Modeling the Effect of Nature-Based Solutions in Reducing Soil Erosion with InVEST ^® SDR: The Carapelle Case Study

Abstract

Soil erosion threatens agricultural sustainability and water quality in Mediterranean watersheds, necessitating effective Nature-Based Solutions (NBSs) for mitigation. This study applied the InVEST Sediment Delivery Ratio (SDR) model to assess erosion patterns and evaluate NBS effectiveness in the Carapelle watershed (506 km²). The SDR model was calibrated and validated using measured sediment yield data from 2007 and 2008. Model validation achieved a 4.3% deviation from observed data after parameter optimization. Four NBS scenarios were evaluated: contour farming (CF), no-tillage (NT), cover crops (CCs), and combined practices (Comb). Baseline soil loss varied from 2.43 t ha⁻¹ yr⁻¹ (2007) to 3.88 t ha⁻¹ yr⁻¹ (2008), with sediment export ranging from 0.86 to 1.30 t ha⁻¹ yr⁻¹. NT demonstrated the highest individual effectiveness, reducing sediment export by 72.2% on average. The Comb approach (NT + CCs) achieved a superior performance with a 75.9% sediment export reduction and a 70.5% soil loss reduction. Spatial analysis revealed that high-retention zones were concentrated in forest and shrubland, while agricultural zones showed the greatest potential for NBS implementation. NBSs significantly enhance sediment retention services in Mediterranean agricultural watersheds. The InVEST SDR model proves to be effective for watershed-scale assessment. The results provide actionable guidance for sustainable land management and soil conservation policy in erosion-prone Mediterranean environments.

1. Introduction

Soils play a critical role in supporting human well-being and protecting the environment, delivering a wide range of essential ecosystem functions [1,2]. Soil also serves as one of the largest carbon reservoirs on Earth, storing twice as much carbon as the biosphere and atmosphere combined [3]. Additionally, they provide habitats for millions of species, supplying the nutrients needed for life to thrive. Healthy soil forms the backbone of agriculture and is an essential resource for meeting human needs across history [4]—a necessity that has only grown more urgent in our increasingly demanding 21st century [5].

Soil erosion has been a pressing environmental issue for thousands of years [6]. Although it is a natural process, soil erosion is aggravated by anthropogenic activities [7,8,9,10]. Consequently, soil erosion by water has become one of the major threats to soils worldwide [11] and in Europe [12,13]. Impacts of soil erosion include negative effects on the delivery of many natural processes such as crop production, water storage and filtration, nutrient cycling, biodiversity, and carbon storage [14,15].

Sediment yield has been increasing in many parts of the world, posing significant challenges to water quality and reservoir management [16]. These changes stem from the interplay between topography, climate, and land use. Because this relationship is highly complex and nonlinear, shifts in land use or management practices can have unpredictable effects [17,18]. As a result, understanding how natural landscapes retain sediment has become a crucial priority for landscape managers. For example, programs that invest in watershed services need to know where sediment originates and how it moves through the landscape. This knowledge helps stakeholders devise strategies to reduce sediment loads, whether by preserving areas that are good at retaining sediment or by directing agricultural activity to less impactful areas [19]. On a broader scale, understanding how different land management strategies can mitigate the impacts of agricultural development is key to making informed decisions [20].

To address these needs, various tools and models have been developed to improve our understanding of sediment dynamics [21]. These tools range in sophistication, from simple empirical models to more detailed physics-based approaches, and their use often depends on the expertise, resources, and data availability for a given project [22,23]. While physics-based models have advanced considerably in the past decade [24,25], they come with two significant challenges: (1) selecting parameters, calibrating models, and testing them requires a lot of data and expertise [26,27,28], making these models less practical for resource-limited projects; (2) many fundamental aspects of sediment transport remain poorly understood. For instance, predicting where sediment will be deposited—either overland or in streams—remains a significant challenge for model integration [25,29].

On the other hand, simpler tools designed for ecosystem service assessments are more accessible. These tools are often used in data-scarce contexts. They focus on understanding the trade-offs and synergies between multiple ecosystem services rather than producing highly accurate predictions. Predicting how land use or climate changes will affect these services is central to these models, which are often used to compare different scenarios [30]. This emphasis on relative comparisons rather than absolute accuracy presents an opportunity to develop models that focus on the dominant processes driving sediment dynamics [31].

One such tool is the InVEST (Integrated Valuation of Environmental Services and Tradeoffs) sediment retention model. InVEST is a suite of tools designed to quantify and map various ecosystem services, including water provision, pollination, and habitat quality [32]. The freshwater services models in InVEST use spatially explicit environmental data and physics-based methods to link the supply of ecosystem services to demand. For sediment retention, this involves estimating sediment delivery and retention across a landscape and then valuing the service in terms of avoided costs—such as reduced reservoir sedimentation or better stream health [33].

The Sediment Delivery Ratio (SDR) model within InVEST integrates the RUSLE alongside studies by [34,35]. It maps the spatial distribution of sediment production in a watershed by calculating the ratio between soil erosion and sediment transport [36,37,38,39]. This approach evaluates a land parcel’s capacity to retain sediment, factoring in geomorphology, climate, vegetation, and management practices [40,41]. Additionally, sediment retention can be assessed through avoided costs, such as sedimentation in reservoirs, stream degradation, and increased water treatment expenses [32].

To address and mitigate erosion, effective control strategies are needed. Traditionally, erosion control has been achieved through engineering interventions such as terraces, dams, and canals, but these solutions have often been expensive and not always sustainable in the long term. Alternatively, Nature-Based Solutions (NBSs) represent an innovative and sustainable approach to erosion control. NBSs are defined as actions that protect, sustainably manage, and restore natural or modified ecosystems while addressing societal challenges effectively and adaptively, simultaneously providing human well-being and biodiversity benefits [42]. In agricultural landscapes, NBSs specifically target soil and water conservation through practices that enhance natural protective mechanisms while maintaining productive capacity [43]. In Mediterranean agricultural systems, recent studies demonstrate that NBSs like no tillage, crop rotation, and contour farming improve ecosystem services, soil and water conservation, and agricultural productivity, often with economic benefits and reduced environmental impacts [44,45,46,47]. These practices address the region-specific challenges of seasonal rainfall intensity and intensive cereal production while maintaining agricultural productivity.

