Performance evaluation of a community-based water management organisation using integrated water quality monitoring

Sustainable water resource management remains a challenge worldwide (Hering & Ingold 2012; Mdee et al. 2022), especially in terms of access to drinking water and sanitation. Nearly one-quarter of the global population lacks access to safely managed drinking water, while over half of the population, ∼4.2 billion people, lacks adequate sanitation services (WHO 2021). The aim of reducing such disparities by 2030 is outlined in the United Nations Sustainable Development Goals (United Nations 2015). This undoubtedly represents a considerable challenge, particularly in developing economies or rural settlements where limited economic, technical, and human resources impede effective water resource management (Mena-Rivera & Quiros-Vega 2018; Patton et al. 2020). To overcome some of these challenges, a community-based water management model has been widely implemented (Hutchings et al. 2015; Tantoh & Simatele 2017; Machado et al. 2019). However, there is a general debate on their capacity to ensure both short- and long-term water security (Stedman et al. 2009; Mihelcic & Schweitzer 2012; Chowns 2015; Machado et al. 2022).

Community-based water management aims to improve water services by delegating responsibility to individuals within the community who work within the framework of a formal committee (Schouten 2003). Community-based water management organisations (CBWMOs) are then responsible for (a) ensuring inclusive decision-making processes, (b) providing training, maintenance, and technical support, and c) ensuring financial stability to adequately sustain the operation of the water distribution system (Day 2009; Mihelcic & Schweitzer 2012). This partially decentralised structure seeks to increase efficiency while also developing a sense of community ownership (Day 2009; Chowns 2015; Kativhu et al. 2018). Despite the wide range of responsibilities, the performance of CBWMOs has been primarily evaluated on the basis of improving infrastructure and administrative capabilities (Madrigal-Ballestero et al. 2013; Machado et al. 2022; Nowicki et al. 2022). The considerations of their long-term capacity to maintain adequate water quality have often been neglected. In addition, the influence of socioeconomic factors and operational monitoring on water quality outcomes within these organisations is still underexplored.

In general, data on drinking water quality are still scarce or inaccessible (WHO 2022). Yet, monitoring efforts have shown widespread contamination in both water distribution systems (Bain et al. 2014) and water sources (Burri et al. 2019; Hoque et al. 2021). The risk of consuming unsafe water is highest in rural areas of middle- and low-income countries, where most CBWMOs operate. In terms of monitoring water quality, CBWMOs often rely on government or external support, as it is expensive and time-consuming, which limits testing frequency, the numbers of pollutants analysed, the response time, and their capacity to ensure safe drinking standards. There is also some uncertainty as to whether the outcome of such programmes is used effectively (Ward et al. 1986; Timmerman et al. 2010; Kumpel et al. 2020). Nevertheless, water quality monitoring programmes are essential to ensure compliance with safe drinking standards, improve the operation of the distribution system, identify pollution sources, and assess water security (Charles et al. 2020; Nowicki et al. 2020; da Luz & Kumpel 2020). To this end, integrated monitoring would require assessment at the point of use, at all the water sources (e.g., wells and springs) and other relevant water bodies (e.g., rivers and lakes). Implementing integrated monitoring programmes, followed by comprehensive data interpretation, is necessary to further evaluate the effectiveness of CBWMOs.

In this study, we present a comprehensive evaluation of the performance of a CBWMO from a water quality perspective, focusing on a rural settlement in Costa Rica. The characteristics of this CBWMO, in terms of capacity and management, make it representative of similar organisations in the country. Our integrated assessment includes seasonal water quality data collected throughout the water distribution system, from sources to the point of use, complemented by a risk analysis of the aqueduct infrastructure. To identify potential anthropogenic pressures, we also analyse water quality in the main river of the catchment. Water quality data is interpreted using hydrochemical and multivariate techniques. We aim to (a) evaluate the capacity of CBWMO to provide safe drinking water, (b) identify the primary driver of water composition in the sources, and (c) identify potential threats that could compromise local water security. We expect to provide a better understanding of the challenges faced by CBWMOs in maintaining adequate water quality and ensuring sustainable water resource management in the region.

