Impacts of land use and climate changes on hydrological responses in a Peruvian Andean watershed Open Access

Rapid economic development, population growth, ecological degradation, and climate change have positioned water resources as a global priority for people around the world, especially when allocating or distributing water for different uses (Aghsaei et al. 2020; Zhang et al. 2021; Afonso de Oliveira Serrão et al. 2022). This problem is greater in the Andean regions because water resources depend on Andean ecosystems, which are mostly vulnerable to changing conditions (Ochoa-Tocachi et al. 2016; Bonnesoeur et al. 2019; Monge-Salazar et al. 2022). These ecosystems play a critical role in generating ecosystem services (ES) (Corvalán et al. 2005; Silvestri et al. 2013; Wezel et al. 2014; Díaz et al. 2015), which are significantly impacted in areas such as water regulation, soil erosion, water and air pollution, and biodiversity loss (Alam 2018; Kay et al. 2019; Hasan et al. 2020; Su et al. 2020).

Worldwide, more than 50 million people depend on the ES provided by the Andean regions (Doornbos 2015; Bonnesoeur et al. 2019). Such is the case of the Bolivian and Peruvian Andes where afforestation can play a role in improving water supply and regulation or, on the contrary, worsen the problems depending on how and where the forestation is planted (Bonnesoeur et al. 2019). Hence, understanding the hydrological processes linked to ES is essential, even more so when considering changing conditions (Chen et al. 2019; Hasan et al. 2020). Land use and climate change have become key factors for the construction of change scenarios, which are fundamental for assessing impacts on hydrological processes (Morán-Tejeda et al. 2015; Tamm et al. 2018; Chen et al. 2019; Hasan et al. 2020; Zhang et al. 2020; Afonso de Oliveira Serrão et al. 2022; Ougahi et al. 2022; Son et al. 2022; Tan et al. 2022).

Counteracting the effects of climate change and land degradation has become a difficult task politically, economically, and socially. In recent years, Payment for Ecosystem Services (PES) programmes have been implemented in several countries around the world (Immerzeel et al. 2008; Scullion et al. 2011; Bhatta et al. 2014; Grima et al. 2016; Lopes Simedo et al. 2020; Oliveira Fiorini et al. 2020; Perevochtchikova et al. 2021), as an instrument of popular public policy (Derissen & Latacz-Lohmann 2013; Perevochtchikova et al. 2021). In Peru, PES obligatorily implemented in the Service Providers Companies (SPC) through of Remuneration for Ecosystem Services (MERESE) programme (D. S. No 009-2016-MINAM 2016), which consists in the conservation and recovery of ES through agreements between contributors (e.g., rural communities) and retributors (e.g., the SPC). In South America, MERESE is a successful experience because: (i) it is a sustainable mechanism that ensures its conservation funds through water service tariffs, and (ii) it has generated a good social impact by strengthening the relationship between the SPC and the communities located in the Andean regions (Dextre et al. 2022).

Future land use change is one of the major anthropogenic impacts on the land surface, with direct effects on hydrological processes and ES essential for human well-being (Steffen et al. 2007; Hasan et al. 2020). Specifically, land use changes affect the components of the water balance, including flow, surface runoff, groundwater, and evapotranspiration (ET) (Hasan et al. 2020; Teklay et al. 2021; Afonso de Oliveira Serrão et al. 2022). Ochoa-Tocachi et al. (2016) found that the impacts of land use change are diverse and their magnitudes are a function of watershed properties, vegetation, and management type. Anthropogenic interventions are primarily responsible for the greatest flow variability and the decrease in the water regulation capacity and yield of watersheds, regardless of the hydrological properties of the original biome. On the other hand, climate change may affect regional precipitation, temperature, and ET conditions, leading to significant changes in runoff (Zhang et al. 2020). These effects are recognizable worldwide, leading to increases in the frequency and intensity of extreme flood and drought events (Pham et al. 2019; Vaghefi et al. 2019; Zarrineh et al. 2020; Martínez-Retureta et al. 2021). Andres et al. (2014) evaluated the impact of climate change in the Peruvian Andean basin of the Vilcanota River, where they found that an increase in total runoff is expected in the rainy season, but less temporary storage (e.g., snow and soil moisture), which produces less water supply, especially in the dry season.