Quantifying soil erosion and sediment yield is vital for effective landscape management, particularly to mitigate sedimentation in downstream reservoirs. Despite Italy experiencing the EU’s highest erosion rates [12] and several areas of the Apulia Region being classified as at risk of erosion [46], the quantification of sediment retention services from specific NBSs (contour farming, conservation tillage, cover crops) is still limited. Although InVEST is one of the most used models for the assessment of the ecosystem services, no peer-reviewed studies have applied InVEST SDR to Italian watersheds [48]. This study aims to fills these gaps by applying the InVEST SDR model (version 3.14.3) to the Carapelle watershed, a Southern Italian agricultural watershed, to quantify spatial sediment yield and retention for multiple NBSs scenarios.

The Carapelle watershed was selected for its representativeness of Mediterranean agricultural erosion, characterized by seasonal climate variability, intensive wheat cultivation, and varied topography [9,10,49]. Its comprehensive monitoring and prior modeling with different hydrological models provide a validated basis for assessing NBS effectiveness. The main aim is to analyze and identify primary contributor areas to erosion, mapping the spatial distribution of sediment dynamics at the sub-watershed scale and evaluating the model’s performance in representing these processes. This study also aims to use the InVEST SDR model to quantify and map sediment retention levels in the watershed as one of the ecosystem services that promote sustainability. Another main objective is to assess the impact of adopting some NBSs to alleviate or mitigate the soil erosion and sedimentation issue in agricultural basins, which is vital for sustainability. To the best of our knowledge, in Southern Italy, the influence of NBSs on sediment retention services has not been previously analyzed using the InVEST SDR model. The findings of the current study aim to guide and prioritize land use decisions within the region.

2. Materials and Methods

2.1. Study Area

The Carapelle (Figure 1) is an important watercourse located in northern Apulia that originates in the Apennine Mountains, crosses the Tavoliere, and then flows into the Adriatic Sea. The 506 km² catchment area is situated at an altitude between 120 and 1075 m a.s.l. with a slope of 8.2% [50]. Precipitation ranges from 450 to 800 mm annually, with flash floods common from June to October. There are hilly and mountainous areas subject to considerable erosion and characterized by flysch formations, and the alluvial plain is composed of clay–sand sediments dating back to the Plio-Pleistocene. The predominant soils in the region are Entisols, characterized by a fine clay loam texture and a low organic matter content. The activity most responsible for the erosion process, as well as the primary economic activity, is agriculture, especially winter wheat cultivation on over 75% of the land, following a 4-year crop rotation with mineral fertilizers applied in December and February. Forests and pastures occupy mountainous areas, while urban lands are minimal. The main erosion processes include sheet and rill erosion, exacerbated by conventional tillage practices, involving plowing up and down slopes to depths of 25–40 cm. The area’s morphology significantly influences soil erosion, emphasizing the need for erosion control policies focusing on land management and topographic factors in source areas. The hilly areas are covered with deciduous oaks and hardwoods, along with coniferous meadows and pastures. The flat areas are mainly dedicated to the cultivation of wheat and to a lesser extent, olive trees and other agricultural crops. The Mediterranean climate has wet autumn and winter seasons and dry spring and summer periods with average annual rainfall ranging between 779 mm (Bisaccia, 1921–2012) and 531 mm (Castelluccio dei Sauri, 1921–2012). The driest month is August (6.4 mm), while March (94.9 mm) and November (81.4) are the rainiest.

The agricultural landscape of the Apulia Region in Southern Italy, where the Carapelle watershed is located, is characterized by predominantly family-run farming operations with an average farm size of 4.7 hectares [51]. The study area’s agricultural economy is dominated by cereal, followed by olives and vineyards [9]. The area faces significant socio-economic challenges including land fragmentation, generational renewal issues, and the need for enhanced agricultural competitiveness through increased farm efficiency. Environmental stressors such as drought (affecting 73% of surveyed farms), occasional flooding, and water scarcity further compound these economic challenges, necessitating adaptive management strategies and potentially innovative solutions like Nature-Based Solutions to ensure long-term agricultural sustainability and economic viability [51].

The monitoring station, in the Carapelle watershed, is equipped with two gauging systems, one for measuring suspended sediment concentration (SSC), and the other one for streamflow measurement (Q). An electromechanical and ultrasound stage meter from the Department of Civil Protection—National Hydrographic Service was utilized for Q measurements, while an infrared optical probe (Hach-Lange Solitax) was employed for SSC measurements. Both systems recorded data at 30 min intervals. Daily Q data, along with daily and sub-daily (30 min) rainfall data, were obtained from the Department of Civil Protection [50]. The integration of instantaneous flux as a product of SSC and Q was used to quantify the sediment load transported by the river. The monitoring report and instrument details can be found in [52].

The analyzed dataset (years 2007–2008), that is composed of 8 significant rainfall–runoff events, was characterized by with peak discharges ranging from 3.22 to 63.86 m³/s and corresponding sediment loads from 153 to 38,744 tons per event.

2.2. SDR Model Description

The InVEST framework (version 3.14.3) represents a sophisticated modeling approach for evaluating the influence of ecosystem modifications on human-derived benefits through production function methodologies [53]. The InVEST suite, developed by the Natural Capital Project, encompasses comprehensive modules addressing both supporting services (habitat quality and pollination) and final ecosystem services (annual runoff, water availability, carbon sequestration, urban cooling, and Sediment Delivery Ratio). These spatially explicit modeling tools generate detailed distribution maps while incorporating scenario planning capabilities that enable stakeholders—including government agencies and non-profit organizations—to evaluate various land use and climate scenarios against service provision levels [32].

Within this comprehensive framework, the Sediment Delivery Ratio (SDR) module serves as a critical component for analyzing erosional processes and sustainable water resource management. Operating at the spatial resolution of the input Digital Elevation Models, the SDR module quantifies annual soil loss for each pixel (defined as described in the following paragraph) and subsequently estimates the proportion of eroded material effectively reaching watercourses. The model assumes complete sediment transport to basin outlets upon stream entry, thereby excluding in-stream fluvial processes from calculations.

Contemporary research has validated the effectiveness of InVEST SDR for watershed-level assessments, demonstrating its utility for sediment retention service evaluation and erosion process analysis [41,54,55,56]. The SDR calculations reflect landscape connectivity through the integration of runoff patterns, sediment sources, and destination pathways, thereby modeling hydrological connectivity based on upstream area characteristics and flow path configurations to watercourses.

Hydrological connectivity assessment for individual pixels incorporates specific connectivity parameters, with theoretical maximum SDR values constrained to 0.8 [41] and calibration factors establishing relationships between SDR and connectivity indices [34]. The sediment retention service assessment utilizes surface sediment mapping methodologies [57] integrated with “Wischmeier’s Revised Universal Soil Loss Equation (RUSLE)”, which is particularly applicable within Mediterranean environments. This approach incorporates five fundamental parameters—rainfall erosivity, soil erodibility, slope length and steepness factors, land cover management practices, and supporting conservation measures—which each exhibit spatial and temporal variability influenced by additional environmental factors [53].