Costa Rica has made significant progress in providing safe drinking water access and sanitation services in recent years (Merino-Trejos 2019). In terms of water supply, almost 99.5% of the population has access to water, and between 89.5 and 91.8% receives water that meets drinking standards (Mora-Alvarado & Portuguez-Barquero 2018; Mora-Alvarado et al. 2023). On the other hand, sanitation coverage is expanding, but at a slower pace (Vaux et al. 2020). Septic tanks are the typical treatment system (75%), with sewerage accounting for only 23.8%. Only 17.6% of wastewater is adequately treated before discharged into fluvial systems (Mora-Alvarado & Portuguez-Barquero 2018). The country faces several challenges that could compromise water security, including inadequate land use, increasing anthropogenic pollution, and global climate change (Bower 2013). Unappropriated agricultural practices and poor urban development have historically contributed to the degradation of water quality (Mena-Rivera et al. 2017, 2018; Sanchez-Gutierrez et al. 2023). Furthermore, the lack of adequate infrastructure (Esquivel-Hernández et al. 2018) and the strong climate variability (e.g., long-term drought and heavy rainfall) in the Central American region (Hund et al. 2018; Sánchez-Murillo et al. 2020) are highlighted as the major threats to water resource management in the country.

In Costa Rica, CBWMOs manage more than 2,200 aqueducts that supply water for ∼1.5 million people. This represents ∼29% of the total drinking water supply in the country (Mora-Alvarado & Portuguez-Barquero 2018). The CBWMOs operate most of the aqueducts under the schemes of the Comités Administradores de Acueductos Rurales (CAARs) and the Asociaciones Administradoras de Sistemas de Acueductos y Alcantarillados Sanitarios (ASADAS). These are both non-profit organisations, but only the latter is previously authorised and audited by the Instituto de Acueductos y Alcantarillados; the government's institution responsible for providing drinking water supply and sanitation services (Madrigal et al. 2011; Madrigal-Ballestero et al. 2013). Although CAARs and ASADAS play an important role in the supply of drinking water, some of these aqueducts are under highly vulnerable conditions. For instance, 58% of the CBWMOs do not take any action to protect the sources of water, although 87% of systems are supplied from springs and surface waters (PEN 2019), which might be poor quality or highly susceptible to increasing pollution. The use of water sources that do not meet the required standards for drinking purposes has been previously reported (Mena-Rivera & Quiros-Vega 2018; Gómez-Cruz et al. 2019; Sánchez-Gutiérrez et al. 2020a, b). Given that 34% of aqueducts do not undertake any control of water quality and 25% do not have continuous disinfection (PEN 2019), there is some uncertainty as to whether CBWMOs provide safe drinking water to the population.

The water supply in Concepción of San Ramón is provided by the local ASADA. The current distribution system comprises 16 springs, four connection points (CPs), one pumping station, and three storage tanks, distributing water to four different sectors comprising 71 homes, farms, and other municipal buildings. The system is divided into two sub-units. In the first sub-unit, at the top of the catchment, water is transported by gravity through a system comprising four CPs and 15 springs. CPs are defined as independent chambers that do not drain any spring; for instance, water from S1 is discharged into the spring chamber S2, the latter being then not considered a CP. Water from this upper system is stored in tank 1 (T1, capacity 150 m³) and distributed to sectors 1 and 2. In addition, a fraction is transported and stored in tank 2 (T2, capacity 75 m³). In the second sub-unit, located in the middle of the catchment, water from S14 is pumped into tank 3 (T3, capacity 300 m³) and distributed to sectors 3 and 4. This system only operates when water levels are low in the upper sub-units (T1 and T2). Water is disinfected prior to distribution.

Accessibility, monitoring, and infrastructure conditions were evaluated each spring between May and June 2017. The risk assessment of the springs was divided between the protection zone (i.e., 100–150 m radius) and the spring chamber (Meuli & Wehrle 2001). A semi-structured instrument was also applied in the farms nearby (n = 13) to gather information about agricultural practices and identify potential sources of pollution.

