Introduction

Water quality deterioration is one of the main worldwide challenges. Water pollution, resulting from human activities, affects aquatic ecosystems not only in structure but also in their functions, by disturbing and modifying the integrity of these systems (UNESCO IIWQ International Initiative on Water Quality 2019). Clear evidences are cultural eutrophication and acidification, exerting more pressure on inland and marine waters (Smith 2003; Le Moal et al. 2019).

It is estimated that by 2050, one-fifth of the global population will face risks from eutrophication and one-third will be exposed to water with excessive nitrogen and phosphorous (WWAP United Nations World Water Assessment Programme 2017). Moreso, by that same year, two-thirds of the world population is expected to be living in urban areas, the urban population likely doubling in developing countries (UN United Nations 2015). Medium-sized cities (also called “intermediate cities, with population between 50,000 and 1 million inhabitants) are the fastest growing cities in the developing world. They are also well integrated within a rural region, playing a crucial role in rural–urban interactions given the usually strong link and complementary relationship with their rural hinterland (Bolay and Rabinovich 2004; Bolay 2018; Gorelick and Moktar 2018). The peri-urban, within these cities, is further characterized by substantial poverty, inequality and social fragmentation associated with changes in land and natural resource rights and access along with exclusionary service provision (Wandl and Magoni 2017; Karpouzoglou et al. 2018).

In particular, in Latin America and the Caribbean region, eutrophication is a growing problem caused by nonexistent or insufficient wastewater treatment, agricultural expansion and overuse of chemical fertilizers, urbanization of watersheds, intensive system of animal husbandry, among other (Rivas et al. 2020). Rapid urban growth in most cities of this region often brings poorly regulated development in the peri-urban fringe, with damaging social and environmental impacts (Bolay 2020). This fact leads to pronounced effects on aquatic ecosystems through changes in hydrological connectivity, habitat alteration, water tables and soil saturation, pollution and ultimately species richness and abundance (RAMSAR Ramsar Convention on Wetlands 2018).

Agriculture is the most important activity in Argentina´s economy, involving extremely large croplands, significant volumes of water abstraction for irrigation and an exhaustive application of pesticides and fertilizers to maximize yields (Vammen et al. 2019). The Central region of the country, known as the Pampean region, is one of the more endangered areas worldwide due to cultural eutrophication and agrochemical pollution (Diovisalvi et al. 2015; Castro Berman et al. 2018; Iturburu et al. 2019). It represents one of the largest lentic system areas of South America (Quirós 2005) and it is where the most of oil crop production (mainly soybean and corn) within Argentina takes place. Furthermore, the small and medium-sized city populations of this Pampean region, serve as intermediate centers between the countryside and the urban network. Populations that are in a constant growth and facing up to specific environmental problems that do require appropriate and responsible responses (Bolay 2020).

Numerous approaches have been developed and proposed for analyzing environmental issues arising from anthropogenic pressures and for improving water management resources. Among these, the Driver-Pressure-State-Impact-Response (DPSIR) framework (OECD Organization for Economic Co-operation and Development 2019; European Commission 2002) has been put into widespread use for integrated water resource management (Gari et al. 2015; Patrício et al. 2016). The integrative nature of the DPSIR framework in the study of human-ecosystem interactions means that a significant range of techniques, methodologies and tools must be used to achieve that end (Mateus and Campuzano 2008; Patrício et al. 2016). Some of the procedures are of a descriptive nature (static) as a means of structuring and analyzing information in management and decision-making across ecosystems. (Wang et al. 2015; Gebremedhin et al. 2018). Others require the development of innovative theory and methodologies that are aimed at transforming a static reporting scheme in a framework for dynamic integrated assessment modeling and evaluation procedures (Giupponi 2002; Bell 2012; Matta 2014; Pinto et al. 2014; Maydana et al. 2020).

The aim of this article is to propose a methodology for assessing water pollution problems by combining a conceptual modeling tool (DPSIR framework) with the development of a quantitative model (Multi-Criteria Decision Model). This study provides a practical and flexible evaluation tool for conducting an integrated assessment of eutrophication and agrochemicals delivered to groundwater-dependent shallow lakes and gives action guidelines for decision-making environmental managers in the context of intermediate cities in developing countries.

Through the environmental assessment of such vulnerable freshwater ecosystems and the design of a multi-criteria decision model as a decision-making tool, we sought to achieve the following results: (i) define and structure the problem, i.e., cause and effect relationships between the lake ecosystem and its surroundings, hence the production of the conceptual model (ii) identify key D, P and S indicators integrated in a decision model (model conceptualization) carried out by a multidisciplinary team; and (iii) enable action guidelines prioritization to prevent water pollution and eutrophication based on the criteria and sub-criteria contributions (model implementation).

