Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Jun 1.
Published in final edited form as: Am J Hum Biol. 2021 Dec 23;34(6):e23715. doi: 10.1002/ajhb.23715

Cross-cultural variation in thirst perception in hot-humid and hot-arid environments: Evidence from two small-scale populations

Asher Y Rosinger 1,2,*, Hilary J Bethancourt 1,3, Zane S Swanson 4, Kaylee Lopez 1, W Larry Kenney 5, Tomas Huanca 6, Esther Conde 6, Rosemary Nzunza 7, Emmanuel Ndiema 8, David R Braun 9,10, Herman Pontzer 4,11
PMCID: PMC9177510  NIHMSID: NIHMS1764956  PMID: 34942040

Abstract

Objectives:

Thirst is an evolved central homeostatic feedback system that helps regulate body water for survival. Little research has examined how early development and exposure to extreme environments and water availability affect thirst perception, particularly outside Western settings. Therefore, we compared two indicators of perceived thirst (current thirst and pleasantness of drinking water) using visual scales among Tsimane’ forager-horticulturalists in the hot-humid Bolivian Amazon and Daasanach agro-pastoralists in hot-arid Northern Kenya.

Methods:

We examined how these measures of perceived thirst were associated with hydration status (urine specific gravity), ambient temperatures, birth season, age, and population-specific characteristics for 607 adults (n=378 Tsimane’, n=229 Daasanach) aged 18+ using multi-level mixed-effect regressions.

Results:

Tsimane’ had higher perceived thirst than Daasanach. Across populations, hydration status was unrelated to both measures of thirst. There was a significant interaction between birth season and temperature on pleasantness of drinking water, driven by Kenya data. Daasanach born in the wet season (in utero during less water availability) had blunted pleasantness of drinking water at higher temperatures compared to those born in the dry season (in utero during greater water availability).

Conclusions:

Our findings suggest hydration status is not a reliable predictor of thirst perceptions in extreme-hot environments with ad libitum drinking. Rather, our findings, which require additional confirmation, point to the importance of water availability during gestation in affecting thirst sensitivity to heat and water feedback mechanisms, particularly in arid environments. Thirst regulation will be increasingly important to understand given climate change driven exposures to extreme heat and water insecurity.

Keywords: Thirst, hydration, DOHaD, water, heat

Introduction

Drinking to thirst is an oft-repeated guideline, as this recommendation suits the majority of healthy adults living in a temperate climate (Millard-Stafford et al. 2012). Thirst is a key central homeostatic feedback system that evolved to tightly regulate body fluid and ensure survival, yet there is a learned aspect to it as well, which may vary by environment (Kenney and Chiu 2001). To date, the vast majority of studies have examined thirst in Western, educated, industrialized, rich, and democratic (WEIRD) countries and populations (Armstrong et al. 2020a; Brannigan et al. 2015; Carroll 2020; Carroll et al. 2019; Kenney and Chiu 2001; Martens and Westerterp-Plantenga 2012; Millard-Stafford et al. 2012; Phillips et al. 1984a; Phillips et al. 1984b; Rolls and Phillips 1990; Rolls et al. 1982); therefore, much of our understanding is based on research with limited biological and cultural diversity (Gurven and Lieberman 2020).

There are two types of thirst: homeostatic and non-homeostatic. Homeostatic thirst sensation generally occurs when plasma osmolality, which is a measure of the number of dissolved particles in the blood plasma and a strong driver of the body’s electrolyte-water balance and a stimulus for thirst, increases 1-2% due to a deficit of body water or to decreased plasma volume (Thornton 2010; Zimmerman et al. 2017). Another homeostatic thirst response is anticipatory thirst, which occurs in relation to eating and consuming solutes (Zimmerman et al. 2016). The anticipatory thirst response is driven by neural circuitry as a way for the body to anticipate changes in body fluid and stimulate thirst to increase water intake to return to homeostasis (Gizowski and Bourque 2018; Zimmerman 2020).

The intrauterine environment may be a critical time for influencing the homeostatic thirst setpoint (Rosinger 2020). A developmental origins of health and disease (DOHaD) framework suggests that plasticity during development helps calibrate an individual to their future environment (Kuzawa and Quinn 2009). Thus, differences in the intrauterine environment in terms of water availability and exposure to extreme heat may create variation in thirst setpoints. In fact, experimental animal studies have demonstrated that maternal dehydration in pregnant rats affects the development of thirst responses of offspring (Perillan et al. 2008). These studies demonstrated that thirst setpoints of offspring exposed to intrauterine water restriction were not activated until they reached relatively higher thresholds of dehydration compared to offspring not exposed to intrauterine water restriction. However, when they were activated, they were experienced more strongly (Ross and Desai 2005). Among humans, the intrauterine environment is critical for nephron development and later-life kidney health (Luyckx and Brenner 2015; White et al. 2009). Kidney function and the baroreceptors detecting water deficits to stimulate thirst are thus also potentially modifiable (Thornton 2010).

Yet, outside of extreme water loss, most drinking behavior is non-homeostatic and environmentally driven, often occurring around meal times (Rolls et al. 1982; Stevenson et al. 2015). Much of thirst may thus be habitual or contextual (Carroll 2020). Thirst is also a type of interoceptive sensation or awareness of internal bodily states. This awareness is learned and also likely subject to environmental and cultural variation (Brannigan et al. 2015; Ma-Kellams 2014; Stevenson et al. 2015). Drinking behaviors and cues for drinking are shaped as early as development and ontogeny (Rolls et al. 1982). In humans, as in all mammals who begin life by breastfeeding, suckling is not solely activated by a response to hydration state but rather as a response for feeding (Rolls et al. 1982). However, early feeding practices have significant implications for plasma tonicity and the kidney as well as shaping preferences for dietary solutes and sodium chloride, especially since water is aversive to newborns (Desor et al. 1975). Breastfed compared to formula and solid food fed 1-3 month old babies had significantly lower plasma osmolality (Davies 1973). Thus, early life experiences in utero may be critical for thirst setpoint development as well as learned drinking behavior (Rosinger 2020).

These early behaviors are shaped by cultural dietary and hydration customs. Every culture has specific hydration strategies which are learned. Among small-scale populations hydration strategies are often derived from the local ecology and environmental resources available, especially in places where market beverages, like sodas, are not regularly consumed (Rosinger and Tanner 2015; Swanson and Pontzer 2020). For example, in the Amazon, consuming chicha, a fermented yucca beverage, is a crucial way to stay hydrated (Bethancourt et al. 2021; Christopher et al. 2019; Rosinger and Bethancourt 2020). Among pastoralists, drinking milk as well as coffee or tea represent major water sources (Galvin 1992; Sagawa 2006). Yet, other stimulants are also frequently used across populations to limit or quench thirst while providing energy and focus during work, like chewing on coca or tobacco leaves (Breithoff 2017; Charlton 2004). Thus, culturally-specific customs may help explain how individuals within populations respond to increasing water needs and thirst.

It is unclear if thirst perception differs for populations living in water insecure environments with extreme temperatures. Extreme heat increases water needs (Sawka et al. 2005), yet differences in water availability in arid compared to humid environments may lead to variation in thirst responses as a way to adapt to the higher water needs. Since lifestyles and ambient environments lead to substantially different daily water turnover (Pontzer et al. 2021), we might expect to see differences in thirst across hot-humid and hot-dry environments. The hot-humid environment presents an additional physiological challenge in thermoregulation, such that sweat is harder to evaporate and stays on the skin longer (Best and Kamilar 2018). Thus, there may be a greater awareness of body water loss in hot-humid environments than hot-dry environments, where body water evaporates much quicker and less noticeably (Kenney and Hodgson 1987). In these hot environments, being aware of one’s hydration status and thirst may be more important because of the potential for rapid water loss. Conversely, it may be possible that thirst perception is blunted due to water scarcity, which may lead to a decoupling of thirst from hydration status. However, little research has examined how thirst perception and the desire to drink water is associated with these factors in diverse hot environments.

