Ecosystem service change from lost surface water for farming in the Middle Drâa Valley, southern Morocco: an economic valuation through a replacement cost approach

Regions with scarce freshwater resources, such as the Middle East and North Africa, face great challenges already today that will increase even more in the future due to climate change and population growth (Johannsen et al. 2016). Agriculture is the mainstay of the national economies of Northern African countries and the majority of the poor population works and lives in rural areas being highly dependent on water availability for sustaining their livelihoods (Africa Progress Panel 2014). Available surface freshwater resources needed to meet this rising demand are increasingly endangered by human pressures, including climate change, land degradation, over-extraction of water, water pollution, deforestation, and urbanization (López-Morales & Mesa-Jurado 2017). Therefore, food provision and water security will rely not only on the remaining surface water, but also increasingly on groundwater, as aquifers constitute the largest available storage of fresh water (Diekkrüger et al. 2012) in the area.

Morocco's Middle Drâa Valley (MDV) faces water scarcity, salinity, and environmental challenges hindering economic development (Karmaoui et al. 2014; Johannsen et al. 2016; Moumane et al. 2021, 2022; Silva-Novoa Sánchez et al. 2022; Kaczmarek et al. 2023). In particular, the MDV oases face severe water supply and management issues. Following the construction of Mansour Eddahbi dam in 1972, water was released periodically with each oasis receiving an allocated amount of water. Initially, four to eight releases were planned annually (Johannsen et al. 2016; Karmaoui et al. 2019). Today, however, the number of releases has diminished to two annually, and an average annual rainfall of 56 mm is observed in Zagora (Schulz et al. 2008; Fico 2022) and 96 mm in Ouarzazate. The drought events are more frequent, where two droughts of four years and one of three consecutive years have taken place in 20 years since the 1980s (Klose et al. 2008; Karmaoui et al. 2014; Ait El Mokhtar et al. 2019; Terrapon-Pfaff et al. 2021). The oases are heavily impacted by water scarcity, with reduced water distribution and rainfall, and frequent drought events. This has severely affected agricultural production and livelihoods, resulting in a loss of ecosystem services, most particularly the provisioning service of water availability, but also of further services like regulating and cultural ecosystem services (Karmaoui et al. 2016; Mahjoubi et al. 2022). To compensate for the lost surface-water resources (i.e. saguia water/dam releases), farmers have increasingly tried to access groundwater resources by digging and constructing wells with diesel-fired pumps for irrigation with varying strategies and outcomes. By lost surface water we refer to water lost for farmers in the sense they do not find it available in the saguia when they need it to irrigate their fields. In other words, we do not mean that water has disappeared completely in the saguia, but we refer to a problem of timing and the increased uncertainty about the availability of this water. As a response to the loss in surface water for irrigation, farmers may dig several shallow wells in different locations where they have partially functioning or non-functioning saguia (Ait Jeddi et al. 2021). They may also choose to dig one deeper well in a specific location for more efficiency and lower costs if enough water is found.

The Moroccan government is developing policies for a preservation of water-related ecosystem services to counteract the threat of water scarcity in the region. Prioritizing water allocation among competing actors poses a significant challenge (Silva-Novoa Sánchez et al. 2022). In order to make well-informed decisions, it is essential for regulators to comprehend the extent of loss in water-related ecosystem services among the traditional population of the Drâa Valley in recent decades, to be in a position to consider the values lost in the decision-making process. The present study aims at contributing to provide such information. We use a replacement cost approach (RCA) (Sundberg 2004; Wyatt 2009; Kang et al. 2015; Horváthová et al. 2021) to assess the costs of technical substitutes farmers put in place to improve or replace the surface agricultural water sources, subject to significant variability, over a period of 20 years. The timeframe was chosen following an initial exploratory interview phase, which also served to verify the method's applicability prior to conducting the survey and assess the farmers’ perceptions of water-related events. This approach acknowledges that while water may be partially lost, efforts are made to improve or replace it through the adoption of alternative measures. The costs voluntarily spent by farmers to replace this surface water partially or fully are then interpreted as the minimum amount of ecosystem services value lost in the past. A survey with randomly selected farms in the MDV was conducted to calculate these replacement costs and understand local factors causing the replacement to take place.

The next section outlines the study area description, followed by the presentation of the general background and application of the RCA method. Then, Sections 5 and 6 show our results and a discussion of them. The main concluding remarks are highlighted in Section 7.