Freshwater ecosystems (e.g., rivers, lakes, and groundwater) support the provision of potential ES for different water uses, e.g., water supply and regulation, and sediment retention (Pham et al. 2019). Pham et al. (2019) evaluated the impact of coupled scenarios of land use change and climate change on potential ES in the Taro River basin in Italy and found that increased ET demand, combined with changes in precipitation patterns, resulted in a 20% reduction in water yield. On the other hand, Tamm et al. (2018) found that there is a strong linear correlation between forest cover change with streamflow, finding that a 5% reduction in forest cover represents a 1% increase in annual runoff, but in monthly variation the changes are insignificant. Chen et al. (2019) found that in the Jinsha River basin, land use change is expected to have little impact on runoff and its associated extreme events, and climate change is expected to produce a small increase. In general, low water production and overpopulation are the main causes of extreme water scarcity, which makes it necessary to strengthen conservation projects and increase water storage mainly in the rainy season (Zhang et al. 2021). This situation makes the understanding of present and future climate conditions relevant in the determination of vulnerability and the development of adaptation strategies to changing scenarios (Anand et al. 2018; Martínez-Retureta et al. 2021). In this context, the present research aims to answer the research questions:

• What is the impact of land use and climate changes on hydrological responses in a Peruvian Andean watershed?

• What is the relevance of the MERESE programme in the context of changing scenarios?

To answer these questions, we used the Soil and Water Assessment Tool (SWAT) hydrological model (Arnold et al. 2012), and climate data from the Rontoccocha Ecohydrological Monitoring System (REMS). The results of this study could help in the integrated and sustainable management of water resources in the study basin (Tamm et al. 2018; Martínez-Retureta et al. 2021).

The CCAY watershed is located entirely within the territory of the Atumpata Community, and has an area of 2.6 km² with an elevation range of 4,076–4,531 m.a.s.l. and an average slope of 37.6%. On the other hand, the RON watershed is located between the Communities of Atumpata and Micaela Bastidas, and for a better evaluation, it was subdivided into three sub-basins called RON1, RON2, and RON3 (see Figure 2), where (i) RON1 has an area of 3.3 km², an elevation ranges of 4,263–4,650 m.a.s.l. and an average slope of 32.7%, (ii) RON2 has an area of 2.3 km², an elevation ranges of 4,255–4,736 m.a.s.l. and an average slope of 34.4%, and (iii) RON3 has an area of 2.5 km², an elevation ranges of 4,255–4,736 m.a.s.l. and an average slope of 30%.

Data for hydrologic modelling in this study area were obtained from various sources (see Table 1). Topography was defined with the Alos Palsar digital elevation model (DEM) with a spatial resolution of 12.5 m. Land use was determined by supervised classification with the Landsat 08 OLI/TIRS satellite image, resulting in the land uses of 91.6% pasture (PAST), 1.9% forest (FRST), 4% wetlands (WETL), and 2.5% water (WATR). Soil type was determined based on the Digital Soil Map of the World (DSMW) – FAO – SOIL with a spatial resolution of 1,000 m. The predominant soil type identified was lithosol, with a clayey–sandy loam texture. Daily precipitation and temperature data were obtained from REMS and extended using the operational PISCO gridded product (Peruvian Interpolated data of SENAMHI's Climatological and Hydrological Observations; Aybar et al. 2020; Huerta 2022a, b; Millán-Arancibia & Lavado-Casimiro 2023) using the bias correction method of QDM (Cannon et al. 2015). Daily flow records were obtained from REMS for the CCAY basin and for the period 2016–2019. Finally, the GCMs are taken from NEX-GDDP-CMIP6 (Thrasher et al. 2012), which consists of 31 models for precipitation and 28 models for maximum and minimum temperature, for the SSP2-4.5 and SSP5-8.5 scenarios.

