Introduction

Forest management is necessary for forest conservation and the safety of natural goods and people. In the mid-twentieth century, in many developed countries, agrosilvopastoral practices were abandoned due to the exodus of the population from the countryside to the city (^{Vélez, 2000}). This led to increasingly dense vegetation in forest areas and the growth of shrub species that were previously used for domestic purposes or as feed for livestock. Gradually, these agricultural areas became forest ones due to abandonment, producing the advance of the forest and its density (^{Benayas, Martins, Nicolau, & Schulz, 2007}).

In Catalonia, the study area, 65 % of the region is occupied by forests; 80 % of forest properties are privately owned and small (^{Cervera, Garrabou, & Tello, 2015}), sometimes less than 25 ha. This combination of factors - high forest mass density, private ownership and small property size - makes forest management complex. In addition, another determining factor, and limiting at the same time, is the low profitability of forest production, which is why most landowners abandon their forests, neglecting the management needs of their plots (^{Martínez-Alier & Roca-Jusment, 2000}). Another important consideration when managing is forest health. Lack of forest management has allowed the abundant growth of trees, although the individuals are often small in height and thickness due to competition among them (^{Benayas et al., 2007}), which means a shortage of water and nutrients (^{Francos, Úbeda, Tort, Panareda, & Cerdà, 2016}).

One problem related to the lack of forest management is that of large forest fires, an unresolved but identified problem. On many occasions, the availability of fuel is a product of the great forest fires of past decades, which turns the forest into a trap, due to the impossibility of intervention by extinguishing services in case of fire (^{Francos et al., 2018}). In addition, sometimes, due to strong winds or heavy snowfalls, many of the weak trees die and fall forming a undergrowth full of dead fuel. The vertical and horizontal continuity of the fuel creates a scenario that, in the event of a forest fire, makes its extinction impossible. The flames can even spread as far as urban centers, with a very severe impact on the environment (^{Peix, 1999}).

At the beginning of the 21st century, Catalonia began to undertake forest management from the autonomous administration, incorporating management into land-use plans (^{Plana, 2011}). The body in charge of this is the Center for Forest Property (CPF) through so-called Forest Planning Instruments (IOFs), applicable in both private and public farms. The IOFs include Technical Plans for Forest Management and Improvement (PTGMF) and Simple Plans for Forest Management (PSGF), organizational tools in forest planning. Through the IOFs, actions that should be carried out in a forest farm in a period not exceeding 10 years are projected, during which time the basic objectives proposed by the landholder or legal representative and the manager must be achieved. The PTGMFs are a planning tool for the management of forest farms with an area equal to or greater than 25 ha. This instrument must ensure the improvement, sustainability and multifunctionality of the forest systems; make a careful forest parceling of the farm; consider and integrate land management plans, mainly in the field of forest fire prevention and control; introduce silvicultural standards to ensure the regeneration of the tree mass; and minimize the risks of erosion and fire. This system is normally used for the management of private forests (^{Centre de la Propietat Forestal [CPF], 2013}).

On the other hand, forest management actions can have a negative effect on some soil properties. Mechanical clearing of the undergrowth can decompose the first horizons, due to the pulling up of roots and dragging of vegetation (^{Johnson, Johnson, Huntington, & Siccama, 1991}). In addition, there may be a hydrologic change at the surface level of the soil, as the interception area changes completely and the soil may be more unprotected against the impacts of raindrops. Consequently, after such forest management, there may be increased runoff and erosion (^{McBroom, Beasley, Chang, & Ice, 2007}). Runoff generation and erosion studies can be carried out at basin level by instrumentalizing riverbeds or at plot level on the slopes of these basins. There are some methods to assess increased runoff and erosion such as plots, sediment traps or rainfall simulations.

In the present study, the objective was to determine whether forest management (PTGMF) in a Mediterranean forest basin, composed of Quercus suber L. and Pinus halepensis Mill., is the cause of increased runoff and eroded material at slope level; the generation of runoff and erosion was compared to that of a dense control or unmanaged forest, a forest road, and a bare or sparsely vegetated area.

Materials and methods

Study area

The study area is located in northeastern Spain in the province of Girona in the Les Gavarres Massif, where for more than 30 years studie have been conducted on hydrology and erosion in different land uses, both at slope and basin level. The experimental 2.5 km² Vernegà stream basin is the study unit (Figure 1). The Vernegà stream is intermittent or seasonal due to the limited and highly variable rainfall and the granitic nature of the substrate that provides high permeability, having a granitic aquifer inside. The climate is subhumid Mediterranean with average annual rainfall of around 650 mm. Most precipitation occurs during autumn and spring. The depth of the Luvisol-classified soil (^{IUSS Working Group WRB, 2006}) is 1 m, with a structure composed of 80 % sand, 18 % silt and 2 % clay. The vegetation is composed of Q. suber and Quercus ilex L. and in some areas Pinus pinaster Aiton., P. pinea L.and P. halepensis, as well as a significant understory of Erica arborea spp. and Arbutus unedo L. as main shrubs. The density of forest roads in the basin is 2.1 km·km^-2.