Water samples were collected during four sampling campaigns (October 2016, February, April, and June 2017) from the springs (n = 60), households (n = 16), and the Cañuela River (n = 16) (Figure 1). These include all the springs that supply the distribution system, households in each distribution sector, and river sites at the top, middle, and bottom of the catchment. Samples for the physico-chemical and microbial analyses were collected in high-density polyethylene (HDPE) bottles and non-reusable sterilised vials, respectively. HDPE bottles were previously washed with 3% m/v HCl and de-ionised water. Before collecting the samples from the households, pipes were purged for at least 5 min to avoid potential stagnant water. All samples were transported to the laboratory within 12 h of collection and stored at 4 °C. Temperature, pH, and conductivity were measured in situ using handheld multi-parameter Hanna Instruments HI98121 and HI98311 (RI, USA), respectively. Turbidity was measured in an Oakton T100 instrument (Ill, USA) and total dissolved solids (TDS) were determined by gravimetry at 180 °C following filtration through 0.45 μm pore filters (Isopore™, Merck Millipore). Hardness and total alkalinity were determined by titration with standard solutions of EDTA and H₂SO₄, respectively. Ion chromatography (Thermo ICS 5000, CA, USA) was used for the analysis of Cl⁻, ⁠, , Ca²⁺, K⁺, Na⁺, and Mg²⁺. Concentrations of heavy metals (Cu, Fe, Mn, and Zn) were determined by atomic absorption spectrometry with an air-acetylene flame, and Pb was measured using a graphite furnace with the Zeeman effect (Perkin Elmer Analyst 800, CT, USA). Microbial analyses (total coliform and Escherichia coli) were carried out at the Laboratory of Biotechnology at the Universidad Nacional, Costa Rica, following the multiple tube fermentation technique.

River water samples were characterised using additional parameters, including dissolved oxygen (DO), total solids (TS), biochemical oxygen demand (BOD), ammonium (⁠⁠), and total phosphorus (TP). Hardness, total alkalinity, and microbial analyses were not included in the river water characterisation. DO and temperature were measured in situ using a YSI probe ProODO (OH, USA), while conductivity was measured with a handheld Thermo Orion Star A222 probe (MA, USA). The concentration of TS was determined by gravimetry at 105 °C. BOD was determined using the 5-days incubation test (20 °C in the dark) and the modified Winkler method. and TP were analysed spectrophotometrically in a Thermo Aquamate 2000E (Cambridge, UK) following the indophenol blue and the stannous chloride methods, respectively. Water samples were acid-digested (H₂SO₄/K₂S₂O₈) for phosphorus analysis. All analytical procedures followed the guidelines of the Standard Methods for the Examination of Water and Wastewater (APHA et al. 2012).

Instruments were calibrated using National Institute of Standard and Technology traceable standards. Procedural blanks, recovery quality controls, and calibration curves were carried out in each batch of analysis. Charge-balance error was estimated for each sample; samples with values greater than 10% were not included in the statistical analysis (springs, n = 4). Descriptive statistics of parameters that included values below quantification limits were calculated using robust regression on order statistics (Helsel 2012). Variables were excluded from the statistical analysis if the percentage of non-detects was above 60%. Differences in mean values per season were estimated using a one factor permutation test (Blair et al. 1994; Good 1994). Hierarchical cluster analysis (CA) and principal component analysis (PCA) were applied to identify spatial similarities in the water composition of the springs and the river, and the common parameters that were influencing the observed grouping. The suitability of the dataset for the PCA was tested using the Kaiser–Meyer–Olkin and Barlett's sphericity tests (p < 0.05). Hydrochemical characteristics were estimated using the Piper diagram (Piper 1944), the Gibbs diagram (Gibbs 1970), ionic ratios, and the chloro-alkaline imbalance indices (CAI1 and CAI2) (Schoeller 1972). Statistical analyses were carried out in R 3.5.1 (R Core Team 2018).

The risk assessment of the springs showed that they are in reasonably good condition (Table S1). Springs are in public areas or land owned by the local ASADA, so there are not any access restrictions other than the difficulty due to the hilly-uneven terrain. None of the springs had fencing around the inner protection zones (i.e., 10–20 m radius), and trees were often very close to the spring chambers, increasing the risk of potential damage. Most of the springs had reasonable drainage and the spring chambers were found well-constructed as no leakage was observed and manhole covers were present. However, none of the chambers was adequately ventilated, and two chambers had some sediments, although not enough to cause any pipe blockage. Two spring chambers could not be accessed, and therefore the evaluation was not conducted. It was reported that a spring chamber is manually disconnected from the water distribution system during the wet season due to the increasing level of contaminants caused by runoff. Water quality and quantity (i.e., flow) can be easily monitored at any spring; however, standard monitoring equipment had not been installed at the time of the fieldwork.