For this study, the SWAT model was run with a daily time step and was configured with a total of 41 sub-basins and 345 HRUs. The Hargreaves method was chosen to calculate reference ET, the Soil Conservation Service curve number was used to calculate surface runoff, and the variable storage method was used for flow routing.

GCMs represent one of the main sources of uncertainties when conducting climate change studies (Chegwidden et al. 2019). The use of multiple GCMs and scenarios allows us to assess and reduce the uncertainties associated with GCMs (Ficklin et al. 2013; Dhakal et al. 2018). In this study, we use 31 GCM models for precipitation and 28 GCM models for maximum and minimum temperature for the shared socioeconomic pathway (Riahi et al. 2017) scenarios of SSP2-4.5 and SSP5-8.5. We then apply the median, 5% percentile, and 95% percentile statistics in order to couple and represent future projections (Ficklin et al. 2013; Goyburo et al. 2023). As a result, we obtain six sets of time series of precipitation, maximum, and minimum temperature variables.

Finally, six climate change scenarios were constructed by applying the change factor or delta change methodology (Keller et al. 2022), based on the comparison between the historical period (1994–2014) and future period (2025–2055) of the six sets of time series described above. These differences were applied to the observed series to obtain the adjusted projections. The change factor was determined monthly, using multiplicative factors for precipitation and additive factors for temperatures (see Table 2).

A total of 21 change scenarios were configured (see Table 3). The SLC11 scenario considers only current climate and land use. Scenarios SLC21 and SLC31 combine the current climate with projected MERESE and pessimistic land use, respectively. Scenarios SLC12, SLC13, SLC14, SLC15, SLC16, and SLC17 integrate current land use together with the six climate change scenarios. Finally, scenarios SLC22, SLC23, SLC24, SLC25, SLC26, SLC27, SLC32, SLC33, SLC34, SLC35, SLC36, and SLC37 combine the two land use scenarios with the six climate change scenarios.

Table 3

Configuration of combined scenarios

Note: Each of the change scenarios is identified with respect to colour with the following configuration: the current land use with current climate scenario (black colour); projected MERESE land use with current climate scenario (green colour); pessimistic land use with current climate scenario (red colour); current land use with climate change scenarios (blue colour range); projected MERESE land use with climate change scenarios (turquoise colour range); and pessimistic land use with climate change scenarios (purple colour range).

View Large

Sensitivity analysis was performed with a total of 15 parameters obtained from the literature (see Table 4). A total of 400 iterations were executed, resulting in a total of seven sensitive parameters through the evaluation of the p_value, these parameters are classified with respect to the processes they influence: two parameters in the runoff process (CN2 and SOL_K), two parameters in the groundwater process (ALPHA_BF and RCHRG_DP), two parameters in the routing process (CH_K2 and CH_N2), and one parameter in the evaporation process (ESCO). Among all these parameters, the most sensitive parameter was the curve number (CN2), which directly influences surface runoff.

