The medium‐term effects of forest operations on a mixed broadleaf forest: Changes in soil properties and loss of nutrients

This study took place in a mixed deciduous forest located in Northern Iran to investigate the effects of two types of silvicultural methods, specifically the single‐selection method and the group‐selection method. The study also included the observation of skid trails in terms of runoff, sediment yield, and the loss of NO3‐N and PO4‐P. The process of recovery was studied by the assessment of these characteristics after the completion of forest operations for a period of 3 years. Although there is a considerable amount of literature on the topic of soil erosion caused by forest harvesting, there is little information regarding the medium‐term effects of different sizes of gaps in the forest canopy. The attempt to gather data was done by designating hillslope plots on the soil directly affected by machinery passage on skid trails, as well as on the felling gaps in correlation with the two types of silvicultural methods. Plots were also marked on an unharvested area as a control. The results confirmed the suitability of retention forestry within the framework of sustainable forest management. A majority of the data gathered showed no difference between the single selection and the control area. What is more, the values obtained from the felling gaps correlated to the group selection resulted to be the same as the control values 2 years after harvesting. However, the disturbance caused by machinery passage on skid trails showed a significantly higher level of impact, and the recovery time needed was at least 3 years.

able amount of literature on the topic of soil erosion caused by forest harvesting, there is little information regarding the medium-term effects of different sizes of gaps in the forest canopy. The attempt to gather data was done by designating hillslope plots on the soil directly affected by machinery passage on skid trails, as well as on the felling gaps in correlation with the two types of silvicultural methods. Plots were also marked on an unharvested area as a control. The results confirmed the suitability of retention forestry within the framework of sustainable forest management. A majority of the data gathered showed no difference between the single selection and the control area. What is more, the values obtained from the felling gaps correlated to the group selection resulted to be the same as the control values 2 years after harvesting. However, the disturbance caused by machinery passage on skid trails showed a significantly higher level of impact, and the recovery time needed was at least 3 years.

K E Y W O R D S
group-and single-selection methods, hydrological recovery, NO 3 -N, PO 4 -P, water quality, wood harvesting

