Land Use Changes and Their Impacts on Soil Erosion in a Fragile Ecosystem of the Ethiopian Highlands

Abstract

Land cover changes have significant implications for ecosystem services, influencing agricultural productivity, soil stability, hydrological processes, and biodiversity. This study assesses the impacts of land use and land cover (LULC) change on soil erosion in the Upper Guder River catchment, Ethiopia, from 1986 to 2020. We analyzed Landsat imagery for three periods (1986, 2002, and 2020), achieving a classification accuracy of 89.21% and a kappa coefficient of 0.839. Using the Revised Universal Soil Loss Equation (RUSLE) model, we quantified spatial and temporal variations in soil erosion. Over the study period, cultivated land expanded from 51.89% to 78.40%, primarily at the expense of shrubland and grassland, which declined to 6.61% and 2.98%, respectively. Forest cover showed a modest decline, from 13.60% to 11.24%, suggesting a partial offset by reforestation efforts. Built-up areas nearly tripled, reflecting increasing anthropogenic pressure. Mean annual soil loss increased markedly from 107.63 to 172.85 t ha⁻¹ yr⁻¹, with cultivated land exhibiting the highest erosion rates (199.5 t ha⁻¹ yr⁻¹ in 2020). Severe erosion (>50 t ha⁻¹ yr⁻¹) was concentrated on steep slopes under intensive cultivation. These findings emphasize the urgent need for integrated land management strategies that stabilize erosion-prone landscapes while improving agricultural productivity and ecological resilience.

1. Introduction

Land use and land cover changes (LULCCs) are crucial in shaping global environmental dynamics and are closely tied to climate change, soil health, water resources, and ecosystem services [1,2]. Observed at various scales, LULCCs significantly affect both natural systems and human activities, altering ecosystems, hydrological cycles, and agricultural productivity [3]. Interactions between LULC changes and climate disrupt ecosystems, water cycles, and agricultural productivity, emphasizing the need for sustainable land management. In particular, soil erosion emerges as a major issue within this context, serving as both a consequence and a driver of further environmental degradation [4,5]. Deforestation, urbanization, and intensive farming worsen soil erosion, harming soil health and polluting downstream water resources, including those used by downstream dams and reservoirs [6]. This degradation of soil and hydrological instability poses severe threats to agricultural productivity, food security, and water quality, stressing the need for integrated management approaches that can reconcile ecological health with growing human demands.

At the global scale, LULC changes significantly affect carbon emissions and agricultural productivity, both of which are directly linked to climate variability and ecosystem dynamics [7]. Studies like Mwanga et al. (2024) illustrate these effects, showing that in Northern Ghana, agricultural land and built-up areas expanded by 20.8% and 27.2% over a 25-year period, leading to a 14% increase in surface runoff and greater pressure on soil and water resources [8]. Similarly, Salman et al. (2023) observed that Turkey’s vegetation cover improved from 2000 to 2021; however, climate threats such as desertification and drought persisted, showing the complex relationship between land cover and climate [9]. Furthermore, regarding the impact of urbanization, Siddik et al. (2022) reported an 80.3% increase in impervious built-up areas, which reduces groundwater recharge notably in urban regions despite minimal basin-scale changes [10]. Tesfay et al. (2022) also report changes in land allocation, with a notable decrease in cultivated land and an increase in forest land, which in turn influences ecosystem service values and soil conservation [11]. These findings show the need for continuous tracking of LULC dynamics and integrating them into strategies that address soil health and climate challenges, thereby strengthening agricultural resilience in the face of global climate change.

In Ethiopia, rapid and widespread changes in LULC are primarily driven by major population growth, increasing the demand for agricultural land to meet rising food needs and urban development [12,13]. This expansion often occurs at the expense of forest and grazing lands, which play a role in maintaining ecosystem services and soil stability [14,15]. While intended to boost food production, the reliance on extensive agricultural practices without modern technologies and inputs worsens soil degradation and fertility loss [4,16]. The conversion of natural vegetation into cultivated land, combined with practices like intensive tillage, disrupts the soil’s physical structure, leading to erosion, compaction, and reduced porosity and water infiltration rates essential for root growth and soil aeration. Over time, these physical changes degrade the soil’s chemical properties, such as nutrient availability, and biological properties, including microbial activity and organic matter content [17,18]. This gradual decline in soil health reduces agricultural productivity, resulting in low crop yields and less nutritious food. The impact extends to local communities, where poor-quality soils produce deficient crops, leading to increased malnutrition and food insecurity, particularly among vulnerable populations [19]. Without sustainable management practices, agricultural expansion into previously fertile areas continues a worsening cycle of soil degradation, reduced productivity, and increased malnutrition.

This study addresses the need to understand and reduce the effects of land use and land cover changes on soil erosion in the Upper Guder River catchment, a key tributary of the Blue Nile that significantly impacts agricultural productivity and water resources in Ethiopia and beyond, including the flow to the Grand Ethiopian Renaissance Dam. By applying high-resolution, detailed analysis over a long period, this research fills important gaps left by prior studies, providing clear insights to inform effective management and conservation strategies. The main objective is to evaluate the impact of LULC changes on soil erosion in the Upper Guder catchment from 1986 to 2020, with specific objectives to (1) analyze LULC changes, (2) estimate annual mean soil losses, and (3) map erosion severity. The findings will support targeted land management strategies and policy decisions, enhancing ecological and agricultural resilience and strengthening the stability of key water systems, including those linked to the Grand Ethiopian Renaissance Dam (GERD).