The results show that, for the SLC21 scenario (representing a 19.8% increase in forested area), the changes in monthly mean streamflow are insignificant compared with the SLC11 scenario (see Figure 5, upper part, dashed green line), which is evidence that afforestation would have a small impact on flow and some hydrological processes (see Figure 7). However, the spatial distribution of changes in flows and states (see Figure 6) reveals areas with a slight increase in ET and GWQ, a slight decrease in SW and WYLD, and insignificant changes in SURQ. On the other hand, the results show that, for the SLC31 scenario (i.e., pasture, forest, and wetland land uses would be changed to bare soil), the changes in monthly mean streamflow are significant compared to the SLC11 scenario (see Figure 5, upper part, solid red line). First, an increase of up to 9% (average of the four sub-basins: 10% (CCAY), 11% (RON1), 8% (RON2) and 8% (RON3)) is observed during the months of September to February, which corresponds to the spring and summer seasons (wet period in Peru). Then a decrease of up to −9% (average of the four sub-basins: −9% (CCAY), −11% (RON1), −9% (RON2) and −9% (RON3)) is observed during the months of March–August, which corresponds to the fall and winter seasons (dry period in Peru). However, the fluxes and states (see Figure 6) reveal an increase in SURQ and WYLD values, and a decrease in ET, GWQ, and SW.

Six climate change scenarios were configured and executed using the calibrated SWAT model. The results were evaluated and compared to the baseline scenario SLC11. In general, all scenarios project an increase in monthly mean flows (see Figure 5, middle part), with the largest increases between the months of October and June, and the smallest between July and September. It is also observed that the largest increases are recorded in December, with approximately 35%, especially in the SLC15 and SLC12 scenarios. These results also evidence a high dispersion in the change values obtained for each scenario and each month, with the dispersion most pronounced between October and March, coinciding with the months with the highest average monthly flow.

Twelve combined climate and land use change scenarios were configured and executed using the calibrated SWAT model. The results were evaluated and compared to the baseline scenario SLC11. In general, increases in monthly mean streamflow are observed; however, between August and September, the changes are insignificant, while between October and July, the increases are more noticeable, with the exception of the months of April, May, June, and July, where reductions in streamflow are recorded for the scenarios incorporating pessimistic land use change (SL3), i.e., SLC32, SLC33, SLC34, SLC35, SLC36, and SLC37. Similar to the results of the climate change scenarios, December presents the largest monthly mean flow increases, reaching up to approximately 53%, mainly in the SLC15 and SLC12 scenarios. In all climate change scenarios, there is a wide dispersion of change values for each month and in each scenario, this dispersion being more accentuated between October and March, coinciding with the months with the highest monthly mean streamflow.

In relation to flows and states (see Figure 7), we found that for the climate change scenarios combined with afforestation land use, there will be a significant change trend, mainly in the increase of GWQ (10–20%) and in the decrease of SURQ (between 10 and 60%). In contrast, with pessimistic land use, there will be a significant trend in decreasing GWQ (−30 to −25%) and increasing SURQ (130–170%).

Figures 8 and 9, show similar behaviours in the four sub-basins analysed, where all scenarios present the same change direction, mainly characterised by an in-runoff. In general, land use change scenarios generate insignificant variations in the annual water balance compared to the current scenario. Specifically, the projected MERESE scenario (SLC21) shows almost no change, while the pessimistic scenario (SLC31) shows a slight increase in runoff and a slight decrease in ET (up to 1% increase in mean annual runoff for the pessimistic scenario, see Figure 8). On the other hand, the climate change scenarios present a notable increase in runoff, accompanied by a slight increase in ET, mainly attributable to the increase in precipitation in the study area (up to 26% increase in mean annual runoff, see Table 2 and Figures 8 and 9). These results are similar to those found by Clerici et al. (2019) in the Colombian Andes.