| INTRODUCTION
There are several types of ecosystem services that forests provide in addition to wooden products (Nocentini et al., 2022). One of the most important is hydrological protection (Titti et al., 2022;Zhang et al., 2021). Forests play a key role in the regulation of water resources (Sheil, 2018). For instance, by increasing the amount of intercepted rainfall the quantity of water which reaches the soil is decreased (Brown et al., 2013). Human interaction affects the role that forests have which is positive and beneficial for the stability of a given area. This is done through forest management (Goeking & Tarboton, 2020). Forest operations are one aspect of forest management that shows a strong influence on hydrological processes (Hawks et al., 2022). Ground-based forest operations, and to a lesser extent also aerial ones, usually cause soil disturbances which can lead to an increased amount of runoff. This is caused by canopy alteration which produces a lower rate of interception (Hawthorne et al., 2013), lesser evapotranspiration (McEachran et al., 2021), and increased throughfall reaching the forest soil (Jourgholami et al., 2019).
A higher amount of runoff causes higher water and sediment yield to be transported to streams (Jautzy et al., 2021;Picchio et al., 2021), as well as soil nutrient loss, such as organic carbon, nitrogen, and phosphorus (Palviainen et al., 2014), thus producing a negative effect on the fertility of the forest stands (Jamroz & Jerzykiewicz, 2022;Kreutzweiser et al., 2008). These conditions also reduce the degree of mitigation in terms of climate change (Ameray et al., 2021). In addition, the modified physicochemical features of the streams can alter the ecological processes of aquatic ecosystems (Davies et al., 2016).
The notable amount of literature on the topic of the influence of forest operations on runoff shows that this has been studied extensively. The reason for this is easily understood considering the importance of this topic (Ogasawara et al., 2021).
However, the results obtained by the various studies showed very high variability among them, as a consequence of the plethora of factors that could influence the response of soil to forest operations (Safeeq et al., 2020). As a general trend, a recent review highlighted how a harvested area and the construction of operations-related infrastructures (forest roads and landing sites) are the major influencing factors for sediment yield . It was also found that the application of best management practices was effective in the reduction of sediment yield, for example, retention forestry, limited cutblock size, and riparian buffers Hatten et al., 2018).
Similar to sediment yield, nutrient concentration in runoff has been proven to be strongly influenced by forest operations (Kastendick et al., 2012).
Soil compaction after ground-based operations leads to changes in soil biochemical processes, resulting in an increase of dissolved nutrient loss into downstream water (Palviainen et al., 2015). The increase of microbial activities in the forest floor and upper soil layers after canopy opening can change the movement of nutrients, thus increasing the transfer of dissolved nutrients from the hillslopes in the harvested areas down to water sources (Nieminen et al., 2017;Shah & Nisbet, 2019).
The Hyrcanian forests are the most important area for the Iranian wood supply chain. This zone is characterized by relatively steep slopes and by the extensive application of semi-mechanized harvesting systems with ground-based mechanized extraction by cable or grapple skidders (Dalir et al., 2022). This combination makes the forest soils of this area very sensitive to compaction and subsequent runoff.
This issue is one of the most important challenges for the forestry sector in this context. The same situation occurs in several zones of the world where the forestry is mainly located in steep areas and where agricultural areas are mostly on flatter grounds. Therefore, several studies have been developed in this area to study the correlations between forest operations, sediment yield, and nutrient loss. The major part of these studies was focused on possible mitigation strategies to decrease runoff Jourgholami et al., 2018Jourgholami et al., , 2021Jourgholami et al., , 2022, but there are no studies dealing with the influence of different silvicultural methods applied to the related skid trails created to extract timber.
In the Hyrcanian forests, mixed broadleaf stands are managed by applying selection logging with single-selection or group-selection silvicultural methods. Moreover, as previously reported, felling operations are done motor-manually by chainsaw, and extraction is carried out with ground-based machineries such as cable or grapple skidders.
Data collection was limited to a 1-year period through observation. The data showed that there were lesser runoff and sediment amount in the application of the single-selection method in comparison to group selection and skid trails .
There is no information on runoff-related nutrient loss for the area, nor any indication of the recovery time over a longer time span.
In light of these facts, the authors developed the purpose of this study with the specific goals of investigating the effects of two silvicultural methods and of the related skid trails in mixed broadleaf stands over a three-year experiment to provide data regarding: (1) quantitative assessment of runoff and sediment yield; (2) quantitative loss of nitrogen and phosphorus concentration; and (3) recovery in the runoff, sediment yield, nitrogen, and phosphorus loss within a 3-year period after logging. It was hypothesized that (i) both silvicultural methods and skid trails have an impact on forest soil in terms of runoff, sediment yield, and nutrient loss; (ii) the impact on skid trails is higher than in the soil not directly affected by machinery passes; and (iii) the magnitude of impact and recovery time are expected to increase as a consequence of the higher amount of canopy removal on skid trails, areas undergoing group selection, areas undergoing single selection, and the unharvested or control area.
Our study is unique in the fact that data collection spans a 3-year period, as well as, comparing different silvicultural methods applicable in mixed broadleaf stands. The understanding of the effects of different silvicultural methods on mixed broadleaf stands was thus enhanced. The unharvested areas serve as a benchmark or control treatment for discussing the data collected.

| Study area
The activities were carried out in cutting block no. 320 of the Gorazbon District in the Kheyrud Forest in the Hyrcanian forest region of northern Iran (51 33 0 12 00 E and 51 39 0 56 00 E and 36 32 0 08 00 N and 36 36 0 45 00 N) ( Figure 1). This block ranges in altitude from 1220 to 1380 m a.s.l., faces south, and has a mean slope of 18%. Based on the weather data ; meteorological data of Nowshahr Synoptic Station), the mean annual rainfall is 1380 mm with two main peaks occurring in summer and autumn. Average daily temperatures range from a few degrees below 0 C in December, January, and February to +25 C during the summer with a mean annual temperature of 7.8 C.

