Spatial Representation of Soil Erosion and Vegetation Affected by a Forest Fire in the Sierra de Francia (Spain) Using RUSLE and NDVI

Abstract

Extreme weather events are increasing the frequency and intensity of forest fires, generating serious environmental and socio-economic impacts. These fires cause soil loss through erosion, organic matter depletion, increased surface runoff and the release of greenhouse gases, intensifying climate change. They also affect biodiversity, terrestrial and aquatic ecosystems, and soil quality. The assessment of forest fires by remote sensing, such as the use of the Normalised Difference Vegetation Index (NDVI), allows rapid analysis of damaged areas, monitoring of vegetation changes and the design of restoration strategies. On the other hand, models such as RUSLE are key tools for calculating soil erosion and planning conservation measures. A study of the impacts on soils and vegetation in the south of Salamanca, where one of the worst fires in the province took place in 2022, has been carried out using RUSLE and NDVI models, respectively. The study confirms that fires significantly affect soil properties, increase erosion and hinder vegetation recovery, highlighting the need for effective restoration strategies. It was observed that erosion intensifies after fires (the maximum rate of soil loss before is 1551.85 t/ha/year, while after it is 4899.42 t/ha/year) especially in areas with steeper slopes, which increases soil vulnerability, according to the RUSLE model. The NDVI showed a decrease in vegetation recovery in the most affected areas (with a maximum value of 0.3085 after the event and 0.4677 before), indicating a slow regeneration process. The generation of detailed cartographies is essential to identify critical areas and prioritise conservation actions. Furthermore, the study highlights the importance of implementing restoration measures, designing sustainable agricultural strategies and developing environmental policies focused on the mitigation of land degradation and the recovery of fire-affected ecosystems.

1. Introduction

Nowadays, climatic phenomena, such as heat waves or drought, are becoming more extreme and more frequent. This, when coupled with the abandonment of rural areas and the consequent accumulation of fuel, increases the incidence, severity and scale of fires. Moreover, recent studies have shown that there is an increasing trend in the number of forest fires in European countries [1,2,3,4]. Thus, fire-induced environmental changes, such as increased erosion, are expected to increase due to increased fire activity and intensified precipitation [5,6,7,8].

Socioeconomic changes and territorial management, along with variations in population density and the advent of climate change, have altered the frequency, intensity, and seasonality of wildfires, often pushing them beyond their historical range, with severe consequences for biodiversity [9].

Wildfires generate a range of impacts, spanning economic, environmental, social, and cultural domains [10], and affecting not only flora and fauna but also water and the physical, chemical, and biological structure of soils [11]. Moreover, it is often underestimated that these fires significantly contribute to environmental pollution, releasing large amounts of greenhouse gases and particles during combustio0n, which intensifies climate change.

A proper assessment of wildfires is crucial to understanding their impact on the environment, ecosystems, and society. It is essential to analyse the extent and geographic distribution of damaged areas shortly after a fire occurs, as this helps identify regions re-quiring management actions to minimize environmental effects and support ecosystem restoration [12,13,14]. Traditional methods for assessing wildfires are resource-intensive and offer limited coverage. In contrast, satellite remote sensing enables efficient and precise analysis of large areas, facilitating the identification of burned areas and monitoring fire spread. This technology is a key tool for monitoring, immediate response, and recovery after fires, utilizing spectral changes observed in forests before and after the event [15,16].

Soil losses due to erosion are a significant concern in areas affected by high-severity wildfires. These areas are characterized by a decrease in organic matter content, an in-crease in soil water repellence, and reduced soil aggregate stability, which decrease infiltration and lead to increased surface runoff [2,3,4]. These fire effects result in negative con-sequences for both terrestrial ecosystems (loss of soil fertility) and aquatic ecosystems (flooding) [1,17].

Likewise, post-fire erosion represents one of the main difficulties for the recovery of ecosystems, affecting the recolonisation of vegetation and also the stability of the landscape. This is due, in the first place, to the loss of topsoil, which is rich in nutrients, making it difficult for vegetation to re-establish itself and altering the stability of the landscape. Without the vegetation cover that retains the soil and regulates water infiltration, runoff increases, favouring water erosion and generating landslides that modify the topography of the land. In addition, seed removal prevents natural regeneration, while colonisation by invasive species leads to competition with native vegetation, further delaying recovery [3,4,17].

Additionally, soil erosion models are key tools for evaluating the potential impact of wildfires on soil loss and designing stabilization strategies [18]. One of the most effective tools for calculating annual soil losses is the RUSLE model, which requires fewer field data compared with other more complex models. The RUSLE model has been extensively evaluated under various conditions [19,20,21,22,23,24]. This model is crucial for understanding how vegetation loss weakens the soil, making it more susceptible to erosion by rainfall, which facilitates nutrient loss and degradation [25]. Erosion prediction becomes a valuable tool for soil management and conservation, enabling the implementation of more effective protection strategies [26].

In this study, we applied the RUSLE model to assess annual erosional losses. RUSLE was chosen because it is based on the use of digital terrain models and geospatial analysis, using remote sensing and ArcGIS. It is considered one of the most successful and widely used models due to its simplicity and data availability, which makes it widely used in erosional loss estimation studies [27,28,29,30]. The cartographies obtained using this model can serve as a useful input with which to establish planning strategies and develop soil erosion prevention and control plans.

This study will focus on the analysis of one of these major wildfires, specifically the one that began on 11 July 2022 in the municipality of Ladrillar (Cáceres), caused by lightning. The flames spread across the border into the province of Salamanca, affecting a total of 8622.65 hectares there (Figure 1), with most of the affected area being forested land. This was the most devastating fire recorded in the province of Salamanca [31].

Remote sensing technologies are commonly used to evaluate vegetation conditions, with the Normalised Difference Vegetation Index (NDVI) being the most relevant today [32]. This index is widely used to monitor seasonal vegetation changes, as its value is directly related to the amount of vegetation present [33]. NDVI allows for the assessment of vegetation health and density before and after a wildfire, observing the relationship be-tween wavelengths in the red and near-infrared bands [32]. By comparing NDVI values before and after the fire, it is possible to determine the level of vegetation cover loss and its recovery capacity, providing a comprehensive view of the ecological impact and resilience of the affected ecosystem.