The MDV in southern Morocco is a hot and arid region covering about 15,000 km² (Figure 1; Ouhajou 1996; Martin 2006; Klose 2009). Water for irrigation in the MDV is mainly supplied by the Mansour Eddahbi dam, with an estimated annual demand of 250 million m³ (Klose 2009). Together with evaporation losses from the reservoir of at least 50 million m³/year, the total water demand for sustainable agricultural use of the MDV sums up to 300 million m³/year (Busche 2013). The Eddahbi dam's capacity has been reduced by approximately 25% due to high erosion rates and siltation (Klose 2009; Diekkrüger et al. 2012).

About 280,000 people live in the Middle Drâa catchment (population count of 2014), distributed over the six oases (Platt 2008). Most oasis inhabitants are self-sufficient farmers who earn little income (Johannsen et al. 2016). They sell cash crops such as alfalfa, henna, and dates at local markets to buy fertilizers and fuel for groundwater pumps (Heidecke & Schmidt 2008). Household water consumption is much lower compared with agriculture (Heidecke & Schmidt 2008).

The RCA estimates the value of an ecosystem service by using actual market prices of a man-made substitute that provides a similar service (Chee 2004; Wyatt 2009; Kang et al. 2015; Horváthová et al. 2021). This approach considers both investment and maintenance costs to determine the replacement cost. The market and environmental goods can be either complements or substitutes (Sundberg 2004). The cost of replacing an ecosystem service with a man-made substitute is used to measure the economic value of the ecosystem service.

The method is based on the possibility of finding (ideally) perfect substitutes for ecosystem services. For this method to be valid, three conditions must be met (Shabman & Batie 1978; Leschine et al. 1997; Bockstael et al. 2000; Shiferaw et al. 2005). The first condition necessitates comparable substitutes in magnitude and quality, such as surface water provision. While finding perfect substitutes is rare, viable alternatives can often be utilized. The second condition involves employing cost-effective alternatives, prioritizing the least expensive man-made substitute to estimate the value of an ecosystem service. To obtain reliable cost estimates, it is practical to focus on a few available replacement techniques and study them in detail. The third condition is that urgent needs exist, and people are willing to pay for alternatives in the absence of the environmental good or service in question. Ensuring that people are genuinely willing to pay for alternatives prevents overestimation of the value of ecosystem services. For example, evaluating drinking water by looking at the costs of water provision by inter-basin transfer as an alternative, which not all the community is willing to pay for or feels the real need for, would be an overstatement of the true value of the service. However, in our case, this condition cannot be violated since we look at alternatives (i.e. for irrigation water) already put in place and for which people have incurred costs and, therefore, have demonstrated their willingness to pay.

In this analysis, we intend to estimate farmers’ losses in the MDV due to reduced surface water for irrigation over the past 20 years. To this end, we assess the costs of alternative irrigation water sources such as wells and tube wells. Our survey includes detailed information on drilling, construction, pumping with gas, fuel, electricity or solar energy, and maintenance expenses. Specifically, we focus on the costs of constructing wells as substitutes for reduced saguia water from dam releases, the traditional irrigation method of that region. The costs of deepening existing and newly constructed wells are considered as part of maintenance expenses. Additionally, we gather data on crop production, prices, and investments made by farmers to replace the lost ecosystem service of surface water provision at the oasis and farm levels. This comprehensive data is obtained through a detailed cost-based survey with farm households.

The average oasis replacement cost per hectares, metres dug, and kg of dates produced across each oasis is then calculated by dividing the AORC by the total hectares, metres dug, and kg of dates in each oasis in the period 2000–2021.

The term ‘mean’ is also used in this analysis to refer to values in the data set (e.g. mean metres dug per farm, mean annual quantities of dates produced, mean annual benefit of dates, etc.). The mean replacement cost per hectare and metre is calculated by dividing the MRC by the number of hectares and total metres dug in each farm. The total RC per farm represents the sum of all primary digging costs and all secondary costs.

During the survey, each farmer reported two mean annual benefits (in Moroccan Dirham, MAD) obtained from producing and selling dates, their main crop and source of income. One benefit was before changes in water availability, and the second after. Two mean annual quantities of dates (in kg) produced before and after were also reported. The difference between the two benefits is the ‘mean annual benefit lost’ (in MAD) and the difference in mean quantities produced is the ‘mean annual quantities of dates lost’ during the period 2000–2021. To show the relative loss perceived by farmers, the replacement cost per unit of dates can be calculated. This allows a comparison of how hard farms were hit by the reduced water availability.