2. Materials and Methods

2.1. Study Site

The Upper Guder River catchment, covering approximately 146,644 hectares in central Ethiopia, is a region of ecological and hydrological importance. Located in the West and Southwest Shewa Zones of the Oromia Region, the catchment stretches between coordinates 37°58′33″ to 38°23′16″ E and 8°53′43″ to 8°58′46″ N (Figure 1). This diverse region includes districts such as Ameya, Wonchi, Dire Inchini, Jibat, Toke Kutayer, Ambo, Chalia, and Midakegn, each adding unique features to the landscape and land use [20].

The catchment experiences a warm climate, mainly falling within the Dega climatic zone, with some areas classified as Weina Dega. This area has a bimodal rainfall pattern, with a main rainy season from June to September and a shorter rainy period in April and May, resulting in an annual average rainfall of 1418.51 mm. Rainfall varies across the region, affecting crop production and natural vegetation patterns. Temperature changes with the seasons, with January’s minimum at 0.6 °C rising to 10.6 °C in June and July, and maximum temperatures from 27.7 °C in January reaching 30.9 °C in April and May (Figure 2).

The catchment’s landscape varies widely, with elevations ranging from 1618.7 to 3268.7 m, and is mostly slopes that strongly affect soil erosion and water flow. Approximately 70.6% of the area features slopes exceeding 5 degrees, with steep slopes (8.1–15 degrees) accounting for 20.7%, and mountainous regions with slopes above 15 degrees comprising 29.8% (

Table 1. Slope range classification and area distribution in degree.

S/N	Slope Range (Degree)	Area (ha)	Percentage (%)	Class Name
1	0–2	16,672.72	11.37	Gentle/Flat to almost flat terrain
2	2.1–5	26,705.6	18.25	Gently undulating to undulating
3	5.1–8	29,276.3	20.01	Rolling terrain/moderate
4	8.1–15	30,330.2	20.73	Steep/Hilly terrain
5	>15	43,659.2	29.84	Steeply dissected mount
Total		146,644.02	100

). The catchment’s varied topography supports a network of rivers and streams, including the Guder, Dumuga, Chole, Boji, and Sora, which are crucial for farming and local water supplies.

2.2. Dataset Collection and Source

The input data sources and tools for LULC and RUSLE analysis are presented in

Table 2. Input data sources and tools for LULC and RUSLE analysis.

S/N	Input Data Types	Sources	Details
1	Climate data	National Meteorological Agency (NMA)	Rainfall and temperature data (34 years)
2	Soil map dataset	Master Plan of Abay (2011), MOA	Digital soil maps
3	Satellite images (Landsat)	USGS Webpage	Path/Row: 169/54, 30 m resolution; Landsat-5 TM (1986, 4 bands), Landsat-7 ETM+ (2002, 7 bands), Landsat-8 OLI-TIRS (2020, 8 bands)
4	DEM (30 m)	USGS Webpage	30 m resolution SRTM used for terrain analysis and LS-factor computation.
5	Ancillary/primary data	Observations, Questionnaires, GPS-based GCPs, Google Earth	331 Ground Control Points (GCPs) across six land cover types; key informant interviews

. Briefly, the analysis of LULC changes in the Upper Guder River catchment was conducted through a detailed workflow that included image preprocessing, classification, accuracy assessment and change detection. Landsat satellite images from 1986, 2002, and 2020, all with a 30 m resolution and less than 5% cloud cover, were downloaded from the USGS Earth Explorer and processed using ERDAS Imagine and ArcGIS. Thirty-four years of climate data on rainfall and temperature were sourced from the Ethiopian National Meteorological Agency (NMA). At the same time, digital soil maps from the Ministry of Agriculture and the Abay Master Plan (2011) supported the erosion modeling (Table 2). A 30 m Digital Elevation Model (DEM) was used for analyzing the landscape. Field validation involved 241 GPSs across six land cover types, verified with high-resolution Google Earth imagery. Additionally, interviews with 20 local communities and experts gave insights into LULC drivers and erosion patterns.

2.3. Data Analysis and Modeling

The estimation of soil erosion hotspots in this study follows a systematic approach integrating multiple environmental factors using the Revised Universal Soil Loss Equation (RUSLE), as outlined in Figure 3. Each component including rainfall erosivity, soil erodibility, topographic influences, land cover, and land management was derived from spatial datasets and processed to reflect its contribution to soil erosion risk. The integration of DEM-based terrain analysis, remote sensing-derived land cover classification, and survey data ensured a comprehensive assessment of the area. By combining these factors, the model effectively highlights areas prone to severe erosion, guiding targeted conservation efforts. The following sections provide a detailed discussion of each methodological component.

