JOURNAL OF ANTHROPOLOGICAL ARCHAEOLOGY 12, 323-385 (1993) Variability in Faunal Assemblages: The Influence of Hunting Skill, Sharing, Dogs, and Mode of Cooking on Faunal Remains at a Sedentary Kalahari Community Susan KENT Anthropology Program, Old Dominion University, Norfolk, Virginia 23529 Received October 2, 1992 Variability in faunal assemblages from different sites and/or from different time periods is often attributed to economic or taphonomic factors. The role of sharing ‘on faunal remain distributions is compared to other factors that have been sug- gested to influence these distributions, such as hunting skill. Faunal species and skeletal elements are compared among three hunter-gatherer camps that form a sharing network. These are contrasted with those of two other hunter-gatherer camps located at the same Kalahari community occupied by an unusual family that is a relative isolate in terms of sharing. The effect of sharing on equalizing variation in hunting success as reflected in the faunal remain inventory is explored from the five camps inventoried in 1990. Complicating factors which tend to affect faunal remain frequencies are also examined, such as cooking technique and dogs. All faunal remains visible on the surface of each camp were recorded according to species, element or fragment portion, age (mature or immature), and, when pos- sible, side, At all but one camp, surface faunal remains were recorded both before and after ethnographic observations during the dry season of 1990. In addition to hunting success, all occurrences of sharing and consumption of meat were re- corded during these observation periods and those conducted on and off between 1987 and 1992. Although participation in a sharing network obscures differences in hunting skill in the archacological record, sharing impacts on faunal assem- biages in interesting ways that are potentially atchaeologically visible. Sharing in strongly egalitarian societies levels unequal hunting skill that could otherwise affect faunal remain frequencies, taxa richness, meat weight, and other indices measured here. In these societies, sharing reinforces social bonding between kin and friends in ways that help unit families from different camps. Zooarchaeologists have become accustomed to high levels of confidence in their inferences about the origins, functions, and responses to stress of animal remains, This confidence rests on the causal and functional links between attributes of these remains and the processes and contexts which generate them. Their investigations are presently moving toward wider inferences about the context and functions of bones in ancient hominids’ behavioral systems. . . . Zooarchaeologists now need a dif- ferent set of inferential strategies than that which characterized their preceding phase of research. (Gifford 1991:215) © 1993 Academic Pres, Ine. 323 0278-4165/93 $5.00 Copyright © 1993 by Academic Press, Inc. All rights of reeroduction in anv form reserved.324 SUSAN KENT INTRODUCTION The reasons postulated for variability in faunal assemblages between sites and/or time periods are often attributed to differences in subsistence procurement strategies, animal availability, preservation, and predator scavenging or other taphonomic processes. More recently, some archae- ologists have suggested that the sharing of meat (on bones) within and between camps is visible in faunal remain frequencies and/or distributions (Enloe in press, Hudson 1990a,b; Parkington 1992; Yellen 1977). This article explores the impact of ethnographically observed patterns of shar- ing (as well as hunter skill) on faunal assemblages, keeping other variables constant. Sharing in forager societies is usually viewed from ecological and eco- nomic perspectives by anthropologists with theoretical orientations that emphasize those variables (see Kent 1993a; also Gould 1982; Hawkes 1991; Kaplan 1983; Kaplan and Hill 1985; Smith 1991; Winterhalder 1986; and others). A growing number of people are recognizing that archaeol- ogy over the past 30 to 35 years has been disporportionately oriented toward ecological/economic models of the past (e.g., see Watson 1991, 1992). Just as the most staunch ecological or economic materialist would not totally disregard social relationships as having no relevance of any kind to the past, I recognize broad parameters imposed by the environ- ment and a necessity for food and shelter. Where ecological/economic archacologists and I differ is in what we consider to be more important in explaining and understanding the past and what we consider to be more epiphenomenal. In my opinion, sharing as more of a social behavior than an economic one better accounts for the patterning of faunal remains at five sites located in a sedentary