In our cost-based survey, respondents reported expenses for seeking alternative sources of surface agricultural water within the saguia system which is partially or no longer supplying dam-release water. Our questionnaire had three parts: (1) personal farming experience with changing water availability over the last 20 years, (2) farming general details (size, products, quantities, income) and costs associated with current water sources using both closed-ended and open-ended questions, and (3) changes in mainly date fruit production (see Supplementary Appendix 1). We asked about historical events and carefully examined the types of costs that needed to be assessed. We also requested written proof of costs whenever possible.

First, we conducted 38 in-depth interviews in MDV oases in April 2022 to gather an initial understanding of the lost ecosystem services or the ones at risk of being lost. Interviews were recorded, transcribed, and coded using MAXQDA 2022 software (VERBI Software 2021). The timeline feature of the software was used to document drought periods and their characteristics, and the ecosystem services categories proposed in the Millennium Ecosystem Assessment (Millennium Ecosystem Assessment 2005) were adopted for coding. Findings of the exploratory phase were used to develop a cost-based questionnaire, which was tested and used in September 2022 in a main survey of 107 households across 14 randomly selected villages across Agdz oasis (according to the locals) which is part of Mezguita oasis, Ternata, Zagora which is located before Fezouata but considered to be a separate small oasis according to the locals, Fezouata, and Ketaoua oases. We use the names Agdz, Ternata, Zagora, Fezouata, and Ketaoua to refer to our sampling sites, referring to them as oases. The questionnaire assessed economic, social, and farming information.

In addition, three multiple linear regression models B1, B2, and B3 were constructed to examine the validity of hypothesis (2), and to explore what the relative costs per hectare, metre of well dug, and kg of dates produced depend on. In this, γ represents the intercept in the three models and A1, A2, A3, A4, and A5 are the coefficients of regression indicating the strength of the independent variables. To test all the relative variables, we write the models as follows:

• (B1): TRC/hectare = γ + A1.Farm size + A2.Mean metres dug + A3.Total metres dug + A4.Mean annual date benefit + ε

• (B2): MDC/metres dug = γ + A1.Farm size + A2.Total metres dug + A3.Mean annual date benefit + ε

• (B3): MDC/kg of dates = γ + A1.Farm size + A2.Mean metres dug + A3.Mean annual date production + A4.(Farm size * Mean metres dug) + A5.(Mean metres dug * Mean annual date production) + ε

Since the monetary information assessed in the survey is over the period from 2000 to 2021, all figures need to be referenced to the same year. We chose 2021 as the reference year and consequently inflated all monetary figures using the average inflation rate for consumer prices in Morocco for that period of 1.48% annually. In addition, the two mean annual benefits from producing dates from before and after the changes in water availability were inflated using the average inflation rates of both periods (i.e. for 2000–2021, the average inflation rate was 1.48%, and for 2012–2021, the average inflation rate was 1.18%) (see Supplementary Table S1).

In this section, we first present the basic characteristics of the surveyed farmers to provide a context for understanding the local agricultural practices and farming systems in the area. Next, we present an overview of people's perceptions of drought events and ecosystem services since the 1980s. Finally, using our cost data, we provide an estimation of the replacement costs of lost surface water first across each oasis, and secondly, at the level of each farm, where we try to compare the investments between oases, and within each of the oases’ farms. In both parts, we use multiple regression modeling to test our hypotheses (see Section 1).

From 107 farmers interviewed, we found that 78% use replacement sources (wells and tube wells), partially in combination with available dam releases, while the remaining 22% fully use wells. Funding for well digging comes from farming income (for 44% of farms), remittances only (19%), or both (33%). Among the farms, we identified a number of 129 wells dug in 212 events and totaling 3,879.5 m (average well depth is 30.07 m). The farms grow various crops, with over half of them focusing on dates, 11% mixing dates and fruits, and some mixing dates with wheat, vegetables, and alfalfa.