Land-use change is one of the most relevant factors affecting the hydrological system, land surface structure, and material and energy flows (Wu et al. 2015; Hasan et al. 2020). These effects depend on the characteristics of each watershed, the original and replacement vegetation, as well as the type of management applied (Ochoa-Tocachi et al. 2016). Our results show that the projected MERESE scenario (afforestation) generates insignificant variations in flow rates (see Figure 5, upper part), which is consistent with the findings reported in the literature review by (Lalonde et al. 2024), where they report that of 13 studies on the impact of afforestation, 10 show a decrease in runoff and 3 conclude that there is no effect. Similarly, Bonnesoeur et al. (2019) note that there is consensus that watersheds with plantations of exotic species, and to a lesser extent those with natural forests, show reductions in water yield of between 20 and 45% compared to watersheds with non-forest uses, which is reflected in a decrease in the runoff coefficient, as also reported by Ochoa-Tocachi et al. (2016). These effects are probably attributed to the fact that forests transpire more than grasslands and that infiltration rates improve significantly when degraded soils are replaced by forest cover (Wongchuig et al. 2023; Lalonde et al. 2024). In this study, the replacement of grasslands by forests could explain the slight increases observed in ET and GWQ (see Figures 6 and 7), considering that grasslands, in addition to having a good water regulation capacity, play a key role in groundwater recharge (Mosquera et al. 2022).

On the other hand, several studies have shown that soil degradation (e.g., reduction of forested and grassland areas) leads to a significant increase in surface runoff, as well as a decrease in ET and base flow or groundwater (Awotwi et al. 2019; Bonnesoeur et al. 2019; Paiva et al. 2023). Likewise, wetland loss generates negative effects on flow regulation in perennial streams of the humid puna (Wunderlich et al. 2023). Consistent with these findings, the results obtained in this study show that the pessimistic land use scenario presents significant changes, reflected in a decrease in ET, base flow (GWQ), and soil moisture (SW), as well as an increase in surface runoff (SURQ) and water yield (WYLD). These results, coinciding with evidence reported by other authors, strengthen the validity and robustness of the findings presented.

Climate change impacts on water resources in the Peruvian Andes are little explored. In the Atlas of Water Production in Peru (Huerta & Lavado-Casimiro 2021), they evaluated the influence of climate change on current water production capacity, focusing on the SPC tributary watersheds in Peru. They find that, for the Lake Rontoccocha tributary watershed, runoff projections were inconsistent, as of the three GCMs evaluated, one estimated a decrease (−16%, ACCESS 1.0) and the other two projected increases (+20% and +3%, HadGEM2-ES and MPI-ESM-LR, respectively). This high variability can be attributed to the limited use of GCMs (only three) and the lack of a methodology to select models that best represent the historical behaviour of the basin. We used 31 GCMs for precipitation and 28 GCMs for maximum and minimum temperatures, also applying the delta change methodology (Keller et al. 2022) using the median and the 5 and 95% percentiles for the SSP2-4.5 and SSP5-8.5 scenarios, which allowed a better representation of the behaviour of the set of models and more robust results. Recent studies in high Andean basins near the study area (Lavado Casimiro et al. 2011; Andres et al. 2014) indicate that climate change increases runoff during the rainy season (January–March), a finding that coincides with the results obtained in this research (see Figure 5, middle part). In our case, an increase in precipitation (except in the dry season) and temperature is projected, which would cause increases in ET, GWQ, SURQ, and WYLD, and a decrease in SW. Although some studies indicate that the most significant changes will occur towards the end of the century (2070–2100), the average period analysed (2025–2055) is considered adequate and sufficient for the purposes of this study.

Finally, the combined change scenarios were analysed, showing an interaction between land use changes with climate change. In the case of monthly mean flows (see Figure 5, lower part), when combining the climate change scenarios with the SLC21 scenario (associated with afforestation), the observed effects are mainly dominated by climate change, showing that the influence of land use is limited. In contrast, when combining the climate change scenarios with the SLC31 scenario (associated with degradation), a joint influence of both factors is observed, resulting in more pronounced changes. This same pattern is identified in the hydrological flows and states (see Figure 7), where the final effects depend on the interaction between land use change and climate change.