| Experimental design and measurements
In this study, the following four experimental treatments were compared: skid trails (with a similar longitudinal slope of 15%-20% and affected by more than 20 machine passages), group-selection method (without canopy cover), single-selection method (50% of canopy cover), and an unharvested area (control). Each treatment was sampled with 3 replicates with one plot per replicate (4 treatments Â 3 replicates Â 1 plot), hence, a total of 12 plots, each measuring 2 m 2 in surface (1 m wide by 2 m long). These were used to measure runoff, soil, and nutrient loss ( Figure 1c). Wooden boards were used to create the runoff plots which were inserted 10 cm into the ground to provide stability and emerging from the soil approximately 20 cm to prevent extra water from either entering or overflowing out of the experimental plot (Jourgholami & Labelle, 2020).
Within 3 years after forest harvesting operations, there were 159 total runoff events from 31 May 2014 to 17 December 2016. A runoff event was recorded when there was a minimum of 6 h of continuous rain. A downstream storage tank measuring 0.04 m 3 in volume was installed and connected to the plot by plastic pipe to collect the surface runoff. The stored volume of water was divided by plot area to determine the runoff in mm. After each event, the solution (water and suspended particles) was stirred to guarantee homogeneity. One liter of the solution was sampled with a plastic bottle previously rinsed with hydrochloric acid (pH <2.0) and distilled water, and then taken to the laboratory. After sampling, the storage tank was removed, drained, washed, and reinstalled for collecting the runoff from the following event.
In the laboratory, samples were filtered through Whatman 42 filter papers with a pore size of 2.5 μm, previously oven-dried at 105 C for 24 h. The liquid fraction collected from the samples was stored in a refrigerator at 4 C. The filters were weighed, and the amount of sediment was assessed by subtracting the filter weight. Sediment yield (defined as the amount of sediment leaving the watershed or catchment) was determined by measuring the sediment concentration. Then, to determine the sediment concentration (g L À1 ), the dryweighed sediment in a 1 L runoff sample was multiplied by the total collected runoff in the storage tank. Finally, the sediment F I G U R E 2 Different experimental treatments including skid trail (a), group-selection method (b), single-tree selection method (c), and unharvested area (d) in the study area. [Colour figure can be viewed at wileyonlinelibrary.com] concentration was divided by the dimension of the plot area to calculate the sediment yield (g m À2 ). The liquid fraction stored in the fridge was further filtered with a 0.45-μm membrane to assess nitrate (NO 3 -N) and phosphate (PO 4 -P) concentrations (reported as g L À1 ) via spectroscopy method using ultraviolet (UV) spectrophotometer (UV-1200) Palviainen et al., 2015;Shah & Nisbet, 2019).
The runoff coefficient for each event was calculated by dividing the runoff volume (mm) by the total rainfall (mm). Gross rainfall was measured with a rainfall gauge installed in an area without canopy cover. The amount of throughfall in each runoff plot was measured using a manual rain collector measuring 9 cm in diameter and 20 cm in height placed beside the runoff plot.
The soil properties of all runoff plots were measured (i.e., soil bulk density, total porosity, organic matter content, litter depth, and soil particle-size distribution) immediately after forest operations. In each plot, two soil samples were taken from two randomly selected areas: a first sample was taken from a depth between 0 and 10 cm using a steel cylinder (56 mm wide and 100 mm long) to measure soil bulk density. Nearby, a second sample of 20 cm wide Â 20 cm long Â 10 cm deep was taken and both were placed in plastic bags, labelled, and taken to the laboratory for analysis of soil properties according to the methods shown in Table 1. Then, litter depth was measured using a tape meter, while ocular observation was performed at three points of each plot to estimate the mean canopy cover (Korhonen et al., 2006). Canopy cover is defined here as the proportion of the forest floor covered by the vertical projection of the tree crowns (Korhonen et al., 2006). The percentage of canopy cover was assessed and averaged using an ocular estimate from three different points to determine the percentage of cover by dominant and codominant canopy according to the following categories: 0%, 1%-25%, 26%-50%, 51%-75%, and 76%-100%.