Interviewees over the age of 45 from Mezguita, Ternata, Fezouata, and Ketaoua expressed that the Drâa Valley was thriving and healthy during the previous period of independence. The ecosystem provided many essential services, such as drinking water, river flow, and scenic beauty, as well as many other cultural services. However, severe droughts from 1982 to 1987 drastically reduced surface water availability, affecting the oasis ecosystem and livelihoods. The interviews noted that aquifers remained charged during the 1982 droughts likely due to the relatively low use of aquifers and water pumps before 1982 and sufficient dam releases. However, some regions still experienced heavy losses in surface vegetation. Respondents revealed that many palm trees were preserved due to the wet and moisturized aquifers. Nevertheless, participants claimed that several regions in the Drâa basin experienced heavy losses in surface vegetation, including almond, olive, and palm trees.

According to the interviewees, between 1982 and 1987, wells in oases increased, mostly manually dug and less than 8 m deep, mainly in Fezouata and Ketaoua. Since 1990, fuel and gas water pumps have been used. A decade after the droughts, aquifers had increased in salinity, especially in Fezouata and Ketaoua. According to the interviewees, there was an extensive use of groundwater, increasing from 2000, and leading to severe impacts from the 2012 droughts on the oases and their ecosystem services. The droughts lasted two years and were followed by floods in 2014. Current droughts have been ongoing since 2015, according to some interviewees, and since 2016 according to others, while only a few mentioned 2017. Interviewees reported the loss of farming-related ecosystem services from the 1982 droughts to the current droughts, with only a few of these services being replaced. Insufficient water for farming is a major concern for most of the interviewees, and while most were able to replace a portion of the lost surface water, there has been an irreplaceable loss of cultural services.

Table 2 exhibits the results of the multiple linear regression model (A) (see Section 3.4). The positive and significant regression coefficients for farm size suggest that an increase in farm size is associated with an increase in TRC. None of the oases coefficients are statistically significant, indicating that the specific four oases do not have a significant standalone effect on the TRC compared with Agdz as a reference. However, while looking at the interaction terms between farm size and the specific oases, it is clear that the effect of farm size on the TRC varies across different oases (not for Ketaoua) as compared with Agdz. The positive effect of farm size on TRC is most pronounced in Agdz, and this effect decreases in the other oases, significantly so in Ternata and Zagora. In Zagora, the farm size effect may turn slightly negative, indicating that larger farms there tend to invest a bit less into drilling wells, which could be due to different local conditions or strategies.

In Figure 7(c), Zagora shows the highest range of MDC per kg of dates produced at the farm level during 2000–2021. Figure 7 reveals that large farms in Fezouata, Ternata, and Agdz invested relatively little per kg of dates produced to replace surface water. Conversely, small farms in Zagora faced greater replacement costs, with the mean loss per kg of dates being the highest at 1,498 MAD. An exception is Ketaoua, where smaller farms also had relatively low investments, with a maximum of 221 MAD per kg of dates produced.

Overall, it appears that large farms, with high number of hectares, of metres dug, or kg of dates produced, invest relatively less than small farms to replace irrigation water sources.

Table 3 presents the results of the regression model B1 (see Section 3.4), which identifies possible factors influencing the total replacement cost per hectare. The negative and significant regression coefficient for farm size indicates that the investment per hectare decreases significantly as farm size increases. The model also shows that the investment per hectare decreases significantly with each additional unit of mean metres dug within the farm. The coefficient for the annual mean benefit of dates on replacement costs is not a statistically significant predictor in this model. Overall, model B1 explains only 20% of the variability in the total replacement cost per hectare, which is relatively low.

Results from regression model B2 (Table 4) indicate that the MDC per metres dug is expected to increase with each unit increase in farm size. However, the significant negative regression coefficient for mean metres dug suggests that the investment per metre decreases with each unit increase in mean metres dug. While the mean annual date benefit might explain the increase in investment per metres dug, this coefficient is not statistically significant. The overall model is significant, indicating that at least one predictor is useful in explaining the variance in MDCs per metres dug on a farm.

Regression model B3 (Table 5) shows the positive significant effects of farm size, mean metres of well dug, and mean annual date production on the MDC per kg of dates produced by a farm. This cost increases significantly for each unit increase in these three independent variables. Unlike the previous two models (B1 and B2), model B3 does not display a negative relationship between the MDC per kg and the mean annual date production of farms. The interaction terms between farm size and mean metres dug, as well as between mean metres dug and mean annual date production, are not statistically significant. This indicates that there is no combined effect of these variables on the MDC per kg of dates produced beyond their individual contributions. Model B3 is significant and explains 11% of the variability in the MDC per kg of dates, which is relatively low.

