Application of the Rural Water Sustainability Index (RWSI) in Amazon rural communities, Pará, Brazil Open Access

Having access to safe drinking water is a fundamental human right of any individual, regardless of birthplace, belief, gender, color, or social class (Li & Wu, 2019). Thus, ensuring an appropriate water supply is essential for the quality of life and human well-being (Pichel et al., 2019). However, the lack of access to potable water has been a concern, especially in poor regions deprived of water supply services, which are also characterized by unfavorable conditions for human life (Marin & Burgel, 2020). There is an agreement among the literature about the need to reduce the sanitary exclusion of minority classes, composed of the population from rural areas and inhabitants of deprived areas of cities (Aleixo et al., 2016).

Although the sustainable development goal (SDG) 6 (water and sanitation) aims to achieve universal and equitable access to safe and accessible drinking water, sanitation, and hygiene by 2030 (UN-Water, WHO, 2017), achieving this goal will be a major challenge. There are still approximately 844 million people living with no access to safe drinking water, particularly those living in rural areas of developing countries (Gomez et al., 2019). In addition, in 2017, more than 2.3 billion people still lacked access to improved sanitation facilities (Bayu et al., 2020).

Brazil holds 12% of the planet's total freshwater (equivalent to 1,488 million m³/s) and 53% of the total freshwater in South America (334,000 m³/s). However, there are problems relating to the spatial distribution of water resources, lack of water quality, and limited access to water sources (Bordalo, 2016). In addition, the coverage of water supply services in rural areas of Brazil is insufficient, which causes a low quality of life, health, and well-being for the inhabitants of these locations, as well as contributing to the spread of diseases by the absence of clean water or inadequate sanitation (Ferreira et al., 2021).

According to information from the 2017 National Sanitation Information System (SNIS), 16.6% of the rural population in Brazil lack access to water, and 47.6% live without a sewage collection system (Brazil, 2019). Furthermore, the National Household Sample Survey (PNAD) highlights that only 33.41% of rural households have a water supply connected to the public network, while 66.59% of the supply comes from other sources such as wells, rivers, and reservoirs (Brazil, 2014).

Pichel et al. (2019) pointed out that to achieve universal access to safe drinking water, it is necessary to ensure the appropriate supply of water, sanitation, and hygiene, due to the various implications that such conditions have on public health, food security, poverty alleviation, and equality. However, proper access to potable water is a goal that is still far from being reached in several places around the world, for example, in the Brazilian Amazon.

Although the Brazilian Amazon region has one of the largest reserves of fresh water on the planet, thousands of inhabitants experience difficulties accessing drinking water and traditional technologies in water treatment and distribution (Monni et al., 2018). For example, in 2015, information from the SNIS showed that approximately 43.1% of the inhabitants of the North region did not have access to treated water (Brazil, 2017). There are communities located in the Amazon that have difficulties accessing drinking water and supply, making it a problem in the region. The fact that the region is rich in water resources does not guarantee that water will meet the basic needs of its population (Monni et al., 2018; Brito et al., 2020).

According to Brito et al. (2020), the lack of drinking water is more aggravating in geographically distant communities. In this context, it is essential to study and reflect on the use and access to water in the Amazon region, to include a concept of qualitative and economic scarcity given the panorama of the difficulty in accessing drinking water faced by the rural population and the islands.

Understanding the lack of availability (physical, infrastructure, or institutional), the qualitative and economic scarcity of water is necessary for the formulation of public policies at national, state, and local scales (Narzetti & Marques, 2021). In recent years, the commitment of several researchers to measure the sustainability of water in the Brazilian Amazon region has been observed (Rego et al., 2013; Brito et al., 2020; Rocha & Lima, 2020) through the development of instruments for assessment based on sustainability indicators.

Previous studies in the Amazon region of Brazil have used multi-criteria decision analysis (MCDA) techniques to assess water sustainability at different spatial scales, such as watershed (Rocha & Lima, 2020), city (Rego et al., 2013), and island (Brito et al., 2020). However, the studies cited did not integrate the MCDA and geographic information system (GIS) to analyze the water situation for a spatial scale of rural communities. Thus, the inexistence or lack of indexes representing rural communities’ scenarios makes it difficult to implement integrated water resources management for this spatiality (Guppy, 2014).

