Assessing the Impact of Engineering Measures and Vegetation Restoration on Soil Erosion: A Case Study in Osmancık, Türkiye

Abstract

The prioritization of preventing soil loss in Türkiye’s watersheds has become a pressing concern for planners. Numerous mathematical models are presently utilized on a global scale for soil erosion prediction. One such model is the Revised Universal Soil Loss Equation (RUSLE), commonly used to estimate average soil loss. Recently, there has been an increased emphasis on utilizing USLE/RUSLE in conjunction with Geographic Information System (GIS) technology, enabling grid-based analysis for predicting soil erosion and facilitating control measures. This study evaluates the effectiveness of erosion and flood control initiatives started in the 1970s within the Emine Creek watershed and its tributary rivers in Osmancık, Türkiye, utilizing RUSLE/GIS technologies. Two distinct maps illustrating the potential erosion risks were produced for two distinct temporal intervals, and a comparative analysis was conducted to evaluate the alterations that transpired. The implementation of various measures such as terracing, afforestation, and rehabilitation in the watershed led to a notable prediction of decreasing soil loss in the watershed. From 1970 to 2020, the rate of estimated soil loss was reduced from 417 to 256 metric tons per hectare per year, demonstrating the effectiveness of soil conservation measures in a semi-arid and weakly vegetated area at reducing potential soil loss.

1. Introduction

Water erosion, which leads to soil loss, poses a significant threat to natural resources and agricultural productivity [1]. Erosion harms soil health, water quality, hydrological systems, crop yields, habitats, and ecosystem services [2,3]. Soil erosion is impacted by various factors, including wind, precipitation, associated runoff processes, soil erosion susceptibility, land cover, and management characteristics [4,5]. The impacts have led to an estimated average annual soil loss of 12 to 15 tons per hectare on erodible lands worldwide [6]. In the countries of the European Union, this figure stands at 2.46 tons per hectare per year [5], while in Türkiye, it reaches 8.24 tons per hectare per year [7].

Various engineering measures and soil conservation techniques are globally used to combat water erosion and restore degraded lands. When degraded by deforestation, overgrazing, inappropriate land use, etc., and subjected to heavy rainfall, degraded lands can experience extensive erosion. In many parts of the world, terracing and engineering measures are taken to control water erosion on sloped lands [8,9,10,11]. Diversely constructed terraces are a barrier that slows the water flow rate and increases water retention. In addition, the method of contouring is used for comparable purposes in various regions of Europe and China [12,13,14,15]. Vegetation on degraded lands can aid in forming plant root systems that prevent wind and water erosion by holding the soil together. In countries with wastelands and hilly areas, such as South America and Nepal, this technique (bioengineering) is used [16,17,18]. In addition to mitigating slope erosion, engineering interventions, such as the implementation of check dams, are employed to manage gully erosion [19]. With these structures, it is essential to reduce the water’s flow rate and hold sediment. These engineering measures aid in preventing water erosion while also enhancing water quality in the watershed [20,21,22]. To combat water erosion and prevent ecosystem deterioration, the riverbanks are protected so that gully erosion damages do not increase [23].

Monitoring soil erosion is an essential component of soil conservation planning. Even in experimental plots, determining soil losses is expensive and time-consuming [24]. Understanding how these erosion processes occur and identifying areas susceptible to soil loss can significantly enhance land management. Several empirical water erosion prediction models have been developed to assess regions with high erosion intensity and to predict regions with limited data [4,25]. The Revised Universal Soil Loss Equation (RUSLE), proposed by Wieschmeier and Smith [26] and Renard et al. [27], is the most widely used experimental model for soil loss estimation. The RUSLE model, which can be evaluated for watershed protection, is adaptable, time- and cost-efficient, and practical for estimating soil losses in areas with insufficient data [28]. By considering relationships between land use and cover, topography, soil type, and precipitation, RUSLE can provide estimates of long-term annual soil loss. Soil loss is determined using the RUSLE model, which involves the multiplication of six parameters: The RUSLE model determines soil loss through the multiplication of six parameters: (1) the erosivity factor (R), (2) the soil erodibility factor (K), (3) the slope length factor (L), (4) the slope steepness factor (S), (5) the cover management factor (C), (6) the support practice factor (P). These parameter values are determined through field and laboratory research [26]. Factors C and P are associated with soil conservation and land use. In contrast, R, K, and LS factors are associated with the ecological characteristics of the study area. In contrast, factors R, K, and LS are linked to the ecological characteristics of the study area [29].

