ABSTRACT
Erosion is a worldwide threat to biodiversity conservation and agricultural yield, and it is linked to deforestation. In this study, we aim to assess soil loss in landscapes of Cachoeira River watershed, in southern Bahia, northeastern Brazil. We estimate the role of forests in diminishing soil erosion using the Revised Universal Soil Loss Equation (RUSLE). We compare real and simulated scenarios in which the forest was replaced by agricultural use, also comparing estimates of erosivity factor (R factor) derived from remote sensing and climatological station data. Real and simulated annual soil losses varied from 0 to 15.95 t/year and from 0 to 33.53 t/year along the watershed, respectively. However, only 0.04 and 1.67% of this area is highly and severely exposed to erosion, using data from climatological stations and remote sensing, respectively. Soil loss in the simulated deforested scenario was approximately two times higher than the real annual soil loss, indicating the importance of forest cover to mitigate soil erosion. Moreover, soil loss was 10.5 times greater when using precipitation data from remote sensing compared to climatological stations. Conclusively, the practice of agroforestry can be used as an alternative to avoid erosion.
HIGHLIGHTS
This paper assessed soil erosion in an important river in Brazil.
Real and simulated scenarios of soil loss were discussed in this study.
Soil loss was 10.5 times greater when using precipitation data from remote sensing compared to climatological stations.
Soil loss is twice as high in non-forested areas as in forested areas.
INTRODUCTION
Forests are crucial for safeguarding environmental sustainability and, thus, attaining key sustainable development goals. Unfortunately, the current pace at which forests are lost indicates that this process is the leading cause of biodiversity degradation and loss (Magioli & Ferraz 2021). Most importantly, such loss also leads to decay in the portfolio of services and goods provided by nature – the so-called ecosystem services – that ultimately contribute to human wellbeing. Soils are vital components of the biosphere, providing resources necessary to sustain food production, nutrient cycling, water storage, and habitats that host an amazing diversity of life on earth (Schröder et al. 2016; De Deyn & Kooistra 2021). Its presence and quality are closely linked with land use changes, with deforestation being largely responsible for soil erosion, i.e., when soil is carried away and lost. In fact, soil erosion is a global threat (Montanarella et al. 2016; Arabameri et al. 2020), particularly affecting agricultural productivity and water quality (Benavidez et al. 2018; Sidi Almouctar et al. 2021). In particular, the transport of nutrients from agricultural lands to river results in water contamination and eutrophication. Nutrients can be transported through river networks from higher to lower elevations (Sarker 2021), often influenced by external factors such as climate change and tectonics (Sarker et al. 2019). Soil type plays a crucial role in both soil water retention and nutrient transport via runoff. The interplay of soil type, slope steepness, and slope length significantly affects nutrient runoff from forested areas, influencing the watershed outlet or groundwater recharge. Indeed, water erosion is globally responsible for the loss of 75 billion metric tons of soil every year (Dabral et al. 2008), causing a loss of 8 billion dollars in the global economy due to the erosion of fertile land and a reduction in agricultural products (FAO 2019). Although soil erosion can be a natural process, human activities, including non-sustainable agricultural practices, implementation of infrastructure, and other forms of land use changes are exacerbating it. Globally, 84% of the land is degraded by soil erosion (Opeyemi et al. 2019), negatively impacting the wellbeing of more than 3.2 billion people (IPBES 2018; Borrelli et al. 2021). This major environmental and agricultural threat (Pimentel et al. 1995) mainly occurs in agricultural areas, particularly on higher slopes (Salem et al. 2014; Ibrahim et al. 2019), with the destruction of vegetation cover increasing surface runoff and decreasing infiltration (Prosser & Williams 1998).