2.3.1. Image Preprocessing, Classification, and Post-Classification Analysis

Satellite imagery from 1986, 2002, and 2020 served as the basis for the LULC analysis. The raw Landsat images, downloaded from USGS Earth Explorer, were preprocessed to correct distortions and match with the study area’s spatial size. Preprocessing included radiometric corrections to minimize atmospheric effects and geometric corrections to ensure spatial alignment using ERDAS IMAGINE 2015 and ArcGIS Pro 3.2 [21]. All images were re-projected from the WGS84 datum to UTM coordinates (Adindan datum), ensuring spatial data consistency across datasets. To enhance image interpretability, False Color Composites (FCCs) were generated using specific band combinations, 2-3-4 for 1986, 4-3-2 for 2002, and 5-4-3 for 2020. These FCCs showed key differences between LULC classes. Additionally, Principal Component Analysis (PCA) was used to optimize spectral separability, with PCA combinations such as 1-2-3 (RGB) enhancing differentiation between shrublands, grasslands, and settlements [22]. The preprocessed images were then classified using the Maximum Likelihood Classifier (MLC), a supervised classification algorithm based on Bayesian probability theory. MLC assumes that the spectral values of each LULC class follow a normal distribution, calculating the probability of a pixel belonging to each class. The pixel is then assigned to the class with the highest likelihood (Equation (1)).

$$P\left(i|x\right)={\displaystyle \frac{P\left(x|i\right)P\left(i\right)}{P\left(x\right)}}$$

(1)

where the posterior probability (P(i∣x)) is the chance that a pixel (x) belongs to class (i). It is calculated by multiplying the likelihood (P(x∣i)), which is the chance of pixel (x) given class (i), by the prior probability (P(i)), which is the chance of class (i). This product is then divided by (P(x)), the total probability of pixel (x) across all classes, to normalize it.

Field data, including 241 verified GPS points from the field, were used to train the model. Data were combined with high-resolution Google Earth imagery and insights from local experts and field observations to define land cover types clearly [23].

Table 3. Description of LULC classes in the Upper Guder River catchment.

S/N	Land Use Class	Descriptions
1	Bare Land	Land covered with quarries, degraded land, and roads.
2	Cultivation Land	This unit includes cropping areas, with characteristic patterns such as sharp edges between fields and houses.
3	Forest Land	Natural forests, woodlands, eucalyptus plantations, and roadside trees and fences.
4	Grass Land	Grass and herbs cover with scattered trees and shrubs, used for livestock grazing.
5	Settlement Area	Areas covered by towns, residential areas, and industries.
6	Shrub Land	Areas dominated by shrubs, bushes, and young tree species, mixed with herbaceous plants.

summarizes the six LULC classes identified: cultivated land, forest, grassland, shrubland, bare land, and settlements. This data-driven approach ensured that the classification accurately represented actual land cover. The integration of FCC, PCA, and MLC, backed by field data, created a reliable LULC dataset for change detection and further analysis.

Before conducting change detection, the accuracy of the classified images was checked to ensure the LULC results were reliable. Accuracy was checked using a confusion matrix, which compared classified land cover types with field data from 241 GPS-verified points. Metrics included producer accuracy, user accuracy, overall accuracy, and the Kappa coefficient (Equation (2)), which measures how well the classification matches reality beyond random chance [24].

$$\kappa ={\displaystyle \frac{P_o-P_e}{1-P_e}}$$

(2)

where P_o is the observed accuracy (overall accuracy); P_e is the Expected Agreement by Chance, calculated as Equation (3).

$$P_e=\sum \left({\displaystyle \frac{row total\times column total}{total {samples}^2}}\right)$$

(3)

where the Expected Agreement by Chance (P_e) is calculated by summing the product of row totals and column totals divided by the square of total samples.

The classification reached an overall accuracy of 89.21% and a Kappa coefficient of 0.839, indicating very strong agreement. Cultivated land and forest classes demonstrated the highest accuracy, due to their distinct spectral features, with producer accuracy rates of 95.56% and 91.43%, respectively (

Table 4. LULC classification accuracy and Kappa statistics.

S/N	LULC Classes	GCP	Reference Data	Classified Data	Number Correct	Producer’s Accuracy	User’s Accuracy	Kappa
1	Bare Land	15	17	15	12	70.59%	80.00%	0.7848
2	Cultivation Land	141	135	141	129	95.56%	91.49%	0.8065
3	Forest Land	35	35	35	32	91.43%	91.43%	0.8997
4	Grass Land	15	18	15	12	66.67%	80.00%	0.7839
5	Settlement Area	15	13	15	13	100.00%	86.67%	0.8591
6	Shrub Land	20	23	20	17	73.91%	85.00%	0.8342
Total		241	241	241	215	Overall Accuracy	89.21%	0.839

). Grassland and bare land classes showed slightly lower accuracy due to spectral overlaps with other land covers. Settlement areas achieved perfect producer accuracy (100%), attributed to their unique spectral and spatial patterns. These results confirm the strength of the classification process and its suitability for further analyses.

After checking accuracy, we compared the classifications to assess LULC changes over three time periods: 1986–2002, 2002–2020, and 1986–2020. This method involves layering classified images from different years and spotting changes between LULC types [25]. We used ArcGIS to analyze the changes, which helped us visualize spatial changes and identify areas prone to erosion.