In this work, these spectral index techniques have been used because they are very useful for understanding fire in an ecosystem. In addition, the NDVI, being one of the most used methods in forest fire studies by which to monitor burned areas, also serves to assess changes in the water content of vegetation. In this way, spectral indices are key to the spatial assessment of the complex effect of fire on ecosystems, having been applied in several studies of this type [34,35,36,37,38,39].

The main objective of the article is to analyse the effects of the forest fire that occurred in the autonomous community of Castilla y León in 2022, specifically in the town of Monsagro, on soils and vegetation. To this end, the impacts on different components of the environment, such as flora and soil will be examined, with a particular focus on the physico-chemical characteristics of the soil. In addition, we seek to undertake the following:

-

To analyse the variation in organic matter content, sands, clays, silts and pH of the soils affected by the fire, before and after the event.

-

To draw up cartographies reflecting the evolution of the soil–flora dynamics resulting from this event.

-

To calculate the percentage of erosion in the study area, using the data obtained in the analyses carried out before and after the fire, thus making it possible to assess whether the consequences of the fire have led to an increase in soil erosion.

Finally, this is a novel study in the Sierra de Francia, as there is currently no research combining satellite imagery and RUSLE in this area. Moreover, given that the recovery of these areas can take decades, such a study is essential when seeking to implement restoration strategies such as reforestation with native species, soil stabilisation and the implementation of erosion barriers, acceleration of regeneration and a minimisation of negative effects on the landscape. Moreover, in terms of the Sustainable Development Goals (SDGs) this study relates to SDG 6 ‘Clean Water and Sanitation’ (erosion affects water quality, while soil restoration can reduce pollution and improve soil structure) and SDG 15 ‘Life of Terrestrial Ecosystems’ (erosion hinders the regeneration of terrestrial ecosystems and soil restoration and reforestation contribute to the conservation and sustainable use of ecosystems).

2. Materials and Methods

2.1. Study Area

The study area is located in the southwest of the province of Salamanca (Spain), which covers an area of 12,350 km². The geography of the province varies from broad plains in the central and eastern areas, interrupted to the south by mountain ranges and to the northwest by the canyons of the Duero River and its tributaries, with an altitudinal variation ranging from 116 m in the Salto de Saucelle valley in the Arribes Natural Park to 2428 m at the peak of Canchal de la Ceja in the Sierra de Béjar.

The province also has five protected natural areas with a total surface area of 210,282 hectares, including the Arribes del Duero and Las Batuecas-Sierra de Francia Natural Parks, the El Rebollar natural area and the Candelario and Las Quilamas mountain rang-es. The province has a continentalised Mediterranean climate, characterized by cold, semi-humid winters and hot, dry summers [40].

The forest fire that is the subject of this study occurred mostly in this province, specifically in the municipality of Monsagro, located in its southern part. Although the total number of hectares burned in the province of Salamanca was 8622.65 in July 2022.

According to the Monsagro Fire Restoration Plan [41], most of the forest area affected by the fire was made up of resin pine (Pinus pinaster Aiton), with a notable presence of Scots pine (Pinus sylvestris L.), holm oak (Quercus ilex L.), Pyrenean oak (Quercus pyrenaica Willd.), and with a lower density of cork oak (Quercus suber L.), alder (Alnus glutinosa L.), willow (Salix spp.) and strawberry tree (Arbutus unedo L.). All of this accounted for 56.76% of the total area of the fire in woodland, while 14.6% was scrubland and 13.67% crops.

In terms of lithology, the substrates found are of Palaeozoic origin, with metamorphic soils such as slate and quartzite predominating in the centre and west of the burnt area, and granitic soils in the eastern part. However, in the north, where the agricultural land is located, the origin is Cenozoic, composed of clays, silts and sandstones. On top of these rocks there are different soils such as Cambrian Umbrisols, Eutric Luvisols and Eutric Al-isols.

Furthermore, the restoration plan indicates that the area affected in the province of Salamanca by this fire includes a Protected Natural Area, the Las Batuecas-Sierra de Francia Natural Park, including the reserve area (966.78 ha), the compatible use area (10.08 ha) and the limited use area (785.44 ha). The fire covered areas included in the Natura 2000 Network, such as the Special Conservation Area (SCA) Las Batuecas-Sierra de Francia, the SCA along the banks of the rivers Huebra, Yeltes, Uces and tributaries, and the Special Protection Area for Birds (SPA) Las Batuecas-Sierra de Francia. It also affects the Sierra de Béjar y Francia Biosphere Reserve.

This work will focus only on the analysis of 932.05 hectares, as indicated in Figure 2. The reason for this delimitation is the availability of two previous studies from 1978 and 2010, from which five profiles located in the study area have been used, the data of which serve as representative of before the fire, and the information obtained can be extrapolated to the rest of the area due to the similarities with the existing soils [42,43]. Subsequently, in January 2023, five more substrate samples were collected from the same locations. In 2024, five additional samples were collected from the same sites in order to carry out this work. The coordinates of these five soil sampling points are listed in

Table 1. Coordinates corresponding to the sampling points.

Sample	Coordinates X	Coordinates Y
S.1	225083,754338	4492011,62344
S.2	224601,153372	4491167,07176
S.3	223413,700998	4492094,17361
S.4	223718,501607	4493357,82614
S.5	222734,249639	4492976,82538

.

Focusing on this study area, the analysis of the geology and pedology determines that the predominant substratum is slate, although Armorican quartzite is also observed, and above them we find mainly Cambrian Umbrisols. Umbrisols are soils with only one diagnostic horizon, called A, which can be ochric or umbric, located above the cambic B horizon with a degree of saturation (by NH₄Ac) of less than 50%. A cámbic horizon is an in-situ alteration horizon, one which can be determined by different physical and chemical properties. These properties, influenced by alteration processes and lithology, manifest themselves in changes of texture, structure and colour in relation to the underlying horizon [43].

In general terms, umbrisols are characterised by a dark surface horizon (A), resulting from a high accumulation of organic matter in the first metre of depth, generally presenting very acidic pH levels (<5.5). These are distributed in temperate or cold areas. Although their natural fertility is usually low, the presence of organic matter can improve their physical and chemical properties [44].

Specifically, Cambic Umbrisols are soils that are distinguished by a 50% saturated Cambic B horizon, located below an A horizon. The latter horizon is ulmic, i.e., with a colour intensity of less than 3.5 when wet and a colour purity of more than 3.5 when wet and 5.5 when dry. In addition, the organic carbon content must be at least 0.6% throughout the soil thickness, with a saturation degree of less than 50% [45].

