AMERICAN JOURNAL OF HUMAN BIOLOGY 10:151161 (1998)

Hydration as a Limiting Factor in Lactation

G.R. BENTLEY*
Department of Biological Anthropology, University of Cambridge,
Cambridge, United Kingdom

ABSTRACT
A relatively untapped area of research concerning lactation
and hydration exists for bioanthropologists interested in the ecology of
breastfeeding. This review details current limited knowledge on the topic
with recommendations about relevant research questions for human biologists. On average, lactating mothers produce >750 ml/day milk for their infants. Breastfeeding thus promotes a powerful thirst stimulus in the lactating mother, resulting in a 1216% increase in fluid intake among Western
women. Thirst during lactation may be mediated by oxytocin release, since
this hormone is structurally similar to the antidiuretic hormone, vasopressin.
Prolactin also may be involved. A few studies among Western women concluded that moderate dehydration does not affect milk production, but it is
not known how lactating women in hot and/or dry climates respond to chronically limited water supplies. Lactating women in such environments may use
both physiological and behavioral adaptations to conserve fluid requirements, such as reducing activities during breastfeeding and carrying extra
fluid supplies while travelling. Given the limited state of knowledge on lactation and hydration, future field studies should incorporate research to determine the importance of adequate fluids in maintaining milk production
and the health of both mothers and infants. Am. J. Hum. Biol. 10:151161,
1998.
1998 Wiley-Liss, Inc.
INTRODUCTION: REVIEWING THE DATA
Although there is a rapidly growing literature on the ecology and anthropology of
breastfeeding (e.g., Maher, 1992; Vitzthum,
1994; Stuart-Macadam and Dettwyler,
1995), there has been little research on the
effects of hydrational stress on human lactation. Why this topic has received less attention is surprising when one considers the
numerous studies by both clinicians and human biologists on the physiology of pregnant
women and their fluid requirements (e.g.,
Davison et al., 1988; Lindheimer et al.,
1991), and the effects of nutritional stress
on human lactational performance (e.g.,
Butte et al., 1984; Brown and Dewey, 1992;
Prentice et al., 1994). There are, however,
many interesting questions concerning hydration and lactation that have not been addressed in studies of human biology.
From the microevolutionary perspective,
certain human populations that have been
long-term inhabitants of desert areas are either better adapted, or better acclimatized

1998 Wiley-Liss, Inc.

PROD #730

to withstand a lower intake of water than

others (Frisancho, 1993). Whether this extends to water requirements during lactation is unknown. In hot, dry areas, water
holes used by our foraging ancestors may
have been both widely dispersed on the
landscape as well as shared with predators.
Visits to such holes may, therefore, have
been limited. Even today, humans frequently inhabit heat- and/or water-stressed
environments ranging from desert habitats
to arid high altitudes where the effects of
such stresses have long been acknowledged
(e.g., Little, 1980; Weiner, 1980). But anthropological studies of human nutrition,
whether in hot or moderate climates, rarely
include information on fluid intake. In 1980,
Weiner pointed out that there were no stud-

Contract grant sponsor: Royal Society Research Fellowship.

*Correspondence to: G.R. Bentley, Department of Biological
Anthropology, University of Cambridge, Downing Street, Cambridge CB2 3DZ, U.K. E-mail: grb20@cam.ac.uk
Received 7 October 1996; Accepted 18 January 1997

152

G.R. BENTLEY

ies of water intake or dehydration among

native savanna inhabitants who have heavy
workloads under hot ambient temperatures,
and the situation does not seem to have
changed since that date. Given their cultural capacity, human populations in waterstressed environments also are likely to
have developed behavioral adaptations to
deal with sparse fluid supplies.
Adequate hydration for mothers is obviously needed to maintain milk yields for infants and young children, particularly
where water supplies that may otherwise be
used for infant supplementation may be
contaminated. Given the additional fluid requirements of lactating women, how do they
deal with limited water supplies in such environments? What kinds of physiological
and behavioral strategies have evolved in
response to these stresses? Such questions
should provide excellent topics for study by
human biologists and anthropologists who
often work among populations where water
may be in short supply.
The purpose of this review is, therefore, to
detail what little is currently known about
lactation and hydration and to make recommendations about the kinds of research
questions that human biologists and/or biological anthropologists could address to illuminate this topic further.
FLUID REQUIREMENTS OF
LACTATING WOMEN
How much extra fluid does a lactating
woman need, given that average milk output during peak lactation, defined at 3
months, before supplementation occurred,
has been reported as 785 ml/24 hours (Prentice et al., 1986; Lawrence, 1994)? In general, fluid intake for nonlactating individuals varies significantly between 600 and
6,000 ml/day (Stumbo et al., 1985). Recommended water intake for adults under average environmental conditions and energy
expenditure are 1 ml/kcal of energy expenditure, with the recommendation that lactating women consume up to 1,000 extra ml/
day based on a mean milk secretion of 750
ml/day during the first 6 months of lactation
(FNB, 1989).
Two recent studies report the average
daily water intake of lactating Western
women, one based on observations of 26 lactating women (Stumbo et al., 1985) and the
other based on survey data of 77 lactating
women and 6,201 nonpregnant, nonlactat-