Therefore, to fill these gaps and examine thirst outside of WEIRD contexts, this study examined perceived thirst in two small-scale populations living in extreme thermal environments. We compared thirst perceptions among Tsimane’ in the hot-humid Bolivian Amazon and Daasanach in the hot-arid Northern Kenya Lake Turkana region. First, we tested how these perceptions of thirst relate to physiological and environmental characteristics that are generally thought to influence thirst, including hydration status, ambient temperature and humidity, and age. Second, since early life environments may shape exposures to water availability, we tested if season of birth moderated the thirst response to ambient temperature across the dry and humid settings. Drawing on research in WEIRD settings and the animal literature, we hypothesized that thirst would increase with higher urine concentration (Armstrong et al. 2020a) and hotter temperatures (Kenney and Chiu 2001), and decrease with age since thirst reportedly declines with age in Western populations (Kenney and Chiu 2001; Rolls and Phillips 1990). Further, we hypothesized that season of birth would modify the relationship between the thirst response and ambient temperatures, such that those who experienced in utero water deficits would have less of a thirst response in relation to higher temperatures (used as a proxy of increasing water needs) (Perillan et al. 2008; Ross and Desai 2005). Finally, we examined how population-specific cultural factors, like traditional hydration practices and stimulant use, were associated with thirst. We hypothesized that increased use of each beverage or stimulant would be associated with lower thirst.

Methods

Ethical approval

The multisite Water Insecurity, Stress, and Hydration (WISH) study was conducted following the principles of the Declaration of Helsinki and was approved by the institutional review boards of Pennsylvania State University and the Kenya Medical Research Institute. Permission was obtained from the Gran Consejo Tsimane' in San Borja, Bolivia, the Director of Health in the county government of Marsabit, Kenya, and community leaders in each the five Tsimane' communities sampled in Bolivia and each of the seven Daasanach communities sampled in Kenya. All participants provided oral informed consent and Daasanach provided written informed consent.

Fieldsites and data collection

The WISH study was an observational cross-sectional study among Tsimane’ in lowland Bolivia and Daasanach in Northern Kenya in 2019 designed to examine how experiences surrounding water and thirst related to hydration and health in Lowland Bolivia and Northern Kenya (Figure 1). For the purposes of this study, we investigated the research questions among adults only as we had limited sample sizes of children with reliable thirst perception responses.

Figure 1:

Figure 1:

Open in a new tab

Fieldsite settings for the WISH study

The fieldsites and comparative data collection procedures are described in detail elsewhere (Bethancourt et al. 2021). In brief, in both settings, data collection procedures were the same; and data was collected in the early part of the dry season in both sites for feasibility reasons and to have comparative seasonal data (April-May in Bolivia; June-July in Kenya). None of the sampled communities in either population had running water or electricity in their homes at the time of study. Both household heads were invited to participate, and each provided urine samples and then rated their thirst independently. Ambient temperature and humidity were recorded at this time. Participants then had their anthropometrics measured and participated in a household survey.

The data collection in the Bolivian fieldsite took place among Tsimane’, an indigenous forager-horticulturalist population that live in the hot-humid Amazon. Approximately 16,000 Tsimane’ live in ~100 villages, of which five communities were sampled for this study (Leonard et al. 2015). We attempted exhaustive sampling in the five Tsimane’ communities and had >90% participation rates.

Tsimane’ have distinct hydration patterns as they consume traditional fermented beverages, like chicha as well as consume water from water-rich fruits (Rosinger and Tanner 2015; Rosinger and Bethancourt 2020). These hydration strategies translate into higher water intake than adults in the US, as one study found they consumed ~4.5 liters of water on a given day (Rosinger and Tanner 2015). Further, prior work has found that approximately 50% of Tsimane’ adults have concentrated urine and ambient temperature was strongly related to hydration status, indicating that it is hard to meet their water needs in this hot-humid environment (Bethancourt et al. 2021; Rosinger 2015).

The data collection in the Kenyan fieldsite was also part of the Daasanach Human Biology Project. Approximately 19,000 Daasanach semi-nomadic pastoralists live in the hot-arid northern Kenya desert across 26 villages around the northeastern margin of Lake Turkana (Kenya National Bureau of Statistics (KNBS) 2019). Similar to many eastern African pastoralist groups, Daasanach rely on milk from cattle, goats, and other livestock as a major caloric and hydration source. Daasanach living in Kenya are distinctly more reliant on these traditional pastoral forms of subsistence than those who reside in the Omo River Basin of southern Ethiopia (Almagor 1978), who have greater access to market and agricultural goods. Additionally, issues of food and water insecurity, which have previously been identified among Daasanach (Bethancourt et al., 2021), are worsened by increased climate variability in the region (MoALF 2017). Further, a tradition of drinking tea/coffee results in a regular consumption of these beverages at least twice per day, including in the morning (Bethancourt et al. 2021).

We sampled 7 Daasanach communities using a random sampling design of every third household, which resulted in sampling 12-28 households per community based on the pre-allocated amount of time the research team had in each location (Bethancourt et al. 2021). Exhaustive sampling in this site was not feasible due to time and resource constraints as several of the Daasanach study communities had >200 households, thus we chose a representative sampling approach.

Outcome: Thirst perceptions

Foundational thirst research demonstrated that thirst sensation is strongly associated with the pleasantness or desire to imbibe water, thus we asked questions using two visual analog categorical scales to measure perceived thirst (current thirst level and perceived pleasantness of drinking water) adapted from Rolls (Phillips et al. 1984a; Phillips et al. 1984b; Rolls et al. 1982). On a laminated piece of paper, we showed participants hybrid visual analog scales with verbal anchors between 1-10 indicating increasing thirst or pleasantness of drinking water (Figure 2). We asked them to point to the ladder to indicate how thirsty they felt in the moment and separately how pleasant it would be to drink water in that moment.

Figure 2:

Figure 2:

Open in a new tab

Adapted visual thirst scales

Prior to data collection, we had extensive discussions with our translators and informants to translate these concepts and words associated with thirst perception in both sites. Among Daasanach, thirst translated into the word ba’are, while the concept of pleasantness of drinking water translated to bie imiiy midhanle. Among Tsimane’, thirst translated to jajriri’, while pleasant to drink water was jäm’dyi’ coi’ ojñi’ tyeij.

Predictors

Hydration status. Participants provided urine samples, which were covered and allowed to cool for ~10-30 minutes. Urine samples were then measured in triplicate with a Pen refractometer (Atago) for urine specific gravity (Usg) with the average used for analysis. Urine specific gravity is the ratio of urine to the density of water. Elevated Usg or concentrated urine, often categorized as hypohydrated was defined as values > 1.020 (Armstrong et al. 2010). Participants were provided with the results of their urine concentration as well as hydration advice and water after they rated their thirst to prevent this from impacting their hydration status and thirst ratings.

Current weather. Ambient temperature and relative humidity were measured in Bolivia with a mercury thermometer and a Fischer hair hygrometer, respectively. In Kenya, they were both measured using a digital Kestrel 3000 Pocket Weather Meter (Kestrel; Boothwyn, PA, USA). We used the ambient temperature (°F) as the primary predictor and adjusted for relative humidity, which can affect heat retention and perceived thirst (Raymond et al. 2020).

Season of birth. As water availability during gestation and rainfall have been shown to affect birth outcomes in humans and thirst setpoints in animals (Godoy et al. 2008; Randell et al. 2020; Rosinger 2020; Ross and Desai 2005), we collected information on whether participants were born in the rainy or dry season, which is commonly known among participants. Being born in the wet season meant that the individual was in utero during the preceding dry season when water was less available to their mother; while being born in the dry season would mean that they were in utero during the rainy season water was more abundant. In Kenya, Daasanach have a major wet season from March through May, when more than half of the annual precipitation falls and a second smaller rainy season in November. In Bolivia, the rainy season lasts from November through March. Using historical monthly rainfall totals, we simulated hypothetical births in the dry and rainy seasons by site to test our assumption of differential water availability by birth season (Supplemental Table 1). Under this hypothetical situation, Daasanach born in the middle of the major wet season would have experienced ~47.5 mm (or 18.9%) less precipitation throughout gestation compared to being born in the dry season – which is consistent with our assumption. Further, in this scenario, the biggest difference for Daasanach was the second trimester where being born in the rainy season might have been associated with 63 mm or 42.4% less rainfall compared to dry season birth. For Tsimane’, we see a similar pattern consistent with our assumption, and the difference in this hypothetical scenario is 702 mm, with the second trimester again being the biggest contributor. Thus, our assumption of differential water availability by birth season holds up if our simulation provides a reasonable approximation of rainfall in the years and seasons in which participants were born.

Age. Since thirst reportedly declines with age in Western populations (Rolls and Phillips 1990), we examined the association for age in years. Many Tsimane' and Daasanach adults do not know their exact age. Because of recent government efforts to provide identification cards, some adults have identification cards on which a birth date is given. For this study, age was estimated by either using the birth date provided on an identification card or by probing about when the participant was born in relation to locally known historical events.

Population-specific predictors

Since both small-scale populations have specific culturally derived traditions for meeting their water needs and dealing with thirst, during surveys we asked adults about the frequency of their intake over the past week or 24 hours depending on the item. For Tsimane’, we inquired about chicha, fruit added to water, and coca leaf use over the past 7 days. Coca leaves are used throughout Bolivia, often as a way to abate thirst and hold off on eating or drinking during labor.