Finally, from a vegetation point of view, in the sampling area and according to field analysis, it has been observed that there are two dominant species that occupy different altitudes. The highest parts are dominated by a scrubland made up of broom and heather. Below this, there are thickets of Quercus pyrenaica Willd. displaced, in the lower parts, by pine groves of Pinus pinaster Aiton and Pinus sylvestris L. Interspersed with these formations are other herbaceous and shrub species in varying densities (Agrostis truncatula Parl. subsp. truncatula, Verbascum thapsus L., Genista tridentata L., etc.). On the other hand, there is an area of scree where vegetation is non-existent.

2.2. Methodology

The methodology used in this study combines field activities with laboratory analyses (Figure 3). In the field work, representative samples of the different types of soils present in the study area were collected. The laboratory work consisted of analysing these samples, defining the parameters necessary to calculate the various factors (such as granulometric analysis, organic matter content, soil structure, among others) that allow RUSLE to be applied to the analysis of water erosion risk. Subsequently, the data obtained both in the field and in the laboratory were processed using graphical methods (such as the Wischmeier nomogram and DTM generation) and empirical methods (parameter calculation formulae, RUSLE calculations, etc.). All of this allowed the construction of a database that was integrated into a geographic information system (ArcGIS 10.8), from which various parametric cartographies and final maps reflecting erosion risk were generated. On the other hand, satellite images (Landsat) were used to carry out the analysis of vegetation loss in the study area during the period 2022–2024.

The data used to carry out this study are as follows:

-

Meteorological data: Average and annual rainfall from the 3 meteorological stations located in the study area. These data require a continuous record of rainfall intensity variations and has been obtained from the database of the Geographic Information System for Agricultural Data (SIGA).

-

Digital terrain models: These have an accuracy of 1 × 1 m per pixel and have been downloaded from the IGN.

-

Satellite images: Obtained from the Spanish Geographic Institute.

-

Soil data: These were obtained from the physico-chemical analyses carried out in the laboratory of the 5 soil profiles taken in the study area. For the pre-fire data, data from two studies [42,43] have been used and have been contrasted with others carried out later that, due to conflicts of interest, cannot be published, showing their accuracy.

The data set used was obtained from databases available on the different platforms or from laboratory analyses. It is a high-precision analysis, provided by DTMs. Furthermore, the chosen study area is a strategic location, as well as being easily accessible, and can be extrapolated to the rest of the affected area.

2.3. Determination of Physico-Chemical Properties

From these five soil profiles, collected at different times (before the fire, immediately after and two years later) three fundamental parameters were determined: the percentage of organic matter, clay, total sands, silt and pH, thus allowing a basic physico-chemical characterisation.

Samples collected in the field are taken to the laboratory, where they are subjected to standardised processes. First, they are air-dried and sieved with a 2 mm mesh, thus separating the fine fraction needed for analysis. This fraction smaller than 2 mm is stored in plastic bags, while the rest is discarded.

The analytical methods used for the determination of these parameters are as follows. For pH, 10 g of soil is added to 25 mL of water and the sample is shaken and left to stand for 30 min to facilitate contact between the water and the soil. After this time, a pH meter is used to measure the pH.

For the analysis of organic matter, a small amount of the samples must be ground to a size of less than 2 mm and a homogeneous consistency, a process that is carried out for one hour for each sample. This organic matter is determined by the standard method described in [35], which consists of wet oxidation with potassium dichromate (K₂Cr₂O₇) in an acid medium and titration of the excess with ammonium ferrous sulphate (Fe (NH₄)₂SO₄)₂ 6H₂O, using diphenylamine as a redox indicator.

Finally, to determine the textural composition of the soil (% sand, % silt and % clay), a particle size analysis is carried out using the Robinson pipette method [46]. After removal of organic matter with hydrogen peroxide (H₂O₂) and the use of sodium hexamethophosphate as a dispersant, the coarser fraction (sand) is separated by sieving, while the finer fractions (silt and clay) are separated by differential sedimentation.

2.4. Determination of Soil Erosion Rate

Apart from these three physico-chemical analyses, the percentage of soil erosion is also calculated through the most common method, the Universal Soil Loss Equation (USLE), due to its simplicity and wider acceptance. This was created by [47,48], who also created its revised form, the Revised Universal Soil Loss Equation (RUSLE) [49].

RUSLE is considered an empirical model by which annual soil losses are estimated and averaged over a representative period of years, considering specific conditions of climate, soil, relief, vegetation and land use. This method comprises five factors: rainfall erosivity index (R), soil erodibility index (K), topographic factor, which comprises slope length (L) and slope steepness (S), vegetation cover factor (C) and soil conservation or management practices factor (P). All of these are represented in the following equation [49]:

A = R × K × LS × C × P

In order to carry out this analysis, in which the current soil loss of the study area is determined, a cartography of actual erosion is established through these five parameters, which have been classified into different grades (

Table 2. Interpretation of soil loss values [50].

Class	t/ha/year	mm/year
Very low erosion and tolerable soil loss	<5	<0.50
Low erosion and tolerable soil loss	5–10	0.50–1.00
Mild erosion level	10–25	1.00–2.50
Moderate erosion level	25–50	2.50–5.00
Severe erosion level	50–100	5.00–10.00
Very severe erosion level	100–200	10.00–20.00
Extreme erosion level	>200	>20.00

).

2.4.1. Rain Erosivity Factor R

The R factor, i.e., rain erosivity, is defined as the aggressiveness of rain on the soil. It represents the energy with which raindrops, falling with a certain intensity, can break the surface soil aggregates into particles that can be transported [51]. The formula proposed by the USLE to calculate this erosive effect of rainfall was too complicated, because it required a very detailed amount of information on rainfall in the study area. Consequently, multiple authors and institutions have attempted to make the calculations easier by relating the R factor to parameters that are easier to obtain and calculate [52].

Firstly, the former Institute for the Conservation of Nature (I.C.O.N.A.) was able to reach similar conclusions by means of simpler calculations, through the different equations that were used depending on the geographical area in which the study was being carried out. Years later, in 1960, Fournier established the Fournier Index (FI), using only the month with the highest rainfall. Finally, in 1978, Arnoldus proposed a correction to the Fournier Index, renaming it the Modified Fournier Index (MFI), the latter being the one used for this study [53].