In Peru, a total of 52 MERESE initiatives implemented by SPCs were identified (until 2020), which shows a clear upward trend (Tristán et al. 2022). However, the lack of scientific evidence to understand the functioning of water ES in MERESE implementation areas represents a latent challenge that has not yet been fully addressed. This limitation becomes even more relevant when formulating projects aimed at the conservation, restoration, and sustainable use of ecosystems (Tristán et al. 2022). In this study, we address this challenge by using scenarios of land use change, climate change, and combined change in a MERESE intervention area located in a Peruvian high Andean zone. With respect to the land use change scenarios, we found that the afforestation scenario compared to the current scenario will produce minor changes in hydrological processes, while the pessimistic scenario will produce more significant but negative changes, with a decrease in ET, GWQ, and SW. These land use change scenarios are analysed here because the MERESE have established afforestation as one of their main and most popular activities, refer to item 2.3.2. for a complete overview of scenario construction and classification.

Many studies find that afforestation mainly produces positive impacts on hydrological processes (Salemi et al. 2013; Ochoa-Tocachi et al. 2016; Bonnesoeur et al. 2019), but it should be noted that these impacts depend on the initial use of the land to be converted to forest (Mosquera et al. 2022; Lalonde et al. 2024). In other words, the impact on hydrological processes is not the same when land use is changed from bare soil to forest as when land use is changed from pasture to forest. In this case, the MERESE scenario to 2050 projects a change in land use from existing grassland to forest, which is probably one of the reasons why we did not find significant impacts in our results, mainly on streamflow. Based on the results found, we can conclude that the proposed afforestation in the study basin as a whole would not be an adequate intervention for the objectives that EPS EMUSAP ABANCAY wishes to achieve. This makes it necessary to evaluate other alternatives that may not necessarily be related to the expansion of more forested areas, such as the conservation and protection of existing ecosystems (e.g., construction of perimeter fences to protect vulnerable ecosystems, and increase conservation and planting of pastures). This type of intervention may be a more appropriate and much more cost-effective alternative, considering the lower costs compared to other intervention alternatives carried out by MERESE. This analysis is supported by the results found by Cervantes Zavala (2022), who concludes that, in the Rontoccocha Hydrographic Unit, wetland and grassland ecosystems are more efficient than the Polylepis Forest in terms of storing and generating the ecosystem service of water regulation per unit area. Also, it is important to consider that there is no single solution or practice to face future challenges, especially in climate change scenarios; therefore, the integration of various adaptation strategies should be evaluated according to the particularities of each context (Qiu et al. 2019).

Climate change scenarios project an increase in precipitation mainly in the summer season (wet period), which would lead to increased water availability. This increase should be adequately managed and regulated to contribute to increased water availability in dry periods. For example, through the construction of natural dams to increase water storage in wet periods. Although we did not analyse sediment in this work, it is clear that increased precipitation would increase water erosion in the Andes, leading to sediment storage in existing reservoirs (Rosas et al. 2020). For this reason, interventions that contribute to sediment control would also be important, for example, the construction of sediment storage lakes, among others.

In the socioeconomic dimension, MERESE still faces significant challenges, which are manifested in different areas (Tristán et al. 2022; Rodríguez Gamarra 2023). At the institutional level, the high rotation of officials hinders the continuity of initiatives, while the lack of political willingness limits the prioritization of investments in conservation and ecosystem recovery. At the economic level, MERESE's financial sustainability is restricted, as most initiatives depend on limited sources of revenue, making it necessary to attract additional funding from international cooperation or the private sector. At the technical level, there are still limitations in the capacity to monitor and evaluate the impacts on ES, as well as limited availability of information on the most effective actions for the recovery and sustainable use of ecosystems. In terms of water governance, there is weak articulation between key stakeholders (e.g., public and private institutions, non-governmental organizations, communities, irrigation users, and universities), which prevents addressing conflicts related to the use of water resources in an integrated manner. In the study area, the MERESE implemented by EPS EMUSAP ABANCAY was supported by the Water for Abancay and its Communities project from 2020 to 2022, funded by the European Union's EUROCLIMA+ programme, which has strengthened the mechanism in four components: (i) supply, through the restoration of ES, (ii) demand, through the optimization of water use and institutional strengthening, (iii) governance, through the strengthening of water management spaces and climate resilience, and (iv) replication, through the design of an intervention model transferable to other cities. For more information, please visit the following link: https://www.helvetas.org/es/peru/lo-que-hacemos/como-trabajamos/nuestros-proyectos/America-latina/Peru/peru-aguaabancay. Despite the achievements of the project, one of the main limitations was the lack of scientific evidence to guide its objectives and interventions more precisely. Currently, the model is being replicated in other cities, and we believe that the results of this study will contribute significantly to generating such evidence, strengthening the scientific basis for informed decision-making.