The Cover management factor (C factor) represents the most easily manageable conditions and is a variable that land planners can readily influence to reduce soil loss rates [30,31,32]. C factor values range from 0 to 1 and are determined by the weighted average of soil loss rates (SLR) [27]. The most significant impact on the C factor arises from changes in land use, particularly deforestation caused by the expansion of agricultural land. Additionally, in other management-related applications, the P factor is also considered. The P factor encompasses mitigation practices that reduce the erosive capacity of water flow by modifying drainage patterns, flow concentration, flow velocity, and hydraulic forces [27]. Globally, soil erosion is effectively controlled by soil and water conservation measures. Even on steep slopes (25°), soil erosion can be reduced by as much as 70% by applying soil conservation measures [33].

Nevertheless, modifying the support application factor (P) typically requires increased monetary investments and soil conservation subsidies [13,31,34,35]. In addition, the C and the P factors can be used in scenario analysis to evaluate the impact of various soil conservation practices on soil loss, determining whether they mitigate or exacerbate the problem [36]. Ozcan and Aytas [37] simulated various soil conservation practices using the RUSLE model to estimate the impact of those measures on soil loss and sedimentation in the Bakkal Dolin Lake (Çankırı/Turkey).

This study aims to determine the variation in the amount and severity of potential soil loss due to engineering measures implemented in a semiarid watershed with sparse vegetation. In addition, it is to evaluate the effectiveness of slope (terracing) and gully (check dams) improvement measures that effectively combat erosion. We aimed to predict the spatial influence of integrating afforestation and support practices on soil loss flux using RUSLE/GIS. The study area chosen was the Emine Creek watershed in Osmancık, known for its sensitive climate and topography prone to erosion. The selection of this watershed is based on its potential for addressing erosion issues in comparable geographical areas. Situated within the Mediterranean basin, where erosion is widespread, this watershed allows the relevant institution to address and mitigate erosion actively. Additionally, this watershed is a valuable research site for studying erosion processes.

As intended, erosion control measures were applied in the study area in 1970. The LS, K, and R factor values were assumed to be stable, allowing us to observe the effects of the measures. This study represents the first examination of the impact of erosion control measures in Türkiye. However, more information is needed regarding the effects of land cover and management techniques. Thus, planning the implementation and evaluation of erosion mitigation measures will be simplified.

2. Materials and Methods

2.1. Study Site

The research site is situated in the northern part of the Central Anatolia Region, positioned at the interface between the humid climate region of the Black Sea and the semi-arid climate region of Inner Anatolia. During the evaluation period of this study, approximately 63.5% of the project area consisted of fertile and degraded forest lands. Residential and agricultural areas accounted for 24.3% of the watershed (Figure 1). The elevation of the watershed ranges from 410 to 1541 m, with an average elevation of 815 m, and exhibits a very rough and faulted topographic structure. The soil belongs to the order of Inceptisols (Typic Xerochrepts) [38], with a depth varying between 20–120 cm. Severe surface erosion occurred in areas with high elevation and low cover, and A horizon was carried. Severe surface erosion occurred in areas with high elevation and low vegetation cover, leading to the loss of the A horizon and exposing the bedrock. The average annual rainfall in the area is 355 mm, with insufficient rainfall and irregular seasonal distribution exacerbating the effect of drought. According to the Thornthwaite method, the climate type is classified as “semi-drought, mesothermal, no excess water or slight excess water, close to maritime climate” [39]. The common forest stands in Central Anatolia consist of oak (Quercus pubescens Willd., Q. cerris L., Q. infectoria G.Olivier), pine (Pinus nigra J.F.Arnold subsp. pallasiana (Lamb.) Holmboe var. pallasiana, P. slyvestris L.), and juniper (Juniperus oxycedrus L. subsp. oxycedrus, J. excelsa M.Bieb., J. foetidissima Willd.). The dominant tree species in the study area’s forests are pine, oak, and juniper, depending on height and aspect.