For Daasanach, who have a strong tea/coffee tradition and who rely on livestock and milk as part of their dietary and hydration strategies, we inquired about the number of times they consumed tea/coffee in the prior 24 hours and the frequency of milk consumption in the prior week (categorized as 0, <1 time per day, and 1 or more times a day). Further, we asked about the number of times in the prior week they chewed tobacco. This traditional behavior is a fairly common practice, which can also affect saliva production.

Covariates

We adjusted for demographic and physiological variables that can influence thirst (Kenney and Chiu 2001; Phillips et al. 1984b; Rolls et al. 1982). During the survey, we collected information on sex (male/female). Time of day was recorded at the time of the urine sample/thirst measurement. For the analysis, the time was rounded to the hour. Finally, as fat and fat free mass have drastically different water composition and potential water reserve (Ritz et al. 2008), we adjusted for fat free mass as measured through a bioimpedance scale (Tanita).

Statistical methods

Analysis was conducted in Stata V15.1 (College Station, TX). Statistical significance was set at alpha at 0.05 for a two-tailed test. We used t-tests to assess the association between bivariate season of birth and thirst outcomes as well as test differences in characteristics across the two populations. Linear fit and functional polynomial fit scatterplots were used to visualize the bivariate best-fit associations between variables of interest and dispersion of the data.

We used three-level mixed effect linear regression models with individuals nested within households within communities to estimate how current hydration status (Usg>1.020), ambient temperature, season of birth, and age were associated with current thirst and perceived pleasantness of drinking water (Models 1.1-1.2). The models adjusted for the primary predictors, as well as fat free mass, humidity, time of day, and included fixed effects for population. Further, we controlled for sex and estimated the regression models combined for men and women because there were no sex differences in thirst and there was no sex-specific effect modification with the main variables of interest. Also, neither pregnancy nor lactating status were associated with either thirst measure among women. We next tested for an interaction term between season of birth and ambient temperature (Models 1.3-1.4).

Second, we stratified the regression analyses by population to examine whether the results were consistent across both populations and to further adjust for population-specific factors. We re-estimated the fully adjusted models with the interaction term for both thirst outcomes, while additionally including amount of chicha, fruit-water, and coca use among Tsimane’ and tea/coffee, milk, and chewing tobacco among Daasanach using two-level mixed-effects linear regression models with individuals nested within households (Models 2.1-2.4). We used marginal standardization of the fully adjusted models to estimate and visualize the interactions and examine the predicted thirst levels by season of birth across the range of ambient temperatures (Korn and Graubard 1999; Williams 2012).

Analytic Sample

Our analytic sample included all adults aged 18 years and older with complete covariate data on the predictors of interest. 49 Tsimane’ adults, all from one study community, were excluded because data on season of birth was not collected; this resulted in data being analyzed from 4 of the 5 Tsimane’ study communities; however, those from the excluded study community had similar characteristics (age, sex, thirst ratings). Five Daasanach adults were excluded because of missing data on fat free mass and one due to an error in data entry. Further, an additional 7 Tsimane’ and 4 Daasanach were excluded because they reported that a doctor had told them they had diabetes, which can significantly increase thirst and confound associations. None of these exclusions (fat free mass or diabetes status) significantly differed in terms of their demographics. This left a total analytic sample of 607 adults: 378 Tsimane’ and 229 Daasanach.

Results

The mean age and sex makeup were not different between the Tsimane and Daasanach samples (Table 1). However, fat free mass, total body water, BMI, % born in the wet season, and Usg were all significantly higher among Tsimane’ than Daasanach. Specifically, 46% of Tsimane’ reported that they were born during the rainy season, whereas 31% of Daasanach reported they were born during the rainy season. Daasanach were significantly taller and leaner. The ambient temperature was lower in Bolivia compared with Kenya (82.0 °F ±5.9 vs 90.1 °F ±4.9, p<0.001), while the relative humidity was substantially higher in Bolivia (78.5% ±9.1 vs 47.4% ± 12.4, p<0.001).

Table 1:

Descriptive characteristics of Daasanach and Tsimane’ participants

Variable Daasanach (n=229)
Mean or % (SD)		 Tsimane’ (n=378)
Mean or % (SD)		 p-value^
Age (years) 39.6 (14.9) 37.3 (16.7) 0.10
Sex (% men) 43.7% 49.1% 0.20
Weight 51.1 (8.2) 59.2 (8.8) <0.001
Fat free mass (kg) 42.3 (6.4) 45.9 (7.2) <0.001
TBW (%) 56.9 (6.0) 55.1 (6.2) <0.001
Height (cm) 167.5 (7.6) 156.1 (7.4) <0.001
BMI (kg/m2) 18.2 (2.8) 24.2 (3.0) <0.001
Current thirst perception (1-10) 5.5 (2.7) 6.0 (2.7) 0.03
Perceived pleasantness of drinking water (1-10) 6.0 (1.9) 7.9 (2.3) <0.001
Birth season (% dry season) 31.0% 46.7% <0.001
Urine specific gravity (Usg) 1.011 (0.008) 1.020 (0.007) <0.001
% Usg>1.020 18.3% 55.4% <0.001
Ambient temperature (°F) 90.1 (4.9) 82.0 (5.9) <0.001
Humidity (%) 47.4 (12.4) 78.5 (9.1) <0.001
Tea/coffee consumption last 24 hours (number of times) 2.1 (0.80) -- --
Milk consumption last 7 days: -- --
0 times per day (%) 42.8%
<1 time per day (%) 20.5% --
1+ time per day (%) 36.7% -- --
Chewing tobacco last 7 days 3.3 (3.5) -- --
Chicha dulce consumption last 7 days (number of days) -- 1.3 (1.4) --
Fruit water consumption last 7 days (number of days) -- 2.0 (2.1) --
Chewing coca leaves last 7 days (number of days) -- 0.60 (1.5) --
Open in a new tab
^

2 sided t-test; TBW: Total body water; BMI: Body mass index

Bivariate analyses

In bivariate analyses, current perceived thirst was 0.5 pts higher (t-test=2.16; p=0.031) and the pleasantness of drinking water was 1.9 pts (t-test=10.5; p<0.001) higher among Tsimane’ in the hot-humid environment than among Daasanach in the hot-arid environment (Table 1). Current perceived thirst and pleasantness of drinking water were significantly correlated (r=0.25; p<0.0001) when both populations were combined, with stronger correlation among Daasanach (r=0.35; p<0.001) than among Tsimane’ (r=0.20; p<0.001) (Figure 3).

Figure 3:

Figure 3:

Open in a new tab

Perceived current thirst by perceived pleasantness of drinking water. Linear fit scatterplots with 95% confidence intervals.

Both current thirst perception and perceived pleasantness of drinking water were not significantly different by season of birth in bivariate examination (t-tests p=0.89).

Among both Daasanach and Tsimane’, Usg was not associated with perception of current thirst (Figure 4a) or perceived pleasantness of drinking water (Figure 4b). In fact, the relation between Usg with perceived thirst was fairly flat for both populations, and only mildly positive for perceived pleasantness of drinking water among Tsimane’.

Figure 4:

Figure 4:

Open in a new tab

Urine specific gravity by a) perceived thirst and b) perceived pleasantness of drinking water. Local polynomial smooth plots with confidence intervals.

Ambient temperature was positively associated with perceptions of thirst and pleasantness of drinking water among Daasanach (r=0.22, 0.24, p<0.001 for both) in the hot-dry environment; whereas there was no association (r=−0.03, 0.03; p>0.50 for both) among Tsimane’ in the hot-humid environment (Figure 5a-b).

Figure 5:

Figure 5:

Open in a new tab

Ambient temperature by a) perceived thirst and b) perceived pleasantness of drinking water. Local polynomial smooth plots with confidence intervals.

Relative humidity was inversely associated with thirst perception among Daasanach and slightly positively associated or curvilinear with thirst among Tsimane’ (Supplemental Figure 1a-b). These associations with thirst are likely due to the inverse relationship between ambient temperature and relative humidity (Supp Figure 2a); and the u-shape relations between temperature, humidity, and time of day (Supp Figures 2b-c). Current thirst appeared to increase from the morning to 12-1PM in both populations and then flatten (Supp Figure 3a); whereas the pleasantness of drinking water did not have any association with time of day (Supp Figure 3b).

