Sex Differences in Food Preferences of Hadza Hunter-Gatherers

Evolutionary Psychology
www.epjournal.net – 2009. 7(4): 601-616
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Original Article
Sex Differences in Food Preferences of Hadza Hunter-Gatherers
Julia Colette Berbesque, Department of Anthropology, Florida State University, Tallahassee, USA. Email:
jcb03c@fsu.edu (Corresponding author).
Frank W. Marlowe, Department of Anthropology, Florida State University, Tallahassee, USA.
Abstract: Food preferences are important for understanding foraging choices. In studying
human foragers rather than other animals, we have the advantage of being able to ask them
which foods they prefer. Yet surprisingly, no studies of systematically collected data exist on
human forager food preferences. The Hadza of Tanzania are full-time foragers in an area where
the hominin record extends back to 3-4 million years ago, so their diet is very relevant for
understanding the paleo-diet. Here, we report on their food preferences, elicited with
photographs of species within the five major food categories in their diet: honey, meat, berries,
baobab, and tubers. There were sex differences in the ranks of two food categories: meat and
berries. While male and female ranks agreed on the other three food categories, females ranked
berries second and meat fourth, whereas males ranked meat second and berries fourth. Theses
similarities and differences are interesting in light of the fact that the sexes target different foods.
We discuss the implications of Hadza food preferences for the origin of the uniquely human
sexual division of foraging labor.

Keywords: food preferences, Hadza, hunter-gatherers, paleo-diet, sexual division of labor.

¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Introduction
Food preferences, regardless of how genetically or culturally influenced, are an integral
part of maintaining an adequate diet in any particular environment. Despite substantial variation
across societies there are broad categories of nutritional needs that apply to all humans. These
include: carbohydrates, protein, lipids, and micronutrients (vitamins and minerals). If taste is the
means by which humans (and other species) gauge the nutritional benefits of their foods, then
food preferences should largely map onto nutritional requirements. If food A is always taken
when foods A and B are both equally available, then we might assume that food A is higher in
nutritional value than food B. However, the adaptive forager should target the food with a higher
net value after subtracting the energetic cost of acquiring it. Food A might then be chosen even if
it has lower nutritional value if it requires so little energy to obtain that it yields a higher net

---

Hadza Food Preferences
intake of energy. On the other hand, if we offer the forager both foods free of cost, we should
find that the forager then chooses food B, the food with higher nutritional value.
The term preference has different meanings in economics and psychology. In economics
and Optimal Foraging Theory (OFT), “preference” refers to a choice made after the cost has
been subtracted from the benefit. In the mate preference literature in psychology, on the other
hand, the term “preference” refers to the mate one would like to have, independent of the cost
(presumably the perceived inherent value for one’s own reproductive success) (Symons, 1979).
In the case of foods we call this perceived inherent value the nutritional value (or cost-free
benefit). Here, we use the term preference in the latter sense only. We use the term choice, rather
than preference, for foods actually taken just as mate choice is used to address real life mate
acquisition. In contrast to the literature on mate preferences, the evolutionary literature on food
preferences is surprisingly small. But just as with mate preferences, natural selection should have
shaped our food preferences, that is, our ability to evaluate the cost-free nutritional value of
foods. The evolved preferences for foods high in fat (Booth, 1982) and sugar (Meiselman, 1977)
are often cited cases of “mismatch.” Preferences for these energy dense foods were presumably
adaptive in an environment where it was nearly impossible to get too much of these foods.
However, these same food preferences are now responsible for the obesity epidemic in
industrialized countries where access to these foods is virtually limitless.
There are a number of studies involving the actual food choices of foragers (and therefore
preference in the economic sense) (c.f. Ginane and Petit, 2005; Hawkes, Hill, and O’Connell,
1982; Hayward, O’Brien, Hofmeyr, and Kerley, 2007; O’Connell and Hawkes, 1984). There are
others that describe general preferences in a given population, such as the following quote, “The
most highly prized components of the Aboriginal hunter-gatherer diet were the relatively few
energy-dense foods: depot fat, organ meats, fatty insects and honey” (O’Dea, Jewell, Whiten,
Altmann, Strickland, and Oftedal, 1991, p. 238), see also Hladik and Simmen (1997).
Ethnographic accounts of food preferences are often based on statements made by key
informants. To the best of our knowledge, there are no previous reports on the expressed
preferences of foragers measured in a systematic study.
The Hadza are foragers who live in Tanzania and their diet still consists mainly of wild
foods (>90% by kilocalories among those we study) (Marlowe, 2003). They eat foods that are
widely available in East Africa today and were present throughout the span of our species and
before, such as baobab, honey, large game, and tubers (Baum, Small, and Wendel, 1998;
Danforth, Sipes, Fang, and Brady, 2006; Stollhofen et al., 2008; Thulin, Lavin, Pasquet, and
Delgado-Salinas, 2004). There must be less mismatch between their current diet and that often
referred to as the paleo-diet (Cordain, 2002; Cordain, Eaton, Miller, Mann, and Hill, 2002;
Eaton, 2006; Irons, 1998). The Hadza rarely mix foods, but instead eat single component foods,
while western populations often consume multiple component foods (e.g. hamburgers). Though
there are data on the American preferences of these multiple component foods relative to each
other (for example pizza ranked against a hamburger), it is unclear how they relate to any one
particular nutritional component (e.g. the meat versus the tomato versus the bun of the
hamburger). This problem does not exist with the Hadza diet.
One straightforward prediction is that the higher the calories per kilogram of food, the
more that food would be preferred. However, other nutritional properties such as protein or fat or
certain micronutrients may also partly predict preferences. It is also possible that the preference
rank of a particular food may fluctuate. For example, after one has eaten one food for several
days running, one may crave a different food. A one food diet does not supply all of the nutrients
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(4). 2009. -602-