2.2. Methods

Soil Loss Estimation

This study used the combined RUSLE methodology to predict sediment flux rates (ton ha⁻¹ year⁻¹) in the Emine Creek Watershed, Osmancık-Çorum-Türkiye (Figure 2). [26,27], (Equation (1)).

$$A=R\times K\times L\times S\times C\times P$$

(1)

The variable A is used to represent the average annual soil loss (t ha⁻¹ year⁻¹); R is the symbol for the rainfall erosivity factor (MJ mm ha⁻¹ h⁻¹ y⁻¹); K signifies the soil erodibility factor (t ha h ha⁻¹ MJ⁻¹ mm⁻¹); L corresponds to the slope length factor, S represents the slope steepness factor, C indicates the cover management factor, and P denotes the support practice factor.

The utilization of various digital databases, including both raster and vector formats, is integral to the implementation of RUSLE technology. These databases encompass digital elevation models (1:25,000), topographical maps (1:25,000), soil maps (1:25,000), forest maps (1:15,000), afforestation maps (1:25,000), and Corine 2006 land use data.

The R factor, also known as the factor of rainfall erosivity, is determined by multiplying the highest rainfall intensity for 30 min (I30) by the product (EI30) of the total rainfall energy (E) [26,40]. In this study, the R factor was directly obtained from the study area by Erpul et al. [41] (

Table 1. The RUSLE-R values (MJ mm ha⁻¹ h⁻¹ year⁻¹) and their annual distribution at the study site [43].

Months	1	2	3	4	5	6	7	8	9	10	11	12	Annual
∑R a	0.00	29.43	15.40	187.67	1743.77	2594.65	1236.99	574.95	310.88	117.09	44.26	69.01	6924.11
Rk b	0.00	2.94	1.54	18.77	174.38	288.29	123.70	63.88	31.09	13.01	4.92	7.67	730.19
%R c	0.00	0.40	0.21	2.57	23.88	39.48	16.94	8.75	4.26	1.78	0.67	1.05	100.00
R d	0.00	2.94	4.48	23.25	197.63	485.92	609.62	673.51	704.59	717.60	722.52	730.19	730.19

^a Flux of total energy flux; ^b Flux of average energy; ^c Flux of average monthly percentage of energy flux; ^d Flux of maximum energy flux.

). Due to the limited number of meteorology stations in the study area for kriging, it is necessary to consider the influence of elevation classes on the current precipitation levels [42], and the Digital Elevation Model (DEM) of the study area was utilized to form the spatial R surface. Adjustments were made to account for the equalization of rainfall levels based on the elevation within the watershed area [42] using Equation (2):

$$Ry=Rr {\left[\frac{Py}{Pr}\right]}^{1.75}$$

(2)

Ry refers to the corrected R factor value of the unknown unit; Py represents the average annual precipitation (mm) of the unknown unit; and Pr represents the average annual precipitation (mm) of the known reference station.

The study area spans an elevation range of 410 to 1541 m. The Osmancık meteorology station, selected as the reference station, is at an elevation of 419 m. R values for the unknown units were computed using DEM and ArcMap 10.6.1, based on Equation (2), which indicates that precipitation increases by 50 mm for every 30p0-m difference in altitude.

The study used a 1:25,000 scale K factor map, representing the soil’s resistance to erosion, obtained from the Turkish Erosion Database [44]. K factors were digitized for soil unit measurements using Equation (3), as suggested by Torri et al. [45]. For the K factor calculation and mapping, 23,000 profile data points from 0–30 cm were utilized [46]. The digitization of soil units involved using the equation suggested by Romkens et al. [47] and revised by Renard et al. [27]. This was done by associating the vector layers containing polygons of soil classes, erosion, and texture combinations, separating different K classes. Furthermore, the results were compared with K factor values from the Türkiye Big Soil Groups [48].

$$K=0.0293\left(0.65-DG+D{G}^2\right)\ast EXP\left[-0.0021\left(\frac{OM}{C}\right)-0.000037{\left(\frac{OM}{C}\right)}^2-4.02C+1.72\right.C^2]$$

(3)

K represents the RUSLE soil erosion sensitivity (ton ha⁻¹ ha MJ⁻¹ h mm⁻¹); whereas the geometrical average diameters of the principal soil components (mm) are represented by DG; OM denotes organic matter, and C is the percentage of clay.

