Using the Soil and Water Assessment Tool (SWAT) to quantify the economic value of ecosystem services

Modeling tools simulate the functioning of ecosystems and their interactions with human activities, helping decision makers understand how interventions impact ecosystems and evaluate management strategies. This leads to informed decisions that balance human development and environmental protection. Among these models, Soil and Water Assessment Tool (SWAT) stands out for its ability to simulate multiple biophysical processes that can be linked to the provision of ecosystem services (ES). Although SWAT has been successfully applied for the evaluation of ES, the development of complementary approaches that translate the results of SWAT into monetary terms is still in its early stages. To narrow this gap, this review article aims to provide a comprehensive assessment of the literature on the relationship between SWAT model results and economic analysis. Specifically, the review summarizes the research conducted on the use of SWAT model results to estimate economic values, including the different methodologies used and the types of economic values estimated. The review will also discuss the limitations and challenges of these approaches, provide a critical evaluation of the strengths and weaknesses of the research in this area, and provide recommendations to strengthen SWAT application for the economic evaluation of management strategies.


| INTRODUCTION
Ecosystem services (ES) refer to the various benefits that humans derive from the natural environment. These services include things like clean air and water, soil fertility, pollination, pest control, and carbon sequestration (Costanza et al., 1997). They are essential for supporting human well-being and economic activities, and are, therefore, in need of careful management. While some ES, such as food production, are more obvious and easily quantifiable, others like climate regulation, water purification, or recreational opportunities are often overlooked because their benefits are indirectly realized, or they are intangible. However, their importance cannot be overstated, as they help to maintain the ecological balance and sustain human activities and well-being. It is crucial to recognize the value of ES and incorporate them into decision-making processes (Edens & Hein, 2013). This means taking into account the impact of human activities on ES, and finding ways to sustainably manage them for the benefit of both current and future generations.
Traditionally, ES management has relied on the use of observed data, such as flow rates, piezometry, water quality, crop production, and other ES-related biophysical variables. This information can be useful to assess the current potential or flow of some ES, or even assessing trends in relation to the same. However, since monitoring this data is costly, these measurements may have very limited spatial and temporal coverage, specifically in developing countries. When it comes to predicting the effects of different strategies on ES provision, relying solely on observed data can be very uncertain. This is because ES are influenced by a wide range of biophysical and socioeconomic factors, many of which may not be fully understood or even observable. To address these uncertainties, modeling tools that simulate the functioning of ecosystems and their interactions with human activities can be useful (Bagstad et al., 2013). By using such models, it is possible to assess the potential impacts of different management strategies on ES provision under a range of plausible scenarios.
However, there is still a wide gap between the need to evaluate the impact of human actions on our ecosystems and the development of modeling applications that can assess it in useful terms for decision-making, particularly with regard to the provision of ES. Models such as SIMGES, RIBASIM, WEAP, AQUATOOL, PATRI-CAL, or SWAT, have the general purpose of predicting how hydrological systems respond to different drivers, such as climate or land use change, and how this affects the provision of ES. There are also spatially explicit ES modeling tools that are used to assess and quantify the monetary value of ES, such as InVEST or ARIES, that have been used in land-use planning, conservation prioritization, and natural resource management (Ochoa & Urbina-Cardona, 2017). Among all these models, Soil and Water Assessment Tool (SWAT) (Arnold et al., 1994;Neitsch et al., 2011) stands out for the following reasons. SWAT is a continuous model capable of obtaining data on biophysical variables, at least on a daily basis, and at the hydrologic response units (HRU) scale, which allows adapting the results to the desired sociospatial evaluation context. SWAT integrates a set of very different ecohydrological process models, which are involved in the provision of various ES, allowing for a more comprehensive assessment of the effects (benefits or trade-offs) of management actions (Glavan & Pintar, 2012). One of its strongest points is the development of various calibration and validation functions of parameters to ensure the accuracy of its predictions of various hydrological (evapotranspiration [ET], flow, runoff, etc.) and not strictly hydrological processes (agricultural production, forest biomass, etc.) (Arnold et al., 2012;Nair et al., 2011).
Although SWAT (or its most updated version SWAT+) has been successfully applied for the evaluation of various ES, the development of complementary approaches that translate the results of SWAT (biophysical variables) into monetary terms and used these results to evaluate management strategies to support decision-making is still in its early stages (Francesconi et al., 2016). To narrow this gap, this review article was aimed to provide a comprehensive and critical assessment of the literature on the relationship between SWAT model results and economic analysis. Specifically, the research that has been conducted on the use of SWAT model results to estimate economic values will be summarized, including the different ES that have been assessed and the types of economic valuation approaches that have been applied. It will also be discussed the limitations and challenges of these implementations and provide future research directions to address these gaps.