community in the Kalahari Desert of Botswana. However, I try to present enough details so that even the most ardent materialist can find something of interest in the faunal inventories collected from the camps of Kalahari hunter-gatherers. Too often anthropologists have lumped together hunter-gatherer soci- eties based on their similar emphasis on wild plants and animals, while ignoring their many differences in social and political organizations, among other facets of culture (Kent 1992a, nda). A more appropriate grouping of forager societies has been proposed by Woodburn (1982, 1988) who separates societies on the basis of egalitarianism. No one so- ciety is egalitarian per se, but all are relatively more than, the same as, or less than one another, Whereas I use his definition of ‘‘assertedly”’ egal- itarian societies to designate those groups located on one extreme of the continuum, I do not agree with his cause of such egalitarianism as en- trenched in an immediate-return economic system. Sharing and extreme egalitarianism are related in that sharing helps maintain an atmosphere ofVARIABILITY IN FAUNAL ASSEMBLAGES. 325 egalitarianism in those societies that have been referred to as “‘assert- edly”? egalitarian. I think the fact that sharing in these societies integrates people, by fostering social relationships, as pan-tribal organizations do in slightly more sociopolitically complex societies is at least as if not more important as economic explanations (Kent 1993b). From the perspective of my theoretical orientation, sharing is linked to the basic organization of egalitarian societies in that creating and main- taining social bonds unite people. In this sense, sharing is the basic ad- hesive that holds the society together. The key to my orientation is that sometimes sharing will have an economic component, but not always; invariably, however, sharing in strongly egalitarian societies will have a social component in that it promotes social bonding (Kent 1993b). To validate this view, | need to show that sharing has less to do with eco- nomics than it has te do with social relations, It should be kept in mind that I am only referring to highly egalitarian societies. Other less egali- tarian societies use different principtes to bind people together, such as clans or age-sets. More complex societies also have solidarities, formal sociopolitical leaders, and hierarchical stratification, as well as other means of achieving group cohesion (e.g., Northwest Coast fisher— hunters-gatherers}. These simply do not exist in extremely egalitarian societies; they instead have sharing within an egalitarian ethos. There- fore, sharing patterns are not necessarily the same nor are they caused by the same factors, such as economics, in all societies. If the social relationship model of sharing has any validity, then data from highly egalitarian societies, in this case the faunal assemblages, should exhibit evidence of sharing as motivated by the need to establish and perpetuate social bonds between camps. Sharing should furthermore be visible as a leveling mechanism to equalize unequal situations, such as those resulting from hunting success rates that vary between individuals, as has been noted by Lee (1979) and others (e.g., Marshall 1976). By viewing only sedentary foragers, disposal patterns in this study have been kept constant (this is important because elsewhere I have suggested that disposal patterns are partly, if not primarily, dependent on antici- pated mobility patterns or how long people plan to occupy a camp; Kent 1993a). It is necessary to also investigate any potential intervening factors in order to separate them from sharing. These include determining the impact of mode of cooking and the influence of dogs on the archaeological visibility of faunal skeletal elements and taxa. HYPOTHESES AND DATA The nature of sharing networks and their impact on faunal remain dis- tributions were studied among hunter-gatherers at Kutse, a sedentary326 SUSAN KENT community situated just outside the Khutse Game Reserve (to keep the community distinct from the game reserve, I use the alternate spelling, “‘Kutse,”’ to refer to the recently sedentary settlement; Fig. 1). About 70-80% of the Kutse community is composed of Basarwa (‘‘Bushmen"’ or San). A few the Basarwa speak KUa, the majority speak G/wi or G//ana (most of the latter two originated in the Central Kalahari Game Reserve just north of Kutse). Bakgalagadi (Bantu-speakers) account for the re- maining population (see Kent 1989 and 1992a for a further description of BOTSWANA Kutse Study Area - Intl Boundaries: —— AFRICA Game Reserves = —_____ NAMIBIA > cf a Khutse Game. Kutse Reserve Study Area { svete N Game Feserve Gaborone ap~* Samak 7 SOUTH AFRICA Pak oT \ LN f . sf