Topography is the most influential factor controlling soil erosion risk. To calculate the LS factor, which accounts for slope length and steepness, the DEM of the watershed was used in conjunction with Equation (4) and the ArcMap 10.6.1 Hydraulic Accumulation tool. Hydraulic Accumulation has the advantage of considering the area contributing to the slope and the slope’s characteristics, allowing for a more comprehensive assessment of the intricate topography [49,50].

$$LS={(\frac{Xn}{22.13})}^{1.3} {(\frac{\sin \mathsf{\theta }}{0.0896})}^{1.3}$$

(4)

In this context, LS represents the RUSLE topographical factor, X denotes the surface flow condensation number, n signifies the size of the cells in which calculations are performed, and θ represents the slope steepness (^o).

The C factor corresponds to a cover management factor that varies depending on the type and coverage of vegetation. Generally, vegetation reduces the kinetic energy of raindrops before they impact the soil surface, significantly influencing erosion. According to other factors, the C factor has a substantial impact on the increase or decrease of soil loss over a brief period. The C factor, which ranges from zero (a well-protected land cover) to one (barren areas), consists of the following five subfactors: prior land use (PLU), soil moisture (SM), surface cover (SC), surface roughness (SR), and canopy cover (CC). Consequently, while C factor values for forest and grassland uses are low, C factor values for settlement areas are high. This study identified land use in 1970 and 2020 using CORINE Maps, afforestation project maps, and forest maps generated from aerial photos and land surveys. The C Factor can frequently be calculated using remote sensing techniques [37]. But the 1970 mapping of Factor C using remote sensing techniques could not be evaluated due to a lack of satellite images. Forest maps were prepared with aerial photos and field measurements, and their accuracy was verified using ground control points every 400 m, resulting in high spatial accuracy and pixel resolutions of the stand information. The land use in the study area was classified into seven categories: forest, grassland, settlement, agriculture, water, sandy area, and rocky area. In addition, according to the forest cover, it was classified into three parts. In

Table 2. C factor values (Arranged according to Wischmeier and Smith [26], Özcan et al. [51], Saygin et al. [52] and Panagos et al. [31].

Land Use Type/Land Cover (LUT/LC)	C Factor Value
Forest (0–30% ground cover)	0.15
Forest (30–70% ground cover)	0.1
Forest (70–100% ground cover)	0.05
Grassland	0.3
Agriculture	0.5
Water	1
Sandy Area	0.5
Rocky Area	1
Settlement	1

, scientists identified C factor values for each land use. Referring to studies reporting values for similar land cover or studies conducted in the same area or region is an easier way to determine the C factor [50]. Ozcan et al. [51], Saygin et al. [52], and Ozcan and Aytas [37] fit C factor values for forest and grassland uses because they were in the same forest region (Pinus nigra subsp. pallasiana var. pallasiana, Quercus pubescens).

Wischmeier and Smith [26], Gabriels et al. [53], Panagos et al. [31], and Ebabu et al. [13] all did the same for agricultural land uses. The C factor value of agricultural areas was found to be 0.5. The most important reason was that agricultural lands were in marginal areas.

Within the scope of the watershed improvement project, slope improvement (terraces) and gully improvement (threshold and reverse dams, etc.) measures were implemented for the P factor. A combined length of 4646 km of terraces was constructed on a land area spanning 2194 ha to mitigate slope erosion. Additionally, 13,464 check dams were built to address erosion in gullies. In this study, the 1970 P factor calculated based on engineering measures was set to 1 (RUSLE-P = 1). The P factor for 2020 was calculated using Wenner’s Method [54,55] (Equation (5)).

$$P=0.2+0.03\times S$$

(5)

In this context, S represents the percentage of slope steepness.