2.3.2. RUSLE Model Parameterization and Analysis

The Revised Universal Soil Loss Equation (RUSLE) was used to estimate soil erosion in the Upper Guder River catchment. This model examines factors that affect soil detachment and movement, including rainfall, soil properties, topography, land cover, and conservation practices [26]. It calculates average soil loss using Equation (4).

$$A=R\times K\times LS\times C\times P$$

(4)

where A is the mean annual soil loss (t ha⁻¹ yr⁻¹), R is the rainfall erosivity factor (MJ mm ha⁻¹ hr⁻¹ yr⁻¹), K is the soil erodibility factor (t ha MJ⁻¹ mm⁻¹), LS is the slope length and steepness factor (dimensionless), C is the cover management factor (dimensionless), and P is the conservation practices factor (dimensionless).

1.

Rainfall Erosivity (R-factor)

The R-factor, which quantifies the erosive force of rainfall, was calculated using Hurni’s formula (Equation (5)) tailored for Ethiopian conditions. This factor considers the intensity and duration of rainfall events, which are critical in determining the potential for soil detachment and transport. By incorporating local rainfall data (

Table 5. Mean annual rainfall and R-factor for selected stations.

S/N	Station	Longitude	Latitude	Elevation (m)	Mean Annual Rainfall (mm)	R-Factor (MJ mm ha−1 h−1 year−1)
1	Ambo	37.839670	8.984667	2068	1044.93	579.13
2	Guder	37.757312	8.966980	2011	1347.98	749.44
3	Tikur Inchini	37.667700	8.836330	2467	1818.53	1013.89
4	Toke Erensa	37.583330	8.983330	2395	1462.63	813.87

), the R-factor provides a precise measure of the impact of precipitation on soil erosion in the Upper Guder River catchment. Table 5 shows the mean annual rainfall and R-factor for the selected stations used in this assessment, while Figure 4 illustrates the spatial distribution of rainfall erosivity (R-factor) across the study area with high R-values indicating areas with greater potential for rainfall-induced soil erosion.

$$R=-8.12+0.562P$$

(5)

where P is the mean annual rainfall (mm). Rainfall data from four stations (1986–2020) were interpolated, showing R-values from 643.52 to 1013.88 MJ mm ha⁻¹ across the catchment, highlighting zones with higher erosion risk.

2.

Soil Erodibility Factor (K-factor)

The K-factor, which measures soil erodibility, indicates the susceptibility of soil to erosion based on its texture and organic content. In the Upper Guder River catchment, the dominant soil types are Eutric Fluvisols and Chromic Luvisols, covering 23.83% and 22.50% of the area, respectively (

Table 6. Soil types and corresponding K-factor in the study area.

S/N	Soil Type	Color	Area (ha)	%	K-Factor
1	Chromic Luvisols	Brown	32,917.60	22.50	0.20
2	Chromic Vertisols	Black	7118.24	4.86	0.24
3	Eutric Cambisols	Red	16,272.00	11.23	0.25
4	Eutric Fluvisols	Yellow	34,873.20	23.83	0.30
5	Eutric Nitisols	Darker	19,494.60	13.32	0.20
6	Leptosols	Red	1339.16	0.91	0.25
7	Orthic Luvisols	Brown	29,786.20	20.36	0.20
8	Pellic Vertisols	Black	4843.02	3.31	0.24
	Total		146,644.02	100

). Eutric Fluvisols, typically found in floodplains (Figure 5), are nutrient-rich but highly prone to water erosion during heavy rains. On the other hand, Chromic Luvisols, known for their high fertility and base saturation, are particularly susceptible to erosion on slopes. The K-values, which range from 0.20 for Chromic Luvisols to 0.30 for Eutric Fluvisols, reflect these differences in erodibility, with higher values indicating greater susceptibility to erosion.

3.

Slope Steepness and Flow Accumulation (LS-factor)

The LS-factor quantifies the topographic impact on erosion. It was calculated using slope steepness and flow accumulation derived from a 30 m DEM, following the formula by [27] (Equation (6)). This model integrates the effects of slope length and steepness, which are critical in determining the velocity and volume of surface runoff, thereby influencing soil erosion rates. The LS values varied from (2.45 × 10⁻⁸) in flatter areas to 50.63 on steeper slopes, with high LS values indicating areas requiring soil conservation measures.

$$LS=({{\displaystyle \frac{Flow Accumulation }{22.13}})}^{0.4}\times {(0.01745 sin slope (in Degree)/0.8960)}^{1.4}$$

(6)

$$Flow Accumulation=\mathsf{\sum }_{i=1}^{n}Flow Input from Upstream Cells$$

(7)

The LS factor equation quantifies the combined effects of slope length and steepness on soil erosion and is used in the RUSLE model to account for topographic influences on soil loss. Flow accumulation (used in Equation (6)) was calculated using Equation (7). Flow accumulation is the cumulative water flow contributing to a grid cell based on upstream flow contributions, derived from DEM data. Figure 6 shows both the topographic and the LS factor distributions across the studied region. It illustrates the variation in erosion potential due to topographic features.

4.

