1. Introduction

Globally, communities and households use water from multiple sources for multiple purposes. For example, in Pakistan, community tanks serve dual purposes, providing water for both irrigation and household domestic needs [1]. Similarly, in Senegal, networked domestic water systems are used not only for household consumption but also to irrigate home gardens and water livestock [2]. To facilitate multiple uses and circumvent insufficiencies with primary water sources, rural and peri-urban communities have developed multiple-use water systems (MUS). MUS is an approach to water service delivery at the community level and reflects the many ways that water is used for domestic and productive livelihood activities [3–6]. Communities design MUS systems to respond to water needs and uses that often lie outside the state or public model for water resource management, particularly because uses transcend sectoral distinctions of domestic and productive. Households also diversify water sources intended for single uses to a range of other productive and reproductive uses to sustain their livelihoods. In this way, communities and households use assorted water supply sources and technologies (e.g., handpumps, boreholes, rainwater systems, public taps, cattle troughs, irrigation equipment) to customize their water delivery services to match their livelihood needs.

Research on MUS has often focused on community-level infrastructure such as boreholes, wells, and canals, emphasizing how single systems support multiple domestic and productive uses. While informative, this focus underplays the broader reality that households routinely rely on multiple water sources, including supplemental and sometimes unimproved supplies, when primary systems fall short. Restricting analysis to designated “main” sources risks overstating safe water access and understating household water insecurity [7,8]. In practice, more than 90% of households in sub-Saharan Africa use at least two water sources, shifting between improved and unimproved options based on cost, taste, reliability, and proximity [7,9]. Yet global monitoring frameworks, and much MUS-oriented analysis, rarely capture this source-switching behavior, potentially overstating safe water access and understating household water insecurity.

It is important to distinguish between multiple-use water services (MUS) and multiple water source use. MUS refers to the use of a single water source or system to meet both domestic and productive needs, either by design or through adaptation. In contrast, multiple water source use describes households’ reliance on different water sources for different purposes or across seasons, often as a coping strategy in response to variability in availability, quality, or cost. While these practices may overlap in practice, they represent distinct analytical concepts. Multiple water source use reflects flexibility in household water portfolios, whereas MUS focuses on how service delivery arrangements enable or constrain the integration of domestic and productive water uses within a single system.

Seasonal switching is therefore not synonymous with MUS. Households may switch between multiple water sources without engaging in MUS, and conversely, MUS may be practiced either year-round or within a single season depending on the reliability of a given source. In this study, seasonal MUS refers specifically to cases where households rely on the same water source for both domestic and productive uses within a defined seasonal period rather than to general diversification or switching between sources across seasons. Assessments focused solely on service provision therefore risk misrepresenting households’ lived access to safely managed water. Addressing this gap requires approaches that move beyond infrastructure to capture experiential dimensions of water insecurity. The Household Water Insecurity Experiences (HWISE) scale offers one such framework, measuring psychosocial and experiential aspects of water access, reliability, and safety across multiple domains.

Seasonality complicates household water use and underscores the limits of static metrics. Nearly 30% of households in sub-Saharan Africa switch their primary source between seasons [8], demonstrating that water use is dynamic rather than fixed. In the rainy season, households expand their water access portfolios to include rainwater harvesting, ponds, and seasonal streams, reducing collection costs but heightening the risks of microbial contamination; water collected in these months is more than twice as likely to contain E. coli compared to dry season months [10,11]. As surface sources diminish in the dry season, households rely more on groundwater and boreholes, which lowers exposure to unsafe water but increases time and financial burdens through longer travel and queues. These seasonal shifts highlight the importance of incorporating experiential measures to better capture vulnerabilities and inform more resilient water policies.

This study addresses critical gaps in two parallel bodies of literature on multiple-use water systems (MUS) and household water security. Research on MUS has largely emphasized technical design, productivity, and economic outcomes [2,4,6], with limited attention to how MUS practices are associated with the lived experiences of household water insecurity. Conversely, scholarship on household water insecurity has focused on experiential measures, such as the HWISE scale [12,13] without examining how specific water service delivery models like MUS shape these experiences. We bring these perspectives together through a household survey of 1,082 rural households in Mali, applying the HWISE scale and four subdomains (worry, hygiene, access, interruption) to examine whether households undertaking MUS report systematically different levels of water insecurity. Importantly, we differentiate between year-round and seasonal MUS practices.

This paper makes two contributions. First, it provides one of the first empirical examinations of how different MUS practices are associated with household experiences of water insecurity. Second, it shows that MUS is not uniformly associated with lower insecurity. While households practicing year-round MUS report significantly lower levels of water insecurity, households engaging in seasonal MUS report higher insecurity across multiple domains, suggesting that seasonal MUS may coincide with, rather than alleviate, underlying vulnerability and may function as an adaptive response to water insecurity rather than a protective intervention.

