Targeting the Hunting Hypothesis: Review of Evidence From the Hadza

ABSTRACT

The hunting hypothesis holds that ancestral human males favored their own mates and children in sharing meat gained from big game hunting, a practice said to have led to the origin of nuclear families and related changes in life history. Data from East African Hadza hunter‐gatherers operating in an environment like that prevalent when and where Homo evolved contradict key elements of this idea. An alternative model, the grandmother hypothesis, holds that senior women's foraging and food sharing led to life history changes that favored mate guarding, not paternal provisioning, in the formation of nuclear family‐like social units. Relevant data and theory are reviewed and evaluated.

1. Introduction

The initial evolution of genus Homo is often associated with the emergence of nuclear families as units of common economic and reproductive interest [1, 2, 3, 4]. According to the widely favored hunting hypothesis, early human females took plant foods to support themselves and cover the costs of carrying and nursing their infants, just as their hominin female ancestors did. Meanwhile, males, prompted by climate and environmental change, began to act as paternal provisioners via big game hunting and aggressive scavenging, practices unprecedented in hominin history. Nuclear families consisting of an adult man and woman and their children, marked by relatively exclusive pair bonding between the adults and a sexual division of labor in subsistence and childcare, commonly seen as the fundamental human social unit, emerged as a result. Other characteristics of the human lineage—delayed maturity, extended longevity, higher fertility, overlapping juvenile dependents, increased brain size, and greater reliance on learned behavior—are said to have been further products of the change.

The female part of this model is based on observations among living primates, including human hunter‐gatherers. It is assumed, not unreasonably but without direct evidence to have been typical of pre‐human hominins as well. The male element draws on archaeological indications of large ungulate carcass acquisition by hominins that began to appear roughly coincident with the earliest fossil evidence of Homo 2.8–2.0 million years ago, and on the idea that nuclear families and paternal provisioning are typical of modern hunter‐gatherers [5, 6, 7]. As it happens, the ethnographic record is mixed on paternal provisioning: fathers’ contribution to their children's nutritional welfare is significant among some modern foraging populations but not others (e.g., [8, 9, 10, 11]).

Given this ambiguity, several questions emerge: Why assume that archaeological evidence of big game hunting and scavenging represents paternal provisioning rather than some other practice? Why suppose that climate‐driven big game hunting and scavenging were catalytic in early human evolution rather than the products of some other process? What do answers to these questions mean about the idea that nuclear families as customarily defined evolved as fundamental units of human social life?

The East African Hadza offer an important perspective on these issues. They live in a part of the world that has provided the earliest fossil evidence for genus Homo. Their homeland has long included sizable areas free of agriculture where wild plant foods and big game have until recently been abundant. Hadza men self‐identify as big game hunters and have been recognized as such by outside observers since first encountered by Europeans more than a century ago [12, 13, 14]. If any modern hunter‐gatherer populations support the paternal provisioning model, Hadza should be among them, yet comprehensive accounts of their foraging and food sharing practices do not provide that support. Some Hadza men hunt big game often, others do not. Even the most active in this realm fail to take large‐bodied prey on the short‐term, say weekly time scales required for effective family provisioning. Moreover, when hunters are successful, the meat and marrow they take are distributed widely within and between base camps, with little clear evidence of special shares to their own households. Though most accounts over the last half century affirm these generalizations, some of the most recent do not, authors insisting instead that Hadza men do indeed practice paternal provisioning. Why the difference between these accounts and what are the implications?

Here we review evidence of Hadza men's food acquisition and sharing across two time periods, 1958–1990 and 1995–2013, giving detailed attention to data collection protocols, resource choice, consumption patterns, and sharing with family members and others. We identify differences in men's prey choices and sharing patterns between these two intervals, showing that they reflect a sharp decline in big game numbers. In neither situation is paternal provisioning strongly supported. We then briefly reference an alternative model of human life history evolution, developed in detail elsewhere, that in combination with evidence of large animal populations at the time early Homo evolved supports the idea that mate guarding, not paternal provisioning led to the origin of nuclear family‐like social units.

2. Background on Hadza

Hadza are a population of about 800–1000 central‐place foragers living near Lake Eyasi, northern Tanzania [15, 16]. They are defined as a group by their common language, Hadzane. At initial European contact [13], they apparently had a 2500 km² area south and east of the lake largely to themselves (Figure 1). A small subset was found north of the lake in Dunduiya, near what is now the Maswa Game Reserve. Neither group is nor has been an isolate. Both have interacted with surrounding pastoral and agricultural populations for centuries ([16]: 42–63, SI 3.5).

Figure 1.

Satellite‐based image of Hadza country showing major terrain features and localities named in text. Dunduiya is out of frame, approximately 60 km NW of Lake Eyasi. Source: NASA‐JPL.

Until recently, most Hadza lived in short‐term base camps occupied by 25–75 individuals, including nuclear families, single adults, and adolescents. Women and children foraged for fruit and geophytes. Men armed with bows, arrows, and small axes targeted big game and Apis mellifera honey. They encountered small game frequently, captured some easily, but usually did not pursue them assiduously. They took fruit and stingless bee honey in small quantities to satisfy their immediate hunger, but generally ignored geophytes.

Since the 1960s, Hadza in Mangola and Siponga have been forced to abandon much of their traditional lifestyle due to habitat changes associated with the development of commercial agriculture, the clearing of woodland in connection with charcoal burning, and the spread of Iraqw farmers and Datoga pastoralists. Repeated Tanzanian government attempts to force Hadza to settle as farmers have also played a part. They have nevertheless persisted in the low hundreds as full‐time hunter‐gatherers in Tli'ika and Han!abi despite ongoing incursions by Datoga, an increase in bush meat trade that limits men's hunting success, and a growing tourist presence [15, 16].

