Introduction

The Amazon rainforest is home to approximately 1.6 million Indigenous peoples, many of whom depend on the forests and waters of this ecosystem to meet their basic needs [1]. While the health of Indigenous populations relying on intact ecosystems and abundant natural resources has historically been robust [2], rapid environmental degradation and social changes are undermining the ability of some Indigenous peoples to meet their food and water needs [3].

Food and water insecurity often co-occur [4–10] and multiple pathways potentially connect these phenomena [4, 11]. As outlined by Frongillo (2023), food and water insecurity coexist in the same populations and households, act as dual stressors, and can be linked through multiple biological and behavioral mechanisms [4]. For example, water insecurity may worsen food insecurity by precluding households from engaging in water-intensive subsistence agriculture [4]. Conversely, water collection requires substantial physical exertion [12], which is contingent on sufficient calories from food to sustain efforts [13]. Water insecurity may also lead households to purchase energy-dense, pre-packaged foods that require less water for preparation [14], prioritize sugar-sweetened beverages due to perceived poor water quality and availability [7,15–18], and reduce dietary diversity [19,20].

Given the potentially bidirectional and synergistic interactions between various resource insecurities, understanding the specific dynamics between food and water insecurity is essential to formulate and support initiatives designed to promote both Sustainable Development Goals (SDG) 6 (Clean Water and Sanitation) and SDG 2 (Zero Hunger) [21]. While there is clear evidence of the connections between food and water insecurity in contexts with known water shortages [9,22–24], our knowledge about this relationship in water-abundant contexts is just emerging. For example, recent findings from a study among the Tsimane’, an Indigenous group living in the Bolivian Amazon, found that water insecurity was associated with food insecurity and linked with changes in dietary consumption [25]. Similarly, our prior research among the Awajún, an Indigenous group living in the Peruvian Amazon, has demonstrated the existence of water [26,27] and food insecurity [28], and the dual existence of under and over-nutrition, amidst rapid lifestyle and dietary changes in the study area [2]. Yet, our understanding of the relationships between these resource insecurities within this population, and more broadly, requires additional investigation. This is particularly important given the escalating need to understand the dynamics of health disparities among Indigenous populations [29] and to identify locally-relevant pathways for the promotion of SDG 6 (clean water) and SDG 2 (zero hunger) amongst vulnerable groups living in water-abundant areas.

Briefly, the Awajún are an ethnolinguistic group with more than fifty-five thousand individuals living in the provinces of Amazonas, Cajamarca, Loreto, and San Martín in Peru [30]. Historically, the Awajún lived in widely dispersed households, practicing small-scale subsistence farming, hunting, and fishing [31,32]. However, major political and environmental changes occurred in Awajún territory in the twentieth century, including the enactment of colonization policies such as the 1969 “Agrarian Reform Act” and the 1974 “Native Communities Law.” [33]. These political-economic shifts pushed Awajún families toward central locations and encouraged intensive agriculture for the legal recognition of communities [33]. Shifts in ecology, market access, and land use have generally resulted in the consumption of lower quality foods and increased obesity and anemia in this population [2], signaling the onset of the nutritional and epidemiological transition in the population more broadly. Despite these changes, the importance of water and its role in daily life among the Awajún cannot be overstated, as local rivers continue to serve as a source of food, drinking water, and transport for many community members [28,34], with implications for physical and mental health [26,27].

In this paper, we examine the relationship among experiences of food insecurity and water insecurity among Awajún adults living in the northern Peruvian Amazon in four ways. First,based on prior studies, [10,35] we hypothesized that higher water insecurity scores would be positively associated with higher food insecurity scores. Second, we investigated the strength of the associations between these variables and explored if the individual constituent scale items, within the larger water insecurity scale, were significantly associated with food insecurity scores. Third, we examined the relationship between composite water insecurity, individual constituent scale items, and dietary diversity–a proxy for diet quality [36]– testing the hypothesis that water insecurity would be associated with lower dietary diversity scores. Finally, the relationship between water insecurity and the reported consumption of ‘water-intensive’ foods was explored, with the hypothesis that water insecurity would be correlated with lower consumption of ‘water-intensive’ foods.

Materials and methods

Data analysis

All analyses were conducted in Stata14 [37]. Descriptive analyses were conducted using Chi-squared and t-tests. Bivariate and multivariable linear regression models estimating robust standard errors were used to evaluate the relationship between food and water insecurity, and between household dietary diversity and water insecurity. Bivariate and multivariable logistic regression models estimating robust standard errors were used to assess the relationship between each food group and water insecurity and between consumption of water-intensive foods and water insecurity. Community-level fixed effects were used to account for potential differences in food and water insecurity between the four sampled communities. To account for potential Type I statistical error, Benjamini-Hochberg false discovery rate (FDR) correction was applied to regression models between composite food insecurity and individual water insecurity scale items.