oO 100 200 300 km Fic. 1. Location of the study area,VARIABILITY IN FAUNAL ASSEMBLAGES 327 Kutse). Depending on the year and season, the population averages around 90-130 persons. Bakgalagadi at Kutse differ from the better known village Bakgalagadi in that, and with only two exceptions, none own any cattle, many do not own any goats, and some never plant a garden. They tend not to be as hierarchically stratified and gender differ- ences do not appear to be emphasized as much as they are by village Bakgalagadi. Kutse Bakgalagadi tend to depend on wild game for a vast majority of their meat (Kent ndb). Quantitative ethnographic observations on sharing and hunting and in- ventories of all surface faunal remains were conducted at five camps selected from the larger Kutse community during the dry season of 1990. The camps were selected based on the fact that all but two belong to one sharing network and all have the same approximate population, except for one camp where it was easy to factor out the influence of others. Gener- ally, the camps were similar—for example, the camps each had roughly the same number of dogs and a similar number of dependents per hunter, and occupants were familiar with me and my work. The two camps cho- sen for detailed study in 1990 that did not participate in a sharing network were selected for that reason and because the occupants represent the only sharing isolate I know at Kutse. At the four occupied camps ob- served in 1990, complete surface inventories were compiled when I first arrived at each camp and again just after I left 2 to 7 weeks later. One 1990 inventory of faunal remains was made at a camp that had been occupied in [987-1988 and therefore separate before and after inventories were not possible. These data are used to evaluate the following hypotheses: 1. participating in a sharing network levels or equalizes discrepancies in hunter success and skill in archaeologically visible ways; and 2. cooking technique, i.e., boiling in containers versus roasting in pits, results in an increase in unidentifiable bones at camps. If hypothesis [ is valid, there should not be any discernable differences in the faunal assemblage that can be attributed to variation in hunting skill among hunters living at different camps that participate in a sharing net- work. That is, MNI (minimum number of individuals), NISP (number of individual specimens), taxonomic richness, and other commonly used measures should be similar at camps, regardless of the skill of resident hunters, within the same sharing network. In addition, if sharing is pri- marily economic in nature, we might expect differences between camps that do and those that do not participate in sharing networks. These differences in sharing patterns should be visible in variables that indi- rectly measure economic factors, such as MNI, NISP, and meat weight328 SUSAN KENT calculated from MNI, In other words, there needs to be some way to measure economic gain. | suggest that if preservation, animal availability, and taphonomic processes are kept constant, higher MNI and NISP val- ues would indirectly indicate higher meat availability. If these higher values were present at camps that participate in a sharing network, I would interpret them as indicating a possible economic motivation to sharing. Conversely, if values remain constant for camps, regardless of whether members are sharing isolates or participants in a sharing net- work, I would interpret it as negating an economic motive for sharing (note that a preliminary content analysis of all possessions at Kutse camps that is still in progress indicates that the sharing isolate does not have more or less possessions than camps that form a sharing network). If hypothesis 2 is valid, there should be more unidentifiable faunal re- mains located at camps where the major mode of cooking was boiling in contrast to roasting due to increased fragmentation to make bones fit into a cast iron pot. The Study Camps Observations were made of 80 hunting trips and of sharing patterns for 36 animats acquired through hunting at the five camps. In addition, 3588 bones, including pre- and posthabitation inventories, were recorded (posthabitation bone count alone is 2126). Analyses reveal consistent trends and give us some understanding of the nature of variability in site faunal remains, particularly taxonomic richness and bone frequencies. Because the sample consists of surface remains from only five camps, the following should be viewed as exploratory, although the relationships are robust enough to formulate general conclusions. [had previously lived with three of the four families who occupied the five camps used in this study (those of Hunters 1, 2, and 3) during the 1987-1989 field seasons. I lived