Management Factor (C-factor) and Conservation Practices (P-factor)

The C-factor represents the impact of vegetation cover and land management on soil protection. Well-protected grazing lands have a low C-factor of 0.01, indicating high soil protection, while cultivated lands have a higher C-factor of 0.17, showing less protection. The P-factor evaluates the effectiveness of conservation practices based on slope. It ranges from 0.55 for flatter areas with contour farming, which significantly reduces erosion, to 1.00 for steep, unmanaged slopes, where erosion risks are highest (

Table 7. C-factor, slope, P-factor, and references for LULC classes.

LULC	Bare Land	Cultivated Land	Forest Land	Grazing Land	Settlement Area	Shrubs Land
C-Factor	0.05	0.17	0.02	0.01	0.03	0.014
Slope (%)	0–7	7–11.3	11.3–17.6	17.6–26.8	>28.8
P-Factor	0.55	0.60	0.80	0.95	1.00
References	[27,28]	[27,29]	[27,29]	[27,28]	[15,30]	[31]

).

3. Result

3.1. Land Use Land Cover Change

The analysis of satellite imagery from 1986, 2002, and 2020 reveals changes in land use and land cover in the Upper Blue Nile Basin, driven by demographic pressures, agricultural expansion, and urbanization (

Table 8. LULC distribution in the Upper Guder River catchment (1986, 2002, 2020).

S/N	LULC Classes	1986		2002		2020
		ha	%	ha	%	ha	%
1	Bare Land	1456.47	0.99	85.23	0.06	266.94	0.18
2	Cultivation Land	76,094.8	51.89	84,519.4	57.64	114,964	78.4
3	Forest Land	19,944.0	13.6	16,469.1	11.23	16,486.6	11.24
4	Grass Land	6828.57	4.66	8067.87	5.5	4367.79	2.98
5	Settlement Area	339.75	0.23	392.67	0.27	871.38	0.59
6	Shrub Land	41,980.3	28.63	37,109.7	25.31	9686.61	6.61
Total		146,644.02	100	146,644.02	100	146,644.02	100

). These changes reflect the complex interactions between human activities and natural resources over time. In 1986, cultivated land was the dominant LULC category, covering 51.89% of the basin, followed by shrubland at 28.63% and forest land at 13.60%. Grassland, bare land, and settlement areas made up smaller portions, underscoring the rural character of the region, where agriculture was the primary land use.

By 2002, the area of cultivated land had increased to 57.64%, driven by the growing population and the conversion of natural landscapes for farming. Shrubland and forest land decreased to 25.31% and 11.23%, respectively, as pressure on natural resources intensified. Settlement areas expanded slightly to 0.27%, reflecting the early stages of urbanization. These changes highlight a growing imbalance between agricultural land use and the conservation of natural ecosystems. The LULC map for 2002 further illustrates this reduction in vegetative cover, specifically forest land, shrubland, and grassland, which represent non-tilled natural vegetation crucial for soil protection and ecological stability (Figure 7). This decline emphasizes the need for sustainable land management practices.

By 2020, cultivated land had expanded to 78.40% of the area, a 35.9% increase since 2002. This rapid growth was accompanied by a decline in shrubland to 6.61% and grassland to 2.98%. Forest land remained stable at 11.24%, likely due to targeted reforestation initiatives, including the establishment of eucalyptus plantations. Settlement areas nearly tripled, reaching 0.59%, indicating accelerated urban growth and infrastructure development. Bare land, which was 0.99% of the basin in 1986, decreased to 0.18% in 2020, reflecting the success of soil and water conservation practices in reclaiming degraded areas (Table 8).

Analyzing changes over the three periods, 1986 to 2002, 2002 to 2020, and 1986 to 2020 (

Table 9. LULC changes and annual rates in the Upper Guder River catchment (1986–2020).

LULC Classes	1986 (ha)	2002 (ha)	2020 (ha)	% Change			Rate of Change (ha/yr)
				1986–2002	2002–2020	1986–2020	1986–2002	2002–2020	1986–2020
Bare Land	1456.47	85.23	266.94	−5.81	+0.66	−2.38	−85.70	+10.095	−34.98
Cultivation Land	76,094.86	84,519.4	114,964.7	+77.93	+115	+35.9	+526.53	+1691.40	+1143.23
Forest Land	19,944.07	16,469.15	16,486.6	−14.81	+0.01	−6.94	−217.18	+0.97	−101.69
Grass Land	6828.57	8067.87	4367.79	+5.20	−14	−4.94	+77.45	−205.50	−72.37
Settlement Area	339.75	392.67	871.38	+0.187	+1.83	+0.97	+3.307	+26.59	+15.64
Shrub Land	41,980.3	37,109.7	9686.61	−20.75	−103.88	−64.74	−304.41	−1523.50	−949.81

), provides deeper insights into LULC dynamics. Between 1986 and 2002, cultivated land expanded by 8424.54 ha, (526.53 ha/year), marking a 77.93% increase. This expansion came at the expense of shrubland, which decreased by 4870.6 ha (304.41 ha/year), and forest land, which declined by 3494.92 ha (217.18 ha/year). Grassland saw a modest increase of 1239.3 ha (77.45 ha/year), while settlement areas grew slightly by 52.92 ha (3.31 ha/year). Bare land decreased by 371.24 ha (85.70 ha/year) due to rehabilitation efforts.