In Brazil, a country with an economy highly dependent on agricultural production, soil erosion is a major concern. Each year, Brazil experiences an estimated loss of 800 million metric tons of soil (Merten & Minella 2013), placing the country among the global erosion ‘hotspots’ (Guerra et al. 2014). Deforestation is closely linked with runoff and erosion issues. Approximately 70% of the country's population, roughly 140 million people, according to Scarano & Ceotto (2015), reside in the 17 states that were originally part of the Atlantic Forest biome. This intense human presence has significantly diminished this once expansive biome, which was formerly the second-largest tropical forest in the Americas (Tabarelli et al. 2005), now reduced to less than 12% of its original size. While this extensive deforestation likely contributes to increased soil erosion (Zwiener et al. 2017), certain land management practices like agroforestry might alleviate some of these environmental impacts. For instance, in the southern state of Bahia, northeastern Brazil, portions of the native forests have been replaced with cacao (Theobroma cacao) agroforests. This type of land use is known for its biodiversity and climate-friendly attributes. However, the effectiveness of such practices in mitigating soil erosion remains a topic that requires further investigation. Herein we aimed to contribute to this discussion by estimating the soil loss associated with erosion in a specific municipality within agroforestry dominated landscapes in the southern state of Bahia. The study area lies within the largest forest remnants in northeastern Brazil (Joly et al. 2014), but most of the forest cover in this region comprises cacao agroforests, comprising a forest mosaic hosting most of the rich native biota (Schroth et al. 2011). This study also presents an opportunity to compare soil loss assessed using two different data sources, namely remote sensing and climatological stations. Specifically, we raised the following questions: (i) How much soil is lost when non-forested areas are compared to forested areas – comprising the above-mentioned mosaic of native stands and agroforests – and when forest areas are replaced by agricultural areas? (ii) What is the role of rain-induced soil erosion, and (iii) whether such soil loss estimates differ when using remote sensing and climatological station data for the soil erosivity (R factor)?
METHODOLOGY
Study area and land use mapping
Within Ilhéus municipality, our study encompasses the Cachoeira River watershed, covered by different vegetation classes, including pasture, mosaic cultures, and grassland. The Cachoeira River watershed originates in the Ouricana mountain range in the city of Itororó and meets the Salgado River before passing along the coast of the municipality of Ilhéus (Torres et al. 2001). It is limited to the north by basins of the Rios de Contas and Almada, to the south by basins of the Pardo and Una Rivers, to the west by the Rio Pardo basin, and to the east by the Atlantic Ocean (Trindade et al. 2010).
Modeling
Land use maps from Mapbiomas (https://mapbiomas.org) were used for automatic classifications to identify the spatial distribution of the following classes: forests (including native stands and cacao agroforestry systems), pastures, forestry, open areas, urban centers, exposed soil, and water bodies (e.g., rivers, ponds). Using the QGIS tools we converted vector images to shapefile layers with a 30 m resolution. Two scenarios were considered: real and simulated. In the real scenario, the actual land use land cover (LULC) was used, whereas in the simulated scenario, the forest class was replaced by the agricultural use.
RUSLE model description
Rainfall erosivity (R factor)
The R factor is the power of rain to cause soil erosion by water (Cardoso et al. 2020). It is the mathematical expression of erosive power used to determine the average impact of rainfall and runoff on soil erosion in a specific locality and period (Farhan et al. 2013). We calculated R factor using two sources of data: (1) Using Google Earth Engine to generate the mean precipitation values in our study area between 1992 and 2021 and (2) actual precipitation data provided directly from climatological stations (Table 1). Notably, the Google Earth Engine (GEE)-based R factor was derived using the Kriging interpolation method, whereas we used the Inverse Distance Weighted interpolation for the R factor based on climatological station data.
Coordinates of climatological stations used in this study
Climatological station . | Code . | Latitude . | Longitude . | Period . |
---|---|---|---|---|
Salvador (Ondina) | 83229 | −13.00583333 | −38.50583333 | 1992–2021 |
Vitoria da Conquista | 83344 | −14.88638888 | −40.80138888 | 1992–2021 |
Salinas | 83441 | −16.154862 | −42.284921 | 1992–2021 |
Vitoria | 83648 | −20.31583333 | −40.31694443 | 1992–2021 |
Guaratinga | 83446 | −16.58081 | −39.783182 | 1992–2021 |
Caravelas | 83498 | −17.73944444 | −39.25861111 | 1992–2021 |
Itamarandiba | 83488 | −17.85972222 | −42.85277777 | 1992–2021 |
Climatological station . | Code . | Latitude . | Longitude . | Period . |
---|---|---|---|---|
Salvador (Ondina) | 83229 | −13.00583333 | −38.50583333 | 1992–2021 |
Vitoria da Conquista | 83344 | −14.88638888 | −40.80138888 | 1992–2021 |
Salinas | 83441 | −16.154862 | −42.284921 | 1992–2021 |
Vitoria | 83648 | −20.31583333 | −40.31694443 | 1992–2021 |
Guaratinga | 83446 | −16.58081 | −39.783182 | 1992–2021 |
Caravelas | 83498 | −17.73944444 | −39.25861111 | 1992–2021 |
Itamarandiba | 83488 | −17.85972222 | −42.85277777 | 1992–2021 |
EI is the monthly mean erosion index (MJ mm ha−1 h year);
r is the average monthly precipitation (mm); P is the average accumulated annual precipitation (mm). Based on previous studies, the calculation of the R factor in this study relied on daily to annual data for extrapolation in space and time (Meusburger et al. 2012; Diodato et al. 2017).