Food insecurity was treated as the dependent variable based on prior literature establishing water insecurity as a driver of food insecurity [10,35], and assessed using four-items from the USDA food security scale (Table B in S1 Appendix) [38]. Food security assessment items included “Which of the following statements best describes the food eaten in your household in the last 12 months?” with response options of (0) always enough to eat and types of food we want; (1) have enough to eat, but not always the types of food we want; (2) sometimes we don’t have enough to eat; and (3) often we do not have enough to eat. Three additional food security items were: “We worried whether our food would run out before we got money to buy more;” “The food that we bought didn’t last and we didn’t have money to buy more;” and “We couldn’t afford to eat a balanced meal.” Response options were (0) “Never”, (1) “Sometimes”, (2) “Often”, and (3) “Always”.

Responses from each item in the scale were summed for a composite score ranging from zero to twelve, with greater scores indicating worse experiences of food insecurity. Internal consistency of the four-item food insecurity scale was modest (Cronbach’s α = 0.56). This falls below commonly used thresholds for acceptable reliability (>0.70) [39], however, Cronbach’s alpha is sensitive to the number of items included in a scale, and lower values are not unexpected for short scales [40]. Because the four items were conceptually aligned and represented related food insecurity experiences, the composite score was used. The maximum score reached by respondents for the food insecurity scale in this study was 10, so ≥5 (the mid-point) was used as a cutoff for “food insecurity”. This was determined using visual inspection of data distribution with a noticeable cutoff before and after 5. However, the continuous score for food insecurity was used in the statistical models where food insecurity was the dependent variable. This was done to account for potential errors introduced with a cutoff point of 5. Sensitivity analyses using lower cutoff points (3 and 4) were also conducted but did not meaningfully effect results.

Water insecurity, the independent variable, was assessed using an adapted seven-item version of a scale with binary (yes/no) response options developed by Stevenson et al., (2012) to assess water insecurity among women in Ethiopia (Table A in S1 Appendix) [41]. This scaled used a recall period of 30-days. Measures of water insecurity validated for the Peruvian context, such as the WISE (Water Insecurity Experiences Scales) [42,43], had not been developed at the time of data collection (Table 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive analysis of food and water insecurity and sociodemographic characteristics of cross-sectional secondary data available for Awajún adults living in the Peruvian Amazon (n = 212)^†.

https://doi.org/10.1371/journal.pwat.0000556.t001

To assess the internal consistency of the water insecurity scale and determine if all items should be retained, tetrachoric correlation was conducted, given the dichotomous response structure of the items [39]. The average inter-item tetrachoric correlation was 0.18, indicating a modest positive association between the water insecurity items. The tetrachoric alpha for the seven-item scale was 0.61, which suggests moderate internal consistency as a composite water insecurity measure. Composite scores ranged from zero to seven, with higher scores indicating more experiences of water insecurity. Distribution was assessed with visual inspection using a normalized histogram, a Shapiro-Wilk test, and a skewness/kurtosis test.

Household dietary diversity scores (HDDS) [44] were collected using 24-hr recall and operationalized as a sum of eleven locally-relevant food groups: cereals, roots/tubers, vegetables, fruits, meat/poultry/offal, eggs, fish/seafood, pulses/legumes/nuts, dairy, sugar/honey, and miscellaneous. Oil/fat was unintentionally omitted from the survey and therefore excluded from the composite score. Food consumption categories included traditional foods (e.g., wild caught fish, hunted animals, foraged fruits) and foods commonly consumed in the region (e.g., beans, rice, cassava). The relationship between water insecurity and preparation of water-intensive foods was also explored by creating a binary variable that captured cooking methods requiring water (stewing, boiling, preparing in water) and any foods that were water-intensive to prepare, including rice, lentils, stew, beans, oats, pasta, and cassava.

Adjusted models included sociodemographic covariates selected a priori based on their conceptual relationship to food and water insecurity and using change-in-estimate assessment. Variables that plausibly influence food and water insecurity were selected and sequentially included in models. These included biological sex (male/female), an index for combined wealth and educational status constructed for this specific population that has been used previously utilized [28], household size, and marital status (married/partnered vs. not). These were chosen based on prior studies showing that these factors are correlated with resource insecurities such as food and water insecurity in the wider literature [45–48]. Data were not collected on water source type, distance to water source, market access, and seasonality, which may be potential confounders.