Figure 1 Location of the study area. Spain, Catalonia, Les Gavarres Massif and Vernegà Basin. 

Two FGWPs (2002-2012 and 2013-2023) have been carried out in the area. Their objectives were to clear the undergrowth to favor he growth of Q. suber (use of cork); clear the undergrowth, especially E. arborea and A. unedo, to favor P. halepensis (wood production); and generally clear the property to reduce the risk of forestfire. The management in this property was done mechanically with a brush cutter.

Methodology

Seven research units were selected: dense forest, recently managed Q. suber forest, Q. suber forest managed two years ago, recently managed P. halepensis forest, P. halepensis forest managed two years ago, a bare area with sparse vegetation and a forest road (Figure 2). In each area, four rainfall simulations were carried out in June 2014; in total, 28 one-hour simulations were carried out. The four simulations, in each of the areas, were carried out in a 0.5 ha area.

The rainfall simulator and the method used are described by ^{Calvo, Gisbert, Palau and Romero (1988)}. During the time that the simulated rainfall lasted, a pressure of 1.7 kg·cm^-2, equivalent to a rainfall intensity of 60 mm·h^-1, was maintained. The experimental area was 0.24 m² (Figure 2b) and the simulated rainfall precipitated from a height of 2 m (Figure 2c). A moisture recorder (Delta T Devices AT HH2) constantly monitored the first 10 cm of the soil.

The 0.24 m2 plot had an outlet where the runoff water and the sediment transported by this water were collected and quantified. In each simulation the following variables were calculated: 1) initial and final moisture of the first 10 cm of the soil, 2) total rainfall (L), 3) total contribution (L·m^-2), 4) runoff coefficient (%), 5) erosion/rainfall ratio (g·L^-1), 6) erosion/runoff ratio (g·L^-1), 6) erosion (g·m^-2). An ANOVA was subsequently made and significant differences were verified by a post-hoc Tukey test (P< 0.05).

Results

According to Table 1, the moisture in the first 10 cm of soil was very low (0.5 to 0.8 %) before the rainfall simulations; in the end, the moisture in the vegetated areas was above 30 %, while in the bare area it was 25 % and on the road only 14.5 %. This indicates that the bare soil and road had less capacity to absorb water in their first centimeters.

Table 2 shows the rest of the variables analyzed after the rainfall simulation. The contribution in the forest road was significantly higher (P < 0.01, 14.59 L·m^-2) than in the rest of the analyzed plots. The P. halepensis and Q. suber plots, both with recent management, the P. halepensis plots with two years of management, and the dense forest ones showed the lowest values (<1 L·m^-2).

The forest road and the bare terrain, areas devoid of vegetation, produced much higher runoff (4 to 12 %) than the more vegetated areas (<1 %). These data are related to those of moisture, since these two land uses had less capacity to absorb water in the first 10 cm of the soil. The subgroups found in the statistical analysis were the same as in the contribution analysis (Table 2).

The erosion/rainfall ratio was significantly higher on the forest road (0.69 g·L^-1) than in the rest of the plots. In this case, the use of bare soil had no significant differences (P < 0.01) with the other uses; apparently, forest management had no influence on the result. Concentration values were really low, from 0.0001 g·L^-1 in Q. suber with recent management to 0.0088 g·L^-1 in Q. suber with two-year management (Table 2). With respect to the erosion/runoff ratio, the forest road had the highest value with 6.5 g of sediment per liter of runoff, then P. halepensis with recent management 2.7 g·L^-1) stood out and then the other uses with values less than 1 g·L^-1 (Table 2).

With respect to the erosion/runoff ratio, the forest road had the highest value with 6.5 g of sediment per liter of runoff, then P. halepensis with recent management (2.7 g·L^-1)stood out and then the other uses with values less than 1 g·L^-1 (Table 2).

Erosion results were similar to those of the erosion/rainfall ratio analysis, as the forest road stood out for having the highest value with a total of 96 g·m^-2. Allother land uses had statistically similar values, being less than 4 g·m^-2 (Table 2).

Discussion

In the Vernegà basin, ^{Úbeda and Sala (2001)} determined erosion on dense forest slopes, using Gerlach-type erosion boxes. These authors indicated that the maximum erosion value was 3 g·m^-2, a higher value than that obtained in the dense forest of the present study, but similar to the bare plot (3.5 g·m^-2). In another research work at Les Gavarres, runoff in the unmanaged forest was similar, as it did not exceed 0.9 L·m^-2 (^{Sala & Rubio, 2000}).