---

Hadza Food Preferences
that we need as omnivores, and seeking different foods helps us get the other nutrients we need.
The pursuit of a varied diet may mean that a food preference is relatively ephemeral, so that a
person craves the essential nutrients that they are most lacking at the time. This, for example,
might explain a craving for salty foods (Morris, Na, and Johnson, 2008). Finally, we analyze
whether individuals would rank foods in line with their own dietary needs or acquisition abilities
(which may vary depending on reproductive status for females). This might mean that males and
females may rank foods differently, and children may rank foods differently from adults. Below
we interpret our results to see if these or other expectations are supported. We also analyze the
preference ranks in relation to the actual acquisition of these foods across the full year and per
foray.
Materials and Methods
The Hadza
The Hadza are hunter-gatherers who number approximately 1000. They live in a
savanna-woodland habitat that encompasses about 4000 km2 around Lake Eyasi in northern
Tanzania. They live in mobile camps which average 30 individuals (Marlowe, 2006). Camp
membership often changes as people move in and out of camps (Blurton Jones, Hawkes, and
O’Connell, 2005). These camps move about every 6 weeks on average. Hadza feel free to move
wherever they like, although most tend to stay in the region where they grew up.
Hadza men usually go foraging alone. They hunt birds and mammals using only bow and
arrows – poisoned arrows in the case of larger game. They use no snares, traps or nets. They
always have their bow and arrows with them, even when they carry an ax to access honey.
While on walkabout they often feed themselves on berries and baobab (description of Hadza
foods below). They take back to camp mainly meat and honey, as well as some baobab. They
may eat much of the honey they find but take back to camp about half of their haul on average,
and about 90% of medium to large game. Grown men rarely dig tubers.
Hadza women go foraging in groups of 3-8 adults plus nurslings and often some older
children. They mainly collect baobab, gather berries (and berry like fruit which we will refer to
as berries), and dig tubers of several species. They use simple, fire-hardened, sharpened branches
as digging sticks to dig tubers almost every day. They roast and eat some of their tubers once
they finish digging and take the remainder (~ 70% of their haul) back to camp to feed others
(Marlowe, 2006). Although men and women mostly target different foods, roles are not strict.
Women do occasionally get honey (without men’s help), and do opportunistically scavenge
meat.
Hadzaland receives considerable rain (300-600mm) during the months of December
through May, and almost no rain from June through November, so there is a marked contrast
between the rainy season and the dry season. Most foods vary seasonally, with the exception of
some tubers and some game animals.
The Hadza diet can be conveniently categorized into five main food types: honey, meat,
berries, baobab, and tubers (plus Marula nuts in one region only). Honey is the most energy-
dense food in nature (Skinner, 1991), and is highly prized by the Hadza. Honey combs often
contain small amounts of bee larvae, which the Hadza consume along with the comb. The berries
in Hadzaland are much drier than the berries we are familiar with and they are mostly seed, but
what little pulp there is has a high sugar content. Also, when in season, berries are super-
abundant. Baobab fruit is common across much of Africa. Baobab fruit has a chalky pulp that is
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(4). 2009. -603-