| Statistical analysis
The normal distribution of variables and the homogeneity of variance were verified through Kolmogorov-Smirnov test and Levene's test (α = 0.05), respectively.
The difference among site characteristics (bulk density, total porosity, litter depth, organic C, canopy cover, and soil texture) in the different experimental treatments (skid trails, felling gaps after group selection, felling gaps after single selection, and unharvested area) were investigated with one-way analysis of variance (ANOVA), applying the Duncan test as a post-hoc.
To compare the effects of skid trails, felling gaps after group selection and felling gaps after single selection on runoff, sediment yield, and nutrient loss, as well as changes over a 3-year time span, a generalized linear mixed model was applied followed by a type III sum of squares (SS) two-way ANOVA. Authors considered "experimental treatment" and "year" as fixed effects and "plot" as random effect.
HSD Tukey test was applied as post-hoc to assess the differences among treatments. To test the relationship among treatments, topsoil properties, canopy cover, runoff, runoff coefficient, sediment yield, and nitrate (NO 3 -N) and phosphate (PO 4 -P) concentrations, the Pearson correlation was applied at p ≤ 0.05. The statistical analysis was performed using the SPSS software package (version 20; SPSS, Chicago, IL, USA).

| Rainfall and soil properties
During the investigated period, a total number of 159 rainfall events >2 mm were recorded (Table 2).
Our results confirmed that the average soil bulk density (respectively, 1.23 and 1.07 Mg m À3 ) in the group selection and single-tree selection methods were higher than the values of the unharvested area (0.95 Mg m À3 ). Moreover, total porosity (5.6% and 8%), organic C (3.3% and 5%), and litter depth (2.9 cm and 5.4 cm) were lower in group and single-tree selection methods in comparison with unharvested (control) area values of total porosity (63.5%), organic C (11.5%), and litter depth (8.7 cm).
Soil's physical properties were affected by the different experimental treatments. As expected, machine passage was the factor that T A B L E 1 Methods of soil properties analysis. created the greatest changes in bulk density, total porosity, organic C, and litter depth. The soil in skid trails had a higher degree of compaction with a higher bulk density and a lower total porosity, as well as impoverished organic C. These effects are remarkable when compared with the values found in the unharvested area.
The soil texture did not differ from the unharvested area significantly, except for the sand concentration in the group-selection method where the value was slightly lower (5.6%) than the unharvested area. Nonetheless, although little variation in soil texture was found, this was not directly related only to the silvicultural methods tested but also to the variability that naturally occurs in forestry soils (Table 3).    Focusing on the correlation between the investigated parameters and soil properties, soil bulk density, runoff, and sediment yield were positively correlated (Table 4). This highlights the importance of reducing compaction in forest soil to decrease runoff and sediment generation. Additionally, phosphorous and nitrogen were also positively correlated with soil bulk density, runoff, and sediment yield.

| Nutrient loss
Furthermore, total porosity, organic C, and litter depth were negatively correlated to the parameters mentioned previously (Table 4).
Thus, the increase in soil disturbance in terms of higher bulk density and lower porosity is associated with a loss of litter, organic matter, and nutrient availability for plants and seedlings.
In contrast, sand, clay, and silt concentrations are not significantly correlated with the other variables. It can then be deduced that the soil texture of the study site was not affected by forest management, at least in the short-term period (Table 4).