Results

Associations between water insecurity and food insecurity

Bivariate linear regression models with robust standard errors indicated that higher water insecurity scores were associated with higher food insecurity scores (Table 2). For every one-unit increase in water insecurity scores, food insecurity increased by 0.29 points on the 0–12 food insecurity scale (β = 0.29; 95% CI: 0.15, 0.44; p < 0.001). This association remained statistically significant in adjusted models controlling for household size, age, sex, education/SES, marital status, and community-level fixed effects, although the magnitude was slightly attenuated (β = 0.24; 95% CI: 0.11, 0.36; p < 0.001). In practical terms, households experiencing the highest level of water insecurity would be expected to have food insecurity scores approximately 1.65 points higher on the 12-point scale than households reporting no water insecurity experiences, after accounting for sociodemographic characteristics. This suggests that water insecurity may represent a distinct and compounding dimension of household hardship linked to food insecurity.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Adjusted and unadjusted linear regression models estimating robust standard errors of food insecurity and water insecurity among Awajún adults living in the Peruvian Amazon (n = 212).

https://doi.org/10.1371/journal.pwat.0000556.t002

In item-level adjusted models, specific water insecurity experiences associated with higher food insecurity included not collecting water because of distance (β = 0.61; 95% CI: 0.19, 1.03; FDR-adjusted p = 0.019) or illness/weakness (β = 0.68; 95% CI: 0.20, 1.16; FDR-adjusted p = 0.009), and not boiling water (β = 0.60; 95% CI: 0.15, 1.05; p = 0.009), after the Benjamini-Hochberg false discovery rate correction was applied (Table 2).

Water insecurity, dietary diversity, and water-intensive foods

Respondents consumed, on average, approximately half of the 11 available food groups (mean±SD of 5.51 ± 1.97); some of the most common items consumed were cassava (94.6%), sugar (72.3%), rice (64.6%), chicken (45.6%), and eggs (33.5%). Food secure households had higher mean±SD dietary diversity scores than food insecure households (food secure: 6.23 ± 2.02; 95% CI 5.66, 6.79 vs. 5.28 ± 1.90; 95% CI 4.98, 5.57; p = 0.002) (Table 1). The most common water-intensive foods consumed were cassava and rice, and others such as oats (21.4%), noodles (31.6%), beans (6.8%), and stew (4.9%). Although 79.3% of respondents reported stewing, boiling, or preparing foods in water; almost all (97.2%) relied on water-intensive foods. Only 4.4% of respondents reported frying foods as the primary method of preparation.

Composite water insecurity was not associated with dietary diversity nor water-intensive food consumption (Table 3; Table B in Sl Appendix). For individual water insecurity items, “going a long time without drinking” because there was no clean water available was significantly associated with dietary diversity scores, which were 0.79 points lower on average, in comparison to individuals not reporting this experience (β = -0.79; 95% CI: -1.43, -0.16; p = 0.01). This is equivalent to nearly one fewer food group consumed (Table 3). Composite water insecurity was not associated with consumption of any food groups, nor the consumption of water-intensive foods (Tables A & B in S1 Appendix).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Adjusted and unadjusted linear regression models estimating robust standard errors of dietary diversity and water insecurity among Awajún adults living in the Peruvian Amazon (n = 212).

https://doi.org/10.1371/journal.pwat.0000556.t003

Discussion

This study examined the relationship between water insecurity and food insecurity (sufficiency and dietary diversity) among Awajún adults living in the Peruvian Amazon. Consistent with prior literature documenting food and water insecurity among rural populations globally [10], we observed a high prevalence of both forms of resource deprivation in this community at the time of the study. Food insecure households, on average, had lower mean socio-economic status, larger households, and higher composite water insecurity scores. Composite food and water insecurity measures were significantly associated. However, one of the most important findings of this study was that specific constituent measures of the lived experiences of water insecurity were more strongly associated with food insecurity and dietary diversity, than the composite water insecurity score itself. Specifically, reports of not collecting water due to distance, being unable to collect water due to illness/weakness, and being unable to boil household water were significantly associated with reported experiences of food insecurity.