2.3. SDR Model Setup and Data Requirements

The InVEST SDR model version 3.14.1 was configured to operate at 20 m spatial resolution, matching the Digital Elevation Model (DEM) resolution. The model requires several spatially explicit input datasets and biophysical parameters that include DEM, Land Use/Land Cover (LULC), rainfall erosivity (R), soil erodibility (K), watersheds, and a biophysical table configuration; a comprehensive biophysical table was constructed containing land use-specific parameters (Land Use Code, Cover Management Factor (C), Support Practice Factor (P). Each of these parameters is explained in detail in the following sections. The model employs several critical parameters that control sediment delivery calculations:

(a)

Threshold flow accumulation (TFA): The minimum upstream contributing area required for a pixel to be classified as part of the stream network. This parameter defines where surface flow transitions from overland flow to channelized flow, effectively determining the stream network density and the points where sediment delivery calculations begin. A smaller TFA value creates a denser stream network (more stream pixels), while a larger TFA results in fewer, more prominent streams. A 150-hectare threshold was selected to accurately represent the perennial and ephemeral stream network observed in the Mediterranean area of the Carapelle watershed.

(b)

Connectivity parameters: Borselli k parameter [34], as reported in the equations below, was initially set to a default value of 2.0. IC₀ parameter: connectivity threshold parameter, initially 0.5. Maximum SDR: set to 0.8 following the recommendations in [35].

$${SDR}_i={\displaystyle \frac{{SDR}_{max}}{1+exp\left(\frac{{IC}_0-{IC}_i}{k}\right)}}$$

(1)

where SDR_max is the maximum theoretical SDR, and k and IC₀ are calibration parameters that define the shape of the SDR-IC relationship.

$$IC={\mathit{log}}_{10}\left({\displaystyle \frac{D_{up}}{D_{down}}}\right)$$

(2)

D_up is the upslope component defined as:

$$D_{up}= C_{th}^{-} S_{th}^{-} \sqrt{A}$$

where: $C_{th}^{-}$ is the average thresholded factor of the upslope contributing area, $S_{th}^{-}$ is the average thresholded slope gradient of the upslope contributing area (m/m). and A is the upslope contributing area (m²). The downslope is given by:

$$D_{down}={\sum }_i{\displaystyle \frac{d_i}{C_{th,i} S_{th,i}}}$$

(3)

where: d_i is the length of the flow path along the ith cell according to the deepest downslope direction (m), C_th,i and S_th,i are the threshold cover management factor and the threshold slope gradient of the ith cell, respectively. Both the upslope and downslope contributing areas are delineated from a multiple-flow direction algorithm.

2.3.1. Soil Loss

The InVEST SDR model was employed to estimate potential soil loss within the watershed, utilizing the Revised Universal Soil Loss Equation (RUSLE; [58]) and incorporating six key factors as follows:

$${RUSLE}_i={\left(R K LS C P\right)}_i$$

(4)

where:

R = rainfall erosivity (MJ mm ha⁻¹ h⁻¹ yr⁻¹);

K = soil erodibility (t ha h (MJ ha mm)⁻¹);

LS = slope length steepness factor;

C = cover management factor;

P = support practice factor.

The input data for the InVEST model were carefully selected to ensure an accurate representation of the analyzed watershed. Additionally, the 20 m resolution Digital Elevation Model (DEM) was used to derive slope (S) and slope length (L) factors, as well as to calculate flow direction and flow accumulation maps essential for connectivity analysis.

2.3.2. Rainfall Erosivity Factor (R)

The R (MJ mm h⁻¹ ha⁻¹ yr⁻¹) value for the year (j) was estimated using the equation by Ferro [59].

$$R_j=0.524 F_{a,j}^{1.59}$$

(5)

where:

$$F_{a,j}={\sum }_{i=1}^{12}{\displaystyle \frac{P_{i,j}^2}{P_j}}$$

(6)

where:

P_i,j represents the precipitation height (mm) for month i of year j;

P_j is the total annual precipitation (mm) for year j;

R was evaluated for 2007 and 2008 and spatialized, using measurements from eight stations. The stations, including Lacedonia, S. Agata, Bisaccia, Monteleone, Rocchetta, Ascoli, Bovino, and Castelluccio, provided R values ranging from 258.17 MJ mm h⁻¹ ha⁻¹ yr⁻¹ (S. Agata) to 630.01 MJ mm h⁻¹ ha⁻¹ yr⁻¹ (Bisaccia station; 2007) and from 471.39 MJ mm h⁻¹ ha⁻¹ yr⁻¹ (Lacedonia station) to 967.27 MJ mm h⁻¹ ha⁻¹ yr⁻¹ (Bovino station; 2008). These values were spatially interpolated using the Thiessen polygon method in QGIS, creating a zoned raster of rainfall erosivity (Figure 2). This method was selected since the gauging stations are well distributed between the mountainous and the lowland areas. This spatially explicit approach ensured that erosion potential—driven largely by rainfall energy—was realistically captured in the model. As shown in the Thiessen-interpolated map, the southern and western parts of the basin are exposed to higher erosivity values, while the central and northeastern regions are less impacted.

2.3.3. Soil Erodibility Factor (K)

The K factor was derived from detailed pedological data on the basin scale and spatialized into a raster layer for input into the InVEST model. The K factor quantifies the susceptibility of soil particles to detachment by rainfall and surface runoff. Soil properties, including organic matter and texture, were obtained from the Agro-ecological Characterization project of the Puglia region, ACLA2, with a resolution of 250 m. From this database, K values were derived using the following equation [60], given that the values of f (% of silt and very fine sand) are below 70%:

K = 2.77 × 10⁻⁷M^1.14(12 − OM) + 4.28 × 10⁻³(SS − 2) + 3.29 × 10⁻³(PP − 3)

(7)

where:

• OM = organic matter (%);
• SS = texture code: (1) very structured or particulate, (2) fairly structured, (3) slightly structured, (4) solid;
• PP = soil permeability class: (1) rapid, (2) moderate to rapid, (3) moderate, (4) moderate to slow, (5) slow, (6) very slow;
• M = soil texture parameter, defined as

$$M=f\left(f+g\right)=f(100-cl)$$

(8)

where:

$f$ = silt content (%);

g = coarse sand content (%);

cl = clay content (%).