ing controls (Ershow et al., 1991). In the

first study, the 26 subjects consumed an average 2,860 108 (SE) ml/day of water, comprising 2,220 110 ml/day from beverages
and 640 48 ml/day from food. There was
greater inter- than intrawoman variation in
fluid consumption. Water intake averaged
1.37 0.08 ml/kcal ingested, but averaged
only 1.06 0.04 ml/kcal of energy expended,
although estimates of energy requirements
were based on indirect calculations of basal
metabolism. There were no seasonal variations in fluid intake, despite the fact that
the study was undertaken in Iowa where
summers are hot. However, some of the
women had air-conditioning available, and
it appears that none were engaged in any
strenuous outdoor activities.
The second study relied on data collected
for the 197778 USDA Nationwide Food
Consumption Survey, which used food diaries maintained for 3 days and where intakes were somewhat underestimated (Ershow et al., 1991). Water intake was calculated from both beverages (including free
water consumption) and solid foods. Lactating women drank a mean 2,242 658 (SD)
ml/day compared to 1,940 686 ml/day for
nonlactating controls. Total water intake
per unit of energy ingested was identical between lactating and nonlactating controls at
1.3 ml/kcal/day, very similar to the findings
in Stumbo et al. (1985). Water intake varied
by region with the lowest intake in the
northeastern United States (1,827 819 ml/
day) and the highest in the west (2,036
1,096 ml/day), presumably related to temperature differentials. Women drank more,
but not significantly more, water during the
summer months.
Unpublished data from the Dunn Nutrition Unit (courtesy of Gail Goldberg and Andrew Coward) also yield information about
the water intakes of lactating women compared to the nonpregnant, nonlactating
state (Table 1) (see Goldberg et al., 1991).
Water intake was calculated from selfreports of food and beverage consumption.
Mean water intake at 12 weeks postpartum
for lactating women was 2,712 793 (SD)
ml/day compared to 2,413 279 ml/day
when lactation had ceased. The differences
are not significant, although the sample is
extremely small (one-tailed, paired t-test, P
4 0.1, n 4 8). Water intake for the women
averaged 1.06 ml/kcal of energy intake and
1.28 ml/kcal of total energy expenditure

153

LACTATION AND HYDRATION

TABLE 1. Data on fluid intake, total body water, and water turnover rates for 10 lactating, Cambridge women
at 12 weeks postpartum and when nonlactating, nonpregnant at 52 weeks postpartum1
Fluid intake (ml/day)
Subject
1
2
3
4
5
6
7
8
9
10
Mean
SD

Lactating
2,462
1,803
2,505
4,775
2,912
2,805
2,310
2,814
2,445
2,286
2,712
793

NPNL2
2,205
2,389
2,984
2,194
2,452
2,091
2,557
2,439
2,414
279

Total body water (liters)

Water turnover rates (ml/day)

Lactating

NPNL

Lactating

NPNL

59.4
70.4
57.0
56.9
57.2
59.3
54.6
56.3
51.9
63.8
58.7
5.2

59.6
78.1
59.1
57.5
59.5
59.9
64.9
54.2
57.4
63.7
61.4
6.6

3,020
2,945
3,120
5,533
3,512
3,428
2,919
3,455
3,464
3,123
3,452*
766

2,981
2,794
2,910
3,611
3,018
2,983
2,532
2,885
2,930
2,749
2,939
277

Data provided by Gail Goldberg and Andrew Coward of the Dunn Nutrition Unit, Cambridge.
2
Nonpregnant, nonlactating.
*Significant differences between the means of lactating and NPNL women (p < 0.001) (one-tailed, paired t-test).

(calculated using doubly labelled water),

compared to 1.11 ml/kcal of energy intake
for the same women when nonpregnant and
nonlactating and 1.06 ml/kcal of total energy expenditure. Differences between water intake per unit of energy expenditure
were significant between the lactating and
nonlactating state (one-tailed, paired t-test,
P 4 0.02).
Other studies of lactating mothers, some
of which are discussed below, found that under normal circumstances lactating women
drank between 1,400 and 2,300 ml/day,
which appears comparable to the data presented above (Olsen, 1940; Lelong et al.,
1949; Duckman and Hubbard, 1950; Illingworth and Kilpatrick, 1953; Dusdieker et
al., 1990). There is little doubt, therefore,
that lactating women (at least in the West)
drink more than nonlactating controls.
To ensure that additional hydrational
needs are met, breastfeeding promotes a
powerful thirst stimulus. A recent clinical
study confirmed this dipsogenic effect, previously reported anecdotally by midwives
and lactating women (James et al., 1995).
Eleven lactating women participated in the
study with each woman acting as her own
control. Thirst was quantified using a visual
analog scale (see Rolls and Rolls, 1982);
measurements of plasma vasopressin,
plasma oxytocin (OT), plasma osmolality,
and hematocrit as well as blood pressure,
milk intake by the babies, and volume of
water ingested by women were also made.
Thirst ratings were evaluated during a nonsuckling control period (average of 44 min)
and a suckling period (average of 56 min).
Thirst ratings were significantly higher

during the suckling period, and all mothers

drank following breastfeeding an average of
540 255 (SD) ml compared to only two
mothers during the control period, 133 and
351 ml, respectively. The mean difference in
water intake between the suckling and control period was 491 75 (SE) ml. There was
no significant difference in mean levels of
plasma AVP between the control and suckling periods, but there was a significant difference in plasma OT. There was also an
evident relationship between levels of thirst
and OT, although significance tests could
not be assessed due to nonindependence of
the data.
Based on the study by Ershow et al.
(1991) and data from the Dunn Nutrition
Unit, under normal conditions lactating
women need to ingest 1216% more fluids
compared to nonlactating, nonpregnant controls. In addition, the act of breastfeeding
raises the thirst stimulus twofold, judging
from the visual analog scale used by James
et al. (1995). Given these findings, lactating
women must either meet these additional
demands by ingesting their apparent fluid
needs, or must make some kind of physiological or behavioral adaptations to reduce
water requirements.
FLUID REQUIREMENTS OF INFANTS
Breast milk not only provides the sole
source of nutrients for exclusively breastfed
infants, but also their only source of fluids.
Infants are particularly susceptible to dehydration since, compared to adults, they have
a much larger surface area in relation to
body size with resulting greater evaporative
loss. In addition, young infants do not pos-