From 2002 to 2020, the cultivated land area expanded by 30,445.3 ha (1691.40 ha/year), representing a 115% increase. During this period, shrubland decreased by 27,423.09 ha (1523.50 ha/year) and grassland by 3700.08 ha (205.50 ha/year). Settlement areas expanded substantially, increasing by 478.71 ha (26.59 ha/year), reflecting rapid urbanization. Forest land showed minimal growth, increasing by just 17.64 ha (0.97 ha/year), indicating limited but ongoing reforestation efforts.

Over the entire study period (1986–2020), cultivated land increased by 38,869.84 ha (1143.23 ha/year), transforming nearly 80% of the basin into farmland. Shrubland declined drastically by 32,293.39 ha (949.81 ha/year), while forest land and grassland decreased by 3457.47 ha (101.69 ha/year) and 2460.78 ha (72.37 ha/year), respectively. Settlement areas expanded by 531.63 ha, reflecting growing urban and infrastructure needs. Bare land showed a significant decline, shrinking by 1189.53 ha (34.98 ha/year), reclaimed mainly through conservation initiatives (Table 9).

3.2. Soil Loss Analysis and LULC Impact

The expansion of agricultural land in the catchment has impacted soil erosion rates. By combining the RUSLE model with GIS tools, we estimated annual soil loss for the years 1986, 2002, and 2020 and mapped the changes over time (Figure 8). Key factors, including rainfall erosivity (R-factor), soil erodibility (K-factor), slope length and steepness (LS-factor), land cover (C-factor), and conservation practices (P-factor), were derived from rainfall records, soil maps, DEM, and satellite imagery. Our analysis demonstrated a strong link between land use changes and the increasing severity of erosion across the catchment.

Total soil loss increased from 3.899 million tons in 1986 to 6.264 million tons in 2020 (

Table 10. Mean annual soil loss in the Upper Guder River catchment.

S/N	Study Period	Mean Soil Loss (t/ha/yr)	Total Soil Loss (t/yr)
1	1986	107.63	3.899 × 106
2	2002	144.16	5.225 × 106
3	2020	172.85	6.264 × 106

). The mean annual soil loss rose from 107.63 tons/ha/yr to 172.85 tons/ha/yr over the same period. Steeper slopes, particularly in the central-western parts of the catchment, faced the highest erosion rates. These changes are closely tied to land use and land cover dynamics, with cultivation land expanding from 1986 to 2020. This increase came at the expense of shrubland, forest land, and grassland, which are crucial for reducing soil erosion.

Between 1986 and 2002, mean annual soil loss increased to 144.16 tons/ha/year, driven by a 77.93% increase in cultivated land and declines of 20.75% and 14.81% in shrubland and forest land, respectively (

Table 11. LULC contributions to soil erosion (1986, 2002, 2020); minimum, maximum, and mean soil loss values for each LULC class.

LULC Classes	1986 (Mean t/ha/yr)			2002 (Mean t/ha/yr)			2020 (Mean t/ha/yr)
	Min	Max	Mean	Min	Max	Mean	Min	Max	Mean
Cultivation Land	0.0	12,433.7	130.3	0.0	34,426.9	212.2	0.0	6273.9	199.5
Forest Land	0.0	12,433	78.7	0.0	6006.9	69.2	0.0	9847.1	34.4
Shrub Land	0.0	12,433.7	82.3	0.0	4350.6	80.9	0.0	13,132.5	120.1
Grass Land	0.0	9400.5	79.7	0.0	34,426.9	95.5	0.0	12,433.7	170.5
Settlement Area	0.0	4047.53	83.9	0.0	4350.61	90.7	0.0	12,433.7	170.5
Bare Land	0.0	5235.3	103.2	0.0	34,426.9	196.9	0.0	34,426.9	183.3

). From 2002 to 2020, cultivated land expanded by 115%, coinciding with a 103.88% reduction in shrubland.

Erosion rates varied by LULC class, with cultivated land showing the highest rates, reaching 199.5 tons/ha/year in 2020. Forest land maintained lower erosion rates, decreasing from 78.7 tons/ha/year in 1986 to 34.4 tons/ha/year in 2020. Shrubland erosion rates increased from 82.3 tons/ha/year in 1986 to 120.1 tons/ha/year in 2020, while grassland erosion rose from 79.7 tons/ha/year to 170.5 tons/ha/year over the same period. Bare land consistently exhibited high erosion rates, averaging 183.3 tons/ha/year in 2020.

The severity of soil erosion was classified into five categories: very slight, slight, moderate, severe, and very severe, across the catchment (

Table 12. Soil loss severity classes across study periods.

S/N	Severity Classes	Severity Classes	1986 (Mean t/ha/yr)	2002 (Mean t/ha/yr)	2020 (Mean t/ha/yr)
1	Very slight	0–5	0.32	0.35	0.19
2	Slight	5–15	9.66	9.57	9.69
3	Moderate	15–30	21.85	21.64	22.09
4	Severe	30–50	34.31	38.21	39.64
5	Very severe	> 50	323.3	399.3	393.4

). The very slight category decreased from 0.32 tons/ha/yr in 1986 to 0.19 tons/ha/yr in 2020 possibly due to the implementation of conservation practices. The categories of slight and moderate erosion remained relatively stable. However, the severe and very severe categories revealed concerning trends. The very severe category increased sharply from 323.3 tons/ha/yr in 1986 to 399.3 tons/ha/yr in 2002, before slightly decreasing to 393.4 tons/ha/yr in 2020. These areas, primarily located in the central, northern, and southwestern parts of the catchment, are characterized by steep slopes, deforestation, and agricultural activities.