with the fourth family (that of Hunter 5) for the first time during 1990, although I had visited them at this and previous camps many times during other field seasons. The three camps that belonged to a sharing network (Camps 173, 177, and 181'*) were within ‘“‘shouting”’ distance of each other whereas the nonsharing camps were not, although all belonged to the larger Kutse community. By far the vast majority of families at Kutse belong to a sharing network based on a combination of friendship and kinship. Network composition varies con- siderably and there are many nonoverlapping sharing networks (Kent 1993a). At the time of the study, four of the camps had been occupied for 1% to 2 years; the fifth had been occupied for approximately 4 to 6 * See notes section at end of paper for all footnotes.VARIABILITY IN FAUNAL ASSEMBLAGES 329 months. To provide a control of yearly fluctuations in hunting activity, I lived with Hunter 2 and family each time I went to Kutse. It is rather unusual to find a family who is basically a sharing isolate. In. fact, Hunter 2’s family is the only sharing isolate I know of at Kutse. They represent just 0.03-0.04% of all adult Kutse residents, depending on the season and year. Hunter 2 and family, who inhabited the sharing isolate camps (Camps 162 and 175), socialized less often than others and rarely formally shared food outside of camp (Kent 1993a,b). They did have relatives living at other camps at Kutse; they just did not belong to a sharing network with them. Nevertheless, they still participated in com- munitywide meetings and other events. It is interesting to note that while being a social isolate did not economically affect this family because the hunter was fairly skillful, it did affect them socially in the amount of interaction they had with other families in the community. They were not ostracized in the community for their less social behavior; they simply were not visited as much as were people at camps within a sharing net- work. If people came by at an opportune time when meat was being consumed, Hunter 2 and family would give the visitors some meat to eat during their visit, although there was no camp with whom they regularly shared food. Actually it is probably not that uncommon to find that some families in a community are more social than others, although Hunter 2 and family were somewhat extreme in their lack of regular sharing part- ners. Hunter 2 and family provide an excellent opportunity to compare faunal remain frequencies at camps that belong to a sharing network with those that do not. The Kutse camps also provide an opportunity to study the impact of cooking methods on faunal remains. Families at two camps regularly cooked wild meat (and wild plants) in roasting pits, while two boiled meat in medium-sized cast iron pots (wild plants are usually roasted). Another camp tended to both boil and roast, with a bias toward boiling (Table 1). As many variables as possible were held constant in order to determine which factors affect the number and kind of bones and species repre- sented in surface assemblages of camps. For example, in addition to the similarities noted above, time spent hunting by different hunters was roughly equivalent, with the exception of Camp 162 (Table 1). None of the observed hunters had special nonhunting skills (e.g., to keep things as constant as possible, none of the hunters in this study was a ritual curer; i.e., there were no nonmeat benefits accrued from sharing with unsuc- cessfut hunters—therefore the suggestion that while less skillful hunters may have provided less meat to the sharing network, they could have contributed to skillful hunters’ well-being in other ways, such as provid- ing medicine, were not present among the hunters examined). Furthermore, faunal remains inventories were conducted while the330 SUSAN KENT TABLE | ‘SUBSISTENCE TIME ALLOCATION (FROM KENT NDB). No. time hunting! Hunter Camp Hunting Mode of + Camp —_No. successful/No. Hunts ID Year (hours/week) cooking population _—_animals killed ‘Hunter 1? 18119902034 Roast 6 16313 Hunter 2 1621987 1988 8.53" Boil 6 9/2 Hunter2 1751990 18.53 Boil 5 14Isi7 “Hunter 3% 1731990 — Boil 5 — “Hunter 5° 1771990 ‘19.94 Roast 7 41nsig Note. Observation days for 1990 unless otherwise noted: Hunter 1, 41 days; Hunter 2 (1987-1988), 33 days; Hunter 2 (1990), 14 days; Hunter 5, 50 days; Hunter 3, 40 days. Camp population includes all residents but excludes visitors (Hunters 3 and 6-8 are not a part of this study; their activities are described in Kent 1993b and ndb). @ Belong to the same sharing network. ° The less time spent hunting in 1987 and 1988 is partly because of the drought that ended ia 1987; animals did not become plentiful until 1989. © T did not tive in the camp of this