Age was not consistently associated with either thirst perception or perceived pleasantness of drinking water (Supp Figures 4a-b). Thirst did not appear to decrease markedly with age. Further, when stratifying the sample by age bins (18-39, 40-59, and 60+), individuals 60+ years exhibited a thirst sensitivity to hotter temperatures indicating higher thirst responses among older adults similar to younger adults (Supp Figure 5).

Regression analyses

Results of the 3-level mixed effects regression models indicated that none of the primary predictors or covariates were associated with current thirst level (Table 2, Models 1.1). Having an elevated Usg was not associated with either current thirst or perceived pleasantness of drinking water adjusted for covariates. Being in the Daasanach sample (compared to the Tsimane’ sample) (B=−2.3; SE=0.46; p<0.001) was associated with lower perceived pleasantness of drinking water. Being born in the wet season (compared to being born in the dry season) (B=−0.31; SE=0.17; p=0.07) and ambient temperature approached statistical significance (B10°F=0.41, SE=0.27, p=0.12) (Model 1.2).

Table 2:

Multi-level linear mixed effects regression models examining predictors of current thirst and perceived pleasantness of drinking water

Model 1.1 Model 1.2 Model 1.3 Model 1.4
Current
Thirst		 Perceived
pleasantness		 Current
Thirst		 Perceived
pleasantness
VARIABLES Beta (SE) Beta (SE) Beta (SE) Beta (SE)
Ambient temperature (per 10°F) 0.34 (0.36) 0.41 (0.27) 0.45 (0.39) 0.68**(0.30)
Birth season: Wet Season (dry season reference) 0.06 (0.20) −0.31*(0.17) 2.19 (2.52) 4.23**(2.11)
Wet season by temperature interaction −0.25 (0.30) −0.54**(0.25)
Usg>1.020 0.24 (0.24) 0.12 (0.19) 0.23 (0.24) 0.10 (0.19)
Humidity (per 10%) 0.0001 (0.18) 0.02 (0.13) 0.01 (0.18) 0.04 (0.13)
Age (years) 0.01* (0.01) −0.004 (0.01) 0.01* (0.01) −0.01 (0.01)
Fat-Free Mass 0.001 (0.03) −0.02 (0.02) 0.0003 (0.03) −0.02 (0.02)
Male (female reference) 0.08 (0.35) 0.42 (0.29) 0.10 (0.35) 0.45 (0.29)
Time of day 0.02 (0.05) −0.09** (0.04) 0.02 (0.05) −0.09** (0.04)
Kenya (Bolivia Reference) −0.72 (0.62) −2.29*** (0.46) −0.72 (0.61) −2.27*** (0.46)
Observations 607 607 607 607
Open in a new tab

Standard errors in parentheses

***

p<0.01

**

p<0.05

*

p<0.1

Note: 3-level mixed-effects model, with random effects (intercepts) for each participant nested 356 households and within a random effect for 11 communities of residence. Results for Models 1.1 and 1.2 indicate the association between current thirst and perceived pleasantness of drinking water in relation to ambient temperature, season of birth, hydration status, and controlling for covariates listed. Models 1.3 and 1.4 repeat the models along with an interaction between ambient temperature and birth season.

There was a significant interaction between season of birth and ambient temperature for perceived pleasantness of drinking water (B=−0.54, SE=0.25, p=0.031) but not current thirst (Table 2, Models 1.3-1.4). Adults born in the wet season, which indicates they were in utero during less water availability, had a reduced thirst response to higher ambient temperatures; in contrast, adults born in the dry season (in utero during greater water availability) had a more robust, linear thirst response to higher temperatures. In this model, the main terms of temperature (B10°F=0.68, SE=0.30, p=0.021) and birth season (B=4.2; SE=2.1; p=0.045) were associated with increased pleasantness of drinking water.

Population specific regression analyses

We re-estimated the full models with the interaction between season of birth and temperature stratified by population (Table 3). First, among Tsimane’, none of the primary predictors nor the interaction between ambient temperature and season of birth were significantly related to measures of thirst (Models 2.1, 2.2). Having an elevated Usg (>1.020) and each 10 percentage points of humidity were positively but not significantly associated with both measures of thirst. Further, the population-specific factors that may affect thirst, chicha, fruit-water, and coca leaf use were not associated with either measure of thirst.

Table 3:

Multi-level linear mixed effects regression models examining predictors of current thirst and perceived pleasantness of drinking water among Tsimane’ and Daasanach samples separately

Tsimane’
 Daasanach
Model 2.1 Model 2.2 Model 2.3 Model 2.4
Current
Thirst		 Perceived
pleasantness		 Current
Thirst		 Perceived
pleasantness
VARIABLES Beta (SE) Beta (SE) Beta (SE) Beta (SE)
Ambient temperature (per 10°F) −0.11 (0.45) 0.54 (0.38) 0.88 (0.87) 1.91*** (0.56)
Birth season: Wet Season (dry season reference) -0.42 (3.60) 2.94 (3.07) 3.84 (6.42) 11.31** (5.08)
Wet season by temperature interaction 0.08 (0.44) −0.40 (0.37) −0.44 (0.71) −1.29** (0.56)
Usg>1.020 0.44 (0.29) 0.28 (0.25) −0.08 (0.39) −0.20 (0.32)
Humidity (per 10%) 0.22 (0.24) 0.21 (0.20) −0.27 (0.28) 0.16 (0.17)
Age (years) 0.02 (0.01) −0.003 (0.01) 0.01 (0.01) −0.01 (0.01)
Chicha consumption (per 1 day) −0.10 (0.11) 0.06 (0.09) -- --
Fruit-water consumption (per 1 day) 0.05 (0.08) 0.01 (0.06) -- --
Coca leaf use (per 1 day) 0.02 (0.09) −0.06 (0.08) -- --
Tea/coffee consumption (per 1 time) -- -- −0.28 (0.23) −0.32** (0.15)
Milk consumption (none: ref) -- -- Ref Ref
<1 time per day -- -- −0.28 (0.51) 0.16 (0.33)
1+ times per day -- -- −1.43*** (0.43) 0.22 (0.28)
Chewing tobacco leaf use consumption (per 1 time) -- -- −0.05 (0.05) −0.04 (0.04)
Observations 378 378 228 228
Number of household clusters 222 222 134 134
Open in a new tab

Standard errors in parentheses

***

p<0.01

**

p<0.05

*

p<0.1

Note: 2-level mixed-effects model, with random effects (intercepts) for each participant nested households. Results for Models 1.1 and 1.2 test the association between current thirst and perceived pleasantness of drinking water for Tsimane’ in relation to ambient temperature, season of birth, with an interaction between ambient temperature and birth season, hydration status, culturally-specific variables, and controlling for covariates listed as well as fat free mass, sex, and time of day. Models 1.3 and 1.4 repeat the models for Daasanach.

Among Daasanach, the primary predictor results were consistent to the pooled models (Models 2.3, 2.4). However, milk consumption level was significantly associated with current thirst perception, as those who consumed milk one or more times per day reported 1.4 pts (SE=0.43; p=0.001) lower current thirst than those who reported not consuming any milk per day (Model 2.3). Further, each instance of tea/coffee consumption in the prior 24 hours was associated with 0.32-pts (SE=0.15; p=0.034) lower pleasantness of drinking water. Chewing tobacco was not associated with either thirst measure. Ambient temperature and season of birth, as well as the interaction (B=−1.3; SE=0.15; P=0.021) between ambient temperature and season of birth were all statistically significant in the Daasanach-specific model for perceived pleasantness of drinking water (Table 3, Models 2.4). This interaction demonstrates significantly different slopes in the relation between ambient temperature and thirst by season of birth for Daasanach adults (Figure 6).

Figure 6: Season of birth interacts with ambient temperature in relation to predicted perceived pleasantness of drinking water among Daasanach.

Figure 6:

Open in a new tab

Notes: Plot generated using marginal standardization of OLS regression results presented in Table 3 Model 2.4.

Discussion

This study aimed to compare perceptions of thirst in two extreme thermal environments -- hot-dry compared to hot-humid environment -- outside of the typically studied WEIRD settings. We examined how environmental, physiological, and culturally-specific factors were associated with current thirst and perceived pleasantness of drinking water. We found that thirst ratings, particularly perceived pleasantness of drinking water, were higher among Tsimane’ in the hot-humid Amazon than among Daasanach in hot-arid northern Kenya. Second, in contrast to our hypothesis of increasing thirst with greater Usg, our findings suggest that current hydration status is not a reliable predictor of thirst or perceived pleasantness of drinking water in non-WEIRD hot environments during times with free-access to water. Third, in partial support of our hypothesis that season of birth moderated the association between ambient temperature and thirst, there was a significant interaction between season of birth and ambient temperature on the perceived pleasantness of drinking water, but not current thirst. This interaction was driven by the association among Daasanach in the hot-arid environment. In contrast to our hypothesis that thirst would decrease with age, we did not find that thirst consistently decreased with age; in fact, it increased among older adults in relation to hotter temperatures. Finally, in partial support of our hypothesis that culturally-specific factors would reduce thirst, we found some evidence that tea/coffee and milk consumption were associated with lower thirst ratings among Daasanach.