Study Area

Two lake watersheds from peri-urban areas were studied, one of which is located in Mar del Plata (the principle city of the General Pueyrredon district) and the other located in Balcarce. These intermediate cities are in the south-eastern Buenos Aires province of Argentina (Fig. 1). La Brava and Los Padres are groundwater dependent-shallow lakes, polymictic and permanent; typical freshwater ecosystems from the pampean landscape. Furthermore, they present distinct land use, topographic and hydrological functioning. In addition, they present distinct management regulations for natural resource protection (Table 1).

Fig. 1
figure 1

Location map and land-cover types in (A) Los Padres and (B) La Brava lake watersheds

Full size image
Table 1 Main environmental characteristics of Los Padres and La Brava shallow lake watersheds
Full size table

Los Padres and La Brava lakes have one inflow stream, which forms in the Tandilia mountain system and flows through agricultural lands before discharging into the lakes (Los Padres and El Peligro streams, respectively); each waterbody discharges into one outflow stream (La Tapera and Tajamar streams, respectively). These two lakes are located in the watershed discharge area, receiving upstream groundwater flow contribution by direct discharge into the lake bed or indirectly through baseflow to the streams that discharge into the lakes; meanwhile downstream, lake discharge occurs by infiltration to the Pampeano aquifer (Romanelli et al. 2014b), an unconfined multilayer aquifer formed by loess-like sediments.

It is important to mention that even though both waterbodies and the inflow streams in their discharge area are considered as protected natural areas, their hydrogeological position regarding their watershed (discharge area) together with the existence of several pollutant sources are very likely to increase their vulnerability to water pollution.

Materials and Methods

The proposed methodological framework for analyzing the water pollution problems and to give action guidelines for water managers consist of three steps, which are described in this section and illustrated in Fig. 2. The DPSIR was used to support the problem scoping and structuring step for further conceptualization of the Multi-Criteria Decision Model (MCDM) and its implementation for priority interventions regarding eutrophication and water pollution prevention.

Fig. 2
figure 2

Proposed methodological framework for analyzing water pollution and eutrophication problems in groundwater-dependent shallow lakes

Full size image

Step 1. Problem Definition and Structure: Conceptual Model and Indicator Selection

This section introduces the basic aspects of the conceptual framework developed in this work to evaluate water pollution and eutrophication in groundwater-dependent lake ecosystems located in peri-urban lands of developing countries. This step aims to identify a pool of descriptors able to characterize as well as possible the D - P - S - I compartment of the DPSIR framework. The DPSIR approach and forty-one indicators that arose from the environmental, social and economic dimensions, were grouped in accordance with type and together with the unit of measurements (Table 2). Five Driver indicators, which are often the driving forces behind ecological degradation, were acquired from public databases such as the National Census and Statistics Institute (INDEC Instituto Nacional de Estadísticas y Censos, National Institute of Statistics and Censuses 2012). Eleven Pressure indicators were obtained from updated land cover maps based on Google Earth images. Given that the 17 State indicators were determined through field work, a brief description of the water sampling and in-situ measurements is mentioned. It is important to highlight that the P and S indicators were incorporated and evaluated in the designed MCDM. The eight Impact indicators were incorporated in the assessment framework in order to: 1. evaluate the model outputs performance and, 2. track the evolution of the natural system according to the implemented water management instruments (proposed for a medium-long time horizon). Considering that responses are the policies adopted and measures taken to address a problem, they can be represented by a set of action guidelines. In this work the analysis of response options consists of evaluating and choosing one (or more) water management instruments via a decision support tool (i.e., MCDM outputs) for the lake watersheds.

Table 2 Classification of selected indicators for assessing water pollution and eutrophication in lake watersheds based on the DPSIR framework and approached in a Multi-Criteria Decision Model
Full size table

Land cover data collection: pressure indicators

High-spatial-resolution satellite images provided by Google Earth were used to update the existing land use maps (Romanelli et al. 2014a) in each lake watershed. Visual interpretation was employed to identify different land covers in Google Earth, and the “Add polygon” tool was used to vectorize the area. To ensure that the identification criteria were consistent, only one person was assigned to do this. Land cover was classified into 13 distinct categories: shallow lakes, broad-leaved forest, natural grasslands, pasture, bare rock, littoral vegetation, intangible zone, extensive agriculture, open-fields crops, shade cover crops, greenhouse crops, sport and leisure facilities and village/residential zone (Fig. 1). The KML files of all land cover types were transformed into shape files, which could be read by ArcGIS (v9.3, ESRI). In ArcGIS, the information gotten from each polygon was marked in an attributes table and all shape files were merged into one. Finally, a land use map was designed (Song et al. 2018). All maps were projected into the Argentine Gauss Kruger system, zone 6 (Campo Inchauspe Datum).