According to Crispim et al. (2021), the application of MCDA and GIS in an integrated manner for the analysis of water resources in rural communities is a relevant tool to assist decision-makers in water management and direct the priority of actions for a given location. Moreover, GIS allows the observation and simultaneous handling of data from different components to compose an index.

In addition to the complexity that comprises the theme and scientific knowledge, water access policies must consider the position of local users concerning water sources (Mahama et al., 2014). This ‘internal’ understanding can collaborate in the selection of appropriate components and strategies that enable access to good quality water. Thus, few studies in the literature use indices that consider the opinion of users to support the decision of water managers with technical expertise (Crispim et al., 2021).

The purpose of this article is to propose the Rural Water Sustainability Index (RWSI), which is a model that includes significant components and indicators for rural communities. The model also considers the participatory and multi-criteria index based on the Delphi method. It is noteworthy that a participatory and multi-criteria approach is pertinent to the inclusion of local social actors. In this context, the rural poor who are excluded from conducting and structuring public policies for water management stand out. Since there is an appreciation of specialized knowledge, the creation of an inclusive management tool for rural communities involves the inclusion of actors in social, cultural, and traditional knowledge spheres. The method, therefore, would assist in water management in stressed communities. Rural communities were the focus of the study due to the various difficulties faced by these rural populations with regard to attempts to access potable water for consumption and agriculture. The RWSI aims to build a holistic water management tool that supports more efficient water management. Using this tool, managers can identify and prioritize locations that require state intervention to develop strategies and guarantee water for communities. The method was applied and validated in 16 rural communities located in the Santa Luzia do Pará, state of Pará, northern region of Brazil.

The RWSI was developed based on the framework of the Water Poverty Index (WPI) suggested by Lawrence & Meigh (2003). The five components used by the original WPI were kept, as they represent the local socioeconomic characteristics and the problems inherent to water management in rural communities, which usually need improvement (Guppy, 2014). In this way, the RWSI method was built around five important components for rural communities. The indicators comprise social, economic, and environmental data related to water scarcity, use, and access to water. Furthermore, these five components describe the main challenges faced by low-income locations linked to water resources (Crispim et al., 2021).

For the suggested RWSI for rural communities in the Amazon, specific adaptations were made to the methodological procedures and small adjustments to the indicators, as local particularities were considered. For example, the structuring of subcomponents and adequate indicators to identify areas that need improvement and identify the use of and access to water in rural communities of the Amazon. In addition, it was necessary to collect primary data on family units in rural communities due to the lack of information for this spatial scale.

Components, subcomponents, and indicators are the main constituents of the RWSI. They were selected by a literature review of sustainability component structures and groups, as well as existing indicators (Juwana et al., 2012; Sun et al., 2018). The literature review provided the components, subcomponents, and indicators recently used in research in rural communities. The selection of components considered their suitability for the spatial scale (studies in rural areas), clarity, simplicity, integrative, comparative particularity, robustness, sensitivity, and understandability of the local and socioeconomic characteristics. This previous set was then improved through conversation with the parties involved (professionals, technicians, or specialists).

The final values of the components and subcomponents were defined in a participative manner by the Delphi method. The RWSI is comprised of 5 components, 21 subcomponents, and 58 indicators (Table 1), selected through participatory processes and consultation with stakeholders (politicians, researchers, and technicians) regarding the capacity of people to manage water, the availability and how water resources are managed, water use, access to water, and the environment (Crispim et al., 2021). Each component contains a few subcomponents and variables, which can eventually be measured, obtained, or measured.