| MATERIALS AND METHODS
I conducted a systematic literature review on the use of SWAT model to estimate the economic value of ES. The purpose of this review is to synthesize the existing literature on this topic, evaluate the methods used in previous studies, and identify knowledge gaps and research needs. A systematic literature review is a rigorous and transparent approach to reviewing existing literature on a specific research topic. It involves a comprehensive search strategy, structured data extraction and analysis, and predefined inclusion and exclusion criteria to minimize bias (Petticrew & Roberts, 2008).
This review was conducted targeting exclusively studies published in peer-reviewed journals, that have been recently published (2010 or later). To identify relevant studies to be considered in this review, 51 papers that use SWAT to model ES were consulted. These have been gathered from June 2021, in the framework of a national project whose objective was to elaborate a methodological approach to evaluate river basin management actions based on the economic value of ES modeled by SWAT ("Integration of ecological status and environmental services for the design and prioritization of management measures [EESAM]" national project [ref. ACA210/18/00028]). These papers were screened and identified 11 that complied the required criteria. They were complemented by searching papers in the online database Web of Science, using the search term "economic* and ecosystem* and service* and SWAT," which retrieved 60 results. From that list of studies, 13 other papers relevant studies were obtained. Finally, other potential contributions were searched through the SWAT Journal Article Literature Database (Gassman et al., 2014). Concretely though papers that have assigned "economic assessment" in the Primary (n = 3) and Secondary Application categories (n = 16). Out of these 19 papers, 4 studies of interest for this review were identified. In total, 28 studies were selected. The review was guided by the following research questions: (a) When and where was the study conducted, (b) what kind of basin was modeled (in terms of land use and size), (c) which ES were estimated, (d) which SWAT output variables were used, (e) how was the model calibrated, (f) which economic valuation approach was applied for each ES, and (g) what was the purpose of the modeling and economic assessment. Below, the formulated questions based on the reviewed literature are responded. Next, it is discussed the advantages and disadvantages found in the use of SWAT for the economic analysis of ES and its use for decision-making, and possible solutions to improve its applicability in the future.

| CHARACTERIZATION OF THE SELECTED STUDIES
3.1 | Location and year of the studies Figure 1 maps the location of the 28 reviewed studies. By countries, the United States is where the most studies have been carried out, with eight in total. China is second in terms of the number of contributions, with four publications. By continents, only Oceania and Central America do not have any selected study. It is worth noting the lack of studies carried out in the Middle East and North Africa countries, despite the scarcity of water in these countries. Regarding the date of publication (Figure 2), it can be said that the largest number of studies were published between 2018 and 2022 (n = 18) and especially in 2021 (n = 6), which shows that the use of SWAT for the economic valorization of ES is becoming increasingly mainstream.