The area of the farms within the study area ranged from 0.5 to 10 ha. The land is used for agriculture (79%), livestock (14%), and conservation/reforestation (7%). Most of the agricultural area is covered by coffee plantations, which usually require NPK fertilisers, insecticides, and herbicides. These products are constantly applied throughout the year. With regards to livestock, the area was predominantly characterised by cattle and chickens. River water was reported not to be usually used in these economic activities, but just occasionally to clean tools or dilute pesticides. Most farmers have been trained in agricultural production (80%). The overall production supplies the local market (52%), the agro-industry (31%), international markets (8%), and self-consumption (8%). Most farms (92%) kept an informal riparian protection zone. This is present several metres along the river channel, but it changes with the area of the farm and the slope of the terrain. The perception of the river status was mostly pristine with little alterations (69%) and the main threads identified included wastewater discharge from farms (38%), runoff (23%), solid waste (15%), agrochemicals (8%), and deforestation (8%). Only 23% of the farms have houses on the property, which are generally used during the harvest season. The number of farmers depends on the time of the year, demand, and the farm's capacity, although a minimum of two people is expected to occupy the land permanently throughout the year. All households have septic tanks, and in just one case, it was mentioned that grey water is directly discharged to the ground nearby. None of the respondents have received any complaints regarding pollution events related to their own economic activity. Of the respondents, 92% consider themselves part of the local ASADA.

Despite the differences observed in water composition between the point of use and the source (Figure 2), they were mostly non-significant (p > 0.05). Significant differences were only found for conductivity (p = 0.0084), Cl⁻ (p = 0.0006), and Na⁺ (p = 0.0003). These parameters were higher in the households and they are likely related to the residuals of chlorination. In regard to the spatial and seasonal variability, spatial differences among households were negligible (p > 0.05) and a significant seasonal effect was observed only for pH (p = 0.0011). On the contrary, most of the water quality indicators in the springs exhibited significant spatial differences (p < 0.05); except for pH (p = 0.597), conductivity (p = 0.073), (p = 0.053), and turbidity (p = 0.3). Seasonal differences in the springs were only observed for pH (p < 0.0001), turbidity (p < 0.0001), and hardness (p = 0.0410), which could be related to the impact of runoff in the spring chambers. Nonetheless, the general results suggest that the water composition remains mostly constant throughout the year.

Figure 2 shows a summary of water quality parameters. No statistical comparison between river water and springs and/or households was carried out because the former is not used for drinking purposes. Additional assessed indicators included DO, TS, BOD, and TP (Table S4). Oxygen levels were high with an average of 8.32 mg L⁻¹ (SD = 0.41). TS ranged from 86.0 to 173.0 mg L⁻¹. BOD and were mostly found below the QLs (<2 mg L⁻¹ O₂ and < 0.07 mg L⁻¹ N, respectively). The mean TP was 0.52 mg L⁻¹. The highest concentration of BOD was observed during a particular sampling campaign in the dry season at the river mouth (31.1 mg L⁻¹ O₂), alongside a high concentration of TP (6.13 mg L⁻¹⁠) and other major ions. Spatial differences were obtained for turbidity (p = 0.046), Mg²⁺ (p = 0.025), K⁺ (p = 0.005), Na⁺ (p = 0.046), (p = 0.021), and TP (p = 0.001). In general, a decreasing trend in water quality was observed downstream. On the other hand, the parameters that showed a significant seasonal variation included DO (p = 0.005439), TDS (p = 0.04398), Ca²⁺ (p = 0.03497), Mg²⁺ (p = 0.03994), and Na⁺ (p = 0.04491). The concentrations of these parameters were lower during the wet season, likely due to dilution. Similar to the springs and households, trace metals (Cu, Fe, Mn, Zn, and Pb) were found below the quantification limits.