3. Sources of Information on Hadza Men's Food Acquisition and Sharing

We consider three James Woodburn [14, 17] worked south of Lake Eyasi from 1958 to 1961, when most of that area was not directly impacted by farming or herding. Savanna woodland was intact; big game were abundant. Woodburn's research focused on social organization, facilitated by his fluency in Hadzane. He collected quantitative data on subsistence practices but never published them. His qualitative observations on the topic are nevertheless useful.

The Utah/UCLA group (Blurton Jones, Hawkes and O'Connell) compiled quantitative data on Hadza anthropometry, demography, time allocation, prey choice, food processing and sharing. Demographic data were drawn from all parts of Hadza country, 1982–2000 [16]; information on other topics was collected mainly in Tli'ika and Han!abi, 1985–1990 (Table 1; see [28] for review).

Table 1.

Principal sources of quantitative information on Hadza men's time allocation, prey choice, food processing, and sharing, Tli'ika/Han!abi, 1985–2013.

Dates	Data
1985‐90	Two hundred fifty‐six days of observation in eight Hadza base camps, including “instantaneous” scans (n = 1700) of in‐camp activity during daylight hours, with notes on the location of each person present when seen and the activities in which they were engaged [18].
	Seventy‐three follows (570 h) on fifteen focal men while away from camp, day or night, with notes on resources encountered, quantities taken, and whether they were consumed as acquired or carried to camp [19].
	Combination of the above, backed by end‐of‐day activity summaries on every camp resident led to accumulation of about 50,000 individual observations on 90 Hadza and construction of 245 time‐budgets for camp residents, in camp and away, organized by activity, age, and sex [18].
	Observations on kill‐site processing, consumption, and subsequent transport of parts to base for 54 large animals acquired by hunting or scavenging [20, 21]. Identities and activities of all Hadza taking part were noted.
	Records of initial intra‐camp distributions of meat shares from 20 large animal carcasses across all nuclear family households present [29].
	Observations on seven hunters over 102 man‐days (including 28 focal‐follow days) charged with taking all small animals they saw on out‐of‐camp forays, the aim being to establish maximum potential acquisition rates for this prey category [21].
1995‐2013	One hundred eighteen follows on 146 focal men based in 29 camps across all parts of Hadza country, 2001–13 [23]; additional data on a 40‐follow subset, 2006–13, including 33 in Tli'ika/Han!abi, seven in Mangola [24, 25]. Weights, prey types acquired, and whether prey were eaten at acquisition or carried to base were recorded.
	Foods brought to base camps over 1496 man‐days in Tli'ika/Han!bai. Seventeen instances involving 1371 kg of big game, 142 involving 185 kg of small game, 401 involving 348 kg of honey, 146 involving 345 kg of fruit ([26]: ESM Table S1).
	From the same data set: resource weights distributed to all types of households: 312 shares ‐ 36 for big game, 54 for smalls, 101 for honey, 21 for fruit, one each for bee brood and tuber. All but big game are reported to have gone to the household of the man responsible for taking the resource.
	Weights and edible portions of all foods brought to five base camps over nine months, 1995–1996 [27].

Frank Marlowe and colleagues (1995–2013) pursued a larger project: more people conducting fieldwork over longer periods of time, collecting information on a broader range of topics. Here we consider only their work on men's foraging and food sharing (Table 1; see [15] for partial review). Data recorded are similar to those collected by Utah/UCLA group.

4. Foraging Times, Party Compositions, Resource Encounters, and Acquisitions

1958–61. Woodburn reports that hunters generally operated alone. He refers to the same general array of resources encountered and pursued as reported by the other two research groups (see below), but provides little detail. One important observation worth noting:

[Big game] hunting, even by a skilled hunter … is always an unpredictable pursuit. … Perhaps as many as half of the adult men may fail to kill even one large animal a year … There are some men who have killed scarcely a single large animal during their entire adult lives (1968: 53–54).

1985–90. In‐camp scans compiled by Hawkes and O'Connell show that men overall averaged about 4 h per day away from base camps, mainly foraging and visiting other camps [18]. Data from focal follows, most on more active hunters, provide further detail ([19], Supplementary Material, Tables S1–S2). These can be summarized in two semi‐seasonal sets, late dry (LD, July–October, 33 follows, 278 h) and wet/early dry (WED, November–June, 40 follows, 292 h), each defined by patterns in men's out‐of‐camp activities. Figure 2 shows focal party compositions; Figure 3 focal party goals. Table 2 shows numbers of prey encountered, pursued, and acquired as well as times invested in pursuit.

Figure 2.

Composition of focal men's follow parties, late dry versus wet and early dry, 1985–86. Vertical scale = number of parties. Solo means focal man operated alone; with 1–2 men or boys means these were focal man's only companions. All other categories may overlap (i.e., those with > 1–2 men or boys may also include women or teen boys and girls). Details reported in [19].

Figure 3.

Primary goals of focal men's follow parties, late dry versus wet and early dry, 1985–1986. Vertical scale = number of parties. Encounter means time spent walking through areas where men's primary targets (big game or Apis mellifera honey) were likely to be sighted. Ambush, practiced in late dry only, is defined as sitting in a blind adjacent to a frequently used big game trail or water source. Tracking means pursuing a large animal thought to have been wounded by a poisoned arrow. Retrieval means traveling to a large animal kill site to collect carcass parts for transport to base. With women means focal man accompanied one or more women searching for their preferred foods—on these follows mainly geophytes. Visit means walking to a nearby Hadza camp, often to share big game meat from a recent kill known to be available there. Apis means the follow goal was to find and collect Apis mellifera honey, a resource encountered and pursued in this data set mainly in the wet and early dry. Details on follow goals are reported by [19]. Additional details on ambush and encounter hunting are presented in O'Connell et al. [20].

Table 2.

Prey encounters, pursuits, pursuit times, and acquisitions, late dry versus wet and early dry seasons, 1985–1986.