In relation to water resource management, there is very little research on this issue, despite the fact that this watershed supplies water to the city of Abancay for human consumption and irrigation. In this sense, the results of this study can contribute to the following implications:

Water resources planning in the Mariño river basin is a pending task, especially if considering the current management problems and conflicts for water that affect the area (CBC & PACC 2011). The results of this study will contribute significantly to such planning by generating knowledge on how the impacts of land use change, climate change, and the combination of both affect hydrological processes in the headwaters of the basin. Furthermore, these results provide key information that serves as a basis for the formulation of projects aimed at the recovery and conservation of high Andean ecosystems in areas with similar climatic conditions. In particular, the most appropriate interventions to mitigate the effects of climate change and the trend towards land use change are identified through the implementation of nature-based solutions, focused on soil conservation and erosion control. The aim is to maintain or improve water quality, regulate river flows, and reduce the risk of water disasters.

MERESE in Peru is a successful experience in the conservation of Andean ecosystems in South America, which has been possible thanks to the application of various policies and regulations at the national level, but there has always been considerable uncertainty about the effectiveness and impacts on water resources of the interventions that are currently being carried out. In this sense, the results of this study will help to provide more evidence to improve existing national and local policies.

MERESE is based on an agreement between contributors (e.g., rural communities) and retributors (e.g., SPC), which allows both actors to establish a collaborative and mutually supportive relationship. However, it is important to highlight that the success in building these agreements is achieved thanks to the Good Governance Platform, which is a driving group made up of representatives of public and private entities, non-governmental organizations and universities. This platform constitutes a space for concertation where agreements are defined and investments are promoted aimed at the sustainable management of ecosystems, prioritising the active participation of all its members. In this context, the results obtained in this study will contribute to making the spaces for dialogue promoted by the platform more objective and based on scientific evidence, thus strengthening informed and effective decision-making.

The evaluation of the land use and climate change impacts on water resources involves several sources of uncertainty that can be the result of: (i) input data to the hydrological model (Renard et al. 2010; McMillan et al. 2011), (ii) decisions on methodological choices at the time of hydrological modelling (Pelorosso et al. 2009; Addor et al. 2014; Dwarakish & Ganasri 2015; Hattermann et al. 2018; Chegwidden et al. 2019; Saavedra et al. 2022), and (iii) errors at the moment of choosing the hydrological model to be used. The latter is because most of the time the choice of a hydrological model is based on the legacy and not on the adequacy of the hydrological model (Addor & Melsen 2019). In this research, we use the SWAT hydrological model, because it was able to adequately represent the most relevant hydrological processes in the basin and because it was successfully used in several investigations with similar objectives to this study (Morán-Tejeda et al. 2015; Wang & Stephenson 2018; Chen et al. 2019; Osei et al. 2019; Afonso de Oliveira Serrão et al. 2022).

The set of parameters was selected from the calibration process carried out with the R-SWAT package, which, in the version used during the development of this study, has only a limited number of pre-configured target functions and does not allow the incorporation of new target functions. This limitation has restricted the possibility of improving the modelling results or applying more specific calibration strategies. In addition, the model simulations were performed using the land use map corresponding to the year 2019 and the winter season, which could generate overestimation or underestimation errors in the results obtained for other years, thus increasing the uncertainty in the ET and SW estimates, especially in relation to seasonal changes throughout the year.