| The effects of different silvicultural methods and related skid trails on runoff, sediment yield, and nutrient loss
Forest harvesting operations can substantially generate a change in the processes and functions of forest ecosystems such as energy and water fluxes, hydrological and biochemical processes, and plant diversities (Palviainen et al., 2015;Picchio et al., 2021;Shah et al., 2022). Accordingly, forest harvesting reduces canopy interception of precipitation resulting from tree removal, decreases evapotranspiration, increases throughfall passing through the canopy cover, and ultimately can result in an increase in runoff Goeking & Tarboton, 2020;Palviainen et al., 2014;Picchio et al., 2021).
After the recorded rainfall events (Table 2), as a consequence of the soil compaction induced by canopy alteration and, mostly, machinery passage on the forest soil during harvesting operations (Table 3), in the second and third year after harvest, the sediment yield and the runoff in the skid trail treatment were significantly higher than in the other treatments (Figures 3-5). This finding is consistent with Jourgholami and Labelle (2020), who reported that sediment yield was significantly higher in the skid trails as compared to unharvested stands, highlighting how the direct effects of the machinery traffic, leading to increased soil bulk density and lower porosity, is generally the most impactful component of forest management. Previous studies revealed that forest management activities such as timber harvesting and forest road construction contributed to increased sediment yield (Cassiano et al., 2022;Shah et al., 2022).
The obtained results reported in Figures 3 and 6  mineralize organic N into inorganic N (i.e., ammonium) (Kreutzweiser et al., 2008), which eventually convert from ammonium to nitrate during the nitrification processes, thus facilitating export through leaching from forest sites (Hume et al., 2018;Schmidt et al., 1996).
Previous studies demonstrated that forest harvesting results in an increase in nutrient concentrations and subsequent export into receiving aquatic environments due to changes in plant communities, root uptake demand, soil temperature, moisture, and microbial activity (Kreutzweiser et al., 2008;Shah & Nisbet, 2019). Mineralization and plant uptake are important processes in nutrient dynamics and nutrient concentrations. These can change drastically in downstream waters after forest harvesting as reported by Kreutzweiser et al. (2008) and Palviainen et al. (2014).
The highest concentration of phosphate was found in the skid trails (Figures 6 and 8). Previous studies have revealed that soil disturbances following forest harvesting caused elevated concentrations of P due to enhanced soil moisture and temperature (Evans et al., 2000;Palviainen et al., 2004). Evans et al. (2000) concluded that soil disturbances following ground-based forest harvesting can accelerate the weathering process of P in bare mineral soil. Some processes such as the leaching of dissolved organic carbon is coupled with the export of P (Kreutzweiser et al., 2008). In Finland, 3 years after logging operations, Palviainen et al. (2004) reported that P concentrations increased in the litter layer created by the leaching process of organic P from logging residues. The elevated soil temperature after tree removal leads to the acceleration of the processes of mineralization and nitrification in the forest soil, which causes the release of nutrients following the decomposition of organic matter and logging residues (Kreutzweiser et al., 2008;Palviainen et al., 2015). Consistent with the current study, previous studies revealed that the rainfall intensity and proportion of the harvested area in the catchment have significant effects on runoff and nutrient concentrations (Ide et al., 2013;Palviainen et al., 2014;Proto et al., 2016). In this study, all the treatments lay in the southern aspect with higher mineralization leading to the increase of nutrient concentrations after treatment as reported by Palviainen et al. (2014). The average concentration of nitrate by 2.89 mg L À1 in the skid trail was higher than the NO 3 -N concentrations of 0.19 mg L À1 at the 100% clearcut boreal catchment reported by Palviainen et al. (2014). The nutrient concentration of N and P in the unharvested treatment can be attributed to the throughfall nutrient fluxes (Salehi et al., 2016) and subsequent biomass flow (as a lateral runoff) through the litter layer, as reported by Kim et al. (2014).
It was observed that 2 years were enough to allow full recovery of the pre-harvesting situation in the felling gaps related to the groupselection method. A different situation was revealed for the portion of forest soil directly affected by machinery passages on skid trails.
The disturbance for this portion just after forest operations, related to machinery-induced soil compaction, is significantly stronger than that of the canopy gaps related to the two silvicultural methods. This aspect is clearly highlighted by the very high correlation between parameters related to soil compaction, as bulk density and microporosity, and all the investigated parameters (Table 4). This confirms that ground-based forest operations are the factor with the greatest impact in the context of forestry interventions as a consequence of increased soil compaction (Labelle et al., 2022;Proto et al., 2016;Tavankar et al., 2021). Furthermore, the recovery time needed to return to the pre-harvest conditions resulted to be at least 3 years in the skid trails for all the investigated parameters.
This finding largely confirms what has been reported in current literature  and further highlights the importance of proper planning of the forest operations, with particular reference to the skid trails or strip roads network (Görgens et al., 2020;Hoffmann et al., 2022;Picchio et al., 2020).
Comparing the results with the current literature it can be stated that the obtained findings are in agreement with previous studies on the topic carried out in similar stands but with a shorter time-span focus , confirming the direct relationship between management intensity (Ogasawara et al., 2021) and erosion, as well as between machinery-induced soil compaction and erosion (Nazari et al., 2021).