	Encounters	Follows	Pursuits	Pursuit t (h)	Acquired
Late dry
Big game	176	27	36	27.0	4
Small game	89	19	11	1.8	1
Honey	38	9	38	3.6	27
Baobab	9	6	6	3.4	6
Berries	5	3	5	0.2	5
Geophytes	8	5	4	1.0	4
Wet/early dry
Big game	222	23	33	18.0	2
Small game	82	12	16	2.0	7
Honey	171	31	168	74.0	134
Baobab	8	4	2	0.2	2
Berries	32	12	32	2.5	32
Geophytes	10	8	1	0.2	1

Most LD follows involved focal men working alone or with 1–2 other men or teenaged boys. Primary goals (85% of LD follow time) were big game‐related, including some combination of search, ambush, pursuit of wounded prey, or carcass retrieval. A total of 176 large animals (78% impala and zebra) were encountered, 36 were pursued over a total of 27 h. Four were taken, representing about 700 kg edible weight, nearly all of it carried to base. Stingless bee honey was an important secondary target, always pursued on encounter. Amounts involved were generally small (< 100 g per nest).

WED follows involved larger, more diverse groups, although men still operated alone or with other men or teenaged boys on 45% of forays. Apis mellifera honey was the primary focus on 21 follows, secondary on another eight (134 encounters and pursuits over 70 h). The balance of encounter totals shown on Table 2 was made up of stingless bee honey. One hundred six Apis hives yielded an estimated 326 kg of honey, and the rest were dry. Big game was an important target (222 encounters, 76% giraffe and impala, 33 pursuits, 18 h) but only one carcass and small fractions of another, a badly decomposed sub‐adult elephant (215 kg edible total) were recovered.

Small game were encountered often during both seasons, more so than the totals in Table 2 indicate (JOC, who recorded the follows, almost certainly did not note all small game that FM saw). The acquisition rate per WED follow‐day was about one‐fourth that earned in the early wet season 1990 experiment, where focal hunters were asked to take all small game they could in the same area covered in the 1985–1986 follow sample ([22]: 246). This suggests that FM in that sample either saw fewer small game or more likely ignored most of those they noticed. Berries were always taken in small quantities on encounter and eaten immediately. Baobab and geophytes were probably seen more often than indicated and with the exception of two incidents, one involving geophytes, the other baobab, were handled entirely by women.

2001–2013. The Berbesque et al. data set (Table 1) is dominated by follows on single individuals: 105 forays of 118 total (89%). The Wood/Marlowe subset (2006–2013) is similar, nearly all on single FM. In our sample, just 18 of 73 follows (25%) were on men operating alone.

The array of resources encountered, pursued, and taken in the Wood/Marlowe sample is like ours but differs in detail (Table 3). The closest resemblances involve Apis honey, baobab fruit, and geophytes. Apis hives yielded honey on nearly all encounters in both datasets. Hives sampled per follow‐hour were also similar: about one every 5 h in both. Geophytes were pursued on only a quarter of encounters in our data set (mostly by women), never in Wood and Marlowe's nor in the larger Berbesque et al. array. As indicated above, we think the pursuit/encounter ratio in our sample is even lower than shown simply because encounters with this resource where women were not present were not always identified and recorded. Baobabs were seen at similar rates in both samples but collected only about a third as often in ours, and even then, rarely handled by focal men, though they were collected by men occasionally in non‐follow situations.

Table 3.

Comparison of encounters, pursuits, and results; this report vs Wood & Marlowe, [24].

Data sets	Utah/UCLA 1985–1986			Marlowe group 2006–2013
Locations	Tli'ika/Han!abi			Tli'ika/Han!abi/Mangola
	Total	per follow		Total	per follow
Follows	73	—		40	—
Follow time (h)	570	7.8		212	5.3
Large game	Total	n/follow	n/follow h	Total	n/follow	n/follow h
Prey seen	398	5.5	0.69	46	1.20	0.22
Taxa seen	14	—	—	8	—	—
Prey pursued	69	0.9	0.12	25	0.63	0.12
% of seen pursued	17	—	—	54	—	—
Pursuit time (h)	45	0.6	0.08	13	0.33	0.06
Time per pursuit (h)	0.7	—	—	0.5	—	—
Prey taken	6	0.1	0.01	1	0.03	0.005
Prey taken per pursuit h	0.1	—	—	0.08	—	—
Small game
Seen	171	2.32	0.30	195	4.88	0.92
Taxa seen	8	—	—	27	—	—
Prey pursued	27	0.37	0.05	153	3.83	0.72
% seen pursued	16	—	—	78	—	—
Pursuit time (h)	3.8	0.05	0.007	5.9	0.15	0.03
Time per pursuit (h)	0.14	—	—	0.04	—	—
Prey taken	8	0.11	0.014	23	0.58	0.11
Prey taken per pursuit h	2.11	—	—	3.90	—	—
Apis honey
Encountered	142	1.95	0.25	42	1.05	0.20
Pursued	142	1.95	0.25	33	0.83	0.16
Pursuit time (h)	71.1	0.97	0.12	nr	nr	nr
Taken	107	1.47	0.19	42	1.05	0.20
Trigona honey
Encountered	67	0.92	0.12	77	1.93	0.36
Pursued	67	0.92	0.12	71	1.78	0.33
Pursuit time (h)	6.4	0.88	0.11	nr	nr	nr
Taken	53	0.73	0.09	77	1.93	0.36
Baobab fruit
Encountered	17	0.23	0.030	8	0.20	0.04
Pursued	8	0.11	0.014	8	0.20	0.04
Pursuit time (h)	3.6	0.05	0.006	nr	nr	nr
Taken	8	0.11	0.014	8	0.20	0.04
Berries
Encountered	37	0.51	0.06	34	0.85	0.16
Pursued	37	0.51	0.06	32	0.80	0.15
Pursuit time (h)	2.7	0.04	0.005	3.5	0.09	0.02
Taken	37	0.51	0.06	32	0.80	0.15
Geophytes
Encountered	18	0.25	0.03	nr	nr	nr
Pursued	5	0.07	0.009	0	0	0
Pursuit time (h)	1.2	0.02	0.002	0	0	0
Taken	5	0.07	0.009	0	0	0

Abbreviation: nr, data not reported.