As a whole, these findings add to the small but growing body of work on the relationships between water and food insecurity in water-abundant, climate-vulnerable contexts [2,25], complement a larger body of work documenting this association in broader contexts [10], and contribute to scholarship on the intersection of poverty and multiple forms of resource insecurity among Indigenous and vulnerable populations [14,49]. These findings have both methodological and policy implications. From a methodological perspective, the stronger associations observed for specific experiences of water insecurity, rather than for the composite water insecurity score, suggest that disaggregated analyses may reveal pathways linking water insecurity, food insecurity, and dietary diversity that are obscured when relying solely on aggregate measures. From a policy perspective, identifying the particular dimensions of water insecurity most strongly associated with food insecurity may help practitioners and policymakers design more targeted interventions that address the specific barriers shaping resource insecurities in particular local contexts. As we describe below, this locally contextualized approach is important for scientists, practitioners, and policymakers interested in supporting Sustainable Development Goal 6 (SDG 6: Clean Water and Sanitation) and Sustainable Development Goal 2 (SDG 2: Zero Hunger) across diverse social and ecological contexts.

For example, we found that the experience of not “collecting water due to distance” was associated with a higher risk of experiencing food insecurity. Water collection in many parts of the Amazon rainforest is a physically demanding and time-intensive task. Barriers, such as illness or long distances to water sources, may directly affect food procurement and preparation. Reduced physical capacity or accessibility to water may also constrain agricultural production food preparation [50] and water-related tasks. Ethnographic research from the larger project from which this analysis was drawn, sheds light on the dynamics underlying the varying distances community members need to traverse to collect water in this context. Specifically, community members recounted a long history of wealthier Awajún community members securing plots of land for homes and gardens closer to the river, while poorer households were forced to live further from this primary water source [26]. Thus, despite community water source homogeneity in this study context, actual experiences of water insecurity likely vary by proximity to the river, the primary water source- creating inequities in the time-cost associated with bringing water back to households. These findings suggest that interventions that increase the accessibility of clean and accessible water, such as installing infrastructure for taps in households, may have positive rippling effects on other outcomes of interest, such as food security.

We also found that “not collecting water due to illness/weakness” was significantly associated with higher food insecurity scores. A number of studies have connected water insecurity and gastrointestinal diseases [44,45], vector-borne diseases [45], and injuries [12,46–48]. However, attention to the linkages between illness, challenges collecting water, and food insecurity has been predominantly conceptual [51]. This is surprising given that the impacts of water insecurity on domestic labor, including water collection [14,52], childcare [14,53], economic productivity [14,52,54], and subsistence agriculture [14], have been well-studied. Using human-centered measurement tools in the future, such as the Water Insecurity Experiences (WISE) scales [42,43], which capture lived experiences of water insecurity among similar populations and in comparable ecologies, could yield insight into how experiences of access differ even when water sources are largely equivalent. For now, the finding that “not collecting water due to illness/weakness” is correlated with reported food insecurity highlights how upstream interventions, such as providing accessible and adequate healthcare in remote communities, may provide a buffer against a wide variety of resource insecurities.

Similarly, the finding that “not boiling drinking water”, the primary treatment modality in the study communities, was significantly associated with higher food insecurity scores suggests that water treatment behaviors may reflect broader household resource constraints. In this context, not boiling water may be less a matter of preference or health literacy, and more an indicator of limited capacity to engage in labor- and resource-intensive water treatment practices. For example, households able to consistently boil water may possess greater access to fuel, a factor that also contributes to improved food security. In contexts without formalized water treatment facilities, boiling water may be the most affordable, or in some cases, the only available drinking water treatment option [55]. However, this requires significant physical labor to collect firewood and other fuels [56]. It also involves a great deal of time to boil water, let it cool, and properly store it [57]. In many contexts, households must make trade-offs between energy-intensive, and potentially dangerous activities such as collecting water or firewood [12,58] versus competing economically-productive activities, domestic tasks, or caregiving. In these cases, time constraints and resources stressors may make people less likely to boil water because of the inherent costs of this process, as has been seen in Ethiopia [59]. This information importantly points to the potential value of increasing the accessibility of water treatment options outside of boiling, including the use of chlorine tablets and water filtration systems, such as activated charcoal systems as one potential method to improve food security. These options do have barriers, such as the taste of chlorine and the cost and upkeep needed to procure and maintain water filters. Future research should assess the efficacy and impacts of interventions such as household water filtration systems on the experiences of resources insecurities and human health more broadly.

Finally, study respondents had a heavy reliance on water-intensive foods such as cassava and rice. Boiling and stewing were the most common methods of food preparation in this context. This finding illustrates that water is critical for food preparation, and that unavailability may drive the healthfulness of food preparation techniques (e.g., frying vs. stewing). However, water insecurity was not significantly associated with variations in the consumption of water-intensive food consumption in this study, likely because this is a water-abundant context. Additionally, while we found that dietary diversity was lower among food-insecure and lower SES households, dietary diversity was not associated with composite water insecurity scores.

