37° 42’ 21” N and 48° 48’ 44” to 48° 52’ 30” E. Elevation of the study area is ranged from 1,150 to 1,350 m above sea level. The mean annual precipitation is approximately 950 mm and the mean annual temperature is 9.1°C. The original vegetation of this area is an uneven-aged mixed hardwood forest. Average growing stock in this forest was 312 m ³ ha ^-1 and average number of trees was about 235 stem ha ^-1 above 7.5 cm DBH (Diameter at Breast Height). The soil type is forest brown and soil texture varies between sandy clay loams to clay loam. The number of selected trees to harvest in the study area (42 ha) was 286 stems that volume of these trees was 546 m ³ . Chainsaw and cable skidder (Timber jack 450C) was machines for cutting and extraction of marked trees. The length of winching cable was 50 m. The weight of skidder was 9.8 t and its width and length were 3.8 and 6.4 m respectively.

The point-transect method was used for assessing of soil disturbance according to the method developed by McMahon (1995) , immediately after logging operation. The disturbance assessment comprises the classification of disturbance types ( Table 1 ) at 5 m intervals within a 30 cm radius along 30 m transects. Two transects were located at grid centers with dimensions of 100 m × 100 m with random start point. A random azimuth was selected for ordination of first transects, and added 90° for the second transect. In addition, 504 points were recorded for assessing visual soil disturbance ( B.C. Ministry of Forests, 2001 ; Solgi & Najafi, 2014 ). Length, width and deep of ruts (larger than 200, 30 and 5 cm, respectively) that intersected by transects were measured. For assessing soil physical properties, 15 soil samples were taken from each identified areas (Undisturbed, winching corridors, deepest point of ruts and skid trails). The cylindrical sampling method was used to determine dry soil bulk density. The soil samples from the 10 cm were collected with a soil hammer and rings (diameter 5 cm, length 10 cm), put in polyethylene bags, and labeled immediately. Surface litter and duff were removed before sampling. The soil samples were dried in an oven under 105 C for 24 hours to obtain dry bulk density ( Lee et al ., 1983 ). Sample volumes and weights were corrected for large roots, wood, or gravel. The dry bulk density was calculated from the equation (1):

ANOVA and Duncan test by SPSS software version 16 were used for analysis of the soil bulk densities in identified areas, for damages in tree and regeneration species, tree DBH, and growing stages of regeneration, after checking normality of data (Kolmogorov–Smirnov test) and homogeneity of variances (Leven test). Regression analysis was applied to test the relation between wound height and wound size.

The frequency of soil disturbance types is shown in Table 1 . About 5.2% of soils were deep disturbed, and the most disturbance type in this disturbance class was ruts with depths of 5 to 15 cm. About 29% of soils were shallow disturbed and 60% undisturbed.

The soil bulk density in the winching corridors was increased 10.7% than controlled area, but the difference was not statistically significant ( Table 2 ). While, the soil bulk density in the ruts and skid trails were significantly increased (20.6% and 32.1%, respectively) than controlled area soils ( Table 2 ).

A 4.9% and 7.1% of total regeneration (1,407 stem ha ^-1 ) were wounded and destroyed respectively following logging operation. The frequency damages in growing stages of different species are shown in Fig. 1 . The density of seedling, small sapling and large sapling were 723, 438 and 246 stem ha ^-1 , respectively. The mean frequency of damage in seedling, small sapling and large sapling were 8.8%, 12.8% and 19.5%, respectively. The mean damages to large sapling were significantly the highest among other regeneration stages ( Fig. 1 ).

Species of regeneration were comprised, 70.6% Fagus orientalis Lipsky , 13.4% Carpinus betulus L., 3.9% Acer insigne Boiss , 3.1% Acer cappadocicum Gled., 3.7% Alnus subcordata C. A. Mey., 1.9% Acer platanoides L. , 1.6% Tilia begonifolia Stev., 1.1% Ulmus glabra Huds., and 0.7% Fraxinus coriarifolia Scheel. The most frequency of wounded and destroyed regeneration was observed on the Alnus subcordata (15.4% and 10.6%, respectively). The damage frequency on the regeneration of Acer and Alnus species were significantly more than the other species, while damage frequency on the regeneration of Fagus orientalis , Ulmus glabra and Carpinus betulus were less than the other species ( Fig. 1 ).