Data on the other four resource categories are very different. FM in our sample saw three times more big game per follow‐hour than those monitored in the Wood/Marlowe report and twice as many taxa overall. FM in our sample pursued the same number of large animals per follow‐hour, invested about the same time in pursuit, but took twice as many: 0.01 versus 0.005/follow‐hour. The contrast in acquisition rate may reflect differences in taxa encountered, greater selectively in pursuit, or greater pursuit skills among FM in our sample. We think greater selectively (17% pursued of those seen in our sample, 54% in Wood and Marlowe's) and greater pursuit skills among FM we followed are most likely indicated.

Small animals show the opposite pattern. Relative to follow‐hour, three times as many were seen in the W/M sample than in ours, more than ten times as many pursued, and four times the effort invested in pursuit. W/M's subjects also took just under twice the number per pursuit‐hour. These differences might mean that small games were more abundant when the W/M sample was collected, but we think not. In our 1990 experiment [22], also in Tli'ika/Han!abi, where Hadza men were encouraged to take as many small animals as possible, their average capture rate—about two acquisitions on every three follows—is roughly the same as that reported by Wood and Marlowe. Similar small prey densities are implied. If so, far more attention was paid to small game in the W/M sample than in our 1985–1986 data set.

Trigona honey was encountered, pursued, and taken about three times as often per follow‐hour by W/M's subjects as by ours. Berries were noted and pursued twice as often and picked more than twice as often by their FM. Again, these numbers may reflect differences in resource availability, but given the overlap in follow locations, we see them as indicating less attention to availability and greater selectively in pursuit on the part of our subjects.

A tally of acquisitions per follow‐hour captures some of these differences (Figure 4). Wood and Marlowe's focal men enjoyed much higher hourly returns by count from small game, Trigona honey, and berries, and much lower returns from big game hunting than did men we monitored. It seems likely that FM in the Wood/Marlowe 2006–2013 sample were reacting to a different array of prey options, specifically those involving big game, than those observed in 1985–1986.

Figure 4.

Comparison of prey acquisition rates (captures per follow‐hour), Utah/UCLA, 1985–1986 versus Wood/Marlowe, 2006–2013. Data from Table 3.

5. Consumption at Acquisition Sites

1958–1961. Woodburn [14] says that small animals and plant foods taken by a hunter were eaten as acquired. He provides no information on the treatment of big animals at kill‐sites other than that they were dismembered for transport to base (see below).

1985–1990. Small game, stingless bee honey, and plant foods taken by FM were generally eaten as acquired (Table 4). Apis honey and big game were treated differently.

Table 4.

Weights of foods eaten at acquisition versus carried to base, Utah/UCLA, 1985–1986.

		Mean	Mean
	Weight	wt/follow	wt/follow‐h
Large game
Edible wt taken (kg)	965	13.22	1.69
Eaten at acquisition (kg)	40	0.55	0.07
Returned to base (kg)	925	12.67	1.62
Small game
Edible wt taken (kg)	4	0.05	0.007
Eaten at acquisition (kg)	4	0.05	0.007
Returned to base (kg)	0	0.00	0.00
Apis honey
Edible wt taken (kg)	328	4.44	0.57
Eaten at acquisition (kg)	181	2.48	0.32
Returned to base (kg)	147	2.01	0.26
Trigona honey
Edible wt taken (kg)	8.2	0.11	0.014
Eaten at acquisition (kg)	8.2	0.11	0.014
Returned to base (kg)	0.0	0.00	0.00
Baobab
Edible wt taken (kg)	52.0	0.71	0.09
Eaten at acquisition (kg)	2.4	0.03	0.004
Returned to base (kg)	49.6	0.68	0.09
Berries
Edible wt taken (kg)	1.8	0.02	0.003
Eaten at acquisition (kg)	1.8	0.02	0.003
Returned to base (kg)	0.00	0.00	0.00
Geophytes
Edible wt taken (kg)	4.7	0.06	0.008
Eaten at acquisition (kg)	1.0	0.01	0.002
Returned to base (kg)	3.7	0.05	0.006

Apis honey was pursued by groups of 2–16, most often a man, his wife, and 1–2 teenaged boys or another adult man. Subteen children were rarely involved. All party members ate honey at the point of acquisition, regardless of their relationships with FM; all party members participated in transport. What they carried away, they mostly kept in their own households on arrival at camp.

Consumption displayed strong intra‐seasonal patterns. On eleven early WED follows, FM and others present ate about 84% of an estimated 129 kg of honey and comb taken from 43 hives. Mean amounts consumed were estimated to have been as high as 3–5 kg per party member (this is not a misprint) per follow. The remaining 16% (c. 20 kg) was carried to base by various party members including FM and largely consumed in their respective households. In contrast, on thirteen late WED follows, party members ate only about 33% of an estimated 187 kg of honey and comb from 52 hives (mean < 1.5 kg/party member), with the balance brought to base. There, 30%–60% of the estimated 126 kg transported was set aside for trade, leaving 50–90 kg to be eaten in camp, most but not all in the households of those who brought honey home, including but not limited to those of focal men. Trade in honey was conducted with Datoga visitors to Hadza camps or taken by Hadza who had acquired it to Iraqw mead‐makers in Yaeda Valley 5–6 h walk away. In‐camp exchanges involved cloth and metal items; those with Iraqw included tobacco, marijuana, maize, millet, or mead itself. In our experience, the last was always drunk immediately by members of the transport party. Other commodities were carried to base. The amounts involved were small, less than a few kilos in any resource category per incident.