| Characteristics of the basins (size and land use)
A simple characterization of the basins that were modeled with SWAT in the compiled studies was conducted by observing the size 1 and the type of basin based on the predominant land use. Of the watersheds, 25 out of 28 clearly reported the size of the modeled watershed (or it was easy to find it on the Internet). On average, the average size of the modeled basins is about 102,130 km 2 . However, the median is only approximately 2800 km 2 . This is explained by the large variability in the size of the basins of the reviewed studies (standard deviation = 227,709.18). The largest modeled basin was the Danube with 802,500 km 2 (Karabulut et al., 2016), and the smallest was the Cierny Hron River basin in Slovakia, with only 2.92 km 2 (Gallay et al., 2021). Based on these size values, the basins were categorized as large, medium, and small, based on the percentiles 1/3 and 2/3. That is, the basins with less than 484 km 2 are small, with more than 13,121 km 2 are large, and the rest are medium-sized.
According to the dominant land cover, the SWAT modeled basins were categorized based on these classes: agricultural, agroforest, agriurban, forest, grassland, and peatland. The main kind of basin of those selected in this study was agricultural (n = 12). This is not surprising as SWAT is an eco-hydrological model created with the goal of simulating how different land management practices in large and complex agricultural watersheds could impact various aspects of the hydrologic cycle (Arnold et al., 1998). However, 11 of the basins' studies were classified as agroforest or forest watersheds, what also demonstrate the wide application of SWAT in study areas where forests have a strong presence (Marin et al., 2020).

| Valued ES and output variables used
Despite this review departs from 28 studies, in total, 38 estimations of the values of ES were reported (Table 1). This is because some studies quantified more than one ES, such as the case of Ricci et al. (2022), which estimated the value of three ES (food provisioning, soil retention, and nitrogen retention). It was a common practice to evaluate more than one ES among the different studies; however, for this analysis, only ES quantifications depending on SWAT output variables were considered. The results on this dividing the ES by provisioning, regulating, and cultural ES will be described.

| Provisioning ES
Out of the 37 ES assessed with economic approached, 14 can be grouped as provisioning. Of these, the most numerous type of ES is water provisioning (n = 8). For this provisioning, ES it was considered the use and find out that, in all eight valuations, water for irrigation was estimated. For instance, Garg et al. (2012), in a large agricultural watershed in India, used the output variables of SWAT in relation to effective rainfall, total irrigated water, and evapotranspiration and the crops yield to estimate the value of this ES in four different types of scenarios with different land uses (crops) and agricultural management practices. On two occasions, the water provisioning was for more than one use, concretely irrigation, residential and industrial. This was done in a small-sized agroforest watershed in China, based on the water yield (mm) outputs of the SWAT model (Tu et al., 2022). Food was the second most common provisioning ES quantified among the reviewed publications (n = 4). This is mostly achieved by the integration within SWAT of the US Department of Agriculture (USDA) Environmental Policy Integrated Climate (EPIC) model to simulate plant growth options (Williams et al., 1989). Using crop yield outputs of SWAT, in this case, corn, durum wheat, alfalfa, vine, olive, winter pasture, alsike clover, and rapeseed (for biofuel purposes), De Girolamo and Lo Porto (2012)