Water quality data collection: state indicators

Sampling stations (wells N = 15, streams N = 6 and lakes N = 7) representative of Los Padres and La Brava lake watersheds were selected for hydrochemical analysis and to determine pesticide concentrations (Fig. 1). Water samples were collected on 16th and 22nd of November 2016 and on 3rd and 4th of July 2017, respectively. Subsurface (0.5 m depth) lake and stream samples were obtained. All groundwater samples were collected from mills, domiciliary and irrigation wells that had been purged sufficiently to remove at least three borehole volumes prior to sampling and were only taken once stable field measurements-pH, electrical conductivity (EC) and temperature-were obtained. Water temperature, pH and EC were in-situ measured and, additionally, water clarity (Secchi transparency) and depth at lake water sampling sites.

A total of 56 water samples (1 l) were collected using pre-cleaned polyethylene bottles and transported in cold (4 °C) and dark conditions to the laboratory. Chemical analysis (N = 28) included major ions and nutrients (nitrate, total phosphorus-TP and phosphate) determination, following standard methods (APHA American Public Health Association 1998). For pesticides determinations (N = 28), once the samples were collected, they were frozen and stored until further analysis. Seventy-seven pesticides including organophosphates, organochlorines, carbamates, pyrethroids, fungicides, herbicides and phenyl-pyrazoles were determined following the EPA SW 846 CAP 8000 method. Shallow lake water samples were analyzed in the laboratory for: 1. Dissolved Oxygen (N = 7) (APHA American Public Health Association 1998); 2. Phytoplankton chlorophyll a, b and c (Chl a, b and c) biomass only in lakes (N = 7, 10 replicates per sample); and 3. Bacterial indicators of water sanitary quality (Total -TC- and Thermotolerant Coliforms -TtC- and presumptive Escherichia coli -EC-; N = 7 water samples). Finally, water clarity, TP and Chl a data were used to calculate the Trophic State Index (TSI) for each shallow lake (Carlson 1977). All analytical procedures are described in Supplementary material.

Step 2. Model Conceptualization: Multi-criteria Decision Model Description

In this paper a Multi-criteria Decision Model was designed and implemented to assess water pollution on a lake watershed scale, by implementing the software Criterium Decision Plus 4.0 Beta software (CDP, Murphy 2014). The P and S selected indicators were inputs of this decision support tool, in terms of criteria and sub-criteria relevant to water resource management. The overall objective of this MCDM was to establish priority areas and responses needed to prevent water pollution and eutrophication according to criteria contributions.

Multidisciplinary expert judgment was used to select and weight each criterion. Weights for each criterion of the decision model were derived from the standard pair-wise comparison procedure of the Analytic Hierarchy Process (Saaty 1992) in which a decision maker/scientist is asked to judge the relative importance of one criterion versus each of the other criteria. We provided the judgments on relative importance for our example application. The model structure based on the D, P and S indicators consists of three primary criteria: pollutant load, land-cover and conservation zones. Each criterion was further decomposed into secondary and tertiary criteria (Table 2). Evaluations of data were performed with the Simple Multi-Attribute Rating Technique (SMART functions) (Kamenetzky 1982), which map observed values into a measure of utility for each attribute (lowest level) of an alternative. The utility function for each attribute was specified by two pairs of x, y parameters that defined a simple ramp, with one pair specifying the condition for complete utility (y = 1, worst situation), and the other the condition for no utility (y = 0, best situation) (Table 3). In the decision model for the lake watersheds under study, upper, middle and lower basins adapted from (Romanelli et al. 2014a) were used as analysis units (alternatives).

Table 3 Description of the representative criteria and sub-criteria used in the Multi-criteria Decision Model (MCDM): definitions, scale information, and data sources
Full size table

Step 3. Model Implementation: Multi-criteria Decision Model Outputs and Application

The multi-criteria decision model was run two times, once per lake watershed. The decision process results in ranking of the competing alternatives that contribute to the decision goal by giving score values. The decision score for an alternative in the model is the sum of all the ratings of that alternative against each lowest criterion, weighted by the importance of that criterion to the decision. The analysis of which criteria actually made the largest contribution in the decision and which made the least was performed based on stacked histograms of the accumulated values of the alternatives at a target criterion. According to the decision scores, priority maps for each lake watershed were obtained and classified into three categories (i.e., high, moderate and low priority). Several water pollution prevention responses were proposed to be applied in each alternative according its priority, through five instruments (FAO/IWMI Food and Agriculture Organization of the United Nations, International Water Management Institute 2018): 1. regulation instruments, 2. economic instruments, 3. education and awareness-raising, 4. cooperative agreements, and 5. research and innovation.