Finally, in model C (Table 6), the analysis shows that there is a notable impact of the mean annual date production and benefit on the TRC. The negative coefficient associated with the mean annual production implies that as the farm produces more dates annually, the total replacement cost decreases. This suggests that higher production of dates tends to lower the overall replacement cost. However, the positive coefficient linked to the benefit from dates indicates that the total replacement costs generally rise as the benefits from dates increase. This suggests that while higher benefits are associated with increased costs, the reduction in replacement costs due to higher date production counterbalances this effect to some extent.

We surveyed 107 farms to assess their improvements after implementing technical substitutes for their agricultural water supply. The majority, 97%, reported recovering only half of their previous water supply, leading to producing two to three times fewer crops like dates, wheat, barley, and alfalfa. In contrast, only 3% of the farms fully recovered their water and crop levels. The 97% included small- and medium-scale farmers who mainly dug and deepened wells, with 23% also building basins. The 3% that achieved full recovery primarily used a combination of tube wells, water basins, and solar energy pumps for irrigation.

Our original hypothesis (1) suggested that changes in water availability would lead to varying losses across the oases of the MDV, increasing from north to south along the aridity gradient. Consequently, investments to offset these losses would follow a similar pattern. Our results confirmed the first part of the hypothesis but showed that losses did not consistently increase or decrease with the aridity gradient of the MDV (see Figures 3 and 4, Section 3.3). Instead, they fluctuated. Our multiple regression model A (Section 4.3.2) indicates that losses along the valley fluctuate due to farm size and other unknown factors. Larger farms tend to invest more to replace lost surface water, resulting in greater average losses compared with smaller farms. This may be due to their higher investment capacity, access to new technologies (e.g. irrigation, water pumping), and subsidies. Prior research on climate-change adaptation in agriculture supports this (Al-Tawaha et al. 2022). Smaller farmers might adapt more easily by diversifying income sources, whereas larger farmers rely more heavily on farming. Our regression analysis revealed that the effect of farm size on total replacement investment varies across the different oases (Fezouata, Ternata, and Zagora). While farm size generally has a positive impact on the replacement cost of farms, this effect diminishes in these oases. This reduction may be related to their ability to invest from alternative income sources or significant remittances. Additionally, the hydrological and geospatial attributes of these oases, as discussed by Cherkaoui et al. (2023), could explain the differing effects of farm size on expenditures. In Zagora, the results even suggested that larger farms might invest less in digging activities, possibly due to local conditions or economic factors such as lower returns from farming. Variations in farmers’ perceptions of irrigation water scarcity across the MDV may explain the fluctuating losses, as prior behavioral economics research suggests (Shogren & Taylor 2008; Robinson & Hammitt 2011). Other studies highlight additional factors such as proximity to water reservoirs and a diversity of income-generating activities that influence farmers’ investments in irrigation water during scarcity (Acquah & Onumah 2011; Uddin et al. 2014; Chen et al. 2018; Fan et al. 2019). However, this analysis only briefly covers these factors. Losses from changes in water availability varied across the MDV between 2000 and 2021 and did not consistently increase in regions with higher aridity. This finding does not imply that drought did not contribute to these losses, but rather indicates that other factors (e.g. farm size, oasis location) also influence the impact of water scarcity, even in less severely affected areas. Specific aspects of the Fezouata oasis significantly determine the average expenses farmers incur to secure alternative water sources. This highlights the importance of these characteristics in agricultural practices and water management strategies, warranting further research.

The cost of replacing lost surface water is similar across different oases (Section 4.4). As a first interpretation, we assume that farms with higher investments on average may experience more losses than those with lower investments. However, we hypothesized (2) that for large farms with more production, hectares, or deep wells, the cost of replacement is relatively cheaper, resulting in a lower potential loss value per unit of farm area compared with small farms. This relative cost advantage means that larger farms may have an advantage in terms of economies of scale, as they can spread their fixed costs over a larger output or productive area, resulting in lower replacement costs per unit of production or area. This was mainly displayed by the regression models B1 and B2 for the TRC per hectare and the MDC per metre of the well dug. Shan et al. (2015) as well as Lapar et al. (2012) similarly stated that larger farms have lower production costs per unit of output and are more technically efficient than smaller farms (see also Alston et al. 1998; Barkley et al. 1999). Mafoua (2002) stated that two-crop farms, as well as three-crop farms, exhibit overall economies of scale that increase with the farm size, as they can lower the cost of producing crops in the same farm by spreading fixed and variable costs over their large output, compared with small farms. Therefore, the ability to invest in deeper wells in a single event could explain the difference in replacement costs between larger and smaller farms, leading to differences in water loss. Accordingly, a single efficient digging event may offer greater benefits than multiple events. The research of Ho & Shimada (2019) in India suggests that larger farms are more efficient in their groundwater use due to their ability to invest in efficient irrigation technologies, resulting in lower losses and greater cost-effectiveness compared with smaller farms. While investing more in replacing surface water may result in overall higher losses, large-scale farms enjoy economies of scale, allowing them to minimize losses per unit of surface and metres dug.