Several RUSLE studies have extensively used this equation [55,56,57,58]. At the scale of the large watershed, it is very difficult to represent conservation measures such as terracing, tillage, and others on the land use map [56]. In these instances, using empirical equations becomes a viable method for calculating the P factor [55]. Wenner’s method presupposes that the P factor relates to topographical features and inclination angle [34,57,59].

Erosion of a slope occurs when runoff transports soil particles and nutrients down a sloped surface. The primary causes of slope erosion are raindrop impact, sheet erosion, rill erosion, and gully erosion. Widespread erosion control techniques for slope erosion include terracing, mulching, cover crops, and contour plowing. When concentrated water flow in a well-defined channel creates deep trenches in the soil, this is called gully erosion. The mechanisms that contribute to gully erosion are headcut erosion and bank erosion. The most common erosion control methods for gully erosion are vegetation restoration, rock check dams, grading and fill placement, diversion channels, gabions, and riprap [60].

In our study, slope erosion rehabilitation engineering measures include terracing with mini excavators, geosynthetic terraces, stone cordon terraces, and afforestation. To prevent gully erosion, masonry check dams, gabion check dams, wicker check dams, stone check dams, and silt fence check dams are utilized.

3. Results

Annual soil losses over the research site, according to the RUSLE model, were estimated using RS/GIS. All parameters were converted into 10 × 10 grids and multiplied. The spatial distribution of predicted soil loss from the study area was then obtained. R and K factors were calculated independently, considering that LS, R, and K factors would not change in a short time from 1970 to 2020 when engineering measures were taken, but C and P factors would change. The RUSLE-R factor layer was calculated by Erpul et al. (2009) [41] using the DEM of the watershed. The RUSLE-R factor at the research site varied between 683.5 and 919.42 MJ ha⁻¹ mm h⁻¹ year⁻¹, with a mean value of 745.02 MJ ha⁻¹ mm h⁻¹ year⁻¹. The LS factor was calculated by Equation (4) with the DEM of the study area, accounting for the interactions between flow accumulation and topography. LS factor values ranged from 0 to 430.64, with a mean value of 73.94. Most of the study area consists of brown forest soils (85%), and their K factor values, depending on texture and soil depth (horizons), varied from 0.02 to 0.04 t ha h ha⁻¹ MJ⁻¹ mm⁻¹ and the mean value is 0.03 t ha h ha⁻¹ MJ⁻¹ mm⁻¹ (

Table 3. RUSLE variables’ parameters.

Factor	Min	Mean	Max	Std. dev
R factor [MJ mm ha−1 h−1 y−1]	683.50	745.02	919.42	51.42
K factor [t ha h ha−1 MJ−1 mm−1]	0.02	0.03	0.04	0.00
LS factor	0	73.94	430.64	61.50
C factor (1970)	0.15	0.32	1	0.14
C factor (2020)	0.05	0.28	1	0.22
P factor (2020)	0.20	0.75	1	0.37

) (Figure 3).

The C factor values are given in

Table 4. LUT/LC areas and rate of changes within the study area.

Land Use Type/ Land Cover (LUT/LC)	C Factor	1970		2020		Differences
		Area (ha)	Area (%)	Area (ha)	Area (%)	Area (ha)	Area (%)
Forest (0–30% ground cover)	0.05	3324.12	27.41	6797.56	56.05	3473.44	28.64
Forest (30–70% ground cover)	0.10	0	0	387.89	3.20	387.89	3.20
Forest (70–100% ground cover)	0.15	0	0	331.90	2.74	331.90	2.74
Grassland	0.3	5314.62	43.82	1703.58	14.05	−3611.04	−29.78
Agriculture	0.5	3395.71	28.00	2293.64	18.91	−1102.07	−9.09
Settlement	1	92.96	0.77	612.84	5.05	519.88	4.29
TOTAL		12,127.41	100.00	12,127.41	100.00	0.00	0.00

by taking the land use and cover into account according to [26,29,51,52,61]. The change in land use type/land cover (LUT/LC) gives the change in the C factor. For that reason, 1/25,000-scale forest maps from 1970 and 2020 were used.