Results

Conceptual Model Definition

Drivers

In the causal chain of the DPSIR conceptual model the key driving forces affecting water resources in this research include agriculture, urbanization, and recreation. Agriculture has been growing spectacularly since 2003 in Argentina. The existence of a competitive exchange rate, together with a favorable international context marked by a sharp rise in prices (especially soy) have favored a major expansion of agricultural activity, which in turn, has lead to an increase in soybean production (Ministry of Economy of Buenos Aires Province, http://www.ec.gba.gov.ar/). Soybean is Argentina’s principle crop and accounts for 52 % of the total crop area (Leguizamón 2016), with 11,331,799.8 hectares cultivated in Buenos Aires Province during the year 2018 (INDEC Instituto Nacional de Estadísticas y Censos, National Institute of Statistics and Censuses 2012). Based on district-level data on soybean harvested area, Balcarce had 54,200 hectares in the year 2010 and 71,539 ha in 2018/2019, whilst in General Pueyrredon District, the harvested area was of 21,550 and 20,706 ha in 2010 and 2018/2019, respectively (Argentine Ministry of Finance, http://www.estadistica.ec.gba.gov.ar/). The last mentioned district also holds the second most important fruit-horticultural area of Argentina, growing in size and productivity (> 1,300 ha and > 300,000 T) over the last 50 years (Gonzalez et al. 2017). From 2005 to the present, there has been an increase in crop systems under cover (i.e., greenhouse) by 233.67 % and in open-field crops by 178.44 % (Daga et al. 2017).

In both districts, the intensive agricultural activity mentioned above shares a complex territory with various discontinuous settlements possessing different socio-economic conditions: Sierra de Los Padres, Santa Paula, Colinas Verdes, Villa La Brava, among others. In recent decades, growth within the urban and peri-urban areas together with the expansion and consolidation of residential land use on rural lands, has been evident in the region. The total population increased from 564,056 to 614,350 inhabitants and from 42,039 to 44,064 inhabitants in General Pueyrredon and Balcarce districts, respectively from 2001 to 2010 (INDEC Instituto Nacional de Estadísticas y Censos, National Institute of Statistics and Censuses 2012). In reference to the settlements situated closest to the lakes, Sierra de Los Padres experienced an increase of 429 % in its population (4249 inhabitants in 2010), compared to the previous population census (INDEC, National Institute of Statistics and Censuses 2001). Contrary to this, Villa La Brava with 115 inhabitants, evidenced a decrease of 34 % on its total population (174 inhabitants in 2001). It must be taken into consideration, that a large number of households within rural areas lack piped water supplies, being of 85 % and 98 % in Balcarce and General Pueyrredon districts, respectively.

Recreational activities usually have less impact on aquatic ecosystems than others (e.g., agriculture, urbanization) and the impact caused is usually more localized, both spatially and temporarily (Kakoyannis and Stankey 2002). Regarding recreational uses of the area, people normally use the lake shoreline in an unpredictable way mainly during weekends and holidays, occasionally getting together for fishing, picnics or playing football (Cardoni et al. 2008). Among the physical pressures, the noise and waves produced by boats, damage to riparian vegetation by trampling, camping, boats and canoes that can stir up lake bottom sediments, leading to an increased water turbidity (Mosisch and Arthington 1998). Moreover, increased levels of fecal coliform bacteria (e.g., E. coli), as a result of an inadequate sewage system from campsite toilets nearby, along with the presence of water bird colonies, put more biological pressure onto these ecosystems. Due to the lack of data in reference to recreational trends in this area, one indirect D indicators was selected, the road network density (i.e., the ratio of total road length and each lake watershed area). The road network around this area gives an easier access to wildland settings (previously remote areas) for outdoor recreational activities, thereby increasing impact (e.g., pollution) on water ecosystems and their surroundings. The values of this parameter are of 0.54 and 0.69 km/km2 for Los Padres and La Brava lake watersheds, respectively.

Input Data to the Model: Pressure and State Indicators

Land-cover and conservation zones

Land-cover and conservation zone information were included in each model and placed in 4 and 5 categories, respectively. In Los Padres lake Watershed (LPW), the Upper basin included extensive agriculture (12.21 %), natural grasslands (5.5 %), broad-leaved forest (4.9 %) and open-fields crops (0.92 %). In relation to the Middle basin, extensive agriculture (47.97 %), village/residential zone (8.1 %), natural grasslands (7.8 %) and broad-leaved forest (4.5 %) were reported. Finally, the Lower basin is characterized by extensive agriculture (41.55 %) and open-fields crops (36.52 %); to a lesser extent, land-covers such as village/residential zone, broad-leaved forest, natural grasslands, intangible zone, shade cover crops, greenhouse crops and sport and leisure facilities were evident in this sector.