To calculate the R factor, both annual and monthly rainfall data from 25 years ago, obtained from three meteorological stations near the study area (La Alberca, Puebla de Yeltes and Serradilla del Arroyo) and extracted from the Geographic Information System of Agricultural Data [54], were used. With all of these data and the coordinates of the meteorological stations, an Excel file was prepared, one which is necessary to generate twelve raster layers containing the average monthly rainfall from January to December, in the ArcGIS^® [ArcMap^®] v. 10.8 [55] programme using the weighted distance method tool (IDW).

Once the monthly average rainfall data have been obtained, it is necessary to calculate the average annual rainfall value (Pt) using the raster calculator tool. This ranges from a minimum of 643.1 mm in Puebla de Yeltes to 1369.7 mm in La Alberca. With all of these data, the Modified Fournier Index (IMF) is calculated, using the following formula:

$$\mathrm{IMF}=\sum _{\mathrm{i}=1}^{12}{\displaystyle \frac{{\mathrm{p}}_{\mathrm{i}}^2}{{\mathrm{P}}_{\mathrm{t}}}}$$

where p_i is the precipitation of each month (mm) and P_t is the mean annual precipitation (mm). For this study, the IMF value ranges from 90.15 to 78.58 mm.

Finally, the R factor has been calculated using the following equation for our study area:

$$\mathrm{R}=2.56·{\mathrm{IMF}}^{1.065}$$

where the values 2.56 and 1.065 are correction values proposed by ICONA corresponding to the location of the study [56].

Finally, a raster was generated using ArcGIS^® [ArcMap^®] v. 10.8 [55], including the values of the rain erosivity index^®.

2.4.2. Soil Erodibility Factor (K)

The K factor is a measure of the vulnerability of the soil to erosion by both detachment and transport of particles, both in quantity and flow in a specific predicted rainfall event [39]. This factor is related to those local characteristics, soil structure and the degree of weathering of the soils that characterise the study area. Thus, to obtain the values of this factor, it is necessary to obtain the values of the soil properties, such as the percentage of organic matter, sand, silt, soil structure and permeability [54]. To obtain this factor, the Wischmeier Nomogram [47] was used with the data obtained from the physico-chemical analyses of the five profiles, extracting the percentages of silt, very fine sand (0.002–0.1 mm), sand (0.1–0.2 mm), organic materials, soil structure and permeability [57].

2.4.3. Topographic Factor (LS)

The slope length (L) and its slope (S) form the factor known as LS in the RUSLE method. To calculate this factor, the formula of [58] has been used, which previously required the raster slope map that is prepared using a LIDAR map of the study area, itself provided by the National Geographic Institute (IGN), with each obtained using ArcGIS^® [ArcMap^®] v. 10.8 [44]. The equation is as follows:

$$\mathrm{LS}=({\mathrm{Flow} \mathrm{accumulation} · {\displaystyle \frac{\mathrm{cell} \mathrm{size}}{22.14}})}^{0.4}·{\left({\displaystyle \frac{\sin \mathrm{slope}}{0.0896}}\right)}^{1.3}$$

where flow accumulation is the number of cells contributing to the flow in a given cell, cell size is the length of the size of one side of the cells and sin slope is the sine of the slope in radians.

To calculate the S factor, a slope raster is first generated from the digital terrain model (DTM) with a resolution of 1 metre per pixel. The slope values are then converted to radians using specialised raster analysis tools. As for the L factor, an initial raster of flow accumulation is assumed, but because its values may be overestimated, an adjustment is applied to set an upper limit of 250 metres of slope length, equivalent to 250 cells at this scale.

The processing of the flow accumulation is undertaken through two auxiliary rasters, called A and B. To generate the A raster, the values of the flow accumulation raster are reclassified: those equal to or greater than 250 are converted to 0, while those below this threshold are assigned as 1. Subsequently, the A raster is multiplied by the original flow accumulation raster, which has the effect that all values greater than 250 are reduced to 0, while values within the allowed range remain unchanged.

The next step is to generate raster B, where again the initial flux accumulation is reclassified. In this case, values less than or equal to 250 are replaced by 250, while those greater than this threshold become 0. Finally, by combining raster A and B by summation, a corrected raster is obtained in which the original values are preserved for pixels within the 250 limit, and those that previously exceeded it are adjusted to precisely this value to avoid distortions in the flux accumulation calculations.

2.4.4. Plant Cover Factor (C)

The C factor indicates how effective the plants are as a protective layer of the soil against the impact of raindrops and the force of surface flow. This factor is estimated by assigning uniform empirical values already available; however, this method fails to capture real spatio-temporal variations in vegetation cover, especially in the case of changes occurring in a short time, as is the case in this study. The use of satellite images to generate maps of the C factor based on the NDVI improves the accuracy of the estimation of C values [59,60]. This method allows one to capture the state of vegetation cover and the spatio-temporal variation in the estimation of values [59]. Different studies have calculated the relationship between the C factor and the NDVI, which will be used in this study [61,62].

Factor C indicates the effectiveness of the plants as a protective layer of the soil against the impact of raindrops and the force of surface flow. The calculation of this factor was made from a vector layer obtained in ArcGIS^® [ArcMap^®] v. 10.8 [55] from the Spanish forest map of the study area, the cartography of the NDVI and the vector layer of all regional citations of plant species collected, which was trimmed to include only the plant species present in our study area and an additional area of 500 metres. This is due to the fact that forest fires not only affect the exact area of the fire, but also nearby areas, such as via smoke, so this distance was taken into account.

It is worth mentioning that the data of the last vector layer of regional citations are relatively old, with the most recent information from the 2002 Detailed Habitat Mapping Project.

With this information, a table of abundances (Appendix A) has been drawn up that includes the species of flora both before the fire and those that have been identified afterwards, given that the habitat is still in the process of recovery and that plant growth is still very limited. From this map, the plant species of the area are obtained, giving them a value of the C factor (

Table 3. Land cover values [49].

Vegetation Cover Type	C Value
Mixed hardwood forests	0.003
Pyrenean oak	0.04
Riparian forest	0.09
Poplar and banana plantation in production	0.09
Ash groves	0.09
Wild olive groves	0.18
Juniper groves	0.18
Cork oak forests	0.19
Evergreen oak groves	0.19
Portuguese oak groves	0.19
Chestnut groves	0.22
Non-wooded	0.24
Mix conifer forest	0.42

) according to the tables of [47].