The remainder of the paper is organized as follows. Section 2 reviews the background literature on MUS and household water insecurity. Section 3 describes the survey, variable construction, and econometric analysis. Section 4 presents results, highlighting how seasonality shapes household water insecurity outcomes. Section 5 discusses these findings, reframing MUS as a conditional household strategy rather than an inherently beneficial model and draws implications for policy and practice.

2. Background

MUS is commonly organized into three forms: multiple use services by design, irrigation-plus, and domestic-plus systems [2,4–6]. In MUS-by-design systems, community members collectively plan services around diverse water needs. Irrigation-plus systems extend water intended for agriculture to other uses, while domestic-plus systems channel water sources developed for household consumption into productive activities such as home garden or livestock watering. These approaches highlight how households and communities adapt available water infrastructure to meet multiple demands.

International development practitioners and local water managers advocate for MUS approaches in community development programs because of both perceived and documented benefits of these systems [6]. Practitioners argue that a MUS approach will improve water delivery to communities in low- and middle-income countries, providing cascading livelihood benefits to households [14,15], like improved agricultural productivity and income. (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Expected and Realized Benefits of MUS (adapted from James 2003; Guba et al., 2023).

https://doi.org/10.1371/journal.pwat.0000600.g001

Research on MUS systems shows that this approach to water use provides direct household benefits through improved access to clean water that supports health, hygiene, and nutrition [5,16–18]. MUS can also reduce women’s labor burdens by reducing their time spent fetching water and thereby creating increased opportunities for diversified economic activities [2,5,17,19,20]. For example, in Senegal, piped systems designed for domestic use are also used to provide water for livestock and home gardens, allowing households to supplement both income and food production [2]. Despite this evidence of clear household-level gains, little research has directly evaluated the effects of MUS practices on household water insecurity.

Over the last decade, research on household water insecurity has gained ground in both academic literature and global development institutions. Grounded in the human capabilities approach, this literature defines water insecurity as the “inadequate, unreliable, and unaffordable water for a healthy life.” [12,21]. Applied to water insecurity, the human capabilities approach expands beyond a strict focus on availability and quality of water and attends to the intersecting socio-economic, cultural, political, and ecological processes that contribute to household water insecurity. This relational approach to understanding water insecurity is attentive to diverse lived experiences. In this way, this scholarship contextualizes household water insecurity, exploring the structures and processes through which water is secured and emphasizing social relations underpinning access to water.

Household-level studies reveal demonstrated relationships between water insecurity and food security, economic outcomes, and associated health benefits to households [22–26]. In India, research has shown that household water insecurity affects child nutrition and that suboptimal water access is significantly associated with a higher likelihood of child stunting [27]. Although this research extends beyond unidimensional assessments of water infrastructure and delivery, infrastructural changes can transform water security outcomes. In Bolivia, one study demonstrated that households with improved water delivery services closer to their homestead experience greater water security. [28]

More recently, studies on household water insecurity have examined the various dimensions of household experiences, addressing multiple facets of water insecurity, including water worry, water access, water supply interruptions, and hygiene. This work underscores that water insecurity is multidimensional, varying across populations and contexts in both space and time [29]. While much of this work emphasizes the measurement and quantification of water security, some researchers have also considered the cultural meanings of water and customary practices that are not easily captured by standardized metrics [30–33]. By attending to the multiple dimensions of and experiences with water insecurity, scholars can generate evidence that supports more targeted and locally relevant policy interventions [29,34].

This shift has practical implications; for example, a study in urban Mexico demonstrated that a unidimensional framing of water insecurity obscures the pathways through which water affects health outcomes. In this study, the drinking water source was not associated with household water insecurity nor predictive of waterborne illnesses such as diarrhea or dengue fever. Instead, temporal factors like water intermittency and predictability, which shape households’ ability to access, store, and manage water, were found to be more strongly correlated with water insecurity [29]. Such findings highlight the need for frameworks that capture the multidimensional and temporally dynamic nature of water insecurity, rather than relying on a standard metric.

Household water insecurity is influenced by seasonality [35,36], which affects water quality, availability, and source use. Seasonal variation often compels households to shift their primary water sources between rainy and dry periods, a process known as seasonal switching [8]. This pattern is especially common among households relying on basic services such as hand pumps, protected dug wells, or springs, which are typically off-premises and not consistently available. Such sources fall below the “safely managed” service threshold, defined by the World Health Organization (WHO) and the United Nations Children’s Fund (UNICEF) [37] as water that is accessible on premises, available when needed, and free from contamination. Situating seasonality within the MUS framework is critical for guiding the timing and targeting of interventions, particularly as climate change intensifies seasonal variability and further strains fragile water delivery systems.

Understanding the relationship between household water security and community-managed water delivery systems designed for multiple uses is equally critical [23]. Examining this relationship at the household level provides insight into how MUS function within diverse livelihood strategies embedded in local hydro-social systems, that is, the interlinked social, institutional, and ecological processes that shape how water is accessed, shared, and governed within a given context [38]. Because MUS reflects both the social organization and physical infrastructure that connect households, extended families, and communities, situating these systems within their local context helps clarify whether they alleviate or exacerbate household water insecurity.