In reference to La Brava lake Watershed (LBW) model, the Upper and Middle basins were represented by extensive agriculture (35 %, 48 %), natural grasslands (9.53 %, 8.52 %) and broad-leaved forest (3.2 %, 0.24 %) in a lesser proportion, respectively. Concerning the Lower basin, extensive agriculture (31.67 %), natural grasslands (8.15 %), shade cover crops (7.37 %), village/residential zone (5.15 %), broad-leaved forest (4.8 %) and littoral vegetation (1.06 %) were identified. Only streamside vegetation was classified for both watersheds by qualitative labels, being very important in the Upper basin, important in Middle basin and unimportant in the Lower area.

Pollutant load

Table 4 shows the results of the physical, chemical, and microbiological analyses conducted in Los Padres and La Brava lake Watershed, together with the limits specified by international guidelines for aquatic life criteria and drinking water for human consumption.

Table 4 Indicators of State detected in samples in Los Padres and La Brava lake watersheds
Full size table

Pesticide levels in groundwater, inflow streams and lakes

Los Padres Lake watershed (LPW) had the highest total pesticide levels in all water sources, with concentrations of up to 0.29, 0.34 and 0.31 µg/l in groundwater, lakes and streams, respectively. Regarding LBW, total pesticide concentrations were found in lake > groundwater > inflow stream (0.21, 0.11 and 0.05 µg/l, respectively).

As a general pattern, the occurrence of pesticides in water can be ranked from highest to lowest concentrations as follows: organophosphates (OP) > pyrethroids > herbicides were found in both lakes, where chlorphyrifos ethyl (CPF-E) was quantified in 41.67 % of the samples, followed by 2,4-D (29.17 %) and permethrin (16.67 %). In groundwater from LPW, OP represented 89.65 % of the total pesticide load, followed by the herbicide 2,4-D (12.07 %). A similar distribution pattern was found in surface water, where OP accounted between 48.4 and 55.9 % followed by pyrethroids (36.8 %) within the lake and herbicides (41.9 %) in streams. In La Brava Lake, OP were predominant in all samples accounting for 43.6, 63.3 and 36.5 % regarding total concentrations in groundwater, streams and lake samples, respectively.

In reference to OP, CPF-E presented the highest levels in Los Padres Lake samples (0.05–0.11 µg/l) with the following distribution: lake = groundwater > streams. In La Brava Lake, this insecticide represented between 36 and 60% of the total pesticide concentration, with the highest levels within the lake and the lowest in the inflow stream (0.08 and 0.05 µg/l, respectively). Permethrin, the main compound among pyrethroids, was only detected in the inflow stream (0.035 µg/l) and within Los Padres Lake (0.043 µg/l on average). Furthermore, bifenthrin was detected only in this lake (0.04 µg/l), while, the insecticides bifentrin and permethrin were found in two samples in La Brave Lake (0.01 and 0.02 µg/l). Finally, surface water samples mainly presented 2,4-D residues in both watersheds, either in the lake and the inflow streams (0.016 and 0.05 µg/l on average, respectively).

Dissolved inorganic nutrient concentrations in groundwater, inflow streams and lakes

In relation to nitrate content found in the water samples (i.e., groundwater, lakes and streams), Los Padres Watershed had the higher values (31.4 ± 28.5 mg/l) in comparison to LBW (24.7 ± 12.5 mg/l). Nitrate mean values in water can be ranked from the highest to the lowest concentrations as follows: groundwater > inflow stream > lake in Los Padres Watershed, and, inflow stream > groundwater > outflow stream > lake in La Brava Lake Watershed.

Phosphate levels are also a serious matter in waters discharging into waterbodies. In relation to phosphate content in water samples, LBW had little highest values (10.52 ± 5.58 mg/l) in comparison to LPW (8.75 ± 6.44 mg/l). Mean phosphate values in Los Padres Lake were very similar to the inflow stream values, while the average upstream groundwater values were the highest. An analysis of phosphate content in groundwater along the watershed, could be ranked from the highest to the lowest concentrations as follows: upper > lower > middle basin.

In the case of LBW, mean phosphate levels were of 7.31 ± 3.54 mg/l for the lake, 5.47 mg/l for the inflow stream, 13.59 ± 5.83 mg/l for upstream groundwater and 10.95 ± 3.75 mg/l for the outflow stream. Mean phosphate values in groundwater decreased from the upper to the lower basin showing the following values: upper > middle > lower basin.