As shown in Figure 3, the K values across the Carapelle watershed ranged between 0.019 and 0.036 t·ha·h·ha⁻¹·MJ⁻¹·mm⁻¹, with higher values concentrated in the central-western and southern parts of the watershed. These areas are characterized by shallow, poorly structured soils and steeper slopes, making them more prone to detachment. In contrast, lower K values occurred in the flatter, alluvial northeastern sectors. The K factor map was based on point observations from the Apulia Region ACLA2 project, which were spatially interpolated using Inverse Distance Weighting (IDW) and included in the regional soil map to create continuous coverage across the watershed.

2.3.4. Cover Factor (C) and Management Factor (P)

The C factor reflects the combined influence of vegetation cover and agronomic practices on soil surface conditions, with values ranging from 1.0 for completely bare soil to as low as 0.01 in areas covered by dense, protective forest canopies. As a critical parameter in assessing soil erosion vulnerability, the C factor plays a pivotal role in soil conservation strategies. Its magnitude is primarily governed by land use patterns and the characteristics of vegetative cover present on the surface. Land use data was obtained by combining the Land Use Map of Puglia and the Agricultural Land Use Map of Campania, both with a resolution of 100 m, extracted from their respective regional geoportals. The predominant land use is represented by winter wheat cultivation, covering 75% of the area, while other significant land uses include deciduous forests (7%), coniferous forests (4%), olive groves (3.3%), and pastures (7%). All land use categories were reclassified according to the Corine Land Cover 2012 classes (lucode represents a unique integer code assigned to each land cover class). A C factor (cover management) representing the soil protective effect of vegetation cover was derived from the literature [61] for each land use category. To reflect the 4-year crop rotation, C factor rotation was averaged at the yearly time scale. A single C factor value is required by the InVEST biophysical table structure. While monthly C factor variations would provide a greater temporal precision, the model framework necessitates annual average values for each land use classification.

The P factor represents the effectiveness of support practices that reduce the erosion potential by modifying flow patterns, flow velocity, or sediment transport capacity [58]. P values range from 0.1 for highly effective conservation practices to 1.0 for areas with no support practices. For baseline conditions (BASE), P was set to 1.0 for all land uses, representing the absence of specific erosion control measures, which reflects the current conventional management practices in the Carapelle watershed [9,62]. The C and P factor parameters were set in the biophysical table, a table (in CSV format) containing biophysical information related to the different land use and land cover types present in the study area.

2.3.5. Topographic Factor (LS)

The LS is derived from the 20 m Digital Elevation Model (DEM, Figure 1) using the algorithm proposed by [63] implemented within the InVEST SDR model. The combined LS (see Supplementary Figure S2) showed spatial variability from 0.02 in flat areas to >40 in steep terrain, with watershed average values of 2.8 reflecting the moderately sloping agricultural landscape of the Carapelle basin.

2.4. Model Calibration and Validation

A comprehensive analysis was conducted to assess the influence of key model parameters, specifically the Borselli k coefficient and IC₀ threshold value, on total sediment exports at the watershed scale. These parameters constitute critical components of the hydrological connectivity index (IC), which governs the SDR calculations within the InVEST framework. The baseline configuration employed a threshold flow accumulation (TFA) of 150 ha, a Borselli k coefficient of 2.0, and an IC₀ value of 0.5, as recommended in the InVEST user manual, while maintaining the maximum SDR at 0.8, following Vigiak [35].

The SDR model was calibrated comparing simulated and measured sediment yield from 2007 to 2008 at the outlet. The performance of the model was assessed by comparing the observed and simulated annual values by means of standard deviation. The selection of 2007 and 2008 was based on their representativeness of the average hydrological conditions for the study area. After that, a 20-year simulation, using observed rainfall data from 2000 to 2020, was carried out, and the model results were analyzed.

The InVEST SDR model generates the following spatially explicit outputs at the pixel level:

a

Primary erosion outputs: (i) soil loss (RUSLE): annual soil loss per pixel calculated using RUSLE (t ha⁻¹ yr⁻¹); (ii) sediment export: annual sediment delivered to streams per pixel (t ha⁻¹ yr⁻¹); to match what we mean by observed sediment yield data, from now on this will be referred to as sediment yield.

b

Retention service outputs: (i) avoided erosion (AVER) which is the difference between potential erosion (bare soil conditions) and actual erosion given current land cover (t ha⁻¹ yr⁻¹); (ii) avoided export (AVEX) which refers to sediment retained within the landscape that would otherwise reach streams (t ha⁻¹ yr⁻¹) [32].

2.5. Scenario Development for NBS Assessment

To evaluate the effectiveness of soil conservation practices in reducing sediment export in the Carapelle watershed, four NBSs scenarios were implemented using the InVEST SDR model starting from the calibrated Baseline Scenario (BASE), comprising I—contour farming (CF), II—no tillage (NT), and III—cover crops (CCs), which all represent individual scenarios, and IV—the combination of CCs and NT (Comb) to investigate the integrated effects of NT + CCs with cumulative parameter modifications. Each NBS scenario involved specific modifications to the C and P factors of the biophysical input table, reflecting realistic reductions in erosion potential and Sediment Delivery Ratios. C and P factor values were selected from both the RUSLE databases [58] and from other studies [12,46,61,64,65] carried out in Mediterranean areas in comparable slope and climate conditions. The scenarios were applied primarily to arable land (specifically wheat which occupies the majority), where erosion risk is highest and conservation potential is most impactful. These scenarios were compared to BASE to assess the effect in terms of changes in soil erosion, sediment yield, sediment deposition, avoided erosion, and avoided export.

• Contour Farming (CF)

CF involves performing field operations—such as plowing, sowing, and harvesting—along the contour lines of a slope. By creating micro-ridges perpendicular to runoff flow, this practice slows down water movement, increases infiltration, and prevents rill and sheet erosion. As described by Himanshu [66] the reduction in flow velocity due to contour ridges promotes water infiltration, stabilizes gentle slopes, and is especially beneficial for areas with 3–7% slopes, which dominate the Carapelle watershed’s cultivated zones. In the model, contour farming was implemented by reducing the P factor from 1.0 to 0.60 for relevant land use classes (wheat cultivations) and slightly adjusting the C factor to reflect more conservative soil disturbance following RUSLE guidelines for contour farming on 3–7% slopes typical of our study area [58,62,67].

II.

No Tillage (NT)

NT systems avoid mechanical soil turning. Instead of plowing, seeds are directly drilled into the soil, leaving plant residues on the surface and roots intact. This promotes soil structure, maintains macropores, and significantly reduces soil detachment and overland flow. Compared with conventional tillage, NT can reduce runoff and erosion while improving soil moisture conservation and organic matter content [65,68]. For the simulation, the C factor for croplands was reduced to 75%, while the P factor was maintained at 1, as these practices primarily affect surface cover (C factor) rather than flow modification [65].