The Delphi method is based on the systematic use of understanding, experience, and practice. The peculiarity of a group of experts (also identified in the literature as specialists, experts, participants, respondents, or panelists) considers that collective thinking is better than the opinion or conception of a single professional when it is properly organized (Santiago & Dias, 2012). In this study, the Delphi method was used to determine the weights of each component and subcomponent, and the scores of the RWSI indicators, as it has been used more frequently for the development of scenarios to increase the quality of scenarios and make them deeper (Brito et al., 2020; Progênio et al. 2020; Crispim et al., 2021).

The participatory approach facilitates the contribution of experts in different contexts, as this is a simple-to-use approach. This method also allowed the integration of technical and participative components. In the latter method, specialists and technicians were requested to assign a numeric score (note) to previously selected indicators, according to their importance in a specific scenario (Gourbesville, 2008). Moreover, this method provided the participation and sharing of responsibilities between specialists and technicians from different regions of Brazil (semi-arid and Amazon), areas of knowledge, and different experiences on the subject. In doing so, subjective issues are mitigated in the process, as more than one specialist was used to construct the method (Crispim et al., 2021). Finally, a structured questionnaire was prepared based on components, subcomponents, and indicators (Table 1). Figure 1 pathways for the method application.

To choose the Delphi method's participants, a list of likely experts was drawn up based on information obtained from the Lattes Platform in the directory of research groups in Brazil. Subsequently, a query was carried out in the Lattes curriculum of the possible participants to see if there were academic productions on the topic addressed in this study. In this context, specialists were selected based on their work and experiences on the local region and study topic. The identification of specialists took place through an initial checklist, which considered their qualifications, academic background, publications, and participation in activities and research on water resources management in rural communities (Table 2).

Pertinent information, such as the research objectives and a preliminary matrix, which contained components, subcomponents, as well as indicators that make up the RWSI, was sent electronically to the participants.

The initial questionnaire was sent to 34 specialists, from which 24 responses were obtained, with a return rate of 70.5%. It is superior to other studies using the Delphi method, as the rate varies between 30 and 50% in the first round. Experts were instructed to select questions relevant to the study regarding water resources, access, capacity, use, and environment. In addition, participants were asked to assign grades to the indicator variables, which ranged from 0 to 10, with 0 being the lowest value and 10 being the highest value for a given variable. The questionnaire responses were tabulated and analyzed using an electronic spreadsheet. Sequentially, the data were analyzed for the coefficient of agreement (CC) and the stability of the answers. The first consultation round was paramount to select and exclude RWSI's components, subcomponents, and indicators.

Table 3 displays the results of the coefficient of agreement (CC) components and subcomponents of the first consultation process referent in the RWSI framework. The results indicated that only the water use component (U) reached the coefficient of agreement (CC) (≥50%). In addition, only 23.81% (n = 5) of the subcomponents obeyed the coefficient of agreement (CC) (≥ 50%), needing a new round the consultation.

An expected scenario in this type of research is the decrease in participants between one round of questionnaire applications. Experts from the social, economic, and accounting areas did not participate in the second round, although they were invited to participate in the first. In the second round, the questionnaire was sent to the 24 specialists who participated in the first round. Twenty-two participants completed questionnaires, which indicated a response rate of 91.6%. Although the number of specialists decreased between the two rounds of the questionnaires, the final group of specialists remained heterogeneous in academic training, geographic spatiality, and balanced experience (Table 4).

Table 5 shows the results of the coefficient of agreement (CC) components and subcomponents of the second consultation process referent in the RWSI framework. The results display that in all weighted components, the coefficient of agreement (CC) (≥ 50%), with detaching for water-use component (U) that reached 86.4% (CC). In addition, in all the subcomponents’ the coefficient of agreement (CC) (≥ 50%), without the need for additional Delphi consultation process.

Although Table 1 presents qualitative and quantitative indicators, it was necessary to create an ordinal scale to convert qualitative measures into quantitative ones to normalize the values included in the method. The data are represented on different scales (e.g., reuse of water ‘yes’ or ‘no’, duration of collection, waiting and water transportation ‘minutes’). For each answer, a numerical value was assigned based on the predefined scale. All measurement units were transformed to a standard 0–10 scale, where 0 indicates the worst condition and 10 denotes the best situation. This proposed method has been utilized and validated in several similar works (Brito et al., 2020; Crispim et al., 2021). Thus, a scoring matrix was elaborated through consultation with experts from several disciplines (agricultural engineering, agronomic engineering, civil engineering, environmental engineering, geography, and sanitary engineering).