Bacterial presence in lake waters

All surface water samples showed evidence of coliform bacterial presence (total coliforms -TC- and presumptive Escherichia coli -EC-). Bacterial concentration was higher in Los Padres Lake and could be related to higher water temperatures, which favor bacterial growth, coinciding with the water sampling period during spring. In both shallow lakes the sites located closest to the inflow stream discharge to the lake (WLP002 and WLB038) evidenced the highest TC and EC values. Coincidentally, the high E. coli were detected in those lake sites located close to the water bird breeding colonies in the inflow stream delta. The confirmation of E. coli in water samples indicated a recent fecal contamination, meaning that there was a greater risk of pathogens being present.

Eutrophication and Water Pollution: Impact Indicators

In this study, eight indicators were used to describe the impact of anthropogenic activities in both lentic systems (Table 2), being these descriptors higher in LPW and showing an exacerbated situation within this waterbody. Related to eutrophication, the calculated Trophic State Indexes (TSI) (mean values of 98 and 84 for Los Padres and La Brava lakes, respectively) confirmed the eutrophic condition of both waterbodies, as reported by Romanelli et al. (2013, 2020). Regarding the phytoplankton community, mean chlorophyll a values registered in both waterbodies permitted a classification of them as lakes in a “non-clear” state (Scheffer and Jeppesen 2007), with a phytoplankton predominance instead of macrophytes and benthic communities and a depth/clarity ratio higher than 1 in both cases. In Los Padres Lake this community was characterized by low algal diversity due to a dominant cyanobacteria, which has caused an algal bloom. In La Brava Lake, phytoplankton showed a greater algal diversity, with the presence of several centric and pennate diatom taxa, along with cyanobacteria in water samples from this lake. However Sphaerospermopsis aphanizomenoides, a filamentous N2-fixing cyanobacteria species capable of living under several light conditions in the water column (Sabour et al. 2009; Budzyńska et al. 2019), dominated the phytoplankton of both lakes in numbers > 10.000 cells/ml and with a relative abundance of 90 % in Los Padres Lake.

Chlorpyrifos-ethil (CPF-E) is one of the most broad-spectrum insecticides used in Argentina in agriculture and domestic animals, therefore increasing its discharge into aquatic environments (Loewy et al. 2011; Pérez et al. 2017). Regarding the aquatic biota protection criteria, CPF-E levels recorded in 33.3 % of stream and 50 % of lake samples in LPW, exceeded the recommended Water Quality Criteria for Aquatic Life (0.041 µg/l, USEPA United States Environmental Protection Agency 1986). Furthermore, 66.7 % of La Brava Lake samples were above the protection limits. These results showed the agricultural impact on both watersheds and the threat of aquatic biodiversity loss. Drinking water quality criteria is an important human health topic. In the present study, 28.6 % of the wells analyzed in Los Padres Lake (Table 4), presented CPF-E concentrations higher than the maximum levels recommended by the European Union (0.5 mg/L; Hamilton et al. 1994). These results suggest a potential human health hazard due to water consumption in both watersheds.

Both lake watersheds had a large number of wells with nitrate concentrations above the drinking water standard (45 mg/l NO3, WHO World Health Organization 2011), being of 62.5 and 28.6 % in La Brava and Los Padres lake watersheds, respectively. Moreover, all surface water samples had total P levels significantly above the U.S. EPA recommendations for streams, lakes and reservoirs (0.1 – 0.025 mg/l) (USEPA United States Environmental Protection Agency 1986; Horne and Goldman 1994).

Among fecal pollution indicators (i.e., thermotolerant coliforms and E. coli), only 25 % of samples collected from Los Padres Lake exceeded TtC maximum levels, classifying it as an ecosystem with a moderate microbial pollution, not apt for recreational use with primary contact (UN Environment 2016). According to USEPA United States Environmental Protection Agency (2002), 25 % of the samples did not comply with the criterion related to E. coli values. In La Brava Lake, all samples analyzed presented TtC and EC concentrations lower than the levels recommended by the afore-mentioned water quality criteria. In-lake concentration of TtC (< 200 CFU/ml) indicated that it has low microbial pollution (UN Environment 2016).

Multi-criteria Decision Model Outputs: Prioritization, Criteria and Sub-criteria Contributions

Based on the weighting scheme of the MCDM (Table 2) for assessment priorities for water pollution prevention, Lower basin alternative was evaluated as high priority, Middle basin as moderate priority and Upper basin as lower priority for the implementation of water protection strategies in both watersheds (Fig. 3). Priority scoring in LPW model ranged from 0.127 to 0.389 and in LBW model, it ranged from 0.123 to 0.322.