Forest areas expanded significantly between 1970 and 2020, reaching 4193.23 ha in the watershed, representing an increase of approximately 35%, reaching a total of 7517.35 ha (61.9%). When this expansion was evaluated according to the classes of forest cover, it was determined that there was an increase of approximately 29% in forest cover with 0–30% cover (C factor 0.05) and about 6% in forest cover with 30–100% cover (C factors 0.10 and 0.15). The most significant loss of cover occurred in grassland, which decreased from approximately 5314.62 ha in 1970 to just under 1703.58 ha in 2020. Additionally, rangelands endured the greatest loss of cover, declining from about 5314 ha in 1970 to approximately 1714 ha in 2020. The residential area within the watershed experienced a notable expansion from 92.96 hectares in 1970 to 520 hectares in 2020, primarily attributed to population growth (Figure 4).

In the study conducted in the year 2020, the P factor value pertaining to the terraces in the designated study area was determined to be 0.2. Consequently, this finding indicates a reduction in the anticipated soil erosion within these regions, irrespective of any additional factors. In areas where afforestation is viable, either through machinery or human labor, the P factor may assume a value of 0.2. However, in regions with limited suitability for afforestation, the P factor remains at 1. Upon conducting a comprehensive assessment, the P factor for the entirety of the watershed was determined to be 0.75 (Figure 3).

All factors in the study area were multiplied for 1970 and 2020 to determine potential soil losses. This value decreased from 417 t ha⁻¹ year⁻¹ in 1970 to 256 t ha⁻¹ year⁻¹ in 2020 on average. Thus, 160 t ha⁻¹ year⁻¹ of potential soil loss has been prevented because of the watershed improvement projects (Figure 5). In addition, the improvement in P and C factors has prevented soil erosion in the area by an average of 1057 t ha⁻¹ year⁻¹ (%71) for fifty years (

Table 5. Predicted soil loss differences according to the RUSLE.

Factor	Min	Mean	Max	Std. dev
Soil Losses (1970) [t ha−1 year−1]	0	417.01	4265.09	398.11
Soil Losses (2020) [t ha−1 year−1]	0	256.24	3023.14	353.03
Differences [(1970)–(2020)]	−3625.32	−160.77	1768.76	272.47
RKLS (without C and P factors)	0	1474.10	10,077.12	1277.87

).

Examining the annual average total potential soil loss by land use type reveals the following:

-

Forest areas (0–30% ground cover) increased by 3473 ha; the total potential soil loss decreased by 80.45 t ha⁻¹ year⁻¹.

-

Since forest areas with 30–100% ground cover did not exist in 1970, comparisons cannot be made between them.

-

Grassland areas decreased by 3611 ha, total potential soil loss in this class has increased by 66.37 t ha⁻¹ year⁻¹.

-

Agricultural land decreased by 1102 ha over the past fifty years, and potential soil loss decreased by 97 t ha⁻¹ year⁻¹.

-

Although there was an expansion of the residential area by 520 hectares, the potential soil loss experienced a reduction of 158 t ha⁻¹ year⁻¹ (

Table 6. Potential soil losses according to land use type (1970–2020).

Land Use Type/ Land Cover (LUT/LC)	Area (ha) 1970	Soil Loss 1970	Area (ha) 2020	Soil Loss 2020	C-Value
Forest (0–30% ground cover)	3324.12	272.25	6797.56	191.8	0.15
Forest (30–70% ground cover)	0	0	387.89	88.14	0.1
Forest (70–100% ground cover)	0	0	331.9	17.35	0.05
Grassland	5314.62	506.5	1703.58	572.87	0.3
Agriculture	3395.71	422.41	2293.64	325.42	0.5
Settlement	92.96	247.89	612.84	89.92	1
TOTAL	12,127.41	1449.05	12,127.41	1285.5

).