A 8.3% of trees were wounded and 2.9% of them were destroyed following logging operation ( Table 3 ). The most frequency of wounded and destroyed trees were observed on the Alnus subcordata (16.9% and 4.8%, respectively). The lowest frequency of destroyed trees was observed in Fagus orientalis with 2.2%.

The frequency of damage types in tree DBH classes are shown in Table 4 . The most frequency of wounded and destroyed trees were observed on the first DBH class (10.4% and 5.5%, respectively), and the lowest frequency of wounded and destroyed trees were observed in the latest DBH class (3.9% and 0%, respectively). The results showed that the frequency of damages to trees was decreased with increasing DBH of trees ( Table 4 and Fig. 2 ).

From all of the damaged trees (92 stems), 31 stems (33.7%) were damaged on the crown area, 50 stems (54.3%) were damaged on the bole area and 11 stems (12%) were damaged on the root area ( Table 5 ). In fact, about 2.8% of residual trees were damaged on the crown area, 4.5% were damaged on the bole area and 1% was damaged on the root area.

The mean (±SD) of wounds size on the bole of residual trees was obtained 173.9 (±121.5) cm ² , and the mean (±SD) of wounds height from ground level was obtained 0.70 (±0.60) m. The regression analysis showed the wound size significantly increased with increasing wound height ( Fig. 3 ) (F = 49.7; P < 0.01).

From all of the wounds (50 wounds), the numbers of bark (code 1), bast (code 2) and wood (code 3) damaged stems were 8, 13, and 29 stems, respectively. According to collected data, 0.7% of residual trees were damaged on the bole area with bark intensities, 1.2% with bast intensities and 2.6% with the wood intensities. The mean (±SD) of wound intensity on the bole of residual trees was obtained 2.42 (±0.76). The mean of wounds intensity on the < 0.3, 0.3-1 and > 1 m of trees bole were 2.62, 2.46 and 1.91, respectively ( Fig. 4 ). The ANOVA test showed that the mean of wound intensity on the height classes of wound are significant different (P < 0.05). The wounds intensity was decreased with increasing wound height ( Fig. 4 ).

The results showed that about 5.2% of the forest surface soils were deep disturbed, and about 89% were undisturbed or shallow disturbed following selection cutting and cable-skidding. The extent of soil disturbance was reported 30% by Solgi & Najafi (2014) . The results indicated that the soil bulk density in the ruts during winching of logs and in the skid trails are significantly higher than the soil bulk density in the controlled areas. Significantly different bulk density between disturbed and undisturbed areas was found by Marchi et al ., (2014) . However in other studies, the winching operations do not every significantly affect the bulk density. This is due to utilization methods and appropriate training of the operators ( Picchio et al ., 2012 ; Marchi et al ., 2014 ). Krause (1998) reported that compaction from harvesting equipment can reduce water infiltration and air permeability which is detrimental to the establishment and growth of regenerating species. The extent of disturbed soils by ground based logging system was reported 218 m ² in Brazil forests ( Verissimo et al ., 1992 ) and 94 m ² in Malaysia forests ( Pinard et al ., 2000 ) for each harvested tree. The extent and degree of soil disturbance associated with skidder extraction are variable and related to slope ( Stuart & Carr, 1991 ), soil texture ( Clayton, 1990 ; Jusoff, 1991 ) and soil moisture content at the time of logging ( Jusoff, 1991 ). Soil compaction resulting from harvesting operation limit the effective rooting depth of plants by restricting access to water, nutrients and reducing gaseous exchange (Gomez et al ., 2002).