2.4.5. Soil Conservation Practices Factor (P)

The P factor is interpreted as soil loss under specific management practices [47,49]. Typical values of this factor vary between 0 and 1. A value of 1 is associated with land with no supporting practices (especially grasslands and bare land). On the other hand, values close to zero represent lands subjected to specialized support practices [53]. According to the available information, no management practices have been employed in the region, so it is assumed that the P factor is 1 for the entire study area. Therefore, it was not necessary to create a raster layer for this factor.

2.5. Normalized Difference Vegetation Index

The Normalized Difference Vegetation Index (NDVI) is of great value in the ecological field, because it evaluates the fraction of photosynthetically active radiation intercepted by vegetation in an adequate way [63]. The scientific basis of this index lies in the presence of chlorophyll in healthy vegetation. Chlorophyll in a healthy plant absorbs solar energy in the red band (RED), being transformed into sugars necessary for its growth. This same chlorophyll reflects most of the sun’s energy in the near infrared spectrum (NIR). In short, vegetation that is in good health has a high ratio between NIR and RED and this can be detected by satellites [64].

The formula is as follows:

$$\mathrm{N}\mathrm{DVI}={\displaystyle \frac{\mathrm{Near} \mathrm{Infrared} \mathrm{band} \left(\mathrm{NIR}\right)-\mathrm{Red} \mathrm{Band} \left(\mathrm{RED}\right)}{\mathrm{Near} \mathrm{Infrared} \mathrm{band} \left(\mathrm{NIR}\right)+\mathrm{Red} \mathrm{Band} \left(\mathrm{RED}\right)}}$$

where NIR is the reflectance of light in the near infrared band and RED is the reflectance in the red visible light band.

The resulting value of this index is between −1 and 1. Negative figures coincide with natural water bodies, i.e., waterlogged areas, ponds, lakes and rivers. Positive figures closer to the value 0 describe those areas in the absence of vegetation. When the resulting values are positive and closest to 1, this means that the density of vegetation in that area is higher [65].

In order to analyse vegetation loss in the study area between 2022 and 2024, that is, before and after the forest fire, images from the Lansat 9 satellite, included in the US Geological Survey (USGS) program and jointly managed by NASA and USGS, were used. The images analysed were downloaded from the US Geological Survey (USGS) using the online Global Visualization Viewer [66].

The selection has been limited to those images with less than 15% cloud cover and in spring period, before and after the fire, choosing those corresponding to 29 May 2022 and 27 May 2024. These images were processed with ArcGIS^® [Arc-Map^®] v. 10.8 [55] using the raster calculator tool, calculating the NDVI through the aforementioned formula.

The resulting cartography has been classified according to the values in

Table 4. NDVI interpretation values [67].

NDVI Value	Interpretation
0 a 0.20	Dead vegetation
0.21 a 0.33	Sick vegetation
0.34 a 0.66	Healthy vegetation
0.67 a 1	Very healthy vegetation

. This is an adaptable classification as, depending on the type of vegetation, biome or phenological state, it can be easily modified to afford the NDVI’s sensitivity to various spatial and climatic parameters.

3. Results

Different geoprocessing processes have been carried out using the tools available in the ArcGis software [55].

3.1. Physical–Chemical Properties

The results of the physical–chemical analyses (

Table 5. Results of physical–chemical analyses in different years in the study area.

	Year 1978 y 2010						Year 2023		Year 2024
Sample	pH	% Organic Matter	Soil Structure				pH	% Organic Matter	pH	% Organic Matter	Soil Structure
			% Coarse Sands	% Fine Sands	% Silt	% Clay					% Coarse Sands	% Fine Sands	% Silt	% Clay
S.1	4.4	17.8	10.4	48.0	11.5	8.6	4.22	8.55	4.56	6.18	13.29	46.53	35.04	5.13
S.2	4.6	18.60	11.5	44.0	15.0	8.0	4.04	25.26	4.60	18.25	17.81	43.70	27.90	10.59
S.3	4.9	13.00	9.0	43.0	23.0	10.5	4.18	17.18	4.63	21.88	12.78	50.62	25.01	11.60
S.4	5.3	8.20	6.0	36.5	35.4	13.9	4.27	5.91	5.09	7.34	5.64	56.24	30.59	7.53
S.5	5.0	4.27	5.5	44.0	31.4	15.8	4.67	9.01	4.67	10.64	12.94	44.94	30.50	11.63

) carried out before and after the fire allow us to determine the types of soils present in the study area. In the case of 2023, only analyses of organic matter and pH were analysed. Although the analyses were carried out in different years, with a considerable time difference, when comparing the granulometric analyses, it is observed that the percentage of sand, total sum of coarse sands (AG) and fine sands (AF), is notably higher than that of silt and clay in both years, though the percentages vary slightly. The presence of sand is essential for drainage and provision of air space in a soil, which is indispensable for root development. This trend, if analysed through the textural triangle, shows that the different samples in both years have a sandy loam texture. However, it is not only sand that is important, silt contributes to nutrient retention and improves soil texture, while clay is key for both nutrient and water retention. However, an excess of clay can make the soil too compact and retain too much water, thus hindering good root growth [68].

With regard to pH, it can be seen that all of the samples show very low values, both before and after the fire, which is consistent with the typical acidity of the Cambrian Umbrisol soils present in this area, probably influenced by the presence of pine trees that tend to acidify the soil. Although it is true that the pH has decreased slightly after the fire. The pH is crucial for soil fertility, as it can cause certain essential nutrients, necessary for proper plant development, to remain precipitated and insoluble, i.e., these nutrients are immobilised. For nutrient assimilation to occur, a pH between 5.5 and 7 is necessary, although there are some plant species that prefer acidic or alkaline soils [68].

Finally, it can be observed how organic matter varies, rising or falling according to the sample, although without notable differences. This soil component improves the structure, water retention capacity and fertility of a soil [68].

3.2. Soil Erosion

As mentioned above in the methodology section, the RUSLE formula was used to determine soil erosion in the area affected by the 2022 forest fire in the province of Salamanca, calculating each factor independently.