Big game were shot with poisoned arrows or seized from lions or hyenas [20, 30]. Having gained control of a carcass, those present often sent someone to base to gather a transport party, then immediately began dismemberment, consuming some parts in the process. Members of the carrying party joined in on arrival, subteens occasionally among them. Distribution of meat and marrow‐bearing bones to those present began at the kill. Fat‐rich organ meat and certain other high value parts (epeme) were set aside for consumption in private by initiated men. Some other flesh was eaten, but special attention was given to within‐bone nutrient‐bearing parts, including the skulls and mandibles of zebra and the skulls, mandibles, and long bones of antelope and giraffe. Half the contents of about 500 such parts from a 54‐carcass subset were consumed at kills [20, 21]. Men responsible for the kills ate some, but most of this tissue was appropriated by others with no apparent favor to members of the credited hunters’ households.

2001–2013. Berbesque et al. [23] say that in their 118‐follow sample berries (99%) and honey (84%, undifferentiated by type) taken by FM were all or nearly all consumed at the point of acquisition, but that significant fractions of baobab (37%) and small game (45%) came home. Direct comparison of Berbesque and colleagues’ data with ours is problematic in that much of theirs is drawn from Mangola and Siponga, where Hadza reliance on wild foods was marginal. Ideally, we would compare the Tli'ika/Han!abi subset with ours from the same area. Berbesque and colleagues' percentage breakdowns suggest that this should be possible, but Brian Wood denied our request to make them available for inclusion here. Big game meat taken in Tli'ika/Han!abi is an exception. Published data [24] indicate that an estimated 3 kg of 120 kg (2.5%) taken from a single kill were eaten at the point of acquisition, similar to the relative amount (40 of 965 kg, 4%) consumed at those locations in our six‐carcass sample (Table 4).

Regardless of complications with comparisons, the differences and similarities between these data sets are obvious. In both samples, big game meat was largely brought to base, stingless bee honey and berries were all or nearly all eaten in the field, as was most of the Apis honey. Baobab in our sample was mostly bought home, and small game never. In W/M's, the patterns for both were mixed.

6. Base Camp Sharing and Consumption Patterns

1958–61. Woodburn's observations are these:

[Having killed a large animal, the hunter is recognized as the owner but] has little influence over who gets the meat. … The people's meat [apart from epeme] is widely distributed among all the men, women and children of the camp unit. … Shares are asked for, even demanded. … People believe that they are entitled to their share and are not slow to make their claims. … The donor has no choice over whether the meat is shared and little influence over who gets what (1998: 49–52).

Woodburn makes no mention of big game meat from a hunter's kill being distributed preferentially to his own household. The partial exceptions involve certain fat‐rich parts assigned to the wife's parents and the bone marrow of certain animals set aside for nurslings while they were still very small.

1985–1990. As many have observed, food that reaches a Hadza camp may be claimed by anyone who sees it. This is especially the case with Apis honey, a highly valued resource. Collectors try to avoid such claims by hiding most of any honey they bring to camp, setting out on arrival only a small portion for non‐household members to eat, sometimes none. In early dry 1986, they adopted two related tactics, first by relocating shelters to double the average inter‐household distance between nearest neighbors, thus limiting camp residents’ ability to monitor other household members’ consumption, especially after dark; second by asking JOC for permission (always granted) to store honey set aside for trade in the project Land Rover and thus under his protection (never challenged) until they retrieved it for carriage on foot to Yaeda.

Big game meat: Over 256 observation‐days (2076 hunter‐days, 1985–1989), Hadza living in the camps where KH and JOC also resided (hereafter ‘home camps’) took 71 large animals, on average about one per hunter‐month [22]. Meat from these kills was always widely distributed across the hunter's camp. Detailed sharing data cover 113 distributions to nuclear family households from 20 kills, 16 made by hunters from home camps, four by hunters in other camps where folks from home camps traveled to take shares [29]. Prey contributing to the twenty‐kill sample included giraffe (n = 5), eland (1), zebra (9), impala (3), and warthog (2). Animals from the lower end of the prey weight range (small and mid‐sized antelope, mean adult weights 40–150 kg) are underrepresented in this set relative to their presence in the overall 1985–1989 big game kill array. Of the four animals killed by men living in other camps, two were giraffe, two were impala.

We focus here on the intra‐camp distribution of 13 prey taken by adult men in home camps (Table 5). These included giraffe (1), eland (1), zebra (7), warthog (1), and impala (3). Shares from these animals were claimed widely in home camps and for larger prey (zebra, eland, and giraffe) by visitors from other camps as well. Our data pertain to shares that went to 4–8 nuclear family households in home camps. Median share distributions per carcass ranged from 1 to 24 kg/household, mid‐point 10 kg. Distributions to households of men identified as having acquired the prey in question ranged from 0 to 50 kg. Eight of thirteen distributions to acquirers ranged from 12 to 50 kg—all above the central tendency by weight for the thirteen‐carcass sample. The percentage of edible shares per carcass that went to the acquirer ranged from 0% to 39%, mid‐point 15%. Five of six prey with values > 15% were zebra and eland. Share weights in these five cases ranged from 8 to 50 kg. Larger prey yielded relatively greater percentages of edible tissue to the acquirer. In other words, acquirers’ households generally got more meat and a greater percentage of available meat from larger prey in this sample than did other nuclear family households.

Table 5.

Meat shares assigned to credited hunters’ households, 1985–1986.

Date	Prey	Acquisition	Available	Number of	Median	Hunter's	Hunter's
		method	weight (kg)	hsld shares	share (kg)	share (kg)	share (% kg)
August 07, 1986	Impala	ambush	24	5	1	0	0
December 11, 1985	Impala	encounter	24	6	2	1	4
October 09, 1988	Zebra	scavenge	96	6	4	3	3
October 22, 1985	Warthog	encounter	27	4	4	4	15
May 19, 1986	Impala	encounter	24	6	2	8	33
October 21, 1985	Zebra	ambush	120	6	12	12	10
October 31, 1985	Zebra	ambush	120	5	21	16	13
June 13, 1986	Zebra	encounter	120	8	12	19	16
October 21, 1985	Zebra	ambush	120	6	15	26	22
May 28, 1986	Giraffe	scavenge	360	6	10	33	9
October 01, 1988	Zebra	encounter	120	5	20	37	31
August 01, 1986	Zebra	encounter	120	7	8	47	39
December 29, 1985	Eland	encounter	204	4	24	50	25

What to make of this relative to the paternal provisioning model? Our initial meat sharing paper says this:

The household meat weights (Data set 5, table 1, detailed in Appendix A) provide a rough estimate of both the fraction of a hunter's kill eaten by members of his own family and the fraction obtained from kills made by other men ([29]: 122).