154

G.R. BENTLEY

sess the ability to control urine osmolality

fully, thereby increasing their risk for dehydration (Paneth, 1980). Many mothers
supplement their infants with bottles of water or other fluids between feeds, although
this is unnecessary, since breast milk provides all the fluid requirements of infants
even in hot and humid environments (Almroth, 1978). Moreover, supplementation
with contaminated fluid significantly increases the risk of infant morbidity (Forman
et al., 1984; Jason et al., 1984; Bhan et al.,
1986). Other cultural factors, such as swaddling young infants in hot climates or mothers work patterns that affect feeding frequencies, may exacerbate evaporative loss
and increase the risk for potential dehydration. Many weanling foods are also likely to
have a high fluid content, and the introduction of food supplements is now recommended at between 4 and 6 months of age
(Lawrence, 1994).
There is, however, significant individual
variation in milk intake among infants,
with a range from 4501,200 ml/day, although exactly what factors control this
variation are unknown (Dewey and Lonnerdal, 1986). Woolridge (1995) has hypothesized that infant regulation of breastfeeding patterns may vary according to climatic
conditions, although the patterns from
available data on this issue are not clear.
Thus infants in hot climates with high
evaporative rates might feed more frequently to maximize water throughput. In
addition, Woolridge (1995) suggests that
milk quality and composition also may vary
with climatic conditions matching the prevailing needs of the infants according to the
environments to which they are exposed.
This kind of co-adaptation would reflect a
plasticity in breastfeeding that would optimize infant survival in different environments. Such co-adaptations may not be evident in Western cultures that have tended
to impose artificial feeding regimens on
nursing mothers and their infants.
In recent years, the clinical literature has
reported a number of cases of severe hypernatremic dehydration of infants, particularly in newborns (e.g., Anand et al., 1980;
Paneth, 1980; Mercier et al., 1986; Thullen,
1988). In some cases, lack of sufficient milk
production by mothers appears to be responsible. In a few cases, the sodium content of
the mothers milk is abnormally high, findings that have not yet been fully explained.

In general, however, such cases should be

regarded as rare.
STUDIES ON LACTATION
AND HYDRATION
There are only six articles in the lactation
literature that deal specifically with the
topic of dehydration and milk production in
human subjects, four experimental and two
observational. Methodologies have varied
considerably from study to study; most were
undertaken during the 1940s. The first experimental study manipulated the fluid intake of 13 mothers for a period of 310 days.
These women were residing in an infants
home attached to a hospital in Denmark
(Olsen, 1940). The lowest amount of fluid
given to the subjects was 600 ml. Urine loss
or other insensible water losses were not
measured. Milk output was determined by
test-weighing of the babies, who were aged
between 1 month and 4 months. Measurements were, therefore, made close to the
time of peak lactational output. Normal
fluid intake of the women was 1,400 ml/day
(although this measurement did not include
fluid ingested from food).
The second experimental study focused on
a single female subject who was a wetnurse
at a hospital in France (Lelong et al., 1949).
The report does not state specifically how
milk output was measured, but the macronutrient content of the milk was analyzed
and urine output was also measured. During two separate phases of the study, the
subjects fluid intake was monitored at normal intake (between 2,000 and 2,600 ml/
day), and then successively reduced over a
15-day period to a minimum of 400 ml/day
for a period of 6 days.
In a third, more recent study designed to
assess the effects of excess and restricted
fluid intake on lactational performance and
serum prolactin (PRL) levels, 11 lactating
women were given a restricted fluid intake
of 800 ml/day for 3 days starting with the
fourth day postpartum (Horowitz et al.,
1980). Urine osmolality was measured as
were serum PRL levels. Milk output was obtained by weighing the babies.
Each of these three studies came to similar conclusions. Milk output is not significantly affected by moderately reduced fluid
intake, even though excessive thirst may be
experienced by the mothers (Table 2).
The fourth study was a somewhat indirect
measure of milk production, since it exam-

155

LACTATION AND HYDRATION

TABLE 2. Studies of effect of fluid restriction on milk yield among human subjects

Source
Olsen (1940)
Lelong et al. (1949)
Horowitz et al. (1980)
Duckman et al. (1950)

No.
subjects

Period of
restriction
(days)

Normal
fluid
intake (ml)

Restricted
fluid
intake (ml)

Normal
milk
yield (ml)

13
1
11
89

310
6
3
8

1,400
2,300

2,000

600
400
800
<1,500

686/day
1,000/day
68/feed

ined the effect of dehydration on breast engorgement among mothers attempting to

stop nursing (Duckman et al., 1950).
Eighty-nine women were given <1,500 ml/
day of liquid compared to two control
groups, one of which drank just as much as
desired (from 1,500 to 2,500 ml/day), and a
third group that drank more than was required (between 2,500 and 5,000 ml/day).
There were no differences between the
groups in the degree of breast engorgement
or swiftness of milk drying up (Table 2).
Therefore, restricting water intake did not
facilitate the process of stopping milk production, or ease the discomfort felt by the
mothers involved.
The consistent results that moderate dehydration does not affect milk production
are perhaps not surprising from an evolutionary perspective. Given the already large
reproductive investment made by a successfully parturient and lactating mother, it
should be expected that physiological adaptations would arise to ensure the survival of
her infant. This perspective is certainly upheld by current knowledge of the effects of
nutritional stress on lactational performance, which shows similarly that lactation
is remarkably resilient against chronic nutritional stress (Butte et al., 1984; Brown
and Dewey, 1992; Prentice et al., 1994). But
for human biologists, this conclusion is not
wholly satisfactory. To begin with, the experimental studies undertaken among
healthy Western women, one of whom was a
professional wetnurse, are not representative of women living in conditions of chronic
water shortage. Nor do they address the
question of exactly how a lactating woman
is able to maintain normal milk production
with a low fluid intake.
There are only two other formal studies of
lactation and dehydration that come close to
examining women in a more traditional setting. The first was an observational study of
milk production conducted in the Gambia by
Prentice et al. (1984). Ten lactating women
and 10 nonlactating, nonpregnant controls,