Figure 4A and Figure 5A, show how the value of the R factor is higher in the regions located to the southeast of the area studied and decreases as one ascends towards the north, suggesting greater soil erosivity as altitude increases. This could be due to the lower protection offered by vegetation at higher altitudes, where vegetation tends to be sparser and less dense, thus providing less protection to the soil. The average value of this factor ranges between 309.24 and 267.14 MJ mm ha⁻¹ h⁻¹ yr⁻¹ before and after the fire and a standard deviation of 9.54. These results indicate that there is no significant variation in erosive potential across the territory analysed, suggesting that the impact of the fire was uniformly aggressive throughout the area. As this factor is based on the precipitation recorded at the meteorological stations in the area, only one map has been produced, the same before and after the fire, as the variations are small.

The interaction between forest fires, precipitation and the R factor is very important in managing and understanding erosion dynamics. During rainy periods, the soil is saturated with water, making erosion more likely, especially in areas with steep slopes. If a forest fire has previously occurred in the region, rainy periods are more likely to be more aggressive in terms of erosion.

The K factor is essential in the RUSLE model, as it plays a key role in the assessment of erosion susceptibility. The values of this factor range from 0 to 1, indicating a lower sensitivity to erosion and, therefore, a higher soil resistance to erosion when those values are closer to 0, while a value close to 1 reflects a higher susceptibility to erosion [69]. In this study, the values of erodibility before the fire ranged from 0.27 to 0.45 T ha⁻¹ MJ⁻¹ mm⁻¹, with a mean value of 0.15 and a standard deviation of 0.03 and a coefficient of variation of 0.22. This is demonstrated in Figure 4B, which shows how the lower altitude areas are the most vulnerable, with values closer to 1. Two years later, after the fire, the values of this factor have changed, ranging from 0.46 to 0.61 T ha⁻¹ MJ⁻¹ mm, with a mean value of 0.52, a standard deviation of 0.06 and a coefficient of variation of 0.11, suggesting an increase in susceptibility to erosion (Figure 5B). This increase is probably attributed to the loss of organic matter, clay and, especially, to the percentage of silt in the surface layer of the soil as a consequence of the impact of the fire. The silt detaches, forming crusts that seal the soil pores, preventing infiltration and increasing runoff, which has a greater erosive power than the direct impact of raindrops [70,71].

Consequently, soils with a high silt content are more susceptible to erosion than other types of soils, and in this study, it is observed that the soils present a more sandy texture. In addition, when comparing the distribution between the two years analysed, a difference in the distribution is observed, although the highest values are still concentrated in the lower altitude areas.

In the study area, the dimensionless values of the LS factor (for before and after the fire) range from 0 to 59.62, with a mean value of 2.28 and a standard deviation of 1.44. As seen in Figure 4C and Figure 5C, most of the area has values very close to zero, except for a small area in the northwest, with the highest values. This is because the topography of the area is not uniform, thus capturing the heterogeneity of the landscape. Those areas with steeper slopes and longer slope lengths are more prone to erosion, especially during periods of heavy rainfall.

Conversely, flatter areas or shorter slope lengths have values close to zero. It is very important to understand how the LS factor values are spatially distributed for erosion management and land use planning.

This understanding helps one to recognize high-risk areas that would benefit from specific erosion control interventions, such as reforestation or terracing, in order to mitigate soil loss and its environmental consequences. On the other hand, those areas with lower values could be suitable for agriculture or other forms of land use.

Finally, vegetation cover, represented in the RUSLE equation as factor C and serving as an indicator of the management of vegetation cover and its influence on soil erosion, shows values ranging from 0.11 to 0.24, before the fire (Figure 4D). After the forest fire, values vary between 0.45 and 0.55 (Figure 5D).

Factor C is especially relevant after the fire in 2022, as both field and computer work have shown how the reduction in vegetation cover has decreased in thickness and type of vegetation, going from having a forest close to its maximum, in terms of its tree, shrub and herbaceous strata, to having a mostly herbaceous stratum. In other words, before the fire there was vegetation with greater protective power against the kinetic energy of raindrops.

During a forest fire, vegetation is consumed and the landscape is altered, resulting in the destruction of the natural barriers that protect the soil from erosion. This leads to an increased risk of landslides and increased soil transport. This increased vulnerability not only affects the soil directly, but also has wide-ranging environmental implications: degradation of water quality, increased risks of landslides and flash floods, and adverse effects on ecosystems and local communities. It is therefore essential to implement soil management strategies to mitigate the increased risk of erosion and its environmental consequences after a fire. This has also been verified through the NDVI, which will be discussed in the next section.

If we evaluate soil loss in the study area through the different components of the RUSLE model, we find that, in 2022, before the forest fire, the maximum soil loss rate calculated was 1551.85 t/ha/year (Figure 6A), which indicates the rate of soil loss under the natural environmental conditions of the region. Meanwhile, in 2024, after the forest fires, this value increased, amounting to 4899.42 t/ha/year (Figure 6B).

These results, interpreted according to the values in Table 4, reveal a large difference between the two years. As can be seen in Figure 5, both before and after the fire, all degrees of erosion are spread over the whole sampled area. However, after the fire, these values increase a lot. It should be noted that there are localised and infrequent points where the erosion rate is very high, exceeding 200 t/ha per year.

Figure 6 shows how the estimated rate of soil loss is classified into five categories according to the severity of erosion, ranging from very slight, in green, to extremely severe, in red. Prior to the fire, the areas most susceptible to erosion were located mainly in the southern region and a small area northwest of the study area. Post-fire analysis indicates that the areas most vulnerable to erosion remain the same, but with a generalized intensification throughout the area. That is, the comparison between the two years explains a general increase in soil erosion, which could be related to the alteration of soil properties after the fire, particularly with respect to water holding capacity, soil porosity and hydraulic conductivity in the burned areas. All of this results in increased runoff and surface sediment transport.

3.3. Normalised Difference Vegetation Index

Satellite sensors capture the radiation emitted by the earth’s surface, which makes it possible to generate various vegetation indices, including the Normalised Difference Vegetation Index (NDVI). This index is calculated using wavelengths in the red band (0.6–0.7 μm) and in the near infrared (0.75–1.75 μm), which are widely known in plant remote sensing studies. Its implementation with existing vegetation data provides a fairly reliable measure of the correlation between NDVI values and vegetation types that are present prior to the fire. In the post-fire situation, the other indices calculated in this study are used to assess the opposite correlation between areas of maximum erosion and those with low NDVI.