Scholarship on household water insecurity demonstrates parallels between the direct and indirect benefits of MUS, suggesting that these practices could be a coping strategy for individuals and household-level water insecurity. This points to a need for studies that test whether MUS could be an arrangement that reduces instances of water insecurity within households [39]. Research on MUS can clarify how households innovate and deploy technologies to close gaps between domestic water needs and productive water uses.

This study integrates literature on multiple-use water services (MUS) and household water insecurity to assess how water service delivery systems that support multiple water uses influence household water insecurity, as measured by the HWISE scale and its subdomains: water worry, hygiene, access, and interruption. The study further seeks to test and refine a theoretical model that positions MUS as a strategy through which households respond to the multidimensional nature of water insecurity. We hypothesize that the association between MUS and household water insecurity depends on whether MUS can be sustained throughout the year or will be seasonal. Specifically, households practicing year-round MUS will report lower household water insecurity than households practicing seasonal MUS or households not practicing MUS.. By demonstrating when and why MUS coincides with heightened household water insecurity, however, this study challenges the assumption that expanding multiple-use access is sufficient to improve lived water security and instead highlights reliability and seasonality as central determinants of household well-being.

3. Methods

3.2. Study region

The study was conducted among 1,082 households across the districts of Bandiagara, Bankass, Bougouni, Kadiolo, Kolondieba, Koro, Koutiala, Mopti, Sikasso, Tenenkou, Yanfolila, and Yorosso, located within the Mopti and Sikasso regions of southern Mali (Fig 2). These two administrative regions are characterized by diverse and ethnic language groups, predominantly rural populations, and economies heavily reliant on agriculture.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Map of study area showing data collection sites and administrative boundaries.

Map produced in ARCGIS Pro Version 3.6.

https://doi.org/10.1371/journal.pwat.0000600.g002

Sikasso, often referred to as Mali’s “granary,” is distinguished by extensive cotton and cereal production, while Mopti combines crop cultivation with livestock rearing and fishing along the Niger and Bani rivers. Environmental variability, including recurrent droughts, irregular rainfall, and land degradation, shapes water access and agricultural practices in both regions. Together, these features provide a representative context for examining household-level experiences with water insecurity in rural, agriculturally dependent settings of southern Mali.

Mopti lies in the floodplains of the Niger River delta, within Mali’s Sahelian agro-ecological zone. The region receives up to 550 mm of annual rainfall and experiences a prolonged dry season lasting 9–11 months [40,41]. While Mopti includes several urban centers such as Mopti City, Sévaré, Djenné, Bandiagara, Bankass, Douentza, and Youwarou, approximately 90% of its population remains rural [40,42]. Livelihoods depend on livestock herding, fishing, and agriculture, with staple crops including sorghum and millet, complemented by root and garden crops such as yams, cassava, and sweet potatoes. Communities draw primarily on the Niger River for flood-recession irrigation of paddy rice [42] and depend on groundwater from shallow, traditional, and modern wells, as well as boreholes, for drinking, domestic, and livestock needs.

Sikasso, located within Mali’s Guinea agro-ecological zone, has a tropical savanna climate and receives an average of 1,200 mm of rainfall annually, with a rainy season lasting five to seven months [40]. The region contains several urban centers, including Sikasso City, Yanfolila, Bougouni, and Koutiala, yet most residents remain rural and depend on agriculture. Farmers cultivate cotton, sorghum, millet, cowpeas, peanuts, and rice, along with household gardens and root crops such as yams, cassava, and sweet potatoes. Despite its high agricultural potential, Sikasso ranks third nationally in the share of residents living below the poverty line, as many households rely almost entirely on agriculture for income [43]. Communities utilize groundwater recharged by the Bani and Sankarani rivers for drinking, domestic use, and livestock watering, drawing on traditional and modern wells and boreholes. Rainfed agriculture dominates, although some farmers use surface water for supplementary irrigation.

Across both regions, surface water supports irrigation and smaller consumptive uses such as municipal supply and industry, though irrigation remains predominant [42]. MUS enables smallholders to manage water across seasons and stabilize food production. While irrigation infrastructure is primarily designed for agriculture, it also provides water for consumptive uses (e.g., drinking, gardening, livestock watering) and non-consumptive uses (e.g., washing, bathing, fishing, religious, and recreational activities). The prominence of such practices in Mopti and Sikasso makes these regions critical to the growing body of empirical research on MUS and household water insecurity in West Africa.

3.3. Household survey

A survey of 1,082 households was conducted by the International Food Policy Research Institute (IFPRI) as part of a broader study under the United States Agency for International Development (USAID) Feed the Future Innovation Lab for Small-Scale Irrigation (ILSSI). The intent of the household survey was to investigate the potential impacts of small-scale irrigation on agricultural and nutrition outcomes and water access in Mopti and Sikasso. The household survey used a stratified random sample to collect data from households in February and April 2021.