Restricted
milk
yield (ml)
682/day
1,000/day
73/feed

all of whom were fasting during Ramadan,

were studied. Although conditions of fluid
shortage during Ramadan are culturally
imposed, one may argue that some human
populations live, and have lived, in environments where drinking may similarly be limited to certain short periods.
The Gambian women fasted from 5:00
a.m. to 7:30 p.m., consuming neither food
nor drink during this time. The data are
somewhat hard to interpret since milk yield
does appear to decline. However, this apparently reflects the fact that study periods
were each separated by a months interval;
that there was a strong seasonal effect since
women were spending more time in their
fields during Ramadan and nursing less;
and that women were also in late stages of
lactation and were probably reducing periods of suckling (Ann Prentice, pers. comm.).
Indeed, the authors concluded that milk
output was not affected by this short but
repeated period of fasting.
The second study compared milk composition, but not total output, among Bedouin
women in the Negev Desert of Israel during
summer and winter (Yagil et al., 1986). The
project was designed to examine the water
content of milk among moderately dehydrated humans and cows. Dairy researchers
had reported that cows milk becomes more
dilute under conditions of moderate dehydration (Yagil et al., 1986). This condition
was also known among camels during periods of prolonged dehydration (Yagil and Etzion, 1980). The researchers wanted to confirm this phenomenon and see if it occurred
among humans. In this study, Bedouin
women acted as their own controls, and
milk samples were taken during winter
when average temperatures were 18C
(66F) and in summer when average temperatures were 40C (110F). Relative humidity was 75% in the winter and 45% in
the summer. To estimate relative degrees of
dehydration, urine samples were collected
to measure sodium and potassium content,
as well as osmolality.

156

G.R. BENTLEY

TABLE 3. Milk composition in winter and summer

among Bedouin women1
Temperature (F)
Humidity (%)
Water (%)
Lactose (g/l)
Fat (%)

Winter

Summer

% Change

66
75
85
57
4

110
45
88
64
2

+67***
+30***
+4**
+12*
42***

1
Source: Yagil et al. (1984).
*P 4 0.05; **P 4 0.01; ***P 4 0.001.

Similar to cows and camels, the human

milk became more dilute under conditions of
moderate dehydration (Table 3). In summer, water content of the milk was significantly higher than in winter; the lactose
content was also significantly higher and
the fat content was lower. In addition, sodium and potassium content of urine as well
as urine osmolality increased significantly
during the summer. The study concluded
that dilution of milk is an adaptive strategy
to ensure adequate hydration of offspring in
hot, dry conditions. Unfortunately, the quality of the data collected is questionable.
Milk samples were collected using spot sampling techniques so that there may have
been differences in whether fore or hind
milk was collected. Moreover, exactly how
much fluid the Bedouin women were ingesting or how much milk their infants were
getting was not known. This study thus
raises more questions than it answers.
In sum, very little is known about the
kinds of physiological and behavioral strategies that lactating women use in heatstressed environments in order to offset the
effects of dehydration and to ensure an adequate milk supply to their offspring.
Clearly, further work on this topic needs to
be undertaken among lactating women in
heat-stressed populations.
ADAPTIVE STRATEGIES TO ENSURE
ADEQUATE LACTATION IN CONDITIONS
OF MODERATE DEHYDRATION
Physiological adaptations
Humans have few adaptive mechanisms
either to store or conserve water loss so that
such losses must be made up relatively
quickly (Rolls and Rolls, 1982). As a result,
thirst rapidly follows water depletion to ensure the maintenance of water balance. In
the absence of available fluids, death will
occur when water deficits reach 20% of total body weight (Adolph, 1947). In hot climates, evaporative losses are high to ensure

against hyperthermia, so that the organism

is required to ingest additional fluids, often
in great quantities (Better, 1987). Thus an
adequate water supply is essential. Some
populations are clearly better adapted and/
or acclimatized to heat stress, such that
evaporative losses as well as mineral and
salt losses are substantially decreased (Frisancho, 1993).
The primary physiological mechanism to
conserve water is through the concentration
of urine, but this mechanism follows the initial loss of both intracellular and extracellular water. Under normal hydration, human plasma has an osmolality of 290
mOsm/kg water (Grossman, 1990). With
fluid loss through evaporation and so forth,
however, the concentration of plasma solutes increases, resulting in the diffusion of
water into plasma from the interstitial
fluid. As the latter also becomes more concentrated, water diffuses into the interstitial spaces from cellular fluid. Increasing
plasma osmolality triggers the release of the
antidiuretic pituitary hormone, arginine vasopressin (AVP), which in turn stimulates
the thirst response (Rolls and Rolls, 1982;
Grossman, 1990). Relative hydration in individuals is thus usually assessed through
measures of plasma and urine osmolality,
sodium and potassium content of urine,
plasma levels of AVP, and degree of thirst.
Thirst, however, is not always an accurate
measure of the degree of dehydration experienced by individuals (Wardlaw and Insel,
1990). The osmolality of human milk also
matches that of plasma (Lawrence, 1994), so
that measurement of milk osmolality can be
used as another indicator of relative hydration.
One of the primary roles of AVP is to act
on the kidney to retain water and to increase the sodium and chloride content of
urine. Urine osmolality can thus vary
widely, from a minimum of 50 mosmol/kg
to a maximum of 1,200 mosmol/kg (Garrow
and James, 1993). AVP has other important
functions: it acts on the cardiovascular system as a potent vasoconstrictor and can increase arterial blood pressure and decrease
cardiac output and heart rate (Cunningham
and Sawchencko, 1991; Share and Crofton,
1993). The plasma osmotic threshold for
both AVP secretion and the stimulation of
thirst are highly correlated (Thompson et
al., 1986). There is, however, considerable
interindividual variation in both the sensi-