4. Discussion

The mean value of the RUSLE-R factor at the research site is 745 MJ ha⁻¹ mm h⁻¹ year⁻¹. Due to the elevated altitude of the study area, reaching a maximum of 1451 m, the R factor values exhibited a notable increase. In contrast to the European R-factor values ranging from 0–900 MJ ha⁻¹ mm h⁻¹ year⁻¹ [62], the observed R-factor value of approximately 100 MJ ha⁻¹ mm h⁻¹ year⁻¹ in Saudi Arabia [63] indicates a notable erosion potential at the research site. Bayramin et al. [64] showed that the R factor is very remarkable for the semi-arid region of Central Anatolia. In these semiarid regions of Central Anatolia, climatic inconsistency is a significant indicator of potential risk, as extreme weather events occur, and rainy and growing seasons rarely coincide.

The spatial analysis on the LS factor revealed that the study area’s topography mainly supported erosion, meaning that steeper slopes collecting more runoffs would result in less erosion in only a small portion of the study. Although the resolution of the digital elevation model (DEM) employed to compute the LS factor within the study region is deemed sufficient at 10 × 10 m, it is worth noting that it could potentially yield elevated values in regions characterized by slopes exceeding 20 degrees [65]. The average value of the LS factor (73.94), calculated for the study area, is approximately ten times higher than Austria, which has the highest average LS factor (6.95) in the European Union [66]. The LS factor is the sole determinant contributing to greater erosion in the study area compared to Europe, irrespective of other influencing factors.

The K factor values indicate that the study area has high soil erodibility. Sandy soils with high infiltration rates have low K factor values, making sediment transport less easy. Clay soils have low K factor values due to their high resistance to soil detachment. Silt soils have high K factor values compared to others because of their high runoff rates [67]. Regarding soil erosion, inefficient agricultural and grassland practices in the watershed pose a moderate to high risk [68]. In addition, due to the semi-arid climate of the research area, the organic matter content of the soil is even lower (0.2%) than that of dry tropical forests [69]. Thus, the low organic matter content is reflected in the high RUSLE K-factor values calculated for the region.

The analysis of landscape change revealed that forest and settlement areas increased while agricultural and grassland areas decreased. In 1970, engineering and afforestation measures were implemented in the study area, particularly on slopes where labor was employed to construct terraces. The significant increase in forest coverage is mainly due to reforestation efforts in the 1980s, which resulted in a substantial expansion of forest areas and their canopy cover. Examining the LUT/LC distributions for 2020 reveals the success of the watershed improvement initiatives. Particularly by controlling grazing, the degraded forest areas were rehabilitated. As a result of the successful afforestation studies, there have been increases in forest areas and their proximity.

Moreover, the control and reduction of grazing allowed vegetation to become more diverse, and the spacing between plants increased. As a result, the proximity of recovering oak and juniper forests has slowed layer and gully erosion. Vegetation can now be observed growing within gullies and cracks, which decreases gully erosion and the amount of transported sediment.

The C-factor influences the effects of all associated cover and management variables [30]. The values of C are subject to variation within the range from nearly zero, indicating well-protected soils, to 1.5, representing surfaces with finely tilled ridges that are highly susceptible to soil erosion. On local scales in Türkiye, Ozcan et al. [29,37,51] and Ozhan et al. [61] studied only forest and grassland for the C factor. In Türkiye, Hacisalihoglu [70] determined the C variable to be 0.01 for coniferous forests and 0.07 for pastures in semi-arid regions, while Ozhan et al. [61] determined the value to be between 0.001 and 0.021 for deciduous forest and 0.13 for inner-forest glade in humid regions. Mati [71] used surface cover and canopy cover to calculate variable C and obtained a value of 0.007 for Kenyan forests, and according to Mahamud et al. [72], the C factors in the Malaysian study area range from 0.01 to 1.00. Our results are comparable to those of Hacisalihoglu [70] for forests and grasslands, while Ozhan et al. [61] found lower C values for grasslands. This difference in the C factor value could be due to variations in topography, vegetation density, and climate. Rehabilitation works such as afforestation and planting carried out in the area for various reasons can reduce soil erosion by at least 6%, and the protection of the forested regions and grasslands can aid in reducing soil erosion [73].

The P factor represents the ratio between soil loss caused by the support practice and soil loss caused by upslope and downslope tillage. The P factor, along with the C factor, is the primary driving factor among the erosion factors [74]. By modifying the flow pattern, gradient, or direction of surface runoff and decreasing runoff volume and velocity, these practices have a proportional impact on erosion [70]. P-factor values range from approximately 0.5 for reverse-slope bench terraces to 1.0 without erosion control measures [26].