The NDVI values show a significant change before and after the fire. As shown in Figure 7A, before the event, the sampled area presented a range of values, where areas with indices between 0 and 1 indicated the presence of vegetation cover, while negative values up to −1 indicate the presence of water. The closeness of these values to 0 or 1 provides information on the health status of the plants. It should be noted that these values also vary according to the percentage of woodiness they contain. In the case of grasses, they have a greater variability according to water stress, while the shrub and tree stratum show more constant values. In short, the values of this index change according to the type of vegetation, its woodiness, species and season of the year.

However, after the fire and after approximately two years (Figure 7B), the analyses reveal a notable decrease in plant activity in line with expectations.

In 2022, the maximum NDVI value is 0.4677, while in 2024 only a value of 0.3085 is reached, indicating a decrease of 34%. If we look at the 2024 map, it can be seen through the colouring that there is a large percentage of values between 0 and 0.20, which indicate the mortality of vegetation, a normal situation after a fire episode. Although it is true that a detailed spatial analysis was not carried out, it was observed that the southern area was the most affected by the loss of activity, coinciding with the fact that it is at the highest altitude and therefore has more eroded and rocky soils and is less prone to intense vegetation development, while the valley areas, where there has been greater soil deposition, have the opposite situation, although in this case this is due to the development of fast-growing herbaceous and shrub formations.

Comparing both maps, it is clear how the decrease in vegetation is especially important in the southernmost area of the study area, which has a higher altitude and, therefore, has a rockier soil that hinders the growth of vegetation. In addition, through the growth values obtained from the NDVI values in 2024, one can see how the terrain is trying to recover.

The organ specializing in capturing light for photosynthesis in vegetation is the leaf. Chlorophyll, the chemical substance that gives plants their green colour, is fundamental to this process, as it absorbs red and blue wavelengths, while reflecting green ones. This percentage of absorption varies throughout the seasons of the year [18]. However, the images used to calculate this index in this work were taken in the same season to minimize this variability and ensure comparability of the data.

3.4. Recovery Measures

Once the 2022 forest fire was under control and extinguished, the succession process began, influenced by the characteristics of the terrain and the recovery measures adopted. Human action through proper management of the burned wood can accelerate this process and the regeneration of the habitat, reducing the impact of the fire.

These actions have the following objectives: the supply and recycling of nutrients, reduction of vegetation mortality during rainy periods, greater protection against herbivory, the prevention of soil erosion, the increase of biodiversity and the facilitation of mutualistic interactions, such as the dispersion of seeds.

The most common procedure in Spain is the removal and extraction of commercially valuable vegetation remains and the shredding of those that cannot be sold. This modifies the landscape, which is stripped of its vegetation cover [72].

In the Monsagro fire, different restoration actions were implemented according to the plan established by the Junta de Castilla y León. These interventions covered the entire area affected by the fire, without focusing solely on our study area. Among the main measures adopted were the restoration of water supply, feeding of affected livestock and hydrological–forestry actions, all with the aim of mitigating the damage caused by the fire and preventing or reducing future damage [73].

To prevent water erosion in flat terrain, activities such as shredding of trees and logging debris were carried out in areas where, due to the severity of the fire, the timber had no commercial value. This shredding helped eliminate potential sources of pests and create a mulch on the ground to prevent runoff and splash erosion during rain events. However, on sloping ground, the work was undertaken manually, leaving the shredded material aligned with contour lines, thereby reducing runoff. In areas of difficult access, helicopters were used, placing cereal straw and creating a mulch. The aim is to achieve a high level of soil protection against erosion [73].

In addition, the reforestation of irrigation channels was carried out and barricades and fajinas, structures perpendicular to the flow of water, were built to reduce the speed of runoff water and act as a barrier for the material carried away. Silvicultural treatments have also been carried out to promote the regeneration and regrowth of some species, such as strawberry trees (Arbutus unedo L.), holm oaks (Quercus ilex L.) and cork oaks (Quercus suber L.).

A monitoring measure that has been applied against the possible appearance of pests is the installation of pheromone traps at strategic points for possible pest dispersal. In addition, as protection and prevention measures for possible events such as this one, firebreaks have been created and water tanks and ponds have been built to provide water for the fauna and as a loading point for fire extinction.

4. Discussion

4.1. Physical and Chemical Properties

The physical–chemical analysis of the soils helps to determine which properties have been affected, such as the alteration of the structure or a decrease in the organic matter content, which cause a lower retention of essential nutrients for the development of vegetation. The granulometry of these soils has also been affected in cases where the percentage of sand has increased, while the percentage of silt and clay has decreased. Similar changes in these percentages have been found in other studies of fire-affected soils [73,74,75]. These variations are mainly attributed to the following mechanisms:

-

Heat-induced dehydration of clay minerals in the soil samples, causing strong interactions between clay particles, leading to a decrease [73].

-

Thermal transformation of iron and aluminium oxides, causing them to act as cementing agents in the clay, forming silt-sized particles in the soil samples [74,76].

-

Aluminium oxides and hydroxides released during clay decomposition act as cementing agents in the formation of area-sized particles, therefore, an increase in sands is strongly associated with a decrease in clay content [74,77,78]. In addition, clay decomposition results from the removal of structural hydroxyl ions, leading to the disintegration of the crystalline structure of clay minerals [77].

4.2. Soil Erosion Discussions

Soil erosion and land degradation processes affect various types and conditions of land use [79]. For this reason, an increasing number of scientists are trying to effectively address soil erosion problems and identify how to mitigate their effect [80]. One effective way of assessing soil erosion is through GIS techniques and satellite imagery, which allow for spatio-temporal adjustments [81]. In this study, the analysis of soil loss was carried out using the RUSLE equation implemented in ArcGIS 10.8, which was very useful for determining erosive losses and which factors were affected, the effectiveness of this method having been demonstrated in different studies [82].

The results referring to the period before the fire show rates of between 10.1 and 100, with the greatest losses located to the south of the study area, coinciding with the steepest slopes. These results are in agreement with those estimated by other studies carried out in the area [21,24].