The scarcity of meteorological and hydrological data is one of the main limitations for assessing the impact of climate and land use change in high Andean watersheds in Peru, as reported in several research studies (Lavado Casimiro et al. 2011; Llauca et al. 2021; Saavedra et al. 2022). In this study, this limitation is even more critical, as the area of analysis is framed within a MERESE project, where meteorological records are usually small. In an effort to mitigate this restriction, we use data from the REMS (6 years of records) and complement them with the PISCO database for precipitation and temperature (Aybar et al. 2020; Huerta 2022a, b; Millán-Arancibia & Lavado-Casimiro 2023), extending the series to 20 years. While this procedure does not completely eliminate uncertainty, the use of in situ observed data contributes significantly to constrain uncertainty more robustly in hydrological models (Exbrayat et al. 2014). We acknowledge that the relatively short periods that were used for calibration and validation (2 years each) might limit the model's ability to capture hydrological variability in the basin. However, we support the findings of Li et al. (2010) and Bai et al. (2021), who conclude that extending the calibration period does not necessarily improve model performance, as it depends mainly on the representativeness of the data series with respect to the hydrological behaviour of the basin. Consequently, we consider that the good performance of the SWAT model in reproducing the observed flows is sufficient to be confident in the robustness and consistency of our results.

This research evaluated the impacts of land use and climate change on the hydrological responses of a high Andean watershed, as well as the relevance of the MERESE programme in the context of changing scenarios. The study was carried out in the Rontoccocha and Ccayllahuasi basins, which are the main water recharge zones supplying water sources for human consumption and irrigation in the city of Abancay. For this purpose, the SWAT hydrological model was used, which was calibrated and validated with data from the REMS and the PISCO database. Furthermore, 21 individual and combined scenarios of land use change and climate change were constructed and simulated in the SWAT model to analyse their effects on the hydrological processes in the basin. The main conclusions are detailed as follows:

• Despite the scarcity of hydrological and meteorological data in the study area, the results obtained show a good correspondence between observed and simulated flows, achieving satisfactory model performances, with an NSE value of 0.81 during the calibration period and 0.66 in the validation period.

• In the Andes of southeastern Peru, the impact of climate change will be greater (up to 26% increase in mean annual runoff) than the impact of land use change (up to 1% increase in mean annual runoff for the pessimistic scenario). In relation to the land use change scenarios, the SLC21 scenario presents little significant changes with respect to the SLC11 scenario, while the SLC31 scenario presents significantly negative changes (i.e., an increase of up to 11% in the wet period and a decrease of up to −11% in the dry period). In the climate change scenarios, there will be increases in the monthly mean flows, where the largest changes are observed in the wet period (up to 35% increase in the month of December).

• In relation to flows and states, we highlight that for the climate change scenarios combined with afforestation land use, there will be a significant change trend, mainly in the increase of GWQ (10–20%) and in the decrease of SURQ (between 10 and 60%). In contrast, with pessimistic land use there will be a significant trend in decreasing GWQ (−30 to −25%) and increasing SURQ (130–170%). This result shows that there is an interaction between land use and climate change, because the impacts depend on the magnitude and dynamics of the scenarios analysed.

• Understanding the behaviour of the main hydrological processes and their impacts under changing scenarios in a basin significantly helps to guide and prioritise interventions planned in MERESE programmes, promoting more efficient and sustainable management of water resources.

The authors thank EPS EMUSAP ABANCAY S.A. and the Water for Abancay and its Communities Project for generously providing us with data from the Rontoccocha Ecohydrological Monitoring System. In addition, the authors also thank the Servicio Nacional de Meteorología e Hidrología del Perú (SENAMHI) for providing the precipitation and temperature data from the PISCO database.

This work was funded by the Ministry of Environment, Peru.

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

The authors declare there is no conflict.