Farmers’ investment strategies in the MDV to access more irrigation water differ according to a few factors. Our findings show that the overall investment of farmers increases with increasing farm hectares and mean metres dug. One explanation is that larger farms can invest more and require more water due to larger cropping areas. Higher annual benefits from growing dates may explain the additional investment in irrigation water. This was displayed by the regression model (C) (see Section 4.5). On the one hand, the negative coefficient for mean annual production indicates that as the farm produces more dates, the total replacement cost decreases. This could be because higher production might lead to economies of scale or more efficient use of resources, reducing overall costs. The findings of a study by Kiprop et al. (2017) with farmers in Kenya concluded that crop income from irrigation significantly influences farmers’ decisions to pay for irrigation water. On the other hand, the positive coefficient for benefits from dates suggests that as the benefits from date production increase, the total replacement costs also tend to increase. This could be due to additional investments or expenses required to maximize the benefits, such as marketing or processing costs. So, while higher production of dates may lead to cost savings, the increase in benefits might also entail additional costs, resulting in a complex relationship between date production, benefits, and replacement costs.

The survey identified two potential factors that could affect farmers’ investment in irrigation water for further investigation. First, farmers who rely exclusively on groundwater for irrigation may be more willing to pay for water, possibly due to its perceived reliability as a consistent water source, compared with those who use both groundwater and dam releases. This is consistent with the work of Biswas & Venkatachalam (2015), who concluded that farmers are willing to spend more on irrigation water if they can predict water availability (see also Bouman et al. 2008). Second, household size may also impact farmers’ willingness to invest in irrigation water. Larger families may be less willing to invest in irrigation water due to high engagement in non-agricultural activities. Alternatively, larger families may rely more heavily on remittances from family members working outside the area. This is supported by the findings of Tang et al. (2013) in a similar study with farmers within the Chinese Loess Plateau, on agricultural practices and sustainable rural livelihoods factors (see also Mustapha 2012; Arshad et al. 2016; Gebretsadik & Romstad 2020). The individual discount rate may also influence the investment decisions of farmers, as when it is high, it could make future costs and benefits associated with water replacement appear less valuable compared with immediate costs, thus increasing the willingness to invest to replace this water. Conversely, a lower discount rate may result in a longer asset lifecycle and slower technological progress. While the impact of the individual discount rate on farmers’ investment is important, it was not the primary focus of the present paper. In summary, our findings highlight the complex interplay between farm size, perceived benefits of crops, and irrigation infrastructure investment in agricultural production. Further research could explore these factors more deeply, as well as examine potential policy implications for promoting sustainable and efficient water use in agriculture.

We used the RCA to estimate losses in the MDV oases due to water availability changes and gain insights into farmers’ behavior in adopting technical alternatives. The RCA method was effective for partially achieving this goal. Our working assumption was that money spent on obtaining additional water was indicative of losses suffered from reduced irrigation water. The ARC and MRC provided reliable estimates of average and farm-level loss values. However, we believe our estimates for losses experienced by farmers undervalued the ecosystem service of surface water for irrigation, indicating one of the flows of the RCA. This was because our analysis only covered detailed costs related to date production, benefits, and crop production reduction and other crops such as alfalfa, wheat, and livestock were not quantified in the same way. This simply means that the real value of water is larger than the replacement cost we have estimated. Most farmers were not able to provide those details. Additionally, most of these products are mainly exchanged in nature and used for subsistence. To make a proper and accurate estimation of the ecosystem service provided by surface water and a comprehensive understanding of farmers’ losses, it is crucial to quantify the non-recovered crops. The RCA is considered suitable if the service in question can actually be replaced, but the farms in this study failed to fully recover their original water and crop levels after investment, with only 3% succeeding in doing so.