| Regulating ES
Most ES estimated by using SWAT output variables were classified as regulating (n = 20). Within this group, the categories of services that have been covered, ranked from most frequent to least frequent, are as follows: Sediment retention (n = 7), nitrogen retention (n = 5), flow regulation (n = 4), phosphorus retention (n = 3), carbon sequestration (n = 1). Note that it was decided to separate nitrogen and phosphorus retention ES into different groups as it was a common practice in the reviewed studies. In this way, results demonstrate that sediment retention in the basin is the most frequently estimated regulating ES. This is made possible by the adoption of the Modified Universal Soil Loss Equation (MUSLE) (Williams, 1975) to estimate Erosion and sediment yield for each HRU. Ashagre et al. (2018) quantified the sediment retention ES in a large and forested basin in Tanzania using the estimated sediment concentration values at specific points in the channel network and the respective changes when applying the conservation scenarios. The second most common regulating ES was nitrogen retention. In a small agriurban water in Belgium, Cools et al. (2011) modeled with SWAT instream concentration of nitrate and total nitrogen (TN) (among other physicochemical variables) to quantify the supply of this ES by implementing several measures. By flow regulation, ES it is referred to the capacity of natural F I G U R E 2 Cumulative frequency of studies according to year of publication. *The value of 2023 does not cover the whole year but just the papers published in January and February.
T A B L E 1 List of ecosystem services that have been estimated using SWAT and its output variables, and the followed approach for economic valuation. ecosystems to regulate the river flows in a watershed, thus ensuring reduced flood risk an increasing water availability. Four studies in this review evaluated this ES. Among them, it can be highlighted the study conducted by Gallay et al. (2021) in a small and forested watershed in Slovakia, in which surface run-off SWAT output variable served as an input to run an HEC-RAS model to map flood hazard. Phosphorus retention ES effect and value were estimated in three occasions. This was the case of a small agricultural watershed in Iowa that used phosphorus concentration changes as a biophysical variable to estimate this service (Johnson et al., 2016). Finally, only Tarigan et al. (2021), in a medium-sized basin predominantly covered with peatlands, used the river flow SWAT output variable to estimate fluvial carbon loads using literature-based concentrations of total organic carbon (TOC) for disturbed and undisturbed peatlands.

| Cultural ES
As identified by Francesconi et al. (2016) in their literature review on research papers quantifying ES using SWAT, studies typically model cultural ES less frequently than other types of services, likely due to the need for an understanding of human perceived benefits. However, biophysical variables that can be obtained from SWAT models can provide valuable information for the economic valuation of cultural ES, as they can help quantify the ecological and environmental aspects that underlie these services. In this review, only three papers were considered to quantify this kind of ES. One of them is Choi and Ready (2021). In a middle-sized agroforest basin in Pennsylvania, they calculated the National Sanitation

| Calibration items
When coupling biophysical models to socioeconomic models to quantify the economic value of ES, it is very important to enhance the performance of the former to provide more reliable and informative estimates of the economic value of ES. This is achieved by calibrating and validating the models against empirical observations. Overall, SWAT is an appropriate model when observed data are available for calibration and validation. Its ability to incorporate observed data and simulate a range of watershed processes makes it a powerful tool for quantifying eco-hydrological processes (Arnold et al., 2012).
Of the compiled studies, three did not report the details of the calibration and validation process of any of the used outcome variables to estimate the ES values. Most of the studies reported to calibrate and validate the hydrological processes, which are considered the foundation of the model (Arnold et al., 2012;Nair et al., 2011), most of them based on observed data from gauging stations. Only one study did not calibrate or validated any of the objective variables used in the analysis, because one of their objectives was to conduct a model performance comparison between SWAT and SWAT-FT models (Wang et al., 2021). Regarding the rest of modeled processes, although not the most common, output variables that have not been properly calibrated are used. In the case of sediments, three out of nine studies did not conduct the corresponding calibration and validation procedure. In the case of nitrogen, it happened in two out of eight cases, and regarding phosphorus, one out of five. In the case of the crop yield, using this output variable without having been calibrated/validated is a more common practice, since this was observed in three out of eight cases. Crop yield observed data used for calibration and validation is often obtained from statistical repositories representing administration limits that might not fit well with the basin delimitations (Ricci et al., 2022). However, in all these cases, authors emphasize the effort made to parametrize correctly the agricultural (or pasture) management practices of the region (tillage operations, fertilization, planting, harvesting, or irrigation) ( Table 2).