^{Sidle et al. (2006)} worked in several locations in South Asia and found that, following forest management, runoff at slope scale increased and was even noted at basin level. This is because management work can change the hydraulic conductivity of the first soil centimeters and break the most superficial structure. In the areas of the present study, there were no major changes in the first centimeters of the soil with respect to the runoff coefficient in the dense forest and managed forest areas, being less than 1.5 %, but it was higher in the bare area (4.1%) and on the forest road (12.5 %). ^{Mohamadi and Kavian (2015}) studied runoff generation in areas without vegetation and obtained a maximum of 5.6 % and a minimum of 4.7 %, values similar to those obtained in the bare plot with a 4.1 % runoff coefficient. In the same work, sediment concentrations in runoff waters (0.58 g·L^-1) also resemble those of the present study in the bare plot (0.62 g·L^-1). ^{Sidle et al. (2006)} state that only long-term work can prove the effects of forest management in terms of erosion, since there is material that can remain loose on the slopes and only after large rainfall events is it mobilized downstream and even earth movements can possibly occur. A triggering event can occur long after forest management. In contrast, ^{Croke, Hairsine, and Fogarty (2001)} did record sediment movement and higher loads of suspended particles in Scottish rivers just after brushcutting. The lack of water absorption by the soil and the movement of roots and surface structure can cause these landslides.

Erosion data in forests without some kind of management can also be disparate; slope, cover, soil type and rainfall intensity are some of the variable that can come into play. For example, in Panama, ^{Zimmermann, Francke, and Elsenbeer (2012)} obtained an erosion rate of 0.01 t·ha^-1 for a year with little precipitation and 0.02 t·ha^-1 for a rainier year, indicating that the changes were minimal due to modifications in precipitation. In the study basin in this article, ^{Úbeda and Sala (1998)} determined that the rates in dense forest slopes were 0.12 t·ha^-1·year^-1. In the present study, the characteristics of the plots are similar in slope, soil type and rainfall intensity, and different in vegetation cover; therefore, it can be deduced that this last variable was the one that affected erosion generation to a greater extent, since the bare plot and the forest road had higher values than soils with some type of cover.

^{Mena, Benavides, and Castillo (2011)} studied the generation of runoff and erosion resulting from simulated rainfall in various agricultural uses in Colombia and compared it to an unmanaged dense forest. The minimum runoff that they obtained in this plot was 0.090 L·m^-2 and the maximum was 0.389 L·m^-2. In our case, the runoff in the dense forest was higher (0.80 L·m^-2), although it cannot be considered to be a very high rate compared to the 14.59 L·m^-2 of the forest road. Regarding erosion, the authors obtained 1.9 g·m^-2 as a minimum and 6.3 g·m^-2 as a maximum, which are higher than those obtained in the unmanaged forest of the present study (0.35 g·m^-2); however, these erosion rates are not considered high.

^{Labrière, Locatelli, Laumonier, Freycon, and Martial (2015)} made an extensive review of erosion in the tropics in 18 types of land uses. In this study, the erosion range varied from 1 g·m^-2·year^-1 to 2 458 g·m^-2·year^-1. Agricultural land uses produced greater erosion. In this review, managed forests had a maximum of 6 g·m^-2·year^-1. It should be mentioned that the data correspond to works that were not carried out with simulated rainfall, but with erosion plots.

The type of machinery used to carry out forestry work is also a determining factor, since it can not only cut down trees and shrubs, but can also pull up roots, which involves a movement of fine and coarse particles that are likely to be mobilized after a rain (^{Stott, Leeks, Marks, & Sawyer, 2001}). The machinery used in the Vernegà basin is also quite aggressive with the soil surface, even capable of pulling up roots; however, the mineral soil was never uncovered, as there are vegetal remains and fresh, organic matter (leaves and branches) and in decomposition (humus) that achieve a cushioning effect and favor infiltration capacity, avoiding accelerated runoff. ^{Hartanto, Prabhu, Widayat, and Asdak (2002)} indicate that there are factors that forest workers must be aware of in order not to cause irreparable damage to the soil surface. These authors indicate the need to leave a litter layer to avoid splashing from raindrops and thus not alter the soil’s bulk density. Another issue they consider vital is to provide some soil surface roughness, so that, in case of erosion, the particles are trapped in the soil.