III.

Cover Crops (CCs)

CCs are planted during the off-season and are not harvested. Instead, they act as a vegetative shield, minimizing raindrop impact, reducing runoff, and enhancing soil organic content through root biomass and nitrogen fixation [67,69,70]. In the model, this was reflected by the following: the C factor for croplands was reduced by 20%, as these practices primarily affect surface cover (C factor) rather than flow modification [67,71].

IV.

Combination of CCs and NT (Comb)

To explore the cumulative benefits of combining conservation techniques, a fourth NBS scenario was introduced by simultaneously applying NT and CCs on arable land. This combined approach was designed to maximize soil surface protection through residue and vegetation cover, minimal mechanical disturbance of soil profile, enhanced root structure, organic matter accumulation, and nitrogen fixation. In the InVEST SDR model, this scenario was parameterized by assigning low C factors (reduced by 80%) to cropland polygons with seasonal cover, and maintaining P factor unchanged, as no mechanical alteration was simulated [67,71].

These scenarios allow for a comparative analysis of how each practice—individually—modifies soil erosion, sediment export, and retention across the watershed. It is worthy to mention here that NBS parameter modifications (C and P factors) were applied to the entire RUSLE-SDR modeling chain, affecting (1) soil loss calculation through modified C and P factors in the RUSLE, (2) sediment delivery through altered connectivity patterns, (3) sediment export through cumulative watershed-scale effects.

3. Results

3.1. Sediment Yield and Calibrated Parameters

This initial parameterization yielded an annual sediment export of 30,687.15 t yr⁻¹ for the entire watershed. The analysis of the effect of the calibration parameters on the results consisted of changing the Borselli k and IC₀ parameters in increments of ±10% within a range of ±50% from their baseline values (

Table 1. Analysis of the effect of the calibration parameters on the results.

Borselli k	IC0	Sediment Yield t yr−1
1	0.5	25,316.89
2	0.50	30,687.15
3	0.50	43,416.70
3	0.25	46,038.50
3	0.55	42,903.40
2.8	0.50	41,366.8
2.8	0.25	45,298.60
3	0.6	62,517.30
3	0.75	68,728.10

). Moreover, the model was executed using these parameters independently and in combination to evaluate their respective and interactive effects on sediment export predictions.

Increasing the Borselli k coefficient from 2.0 to 3.0 while maintaining IC₀ = 0.5 resulted in a sediment export of 43,416.7 t yr⁻¹, representing a substantial increase of 41.5% relative to the Baseline Scenario. Further reducing the IC₀ to 0.25 while maintaining Borselli k = 3.0 enhanced hydrological connectivity, producing sediment exports of 46,038.5 t yr⁻¹, whereas increasing IC₀ to 0.55 marginally reduced connectivity, resulting in moderate decreases in sediment export. Other parameter configurations yielded sediment exports ranging between 42,662.4 t and 68,728.1 t yr⁻¹, depending on specific parameter combinations and biophysical table specifications, with the optimal parameter combination producing 43,416.7 t yr⁻¹ for the 2007 calibration year. To validate the model’s predictive capability, simulations were performed for 2008 using the calibrated parameters, with observed sediment yield data from the Carapelle watershed outlet monitoring program providing a reference value of 62,594.7 t yr⁻¹ (equivalent to 1.237 t ha⁻¹). The initial baseline simulation produced 46,211.69 t yr⁻¹, underestimating observed values by approximately 26%. Multiple simulation scenarios with varying parameter combinations produced export values ranging from 50,860.5 t to 67,983.5 t yr⁻¹, with the most accurate configuration employing Borselli k = 3.0 and IC₀ = 0.5, yielding 65,267.3 t yr⁻¹ and achieving only 4.3% deviation from observed data.

3.2. Soil Loss Rate Using the RUSLE

The InVEST SDR model first quantifies gross soil loss (computed via RUSLE), then through the Borselli [34] connectivity principle and SDR value it calculates the sediment actually delivered to streams.

The analysis of the gross soil loss (Figure 4) revealed substantial temporal variability between 2007 and 2008, with median erosion rates increasing from 2.43 t ha⁻¹ yr⁻¹ in 2007 to 3.88 t ha⁻¹ yr⁻¹ in 2008, representing a 60% increase in typical soil loss conditions. The 2007 erosion distribution exhibited a relatively compressed interquartile range (1.05–3.39 t ha⁻¹ yr⁻¹) and maximum value of 61.8 t ha⁻¹ yr⁻¹, while 2008 demonstrated both elevated median values and greater variability with an expanded interquartile range (1.71–5.28 t ha⁻¹ yr⁻¹) and maximum erosion reaching 94.5 t ha⁻¹ yr⁻¹. Both years maintained identical minimum values (0.000 t ha⁻¹ yr⁻¹), indicating the presence of areas with a negligible erosion risk, while the substantial increase in upper quartile values (3.39 to 5.26 t ha⁻¹ yr⁻¹) and maximum extremes underscore the heightened erosional intensity characterizing the 2008 conditions. The substantial differences in soil loss between 2007 and 2008 reflect contrasting meteorological conditions. Indeed, in 2007 the average rainfall for the wet season (Nov–March) was 286.4 mm, while for 2008 it was 387.4 mm. This rainfall variability explains the 60% increase in median soil loss from 2.43 t ha⁻¹ yr⁻¹ (2007) to 3.88 t ha⁻¹ yr⁻¹ (2008), demonstrating the strong climate dependency of Mediterranean erosion processes.

The spatial patterns of soil erosion (Figure 5) estimated by RUSLE for 2007 and 2008 show broadly similar distributions, with erosion hotspots concentrated along steep slopes and hilly sectors, while flat and vegetated areas remain largely stable under a low erosion risk (<10 t ha⁻¹ yr⁻¹).

To assess the reliability of the modeled output, Baseline Scenario (BASE) erosion rates (RUSLE), based on measured erosivity “R” and erodibility “K” data were compared to other models already used in the area (i.e., SWAT, AnnAGNPS). Specifically, Abdelwahab et al. [9] report average soil erosion rates of 8.8 t ha⁻¹ yr⁻¹, ranging from 0.0 to 54.7 t ha⁻¹ yr⁻¹ (99° percentile) for SWAT (MUSLE) and of 5.59 t ha⁻¹ yr⁻¹, ranging from 0 to 30.1 t ha⁻¹ yr⁻¹ (99° percentile) for AnnAGNPS (RUSLE and HUSLE).