The sampling frame for this survey was defined using key project suitability criteria, including irrigation potential. The irrigation suitability ranking of districts was developed by IFPRI using a methodology originally designed for the World Bank [44], which incorporates biophysical and economic factors such as soil type, slope, and water availability. Both Mopti and Sikasso were purposively selected to capture contrasting agro-ecological zones and to align with USAID operational areas while accounting for security conditions. Within each region, districts were stratified by irrigation suitability, and a corresponding sampling frame of villages was created. Seven villages per district were then randomly selected. At the village level, a complete household listing was compiled and used to randomly select participating households.

To ensure sufficient statistical power to detect differences between irrigators and non-irrigators, IFPRI conducted power calculations based on expected income effects from irrigation. Assuming a 30% increase in household income, a power of 90%, and a 5% significance level, the analysis indicated a required sample of 587 irrigating households and 587 non-irrigating households, for a total of approximately 1,174 households.

The household survey collected data on agricultural production; household water sources; water insecurity, hygiene, and sanitation; the effects of agricultural water management technologies on food and nutrition; and access to rural services and market participation. For this study, data on household water insecurity, potable water security, and responses to the HWISE Scale [13] were analyzed. The HWISE Scale is a validated, cross-cultural instrument that measures household-level experiences of inadequate, unreliable, or unsafe water access over the previous four weeks. It consists of 12 items that capture multiple dimensions of water insecurity, availability, accessibility, use, and reliability, by asking how frequently households encountered specific challenges such as worrying about water, altering daily plans due to water problems, being unable to wash hands, body, or clothes, lacking sufficient water for cooking or drinking, using unsafe water, experiencing interruptions in supply, or spending excessive time collecting water. Responses are recorded on a four-point frequency scale (never, rarely, sometimes, often/always), enabling the calculation of a composite score that quantifies the severity of experiences with household water insecurity.

3.3.1 Statistical analysis.

Analyses were conducted in Stata/MP version 17.0 [45]. To test whether MUS households experience lower water insecurity than non-MUS households, we estimated five multiple linear regression models. Dependent variables included the 12-item HWISE Scale (scores range 0–36) and four subdomain scores, worry, hygiene, access, and interruption (each 0–9). Modeling both the composite scale and its subdomains provided a more detailed understanding of how MUS relates to different facets of water insecurity.

Following prior HWISE applications [13,22,29,36], we treated the HWISE score and its subscores as continuous variables and analyzed them using multiple linear regression. Although these scales are technically bounded count variables derived from ordinal items, prior validation studies demonstrate that treating them as continuous variables yields consistent and interpretable estimates. We assessed model assumptions by examining residuals for normality and homoscedasticity; no severe violations were observed. This approach allows comparability with other studies using the HWISE scale. To assess robustness given the ordinal and bounded nature of these scales, we additionally estimated ordered logit models, which produced substantively similar results. Detailed results from these models are presented in S4 Table.

Our independent variables for the multiple linear regression analysis are grouped into four categories (family characteristics, potable water source, water storage and treatment, and multiple-use households) and include factors known to contribute to, or mitigate, water insecurity. Variables are described in Table 1. We assessed potential multicollinearity among predictor variables prior to model estimation. Variance inflation factors (VIFs) were calculated for all independent variables. Following conventional guidelines (VIF > 10 indicating severe multicollinearity [46], all variables exhibited acceptable values (range: 1.2–3.4), suggesting that multicollinearity was not a concern. Of the 1,082 households surveyed, complete data were available for 1,069 households included in the regression analyses. Observations with missing data on one or more model variables were excluded using listwise deletion. Missing data were minimal (< 2%) and did not appear to follow a systematic pattern across key variables.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Variables included in the multiple linear regression models.

https://doi.org/10.1371/journal.pwat.0000600.t001

3.3.2. Variable development and data transformation.

We used existing survey data to derive both dependent and independent variables for multiple linear regression analysis. The dependent variables included overall and subdomain household water insecurity scores calculated from the HWISE Scale. Independent variables captured household and infrastructural characteristics, including MUS status, access to an improved primary water source, and a wealth index based on asset ownership. Together, these variables allowed us to assess how MUS relates to household experiences with water insecurity.

Multiple-Water Use Households: Multiple-Water Use Households describe households that rely on the same water sources for both domestic needs (such as drinking, cooking, and cleaning) and productive activities (such as irrigation, livestock, or small enterprises). This approach reflects how rural households often rely on single or limited water sources to meet multiple needs. Because the original survey did not include a dedicated question identifying types of MUS households, binary variables were constructed to classify households that practiced MUS year-round, those that practiced MUS only during the rainy season, and those that did so only during the dry season.