Fig. 3
figure 3

Priority map and criteria contributions (level 1) for water pollution prevention in (A) Los Padres and (B) La Brava lake watersheds. Numbers and letters indicate the recommended actions (responses) for each alternative (see Results section)

Full size image

In reference to the primary criteria contribution (Fig. 3), the Lower basin alternative was determined as high priority class by the contribution of conservation zones (0.18), pollutant load (0.12) and land-cover (0.08) in LPW model, whilst in the LBW model, the high priority class presents a criteria contribution of conservation zones (0.16), pollutant load (0.11) and land-cover (0.04). Middle Basin shows a similar criteria contribution in both models: conservation zones (0.07), land-cover (0.06) and pollutant load (0.05). In the case of lower priority, pollutant load contribution was higher in LPW than LBW; land-cover contribution was higher in LBW than LPW whilst conservation zones contribution was similar in both watersheds. The sub-criteria contributions (secondary and tertiary) to prioritize alternatives are shown in Fig. 4 and are summarized in Table 5.

Fig. 4
figure 4

Sub-criteria contributions (levels 2 and 3) to prioritize alternatives in (A) Los Padres and (B) la Brava lake watersheds. IS: inflow stream, SL: shallow lake, GW: groundwater

Full size image
Table 5 Sub-criteria contributions for each alternative in the prioritization models for Los Padres and La Brava lake watersheds
Full size table

Responses: Provision of Scientific Evidence-based Recommendations through Water Management Instruments

In accordance with the MCDM outcomes, several water pollution prevention responses were proposed to be applied in each alternative through five instruments (FAO/IWMI Food and Agriculture Organization of the United Nations, International Water Management Institute 2018): 1. regulation instruments, 2. economic instruments, 3. education and awareness-raising, 4. cooperative agreements, and 5. research and innovation.

Figure 3 shows the recommended actions (Responses) for each alternative according to primary, secondary, and tertiary criteria contributions to water pollution and eutrophication. Regarding the alternative with high Priority in the LPW model (Lower basin, Fig. 3 A), the contribution to the water pollution was mainly determined by conservation zones, pollutant load, and land-cover. Within these criteria, littoral vegetation, pesticides, and extensive agriculture were the secondary criteria with the highest contribution. Concerning the alternative with high priority in the LBW model (Lower basin, Fig. 3 B), there was a predominance of the conservation zone and pollutant load criterion mainly determined by littoral vegetation, bacterial load and nutrients, as secondary criteria.

According to the LP model outcomes (i.e., criteria contributions), the recommended actions are the following: (1.a) reduction of synthetic agrochemical (pesticides and fertilizers) application as well as a promotion in the usage of organic compounds, mainly fertilizers, (1.b) guidelines on Integrated Pest Management and utilization, (1.c) promote crop biodiversity, (1.d) improvements in irrigation management or an erosion control. These actions would reduce or eliminate leaching and drainage or sediments transport off-farm, leading to reduce nutrients and pesticide export, (1.e) better provincial and local district authority control of EIAs regarding new developments close to the lake and its surroundings, and finally (1.f) assessment of groundwater pollution hazard as a preventive tool for aquifer pollution.

In the case of LB model results, the suggested strategies are: (1.g) definition and conservation of buffer zones including the shallow lake, the inflow stream and delta, (1.h) an increase in data collection to establish nutrient criteria and to develop water quality models in order to translate their results into better water policies, and (1.i) promote the adoption of good agricultural practices by education, environmental awareness and economic incentives to farmers, (1.j) provide the necessary information to local residents regarding septic tank location and construction, and promote Eco-Friendly wastewater treatment systems in order to reduce the bacterial inputs to the lake, (1.k) a revision of the land use code for protected natural area (e.g., definition of protected area zoning considering critical areas for conservation, areas for recreation, etc.).

Some recommendations related to economic, education and research instruments are given for both lake watersheds. From an economic point of view, 2. to provide financial support to develop and apply “environmentally friendly” techniques for crop production and a reduction in taxes for products grown within an agroecology or similar systems. Regarding cooperation agreements, 3. including partners external to government by creating lake-watershed committees and other governance structures that will enhance the current level of coordination. Finally, in relation to education and research instruments: 4. promotion of environmental education and awareness in order to involve stakeholders in the land-use planning process, and 5. increase financial support to the scientific research related to the remediation of polluted soils with agrochemicals and biodegradation of pesticides.

Discussion

Water managers still need operational tools to assess the influence of human activities on water resources. Although the DPSIR framework is considered a valuable tool, it has been proposed that it could be improved by combining it with other methods (Gari et al. 2015; Pinto et al. 2014; Ramos-Quintana et al. 2018). Water management issues require the use of both conceptual and numerical models to inform about water planning and decision-making, beginning with the essential step of scoping and identifying the problem of interest before further analysis and modeling (Zare et al. 2019). The proposed methodological framework has several strengths: First, the identification of a pool of 41 descriptors able to characterize as well as possible the D - P - S - I compartment of the DPSIR framework, giving structure to the environmental problem under study. Secondly, the use of DPSIR helped to define the modeling elements and identify the key criteria and sub-criteria to be included in the analysis, as modeling input data. Finally, the implementation of a coherent quantitative assessment of environmental variables (i.e., drivers, pressures, and stressors) transformed into different weighted criteria (relative importance of one criterion versus each of the other criteria) thereby facilitating their interpretation within a dynamic process to support structured decision-making, that is based on a multi-criteria decision analysis.