| Economic valuation methods
The benefits provided by ES can be difficult to quantify and value, as they often involve complex ecological and social interactions. The application of different economic valuation approaches serves to capture the full range of values associated with ES, and to tailor their valuation methods to the specific context and research question (Pascual et al., 2010). In this review, it was overviewed which are the most common valuation approaches used to valuate ES services based on SWAT model results (Table 1). Starting with the water provisioning services, the most common valuation method applied was the market price-based approaches, that is, estimating the value of a good or service by observing prices paid by buyers and sellers in a market (Pascual et al., 2010) (five out of nine times). In three studies, it was estimated economic water productivity, that is, the amount of economic value generated per unit of water used in a production process (Garg et al., 2012;Liu et al., 2022;Xue et al., 2021). Economic water productivity is not a direct valuation method for this kind of services, but it was considered in this review as it can be relevant for decision-making about water management and allocation.
All the four food provisioning ES valuation exercises in this review were based on market price valuation.
Regarding sediment retention ES, the most common valuation method used was avoided cost approach, that is, the value of ES by calculating the costs that would be saved through their provision, compared to the costs that would be incurred in their absence. The second most common valuation approach for this kind of ES (two out of seven) is the replacement cost, that is, approximating the expenses associated with replacing natural ES using artificial technologies. This is an expectable result as avoided cost and replacement cost methods are frequently employed to assess regulation services (Gómez-Baggethun et al., 2013). to assess the economic value of sediment retention, quantifying the costs of increasing a dam's height to maintain the reservoir capacity or the number of downstream people that would be affected by pump failure and would have to find more expensive water supply sources. In the case of the nitrogen retention ES, three out of the five studies use either avoided costs (n = 2) or replacement costs (n = 1). Two other studies applied mitigation or restoration cost methods, which are based on expenses incurred in mitigating the adverse effects caused by the depletion of ES or in restoring those services (Pascual et al., 2010). For instance, Bawa et al. (2023), in a small agricultural watershed in Georgia, used the opportunity cost of food production to rate the different scenarios to achieve water quality regulations (regarding nitrate and phosphorus).
Regarding flow regulation ES, the four cases applied different valuation methods (replacement cost, avoided cost, mitigation or restoration cost methods, and production function). The later was applied by de Oliveira Serrão et al. (2023) to estimate how climate change would affect the hydropower production (in economic terms) in one hydropower plant. The three phosphorus retention economic valuations were done by using avoided cost, mitigation, or restoration cost methods, and willingness to pay from a benefit transfer. The later was used by Johnson et al. (2016) by using a meta-analysis from the United States that estimated annual average per household values for improvements in surface water quality, based in the changes in a water quality ladder (Johnston et al., 2005). Of the cultural ES valuation experiences reflected in this review, two approaches applied based on SWAT modeling results are highlighted. One of these is Choi and Ready (2021), who applied the SWAT output water quality results to estimate the WQI values obtained by applying different scenarios to conduct a choice experiment survey. The other relevant valuation of cultural ES was conducted by Liu et al. (2019), who used Secchi disk depths obtained from SWAT to conduct a hedonic pricing method and estimate the economic impact in terms of esthetic appreciation of the improvement of water clarity.