individual in 1990 (I lived in the camp next to his). sites were occupied, with the exception of Camp 162 which had been abandoned for approximately | year. Differential bone survivorship was held constant because all of the sites with one exception were approxi- mately the same age. This is important for intersite comparisons as noted by Lyman (1984, 1985, 1991) and Stiner (1991, 1992). Lyman (1991) has documented that when comparing archacological sites that vary in age by many hundreds or thousands of years, differences in element densities can result in differential bone survivorship (also see Grayson 1989a). In addition, postabandonment carnivore activity has been suggested to dif- ferentially affect bones based on element marrow and grease content (e.g., Yellen 1991a) and to bias MNI comparisons (Marean and Spencer, 199f and references therein). Since, with the exception of Site 162, the camps examined here were all the same age and were subjected to ap- proximately the same amount carnivore activity (by a similar number of dogs), differences or similarities in the faunal inventories compared here were not the result of taphonomic processes. Ethnographic Data Collection The full study is based on research gathered during the dry season (May-August) over a 5-year period—1987-1991. During this period, one nondrought rainy season was also observed (December-January, 1989— 1990). This research includes 175 separate hunting trips of 6 hunters over 291 days between 1987 and 1991. In addition, observations of hunting at several other camps through spot checking brings the total number ofVARIABILITY IN FAUNAL ASSEMBLAGES 331 observation days to 369 and the total number of hunters observed to 8, which is 33.3% of all hunters who have lived at Kutse between 1987 and 1991 (not all live at the community at the same time). Hunting data are based on direct observations of when hunters left and returned to camp, what they acquired, who they shared with, etc. Interviews added details, such as if an animal had been caught in a snare but escaped or was eaten by a predator. It is important to note that the 369 observation days rep- resent only the times when the quantitative data were collected; addi- tional observation days were spent at different camps collecting more qualitative data, including interviews and comparing hunters (¢.g., to de- termine whether during the drought or the rainy season, hunters with comparable skill were bringing in comparable numbers of animals). My spot checks with different hunters at other camps (including 24 hunters), interviews (with 80 residents), and qualitative observational data all sup- port the quantitative data collected in 1990 presented below. Since the study project is continuing, additional dry seasons have been observed during the writing and publication of this article, increasing the amount of time I have recorded hunting and gathering and faunal remains at a particular community. The data from 1992 agree with the interpreta- tions presented here but only the quantitative data from 1990 are used, although in conjunction with the ethnographic data collected over a6 year period. The ethnographic data were collected a little differently than in many ethnoarchaeological studies in that I lived with a hunter and his family and therefore, except when a hunter was foraging or when a hunter was vis- iting or otherwise not in camp, I basically observed the hunter and family 24 hr a day. I did not have a separate hearth or windbreak or hut; I used the family’s windbreak hearth. Late when and where others did (as with any visitor to the camp, cooked meat was informally shared with me), I slept within meters of their hearth (j.e., 1 was within visual and hearing distance at night—I often woke up during the middle of the night because two people were whispering, an activity that was then recorded). I woke up when they did and generally went to sleep when they did (of course exceptions occurred when I was ill). Although an extremely demanding and difficult fieldwork situation, | thought it important to be able to ob- serve and record (through time allocation studies) what occurred in a camp during all hours and not just when an anthropologist happened to visit or just between 8:00 am and 6:00 pm. Living at a camp for weeks at a time meant that the informants habituated to my presence and that activities after the first day or two were more or less typical of when I was not present (I also conducted tests to monitor observer affect throughout my stay at a Kutse camp). At all times while at Kutse, I had a notebook and pen in my hands and332 SUSAN KENT I was usually writing notes. After the first field season, people got used to my constant writing. Since then, even the children have acclimated