LACTATION AND HYDRATION

tivity and threshold of the osmoregulatory

system, which appears to be genetically
based (Baylis, 1987; Zerbe et al., 1991).
Such genetic variation suggests that microevolutionary changes could contribute to interpopulation differences in response to hydrational stress.
Are there mechanisms beyond these normal responses to dehydration that are specific to lactating women? One answer may
lie in the structural similarity between AVP
and OT, the latter being primarily associated with the letdown reflex during suckling. But OT also has complementary roles
to AVP, since it has an independent antidiuretic response to plasma hypovolemia and
plasma osmolality and enhances the natriuretic effect of AVP on the kidney (Cunningham and Sawchencko, 1991; Windle et al.,
1995). This hormonal similarity has stimulated considerable research on the relationship between these two neuropeptides (Pedersen et al., 1992; North et al., 1993; Gainer
and Wray, 1994). During lactation, OT may
in fact substitute for AVP in stimulating the
thirst response in lactating women. For example, in a study of thirst during breastfeeding, James et al. (1995) found a significant increase in plasma OT levels during
the nursing periods, but no increase in
plasma AVP or osmoregulatory factors.
There was also a significant correlation between the thirst response of lactating mothers and OT release. The authors suggest
that OT may stimulate thirst during lactation via an afferent from an OT cell. Health
workers have also noticed the strong association between thirst and the letdown reflex that accompanies the release of OT in
nursing mothers (Mulford, 1990).
In addition, PRL, which rises sharply during suckling and plays a critical role in the
maintenance of milk production (Tucker,
1994), may be active in areas of fluid and
electrolyte metabolism, although this area
of research remains highly controversial.
PRL was once claimed to have a independent antidiuretic effects on the kidney
analogous to the role of AVP (Horrobin et
al., 1971; Miller and van Gemert, 1976).
These early studies, however, involved
ovine preparations of PRL that were contaminated with AVP (Berl et al., 1976;
Keeler and Wilson, 1976; Carey et al., 1977;
Vorherr et al., 1978). Nonetheless, some researchers still maintain an independent osmoregulatory role for PRL (Horrobin, 1980;

157

Loretz and Bern, 1982; Nagy et al., 1992).

Others claim that PRL has a potentiating
effect on AVP (Horrobin, 1980; North et al.,
1981) and perhaps an independent dipsogenic effect (Kaufman et al., 1981; Adler et
al., 1986).
PRL is also known to rise during exercise
(Sanfilippo and Nealon, 1985; Smallridge et
al., 1985), an effect that is increased in
warm conditions and is thought to be related to heat stress (Melin et al., 1988). This
increase may be related to the apparent role
of PRL in regulating electrolyte balance in
sweat secretions (Robertson, 1989; Falk et
al., 1991). Given these potentially important actions, the naturally high circulating
levels of PRL produced during lactation may
play a subsidiary role in maintaining water
and/or electrolyte balance, particularly for
women living in hot ambient temperatures.
Kaufman and MacKay (1983) and Kaufman
et al. (1981) argue that PRL is responsible
for the dipsogenic effect of lactation. More
research, however, is needed in this area.
One problem in trying to assess physiological adaptations to hydrational stress
among lactating women is the lack of nonlactating controls in the six studies described in the previous section. It is, therefore, difficult to separate out effects that are
due simply to dehydration, or effects that
are specific to dehydrated lactating women.
The Gambian study did include nonlactating controls, but this study is also confounded by the fact that women were restricting their food intake in addition to fluids, which is likely to have an independent
effect (Prentice et al., 1984).
Several clues to the types of physiological
adaptations to low fluid intake among
breastfeeding humans can be derived from
the literature on other lactating mammals.
Unlike the sparse literature on hydration
and lactation among humans, the dairy industry has promoted a great deal of research on this topic among milk-producing
animals. For example, the desert-adapted
Bedouin goat uses a number of different
strategies to maintain an adequate milk
supply. Some of these are obviously inapplicable to humans, such as the ability to store
water in the rumen for long periods of time,
but others may be more generalized and
amenable to research. Specifically, during
lactation this species of goat increases total
body water (TBW) and plasma volume in
comparison to nonlactating controls in order

158

G.R. BENTLEY

TABLE 4. Effect of fluid restriction on body

composition among lactating and nonlactating
Gambian women during Ramadan1

% Body weight loss

(07001900 hr)
% Total body water
Water turnover rate
(liter/24hr)

Lactating
women

Nonlactating
controls

4.9 0.2
64.3 1.5

3.8 0.3*
61.0 2.0

6.4 0.3

4.44 0.3*

Source: Prentice et al. (1984); results show mean values SE.

*P < 0.01.

to increase water storage (Maltz and Shkolnick, 1980). In addition, total water turnover including milk output of the lactating
black Bedouin goat is more than twice that
of controls. The lactating goats also lose less
water in urine and through evaporation
compared to nonlactating females. In the
dehydrated, lactating black Moroccan goat,
plasma osmolality and sodium content rise
to higher levels among lactating animals
than controls. This, of course, may simply
reflect a greater degree of stress among the
former (Hossaini-Hilali et al., 1994).
Are any of these findings applicable to lactating women? In the Gambian study, serum osmolality and sodium concentration
increased significantly only for the lactating
women during the fasting period, matching
the findings for the black Moroccan goat. In
addition, lactating women lost 7.6% of TBW
during the fasting period, whereas controls
lost significantly less, suggesting that TBW
in humans also may act as an important
reservoir for lactational needs (Prentice et
al., 1984). Unfortunately, there is so much
variation in measures of TBW that it is impossible to compare other studies of lactating and nonlactating women (but see below). The lactating Gambian women also
had a significantly higher water turnover
(by two liters) compared to controls, which
matches the findings for hydrationally
stressed goats (Table 4).
Lactating women in general appear to
have a higher water turnover rate than nonlactating women, no doubt reflecting their
increased fluid intake and milk production.
The following figures come from longitudinal data collected by researchers at the
Dunn Nutrition Unit for 10 Cambridge
women who were followed from pregnancy
through lactation to the nonlactating, nonpregnant state (Goldberg et al., 1991).
There was a highly significant difference in
water turnover rates (one-tailed, paired t-

test, P < 0.001, Table 1), but there was no

significant difference in TBW between the
lactating and nonlactating women in this
study.