4. Discussion

4.1. Trends and Drives of LULC Changes

The Upper Guder River catchment has undergone significant transformations in land use and land cover from 1986 to 2020, driven by human activities, socio-economic forces, and environmental conditions. During this period, cultivated land expanded from 51.89% to 78.40% of the catchment’s total area, corresponding to an annual increase of 1143.23 ha. While this growth supports the region’s growing population and addresses food security needs, it has led to a decline in natural vegetation. Shrubland decreased from 28.63% in 1986 to just 6.61% in 2020, representing an annual loss of 949.81 ha. Similarly, grassland, which provides vital grazing resources and ecosystem services, shrank from 4.66% to 2.98% over the same period.

The expansion of cultivated land reflects the region’s dependence on agriculture to meet growing food demands [32,33]. However, this growth has come at a significant cost: the loss of shrubland, grassland, and forest land raises concerns about biodiversity, ecosystem services, and soil health. Although reforestation efforts have been limited, they demonstrate the potential to mitigate some of these losses [34]. These trends are consistent with findings in other Ethiopian watersheds [35,36], emphasizing the urgent need to balance agricultural expansion with sustainable land management practices. Without further interventions, the intensification of agriculture threatens the region’s ecological resilience and long-term productivity.

The decline in shrubland and grassland reflects the `prioritization of short-term agricultural gains over long-term ecological health [35,36]. These land cover types play crucial roles in preventing soil erosion, regulating water cycles, and maintaining biodiversity, all of which are compromised by their conversion into cultivation [30]. Such trends mirror global patterns where agricultural intensification often leads to the loss of natural landscapes, particularly in resource-limited settings [37,38]. However, the persistence of these unsustainable practices in the Upper Guder River catchment underscores the need for stronger policy frameworks and community-based conservation efforts to address these challenges.

Forest cover in the catchment showed a more complex trend, declining gradually until 2002 before stabilizing in the subsequent decades, reflecting a shift from deforestation to targeted reforestation efforts [34,38]. This relative stability is largely attributed to extensive eucalyptus plantation programs initiated to curb land degradation and meet reforestation goals [14,39]. Eucalyptus trees offer notable economic and reforestation advantages due to their rapid growth and adaptability. However, these benefits come with important ecological trade-offs. Unlike native forests, eucalyptus plantations often support lower biodiversity, deplete soil nutrients over time, and have significantly higher water demands, which may alter local hydrological balances and reduce streamflow, especially in highland ecosystems [40,41]. Consequently, while eucalyptus plays a role in reducing visible deforestation, it may not fully restore ecological integrity or sustain ecosystem services critical to long-term soil and water conservation [42]. Protecting and expanding native forest cover remains essential to improving biodiversity, soil structure, and landscape resilience, particularly in erosion-prone areas such as the Upper Guder catchment (Table 11).

Urbanization has also been an important factor driving LULC changes. Settlement areas experienced expansion between 1986 and 2020, with an annual growth rate of 15.64 ha, reflecting increasing urbanization and infrastructure development. While the total area under settlements remains relatively small, its impact is disproportionate, often encroaching on agricultural and natural lands [42]. The trend reflects rural-to-urban migration and population growth, as well as inadequate urban planning frameworks [43]. Unregulated urban sprawl not only reduces the availability of productive land but also exacerbates environmental degradation, highlighting the need for integrated urban and rural land use planning [44].

The drivers of these LULC changes are multifaceted. Population growth remains the primary driver, increasing demand for food, fuelwood, and construction materials [43,45]. Agricultural expansion has been further incentivized by economic pressures to cultivate cash crops, which often leads to the conversion of ecologically important lands. Additionally, advancements in farming technologies have enabled for greater land use intensification but have not always been accompanied by sustainable practices [12,41]. As a result, marginal lands, including steep slopes and areas prone to erosion, have been brought under cultivation, compounding soil degradation. Environmental factors such as rainfall and topography also play a role in shaping land use decisions. The high rainfall intensities observed in areas like Tikur Inchini, with an R-factor of 1013.89 MJ mm ha⁻¹ h⁻¹ year⁻¹ (Table 5), increase the susceptibility of exposed soils to erosion. Similarly, steep slopes, which account for nearly 30% of the catchment, are often targeted for cultivation despite their vulnerability. The interaction of these environmental factors with human land use practices amplifies the degradation of already fragile ecosystems.

Analysis of the catchment’s LULC trends reveals the need for integrated land management strategies. Expanding agricultural productivity to meet population demands must be balanced with the preservation of ecological health [46,47]. Sustainable practices, such as conservation agriculture, agroforestry, and reforestation can mitigate the adverse effects of land degradation while enhancing the resilience of natural systems [48]. For instance, agroforestry, which integrates trees into farming systems, can restore degraded lands, reduce soil erosion, and improve biodiversity. Reforestation efforts, particularly those focusing on native species, can further stabilize soils and enhance water retention [49].