The two measures of thirst, current thirst sensation and perceived pleasantness of drinking water, were significantly correlated (r=0.25; P<0.001) but not as strongly related (with the former explaining only 6% of the variation in the latter in adjusted models) as prior work has demonstrated (Phillips et al. 1984b). This implies that in these non-WEIRD settings, they are potentially more independent constructs than previously assumed. As thirst is driven by homeostatic and non-homeostatic elements, it is possible that in these settings, these two questions were useful at capturing the two components of thirst, but future work is needed to assess this hypothesis. That the perceived pleasantness of drinking water was substantially higher (~2.4 points) among Tsimane’ than Daasanach is important to explore.

Urine concentration, as well as humidity levels, were both significantly higher among Tsimane’ than Daasanach, yet Usg and humidity were not associated with either measure of thirst among Tsimane’ in either the pooled and or the stratified models. Further, it is unclear why Usg and humidity would be driving the higher perceived pleasantness of drinking water but not current thirst. We adjusted for culturally-specific factors that may affect current thirst sensation and perceived pleasantness of drinking water in the population-specific models as a way to capture some of the cultural variation and behaviors that may modify thirst perception. We found no evidence that the number of days in the prior week of consuming chicha, fruit-water, or chewing coca leaves was associated with thirst among Tsimane’. However, we did find among Daasanach that daily milk consumption was associated with lower current thirst, while greater amount of tea/coffee consumption in the prior day was associated with lower pleasantness of drinking water (indicating less of a desire to consume water). Since the tea/coffee they drink is lower in caffeine than in Western settings as it is made with coffee husks, it would not reach the level needed to cause diuresis and serves as an important hydration source (Killer et al. 2014; Ruxton et al. 2015). These findings imply that those specific hydration strategies may be useful to lower thirst and the desire to imbibe water.

Hydration, thirst, and interoception

None of the physiological factors, including hydration status, we examined were associated with thirst. Our finding that current hydration status as measured through Usg was not associated with thirst is surprising only when compared to experimental studies of induced dehydration. Normal plasma osmolality ranges between 275-295 mOsm/kg with values above 295 considered dehydrated, yet a meta-analysis found that thirst sensation is activated at a plasma osmolality mean of 285.2 mOsm/kg and that this threshold overlaps with increased secretion of arginine vasopressin (Hughes et al. 2018). One twin study found that thirst threshold, the plasma osmolality point at which the thirst sensation kicked in, was heritable; but that thirst sensitivity, the slope of the relation between osmolality and thirst, was not (Zerbe et al. 1991). Participants in experimental studies which induced water losses to a hypohydrated state consistently reported higher thirst ratings (Armstrong et al. 2020a; Carroll et al. 2019). Recent work demonstrated that thirst perceptions are significantly higher beginning prior to the 1-2% body mass decrease from water restriction generally associated with the thirst response, and that changes in Usg and Posm during water restriction were associated with greater perceptual thirst and pleasantness of drinking water (Armstrong et al. 2020a). However, prior work has indicated that during free access to water, rather than in response to water challenges, body-fluid variables like plasma osmolality and angiotensin II concentrations did not relate to thirst perceptions but rather were influenced by anticipatory thirst as well as surrounding eating (Phillips et al. 1984b). Recent experimental work has demonstrated that after rehydration following water restriction, Posm was dissociated with thirst ratings which suggests that thirst is more perceptual and not as directly tied to hydration status during ad libitum water drinking (Armstrong et al. 2020b). Those findings parallel our own that hydration as measured via Usg during an ad libitum state in extreme thermal environments were not associated with thirst perceptions.

Interestingly, our prior work in these populations indicated that current hydration status was associated with temperature and thus was sensitive to changes in the environment (Bethancourt et al. 2021). These results imply that Usg may not be physiologically coupled with thirst in ad libitum hot environments. This also demonstrates the importance of behavioral, cultural, and environmental cues surrounding drinking water and how thirst is a multi-faceted construct driven by all of these cues. When water is not restricted, thirst may result from any number of these factors and mask our ability to detect the homeostatic thirst relationship.

Variation in interoceptive accuracy is another potential explanation for the finding that thirst and Usg were not associated in these extreme thermal environments. Interoception is one’s awareness and ability to detect changes in internal bodily states, in this case thirst (Brannigan et al. 2015; Stevenson et al. 2015). The perception and sensitivity to the thirst sensation is what drives individuals to imbibe fluid. Evidence exists that learning and development within social groups and early life experiences can shape a wide range of behaviors, including interoception (Harshaw 2008; Ma-Kellams 2014).

While we could not find studies comparing thirst across WEIRD and non-WEIRD settings, prior work has found that Western and non-Western groups exhibit varying levels of interoceptive awareness and accuracy (Ma-Kellams 2014). For example, Ma-Kellams’ (2014) review of interoception found that both West-African and East-Asian groups had higher levels of somatic sensitivity and awareness than European-Americans. In those examples, the West-African sample studied placed greater emphasis on bodily cues, like heart rate and fatigue within their language and medicine – with more holistic explanations for emotions and bodily states. However, there is relatively scant literature comparing interoceptive accuracy (i.e., how well a person is able to detect changes in bodily states). In one of these studies, West-African groups showed lower interoceptive accuracy in identifying changes in heart rate than their European-American counterparts (Chentsova-Dutton and Dzokoto 2014). This could be attributed to incorrectly identifying the cause of their bodily changes or that they are more aware of their bodies at all times and thus the greater awareness masks acute changes (Ma-Kellams 2014). Thus, it is also possible that Tsimane’ and Daasanach vary in both their interoceptive accuracy and awareness as it relates to thirst and their urinary markers of hydration status. Further, there could be differences in expression across their languages that may affect the numerical score they provide. Yet, our use of the relative visual ladder and that the perceived thirst scores were more similar than pleasantness of drinking water scores help mitigate this potential limitation. Future work should measure thirst multiple times throughout a day alongside changes in environmental and hydration variables to better untangle how cultural understandings of thirst perception, including both interoceptive sensitivity and accuracy, changes in dynamic states.

Developmental origins of thirst

To examine homeostatic thirst and whether there was a developmental driver of the thirst setpoint, we tested how thirst was associated with ambient temperature and humidity and whether season of birth modified this response. Ambient temperature, but not humidity, was associated with perceived pleasantness of drinking water. Further, we found a significant interaction between season of birth and ambient temperature, which was driven by the hot-arid environment where water scarcity is a bigger problem. Daasanach born in the wet season (i.e., in utero during less water availability) reported a positive, yet blunted perceived pleasantness of drinking water with rising temperatures compared to those born in the dry season (in utero during greater water availability), translating to 1-2 point lower perceived pleasantness ratings at 95-105°F.

One potential reason why we may have found this interaction between birth season and ambient temperature on pleasantness of drinking water among Daasanach but not Tsimane’ is that the Bolivian Amazon has substantially more water year-round even in the dry season than in Northern Kenya. In San Borja, Bolivia, the closest weather station, the mean annual rainfall of 2205 mm is nearly seven times higher than the likely overestimated 333 mm in Northern Kenya from a combination of the Lodwar weather station and historical local weather data from Turkana, Kenya on the other side of Lake Turkana (Leslie and Fry 1989) (Supplemental Table 1). Therefore, among Tsimane’, differences in season of birth may matter less and exert less of an effect in utero, i.e., not be as significant a stressor as they are in an arid environment, on thirst responses to heat because of widespread water availability.

Our findings of a blunted thirst response to a situation of higher water needs (i.e., hotter temperatures) are similar to animal models (Perillan et al. 2008; Ross and Desai 2005). Animal studies have demonstrated that maternal dehydration in rat dams affects the development of thirst responses of offspring, such that those exposed to in utero dehydration did not exhibit higher thirst responses in response to their own dehydration following birth (Perillan et al. 2008). This indicates that the adaptive thirst response may be blunted among those who experience water scarcity in utero.