^{Ehigiator and Anyata (2011}) compare erosion produced by heavy machinery and manual clearing work at slope level and quantify erosion rates of up to 13.8 t·ha^-1·year^-1 with tree-pusher/root-rake attachments and tilled conventionally and 2.5 t·ha^-1·year^-1 when management is less aggressive. These authors, whose area of study focuses on tropical forests, found high erosion rates due to slope formation. The objective of these slopes was precisely to avoid erosion so that the effect of management could be noticeable in terms of particle movement, even 18 years after forest treatment. The authors considered that slope was the key factor in understanding this type of process. In the current case study, the gradient is very low and even zero so it did not influence the results. The forest road recorded significantly higher erosion values than the rest of the land uses, so the absence of vegetation cover is considered to be the determining factor for erosion.

^{Croke et al. (2001)} and ^{Stott et al. (2001)} studied slope form as an important variable for erosion generation in Scotland and Wales. Convex and steeper slopes experienced greater particle movement. ^{O’Farrell, Heimsath, and Kaste (2007)} arrive at the same conclusion. This type of slope is also found in the Vernegà basin; in one of the areas where strawberry tree and heather clearing work was carried out, the gradient is very steep and the slope is convex. In the Vernegà basin, 43.6 % of the area is convex, 40.6 % is concave and only 15.8 % is flat. The flattest areas of the basin correspond to the agricultural areas (^{Pacheco, Farguell, Úbeda, Outeiro, & Miguel, 2011}).

^{Neary, Ice, and Jackson (2009)}, after research in several watersheds in the United States, conclude that forest management always produced quantitative and qualitative changes in hydrology. ^{Neary et al. (2009)} indicate that the soil is a great filter, from the litter to the deepest layers, and that work in the forest must be prevented from damaging these layers in order for water purification to be effective. For this reason, it is extremely necessary to protect the soil and manage it by trying to avoid, as far as possible, the damage resulting from management and erosion generation. In our case, the silvicultural treatments did not significantly affect the managed forest mass, thus protecting the soil system.

One question that could be asked is: What recurrence can forest management have to avoid damaging the soil? ^{McDonald, Healey, and Stevens (2001)} verified how soil quality is affected when forest management is very recurrent, and more so if the slopes are steep. Forestry works can affect the organic and chemical properties of the soil, not only because of the mechanical effect of the instrumentation used, but also because of the increase in soil washing (^{Francos, Úbeda, & Pereira, 2019}). In these cases, it should be taken into account that assessing the possible impact of each management practice on the soil is essential to make the decision whether to do it or not. In addition, steep slope areas are prone, due to their characteristics, to have greater soil washing and be subject to erosion. Therefore, it is essential to introduce the “soil quality” factor to determine the frequency and place where a specific type of management is carried out. In the Vernegà basin, the indicator that determines the frequency at which clearing should be performed is the height of the undergrowth and not the soil parameters. ^{Wakiyama, Onda, Mizugaki, Asai, and Hiramatsu (2012)} have used erosion plots to determine critical erosion points in Japan and recommend not using aggressive management in these places. The authors conclude that the most fragile places are the points where the soil has a lower amount of organic matter. Therefore, soil quality studies, prior to forest management, are necessary to avoid high rates of soil erosion and degradation.

Regarding the generation of runoff and erosion that occurs on forest roads, ^{Zemke (2016)} also used simulated rainfall in a study in Germany in several forest road scenarios and obtained a maximum of 62.76 L·m^-2 runoff and a maximum of 272.2 g·m^-2 erosion. These results compare them to an unaltered soil surface and the data are 2.5 L·m^-2 and 4.7 g·m^-2, respectively. In this case, according to the author, the passage of machinery over forest roads to do forest management tasks favors erosive processes. Also using simulated rainfall, ^{Butzen et al. (2014)} obtained disparate erosion values on forest roads in Luxembourg and Germany, with minimums of 2.6 g·m^-2 and maximums of 122.5 g·m^-2. The values reported by ^{Zemke (2016)} and Butzen et al. (2014) are high compared to those obtained in the present study.

Use and management influence the magnitude of soil loss (^{Panagos et al., 2015a}). Among the risk factors for soil erosion, cover management is the factor that policy makers and landowners can easily influence to help reduce soil loss rates. At the European scale, forests and shrublands have the lowest soil loss values (^{Panagos et al., 2015b}) with 7 g·m^-2·year^-1 on average, although plot- and basin-level studies are needed to obtain accurate data.

Conclusions

This study confirms that there is no significant difference in the generation of runoff and erosion in the managed areas, both pine and cork oak, compared to dense or unmanaged forest. Forest management with mechanical tools did not have a major impact at ground level with respect to the generation of runoff and erosion, as the amounts were not high. When carrying out this type of action, it is essential that the litter is preserved, that the humus or the most organic centimeters of the soil are not altered, and that the finest and most easily biodegradable vegetal remains are left on the surface to protect the soil from rain and external agents. Forest management is necessary for the use of forest products (cork and wood in this case), the minimization of possible effects in the event of a forest fire and the facilitation of the entry of firefighters, as set out in the technical plans for forest management and improvement.