As shown in Fig 3 below, proxy variables for households practicing MUS year-round or seasonally were developed using survey data on irrigation practices and the types of water sources used for productive and domestic activities. The process followed a stepwise approach. First, households were classified according to whether they used the same type of water source, surface water or groundwater, for both productive and domestic purposes. Next, productive water use was defined by the presence of irrigation, and irrigating households were categorized by seasonal practice as year-round, rainy-season-only, or dry-season-only. Each irrigation category was then further divided by source type. Domestic water use was evaluated across three key functions: drinking, cleaning, and cooking. and coded by season, irrigation status, and source type. To ensure comparability, domestic-use classifications followed the same seasonal and source-based criteria used for productive uses.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. MUS Variable Construction.

https://doi.org/10.1371/journal.pwat.0000600.g003

Final MUS variables were generated by linking productive and domestic water uses within each temporal frame. Year-round MUS households were those that consistently relied on the same source for irrigation and all domestic uses in both seasons. Seasonal MUS households (rainy- or dry-season-only) met the same alignment criteria but within a single season. This classification framework provided a systematic basis for comparing household water insecurity across the three MUS categories. MUS classification was derived from existing survey responses rather than a dedicated MUS instrument; therefore, some degree of misclassification is possible. Nevertheless, the operational definition closely aligns with the conceptual definition of MUS as the use of a single water source for both domestic and productive purposes.

Household Water Insecurity Scores: Household water insecurity was measured using the 12-item HWISE Scale [13], implemented as part of the household survey. The items captured experiences related to water availability, quantity, hygiene, and psychosocial distress, including situations such as interrupted supply, insufficient water for washing or drinking, and worry or frustration over water problems. Each item was scored from 0 (“never”) to 3 (“often/always”) based on how frequently the experience occurred in the previous four weeks. Individual scores were summed to produce an overall range of 0–36, where higher values represent greater water insecurity.

Subdomains of household water insecurity: As in previous studies [29,32,34], we examined subdomains of household water insecurity to assess how different dimensions operate within the MUS context. Subdomains represent clusters of related experiences within the broader HWISE Scale that capture specific aspects such as emotional distress, access challenges, and disruptions in use. As shown in Table 2, we derived four subdomain scores (each ranging from 0 to 9) from the 12 HWISE items: worry, reflecting emotional distress related to water scarcity; hygiene, measuring the inability to maintain personal and domestic hygiene; access, representing challenges in obtaining sufficient water; and interruptions, capturing disruptions in water availability or supply reliability.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Items used in each HWISE subscore construct.

https://doi.org/10.1371/journal.pwat.0000600.t002

Improved and unimproved water sources: A binary variable was created using existing data on potable water sources. This data were aggregated using the WHO Joint Monitoring Program definitions, improved water sources (piped water, protected borehole, protected well, and protected springs), and unimproved sources (unprotected spring, water vendor, tanker, truck, bottled water, and bagged water) [37].

Wealth Index: A wealth index was constructed to assess household socioeconomic status using available non-monetary indicators. Because the survey did not collect household income, expenditure, or comprehensive asset ownership data beyond livestock, livestock holdings and educational attainment were used as proxy indicators of relative wealth. In the rural farming context of this study, livestock function as both productive and financial assets, representing a key store of value, buffering capacity, and source of livelihood. In settings where cash income and durable goods are limited or inconsistently distributed, livestock ownership provides a stable and contextually meaningful indicator of long-term economic well-being [47–49]. Educational attainment was included to capture structural socioeconomic advantages related to income opportunities, access to information, and social position.

Principal Component Analysis (PCA) was applied to summarize variation across household livestock ownership and educational attainment. Survey questions included the education level of the household head (none, secondary, post-secondary) and ownership of specific livestock types (cattle, oxen, camels, bulls, and poultry). Each variable was coded as binary, with a value of 1 indicating ownership of the asset or attainment of the specified education level and 0 otherwise. PCA was selected over Multiple Correspondence Analysis (MCA) because all indicators were numeric (binary) and could be represented on a continuous scale, making PCA appropriate for extracting a single underlying dimension of relative socioeconomic status. Following Loewen and Gonulal (2015) [50], variables with factor loadings below 0.1 in the initial PCA were dropped, and the analysis was re-run using the remaining indicators. The first principal component, which explained the largest share of variation across households, was retained and used as the composite wealth score. Each household received a continuous index value representing its relative socioeconomic position within the sample. These scores were ranked and divided into quintiles, with the lowest two quintiles classified as “poor” and the upper three classified as “not poor.” The PCA results were used exclusively to construct the wealth index and were not applied to any additional analyses. Although this index does not capture short-term income, liquid assets, or consumption, it provides a concise and seasonally stable measure of relative household wealth suitable for controlling socioeconomic differences in the analysis of household water insecurity.

4. Results

This section presents the results of an empirical analysis examining the relationship between MUS and experiences with household water insecurity in Mopti and Sikasso. The analysis first characterizes the socio-economic and farm-level attributes of MUS households and then reports the frequency of experiences of water insecurity across four subdomains of HWISE: worry, interruption, hygiene, and access. A multiple linear regression model is then employed to identify the significant predictors of household water insecurity. Collectively, the findings illustrate how seasonal MUS practices shape household vulnerability and resilience to water insecurity, helping to inform policy development and development interventions aimed at improving water delivery services in low- and middle-income countries.