The only significant association that emerged was that respondents who reported going “extended periods of time without drinking water” had lower dietary diversity scores. This contrasts with prior findings from a study of 27 sites in low- and middle-income countries, which found that composite water insecurity scores were associated with consumption of lower perceived quality foods [10], and other studies documenting associations between water insecurity and lower dietary diversity [20,60]. The divergence in findings is interesting as it may suggest that only the most severe forms of water deprivation, such as going for “going a long time without water”, are relevant for dietary diversity in water-abundant contexts. These findings indicate that while chronic water insecurity (potentially driven by concerns with water safety and ease of access) may be pervasive, it is the intensity and immediacy of water deprivation that may directly constrain dietary behaviors. These findings highlight the importance of comparative studies and the collection of locally specific data to inform policy and infrastructural investments to support SDG 6 (Clean Water and Sanitation) and SDG 2 (Zero Hunger) in particular places

Conclusions

This study contributes new evidence to understanding food and water insecurity in a water-abundant but highly vulnerable environment with implications for initiatives aiming to achieve SDG 6 (Clean Water and Sanitation) and SDG 2 (Zero Hunger). Specifically, both forms of resource insecurity were widespread among the Awajún, likely reflecting the compounding pressures of environmental degradation, socioeconomic marginalization, and ongoing political-economic transformations in this region of the Peruvian Amazon. Importantly, constituent lived experiences of water insecurity—including barriers to water collection, water treatment practices, and extended periods of water deprivation—were more strongly associated with food insecurity and dietary diversity than composite water insecurity scores, suggesting that attention to specific dimensions of water insecurity may improve efforts to identify, monitor, and address resource insecurity in vulnerable populations.

Furthermore, our research importantly echoes prior findings that food and water insecurity are co-occurring, overlapping vulnerabilities. Interventions focused on promoting food security would benefit from addressing water insecurity. Additionally, our findings that specific dimensions of water insecurity were associated with food insecurity scores highlights the need to examine the nuanced relationships between water and food across different contexts. Future analyses will build on the present study using data collected in 2024 following major infrastructural changes in the region, including the installation of a community water system. These data include both the original water insecurity questions used in 2013 and the Household Water Insecurity Experiences (HWISE) Scale, which will allow examination of longer-term changes in water insecurity, assessment of potential improvements associated with infrastructure development, and a comparison of newer experiential measures with earlier approaches to quantifying water insecurity.

In terms of policy implications, investigations of site-specific, and longitudinal, dynamics can be used to understand how experiences of resource insecurities change over time and to design interventions that simultaneously support food and water security. Indeed, policy-makers considering the potential synergies and trade-offs in achieving the SDGs of “Zero Hunger” and “Clean Water and Sanitation” have highlighted the need for, “complementary policies to reconcile lines of action that can make progress toward one SDG while inadvertently moving away from another SDG, as well as identify potential win-wins, where one line of action can make progress toward multiple SDGs simultaneously.” [61] (pg. 280). These types of policies will require locally-specific knowledge of the water-food nexus.

Finally, the finding that specific water insecurity scale items, particularly not collecting water because of distance or illness, were associated with food insecurity in this context may underscore that the pathways linking water and food insecurity are mediated not only by access, but by the time, labor, physical capacity, and risks embedded in daily water collection. Similarly, the association between going “prolonged periods without drinking water” and reduced dietary diversity suggests that acute water deprivation may have more immediate effects on dietary quality than chronic, everyday water stress. Examining the constituent parts of the composite water insecurity scores also revealed association between being unable to boil drinking water and food insecurity, highlighting the possibility that water treatment behaviors may be protective or serve as broader markers of household resilience connected to access to fuel, knowledge, and other resources that support health-related decision-making. Together, these findings reinforce the idea that progress toward SDG 6 (Clean Water and Sanitation) and SDG 2 (Zero Hunger) cannot be pursued in isolation. Clean, accessible, and safely managed water is not only a basic human right but is foundational to achieving nutritional well-being, dietary diversity, and stable food systems. For Indigenous populations such as the Awajún, whose livelihoods, diets, and cultural practices remain closely linked to local ecologies, investments in safe water access, water treatment options, and reliable infrastructure are likely to yield cascading benefits for food security, health, and resilience. Future interventions should integrate experiential measures of water and food insecurity, include human-centered assessments of water access and labor demands, and consider the socio-political drivers that ultimately shape resource inequalities within and between communities.

Supporting information

Acknowledgments

Our deepest appreciation goes to the community members, local partners in Peru who participated in this research, and to the Wenner-Gren Foundation for supporting our work.