---

Hadza Food Preferences
high in vitamin C, and hard seeds that are high in fat. The seeds are only eaten when baobab is
taken back to camp to pound into flour. Many Hadza tubers are continuously available
throughout the year, and are a source of carbohydrates. Tubers vary much more in relation to
region than season. The species eaten most frequently by the Hadza is //ekwa (Vigna frutescens).
All of their tubers have high fiber content but it is so high in //ekwa that one cannot swallow it
but must spit out the quid after chewing it for a while. Figure 1 shows three species of berries
(Ngwilape, Embilipe, and Congolobe), baobab, and the tuber //ekwa. Table 1 shows basic
macronutrient contents of Hadza foods, although it must be noted that the nutritional value of
most foods vary by region, season, and in the case of meat by age and sex of the animal as well.

Figure 1. Examples of Hadza foods.


Table 1. Composition of Hadza foods
Carbs

Fat
Protein
(starch)
Sugars
Kcal
87.71-
Honey
<8.01
<4.01
Trace1
96.01
403-4391
Meat
1.3-9.04
57.3-68.24
Trace5
Trace5
2166-6097
Tuber
0.63-3.43
2.32-10.43
19.42-61.33
6.22-48.32
73-858, 146-2982, 177-2793
Berries
<2.0 1
7.11-15.21
Trace
61.1-72.71
108-1458,318-3421
Baobab*
15.01
19.41
5.51
23.41
328.51

1 (Murray et al., 2001) 2 (Schoeninger et al., 2001) 3 (Vincent, 1985) 4 (Van Zyl and Ferreira, 2004) 5 (Cordain et
al., 2002) 6 (Weiner, 1973) 7 (McCullough and Ullrey, 1983) 8 (Galvin et al., ND)
All values based on grams per 100 grams dry weight. All values are averages; foods vary by species, location, and
season. Meat macronutrient values large based on springbok, blesbok, and impala samples. Meat kilocalories are
based on deer species.
* Baobab composition is based on a ratio of 50% pulp and 50% seed by weight of a pod.
Note. Bold numbers indicate which food has the highest value for each macronutrient.
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(4). 2009. -604-

---

Hadza Food Preferences


Food taboos among the Hadza are sex specific. Hadza men who have become “real men”
by virtue of killing one of the very large game animals (or else are over about 30 years of age)
take particular portions of meat from large game animals. This consists primarily of organs
including the genitals, kidney, heart, throat, and tongue. The men eat these portions hidden away
from the women and children, and it is considered dangerous for women or children to eat these
parts. Perhaps a less stringent taboo (FM has witness a violation of this taboo) is the eating of
tortoises by men. Men say that the poison on their arrows will become weak and not work if they
eat tortoise. Women and children, however, eat the tortoise without any repercussions. Both men
and women avoid eating fish and reptiles. This avoidance does not rise to the level of a taboo,
but when some food does not taste good to them they say that it “tastes like snake,” which is
what they say about fish and all reptiles (except for tortoise).