4.2. Cross tabulations: Household water insecurity

MUS underscores the central role of water in diverse livelihood strategies, illustrating how resource-constrained communities secure and utilize water for both domestic and productive purposes. Existing literature highlights the benefits of MUS, including improved access to water and expanded opportunities for income generation. Given these advantages, it is important to assess how experiences of water insecurity vary across households practicing MUS year-round, during the rainy season only, or during the dry season only. Table 3 shows that, when considered in aggregate, households report relatively low levels of water insecurity, with most respondents indicating that they “never” or “rarely” experienced challenges related to water worry, hygiene, access, or supply interruptions. These aggregate patterns suggest broadly low levels of reported water insecurity across the sample.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Frequency of experiences with household water insecurity for all households surveyed.

https://doi.org/10.1371/journal.pwat.0000600.t003

The analysis included non-MUS households and households practicing MUS year-round, exclusively during the dry season, or exclusively during the rainy season. A Kruskal–Wallis test was used to compare HWISE scores across household groups, with detailed results presented in Table S2. This nonparametric approach was selected because HWISE scores are ordinal and deviate from a normal distribution, making it appropriate for assessing median differences across multiple independent groups. Households practicing MUS year-round exhibited mean scores slightly above one point across all HWISE subdomains, with no statistically significant differences observed across domains. Dry-season MUS households demonstrated intermediate outcomes, with no statistically significant differences relative to year-round MUS households. In contrast, rainy-season MUS households exhibited consistently higher mean scores across all subdomains; although the absolute magnitudes were comparable, these differences were statistically significant relative to both year-round and dry-season MUS households, suggesting higher levels of water insecurity.

4.3. Correlates for water insecurity

To evaluate the factors influencing households’ experiences with water insecurity in Mopti and Sikasso, five multiple linear regression models were developed. These models examined how engagement in MUS, including year-round (MUS-Y), dry-season-only (MUS-D), and rainy-season-only (MUS-R) practices, was associated with overall and domain-specific dimensions of water insecurity. The dependent variables included the total HWISE score and four subdomains: worry, hygiene, access, and interruption.

Explanatory variables were drawn from the survey data based on prior research highlighting their relevance to household water insecurity and water-use behaviors. The analysis focused on variables capturing both water-supply constraints and household characteristics. In particular, water source type, time to collect water, and seasonality have been shown to influence access and reliability of supply [8,12,32,51], whereas gender, age structure, and household size shape vulnerability and water management roles [23,28]. The S3 Table summarizes all variables and their descriptions. Collectively, these variables represent key socioeconomic, demographic, and water-access dimensions relevant to rural Mali. Table 4 presents the regression coefficients, 95% confidence intervals, and significance levels for each model. As is common in models of experiential and perception-based outcomes such as household water insecurity, overall model fit (R²) is expected to be modest; accordingly, the regression results are interpreted as conditional associations rather than comprehensive explanations or predictions of household water insecurity.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Multiple linear regression models predicting the influence of multiple water use on Household Water Insecurity.

https://doi.org/10.1371/journal.pwat.0000600.t004

Households not classified as MUS users serve as the reference group in the regression models. Relative to this group, year-round MUS is associated with significantly lower levels of water insecurity across all HWISE subdomains, whereas seasonal MUS (rainy- or dry-season only) is associated with significantly higher levels of water insecurity. Non-MUS households in the sample typically relied on a single domestic water source year-round and were less likely to shift water sources across seasons for productive uses.

5. Discussion

Our findings suggest that the benefits commonly attributed to multiple-use water services depend less on the existence of multiple-use practices themselves than on the reliability and seasonal continuity of the water systems that support them. Year-round MUS households experienced significantly lower water insecurity, reducing their HWISE scores by nearly two-thirds relative to the sample mean. In comparison, seasonal MUS households, whether in the rainy or dry season, reported significantly greater experiences with household water insecurity, with scores nearly half as high as the mean. These differences indicate that households practicing different forms of MUS experience systematically different levels of household water insecurity. However, these associations should not be interpreted as evidence that MUS itself determines household water insecurity. Rather, they suggest that the environmental and infrastructural conditions that enable or constrain year-round MUS, particularly water availability, reliability, and seasonality, also shape households’ experiences of water insecurity.

Moreover, our findings suggest that the relationship between MUS and household water insecurity is conditional, rather than uniform. Year-round MUS households consistently showed lower experience with household water insecurity, while seasonal MUS households, particularly those relying on rain-fed or surface water sources, reported higher experiences with water insecurity. These results should not be interpreted as evidence that MUS practices themselves are harmful; rather, they highlight how the seasonal availability, quality, and reliability of water sources mediate the outcomes of MUS and shape household vulnerabilities.