Hydrogeochemical and multivariate analyses were carried out to better understand the spatial variability in water composition. Mean values were used because seasonal differences were mostly non-significant. At all springs, the abundance of major cations followed: Ca²⁺ > Mg²⁺ > Na⁺ > K⁺. The abundance of major anions was > > > Cl⁻ for 53% of the samples and > > > Cl⁻ for the remaining 47%. The sequence of major ions is typical of groundwater with short residence times (Chebotarev 1955); however, the inclusion of is more likely to reflect the impact of anthropogenic activities. Piper diagram was used to characterise the groundwater type, resulting in Ca²⁺–Mg²⁺– classification (Figure S1). This water composition is mainly influenced by rock weathering processes as indicated by the concentration of TDS (M = 142.7 mg L⁻¹, SD = 16.2) versus the Na⁺/[Na⁺ + Ca²⁺] ratio (M = 0.218, SD = 0.015), or the Cl⁻/[Cl⁻ + ] ratio (M = 0.058, SD = 0.020) (Table S5).

The ion exchange reactions between the groundwater and the underlying material were evaluated using CAI. Negative values are an indication of normal ion exchange where Ca²⁺ and Mg²⁺ are adsorbed on mineral surfaces, releasing Na⁺ and K⁺. On the contrary, positive values indicate the sorption of Na⁺ and K⁺ following the release of Ca²⁺ and Mg²⁺ (reverse ion exchange) (Madrigal-Solís et al. 2022). The negative values obtained suggest normal ion exchange (Figure 3(e)). This process also leads to an increase in the concentration of Na⁺ and/or K⁺, which could explain the high concentration of Na⁺ relative to Cl⁻. The contribution of ion exchange to the hydrochemistry of the springs is derived from the relationship between [Ca²⁺ + Mg²⁺ − − ] and [Na⁺ + K⁺ − Cl⁻] (Figure 3(f)). Samples do not follow the −1 slope, which is usually an indicator of significant cation exchange (Fisher & Mullican 2012). Instead, a slope of −0.494 was observed. This suggests a partial contribution of cation exchange processes influencing water composition. These results, along with those of the Ca²⁺/Mg²⁺ and Na⁺/Cl⁻ ratios, suggest that the sources of cations in the water are predominantly the dissolution of carbonate-bearing minerals with a perhaps minor contribution from silicate weathering followed by normal ion exchange.

Water supplied by the ASADA of Concepción of San Ramón presented high levels of compliance with local and international standards for safe human consumption (Table S2). This is achieved by ensuring adequate infrastructure, operation, and maintenance of the distribution system, and by using water from multiple sources. The infrastructure of the spring chambers was acceptable, but it is necessary to define a protection zone and to install fences to minimise the risk of interference and/or to protect instrumentation if it is installed. The spatial and temporal variations in the water composition in the households were not significant and were smaller than the variations within the springs (Figure 2). Microbial pollution was not present in the households, highlighting the efficiency of the disinfection treatment. However, total coliforms and E. coli were found in the springs, so they alone cannot be considered as a source of safe water. E. coli is an indicator of faecal pollution and, if pathogenic, it can cause illness (Edberg et al. 2000). Similar concentrations of E. coli have been reported in other drinking water distribution systems in both rural and urban settings throughout the country (Sánchez-Gutiérrez et al. 2020b; Barrantes et al. 2022).

Water composition in the springs is mainly driven by rock–water interaction processes during groundwater recharge. However, there were significant differences in the concentration of most parameters. This was partially unexpected because, as mentioned above, they are located in a relatively small area and the infrastructure of the spring chambers was adequate, minimising external factors. Most springs are clustered together in the northeast area of the catchment (Figure 2), but several have shown early signs of pollution from anthropogenic origin. Potential sources include fertilisers and faecal pollution, as demonstrated by the concentrations of and E. coli, respectively. The impact of these was also noted in the decreasing water quality of the river where the concentrations of TP and most cations increased downstream; in particular, K⁺. Loss of both TP and K⁺ from soils into surface waters can occur, via leaching and runoff, in areas of intensive agriculture where excessive amounts of fertilisers are used (Skowron et al. 2018; Liu et al. 2021). This coincides with the current land use as coffee plantations and livestock are abundant in the catchment (Section 3.1), but additional approaches are needed to quantify the impact of the potential sources on water quality. The environmental conditions in the Cañuela River catchment appear to be representative of other areas where CBWMOs operate in Costa Rica. Similar aquifer structures, recharge mechanisms, and environmental pressures have been reported in the Central Valley, Central Pacific and Northern Pacific regions (Madrigal-Solis et al. 2020; Sánchez-Gutiérrez et al. 2020a, b). Additional local pressures also exist; for instance, in more urbanised catchments where high levels of have been also related to the lack of centralised wastewater treatment facilities rather than only to agricultural practices (Sanchez-Gutierrez et al. 2023).