These results were consistent with those obtained from InVEST (RUSLE). The average erosion rates were 4.4 t ha⁻¹ yr⁻¹ in 2007 and 6.6 t ha⁻¹ yr⁻¹ in 2008, (99° percentile), while the ranges were from 0 t ha⁻¹ yr⁻¹ to 40.52 t ha⁻¹ yr⁻¹ (99° percentile) for 2007 and from 0 t ha⁻¹ yr⁻¹ to 67.70 t ha⁻¹ yr⁻¹ (99° percentile) for 2008.

It can also be highlighted that, despite the differences in magnitude, due to the different input data used (i.e., climate), the European RUSLE2015 map produced by Panagos [12] yielded a similar range of soil erosion (0–53.5 t ha⁻¹ yr⁻¹) for erosion-prone zones located mainly in cultivated uplands.

3.3. Analysis of Sediment Dynamics Under NBS Implementation at the Watershed Scale

The InVEST SDR model analysis, using 20-year rainfall data, revealed substantial improvements at the watershed scale in sediment dynamics across all investigated parameters following NBS implementation (

Table 2. InVEST SDR annual average (20 years) watershed results for BASE and for all the NBS scenarios (Wet year: 2000; dry year: 2003).

		RUSLE (t ha−1) Gross Erosion	Sediment Yield (t ha−1)	Sediment Deposition (t ha−1)	Avoided Erosion (t ha−1) *	Avoided Export (t ha−1) **
Baseline	Average	6.58	1.45	5.09	24.18	8.65
	Wet year	10.15	2.23	7.85	37.27	13.34
	Dry year	4.24	0.94	3.28	15.64	5.60
CF	Average	4.26	0.94	3.29	26.50	7.77
	Wet year	6.56	1.44	5.07	40.86	11.98
	Dry year	2.74	0.61	2.11	17.15	5.03
CCs	Average	5.42	1.15	4.23	25.34	8.10
	Wet year	8.35	1.77	6.52	39.06	12.48
	Dry year	3.49	0.75	2.72	16.40	5.24
NT	Average	2.23	0.41	1.81	28.53	6.05
	Wet year	3.42	0.62	2.77	44.00	9.31
	Dry year	1.42	0.26	1.15	18.46	3.92
Comb	Average	1.94	0.35	1.58	28.82	5.77
	Wet year	2.97	0.53	2.42	44.45	8.88
	Dry year	1.23	0.22	1.00	18.65	3.74

Notes: * Vegetation contribution in reducing erosion from a specific area with respect to the bare soil. ** Vegetation contribution in trapping sediment originated upslope to a specific area.

). Under BASE, the watershed exhibited considerable sediment production, with normalized soil loss RUSLE values of 6.58 t ha⁻¹ yr⁻¹. Correspondingly, baseline sediment yield rate was 1.45 t ha⁻¹ yr⁻¹, while sediment deposition contributed 5.09 t ha⁻¹ yr⁻¹.

NT demonstrated the most pronounced individual NBS effectiveness across all sediment parameters as an individual scenario. RUSLE reductions averaged 66.1% over the study period, while sediment yield reductions reached 72.7%.

CCs provided consistent moderate effectiveness with normalized soil loss reductions of 17.6% and sediment yield reductions of 20.7%.

CF achieved intermediate effectiveness with average RUSLE reductions of 35.2% and sediment yield reductions of 35.1%.

The Comb approach delivered a superior performance across all scenarios, achieving the highest average RUSLE reduction of 70.5% and sediment export reduction of 75.9%.

The AVER demonstrated consistent increases across all NBS scenarios, indicating an enhanced soil conservation capacity. NT achieved the second-highest avoided erosion increases of 17.9% averaged across both years, while the Comb approach delivered the greatest avoided erosion benefits with 19.1% increases.

In both the wet and the dry year (rainiest year, 2003; driest year 2000), all the NBSs showed the same behavior in terms of percentage reduction in sediment. Specifically, in the wet year, where the average gross erosion value was higher than the extreme erosion rate of 10 t ha⁻¹, all the scenarios were effective in reducing the value back below the threshold.

3.3.1. Sediment Yield Spatial Distribution

A comparative evaluation of sediment yield (SY) distribution in the watershed for BASE using the InVEST SDR model and under varying NBS scenarios—cover crops (CCs), conservation/no-till farming (NT), contour farming (CF), and a combined approach CCs + NT (Comb)—reveals consistent mitigation effects that NBSs could have if implemented (Figure 6).

It is worthy to mention here that, since the InVEST SDR model operates at the pixel level rather than as a distributed hydrological model, we utilized existing sub-watershed boundaries based on topography [47] to better spatially analyze the results. Zonal statistics were applied to aggregate pixel-based soil loss and sediment export values within each sub-watershed.

Generally, BASE showed several sub-watersheds that overtook the threshold of severe erosion of 10 t ha⁻¹. Regarding the NBS scenarios, NT showed the highest effect with an average reduction in SY of about 72% for both years. CF and CCs reduced the SY by about 35% and 21%, respectively. The Comb scenario, combining NT and CCs, was found to be the most effective; the average reduction in SY was about 76%.

The analysis carried out in Figure 6 confirms that BASE exhibited the highest median values (1.43 t ha⁻¹ yr⁻¹), while the Comb scenario consistently yielded the lowest median values (0.76 t ha⁻¹ yr⁻¹), affirming the additive benefits of integrating conservation practices. Notably, the spread between median and maximum values was 2.52 t ha⁻¹ yr⁻¹, which reflects a heightened susceptibility to extreme erosive episodes possibly linked to inter-annual climatic variation—a trend frequently observed in erosion modeling studies [72,73,74].

Individual measures such as CCs, NT, and CF substantially reduced the median SY relative to BASE. This is in line with empirical and simulation studies showing that CCs and NT can lower sediment yield by 25–50% or more depending on the landscape and rainfall context [75,76]. The even lower median values in the Comb scenario underscore the value of stacking NBS interventions for optimal erosion control, echoing the findings from multi-practice field trials and watershed modeling. However, the persistence of high-magnitude outliers under all scenarios highlights the limitations of NBSs under extreme weather and emphasizes that, while these interventions mitigate the central tendency and frequency of erosion events, they do not fully eliminate the risk under exceptional conditions [77]. This agrees with contemporary literature emphasizing the context-specific efficacy of management practices and the need for integrated, adaptive strategies to manage rare but intense sediment export events in agricultural systems [78]. The overall findings reinforce both the efficacy and challenges of NBSs for sediment management, recommending their continuous implementation and the necessity of monitoring to adaptively respond to annual climatic fluctuations and system outliers.