Materials and Procedure

Design
We interviewed individuals one-on-one in the privacy of a Land Rover in the morning.
The Hadza were never interviewed after a meal, because being sated by a particular food might
affect their preferences. We used photographs of foods on index cards for visual aids.
Photographic methods have been used in Western populations to elicit emotions about complex
foods as well as food preferences (Corney, Eves, Kipps, and Lumbers, 2001; Hinton, Holland,
Gellatly, Soni, and Owen 2006; Rousset, Deiss, Juillard, and Schlich, 2005), and two studies
found that photographic methods were very concordant with taste test preference ranks (Guinard,
2000; Guthrie, Rapoport, and Wardle 2000).
We attempted to get at cost-free preferences by specifically stating that the foods should
be considered as if they were already in front of the subject, ready to eat, pre-processed, and the
subject could eat as much of that food as they wanted at one sitting. This was a two tiered
process. The first tier was the ranking of five species from each of three of the food categories:
tubers, berries, and game. Only the five species photographs within a particular food category
(e.g. tubers) were shown at one time. The first tier species level sorting was done primarily to
ensure that any individual’s bias against a particular species within a category did not affect their
overall ranks of the five broader food categories (see Figure 2).













Evolutionary Psychology – ISSN 1474-7049 – Volume 7(4). 2009. -605-

---

Hadza Food Preferences
Figure 2. Forced-choice preference ranking of Hadza foods.

Note. This schema shows an individual that selected species A for meat, C for berries, and E for tubers in tier 1.
Other individuals may rank different species as “favorites” in the three categories in tier 1. Once they have chosen
the favorite in the three categories in tier 1, we asked them to rank those species against baobab and honey in tier 2.

The second tier included the favorite species from the three previous categories, plus a
picture of honey, and a picture of baobab. Presumably if this were not a two tiered sort, three
species of berries could be ranked before any other food type. The nutritional differences
between those berry species is small compared to the differences between those berry species
and any other food category. Because we wanted to investigate whether preferences mapped onto
different nutritional attributes, we were interested in the overall ranks of these nutritionally
distinct categories which are salient to the Hadza and comprise approximately 95% of their diet.
Honey was represented by a single picture (of ba’alako) because recognizing the honey
from the different species of bees by photograph would be difficult. In addition, ba’alako (the
honey from the African bee--Apis unicolor adansonii) is the most common, provides the most
kilograms of honey, and is highly prized. A single photograph of baobab was used because there
is only one species, but it is such a major fruit in terms of kilograms in the Hadza diet that it
deserves a category unto itself. In addition, the Hadza do not refer to it with the same term “fruit”
as they use for all species of berries. Animal pictures were used for their recognizability over
pictures of meat itself. The species chosen for the categories tuber, berries, and meat are by no
means comprehensive because we wanted to keep the interviews relatively brief. All of the
species chosen for each of these categories are very well represented in FM’s acquisition data,
and were well-known across regions. We chose game animals ranging from small to large to
capture the range taken. The food species used can be seen in Table 2.





Evolutionary Psychology – ISSN 1474-7049 – Volume 7(4). 2009. -606-

---

Hadza Food Preferences
Table 2. Hadza food species used in this study.
Category Hadzane
name
(English name)
Scientific name

Matukwayako
Coccinea surantiaca or aurantiaca

Penzepenze
Vigna sp. (Papilionoidea Leguminosae)
Tuber
//Ekwa hasa
Vigna frutescens
Do’aiko/Shakeako
Vigna macrorhyncha
Shumuwako
Vatoraea pseudolablab

Embilipe
Grewia flavescens Juss., Grewia platyclada

Congolobe
Grewia bicolor Juss.
Berry
K’alahaibe
Opilia campestris Engl.
Undushipi
Cordia gharaf Ehrenb.
Ngwilape
Grewia similis K. Schum.

Nakomako

(Buffalo)
Syncerus caffer
Meat
Gewedako
(Dikdik)
Madoqua kirkii
Komati (Eland)
Taurotragus oryx
Molola (Jackal)
Canis adustus
Kwa’i (Warthog)
Phacochoerus aethiopicus
Baobab
N//obabe
Adasonia digitata L.
Honey
Ba’alako
Apis unicolor adansonii (bee species)