In this work, the conceptual and quantitative assessment proposal includes updated information obtained through satellite images, field work, and public datasets (e.g., population census). All this information related to environmental issues (social, economic and environmental data) is originally collected for different purposes, at different scales, within different time frames and with different underlying assumptions about the nature of the phenomena, causing technical difficulties to the integration of social and environmental data. Other limitations of the proposed methodology are: the amount of input data necessary at every step of the procedure in order to accurately record the decision-making problem, the subjectivity in the selection of the key indicators and the assignment of weights involved in the task.

Small urban centers and villages around cities are turned into peri-urban areas due to changes of city and municipal boundaries in the developing world (Nuhu 2019; Bolay 2020). In most of these countries, the peri-urban is often a zone of chaotic urbanization (Ravetz et al. 2013), used for a multitude of purposes, leading to social and environmental problems (Allen 2003; Busck et al. 2006). In addition, developing countries are commonly hit hard by financial and economic crises; in such cases, economic crisis has important implications for policy solutions to environmental issues due to a lack of resources. Given the complex and multidisciplinary nature of many of the spatial decision problems, such as water pollution assessment and management, the proposed methodology allowed prioritizing areas and identifying specific actions to prevent water pollution-induced problems in accordance with the main criteria contributions, in critical socio-ecological contexts such as peri-urban lands in middle-sized cities from the Global South. Financial, human and infrastructure resources for water pollution management are scarce here and the combined use of the DPSIR conceptual framework with the MCDM analysis can help to ensure that we get the greatest improvement with the available resources.

Knowing criteria and sub-criteria contributions to prioritize alternatives is a benefit when assessing water pollution and eutrophication, since it allows identifying which aspect (environmental or socio-economic criteria) dominates in the final result. As a consequence, this fact helps decision makers focus on which sub-criterion should be evaluated and monitored, and also which management instrument should be applied.

The decision support tool implemented in both lake watersheds, selected as case studies from peri-urban lands in the Pampean Region, therefore providing a common basis of understanding of the water pollution problems and a quantitative ranking (i.e., decision scores) for defining specific actions (responses) for human-induced stresses on such natural systems. Several differences related to the water pollution assessment in each lake watershed were highlighted within the final resolutions. In the case of LPW, the main criteria contributing to these water problems were pesticides, extensive agriculture, nutrients, herbicides in the inflow stream and nitrate in upstream groundwater. On the other hand, in LBW, the same issues were characterized by nutrients, extensive agriculture, phosphate and nitrate in upstream groundwater. In addition, decision scores defined the lower basins, where both groundwater-dependent shallow lakes are located, as those with the highest priority for intervention; the greater contribution of the primary criteria could be related to its location in the regional discharge area and the receiving upstream pollutant loads (from the inflow streams and groundwater).

Conclusions

This study evidences the impact of anthropogenic activities on lentic systems, causing water quality deterioration in both lake watersheds, being exacerbated in LPW. Nutrient and pesticide inputs from agricultural lands play a significant role in water quality deterioration and eutrophication in both lakes. This fact may represent an environmental problem by impeding recreational uses, threatening drinking water sources (downstream groundwater) and degrading the overall ecological health of the lakes.

This research contributed to the proposal of 41 indicators following the DPSIR conceptual framework with the development of a quantitative assessment of water pollution and eutrophication in freshwater systems from peri-urban areas in developing countries. Pressure and State indicators were included in a multi-criteria decision analysis/model (MCDA) performed by a multidisciplinary team. The obtained decision scores for each lake watershed allowed the establishment of a ranking of priority areas and therefore identify which criteria to focus on in order to establish action guidelines (responses) to minimize water pollution problems, particularly in emerging countries, were the low availability of financial, human resources and infrastructure have serious implications in providing solutions to specific environmental issues.

This decision-support DPSIR framework for water pollution assessment could easily and successfully be extrapolated to further regions or other environmental issues, especially in the context of developing countries. It provides a common basis of understanding and encoding of socio-ecological information in relation to an environmental problem, which can be spatially and temporally analyzed. It also enables direct correlation to adjacent basins, or comparative studies of the same basin at multiple time scales, whilst at the same time contributing to effectively designing and implementing specific measures toward the protection and restoration of any given hydrogeological environment.