The word eaten is not only wrong in this context but is contradicted by comments at several other points in the same paper. Here we reaffirm those contradictory comments. First, as noted above, consumption at the kill often included significant amounts of high‐value bone marrow and brain cavity contents. Anyone present could have had access to some of this; many did. The other within‐bone nutrient parts were claimed by members of various households present at the kill, not just those of the men who acquired the carcasses, and taken to base. And as we said, certain fat‐rich parts were always set aside for consumption by initiated men. Thus, only a fraction of the most valuable tissue went to the acquirers’ households and ultimately a still smaller fraction to their nutritionally most vulnerable subteen members.

Second, the largest amounts of meat from larger prey in the household share sample that reached home camps were beyond the capacity of acquirers’ households to consume before others came to take part. Word of big kills spread quickly beyond home camps prompting residents of other camps to arrive within hours, claiming shares. This pattern is reflected in the fact that 20% of our share sample came from kills made by nonresident hunters. Our campmates had gone to collect it. Our men's focal follow data show six cases (EW‐10, EW‐13, LW‐7, LW‐9, ED‐3, LD‐19 [19] Tables S1–S2) in which 1–18 adults from our home camps visited other camps soon after hearing of large kills. Not only did they stay for several hours eating meat, but in at least two instances, they carried meat away. Disproportionately large shares initially assigned to an acquirer inevitably drew special visitor attention [31]. While the acquirer could not deny such claims, the fact that he had “extra” to surrender meant that he got credit for doing so, which is important in status assessments related to big game hunting [22, 29]. In short, disproportionate shares to those responsible for a kill were not necessarily eaten in their own households. Contrary to assertions by other analysts, initial distributions are not a prima facie marker of differential paternal provisioning ([32]; cf [24, 33]).

In short, support in these data for differential paternal provisioning via big game meat is at best questionable.

1995–2013. As indicated above, most data on resources carried to base in the Wood/Marlowe 40‐follow sample are not available to us. Comparison with data from our 73‐follow sample (Table 4) would have been useful.

Wood and Marlowe [26] report non‐follow data on foods brought to base camps over 1496 man‐days in Tli'ika/Han!abi, 2006–2009 (Table 1). Presented as returns per man‐day, man‐days per return, and weights per man‐day returned, the constituent values differ sharply from those recorded on our 1985–1986 focal follows (Table 6, Figure 5). Men in the Utah/UCLA sample brought big game meat to base eight times as often as those in the W/M set; associated weights were 12 times greater. Honey was brought back at similar rates in both samples, but the weights involved were four times greater in our set. The values were reversed for small game and fruit: never in the Utah/UCLA sample, in the Wood/Marlowe tally one and two kg every 10 man‐days for small game and fruit, respectively. W/M cite numbers of times small game, honey, and fruit were returned to base, but the weights involved were meager.

Table 6.

Number of instances and associated weights of foods returned to base by adult male residents.

Items returned to base, mean per man‐day
	Large game	Small game	Honey	Fruit
Utah/UCLA 1985–1986	0.08	0.00	0.26	0.00
Wood/Marlowe 2006–2009	0.01	0.09	0.27	0.10
Mean man‐days per item returned to base
	Large game	Small game	Honey	Fruit
Utah/UCLA 1985–1986	12.2	> 73	3.4	> 73
Wood/Marlowe 2006–2009	88.0	10.5	3.7	10.2
Mean wt (kg) returned to base per man‐day
	Large game	Small game	Honey	Fruit
Utah/UCLA 1985–1986	11.71	0.00	0.96	0.00
Wood/Marlowe 2006–2009	0.92	0.12	0.23	0.23

Figure 5.

(a) Mean number of acquisitions per man‐day returned to base, (b) mean weight (kg) of acquisitions per man‐day returned to base. Data from Table 6.

The same W/M data set also monitors resource weights distributed to individual households. The sample is sizable: 312 shares—36 for big game, 54 for smalls, 101 for honey, 21 for fruit, one each for bee brood and tuber. All but big game are recorded as having gone to the household of the man responsible for taking the resource. For big game, the pattern resembles the one we saw: hunters initially kept significantly larger shares of the largest animals. Note again, however, that initial allocations do not necessarily monitor consumption, especially when disproportionately large shares are involved. W/M claim to control consumption for resources carried off by visitors, but data on big game are not reported ([26]: 293, Table 3).

In another non‐follow sample, Marlowe [27] reported observations on foods brought to camp during his 1995–1996 fieldwork. He said that meat and honey were the main items men carried home, but that berries, baobab, and small game also came back frequently. Quantitative data comparable to ours cannot be extracted from his account, but men's behavior with respect to the latter resources does show that they brought back more of these foods than we saw a decade earlier. Marlowe also documented the differential distribution of some of these resources (but not big game meat) to nursing wives and those with children younger than age eight ([27], 221–24). Stibbard‐Hawkes and colleagues’ [34] analysis of relationships between a nutritional marker of meat consumption in hunters’ diets and those of their wives vs. men's big game hunting reputations shows no correlation (cf [35]). This result is consistent with Marlowe's [27] point about the relative unimportance of big game vs. other resources as paternal provisioning items. It also matches our own observations on big game meat and paternal provisioning: no unambiguous differential benefit from that resource to successful hunters’ households.

Limited paternal provisioning at base camps is apparent in the Marlowe group's data. The small game and fruit amounts shown to be shared with families are greater than those we and Woodburn observed, but are not part of standard versions of the hunting hypothesis.