Soil erodibility (K factor)
The K factor determines the natural susceptibility to erosion, which is particularly influenced by soil's properties, such as texture, permeability, shear strength, organic matter, and chemical composition (Efthimiou 2020). The K factor is determined by two key factors: the first is the soil's infiltration capacity to withstand detachment and transport by rainfall, and the second is the runoff process (Wischmeier & Mannering 1969). Obtaining K factor values is challenging due to the need for long-term field experiments under natural precipitation conditions (Schick et al. 2014). For this reason, we utilized a soil map provided by Mapbiomas to identify soil types prior to gathering K factor values from the scientific literature. As these values particularly depend on soil classification (Podhrázská et al. 2015) and pedology maps, the K value for each soil type was inserted into the attribute table in the soil map layer, thus converting this vector layer into a raster type. The higher the coefficient of soil erodibility, the higher the soil loss.
Topographical factor (LS)


Vegetation cover, the management factor (C) and erosion control practice factor (P)
Vegetation cover and the management factor (C) are dimensionless indexes that depend on the extent to which soils are covered by vegetation, as well as the land topography and previous models of land use (Renard et al. 1997). Similarly, the P factor is a dimensionless index that accounts for control practices that reduce the erosion potential of runoff (Renard et al. 1997). It expresses the ratio between expected soil erosion for a given soil conservation practice for up and down tillage (Wischmeier & Smith 1978). Herein the C and P factors were lumped together (as CP), with values ranging from 0 to 1, where 0 corresponds to soil with a high vegetation cover and 1 means the soil is entirely exposed to erosion (Ajibade et al. 2020). The LULC map for 2021 was used to extract information on vegetation class, which was then searched for in the scientific literature to find the values of CP.
RESULTS AND DISCUSSION
R factor based on remote sensing and climatological station data
(a) R factor map based on precipitation data from remote sensing and (b) R factor map based on climatological stations.
(a) R factor map based on precipitation data from remote sensing and (b) R factor map based on climatological stations.
The results displayed in Figure 3 indicate that the values of the R factor are influenced by the sources of precipitation. Similarly, findings from a previous study confirmed that soil erosion is driven by the R factor, particularly during the rainy season (Yao et al. 2016). Notably, this study corroborates a previous finding that the R factor greatly influences the soil erosion potential in a GEE-based study (Sud et al. 2024).
Soil erodibility (K factor)
The values of K factors of the Cachoeira River watershed
Soil types . | K factor . | References . |
---|---|---|
Chromic Luvisols | 0.024 | Embrapaa |
Thiomorphic Gleysoil | 0.004 | Embrapa |
Humiluvic Spodosol | 0.03 | Embrapa |
Dystrophic Yellow Latosol | 0.017 | Embrapa |
Dystrophic Red Latosol | 0.023 | Pereira & Cabral (2021) |
Soil types . | K factor . | References . |
---|---|---|
Chromic Luvisols | 0.024 | Embrapaa |
Thiomorphic Gleysoil | 0.004 | Embrapa |
Humiluvic Spodosol | 0.03 | Embrapa |
Dystrophic Yellow Latosol | 0.017 | Embrapa |
Dystrophic Red Latosol | 0.023 | Pereira & Cabral (2021) |
Note: Embrapa is part of the national agricultural research system in Brazil.
Table 2 describes the values of the K factor along with different soil types in the study areas. The values of the K factor varied from 0.004 to 0.03, which explained that the degrees of limitation of K factor varied from null to moderate (Giboshi 1999).
Topographic factor (LS)
The LS factor shows how slope length and steepness influence the process of soil erosion (Kadam et al. 2018). In this study, the LS factor varied from 0.03 to 12.53. It is noted that higher LS values led to higher soil erosion (Fayas et al. 2019).
CP factor
(a) Real CP factor map and (b) simulated CP factor map of the Cachoeira River watershed.
(a) Real CP factor map and (b) simulated CP factor map of the Cachoeira River watershed.
CP values are low as they are close to zero. This means that the soil is covered with vegetation, which is important for reducing the erosion risk. This result is supported by previous studies that have highlighted how agroforestry improves physical, biological, and chemical properties (Fahad et al. 2022; François et al. 2023). The real and simulated CP factors had different effects on soil loss (Figures 5 (a) & (b)). Therefore, the utilization of cacao agroforestry systems can be a promising alternative for soil erosion reduction and soil conservation in the study areas. It is noted that the LULC has a low contribution to soil erosion in the study area and, therefore, does not represent a potential risk for erosion.
Annual soil loss
The case of the soil in the Cachoeira River watershed using data from remote sensing
(a) Real soil loss map and (b) simulated soil loss based on precipitation data from remote sensing.