Forest regeneration is the growth of tree seedlings into large mature trees, an important indicator for forest health. The long term health and economic viability of forest ecosystems is dependent on sufficient tree regeneration. The results of this study showed 4.9% and 7.1% of regeneration was damaged and destroyed by cable-skidding, respectively. Forests depend on adequate regeneration of tree species to be healthy and sustainable. Lotfalian et al . (2008) reported damage percentage to regeneration during cable-skidding about 8.0% in the Caspian forests of Iran. Hosseini et al . (2000) compared damages to natural regeneration by two logging systems (skidder and arial cable) in the Caspian forests of Iran. They reported that damages to regeneration in skidder system were significantly higher than aerial cable system (11% vs. 5%). Winching is the main cause of damages to regeneration during cable skidder logging ( Nikooy et al ., 2010 ; Picchio et al ., 2011 ). Results of study showed damage to regeneration was increased by increasing of height of regeneration, so damage percentages on seedling, small sapling and large sapling was 8.8, 12.8, and 19.5%. This may be due to reduction of regeneration flexibility with increasing their heights. Nikooy (2007) studied logging damage on regeneration in the Caspian forest and reported that damage rate on the seedling; small sapling and large sapling stage were 20.3, 20.6, and 26.5% in felling gaps. Whitman et al . (1997) reported damage level of 15% to seedlings in a single tree selection cutting operation in northern Belize. The logging methods have a big influence on effects on regeneration disturbance. The regeneration survival is significantly higher when using a snatch block in the winching. When block are not used, the impacts are bigger ( Picchio et al ., 2012 ; Marchi et al ., 2014 ).

The results of this study indicated 11.2% of residual trees were damaged by timber harvesting operation in the study area. This amount of damage to residual trees is considerable. Residual tree damage following selection cutting in the Caspian forests reported 15.5% by Lotfalian et al . (2008) , 19% by Naghdi et al . (2009) , 19.7% by Nikooy et al . (2010) and 16.9% by Tavankar et al . (2015) . Differences in the percentage of damages are probably due to ground slope and harvesting intensity. Yilmaz & Akay (2008) in Turkey forests showed 14% of residual trees damaged during felling and skidding operation. Hartsoug (2003) in the north eastern California forests showed that 23% of residual trees damaged during ground based logging operation and smaller trees were more likely to be damaged. The study conducted by Ficklin et al . (1997) indicated that skidding operation using wheeled skidder damaged about 22% of the residual trees. The levels recorded in this study are near to those reported for mechanized in Mediterranean European operations with wincing ( Vasiliauskas, 2001 ). Marchi et al . (2014) reported that the frequency of wounded trees is highest when winching without a snatch block and decreased when using a snatch block. Operation methodologies with a snatch block allows reducing by one quarter the frequency of wounded trees, which dropped from 50% to about 36% compared to without a snatch block methodology.

Our results indicated that 90% of destroyed trees had diameter less than 40 cm. Studies in West Virginia hardwood stands have shown that most of the damaged and destroyed trees are less than 5.0 inches in diameter ( Lamson et al ., 1985 ), so the net loss of residual basal area was only 6%. Our results also showed the shade-intolerant species such as Alder and Maple were more damaged than other tree species. One of the main goals of selection cutting management is the conservation of species diversity. About 2.6% of residual trees were intensive damaged in the crown area. All of crown damages were occurred in the tree felling stage. The results of this study showed 78% of wounds were occurred in the heights of less than 1 m from the ground level on the boles of residual trees. This finding has been supported by researchers by Bettinger & Kellogg (1993) , Solgi & Najafi (2007) , Lotfalian et al . (2008) , Naghdi et al . (2009) , Nikooy et al . (2010) , Jourgholami (2012) , Marchi et. al (2014) . The highest risk for decay is given for trees with injuries in the area of the felling cut and the root collar.

About 1% of residual trees were wounded on root area. The root wounds can significantly reduce tree growth, especially if they are intensive ( Meadows, 1993 ). Tavankar & Bonyad (2014) , reported 67.1% of logging wounds on residual tree boles were closed and 32.8% of the damaged trees were not able to improve their wounds, about 4.2% of wounds caused to tree mortality. Our results indicated that 3.2% of residual trees were damaged with injury sizes of more than 200 cm ² . The residual trees can not able to repair these wounds ( Tavankar & Bonyad, 2014 ).