These findings align with prior MUS scholarship demonstrating that reliable, year-round water access can support livelihood stability, reduce labor burdens, and enhance household well-being [4,5]. However, our findings also suggest that these benefits should not be attributed to MUS practices alone. Rather, they indicate that the reliable and year-round availability of water may enable both sustained MUS practices and lower experiences of household water insecurity. In this way, our results extend the MUS literature by showing that the advantages associated with year-round MUS are reflected not only in productive or economic outcomes, but also in lower experiential dimensions of household water insecurity, as measured by the HWISE scale. These relationships should be interpreted as associations rather than evidence that MUS itself causes improvements in household water security.

The finding that households practicing MUS only seasonally experience higher water insecurity than both year-round MUS households and non-MUS households warrants further reflection. One explanation is that seasonal MUS often reflects an adaptive response to constrained and unstable water access rather than a stable service arrangement. During the rainy season, households engaging in MUS frequently rely on surface water or rain-fed sources that are more accessible but less reliable and more vulnerable to microbial contamination. In the dry season, seasonal MUS may require households to depend on distant irrigation canals, receding streams, or shared groundwater sources, increasing time burdens, competition between users and uses, and exposure to supply interruptions. In contrast, non-MUS households may rely more consistently on domestic water sources designed primarily for consumptive use, even if those sources are limited in quantity. Seasonal MUS thus introduces additional tradeoffs, between accessibility, quality, and reliability, that can amplify experiences of water worry, interruption, and access constraints.

Rather than indicating a failure of MUS per se, these patterns suggest that seasonal MUS functions as a coping strategy under conditions of heightened vulnerability, whereas year-round MUS reflects more stable and resilient water access arrangements. This interpretation is consistent with research on multiple water source use and seasonal switching, which shows that households often diversify or reallocate water sources in response to a range of risks, such as seasonal scarcity, quality, and service unreliability [7,8]. Interpreting seasonal MUS as an adaptive response also raises questions about the mechanisms driving seasonal switching. In some cases, seasonal MUS may reflect the temporary loss or physical unavailability of a primary multiple-use source, while in others it may reflect increased strain on a source that remains available, prompting households or communities to prioritize domestic uses and reallocate productive uses elsewhere. While the survey data do not allow us to distinguish conclusively between these mechanisms, both are consistent with the observed association between seasonal MUS and heightened experiences of household water insecurity. More broadly, observed differences related to seasonality reflect household vulnerability and resilience. Rainy-season MUS households face heightened risks of microbial contamination and supply interruptions when relying on rain-fed or surface sources, while dry-season MUS households encounter significant time and cost burdens as they access irrigation canals or distant groundwater during periods of scarcity. These seasonal dynamics reinforce the importance of moving beyond static indicators of water access; infrastructure-based metrics alone cannot capture distinct insecurity such as variability in availability, quality, and reliability. The patterns that emerge from this study echo a growing body of work that emphasize seasonality as a central driver of household water insecurity across diverse contexts, particularly related to contexts where households rely on variable surface or groundwater sources [35,36].

Although this study focuses on seasonal variability rather than rare hydrological extremes, the observed relationships between seasonal MUS and heightened household water insecurity have implications for resilience to more extreme events such as droughts or floods. Research on rural water supply performance during drought has shown that communities are often forced to switch water sources as infrastructure fails or becomes unreliable under stress, with resilience hinging on the continuity and robustness of available water services rather than nominal access alone [52]. In this context, the higher water insecurity observed among households practicing seasonal MUS likely signals underlying vulnerability to more extreme disruptions, particularly where households depend on surface water or seasonally intermittent sources. Conversely, the lower insecurity reported by year-round MUS households suggests that consistent access to a reliable source capable of supporting multiple uses may enhance household resilience when confronted with prolonged scarcity or system shocks. While future research is needed to explicitly examine MUS performance under extreme events, these findings underscore the importance of designing water services that prioritize reliability and continuity as core components of climate resilience.

These results challenge the assumption that MUS is inherently protective against household water insecurity, demonstrating that benefits depend critically on continuity and reliability across seasons. Consistent with these resilience dynamics, prior research has emphasized MUS as a driver of economic and livelihood gains, but our analysis demonstrates that seasonal MUS might well be a sign of water insecurity. At the same time, this study extends household water insecurity scholarship by explicitly linking seasonal switching to experiential measures of water insecurity captured by the HWISE scale. In this way, we show that seasonal MUS elevates not only risks of microbial contamination or travel time but also subdimensions such as worry and interruption. When unexamined, prevailing assumptions that MUS is uniformly protective may lead policymakers and practitioners to overlook seasonal vulnerabilities, misclassify at-risk households, and overestimate the resilience benefits of MUS interventions.