While other methods, such as the defensive expenditure method, can also estimate loss value, we deemed the RCA method more suitable for our analysis (Sundberg 2004). It is worth noting that although the RCA method has been used in other contexts, it had not been utilized to estimate loss values due to lost ecosystem service elements. Therefore, there was no previous reference with which to compare the application of the RCA method in this particular context. This lack of similar applications could have potentially affected the accuracy and reliability of our results. Nonetheless, the RCA method was still a suitable choice because it allowed for the partial estimation of the value of the environmental good (i.e. irrigation surface water) using the costs of man-made substitutes, despite the non-quantified aspects that were missing. While our sample is representative of the population in the oases studied, the findings may still be difficult to generalize to other similar regions. Further research should consider broader geographic samples for this purpose. Overall, the RCA method provided us with valuable insights into how farmers adapt to changing environmental conditions and the costs associated with these adaptations. This highlights the usefulness of the RCA method in such contexts.

The application of RCA in the MDV case brought about several advantages, but also some challenges. One strength was the availability of reliable data. However, estimating costs from 20 years ago proved to be a challenge, particularly when dealing with farmers who had no written records of their expenses. To deal with this challenge, we adopted various approaches (see Section 3.3) when approaching the farmers. The meticulousness in bringing past information and numbers into sharp focus played a vital role in producing highly reliable results. Thus, by implementing these methods, we were able to gather reliable data on the costs incurred by farmers over the past two decades. Although the process had its challenges, we overcame them by utilizing effective techniques, ultimately resulting in valuable insights into the costs associated with adapting to changing environmental conditions.

Our analysis demonstrates the potential and shortcomings of using the RCA method to value the losses of ecosystem services experienced by farmers in the Drâa Valley, southern Morocco, due to lost surface-water availability in the past decades. We estimated the costs farmers incurred to invest in technical installations (wells) to replace the decreasing irrigation water availability in the Drâa River. Those replacement costs reflect the value of irrigation water in the area. We identified how farmers reacted to surface water reduction and showed the differences among and within oases through cost assessments. However, the lack of essential quantifiable data, mainly for non-recovered crops and cattle, limited the accuracy of our estimations. This information can be a valuable tool for understanding and forecasting farmers’ behavior and reasoning under water scarcity assuming their varying utility and helps in understanding how they may be willing to adapt to changes in water availability as future water deficits increase. Contrasting the losses experienced by farmers across different oases in a water scarcity context can help identify the areas suffering the most. In addition, the analysis showcases that large farms may have an advantage in terms of economies of scale, as they can spread their costs over a larger output or productive area, resulting in lower replacement costs per unit of production or area. Understanding the implications of an economy of scale on water management and production costs can be crucial for policymakers, farmers, and researchers. This can also contribute to developing water allocation incentives and programs to support effective water resource management, particularly for small farmers, as those are often neglected in research on water scarcity and agricultural development (Perret & Stevens 2006). Our results highlight the factors playing an important role in shaping the impacts of water scarcity on farmers. This can help inform policymakers and stakeholders focusing on promoting sustainable and efficient water use in agriculture, considering the complex interplay between various factors.

This paper contributes to the literature on ecosystem services valuation by providing a practical example of how the RCA method can be used to estimate the value of lost ecosystem services, such as irrigation water supply. By using this method, we were able to gain valuable insights into the socioeconomic impacts of water scarcity on rural communities in semi-arid countries. However, further studies and applications of economic valuation methods with similar objectives are necessary to compare and improve the approach and obtain better outputs. Our research can inform future studies on ecosystem services valuation and its implications for sustainable development. Overall, the present analysis provides insights into the socioeconomic impacts of water scarcity on rural communities in semi-arid countries and can further contribute to a better understanding of the challenges they face. It also provides insights into the potential strategies and policies that can be developed to promote sustainable water resource management and enhance agricultural productivity in water-scarce regions. The analysis underscores the need for more applications of economic valuation methods that can contribute in this sense.

We thank the local farmers and governmental actors for their contribution to the data assessment used in the study. Further, we thank our colleagues for critical and helpful contribution. We also thank the German Ministry of Education and Research (BMBF) for financial support (SALIDRAAjuj-01-FKZ UU1906).

I.M. Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Resources; Software; Supervision, Visualization, Writing – original draft, Writing – review and editing. O.F. Conceptualization; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Supervision; Validation; Visualization; Writing – review and editing.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.