Prior work examining developmental influences on nutrition and health, notably work in the developmental origins of health and disease (Kuzawa and Quinn 2009), demonstrate how critical the maternal nutritional environment is to later life health (Thayer et al. 2020). The fetal environment and maternal exposure to heat and precipitation have been shown to affect birth outcomes (Randell et al. 2020). For example, in Ethiopia, hotter temperatures throughout pregnancy, particularly in the first and third trimesters, were associated with greater odds of stunting, while greater rainfall in utero was associated with less stunting (Randell et al. 2020). Further, the in utero environment, i.e., water restriction particularly in the second trimester onward, is linked to changes in thirst setpoints and water intake behaviors of offspring (Ramirez et al. 2002; Rosinger 2020; Ross and Desai 2005; Xu et al. 2000). Animal studies have shown that this manifests in offspring through a lower plasma osmolality threshold at which arginine vasopressin is excreted to correct body water losses (Ross and Desai 2005). The thirst setpoint and sensitivity to external environments is a likely mechanism driving the water intake behaviors. Our data suggest that broad categorization of water availability in utero, particularly in water scarce environments, may have critical implications for how thirst, and consequently, water drinking patterns develop in later life. However, our findings highlight the need for more careful studies which examine dosage effects, linkage of climate data to individual records, and examination of effects by trimester of exposure to further untangle the developmental origins of thirst and human water needs.

Aging, thirst, and heat outside the Western context

In WEIRD settings, the sensation of thirst (Rolls and Phillips 1990) and water drinking (Rosinger and Herrick 2016) changes with age. Older adults exhibit reduced thresholds and sensitivity to somatic changes and drink less fluid compared to younger persons (Kenney and Chiu 2001; Rolls and Phillips 1990). We did not find that age was significantly associated with thirst ratings. However, among Daasanach, there was some evidence of a quadratic relationship with age – peaking around 30 years of age and then slowly declining, whereas, thirst was slightly positively associated with age among Tsimane’ (Supp Figure 4a-b). Kenney and Chiu (2001) demonstrated that osmoregulation is still present in older adults, but it just occurs later than younger adults due to a reduced ability to detect a volume deficit. In our study, as it was ad libitum and not under conditions of water restriction, the only test of water stress was to examine how thirst varied in response to ambient temperature. When the relation of ambient temperature to thirst and perceived pleasantness of drinking water was examined stratified by age group, adults aged 60+ years demonstrated increasing thirst with hotter temperatures (Supp Figure 5). This indicated higher thirst responses among older adults similar or even greater than among younger and middle-aged adults.

Prior work has shown that other physiological processes connected to thirst, like urine concentration (Rosinger et al. 2019) and blood pressure (Gurven et al. 2012; Raichlen et al. 2017), do not decline or increase, respectively, at the same rate in small-scale societies as in WEIRD settings due to more active lifestyles. It is possible that thirst sensitivity remains more tightly connected to ambient temperatures throughout the lifecourse in small-scale societies in comparison to WEIRD settings due to their non-industrial lifestyles and lack of reliance on climate control.

Given ongoing climate change and record-setting high temperatures, a greater understanding of thirst in extreme thermal conditions is critical, particularly among vulnerable groups like older adults whose thirst is more uncoupled from hydration status (Kenney et al. 2014). Extreme heat and rising temperatures will likely affect thirst and human water needs, and lead to higher risk of dehydration before thirst signals kick in. Prior work demonstrated the tight connection between hydration status and ambient temperatures among Tsimane’, illustrating the challenge of keeping up with water needs as the day progressed and generally not rehydrating until the evening (Rosinger 2015). Yet, hotter projected night-time temperatures will make this rehydration and cooling off even more challenging. Further, problems with water availability (Konapala et al. 2020) and water insecurity may increase stress surrounding how one will meet their water needs and result in changes to thirst perception (Rosinger and Young 2020). It will be important for future work to untangle the effects of extreme heat and water insecurity on thirst regulation patterns, especially in light of how extreme heat and dehydration contributes to kidney disease (Johnson et al. 2019).

Limitations and future directions

As this is a cross-sectional, observational study, we cannot infer causality and all relationships should be viewed as associations. Water intake was ad libitum and we did not measure thirst in response to water deprivation or induced dehydration. While we did not induce water restriction, our sampling over a range of ambient temperatures allowed us to examine how thirst varied in relation to ambient environments as hotter temperatures increase water needs (Kenney and Hodgson 1987). Further, we do not have repeat measurements of thirst in different contexts among the same people, which limits our ability to infer how individuals’ thirst within these populations respond to changing conditions. We also did not measure an individual’s physical activity levels over the prior 24 hours, which is associated with water needs. Both small-scale populations are moderately active, and the study was conducted with ad libitum access to water and food – meaning that if participants were particularly active prior to our study, they would have had access to the water and food in their homes before coming to the interview site. Thus, there should not be systematic differences of increased physical activity associated with participating in our study by community or population. Nor did we measure their intake of liquids or hydrating foods in the previous 24 hours. However, we do control for time of day when an individual was sampled, which adjusts for potential differences in time to the last meal as well as broad physical activity patterns. Future work in these settings should examine repeated assessments of thirst in relation to different ambient environments and control for physical activity and recent water intake.

While Usg is a reliable measure of hydration status, it may lag behind plasma osmolality by 30 minutes (Oppliger et al. 2005). This would matter more during periods of rapid water loss or gain rather than in ad libitum settings. Future work should examine other markers of hydration status like plasma osmolality, but if possible do so in more non-invasive ways than venous blood draws to minimize discomfort associated with the research.

Different dry and wet seasons vary in water availability, temperature, and duration over time. Thus our measure of birth season may be subject to some non-differential misclassification bias as we do not have the precise water availability or heat stress for each individual, nor in which trimesters of gestation they experienced the majority of conditions, which may bias our estimates towards the null (Szklo and Nieto 2014). Future work in the DOHaD field should try to utilize natural experiments with higher resolution data of temperature and water availability during gestation and trimester and subsequent thirst perceptions. These relationships should also be explored among children and in longitudinal studies to examine how thirst changes throughout development.

Finally, it is hard to disentangle environmental effects (e.g. humidity) from culturally-specific behaviors and norms of thirst, which may be specific to each population in explaining the population level differences between Tsimane’ and Daasanach. For example, the morning tea drinking ritual among Daasanach, which is not present among Tsimane’, was associated with lower thirst. Finally, water temperature has an impact on palatability of drinking water in terms of how much water volume is consumed (Rolls and Rolls 1982), and may have unconsciously affected thirst perception scores if there were differences in envisioned water temperature across individuals or populations. However, as our scales were validated as measures of thirst and hypothetical water intake, there’s no mention of water volume or what the water temperature would be. Further, since both settings are without refrigeration, water would be ambient temperature and thus this potential bias is unlikely this would have influenced our findings. However, we can at least be sure that our measures of thirst are not driven by anticipatory thirst as participants did not eat food during the hour of data collection.

Conclusions

This study presents some of the first research of thirst in non-WEIRD settings and contributes to an emerging framework in the human biology of water to better understand how water needs are shaped, vary, and met (Rosinger and Brewis 2020). This study demonstrated that thirst varied cross-culturally and was greater in a hot-humid setting than a hot-arid environment among small-scale populations. In the hot-dry environment, we found that milk and tea/coffee consumption were associated with lower thirst. Our findings suggest that thirst sensation and the perceived pleasantness of drinking water are not coupled to current hydration status in ad-libitum conditions. Further, we did not find that thirst decreased with age. Rather, thirst may be more responsive to current ambient environment depending on water availability and exposure to heat in utero which may affect sensitivity to heat and water feedback mechanisms throughout the lifecourse, particularly in arid environments, but more work is needed to confirm this. Thirst regulation will be increasingly important to understand given climate change driven exposures to extreme heat and water insecurity.

Supplementary Material

supinfo

Acknowledgments and funding information

The authors would like to thank the Gran Consejo Tsimane’, the Kenya Medical Research Institute, the National Museums of Kenya, the Illeret Health Clinic, our translators (Manuel Roca Moye, Elias Hiza Nate, Robin Nate Roca, Luke Lomeiku, Samuel Esho, and Joshua Koribok), research assistants (Jessica Saunders, Shiva Dhanasekar, Celine LaTona, Alysha Kelyman, Kaitlyn Barnhart, Jason John), community leaders from each participating community, and all study participants. Thanks also to Barbara Rolls for helpful conversations around thirst.

This work was funded by the National Science Foundation (NSF ARCH #1624398; NSF REU #1930719; NSF CNH2-S #1924322) and a Pennsylvania State University Social Science Research Institute (SSRI) Human Health and Environment Seed Grant. This work was supported by the Ann Atherton Hertzler Early Career Professorship funds and Penn State’s Population Research Institute (NICHD P2CHD041025). The funders had no role in the research or interpretation of results.

Footnotes

Conflicts of Interest

The authors report no conflicts of interest.