3.3.2. Avoided Erosion Evaluation

Overall, the combined NBS scenarios consistently maximize AVER (Figure 7) across diverse topographic and hydrologic settings, but the relative spatial efficacy of individual practices shifts with annual hydroclimatic conditions: NT is most effective in downstream depositional zones, CCs in mid-watershed hillslopes, and CF in headwater steep slopes. These findings suggest that optimizing sediment control requires not only integrated practice packages but also the strategic placement of specific NBSs according to local slope–runoff dynamics and anticipated climatic variability.

The spatial patterns of AVER across the watershed (Figure 7) reveal distinct hotspots of conservation benefit that vary by practice, underscoring the importance of tailored NBS deployment. Under the Comb scenario, AVER reached high values (84.14 t ha⁻¹ yr⁻¹) predominantly in the southwestern and central sub-catchments, reflecting the synergistic effects of CCs and NT in areas with steep slopes and intense rainfall. NT alone produced a similarly extensive band of high AVER, while CCs and CF exhibited the more moderate spatial coverage of erosion reduction, particularly in the watershed’s northeastern uplands.

4. Discussion

4.1. Model Calibration and Performance

The application of the InVEST SDR model in the Carapelle watershed confirms that this model can be applied in Mediterranean areas [39,79]. The parameter analysis confirms that the Borselli k coefficient exhibits a direct proportional relationship with sediment export, while IC₀ demonstrates an inverse proportional relationship, consistent with previous research by [37,39,41], and [53]. The analysis reveals that minor parameter adjustments can induce significant variations in model outputs, emphasizing the critical importance of appropriate parameter calibration when applying the InVEST SDR model in Mediterranean environments characterized by complex topography and intensive land use pressures. In this study, the calibrated Borselli k was 3. These results align with the methodology employed by [39], who performed systematic calibrations of the k coefficient starting from a baseline value of 2.0 with increments of 0.2, identifying optimal k values ranging from ~3 to 8 for the Koga and for the Andassa watershed, respectively. The value of the calibrated IC₀ was 0.5. This confirms what was stated by Vigiak [35]: that this parameter is mostly landscape-independent.

The initial underestimation of observed sediment yield by approximately 26% highlights the importance of parameter optimization and accurate land use characterization in heterogeneous Mediterranean catchments, where complex interactions between topography, climate, and anthropogenic factors influence sediment transport processes. The successful calibration that reduced the deviation to only 4.3% demonstrates the model’s capacity to provide reliable predictions when appropriately parameterized. The substantial sensitivity observed, particularly with extreme parameter values, underscores the necessity for rigorous calibration procedures and reinforces the importance of field-based validation data for model parameterization. Overall, the model demonstrated reasonable agreement with observed sediment data following appropriate parameter calibration, reinforcing previous findings [36,39] regarding the critical importance of biophysical table accuracy and hydrological connectivity parameter optimization for reliable sediment delivery predictions in complex Mediterranean watersheds. The comparison between InVEST SDR and the previous model applied in the study area revealed important trade-offs. InVEST SDR was able to identify the sediment source area with a distribution and a magnitude of values comparable with those obtained from SWAT and AnnAGNPS [9]. The mismatch between the results may be attributed to differences in the fact that the SWAT model in the MUSLE equation does not account for the rainfall erosivity, but rather the surface runoff and the peak flow. AnnAGNPS instead uses a different equation to assess the R factor [80,81]. Moreover, both SWAT and AnnAGNPS were calibrated at the daily time scale.

With respect to conceptual models, which require extensive input data (meteorological, soil, land use, management practices) and significant calibration effort, InVEST SDR operates with more accessible inputs and is easier to implement by the user. For these reasons, NBS assessment in Mediterranean agricultural areas, where sheet and rill erosion dominate, can be successfully carried out using InVEST SDR, which provides adequate accuracy and greater accessibility. However, InVEST SDR cannot capture some processes that SWAT and AnnAGNPS do, such as stream flow, overall hydrological balance, and in-stream sediment dynamics.

4.2. Temporal and Spatial Soil Erosion and Sediment Dynamics Assessment

The temporal variation observed in this study aligns with established patterns of Mediterranean soil erosion variability, where rainfall erosivity serves as the primary driver of inter-annual soil loss fluctuations, with studies demonstrating that over 90% of temporal erosion variation can be attributed to changes in precipitation intensity and distribution patterns [23].

The BASE scenario soil loss rates median of 7.35 t ha⁻¹ yr⁻¹ falls within the range of Mediterranean erosion assessments. Mediterranean studies demonstrate considerable regional variation, with annual soil loss rates ranging from 2.3 t ha⁻¹ yr⁻¹ in Spanish pre-Pyrenees catchments [82] to 20.5 t ha⁻¹ yr⁻¹ in western Crete watersheds [83]. Intermediate values include 3.64 ± 5.14 t ha⁻¹ in Spain’s Mijares watershed [84] and 15.1 t ha⁻¹ yr⁻¹ in Portugal’s Alqueva Dam watershed [85]. Comparative InVEST SDR model applications in other semi-arid regions demonstrate the potential for substantially higher erosion rates under different climatic and topographic conditions. For instance, Gashaw [39] reported mean annual soil loss rates of 58.2 and 27.3 t ha⁻¹ yr⁻¹ for watersheds in the Ethiopian Highlands (Andassa and Koga, respectively), while Ismaili Alaoui et al. [79] found highly variable soil loss in Morocco’s Oued Beht watershed (0–184 t ha⁻¹ yr⁻¹) with a watershed average of 16.4 t ha⁻¹ yr⁻¹.

Regional studies in Apulia reveal similarly variable erosion patterns. Using the RUSLE model, Petito et al. [46] reported average soil loss rates of 2.3 t ha⁻¹ y⁻¹ under conventional management and 2.1 t ha⁻¹ y⁻¹ with conservation agriculture in annual croplands. Conversely, Karydas et al. [86] found lower mean annual erosion rates of 0.87 t ha⁻¹ yr⁻¹ in the Candelaro River Basin using the G2 model with Sentinel-2 imagery, with peak erosion occurring during the autumn months. Moreover, studies carried out with conceptual models based on USLE-derived equations (i.e., RUSLE, MUSLE) showed similar values [87,88] thus confirming the reliability of using the InVEST SDR module.

4.3. Nature-Based Solutions: A Paradigm Shift in Erosion Control

The findings of this study strongly support the NBS framework for watershed management. Recent assessments of NBSs in Southern Italy have demonstrated that stakeholders recognize the critical importance of sustainable soil and water management practices, with drought, floods, and water pollution identified as primary challenges, and soil erosion as a major concern [51]. NBSs (such as CF, NT, CCs) help mitigate runoff, reduce erosion, and ensure a sustainable water supply by retaining water on-site, allowing it to slowly seep into the ground or be used for irrigation and groundwater replenishment. The ecosystem service approach employed in this study aligns with the NBS philosophy of working with natural processes to achieve multiple benefits simultaneously.