| Purpose of ES economic valuation
Monetary valuation of ES can help decision makers understand the economic trade-offs of different landscape decisions and policies (Marre et al., 2016). This section aims to summarize the purposes of conducting economic valuation of different ES using SWAT-modeled results. For this purpose, only articles that go beyond ES economic assessment with SWAT results and analyze these results to support decision-making in a more direct way were considered (23 out of 28). In this sense, studies can be classified into two groups: those that compare a set of plausible scenarios (n = 18) and those that apply optimization processes to find the best possible scenario (n = 5). The first group is more diverse in the kind of scenarios that were evaluated. Most of them imply changes in the land use/land cover input data (LULC), either to carry out a retrospective analysis of the provision of ES based on LULC layers from previous times or to compare the current situation with land conservation/restoration scenarios or counterfactual scenarios. An example of the first case is provided by Zhang et al. (2022). They modeled a middle-sized agroforest basin in China to evaluate the anthropogenic impacts in terms of LULC modification on the flow regulation services, using LULC data and models from 2008, 2013, and 2018. Instead, in a small-sized and forested basin in Brazil, Kroeger et al. (2019) analyzed the potential economic impact, in terms of sediment retention ES, of a LULC counterfactual scenario obtained from a land change model based on previous trends. Apart of LULC modifications, some other studies applied management actions in different scenarios. Xu et al. (2019), in a large-sized agroforest basin in the United States, implemented the vegetative filter strip (VFS) model incorporated in SWAT to evaluate the nitrogen and phosphorous retention ES benefits of implementing this kind of actions.
Other management practices such as contour farming, notillage, and reduction in fertilizer application have been tested against current management scenarios to elucidate the economic feasibility of such practices in terms of ES. When comparing between scenarios, some studies just consider trade-offs between ES benefits (in economic terms) (e.g., Tarigan et al., 2021), and others consider also the capital, management, and operational costs of implementing the restoration/conservation scenario in the basin. In the case of Ovando et al. (2019), who evaluated the ES benefits obtained by SWAT and other approaches, they concisely considered the costs of active forest management (considering forest types, structure, yield, and dynamics) versus a forest abandonment scenario. Finally, some other studies conducted optimization processes to find the most optimal set of land use strategies. Therefore, in these cases, the modeling of multiple SWAT model scenarios was required. One example of this is the study of Liu et al. (2022), conducted in an agricultural watershed in China, where they searched for the optimal regional cropping distribution based on the maximum economic water productivity (and other two objectives) and establishing plant area, neighbor, and irrigation constraints. Another example is (Kaim et al., 2021), who in a small-sized agroforest in Germany, found the optimum combination of agricultural production and biodiversity conservation, establishing land use transition rules and maximum land cover restrictions.

| DISCUSSION
Leaving aside the comparison with other models, this review positions SWAT as one of the most useful tools for assessing the provisioning of ES in economic terms. This makes SWAT a tool to be considered to evaluate best management practices, land use scenarios, the effects of climate change, the current flow of ES in data missing areas, among other decision-relevant applications. By complementing SWAT with economic valuation approaches, the results of these evaluations can be translated into monetary terms, allowing decision makers to more fully understand the economic implications of different management strategies (Francesconi et al., 2016). The results of this review support this claim for various reasons.
First, the SWAT model has been implemented in basins of very different sizes and land use dominance, although agricultural watersheds were the most common, as SWAT can model various agricultural-related processes (Arnold et al., 1998). Results also demonstrate that SWAT output variables are most frequently used to estimate the value of regulating ES, ahead of provisioning services. This is important because often this kind of ES are not taken into account in the assessment of management strategies as their benefits are indirect and are not always easily observable or measurable in economic terms (Barbier, 2013). It can also be highlighted that the use of SWAT output variables applied for economic valuation of ES, has been preceded, in most of the cases, by a calibration and validation process of the used output variables, as it is recommended to enhance model's accuracy (Arnold et al., 2012). However, it can be found in the literature cases in which the authors have relied on the use of output variables that have not gone through this process, especially regarding crop yields. This argues in favor of using SWAT for these situations where there is not enough valid observable data available to calibrate and validate the results, although this requires appropriate parameterization of management practices (e.g., Johnson et al., 2016;Roebeling et al., 2014). The SWAT output variables have successfully been used to apply different economic valuation methods depending on the ES valued, as market prices, avoided costs, replacement costs, mitigation or restoration cost, or even less common valuation methods such as choice experiments or hedonic pricing (Pascual et al., 2010). This demonstrates the versatility with which the results of this model can be used for the economic valuation of ES (Liu et al., 2019). Finally, this review has also collected a set of cases of economically based evaluations of different types of management scenarios that have been based on the use of SWAT. These have ranged from the simple assessment of the impact of land use changes, climate change effects, management practices, to the application of optimization processes with different objectives and constraints.