(a) Real soil loss map and (b) simulated soil loss based on precipitation data from remote sensing.
The highest value in t/ha/year of the simulated scenario of soil loss was 2.1 times greater than that of the actual soil loss scenario in the Cachoeira River watershed. This suggests that forests/agroforests can significantly contribute to soil protection against erosion. The results of this study show the sensitivity of CP values used in soil loss assessment, in which the soil erosion rate is categorized as very high. Comparably, the soil loss values of this study are greater than those of a study in Nigeria, where soil erosion ranged from 0 to > 756.60 t/ha/year (Opeyemi et al. 2019) and in the extreme south of Bahia, Teixeira de Freitas, with Argisol, which reported a total soil loss of 686.85 Mg ha−1 (Guimaraes et al. 2017). Another study conducted in the municipality of Teixeira de Freitas, southern Bahia, indicated that the area covered by eucalyptus had significantly less soil erosion compared to the uncovered soil area (Ferreira et al. 2019). As expected, simulated CP factors led to higher soil erosion than the real scenario. In this study, the topography seems not to have a great influence on soil erosion. Kadam et al. (2018) highlighted that the higher the slope area, the more intense the runoff and flow produced.
The case of the soil in the Cachoeira River watershed using data from climatological stations
(a) Real soil loss map and (b) simulated soil loss based on precipitation data from climatological stations.
(a) Real soil loss map and (b) simulated soil loss based on precipitation data from climatological stations.
Based on the findings of Figures 6(a) & 7(a), soil erosion is influenced by the precipitation source used for R factor calculation. Notably, higher precipitation values resulted in greater soil erosion, as evidenced by the fact that precipitation data from remote sensing sources led to greater soil erosion compared to data from climatological stations. Importantly, it was found that the mean soil erosion calculated using data from remote sensing sources was 10.5 times greater than that assessed using climatological precipitation data. The findings of this study corroborate the results of another study, indicating that higher R values lead to higher soil erosion (Fayas et al. 2019). However, at this stage, we cannot assert that one precipitation source is better than the other. This result only indicates that the soil loss estimation is a nominal value that can depend on the source of precipitation data used for the R factor calculation.
Severity of soil loss in the Cachoeira River watershed when using data from remote sensing and climatological stations
In terms of classes of soil loss severity, only the real scenarios were considered in this study. It was discovered that the erosion risk ranged from none or slight to very high in the Cachoeira River watershed when precipitation data from remote sensing sources were used. Soil erosion is classified as very low, low, moderate, high, very high, and severe (Table 3), according to the classification used by İrvem et al. (2007). Most of the areas were classified as having no or slight soil erosion. Notably, more than 85% of this river watershed had low soil erosion. Only 1.67% of the Cachoeira River watershed experienced severe erosion (Table 3). Similarly, 96.06 and 0.04% of this river watershed experienced very low and very high soil erosion, respectively (Table 3).
Severity range and severity class of soil loss in the Cachoeira River watershed when using remote sensing and climatological station data
Severity range . | Severity class . | Cachoeira River watershed/R factor data from remote sensing . | Cachoeira River watershed/R factor data from climatological stations . | ||
---|---|---|---|---|---|
Area (ha) . | Area (%) . | Area (ha) . | Area (%) . | ||
≤5 | Very low | 21,807.09 | 88.67 | 23,671.35 | 96.06 |
5–12 | Low | 1,385.28 | 5.63 | 342.36 | 1.39 |
12–50 | Moderate | 462.87 | 1.88 | 531.54 | 2.16 |
50–100 | High | 246.87 | 1.00 | 85.59 | 0.35 |
100–200 | Very high | 305.64 | 1.24 | 10.35 | 0.04 |
>200 | Severe | 412.56 | 1.67 | — | — |
Severity range . | Severity class . | Cachoeira River watershed/R factor data from remote sensing . | Cachoeira River watershed/R factor data from climatological stations . | ||
---|---|---|---|---|---|
Area (ha) . | Area (%) . | Area (ha) . | Area (%) . | ||
≤5 | Very low | 21,807.09 | 88.67 | 23,671.35 | 96.06 |
5–12 | Low | 1,385.28 | 5.63 | 342.36 | 1.39 |
12–50 | Moderate | 462.87 | 1.88 | 531.54 | 2.16 |
50–100 | High | 246.87 | 1.00 | 85.59 | 0.35 |
100–200 | Very high | 305.64 | 1.24 | 10.35 | 0.04 |
>200 | Severe | 412.56 | 1.67 | — | — |
The areas of high, very high, and severe vulnerability along the Cachoeira River watershed were 246.87, 305.64, and 412.56 ha, respectively (Table 3). These vulnerabilities are likely associated with agricultural practices. Based on the results displayed in Table 3, it is noted that the precipitation source used for the R factor in the RUSLE model affects the severity class of soil erosion.