The general water quality information is of interest to the CBMWO not only for regulatory and operational purposes but also for future planning and management. For instance, household data indicated that the CBMWO provides safe managed drinking water. A comparison of the water quality of the springs to drinking water guidelines highlights the need for disinfection treatment. In its role as provider of water services, the CBMWO is then expected to primarily undertake the necessary steps to ensure satisfactory operation of the disinfection system. This type of recommendation can arise from the routine monitoring that is provided by the government; usuallyevery three years (for a similar number of parameters as presented here). However, a more holistic interpretation of the water quality data revealed insights into the complex mechanisms that are driving the water composition of the springs, and the potential sources of pollution. In this regard, the assessment of river water quality was instrumental to identifying anthropogenic pressures. This suggests that CBWMOs should also assume a role that is more orientated towards a broader management of water resources and conservation. In this case, for instance, to reduce the impact of agriculture on groundwater. It is important, however, to consider the legal framework that regulates the scope of the CBWMOs. This is because a multi-sector participatory approach involving the state, local governments, other CBWMOs, stakeholders, farmers, industry, and academia would eventually be needed to address current environmental challenges. Having provided evidence of the need to protect water sources in the long term, the involvement of CBWMOs in the planning and development of such initiatives should be strongly considered.

This study has demonstrated how integrated water quality monitoring is necessary to effectively evaluate the performance of CBWMO to make informed management decisions. In the case of the ASADA of Concepción of San Ramon, Costa Rica water quality data showed that (a) safely managed drinking water is provided to the local population, (b) the main composition of spring water derives from rock–water interaction processes during groundwater recharge, and (c) early signs of pollution from anthropogenic activities, mainly agriculture, are likely impacting the water quality of the springs and the river. It is important to note that the monitoring in this study is limited to two campaigns per season over one hydrological year. Water quality can change significantly at different spatial and temporal scales; especially during periods of heavy rainfall or drought. Such changes are expected to become more notorious in the river than in the springs or in households. Although the characteristics of the ASADA of Concepción are representative of other CBWMOs in the country, we recognise that the capacity to carry out monitoring of water quality is still limited for some CBWMOs; in particular, at adequate temporal and spatial resolutions. Additional topics related to water quality and risk assessment that should also be considered in the context of community-based management include the perception of the community about the water services provided by the CBWMOs, continuity of water service (e.g., shortages or interruptions), the effective communication of water quality data, and the proactive use of this type of information to support water safety plans. Nevertheless, we emphasise the importance of environmental data in supporting evidence-based decision-making regarding water resource management.

The authors gratefully acknowledge the collaboration of the ASADA of Concepción of San Ramón, Costa Rica. We thank Cristina Benavides-Benavides and Viviana Salgado-Silva for their contribution to the sampling campaigns and project design. We also thank undergraduate students who participated in fieldwork campaigns. This project was funded by the Research Office of the Universidad Nacional, Costa Rica, under grant SIA 0129-14. T.H.A.S. is grateful for financial support from the Faculty of Exact and Natural Sciences of the Universidad Nacional, Costa Rica.

M.L.-E. conceptualised the whole article, developed the methodology, rendered support in formal analysis, investigated and visualised the whole process, and wrote the original draft. L.M.-R. conceptualised the whole article, developed the methodology, rendered support in formal analysis, investigated and visualised the whole process, and wrote the original draft. R.S.-G. rendered support in formal analysis and wrote the review the edited the article. I.V.-G. conceptualised the whole article and wrote the review the edited the article. T.H.A.S. conceptualised the whole article, developed the methodology, investigated the whole process, and wrote the review the edited the article.

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

The authors declare there is no conflict.