| Implementation gaps and improvement recommendations
By reviewing the different studies, it has been identified a series of implementation gaps in the application of SWAT to economically evaluate ES and for using these results to support watershed management decisionmaking.
Carbon sequestration measures seek to mitigate climate change, and so it plays a crucial role in the transition to a low-carbon economy and the achievement of global climate goals, such as limiting global warming to 1.5°C above preindustrial levels, as outlined in the 2015 Paris Agreement. However, despite the need to develop these strategies, only one of the reviewed articles dealt with carbon sequestration, and used SWAT results only to estimate the carbon exports using simulated flows and export parameters that vary according to the peatland conservation estate (Tarigan et al., 2021). A major stock of carbon that is sequestered in ecosystems is found in terrestrial ecosystems, particularly in soils. However, SWAT cannot model soil organic carbon or carbon exports. SWAT for Carbon (SWAT-C) model can estimate the carbon budget of the watershed, including carbon inputs from land use and management practices, carbon storage in various pools such as soils and vegetation, and carbon exports via surface and subsurface flows (Zhang et al., 2013). The model also includes modules that simulate processes that affect carbon dynamics, such as plant growth, residue decomposition, erosion, leaching, and denitrification. However, none of the studies reviewed used this version of the model for such purpose.
Among the different studies reviewed, there are several cases of SWAT models in mostly forested or agroforestry catchments (e.g., Ashagre et al., 2018;Gallay et al., 2021;Ovando et al., 2019). However, the estimation and valuation of forest provisioning services have been only considered in one case, by Bawa et al. (2023), where softwood and hardwood production were estimated and valued based on market prices. Others such as Ovando et al. (2019) conducted the quantification of forest provisioning services using alternative approaches and not depending on SWAT results. This could be an indication of the poor ability of the model to simulate forest management processes effectively, and hence of the quantification of such forest provisioning ES such as timber, other materials, or food (leafy vegetables, nuts, seeds, etc.).
Recently, there has been a growing concern about the impacts of hydroclimatic extremes on socioeconomic sectors, as these events, such as droughts and floods, are likely to have intensified due to climate change. From a review of SWAT studies focused on the simulation of hydroclimatic extremes, it can be concluded that this model has been applied in various occasions to produce flood and drought simulations (Tan et al., 2020). However, among the complied studies, only one of it estimated the flow regulation services economic value in terms of flood risk minimization, and none about drought mitigation (Gallay et al., 2021). Tan et al. (2020) suggest that they might be room for improving SWAT hydroclimatic extremes simulations such as through the modification of SWAT's internal algorithms. But a potential explanation for this implementation gap might be the lack of economic valuation approaches based on biophysical variables that monetize these ES (Logar & van den Bergh, 2013).

| CONCLUSIONS
The SWAT is a valuable tool for assessing the economic value of ES and evaluating the economic implications of different management strategies. Results of this review demonstrate that the SWAT model has been implemented in basins of different sizes and land use dominance, with output variables most frequently used to estimate the value of regulating ES. SWAT output variables have successfully been used to apply different economic valuation methods, demonstrating the model's versatility. However, there are gaps in the application of SWAT to economically value ES and support decision-making, such as a lack of studies focused on carbon sequestration and forest provisioning services. Additionally, while SWAT has been applied to simulate hydroclimatic extremes, there is a need for more studies focused on estimating the economic value of flow regulation services in terms of drought mitigation. Overall, SWAT is a useful tool for assessing the economic value of ES, but further research is needed to better understand its limitations and how it can be improved to support watershed management decisionmaking.

ACKNOWLEDGMENTS
This work was supported by the MERLIN project (Mainstreaming Ecological Restoration of freshwaterrelated ecosystems in a Landscape context: INnovation, upscaling and transformation), funded by the European Union's Horizon 2020 research and innovation programme under grant agreement ID 101036337.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study.

ETHICS STATEMENT
None declared.