Interpreting the index is fundamental to understanding the logic of the components and values of the index (Crispim et al., 2021). For the RWSI, the interpretation for components and their aggregated index is based on the quartile scale (Table 6), creating four performance ranges: Poor (0.0–5.8), Poor-Medium (5.81–6.0), Medium-Good (6.01–6.30), and Good (6.31–10.0). The quartile classes were determined based on values of the RWSI and according to the method used by Juwana et al. (2012) and Crispim et al. (2021). Levels of performance were incorporated into the maps that illustrate the condition of the components and the index to a specific time of assessment and would be used as the basis for relevant ‘Priority of Action’ to improve the water sustainability at the rural communities’ scales.

MS-Excel 2016, BioEstat version 5.3, and Program (R) version 3.6.1 were used to perform the statistical analysis. For elaborating the maps of the RWSI and components, the GIS software was used with the inverse distance weighting interpolation method.

The proposed method was applied in 16 rural communities in Santa Luzia do Pará, the northern region of Brazil (Figure 2). The municipality of Santa Luzia do Pará is in the Mesoregion of Nordeste Paraense and the microregion of Guamá, about 173.7 km from Belém, capital of the state (De Farias Neto et al., 2013). The Brazilian Institute of Geography and Statistics – IBGE (2010) estimates that the total population of the municipality in 2010 was 19,424 inhabitants, with 8,693 inhabitants in the urban area and 10,731 in the rural area, with a demographic density of 14.32 (inhabitants/km²) and a territorial extension of 1,356,124 km².

According to the classification by Köppen (1936), Santa Luzia do Pará has a tropical monsoon climate, with an average temperature of 28 °C. The municipality has two well-defined seasons, a rainy season and a less rainy season (De Farias Neto et al., 2013). The average annual precipitation in the municipality's territory reaches 1,750–3,000 (mm), for 30 years (De Andrade et al., 2017).

This municipality is in the Hydrographic Region of the Atlantic Coast – Northeast. Part of its territory is within two relevant hydrographic basins in the State of Pará, namely the Guamá River (Rocha & Lima, 2020) and the Caeté River (Dias & Cirilo, 2018). The rivers Caeté, Guamá, Peritoró, and Sujo make up the drainage network present in the municipality's territory.

The income of the inhabitants of rural communities in Santa Luzia do Pará comes from activities related to family farming, through various forms of production, such as the exploitation of the soil for crops, as well as plant extraction, agricultural activities, and social programs family grant and pensions rural (IBGE, 2010).

The primary dataset was found for each rural community, and household surveys were carried out among the local population residing in the study area. The questionnaire used was closed and structured with questions related to people's ability to obtain and manage water, water sources, domestic consumption, access to drinking water; water management conditions were the delimiters used to collect information. Surveys were carried out from March to April 2021, covering 16 rural communities.

The principle of randomness was used to select individuals who participated in the research. ​In this way, residences in the central, extreme, and more dispersed parts of rural communities were visited. Each respondent was required to answer a series of questions based on the indicators listed in Table 1.

In total, 380 interviews were conducted in 16 rural communities in proportion to the population domiciled in each community. The family unit was the source of information for data acquisition for the study. In each selected household, an adult member of the family (male or female) aged 18 or over, preferably the person responsible for managing water in the household, participated in the survey. However, in the absence of the water manager, the available adult member participated in the interview. The group of people interviewed comprised community health agents (ACS – Agentes comunitários de Saúde, in Portuguese), farmers, merchants, students, landless settlement leaders, rural association presidents, Quilombola community leaders, cooperative leaders, politicians, religious leaders, and teachers; all participants either worked or resided in the areas.