4.2. Impacts of LULC Change on Soil Erosion

The analysis of soil loss in the Upper Guder River catchment from 1986 to 2020 reveals a significant increase in erosion rates, rising from 107.63 t/ha in 1986 to 172.85 t/ha in 2020 (Table 10). These values exceed Ethiopia’s tolerable soil loss threshold, highlighting severe land degradation in the region. The primary factor driving this escalation is the extensive conversion of shrubland and grassland into cultivated land. Between 1986 and 2020, cultivated land expanded from 51.89% to 78.40% of the catchment area, while shrubland declined. This shift stripped the landscape of protective vegetation, exposing large areas of soil to erosion, particularly in steep and highly erodible zones [50].

The severity of soil erosion is further reflected in the increasing proportion of areas classified as experiencing “very severe” erosion. In 1986, mean soil loss in this category was 323.3 t/ha/year, which rose to 399.3 t/ha/year in 2002 before slightly stabilizing at 393.4 t/ha/year in 2020. This trend highlights the compounding effects of LULC changes and biophysical factors, such as slope and soil type, on erosion dynamics. Steeply dissected terrains, which account for 29.84% of the catchment area, are particularly vulnerable to erosion when converted to agriculture. The interaction between steep slopes and intensified cultivation exacerbates runoff and sediment displacement, accelerating land degradation [41].

The RUSLE model parameters provide further insights into the mechanisms linking LULC changes to soil erosion. The C-factor, which represents land cover and management practices, indicates cultivated land (C = 0.17) as a major contributor to erosion, compared to forest (C = 0.02) or shrubland (C = 0.014). The dramatic reduction in shrubland, a land cover type that effectively mitigates erosion, has significantly increased the catchment’s vulnerability. Similarly, soil types with high K-factor values, such as Eutric Fluvisols (K = 0.30), increase erosion risks in areas where protective vegetation has been removed. These biophysical parameters underline the effect of LULC on determining erosion susceptibility. The spatial and temporal patterns of erosion observed in this study align with findings from other regions in the Ethiopian highlands, such as the Chemoga Watershed, where erosion rates average 93 t/ha/year [51]. Notably, the mean annual soil loss in the Upper Guder River catchment exceeds the national average of 100 t/ha/year [52] and is comparable to the extreme rates recorded in the Northwestern Highlands (243 t/ha/year; [53]). These comparisons emphasize the alarming scale of erosion in the study area and its implications for agricultural sustainability and water resource management.

The impacts of erosion extend beyond soil degradation to affect agricultural productivity, water quality, and reservoir capacity [50,54]. Sedimentation in downstream reservoirs, including those supporting Ethiopia’s hydropower initiatives, reduces water storage capacity and operational efficiency. Furthermore, the loss of fertile topsoil diminishes crop yields, exacerbating food insecurity in a region already under pressure from population growth and climatic variability [55,56]. Addressing these challenges requires a multifaceted approach. Reforestation, particularly with native species, can restore degraded areas and reduce runoff. Terracing and contour farming are essential for stabilizing steep slopes, while agroforestry systems can enhance soil health and biodiversity simultaneously [57]. Policies promoting conservation agriculture, including minimal tillage and cover cropping, can mitigate soil loss on cultivated lands. Strengthening institutional frameworks and community engagement will be crucial to ensuring the successful implementation of these practices.

Despite the strengths of this study, several limitations and uncertainties should be acknowledged. The use of RUSLE, while widely adopted, does not account for gully erosion or sediment transport processes, which may result in underestimations in complex terrain. Satellite imagery, although filtered for low cloud cover, may still introduce classification errors due to seasonal variations in vegetation and spatial resolution constraints. Furthermore, certain land management practices and site-specific interventions could not be fully captured due to the generalization of C and P factors across LULC classes. These methodological limitations are common in large-scale erosion assessments and underscore the importance of integrating field-based validation and dynamic modeling approaches in future research [58].

5. Conclusions

This study assessed the impacts of land use and land cover changes on soil erosion in the Upper Guder River catchment using the RUSLE model from 1986 to 2020. The findings show that (1) substantial land conversion occurred, with cultivated land expanding to over 75% of the catchment, primarily replacing shrubland, forest, and grassland. These changes were primarily driven by population growth, agricultural expansion, and inadequate policy enforcement. (2) Erosion hotspots intensified, particularly in steep slope areas where vegetation loss exposed vulnerable soils. (3) Average annual soil loss increased sharply, rising from 107.63 t/ha in 1986 to 172.85 t/ha in 2020, highlighting the compounding effects of land conversion and topographic vulnerability. (4) Soil erosion patterns were strongly influenced by terrain, rainfall variability, soil type, and vegetation cover. These results underscore the urgent need for targeted land management strategies, especially in high-risk areas. Approaches such as reforestation, terracing, and sustainable cultivation practices can significantly reduce erosion and enhance ecosystem resilience. Additionally, community participation remains vital, as local engagement enhances the sustainability and success of conservation initiatives. Policy support is equally critical. Practical interventions should include promoting agroforestry, providing incentives for soil conservation on steep lands, and integrating land use plan into local agricultural extension services. Future research should aim to improve erosion prediction by incorporating indigenous land use knowledge and informing policies that promote both agricultural productivity and ecosystem resilience.