The simulated NBSs in this study—namely contour farming (CF), conservation/no tillage (NT), cover crops (CCs), and the combination of NT and CCs (Comb)—demonstrated variable efficacy in reducing sediment export and enhancing sediment retention. These results emphasize the importance of selecting NBSs that disrupt sediment connectivity and modify overland flow pathways, particularly in erosion-prone Mediterranean agroecosystems.

The superior performance of NT aligns with extensive field validation studies demonstrating that NT systems fundamentally alter watershed hydrology and sediment dynamics. Prasad et al. [89] reported that NT with residue management achieved sediment loss reductions of up to 86% through enhanced surface roughness, improved infiltration rates, and the preservation of soil aggregates [68]. Our results (72% average SY reduction) fall within this range.

CCs demonstrate significant effectiveness in reducing sediment losses from agricultural systems through multiple mechanisms. Field studies show that CCs can reduce sediment loads by 43–56% [90,91], with some studies reporting up to an 85% reduction in total suspended solids [92]. Cover crops achieve these benefits by intercepting rainfall kinetic energy and reducing runoff velocity [93]. Our normalized results (1.15 t ha⁻¹ yr⁻¹ SY under CCs vs. 1.45 t ha⁻¹ yr⁻¹ BASE) closely match these experimental findings.

CF demonstrated consistent intermediate effectiveness across all sediment parameters in our study, achieving average reductions of 35.2% in SY. These results align closely with recent NBS effectiveness studies using similar modeling approaches. Zantet Oybitet et al. [94] evaluated contour farming effectiveness using the SWAT model in Ethiopian watersheds and reported sediment yield reductions of 48% for contour practices, which falls within the upper range of our findings.

The Comb scenario yields synergistic benefits, achieving up to 76% of SY with respect to BASE. These results are consistent with watershed-scale research by [95], who documented that combined practices, particularly crop rotation with cover crops, achieve the greatest erosion reductions of 38–46%. The mechanistic basis for synergy involves complementary temporal and spatial protection mechanisms. NT provides year-round structural protection through residue retention, while CCs offer additional protection during critical periods between cash crops. This temporal complementarity is supported by studies demonstrating that CCs are most effective during winter months when NT residues may be decomposed or displaced [96,97].

4.4. Sediment Retention and Water Quality: Ecosystem Service Quantification

The quantification of sediment retention as an ecosystem service provides critical insights for watershed management and policy development. The substantial avoided erosion values documented in this study represent tangible benefits that can be valued economically and integrated into the payment for ecosystem services schemes. The InVEST framework’s capacity to link ecosystem service supply to demand, integrating information on where people live and the areas requiring protection, provides a robust foundation for such valuation. The economic valuation of NBSs remains a significant gap in current assessments, with the need for comprehensive evaluations that incorporate livability improvements, water quality enhancement, and socio-cultural benefits [51,98]. This study contributes to filling this gap by providing quantitative estimates of sediment retention services that can inform cost–benefit analyses of conservation investments. The spatial distribution of retention services, with the highest values in forested and shrubland areas, underscores the critical importance of protecting existing natural vegetation and strategically restoring degraded areas. The consistency of retention services across different years, despite rainfall variability, demonstrates the reliability of natural systems in providing ecosystem services under Mediterranean climatic conditions [99].

This study complements our previous comprehensive modeling efforts in the Carapelle watershed [9,10,47,50,100] by specifically addressing the practical need for accessible NBS assessment tools. While our previous work with SWAT and AnnAGNPS provided detailed process understanding, this study evaluates whether simplified ecosystem services frameworks can provide reliable guidance for conservation practitioners and policy makers with limited technical resources or extensive datasets.

4.5. Study Limitations and Future Research

While the InVEST SDR model provided valuable insights into erosion patterns and NBS effectiveness, several limitations should be acknowledged. The model’s focus on sheet and rill erosion may underestimate the total sediment yield in areas where bed and bank erosion are relevant, especially during extreme rainfall events, such as in the area studied by Ricci et al. [10]. Moreover, it is known that models can be affected by uncertainty, which can be related to input data (i.e., the integration of data from different sources, low resolution, interpolation), to model uncertainty (i.e., missing processes, the simplification of processes), and parameter uncertainty (i.e., non-uniqueness) [101]. Specifically, in this work, input uncertainty can be influential, since the data used had different sources and resolutions.

While these limitations should be acknowledged, they do not diminish the value of our findings for Mediterranean watershed management and provide important considerations for model applications in similar environments.

5. Conclusions

This study showcases the effectiveness of the InVEST SDR model when applied to the Carapelle watershed, providing critical insights into Mediterranean soil erosion and sediment retention under alternative management practices. The calibrated SDR model accurately captured spatiotemporal sediment dynamics, validated by both local measurements and performance metrics.

Nature-Based Solutions, particularly the combination of no tillage and cover crops, yield the greatest and most consistent reductions in sediment export (up to 75.9%) and soil loss (up to 70.5%). These outcomes are robust across variable rainfall conditions, with spatial hotspots identified for targeted intervention.

The results have direct implications for watershed management policy, emphasizing the strategic placement of NBSs in terrain-vulnerable zones, the economic evaluation of ecosystem service gains, and integration with participatory local management. Future research should address gully erosion, long-term climatic variability, and the economic valuation of ecosystem services to optimize landscape conservation investments.

These findings provide a replicable framework and evidence base for Mediterranean regions seeking to balance agricultural productivity with environmental sustainability in the face of accelerating climate and land use pressures.

Several research priorities to advance our understanding of NBS effectiveness in Mediterranean watersheds are recommended:

▪

Long-term monitoring programs to validate NBS implementation effectiveness over extended periods and varying climatic conditions.

▪

LClimate change integration through scenario modeling to project future erosion risks and NBS performance under changing precipitation patterns and temperature regimes.

▪

LEconomic valuation frameworks for comprehensive cost–benefit analyses of ecosystem services, including avoided damages from erosion and enhanced water quality benefits.

▪

LParticipatory approaches that engage local farmers and stakeholders to enhance NBS adoption rates and ensure sustainable implementation.

▪

LMulti-scale analysis extending from plot-level experiments to regional assessments for broader policy applications.

These research directions will strengthen the scientific foundation for watershed management decisions and support the transition toward sustainable agricultural practices in Mediterranean environments.