Food Acquisition
Data on food acquisition were obtained by dividing the kilocalories of each food returned
when those foods were acquired by the total time spent foraging out of camp on those forays.
Because the Hadza are opportunistic foragers and will rarely search for a particular item to the
exclusion of other possibilities, it is very difficult to categorize the targeted item and calculate a
search time on any given foray for this item. Hourly kilocalories does not perfectly capture the
time necessary to acquire a particular food, but is probably the best possible proxy. Because
Hadza men usually forage alone, it takes years of focal individual follows to calculate a
percentage of diet eaten out of camp for males. FM has estimated from preliminary data analysis
that roughly 30% of the diet is eaten out of camp. We use acquisition of foods brought back to
camp as a proxy for the complete diet. Data on kilocalories per hour of foraging were collected
from six camps in 1995-96 on adults (ages 18 and older). We also show the percentage of the
diet that each food category constitutes. Data on percentage contribution to diet were collected
from 16 camps from 1995-2005. We present both kilocalories per hour of foraging and
percentage contribution to the diet because our data on kilocalories per hour of foraging are not
affected by food availability (which varies by region and season), whereas percentage
contribution to the diet is. For instance, because honey is available seasonally, percentage
contribution to diet is probably less relevant to taste preferences than hourly kilocalories. If
honey is not often available, it will not constitute a high percentage of the diet. Nevertheless
when it is available it provides men with their highest return per hour.

Although food preference data were acquired in 2007-2008 in three regions, we do not
believe that there were significantly different foraging patterns in those years than in the years in
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(4). 2009. -607-

---

Hadza Food Preferences
which other data on kcal/hour acquisition (1995-1996) or percentage contribution to the diet
(1995-2005) were collected. All three datasets were collected in the same three regions in camps
where the residents were foraging full-time; many of them were also in the food preference data.
Results
The total sample consisted of 94 individuals. Males ranged in age from 16-70 years with
a mean age of 38.73 (SD = 13.15, n = 45). Females ranged in age from 19-78 years with a mean
age of 37.16 (SD = 16.98, n = 49). The foods were ranked from first to fifth by the total sample
as follows: honey, meat, baobab, berries, and tubers.
Both males and females ranked honey first, baobab third, and tubers last. There was no
significant age variation in preference ranks within sex when analyzed by ten year age cohorts.
Because our data are ranked, we used non-parametric Mann-Whitney tests to see if there were
sex differences. Although honey was ranked first by both sexes, males had a higher mean rank
for it (4.51) than females (3.67) (Mann-Whitney U = 753, p = .003, n1 = 45, n2 = 49) (see Table
3). Females ranked berries second and meat fourth, while males ranked meat second and berries
fourth (Fig. 3). The mean rank of meat was higher by males than females, but significant only
with a one-tailed test (U = 889, p = .099, n1 = 45, n2 = 49). Females ranked berries significantly
higher than males (U = 679.5, p = .001, n1 = 45, n2 = 49). Women who were noticeably pregnant
(n = 4) did not have different preferences than other females, nor did nursing women (n = 13).

Figure 3. Hadza food preference ranks by sex.







Evolutionary Psychology – ISSN 1474-7049 – Volume 7(4). 2009. -608-

---

Hadza Food Preferences
Table 3. Descriptive statistics of food preference rankings by sex.
Sex

Honey
Meat
Baobab
Berries
Tubers
Males
Rank
1
2
3
4
5
n = 45
Mean Score
4.51
3.40
2.67
2.38
2.04

SD
0.84
1.27
1.23
1.07
1.15
Females
Rank
1
4
3
2
5
n = 49
Mean Score
3.67
2.92
3.06
3.20
2.14

SD
1.44
1.46
1.33
1.21
1.24
Total
Rank
1
2
3
4
5
n = 94
Mean Score
4.07
3.15
2.87
2.81
2.10