The findings of this study should be interpreted with attention to context and scope. While the analysis draws on data from rural communities in Mopti and Sikasso, the mechanisms identified, particularly the conditional role of seasonality, reliability, and source continuity in shaping household water insecurity, are likely relevant to other settings characterized by seasonal variability and mixed water-use practices. At the same time, the specific magnitudes and configurations of MUS observed here are shaped by local hydrological conditions, governance arrangements, and livelihood strategies, and should not be assumed to translate directly to urban contexts or regions with fundamentally different water infrastructures. In this sense, the contribution of this study lies less in generalizing outcomes and more in clarifying processes through which MUS may reduce or reinforce household water insecurity under variable conditions.

Our findings suggest that policies should move beyond generic promotion of MUS and instead prioritize designs and management approaches that ensure continuity and reduce seasonal vulnerabilities. First, interventions should strengthen year-round reliability through measures that explicitly account for intra-annual variability, such as dry-season water storage, conjunctive use of groundwater and surface water, and investment in maintenance timed to periods of peak seasonal stress. Second, for households reliant on seasonal MUS, program design should explicitly address seasonal risks by promoting low-cost water treatment to mitigate microbial contamination during rainy-season reliance on surface water and by reducing dry-season burdens through improved scheduling and water user committees. Third, institutional frameworks should recognize MUS not only as an infrastructure model but as a household survival and resilience strategy. This requires integrating MUS into national water monitoring frameworks and pairing infrastructure indicators with experiential metrics such as the HWISE scale to capture seasonal fluctuations in reliability, quality, and access and to identify households facing hidden or intermittent water insecurity. Together, these measures can help ensure that MUS functions as a resilience-building strategy rather than inadvertently reinforcing seasonal water insecurity.

Conceptually, our study reframes MUS from a primarily community and infrastructure-oriented model to one evaluated through a household lens with experiential outcomes. Using HWISE to evaluate the water insecurity implications of MUS illustrates the value of integrating tools like HWISE into the evaluation of water service delivery models. In doing so, we highlight how MUS functions not as a uniform good but as a conditional strategy.

Overall, our findings show that MUS is not inherently protective against water insecurity. Its success depends on whether MUS systems are reliable and continuous across seasons. When practiced year-round, MUS enhances resilience and reduces vulnerability. When practiced only seasonally, the HWISE scales revealed that MUS may amplify burdens and risks. These results call for a reframing of MUS models. Integrating tools like HWISE is essential for designing systems that align with household realities and contribute to improved experiences with household water insecurity.

6. Limitations

This study is subject to several limitations. First, the household survey was not explicitly designed to measure MUS. We constructed MUS categories indirectly from irrigation and domestic water use data, applying harmonized seasonal and source-based criteria. While this approach allowed for a systematic analysis of MUS within the available dataset, it represents a proxy classification rather than direct household reporting. As such, the prevalence of MUS may be under- or overestimated. Future studies should incorporate dedicated survey modules on MUS to enable more accurate measurement and validation.

Second, data collection occurred during the COVID-19 pandemic and relied on local enumerators. This strategy enabled fieldwork despite mobility restrictions but limited our direct engagement with the study context. Consequently, some interpretations necessarily involved inference, which may have shaped our understanding of the findings.

Third, the study is based on a single geographic setting. The hydrological, cultural, and institutional conditions of this area may not reflect those in other regions, limiting the generalizability of the results.

Fourth, several measures, including household water use and perceived water quality, were self-reported. These indicators are vulnerable to recall bias and social desirability effects, which may introduce uncertainty into the analysis [52,53].

7. Conclusion

Our study advances scholarship on MUS and household water insecurity by directly linking them through experiential measures. While prior MUS research has emphasized technical design, productivity, or economic benefits, and household water insecurity scholarship has focused on experiential outcomes without considering service models, our analysis bridges these perspectives. By applying the HWISE scale to household survey data from Mali, we provide one of the first empirical tests in sub-Saharan Africa of how MUS practices are associated with household water insecurity. Across the two study regions of Mali, year-round MUS is associated with lower household water insecurity, whereas seasonal MUS amplifies vulnerabilities. The results demonstrate that MUS is not uniformly protective. Year-round MUS significantly reduces insecurity, while seasonal MUS amplifies vulnerabilities. By reframing MUS as a resilience and survival strategy rather than an inherently beneficial model, this study challenges prevailing assumptions and reinforces the importance of seasonality in evaluating water service delivery. Ultimately, our findings show that MUS strengthens resilience only when practiced year-round, highlighting that seasonality determines whether multiple water use alleviates or signals household water insecurity.

Supporting information

Acknowledgments

This publication was partly made possible through support provided by the Feed the Future Innovation Lab for Small-Scale Irrigation through the U.S. Agency for International Development under the terms of Contract No. AID-OAA-A-13–0005. The opinions expressed herein are those of the authors and do not necessarily reflect the views of the United States Government. The authors also acknowledge the support of the Texas A&M University Chancellor’s EDGES Fellowship Program, whose fellowship support contributed to the completion of this research.

Data: Cleaned data can be accessed through the USAID Feed the Future Innovation Lab for Small-Scale Irrigation Mali Dataverse.