Behavioral adaptations
One behavioral strategy that many
women may employ in heat-stressed environments where access to water may be limited during the day is to superhydrate themselves during their limited drinking opportunities. In the Gambian study, the
lactating, fasting women did indeed superhydrate themselves in the evening before
fasting (Prentice et al., 1984). Urinary osmolality and concentrations of sodium,
urea, and creatinine were, therefore, lower
among the lactating Gambian women compared to controls. As a result, even though
women lost a significantly higher proportion
of TBW compared to controls, dehydration
was not severe judging by the urinary indicators. Moreover, although the serum indicators of dehydration changed significantly,
none of the values fell outside the normal
range.
The Gambian study clearly shows that superhydration can be an effective strategy to
offset short-term fluid deprivation. Are
there other examples of this strategy among
women in areas of limited water supplies?
The !Kung and G/wi foragers who inhabit
desert areas in Botswana and Namibia usually visited waterpans once a day, often in
the late afternoon, where they may have superhydrated, although water was also kept
and stored in various containers (Lee, 1979;
Silberbauer, 1981). Pregnant and lactating
G/wi women who did not travel to the water
pans were given more to drink than others
when supplies came back to camp (Silberbauer, 1981). Lactating Turkana appear to
behave similarly (Bettina Shell-Duncan and
Sandra Gray, pers. comm.). These women
appear to fill up on fluids during daily visits
to the watering holes where they drink extremely large amounts of water (estimated
at 2 liters/time).
Carrying around water is another obvious
behavioral strategy that lactating women
can employ to offset dehydration. Turkana
women carry small containers of fluid with
them during the day and drink small
amounts of tea and buttermilk when they
can. The !Kung and G/wi likewise used ostrich eggs, springhare bladders, and dried
antelope stomachs to store and carry water,

159

LACTATION AND HYDRATION

with the former being more popular for its

durability (Lee, 1979; Silberbauer, 1981).
Each ostrich egg contained 1,000 ml on average, and married women supposedly
owned between 5 and 10 shells each. The
disadvantage was their weight and bulk,
empty shells weighing 250 g. Dried animal
organs held more but were also more fragile.
There is, however, a limit to how much water a lactating woman could carry, particularly if also burdened with a child and other
possessions.
Women also may be able to conserve on
heat stress and water loss by cutting down
on their subsistence-related and other activities. Most studies relevant to this issue
have examined the activities and energy
budgets of lactating women from the viewpoint of energy conservation rather than hydration conservation (Lawrence and Whitehead, 1988; Panter-Brick, 1989, 1993; Peacock et al., 1992), and there have been
virtually no studies of this kind among water-stressed groups.
SUGGESTIONS FOR
FURTHER RESEARCH
Given the data presented, the topic of lactation and hydration requires a great deal
more study, particularly in field settings.
The kinds of research required admittedly
would be difficult to undertake in field situations. Careful studies of fluid intake and
milk production among lactating women
who may be hydrationally stressed are
needed. Serum and/or urine indicators of
dehydration should be collected, as well as
data on the kinds of environmental stresses
to which lactating women are exposed. More
data on water turnover rates, body composition, and milk composition among potentially dehydrated lactating women and the
nursing patterns of their infants and children are needed. Given that lactating
women generally consume more fluids, the
potentially increased health risks in areas
where water supplies may be contaminated
needs further study.
In addition, it would be useful to have
breast milk samples, especially from different seasons, to examine composition and
water content among these kinds of populations. Since no one has yet attempted to replicate the research cited above on Bedouin
women, it is difficult to state with any certainty whether dilution of milk is a wide-

spread human adaptation to fluid restriction.

In addition, more hormonal data among
lactating mothers are essential, specifically
on amounts of circulating AVP, OT, and
PRL in relation to thirst and dehydration
indicators during nursing periods. With the
advent of new techniques to collect blood
samples, such as spot sampling that allows
relatively long-term storage in the field,
such analyses are increasingly possible.
In sum, a relatively untapped area of research exists for human biologists interested in the ecology of breastfeeding. Future
field studies of lactation, particularly those
undertaken in hot climates, should incorporate research on the issue of adequate hydration in maintaining milk production and
the effects of dehydration on the health of
both mothers and infants.
ACKNOWLEDGMENTS
Justin Aunger stimulated the initial idea
to research this topic. I am grateful to Gail
Goldberg and Andrew Coward at the Dunn
Nutrition Unit for permission to use unpublished data on Cambridge women. Robert
Aunger, Sandra Gray, Bettina ShellDuncan, and Virginia Vitzthum provided
valuable comments on earlier drafts, as well
as two anonymous reviewers. Research for
this work was supported by a Royal Society
Research Fellowship.
LITERATURE CITED
Adler RA, Herzberg VL, Brinck-Johnsen T, Weyl Sokol
H (1986) Increased water excretion in hyperprolactinemic rats. Endocrinology 118:15191524.
Adolph AF (1947) Physiology of Man in the Desert. New
York: Interscience Publishers.
Almroth SG (1978) Water requirements of breast-fed
infants in a hot climate. Am. J. Clin. Nutr. 31:1154
1157.
Anand S, Sandborg C, Robinson RG, Lieberman E
(1980) Neonatal hypernatremia associated with elevated sodium concentration of breast milk. J. Pediatr. 96:6668.
Baylis PH (1987) Osmoregulation and control of vasopressin secretion in healthy humans. Am. J. Physiol.
(Regulatory Integrative and Comparative Physiology)
253:R671R678.
Berl T, Brautbar N, Ben-David M, Czaczkes W, Kleeman C (1976) Osmotic control of prolactin release and
its effect on renal water excretion in man. Kidney Int.
10:158163.
Better OS (1987) Impaired fluid and electrolyte balance
in hot climates. Kidney Int. 32:S97S101.
Bhan MK, Arora NK, Ghai OP, Ramachandran K, Khoshoo V, Bhandari N (1986) Major factors in diarrhoea
related mortality among rural children. Indian J.
Med. Res. 83:912.
Brown KH, Dewey KG (1992) Relationships between