Interestingly, the demonstrated 1-3 years recovery time is in line
with what was reported for runoff recovery after retention forestry interventions in mixed hardwood stand in other parts of the world (Oda et al., 2018), and much faster than the 14-20 year period found for clearcutting on large surfaces (Oda et al., 2018;Palviainen et al., 2014), which was attributed to the applied method, area, and the typology of the forest stand. Similarly, the recovery time for nutrient loss after clearcutting was found to be much longer (Brais et al., 2002;Ide et al., 2013;Palviainen et al., 2015).
The recovery of hydrological characteristics can take from a few years to several decades after forest harvesting. Regeneration of trees, the establishment of pioneer species, and understory plants have a crucial role in maintaining canopy cover and retaining nutrients, increasing evapotranspiration, transpiration, and plant uptake to accelerate the process of restoration of the runoff, sediment yield, and nutrients to the pre-harvest level (Kreutzweiser et al., 2008;Palviainen et al., 2015;Picchio et al., 2021;Safeeq et al., 2020). Unlike C and N, P is controlled more by abiotic than biotic factors, the restoration of P after forest harvesting may need more time than C and N to return to the unharvested level (Cleveland & Liptzin, 2007).

| Study limitations and future research
There are some limitations to this study. However, it gives indications toward which objectives future research should be directed, particularly in the context of the study area. Firstly, this study did not consider the influence of silvicultural methods and skid trails on subsurface runoff. Considering that this component of runoff could provide some important information, future research in the study area should also be addressed toward evaluating the sub-surface runoff after forest operations (Puntenney-Desmond et al., 2020).
Furthermore, it is worth mentioning that the experimental design used small plots of 2 m 2 each. In a forested landscape, there is a considerable spatial heterogeneity which allows for the surface water to flow in the direction of unharvested sections within a catchment before reaching a stream. On a landscape scale, the applied design is essentially a 'worst-case' scenario for runoff evaluation. We suggest, therefore, that future studies be developed as pair catchment evaluation rather than focusing only on hillslopes of a limited dimension .
According to the results of this study, we can conclude that the gap size related to the single-tree selection method has a substantial effect on the recovery values of hydrological properties. Hence, mimicking the natural canopy gap remarkably has less impact than other methods limiting the impact on forest soil and improving water quality.
Moreover, some best management practices are advised to maintain water quality after harvesting operations including: • Postponing the logging operations during heavy rainfall.
• Avoid creating a passageway to the skid trails from streams and water sources.
• Spread brash/slash on the skid trails to increase the ground cover of bare soil.
• Applying water diversion structures such as contour-felled log erosion barriers on the skid trail after logging operations.

| CONCLUSIONS
This study discusses the impacts of two different silvicultural methods and machinery traffic on skid trails in terms of runoff, runoff coefficient, sediment yield, and the loss of nitrate (NO 3 -N) and phosphate (PO 4 -P). These were monitored for 3 years after logging in a mixed deciduous forest. Skid trails showed a stronger impact on the studied parameters and, although the disturbed areas were in a process of a recovery year by year after the timber harvesting, the initial conditions were not restored by the end of the experiment for all the investigated variables. It was also found that single-tree selection and group-selection silvicultural methods caused a lighter disturbance to the environment, demonstrating less invasiveness and potentially shorter recovery periods of the disturbed areas. This aspect can probably be related to the lower amount of litter removal and topsoil disturbance which occurred in the areas which were not directly affected by machinery traffic.
Achieving sustainable forest management encompasses a comprehensive view of the ecosystem and its resilience capacity. The obtained results showed that retention forestry (single selection and group selection) has a low impact on the hydrological characteristics in mixed deciduous forests, with expected recovery times of a few years, thus suggesting the applicability of retention forestry within the framework of sustainable forest management.

ACKNOWLEDGMENT
Open Access Funding provided by Universita degli Studi della Tuscia within the CRUI-CARE Agreement.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.