Data Availability Statement

Data available upon reasonable request due to privacy/ethical restrictions.

References:

  1. Almagor U 1978. Pastoral partners: affinity and bond partnership among the Dassanetch of South-West Ethiopia: Manchester University Press. [Google Scholar]
  2. Armstrong LE, Giersch GEW, Colburn AT, Lopez V, Sekiguchi Y, Munoz CX, and Lee EC. 2020a. Progression of human subjective perceptions during euhydration, mild dehydration, and drinking. Physiology & Behavior:113211. [DOI] [PubMed] [Google Scholar]
  3. Armstrong LE, Giersch GEW, Dunn L, Fiol A, Muñoz CX, and Lee EC. 2020b. Inputs to Thirst and Drinking during Water Restriction and Rehydration. Nutrients 12(9):2554. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Armstrong LE, Pumerantz AC, Fiala KA, Roti MW, Kavouras SA, Casa DJ, and Maresh CM. 2010. Human hydration indices: acute and longitudinal reference values. Int J Sport Nutr Exerc Metab 20(2):145–153. [DOI] [PubMed] [Google Scholar]
  5. Best A, and Kamilar JM. 2018. The evolution of eccrine sweat glands in human and nonhuman primates. Journal of Human Evolution 117:33–43. [DOI] [PubMed] [Google Scholar]
  6. Bethancourt HJ, Swanson ZS, Nzunza R, Huanca T, Conde E, Kenney WL, Young SL, Ndiema E, Braun D, Pontzer H et al. . 2021. Hydration in relation to water insecurity, heat index, and lactation status in two small-scale populations in hot-humid and hot-arid environments. American Journal of Human Biology 33(1):e23447. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Brannigan M, Stevenson RJ, and Francis H. 2015. Thirst interoception and its relationship to a Western-style diet. Physiology & Behavior 139:423–429. [DOI] [PubMed] [Google Scholar]
  8. Breithoff E. 2017. The ‘white death’: Thirst and water in the Chaco War. Modern Conflict and the Senses: Routledge. p 213–228. [Google Scholar]
  9. Carroll H 2020. Redefining thirst: A conceptual four-compartment model characterising types of thirst, and their underlying mechanisms and interactions. [Google Scholar]
  10. Carroll HA, Templeman I, Chen Y-C, Edinburgh R, Burch EK, Jewitt JT, Povey G, Robinson TD, Dooley WL, and Buckley C. 2019. Hydration status affects thirst and salt preference but not energy intake or postprandial ghrelin in healthy adults: A randomised crossover trial. Physiology & behavior 212:112725. [DOI] [PubMed] [Google Scholar]
  11. Charlton A. 2004. Medicinal uses of tobacco in history. Journal of the royal society of medicine 97(6):292–296. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Chentsova-Dutton YE, and Dzokoto V. 2014. Listen to your heart: The cultural shaping of interoceptive awareness and accuracy. Emotion 14(4):666. [DOI] [PubMed] [Google Scholar]
  13. Christopher L, Madimenos FC, Bribiescas RG, Urlacher SS, Snodgrass JJ, Sugiyama LS, and Pontzer H. 2019. High energy requirements and water throughput of adult Shuar forager-horticulturalists of Amazonian Ecuador. American Journal of Human Biology 31(2):e23223. [DOI] [PubMed] [Google Scholar]
  14. Davies DP. 1973. Plasma Osmolality and Feeding Practices of Healthy Infants in First Three Months of Life. British Medical Journal 2(5862):340–342. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Desor J, Maller O, and Andrews K. 1975. Ingestive responses of human newborns to salty, sour, and bitter stimuli. Journal of comparative and physiological psychology 89(8):966. [DOI] [PubMed] [Google Scholar]
  16. Galvin KA. 1992. Nutritional ecology of pastoralists in dry tropical Africa. American Journal of Human Biology 4(2):209–221. [DOI] [PubMed] [Google Scholar]
  17. Gizowski C, and Bourque CW. 2018. The neural basis of homeostatic and anticipatory thirst. Nature Reviews Nephrology 14(1):11–25. [DOI] [PubMed] [Google Scholar]
  18. Godoy R, Tanner S, Reyes-García V, Leonard WR, McDade TW, Vento M, Broesch J, Fitzpatrick IC, Giovannini P, Huanca T et al. . 2008. The effect of rainfall during gestation and early childhood on adult height in a foraging and horticultural society of the Bolivian Amazon. American Journal of Human Biology 20(1):23–34. [DOI] [PubMed] [Google Scholar]
  19. Gurven M, Blackwell AD, Eid Rodríguez D, Stieglitz J, and Kaplan H. 2012. Does Blood Pressure Inevitably Rise With Age? Longitudinal Evidence Among Forager-Horticulturalists. Hypertension 60 (1):25–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Gurven MD, and Lieberman DE. 2020. WEIRD bodies: mismatch, medicine and missing diversity. Evolution and Human Behavior 41(5), 330–340. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Harshaw C 2008. Alimentary epigenetics: A developmental psychobiological systems view of the perception of hunger, thirst and satiety. Developmental Review 28(4):541–569. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hughes F, Mythen M, and Montgomery H. 2018. The sensitivity of the human thirst response to changes in plasma osmolality: a systematic review. Perioperative Medicine 7(1):1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Johnson RJ, Sánchez-Lozada LG, Newman LS, Lanaspa MA, Diaz HF, Lemery J, Rodriguez-Iturbe B, Tolan DR, Butler-Dawson J, Sato Y et al. . 2019. Climate Change and the Kidney. Annals of Nutrition and Metabolism 74(suppl 3)(3):38–44. [DOI] [PubMed] [Google Scholar]
  24. Kenney WL, and Chiu P. 2001. Influence of age on thirst and fluid intake. Medicine & Science in Sports & Exercise 33(9):1524–1532. [DOI] [PubMed] [Google Scholar]
  25. Kenney WL, Craighead DH, and Alexander LM. 2014. Heat Waves, Aging, and Human Cardiovascular Health. Medicine & Science in Sports & Exercise 46(10):1891–1899. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kenney WL, and Hodgson JL. 1987. Heat Tolerance, Thermoregulation and Ageing. Sports Medicine 4(6):446–456. [DOI] [PubMed] [Google Scholar]
  27. Kenya National Bureau of Statistics (KNBS). 2019. Ethnic Affiliation. Census. Nairobi, Kenya: Kenya National Bureau of Statistics (KNBS) [Google Scholar]
  28. Killer SC, Blannin AK, and Jeukendrup AE. 2014. No evidence of dehydration with moderate daily coffee intake: A counterbalanced cross-over study in a free-living population. PLoS ONE 9(1):e84154. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Konapala G, Mishra AK, Wada Y, and Mann ME. 2020. Climate change will affect global water availability through compounding changes in seasonal precipitation and evaporation. Nature Communications 11(1):3044. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Korn EL, and Graubard BI. 1999. Predictive margins (direct standardization). Analysis of health surveys. New York, NY: John Wiley & Sons, Inc. p 126–140. [Google Scholar]
  31. Kuzawa CW, and Quinn EA. 2009. Developmental Origins of Adult Function and Health: Evolutionary Hypotheses. Annual Review of Anthropology 38(1):131–147. [Google Scholar]
  32. Leonard WR, Reyes-García V, Tanner S, Rosinger A, Schultz A, Vadez V, Zhang R, and Godoy R. 2015. The Tsimane’ Amazonian Panel Study (TAPS): Nine years (2002–2010) of annual data available to the public. Economics & Human Biology 19:51–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Leslie PW, and Fry PH. 1989. Extreme seasonality of births among nomadic Turkana pastoralists. American Journal of Physical Anthropology 79(1):103–115. [DOI] [PubMed] [Google Scholar]
  34. Luyckx VA, and Brenner BM. 2015. Birth weight, malnutrition and kidney-associated outcomes—a global concern. Nature Reviews Nephrology 11(3):135. [DOI] [PubMed] [Google Scholar]
  35. Ma-Kellams C 2014. Cross-cultural differences in somatic awareness and interoceptive accuracy: a review of the literature and directions for future research. Frontiers in Psychology 5(1379). [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Martens MJI, and Westerterp-Plantenga MS. 2012. Mode of consumption plays a role in alleviating hunger and thirst. Obesity 20(3):517–524. [DOI] [PubMed] [Google Scholar]
  37. Millard-Stafford M, Wendland DM, O'Dea NK, and Norman TL. 2012. Thirst and hydration status in everyday life. S147–S151 p. [DOI] [PubMed] [Google Scholar]