FUTURE RESEARCH AND LIMITATIONS
Soil erosion assessment is crucial for environmental management and ecosystem services. Various factors, including the LS factor and CP factor, can be sources of uncertainty in soil erosion rate estimation. A previous study indicated that errors and uncertainties in the DEM influence soil erosion rate estimation (Abd Aziz et al. 2012). The LS factor influences soil loss over space and time (Wang et al. 2002) due to overlaying in a geospatial environment (Luvai et al. 2022). Similarly, Anjitha Krishna et al. (2019) underlined that the major uncertainty in soil loss assessment is derived from the LS factor. These challenges can be overcome by analyzing the global sensitivity of RUSLE and using different equations for LS factor calculation to be used in the RUSLE model. Additionally, DEM from different sources, including the Advanced Spaceborne Thermal Emission and Reflection Radiometer, with multiple resolution grids, can be used to partially clarify these issues. A previous study outlined that different DEM sources and resampled grid sizes led to different soil loss estimations (Pandey et al. 2021). Moreover, the input model and method used for soil loss estimation can both be sources of uncertainty (Schürz et al. 2020). Furthermore, another limitation of this study is that K and CP values were collected from scientific literature. These factors may influence the soil erosion rate, as it is better to collect data from the study field. Importantly, the utilization of RUSLE model may overestimate soil loss and has some limitations as it does not predict sediment yield or other elements, including gullies (Croke & Nethery 2006). Further research is needed to collect data from the study field for soil erosion assessment while performing global sensitivity and uncertainty analyses of the RUSLE model.
IMPLICATION OF THIS STUDY FOR ENVIRONMENTAL PROTECTION
Anthropogenic activities (e.g., deforestation) are among the drivers causing climate change, erosion, and contamination of natural resources (e.g., water and air quality). The world today is about 0.85 °C higher compared to the average temperatures from the period 1950–1980 (Sarker 2022). In terms of environmental studies, this study shows that forests can not only be crucial for reducing soil erosion but also for mitigating climate change through carbon sequestration. In other words, the findings of this study demonstrate that forest restoration, afforestation, or tree plantations can serve as techniques for soil conservation and reclaiming damaged lands. Additionally, forests are fundamental for providing a range of services, including preserving biodiversity, wildlife habitat, and freshwater supply (Jackson et al. 2008). Indeed, this study demonstrates that anticipatory planning of reforestation is a plausible alternative that can greatly contribute to environmental protection and sustainable soil management.
CONCLUSION
Managing soil is crucial for soil conservation, environmental sustainability, sustainable development, and human wellbeing. This study was carried out in the Cachoeira River watershed in the Atlantic Forest in Bahia State, northeastern Brazil, to assess soil erosion using the RUSLE model. Soil erosion ranged from 0 to 167.87 t/year in the real scenario in the Cachoeira River watershed, whereas it ranged from 0 to 351.81 t/year in the simulated scenario when using remote sensing data. Conversely, when using data from climatological stations, soil losses ranged from 0 to 15.95 t/year in the real scenario and from 0 to 33.53 t/year in the simulated scenario. The novelty of this study is that soil erosion in simulated scenario was two times higher than in the real scenario. Similarly, soil erosion calculated using precipitation data from remote sensing for R factor calculation was 10.5 times higher than the soil loss assessed when climatological stations were used for the calculation of the R factor. Consequently, the sources of precipitation data used for R factor calculation are one of the parameters that can significantly influence soil erosion assessment. These findings highlight the importance of forests for soil conservation and the choice of precipitation source used for R factor calculation. In conclusion, the findings of this study may assist decision-makers in adopting a more effective approach to conservation and ecosystem services in the Cachoeira River watershed.
ACKNOWLEDGEMENTS
This study was financed in part by the Brazilian National Council for Scientific and Technological Development (CNPq Grant 142018/2020-1).
AUTHOR CONTRIBUTIONS
M.F. contributed to conceptualization, methodology, design, material preparation, data curation, writing-original draft, writing-review and editing, validation. E.M.-N. contributed to conceptualization, methodology, supervision, writing-review and editing, validation. D.F. contributed to conceptualization, supervision, writing-review and editing, validation. M.C.G.P., U.C.d.O., and H.P.d.S. contributed to writing-review and editing, validation. R.N.d.V. contributed to data curation, writing-review and editing, validation.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.