Sensitivity analysis helps to improve the accuracy and understanding of the final results, thus reducing the risks of elaborating a meaningless tool (Juwana et al., 2012). A sensitivity analysis was conducted to test the robustness of the RWSI and check redundancies among components. To do so, a correlation matrix and simple linear regression were employed to verify the relationships between the independent variables (components) with the dependent variable (RWSI). The following precision criteria were applied: correlation coefficient (r), the determination coefficient (r²), and standard error of the estimate (Se).

Figure 3 shows the rural water sustainability scores for each community in the study area. The map exhibits that rural communities had different scores in the water sustainability index. Therefore, rural communities in the weakest levels and with difficult or limited access to water resources must be prioritized on government actions toward water management initiatives. Decision-makers can identify which communities are under greater water stress at the household levels and allocate resources to mitigate negative conditions. In addition, the comparison of these indices for the 16 rural communities can be employed as a starting point by the municipal government of Santa Luzia do Pará-PA for improving water management planning and policymaking.

Predictably, rural communities distant to the Caeté, Grande, Guamá, and Peritoró rivers showed poorer results than the communities that are close to these sources of surface water, e.g., Areia Branca (P01), Cantã (P03), Mucurateua (P06), and Piracema (P08); similar to results obtained by Crispim et al. (2021) in a case study in rural communities in the municipality of Pombal, state of Paraíba, northeast of Brazil.

Rural communities close to rivers still depended on groundwater supply. Broca (P02), Cantã (P03), and São João do Caeté (P12) scored poorly or medium in the sustainability index. RWSI analysis showed that Broca (P02), Cantã (P03), and São João do Caeté (P12) communities were unable to catch water from perennial sources due to the absence of water infrastructure. Even though most of these communities are close to rivers, the majority of the inhabitants depend on dug wells or drilled wells for water supply. However, in the study carried out by Guimarães et al. (2009) in rural communities in the municipality of Santa Luzia do Pará, it was observed that water from groundwater wells is on-suitable for human consumption. Although Brazilian Amazon rainfall is approximately 2,095 mm/year, with potential for capture and utilization (Almeida et al., 2017), there exists a low use of rainwater to supply human water in the study area. For Crispim et al. (2021), the existence of rainwater harvesting systems decreases the total time spent with water collection, waiting, and water transportation to the household as well as the number of times during the day to collect water.

RWSI application at the 16 selected areas in Santa Luzia do Pará demonstrated that the index was able to verify similarities and dissimilarities between the rural communities based on the overall situation of water resources (final index values). The mode also identified issues linked with water condition (components values). The map shows diversified spatial scenery between rural communities of Santa Luzia do Pará, enabling policy planners to quickly verify the communities in need of priority actions and efforts to maximize water sustainability.

Figure 4 shows the individual RWSI component's performance in each rural community. These values enable the observation of local strengths and weaknesses of the municipality's water situation as well as the comparison between different regions. Moreover, when water management decisions require prioritizing actions, an essential factor is identifying the most vulnerable community. We observed that the capacity, use, and environment components presented the lowest scores. In contrast, water resources and water access components presented the highest scores.

The low score related to ‘capacity’ is associated with the lack of capacity of people to manage water for human consumption. Other factors include socioeconomic issues, institutional aspects, and difficulty accessing drinking water in residences. Similarly, the water ‘use’ component also scored poorly as most of these rural communities lack water to irrigate crops and supply the livestock demand. Therefore, available water volumes do not allow multiple uses. Finally, for the ‘environment’ component, the model identified problems concerning the use of chemical fertilizers and solid waste disposal that can affect the area's ecological integrity.

The results in the component ‘Water resources’ demonstrate high values in rural communities and were classified as ‘good’ (except for Areia Branca). Furthermore, this component incorporates two subcomponents of debatable reliability when analyzed at the family unit: water source and quality. The studied population assessed the water quality based on characteristics such as color, odor, and taste. However, the analysis did not take into consideration laboratory analysis to identify physical–chemical characteristics or bacteriological contamination. Thus, this component might fail to show local situations of water resources accurately.