Implementing afforestation initiatives aimed at converting low-slope grasslands into forested regions has resulted in a notable rise in the average erosion rate experienced by these grasslands. The study conducted by Waseem et al. [75] in India, a country with topographical similarities to Türkiye, revealed that despite two-thirds of the watershed being cultivated and vegetated, approximately one-third of the watershed experiences a significant erosion rate due to its slope exceeding 30%.

While there was an annual average loss of 417 t ha⁻¹ year⁻¹ in the research area as of 1970, it was calculated that this loss amount decreased to 256 t ha⁻¹ year⁻¹ in 2020 with the watershed improvement works. The implementation of terracing and afforestation activities as part of the improvement project has primarily targeted unproductive pastures and agricultural lands. Consequently, the erosion-induced soil loss in the area has been mitigated by an estimated reduction of approximately 40%. Without the implementation of engineering studies, which are essential for addressing erosion in the region, the estimated average soil loss would have increased significantly from 417 to 1474 t ha⁻¹ year⁻¹ within 50 years. This value is equivalent to the amount of soil carried away by erosion after a fire in Greece, which has a similar ecosystem [76].

The RUSLE method is specifically designed to forecast soil erosion in agricultural regions [77]. Therefore, its applicability may be limited to predicting erosion in various ecosystems, such as forests, scrublands, and urban areas [78,79]. Also, in some regions, it may be difficult to obtain the necessary data, making it challenging to make precise estimates [80]. Uncertainties in meteorological data, such as precipitation amount and intensity, can impact the R factor and, consequently, the accuracy of forecasts [81,82]. In addition, factors affecting erosion processes, such as soil characteristics (calculating the correct K factor) [83] and incorrect identification of vegetation, may also influence the results [84,85]. Certain parameters can be directly measured, while others are approximated. Insufficient, constrained, or untrustworthy input data may hinder the ability to make precise predictions. In regions with limited and incomplete data, the imprecise estimation of extreme values can substantially influence soil loss estimates [83,86]. Field-measured data may be insufficient or unmeasured to validate against RUSLE results.

Consequently, there may be issues with the verification and dependability of RUSLE estimates. The application of RUSLE at various scales can lead to varying outcomes [87]. Despite the possibility that different results can be calculated in some local areas of the study area, this difference was not significant enough to affect the dependability of potential soil loss results throughout the entire watershed.

5. Conclusions

Soil erosion is a worldwide environmental issue that falls under Goal 15 of the United Nations’ Sustainable Development Goals, and it holds significant importance in the context of a changing and growing global population. The Revised Universal Soil Loss Equation (RUSLE) provides an interactive and dynamic model for assessing vegetation-erosion interactions, as it incorporates physical subfactors representing different vegetation properties to calculate Soil Loss Ratios (SLR). Recently, RUSLE and GIS have gained widespread usage in calculating erosion losses in small valleys and identifying the contributing factors. This study demonstrates the significance of factors C and P in combating soil erosion, with the objective of stabilizing, halting, and reversing land degradation while promoting the sustainable use of land resources. Land use and land cover (LUT/LC) have changed in the Emine Creek Watershed because of soil conservation measures (P factor). Based on the RUSLE model, the alterations resulted in a notable decrease of 38% in the potential soil erosion within the watershed region for 50 years. If the engineering measures had not been implemented, there would have been a 353% increase in potential soil loss over a period of 50 years.

Consequently, due to enhancements in the P and C factors, the potential soil loss of the watershed after 50 years has been diminished by approximately four. This study shows how effectively applying soil conservation measures in a semi-arid and weakly vegetated area reduces potential soil loss. Further investigation is warranted to examine the reliability of projected soil erosion levels in relation to different precipitation patterns, severity conditions and support practice factor activities. In addition to focusing on the complex dynamics and interactions between R, K, LS, C, and P factors and considering cost-benefit analyses related to engineering activities carried out in marginal areas, realistic solutions can be attained.