  38. MoALF. 2017. Climate Risk Profile for Marsabit County. Kenya County Climate Risk Profile Series. . In: The Ministry of Agriculture LaFM, editor. Nairobi, Kenya. [Google Scholar]
  39. Oppliger RA, Magnes SA, Popowski LA, and Gisolfi CV. 2005. Accuracy of Urine Specific Gravity and Osmolality as Indicators of Hydration Status. International Journal of Sport Nutrition and Exercise Metabolism 15(3):236–251. [DOI] [PubMed] [Google Scholar]
  40. Perillan C, Costales M, Vijande M, and Arguelles J. 2008. In utero extracellular dehydration modifies thirst in neonatal rats. Appetite 51(3):599–603. [DOI] [PubMed] [Google Scholar]
  41. Phillips PA, Rolls BJ, Ledingham JGG, Forsling ML, Morton JJ, Crowe MJ, and Wollner L. 1984a. Reduced Thirst after Water Deprivation in Healthy Elderly Men. New England Journal of Medicine 311(12):753–759. [DOI] [PubMed] [Google Scholar]
  42. Phillips PA, Rolls BJ, Ledingham JGG, and Morton JJ. 1984b. Body fluid changes, thirst and drinking in man during free access to water. Physiology & Behavior 33(3):357–363. [DOI] [PubMed] [Google Scholar]
  43. Pontzer H, Brown MH, Wood BM, Raichlen DA, Mabulla AZP, Harris JA, Dunsworth H, Hare B, Walker K, Luke A et al. . 2021. Evolution of water conservation in humans. Current Biology 31(8), 1804–1810. [DOI] [PubMed] [Google Scholar]
  44. Raichlen DA, Pontzer H, Harris JA, Mabulla AZP, Marlowe FW, Josh Snodgrass J, Eick G, Colette Berbesque J, Sancilio A, and Wood BM. 2017. Physical activity patterns and biomarkers of cardiovascular disease risk in hunter-gatherers. American Journal of Human Biology 29(2):e22919. [DOI] [PubMed] [Google Scholar]
  45. Ramirez BA, Wang S, Kallichanda N, and Ross MG. 2002. Chronic in utero plasma hyperosmolality alters hypothalamic arginine vasopressin synthesis and pituitary arginine vasopressin content in newborn lambs. American Journal of Obstetrics and Gynecology 187(1):191–196. [DOI] [PubMed] [Google Scholar]
  46. Randell H, Gray C, and Grace K. 2020. Stunted from the start: Early life weather conditions and child undernutrition in Ethiopia. Social Science & Medicine 261:113234. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Raymond C, Matthews T, and Horton RM. 2020. The emergence of heat and humidity too severe for human tolerance. Science Advances 6(19):eaaw1838. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Ritz P, Berrut G, Tack I, Arnaud M, and Tichet J. 2008. Influence of gender and body composition on hydration and body water spaces. Clinical nutrition 27(5):740–746. [DOI] [PubMed] [Google Scholar]
  49. Rolls B, and Phillips P. 1990. Aging and disturbances of thirst and fluid balance. Nutr Rev 48(3):137 – 144. [DOI] [PubMed] [Google Scholar]
  50. Rolls BJ, Rolls BJ, and Rolls ET. 1982. Thirst: CUP Archive. [Google Scholar]
  51. Rosinger A 2015. Heat and hydration status: Predictors of repeated measures of urine specific gravity among Tsimane’ adults in the Bolivian Amazon. Am J Phys Anthropol 158(4):696–707. [DOI] [PubMed] [Google Scholar]
  52. Rosinger A, and Herrick K. 2016. Daily Water Intake Among U.S. Men and Women, 2009–2012. NCHS data brief Hyattsville, MD: National Center for Health Statistics; 242. [PubMed] [Google Scholar]
  53. Rosinger A, and Tanner S. 2015. Water from fruit or the river? Examining hydration strategies and gastrointestinal illness among Tsimane’ adults in the Bolivian Amazon. Public Health Nutr 18(06):1098–1108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Rosinger AY. 2020. Biobehavioral variation in human water needs: How adaptations, early life environments, and the life course affect body water homeostasis. American Journal of Human Biology 32(1):e23338. [DOI] [PubMed] [Google Scholar]
  55. Rosinger AY, and Bethancourt HJ. 2020. Chicha as Water: Traditional Fermented Beer Consumption Among Forager-Horticulturalists in the Bolivian Amazon. Alcohol and Humans: Oxford University Press. p 147–162. [Google Scholar]
  56. Rosinger AY, and Brewis A. 2020. Life and death: Toward a human biology of water. American Journal of Human Biology 32(1):e23361. [DOI] [PubMed] [Google Scholar]
  57. Rosinger AY, Pontzer H, Raichlen DA, Wood BM, Tanner SN, and Sands JM. 2019. Age-related decline in urine concentration may not be universal: Comparative study from the U.S. and two small-scale societies. American Journal of Physical Anthropology 168(4):705–716. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Rosinger AY, and Young SL. 2020. The toll of household water insecurity on health and human biology: Current understandings and future direction. WIREs Water. [Google Scholar]
  59. Ross MG, and Desai M. 2005. Gestational programming: population survival effects of drought and famine during pregnancy. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology 288(1):R25–R33. [DOI] [PubMed] [Google Scholar]
  60. Ruxton C, Phillips F, and Bond T. 2015. Is tea a healthy source of hydration? Nutrition Bulletin 40(3):166–176. [Google Scholar]
  61. Sagawa T 2006. Wives' Domestic and Political Activities at Home: The Space of Coffee Drinking among the Daasanetch of Southwestern Ethiopia. African Study Monographs 27(2):63–86. [Google Scholar]
  62. Sawka MN, Cheuvront SN, and Carter R. 2005. Human water needs. Nutr Rev 63(suppl 1):S30–S39. [DOI] [PubMed] [Google Scholar]
  63. Stevenson RJ, Mahmut M, and Rooney K. 2015. Individual differences in the interoceptive states of hunger, fullness and thirst. Appetite 95:44–57. [DOI] [PubMed] [Google Scholar]
  64. Swanson ZS, and Pontzer H. 2020. Water turnover among human populations: Effects of environment and lifestyle. American Journal of Human Biology 32(1):e23365. [DOI] [PubMed] [Google Scholar]
  65. Szklo M, and Nieto FJ. 2014. Epidemiology: beyond the basics: Jones & Bartlett Publishers. [Google Scholar]
  66. Thayer ZM, Rutherford J, and Kuzawa CW. 2020. The Maternal Nutritional Buffering Model: an evolutionary framework for pregnancy nutritional intervention. Evolution, Medicine, and Public Health 2020(1):14–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Thornton SN. 2010. Thirst and hydration: Physiology and consequences of dysfunction. Physiology & Behavior 100(1):15–21. [DOI] [PubMed] [Google Scholar]
  68. White SL, Perkovic V, Cass A, Chang CL, Poulter NR, Spector T, Haysom L, Craig JC, Salmi IA, Chadban SJ et al. . 2009. Is Low Birth Weight an Antecedent of CKD in Later Life? A Systematic Review of Observational Studies. American Journal of Kidney Diseases 54(2):248–261. [DOI] [PubMed] [Google Scholar]
  69. Williams R 2012. Using the margins command to estimate and interpret adjusted predictions and marginal effects. The Stata Journal 12(2):308–331. [Google Scholar]
  70. Xu Z, Glenda C, Day L, Yao J, and Ross MG. 2000. Osmotic threshold and sensitivity for vasopressin release and Fos expression by hypertonic NaCl in ovine fetus. American Journal of Physiology-Endocrinology and Metabolism 279(6):E1207–E1215. [DOI] [PubMed] [Google Scholar]
  71. Zerbe R, Miller J, and Robertson G. 1991. The reproducibility and heritability of individual differences in osmoregulatory function in normal human subjects. The Journal of laboratory and clinical medicine 117(1):51–59. [PubMed] [Google Scholar]
  72. Zimmerman CA. 2020. The origins of thirst. Science 370(6512):45–46. [DOI] [PubMed] [Google Scholar]
  73. Zimmerman CA, Leib DE, and Knight ZA. 2017. Neural circuits underlying thirst and fluid homeostasis. Nature Reviews Neuroscience 18(8):459–469. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Zimmerman CA, Lin Y-C, Leib DE, Guo L, Huey EL, Daly GE, Chen Y, and Knight ZA. 2016. Thirst neurons anticipate the homeostatic consequences of eating and drinking. Nature 537(7622):680–684. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supinfo
NIHMS1764956-supplement-supinfo.docx (2MB, docx)

Data Availability Statement

Data available upon reasonable request due to privacy/ethical restrictions.

RESOURCES