160

G.R. BENTLEY

maternal nutritional status and milk energy output of

women in developing countries. In MF Picciano and B
Lonnerdal (eds.): Mechanisms Regulating Lactation
and Infant Nutrient Utilization. New York: WileyLiss, pp. 7795.
Butte NF, Garza C, Stuff JE, OBrian Smith E, Nichols
BL (1984) Effect of maternal diet and body composition on lactational performance. Am. J. Clin. Nutr.
39:296306.
Carey R, Johanson A, Seif S (1977) The effects of ovine
prolactin on water and electrolyte excretion in man
are attributable to vasopressin contamination. J.
Clin. Endocrinol. Metab. 44:850858.
Cunningham ET, Sawchenko PE (1991) Reflex control
of magnocellular vasopressin and oxytocin secretion.
Trends in Neuroscience 14:406411.
Davison JM, Shiells EA, Philips PR, Lindheimer MD
(1988) Serial evaluation of vasopressin release and
thirst in human pregnancy: Role of human chorionic
gonadotropin in the osmoregulatory changes of gestation. J. Clin. Invest. 81:798806.
Dewey K, Lonnerdal B (1986) Infant self-regulation of
breast milk intake. Acta Pediatr. Scand. 75:893898.
Duckman S, Hubbard JD (1950) The role of fluids in
relieving breast engorgement without the use of hormones. Am. J. Obstet. Gynecol. 60:200204.
Dusdieker LB, Stumbo PJ, Booth BM, Wilmoth RN
(1990) Prolonged maternal fluid supplementation in
breast-feeding. Pediatrics 86:737740.
Ershow AG, Brown LM, Cantor KP (1991) Intake of
tapwater and total water by pregnant and lactating
women. Am. J. Public Health 81:328334.
Falk B, Bar-Or O, MacDougall J (1991) Aldosterone and
prolactin response to exercise in the heat in circumpubertal boys. J. Appl. Physiol. 71:17411745.
Food and Nutrition Board (FNB) (1989) Recommended
Dietary Allowances, 10th ed. Washington DC: National Academy Press.
Forman MR, Graubard BI, Hoffman HJ, Beren R, Harley EE, Bennett P (1984) The Pima Infant Feeding
Study: Breast feeding and gastroenteritis in the first
year of life. Am. J. Epidemiol. 119:335349.
Frisancho AR (1993) Human Adaptation and Accommodation. Ann Arbor, University of Michigan Press.
Gainer H, Wray S (1994) Cellular and molecular biology
of oxytocin and vasopressin. In E Knobil and JD Neill
(eds.): The Physiology of Reproduction, Vol. 2, 2nd ed.
New York: Raven Press, pp. 10991129.
Garrow JS, James WPT (eds.) (1993) Human Nutrition
and Dietetics, 9th ed. New York: Churchill Livingstone.
Goldberg GR, Prentice AM, Coward WA, Davies HL,
Murgatroyd PR, Sawyer MB, Ashford J, Black AE
(1991) Longitudinal assessment of the components of
energy balance in well-nourished lactating women.
Am. J. Clin. Nutr. 54:788798.
Grossman SP (1990) Thirst and Sodium Appetite:
Physiological Basis. New York: Academic Press.
Horowitz M, Higgins G, Graham J, Berriman H, Harding P (1980) Effect of modification of fluid intake in
the puerperium on serum prolactin levels and lactation. Med. J. Aust. 2:625626.
Horrobin DF (1980) Prolactin as a regulator of fluid and
electrolyte metabolism in mammals. Fed. Proc.
39:25672570.
Horrobin D, Lloyd I, Lipton A, Burstyn P, Durkin N,
Muiruri K (1971) Actions of prolactin on human renal
function. Lancet 2:353354.
Hossaini-Hilali J, Benlamlih S, Dahlborn K (1994) Effects of dehydration, rehydration, and hyperhydration

in the lactating and non-lactating black Moroccan

goat. Comp. Biochem. Physiol. 109A:10171026.
Illingworth RS, Kilpatrick B (1953) Lactation and fluid
intake. Lancet 2:11751177.
James RJA, Irons DW, Holmes C, Charlton AL, Drewett
RF, Baylis PH (1995) Thirst induced by a suckling
episode during breast feeding and its regulation with
plasma vasopressin, oxytocin and osmoregulation.
Clin. Endocrinol. 43:277282.
Jason JM, Nieberg P, Marks JS (1984) Mortality and
infectious disease associated with infant-feeding practices in developing countries. Pediatrics 74:702727.
Kaufman S, MacKay BJ, Scott J (1981) Daily water and
electrolyte balance in chronically hyperprolactinaemic rats. J. Physiol. 321:1119.
Kaufman S, MacKay BJ (1983) Plasma prolactin levels
and body fluid deficits in the rat: Causal interactions
and control of water intake. J. Physiol. 336:7381.
Keeler R, Wilson N (1976) Vasopressin contamination
as a cause of some apparent renal actions of prolactin.
Can. J. Physiol. Pharmacol. 54:887890.
Lawrence M, Whitehead RG (1988) Physical activity
and total energy expenditure of child-bearing Gambian village women. Eur. J. Clin. Nutr. 42:4560.
Lawrence RA (1994) Breastfeeding: A Guide for the
Medical Profession. St. Louis, MO: Mosby-Yearbook.
Lee RB (1979) The !Kung San: Men, Women, and Work
in a Foraging Society. Cambridge: Cambridge University Press.
Lelong M, Alison F, Vinceneux J (1949) Secretion lactee
humaine et alimentation hydrique. Lait 29:237246.
Lindheimer MD, Barron WM, Davison JM (1991) Osmotic and volume control of vasopressin release in
pregnancy. Am. J. Kidney. Dis. 17:105111.
Little MA (1980) Designs for human-biological research
among savanna pastoralists. In DR Harris (ed.): Human Ecology in Savanna Environments. New York:
Academic Press, pp. 479503.
Loretz CA, Bern HA (1982) Prolactin and osmoregulation in vertebrates: An update. Neuroendocrinology
35:292304.
Maher V (ed.) (1992) The Anthropology of Breastfeeding: Natural Law or Social Construct. Washington,
DC: Berg.
Maltz E, Shkolnik A (1980) Milk production in the desert: Lactation and water economy in the black
bedouin goat. Physiol. Zool. 53:1218.
Melin B, Cure M, Pequignot J, Bittel J (1988) Body temperature and plasma prolactin and norepinephrine
relationships during exercise in a warm environment:
Effect of dehydration. Eur. J. Appl. Physiol. 58:146
151.
Mercier J, Outin S, Paradis K, Hartmann J, Lescoeur B,
Lenoir G, Beaufils F (1986) Allaitement maternel et
tude de 3 observadeshydration hypernatremique: E
tions. Arch. Francaises de Pediatrie 43:465470.
Miller M, van Gemert M (1976) Antidiuretic action of
prolactin in the rat with diabetes insipidus. Horm.
Res. 7:319332.
Mulford C (1990) Subtle signs and symptoms of the milk
ejection reflex. J. Hum. Lact. 60:177178.
Nagy GM, Arendt A, Banky Z, Halasz B (1992) Dehydration attenuates plasma prolactin response to suckling through a dopaminergic mechanism. Endocrinology 130:819823.
North WG, Gellai M, Sokol HW, Adler RA (1981) Demonstration that rat prolactin has no intrinsic antidiuretic activity in the rat. Horm. Res. 15:5564.
North WG, Moses AM, Share L (eds.) (1993) The Neurohypophysis: A Window on Brain Function. Ann.
N. Y. Acad. Sci. 652.