7. Discussion

Four issues are raised by this review: Why does Hadza men's foraging and food sharing behavior differ so sharply 1958–1990 vs. 1995–2013? Do Hadza men practice paternal provisioning? If not, or more precisely if their selective contribution to their younger children's welfare is as limited as it seems, why do women pair up with them? What do the results of this comparison imply about men's behavior toward their mates and offspring at the origin of genus Homo?

The main difference between the pre‐ and post‐1990 samples is the acquisition rate for big game. Our figure from Tli'ika/Han!abi was one every 29 hunter‐days over 2072 hunter‐days ([36]: Table 1). Writing on fieldwork with Hadza in 1978, McDowell [37] reported average large game acquisition rates around Mangola of one animal every 35 hunter‐days on a sample of 3400 hunter‐days, a similar value. In contrast, Wood and Marlowe [26] cite a rate of just one large animal every 103 hunter‐days over 1952 hunter‐days for areas east of Eyasi, 2007–2009. The rate in earlier decades was almost three times higher. For wet and early dry seasons, our 1985–1989 data show one large animal taken every 39 hunter‐days (total hunter‐days 732), Wood and Marlowe's 2007–2009 data show one every 241 hunter‐days (total hunter‐days 1447). Our rate was more than six times higher.

Why the difference? Tanzania witnessed major changes in demography and economy from the late 1980s onward [38]. Country‐wide, population grew at an annual average rate of three percent; agriculture spread into areas marginal for the practice ([16]: 58–60). Population in four census districts bordering Hadza country (Karatu, Mbulu, Meatu, Ngorongoro) more than doubled between 1988 and 2012, from about 500,000 to > 1,000,000 ([16, 39]: 195–96). A parallel increase in bushmeat trade from the 1990s onward, mostly involving big game, is also reported ([15]: 285–289 [16]: SI 3.5). Aerial counts across Hadza country in 1977 showed that big game densities were low but consistent with annual average rainfall in comparison with 15 East African game reserves ([30]: Figure 1). Aerial counts in 1989 and 1992 show that they had fallen east of Lake Eyasi by > 50%, although the fact that they also show a drop in domestic livestock numbers makes us skeptical of their accuracy. Differences in the altitude and speed at which later counts were run (higher, faster) may be partly responsible. Nevertheless, a comparison of on‐the‐ground encounter rates, ours versus Wood and Marlowe's, points to a sharp decline in big game numbers over time between the two samples, which is almost certainly related to local human population growth. In 1985–1986, we saw 398 large game animals (14 taxa) on 73 focal follows—0.69/follow‐hour. Members of the Marlowe group operating in the same area 20–28 years later saw 46 big animals (eight taxa) on forty focal follows—0.22/follow‐hour. Our rate was three times higher.

Do Hadza men practice paternal provisioning? Neither Woodburn's observations nor those of the Utah/UCLA group show much evidence of this. Big game meat was distributed as a public good, with no unambiguous advantage to successful hunters’ households. At an average acquisition rate of one large carcass every month in our sample, it was an important food source, but one that was unreliable on a day‐to‐day basis. Our 1985–1988 data show periods of several weeks in which no big game meat was available in camps where we were living ([19], Supplementary Material). Apis honey did play a role in family diets, but roughly 80% of the amount collected on our focal men's follows was either eaten by members of collecting parties, where nutritionally dependent children were seldom represented, or sent away in trade. Most of what came back was available only in the wet and early dry, and even then, unpredictably day‐to‐day.

In the Marlowe group's focal follow samples, honey was even less important. Berbesque and colleagues’ 118‐follow sample shows that about 85% of the honey taken (undifferentiated by type) was consumed at the point of acquisition, mostly by the single men who collected it. Wood and Marlowe's 40‐follow subset (2014: Table 1) includes 42 encounters with Apis hives, 77 with Trigona nests (ratio 1:1.8). Our 73‐follow sample ([19]: Supplementary Material, Tables S1–S2) shows the reverse: 147 encounters with Apis hives, 67 with Trigona nests (ratio 2.2:1). Mean weights taken per party member‐day were also very different: mean 0.3 kg in the W/M data set, mean 3.0 kg in ours, a function of the differences in types of honey taken. Apis hives generally yield weights measured in kilos, Trigona nests weights measured in grams. This contrast is reflected in comparison of counts versus weights of honey return to base (Table 6, Figure 5), similar across the Utah/UCLA and Marlowe group samples by count, but four times greater in the former by weight. Wood and Marlowe's insistence notwithstanding, nutritionally effective paternal provisioning is not apparent in their data.

Why pair bond? As documented elsewhere, children's nutritional needs into adolescence are covered mainly by women, initially by nursing, then by resources that women collect nearly every day [18, 40]. Unlike on men's out‐of‐camp forays, weaned subteens routinely accompany women's foraging parties targeting geophytes and berries. Women also bring sizable amounts of these foods to camp to feed youngsters left behind. Children themselves begin collecting seasonally significant amounts of foods on their own by age five (e.g., [41, 42, 43, 44]). Importantly, recall that women need not be married to successful hunters to claim shares of big game meat.

The answer to the question of pair‐bonding lies in elements of human life history. Women's fertility ceases before menopause, generally by ages 45–50 years, while men's continues for decades thereafter. As a result, mating age sex ratios are male‐biased—more men than women in their fertile years. In the Hadza case, that ratio is 1.6:1 ([45]: 11801, Table 2). Competition among men for paternities is inevitable and favors a pattern of mate guarding [45, 46, 47], which plays out with pair bonds forming nuclear families. But where husbands and wives spend most of their waking hours apart, why are men's proprietary claims on their wives respected? Although it is not men's only behavior that does so, big game hunting helps set the groundwork. It allows men to develop reputations for providing their communities with large quantities of an unpredictably acquired but highly valued, widely shared resource, a practice that earns respect and deference ([48, 49]; see also [50]). The fact that big game hunting and aggressive scavenging are dangerous practices involving confrontations with large carnivores enhances the appeal of successful hunters as attractive allies and potentially formidable antagonists. The fitness‐related benefit men gain in the process is not the differential nutritional welfare of their wives and children, but the advantage enjoyed in mating competition ([16]: 274–76 [9, 51]). Woodburn [52] shows that a husband is essential to his wife and children's well‐being, not least in protecting the wife from other men's interference. Hunting reputations earn others’ deference to the proprietary mate guarding that allows better hunters to claim and hold fertile, more productive wives through more of their adulthood. The basic social unit consists of adult women of reproductive age, their children, and their own mothers or close senior female kin. Men attach themselves to the fertile age women while avoiding direct contact with their senior helpers [17]. The resulting arrangements resemble nuclear families but not as these entities are customarily defined.