The final score of the RWSI for the 16 communities and the score for each component are shown below (see Table 8). A total of 380 interviews were conducted to reach RWSI performance. The water sustainability in rural communities of Santa Luzia do Pará-PA differ depending on the components verified. The final RWSI scores demonstrate that about 25.0% of the rural communities were classified under ‘poor water sustainability’ condition; 25.0% scored as ‘poor-medium sustainability’; 37.5% were obtained ‘medium-good condition’ results; while only 12.5% fell under the ‘good water sustainability’ classification.

Each community presented various strengths and weaknesses concerning their water sustainability needs. For example, ‘capacity’ was the highest component for Areia Branca (6.5) but the lowest for the communities of Cantã and Vila Caeté (4.6), respectively. On average, capacity was the lowest component among all indicators, showing that the institutional framework to support the population's ability to obtain and manage water is far from adequate. In addition, the results indicate that investments are needed to improve water sustainability in these communities. On the other hand, the ‘water resources’ component obtained higher values than the other indicators due to good results in the ‘quality’, ‘water sources’, and ‘water resource management’ subcomponents. The results are similar to Crispim et al. (2021) in a case study in Pombal-PB, Brazil.

To validate the RWSI, a sensitivity analysis was performed to test the model's robustness. A multivariate analysis was applied to verify redundancies between the RWSI components. The results in Table 9 confirmed that the selected RWSI components did not have a very strong or strong positive correlation, showing that they are a good set of components for the multi-criteria analysis. This analysis improved the accuracy and understanding of the results, reducing the risks of developing a meaningless tool. The lack of correlation is an expected characteristic (Giné Garriga & Pérez Foguet, 2010; Crispim et al., 2021), showing that each indicator is a statistically distinct component in the data.

As seen in Table 9, there is a moderate positive correlation between ‘Water Resources’ and ‘Environment’ (r = 0.59) because of their natural correlation. Still, less than 0.70, ensuring no significant redundancies or double counting and avoiding their elimination. They observed that the capacity component has a weak negative relationship with water resources (r = −0.31), access (−0.06), and the environment (r = −0.18), not being statistically significant at a 95% confidence level.

Table 10 brings the results of the simple linear regression and the precision criteria. They indicated the RWSI has a strong positive relationship with access (r = 0.71), which is statistically significant at a 95% confidence level. If the access to water increases in the study area, the RWSI also increases. Therefore, the RWSI has a moderately positive relationship with water resources (r = 0.61) and the environment (r = 0.63).

The coefficients of determination (r²) of the components ranged from r² = 0.05 for ‘capacity’ to r² = 0.50 for ‘access’. The water access component had the greatest impact on the suggested RWSI, with a 50.0% influence. Other components only impacted 50.0% of the RWSI variation index. The standard error showed low values for the components access (0.19), water resources, and environment (0.21), respectively.

Ward's hierarchical agglomerative clustering method was used to analyze the similarity between the 16 rural communities in Santa Luzia do Pará. The method considered the water situation using the values found for the five components of the RWSI (capacity, water resources, use, access, and environment). The dendrogram, Figure 5, illustrates the overall clustering of 16 rural communities. The Euclidean distance (dissimilarity measure) was utilized to determine differences between elements (e.g., rural communities). The cut made on the axis of the distance of the dendrogram was at the height of 1.6, showing the formation of five probable clusters (Figure 5). It is noteworthy that this decision was subjective, and the researcher's discretion was used to define the best distance for the cut (Brito et al., 2020; Crispim et al., 2021). This choice aims to classify groups for rural communities and enable adequate local water planning.

We illustrate a spider diagram in Figure 6 to summarize the differences in the means between the five clusters, presented in Table 10. The results aim to assist the understanding of the specificities of the five groups. Such knowledge allows policy planners to identify priority groups, define particular intervention strategies, and direct resources for the water urgencies in priority areas. Among the five clusters, the RWSI is the most significant for cluster 5 and least significant for clusters 1 and 2, respectively. Cluster 5 has obtained the highest values on the components ‘water resources’, ‘use’, and ‘environment’. The ‘capacity’ component is the highest for cluster 1, while the ‘access’ component is the highest for cluster 4 (Table 11).