In the results obtained after the fire, a clear increase in erosion rates was observed, with the greatest losses concentrated in the southern zone, with rates of between 50.1 and 200 Tm/Ha/year. These results are similar to those of other studies [82]. Within the RUSLE factors, the most affected were factor K and factor C, which increased significantly. This reflects a greater exposure of the soil to erosive processes, which favours sediment transport and the loss of essential nutrients for vegetation regeneration. As for the latter factor, this model shows a high sensitivity to alterations in factor C, particularly when modelling post-fire erosion [83,84]. To avoid this, the analysis of the NDVI factor, which indicates whether there are substantial changes in vegetation cover, has been used [10,85,86].

4.3. Normalised Difference Vegetation Index (NDVI)

The NDVI is a key indicator in vegetation recovery, allowing the assessment of photosynthetic activity through satellite imagery. Satellite sensors capture the radiation emitted by the earth’s surface, which allows the generation of various vegetation indices, including the Normalised Difference Vegetation Index (NDVI). This index is calculated using wavelengths in the red (0.6–0.7 μm) and near-infrared (0.75–1.75 μm) bands, which are widely known in plant remote sensing studies [57,60,87].

The results obtained show a clear decrease in NDVI values, which means that plant recovery is slow, especially in the most affected areas. Moreover, this indicator is very useful, as it allows monitoring the progress of restoration in the long term and adjusting soil management strategies as necessary.

The analysis of the evolution of NDVI is descriptive in general terms and no detailed correlation with other studies has been considered. It remains simply as a clear statement of an observable change in the satellite images between periods before and after the fire. However, there is presumably a clear difference between herbaceous, shrub and tree formations, especially because the forest cover practically disappears, giving way to an active regeneration of scrub and herbaceous communities.

Furthermore, the decrease in NDVI values is due, on the one hand, to the fire, but also to the effects of soil compaction and soil movement, as a consequence of the removal of burnt trees and subsequent reforestation. In any case, these would always be very localised activities and much less significant at the scale of the work proposed than the alterations derived from the fire itself, and can therefore be ruled out.

On the other hand, previous studies have made extensive use of the relationship between NDVI and RUSLE to analyse how soil erosion evolves as a function of vegetation cover. Several studies have indicated that the combination of both indices not only allows the immediate impact of a fire on the ecosystem to be assessed, but also facilitates the monitoring of recovery over time. This information is key to identify critical areas where restoration strategies should be implemented [86,87].

4.4. Implications

The combined use of satellite imagery and the RUSLE model provides a comprehensive and accurate approach to post-fire impact assessment, allowing detailed estimation of soil erosion and vegetation cover changes. Satellite imagery provides spatial and temporal monitoring to analyse the severity of the fire and its effects on the landscape before and after. On the other hand, RUSLE allows the quantification of soil loss by considering factors such as rainfall erosivity, soil erodibility, slope, vegetation cover and conservation practices. This methodological combination allows large areas to be analysed efficiently, reducing the need for extensive field measurements and providing high resolution data. In addition, it facilitates the identification of critical areas with increased vulnerability to erosion, which contributes to the planning and implementation of soil restoration and conservation strategies. Overall, this approach improves land management and decision-making in the recovery of ecosystems affected by forest fires, providing key information to mitigate their effects and promote environmental sustainability.

The cartographies obtained can be useful to locate the affected areas and to implement restoration measures, according to their degree of affection, to mitigate the damage caused and prevent future catastrophes. The objectives of these measures are the supply and recycling of nutrients, reduction of vegetation mortality during rainy periods, increased protection against herbivory, prevention of soil erosion, increase of biodiversity and facilitation of mutualistic interactions, such as seed dispersal [87].

It can also be useful for soil planning and conservation, facilitating decision-making on conservation measures. In addition, it allows the effectiveness of sustainable agricultural practices aimed at reducing erosion to be assessed.

In terms of the implications for agricultural productivity, the RUSLE model helps to design strategies to minimise soil degradation and maintain soil productivity in the long term.

Finally, it can support the development of environmental policies, providing scientific data for the formulation of conservation and sustainable land use policies. It can also be integrated with other models to forecast future land degradation scenarios.

5. Conclusions

The Monsagro forest fire left a significant impact on the environment, being considered an ecological disaster that altered the local vegetation and fauna, as well as degrading the soils, affecting their fertility and structure, increasing erosion.

This study has made it possible to analyse the effects of the fires on the physico-chemical properties of the soil, erosion, vegetation cover and the implications for soil management and conservation. The results obtained show the magnitude of the changes induced by fire and the importance of implementing adequate strategies for the restoration and mitigation of its negative effects.

Regarding the physical–chemical properties of the soil, a significant alteration in the granulometry was observed, characterised by an increase in the sand content and a reduction in the percentages of silt and clay. This phenomenon is attributed to heat-induced dehydration of clay minerals, as well as thermal transformation of iron and aluminium oxides, which act as cementing agents in the formation of larger particles. These alterations can affect the soil’s ability to retain nutrients and water, compromising its fertility and the regeneration of vegetation.

In relation to soil erosion, analyses based on the RUSLE equation revealed a significant increase in post-fire soil loss, especially in areas with steeper slopes. Factors K and C of the RUSLE equation were the most affected, reflecting a greater vulnerability of the soil to erosion processes. Erosion modelling shows that the loss of essential nutrients can compromise vegetation regeneration, so early intervention is required to stabilise affected areas and reduce erosion. The maximum rate of soil loss before is 1551.85 t/ha/year, while after this value is 4899.42 t/ha/year.

The Normalised Difference Vegetation Index (NDVI) showed a marked decrease in the most affected areas, indicating slow vegetation recovery, with a maximum value of 0.3085 after the event and 0.4677 before the event. This index has been used as a key tool with which to monitor the long-term regeneration process, allowing soil management strategies to be adjusted. Furthermore, it has been shown that the relationship between NDVI and RUSLE can provide valuable information to assess the impact of fires on the ecosystem and to plan the restoration of degraded areas.

The implications of the study highlight the importance of generating detailed cartographies to identify the most affected areas and prioritise restoration actions. The implementation of measures that favour the recycling of nutrients, the protection of vegetation in early stages of regeneration and the prevention of soil erosion are recommended. Furthermore, the results obtained may be of great use for soil planning and conservation, helping in the design design sustainable agricultural strategies and environmental policies focused on mitigating soil degradation. Furthermore, the integration of these analyses with other predictive models can improve the capacity to respond to future land degradation events, ensuring a more effective and sustainable management of fire-affected ecosystems.