How do these observations play out at the origin of Homo ? The answer turns on the model of life history evolution presented in the grandmother hypothesis. Developed in detail elsewhere ([18, 40, 48, 53, 54, 55]; see [56] for recent summary, also Charnov [57] for theoretical background), the main elements are these:

As in the hunting hypothesis, the hominin baseline was a Late Pliocene socioeconomic arrangement like that known among living chimpanzees. Diets were broadly similar between the sexes, but adult males may have hunted and consumed small animals. Infants began to pick food for themselves while still nursing and became fully independent feeders by age 4–5 years. Sexual maturity was reached at about age twelve. The reproductive age sex ratio was biased toward females. High ranking males monopolized copulations with older females when conception was most likely but otherwise maintained no close connections with them apart from living in the same community. Lifespans seldom exceeded 45 years. Aging faster than modern humans, females usually died before menopause. The evolution of Homo beyond this baseline was provoked by Early Pleistocene changes in climate and environment that restricted tropical forests and fostered the spread of savannas. Foods that weaned children could collect on their own became less available. Other foods, accessible to adults and better able to survive cooler, drier, more seasonal conditions became more abundant. Women began provisioning those children with these resources. Senior women past menopause contributed by feeding their daughters’ children. More active/physically capable seniors contributed more effectively, enhancing their grandchildren's survivorship and favoring the transmission of traits that contributed to that enhancement. Extended adult life expectancy, earlier weaning, shorter birth intervals, overlapping juvenile offspring, and delayed maturity all followed.

Brain size in placental mammals depends on the duration of development, which in turn varies with average adult lifespan [58, 59], allowing changes in the latter to be tracked in the fossil record. In humans, the shift toward extended longevity began with the origin of Homo, 2.8–2.0 Ma [5, 7]. The transition toward a male bias in the mating age sex ratio leading to an increase in mating competition and the onset of a pattern of mate guarding started at about the same time, probably becoming significant by 2.0 Ma. These developments occurred at a time when big game hunting opportunities were markedly greater than they have been in the recent past [60, 61]. Ungulate taxa were more numerous, especially in larger body size ranges. A larger predator guild created more scavenging opportunities [62]. Archaeological data from 2.0 to 1.6 Ma show that early humans acquired large animals in complete or nearly complete condition by this time [63, 64], possibly before [65], arguably due to an increase in mating competition.

Could differential paternal provisioning via the products of this success have been practiced? We think not. Blurton Jones’ [31] model of tolerated theft shows that where resources are taken unpredictably in large packages, as would likely have been the case in these early days, the products were probably distributed widely across members of local groups, regardless of who was responsible for acquiring the kill in the first place. Large carcasses could not have been defended against claims from those who had none. The distribution of big game meat within those local groups should have followed the pattern seen among the modern Hadza: meat and marrow distributed as a public good, with no advantage to the close kin of men responsible for its initial acquisition. Nuclear family‐like formations, defined in part by males establishing semi‐exclusive relationships with reproductive age women, and indirectly the latter's senior female relatives, should have been the emergent social form. Paternal provisioning will have had little to do with it.

8. Conclusion

Nuclear families are widely identified as fundamental elements of the human condition. The hunting hypothesis implicates them in the origin of Homo. Given their presence in an apparent homeland of that genus, one might have expected the Hadza to provide data or at least the basis for a theoretical argument supporting that hypothesis, but they do neither. Until recently, big game hunting has been a major food source for Hadza, yet evidence from those days past shows that it does not offer the differential advantage to successful hunters’ families essential to the big game hunting argument. This doesn't mean that hunter‐gatherers in general never practice paternal provisioning. They certainly do in some situations (e.g., [10, 11]) but apparently not in the modern environment most suitable for comparison with the ones in which early Homo evolved. Our report on men's focal follows [19] leads us to confirm observations made by Woodburn as well as our own. Paternal provisioning is effectively absent in these data, with limited exceptions of uncertain significance in some of the big game meat distribution data and in Apis honey, which is seasonally limited and largely consumed away from nutritionally dependent subteens. It also allows us to reaffirm the contrast with Wood and Marlowe's data gathered since 1995 [32, 36]. The differences between these accounts reflect significant changes in regional ecology that affected hunters’ foraging success. Though paternal provisioning is indicated in the Wood/Marlowe data, it is not the substantial family food source that the hunting hypothesis requires. Women's work is what supports nutritionally dependent children day‐to‐day, year‐round (e.g., [17]: 249 [9]; cf [3]). A better evolutionary argument can be made for nuclear families as products of mate guarding, with advantages for men in mating competition and for women in protection from harassment provided by publicly consenting to paired status.

Ethics Statement

Approval for fieldwork and reports by Blurton Jones, Hawkes, and O'Connell cited here was granted by Institutional Review Boards at University of Utah and UCLA; for fieldwork reports by members of the Marlowe group (Marlowe, Wood, Berbesque et al.) approval came from IRB at Harvard, Florida State and UCLA.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

All Hadza data reviewed here have been published elsewhere ([9, 14, 16, 19, 20, 21, 22, 23, 24, 25, 26, 27, 29, 30, 36, 66]). The authors have nothing to report.