Table 11 reveals that the first cluster comprises three rural communities, Broca (the largest) and two medium-size rural communities (Areia Branca and Tamancuoca). These communities scored a ‘Poor-Medium’ performance in the RWSI (RWSI: 5.9) because those areas have no water supply system. Alternative water sources are used in such cases, such as artesian wells or dug wells. Therefore, the first action would be to install the water supply systems across those regions. The communities in cluster 2 also presented a ‘Poor-Medium’ (RWSI: 5.9) score.

Similarly to cluster 1, the population in those areas uses the same water sources (dug wells, drilled wells, and rivers) without applying an adequate water treatment. The cluster 3 groups rural communities with ‘Poor-Medium’ (RWSI: 6.0) performance. In addition, 66.6% (n = 2) of rural communities in this group lack access to a water supply system, except for Fuzil. However, these two communities present good water levels. Clusters 2 and 3, on the other hand, are constituted of rural communities dissipated within the territory of Santa Luzia do Pará. This situation presents a low capacity to manage water for human consumption, with RWSI's scores classified as poor (5.2) and (5.0), respectively. Therefore, proper attention needs to be given to the population's capacity to manage water and change cultural behaviors.

Cluster 4 comprises rural communities near the municipality's area, except Pitoró. The latter community is farther away but presents a good performance in the RWSI (6.3) model and high scores in the access component (7.4). The results indicate that most of the population of this area has piped water in residences. In addition, the rural communities of this group have access to public services such as schools, healthcare facilities, and comprise the area with the largest number of families. Lastly, cluster 5 performs substantially higher in the RWSI (6.4), presenting the best situation considering the water resources components relating to the quality of source and quantity of water. In the ‘use’ component, the population in this area uses water for multiple activities. Furthermore, the quantity of water meets demands (domestic consumption and livestock demand). As for the ‘environment’ component, the rural communities in this group prioritize the local environmental integrity of the water as well as ecosystem goods and services from aquatic habitats in the region. However, most rural communities have low political participation through the rural association.

The adaptation of the RWSI in Amazon rural communities to evaluate water sustainability locally was a valid application for calculating the components and index. The tool served to identify areas with water stress and assist policymakers in verifying the weaknesses and strengths of components requiring more immediate attention. However, future studies must consider the results obtained through sensitivity analysis to determine components’ weights of the RWSI.

The RWSI was able to verify similarities and differences between the locations as well as identify places that need high priority to enhance the management of water resources at a local scale. Policymakers can then allocate resources to improve water sustainability. In addition, the method is easy to operate and needs minor adjustments in the indicators concerning socioeconomic and culture related to water management for replication in other places of the Amazon region.

The RWSI in 16 rural communities in Santa Luzia do Pará-PA ranges from 5.7 to 6.5, with the lowest value for Areia Branca and Cantã, and the highest value for Quilombola Jacarequara and Tentugal. These locations fall into the ‘poor’ and ‘poor-medium’ water categories. This result indicates a water stress scenario, helping policymakers evaluate difficulties to the water sector. Finally, to make the RWSI highly useful for the municipal Government of Santa Luzia do Pará-PA, the comparison of water sustainability issues across rural communities needs to be extended to all municipality regions, which will be undertaken in a future study.

The proposed methodology can be replicated in other developing countries and territories inhabited by vulnerable communities with social, economic, and participation needs. For that, there need to be adaptations to the subcomponents and indicators that represent local context because countries may have different ways of assessing the situation of water. However, it is worth mentioning that using other participatory methods (Analytic Hierarchy Process, Data Envelopment Analysis, and the Benefit of Doubt or statistical-based methods) to determine or adjust weights and grades can change the result's the RWSI.

The authors would also like to thank the Coordination for the Improvement of Higher Education Personnel – Brasil (CAPES) – Finance Code 001, for sponsoring the first author of this paper. This research would not have been possible without the funding provided.

The authors declare none.

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