SD
1.26
1.38
1.29
1.21
1.19

Both sexes ranked the five species of berries the same: 1) Ngwilape, 2) Undushupe, 3)
Congolobe, 4) Embilipe, and finally 5) K’alahaibe. It is worth noting that during certain berry
seasons, men and women often forage for berries together, and even when foraging separately
men also feed on berries, though it is women who tend to gather enough to take back to camp.
Baobab is the other food that both sexes gather in substantial quantities and they both ranked it
third. In contrast, males and females ranked the five tuber species differently, and grown men
very rarely dig tubers. Women and men also ranked the five animal species differently, which
women rarely take.
The concordance, or inter-rater agreement in this scale for the entire population was
tested with Kendall’s W (W = .205, df = 4, p = .000). Males were more concordant in their
ranking (W = .385, df = 4, p = .000) than were females (W = .124, df = 4, p = .000) (for tests of
concordance for particular food categories by sex, see Appendix 1).
Males ranked higher those foods that they tend to target (see Table 4 for acquisition by
sex). Male hourly kilocalorie acquisition by food type perfectly predicts male food preference
ranks (r = .993, n = 5, p = .001). Females ranked male acquired foods lower than males ranked
them (significantly lower for honey). Female hourly kilocalorie acquisition by food type (Table
4) does not predict their stated food preferences. Honey, meat, and tuber foraging returns deviate
from this pattern in female food preferences. Females get little honey but prefer it most, but they
also get very little meat (it yields the lowest kcals/hour for them) and yet they do not rank meat
high. Women get great quantities of tubers but prefer them least.

Table 4. Hourly acquisition and percentage contribution to diet of Hadza men and women.






Evolutionary Psychology – ISSN 1474-7049 – Volume 7(4). 2009. -609-

---

Hadza Food Preferences
Discussion
Our results call into question some common assumptions about human foragers. In
humans, males and females forage apart and target different foods much of the time (Kelly,
1995). Jochim (1988) points out that OFT models have been developed for species in which the
generic forager is analyzed, rather than two very different foragers (one male and one female).
Yet despite the distinct costs for males and females in acquiring certain foods (due to strength,
size, or risk), the preferences of males and females are often assumed to be the same. We found,
however, there are sex differences in food preferences among the Hadza. In addition,
ethnographers often claim that meat is the most preferred food among foragers (including the
Hadza) (Bunn, 2001), yet our empirical data on individual preferences demonstrates that honey
is more preferred than meat among the Hadza.
Caloric value predicted the most preferred (honey) and least preferred foods (tubers) for
the Hadza. The macronutrient differences in male preferred foods and female preferred foods
may be characterized as males preferring protein (meat) more and females preferring sugar
(berries) more. What an individual had recently eaten did not seem to substantially affect
preference ranks because there was a high degree of concordance in several food category ranks.
If recent consumption affected preferences we would have expected a difference in preference
ranks from one season or region to the next due to changes in the availability of each food item.
For example, because we collected the data in three different regions over two years (one year
without any berries available and one with several species of berries available), we would not
expect the significant agreement that we found on berry preference ranks in both years. This
consistency in preference ranks leads us to conclude that preferences for a food item are not
heavily dependent on the availability of that food item.
It is interesting that men bring back to camp more calories of honey per hour gone from
camp than any other food type, followed by meat, baobab, berries, and finally tubers. This
happens to be exactly the same as their preference ranking. This suggests that men might devote
more effort to getting the foods that they like the most. Women’s acquisition, on the other hand,
does not match their preference ranks.
Both sexes ranked tubers last, which is what we would expect if they are fallback foods
that are taken when more preferred foods are not available in sufficient quantities (Marlowe and
Berbesque, in press). Although meat and honey are calorically much more valuable than female
foods, females did not place the premium on them that males did, ranking honey first but
significantly less frequently than men and ranking meat 4th. Females ranked berries high (2nd),
despite their relatively low caloric value per kilogram (as compared with meat). In addition,
tubers yield the highest kilocalories per female hour of foraging and yet females ranked tubers
last.
It is not clear what we should expect female preferences to be if they are constrained in
their foraging. If females had the same dietary requirements as males we might expect both sexes
to have the same preferences regardless of who acquires the food. According to one view, the
risk aversion of females to the less reliable, higher variance foods (such as honey or meat) may
be due to constraints related to infant care (Brown, 1970; Jochim, 1988; Marlowe, 2007).
The view that males and females are pursuing different foraging strategies for the benefit
of the household has been challenged in the past two decades. Females may simply avoid
hunting and honey collection in favor of more predictable returns which are better for
provisioning children. Males may target high variance foods to increase their number of mating
Evolutionary Psychology – ISSN 1474-7049 – Volume 7(4). 2009. -610-

---