LACTATION AND HYDRATION

Olsen A (1940) Nursing under conditions of thirst or

excessive ingestion of fluids. Acta Obstet. Gynecol.
Scand. 20:313343.
Paneth N (1980) Hypernatremic dehydration of infancy.
Am. J. Dis. Child. 134:785792.
Panter-Brick C (1989) Motherhood and subsistence
work: The Tamang of rural Nepal. Hum. Ecol. 17:205
228.
Panter-Brick C (1993) Seasonality of energy expenditure during pregnancy and lactation for rural Nepali
women. Am. J. Clin. Nutr. 57:620628.
Peacock NR, Sydes KA, Wrangham RW, Ross EAM,
Jenike MR, Aunger RJ (1992) Do marginally nourished women spontaneously reduce energy expenditure during pregnancy and lactation? Evidence from
rural Zaire. Unpublished paper presented at the 91st
Annual Meeting of the American Anthropological Association, San Francisco, CA.
Pederson CA, Caldwell JD, Jirikoswki GF, Insel TR
(eds.) (1992) Oxytocin in Maternal, Sexual, and Social
Behaviors. Ann. N. Y. Acad. Sci. 689.
Prentice AM, Goldberg GR, Prentice A (1994) Body
mass index and lactation performance. Eur. J. Clin.
Nutr. 48, Suppl 3:S78S89.
Prentice AM, Lamb WH, Prentice A, Coward WA (1984)
The effect of water abstention on milk synthesis in
lactating women. Clin. Sci. 66:291298.
Prentice A, Paul A, Prentice A, Black A, Cole T, Whitehead R (1986) Cross-cultural differences in lactational
performance. In M Hamosh and AS Goldman (eds.):
Human Lactation 2. New York: Plenum, pp. 1344.
Robertson M (1989) Prolactin, human nutrition and
evolution, and the relation to cystic fibrosis. Med. Hypoth. 29:8799.
Rolls BJ, Rolls ET (1982) Thirst. New York: Cambridge
University Press.
Sanfilippo J, Nealon N (1985) Relationship between
beta-endorphin and prolactin in female runners. Infertility 8:59.
Share L, Crofton JT (1993) Interactions between the
gonadal steroid hormones and vasopressin and oxytocin. Ann. N. Y. Acad. Sci. 652:438454.
Silberbauer GB (1981) Hunter and Habitat in the Central Kalahari Desert. Cambridge: Cambridge University Press.
Smallridge R, Whorton N, Burman K, Fergusson E
(1985) Effects of exercise and physical fitness on the

161

pituitary-thyroid axis and on prolactin secretion in

male runners. Metabolism 34:949954.
Stuart-Macadam P, Dettwyler KA (eds.) (1995) Breastfeeding: Biocultural Perspectives. New York: Walter
de Gruyter.
Stumbo PJ, Booth BM, Eichenberger JM, Dusdieker LB
(1985) Water intakes of lactating women. Am. J. Clin.
Nutr. 42:870876.
Thompson C, Bland J, Burd J, Baylis P (1986) The osmotic thresholds for thirst and vasopressin release
are similar in healthy man. Clin. Sci. 71:651656.
Thullen J (1988) Management of hypernatremic dehydration due to insufficient lactation. Clin. Pediatr.
27:370372.
Tucker HA (1994) Lactation and its hormonal control.
In E. Knobil E and JD Neill (eds.): The Physiology of
Reproduction, Vol. 2, 2nd ed. New York: Raven Press,
pp. 10651098.
Vitzthum V (1994) The comparative study of breastfeeding structure and its relation to human reproductive
ecology. Yrbk. Phys. Anthropol. 37:307349.
Vorherr H, Vorherr U, Solomon S (1978) Contamination
of prolactin preparations by antidiuretic hormone and
oxytocin. Am. J. Physiol. F234:318324.
Wardlaw GM, Insel PM (1990) Perspectives in Nutrition. St Louis: Times Mirror.
Weiner J (1980) Work and wellbeing in savanna environments: Physiological considerations. In DR Harris
(ed.): Human Ecology in Savanna Environments. New
York: Academic Press, pp. 421437.
Windle R, Judah J, Forsling M (1995) Do vasopressin
and oxytocin have synergistic renal effects on the conscious rat? J. Endocrinol. 144:441448.
Woolridge M (1995) Baby-controlled breastfeeding: Biocultural implications. In P Stuart-Macadam and KA
Dettwyler (eds.): Breastfeeding: Biocultural Perspectives. New York: de Gruyter, pp. 217242.
Yagil R, Amir H, Abu-Rabiya Y, Etzion Z (1986) Dilution of milk: A physiological adaptation of mammals
to water stress. J. Arid Environ. 11:243247.
Yagil R, Etzion Z (1980) Effect of drought condition on
the quality of camel milk. J. Dairy Res. 47:159166.
Zerbe RL, Miller JZ, Robertson GL (1991) The reproducibility and heritability of individual differences in
osmoregulatory function in normal human subjects. J.
Lab. Clin. Med. 117:5159.