to my presence and continuous writing (although visitors to Kutse, including children, who have not seen me before are fascinated by it). As mentioned above, the members of the 1990 sharing network occu- pied camps within sight and shouting distance of each other. Therefore, I was able to observe when hunters from the other camps left and returned, as well as their success. The occasional instances when I misjudged when a hunter would return from foraging and was not in camp when he arrived, I would reconstruct the time of day through interviews concerning the position of the sun, etc. I also always inquired about hunting success after each hunter returned to ensure that very small animals that I may have missed, such as squirrels, were recorded, particularly at the monitored camps where I did not live. Interviews concerning butchering and sharing were also conducted whenever anyone obtained an animal, either through the hunter’s own efforts or through meat sharing from someone (the latter would then include females). After butchering and meat distribution, I would ask the hunter who he gave meat to and why (or sometimes why he did not give meat to specific people) in order to see if my observations matched the hunter’s explanations and to determine how the hunter viewed the distribution of meat. | also inquired if people consumed meat while visiting other camps. The quantitative data from 1990 used here are based on ethnographic observations which occurred over 7 weeks at Camps 173, 177, and 181; 5 weeks at Camp 162; and 2 weeks at Camp 175.? In addition to hunting success, all sharing and meat consumption were recorded during these periods. Bone Data Collection Inventories included all surface bones at each camp, regardless of size. A complete inventory was made when I first arrived at a camp and then again when I left (bones were replaced after examination). My field as- sistant and I walked transects across an entire camp to record bones. Location in a feature (e.g., ash area, windbreak, etc.) or not in a feature was noted. Faunal remains were recorded according to: 1. species, 2. skeletal element and/or fragment part (distal or proximal, anterior or posterior, shaft including what part of the shaft, articular end, etc.), 3. age (mature or immature), 4. presence of charring, 5. and, when possible, side.VARIABILITY IN FAUNAL ASSEMBLAGES 333 1 worked with the same field assistant during all my research at Kutse, with the exception of the 1992 field season, a situation which provided continuity in identifications. He and I both inspected every bone regard- less of size or condition. All bones, including shaft fragments, vertebra, and ribs were identified to element, even if the species was not known (except for coccyx, atlas, and axis, most vertebra were not identified further unless it was obvious that one was a cervical, thoracic, or lumbar; ie., I did not usually distinguish the second thoracic from the third). We both had to agree on the species and element identification before I re- corded the information. When we were uncertain, we sometimes would ask an older hunter or his wife for an opinion. After examining a bone, it was returned to the location at which we picked it up so that I could monitor the faunal assemblage during the life history, and later abandon- ment, of a camp. Although the recording was systematic and careful, bones were un- doubtedly missed. For instance, steenbok and larger taxa dominated the inventories. Smaller animals, such as fox or hare were probably under- represented due to the effects of trampling in the unconsolidated Kalahari sand. In fact, Yellen (1977a), working in Ngamiland, noted that small bones were often located beneath the surface, within the top several centimeters of soil. The manufacture of bone artifacts alters what is considered bone refuse. However, bone artifacts are not often used today, although pre- sumably they were more common in the past. Three bone artifacts still in common use at Kutse are antelope or goat tibia smoking pipes, antelope or goat cranial scrapers used to process small animals’ hides, and ostrich tibia used to mash up melon flesh to make a liquid for drinking (Figs. 2-4). These bone tools were not included in MNI or other calculations because they are usually curated. As such, they are not part of the faunal refuse assemblage and do not represent subsistence activities or food. Cranial fragments cannot be simplistically tied to MNI or amount of meat intro- duced to a camp until more research is conducted documenting the influ- ence of cranial scrapers on the abundance of cranial fragments. For ex- ample, we do not know if using crania as scrapers creates more fragments than if the bone were not used as a tool (e.g., are there more fragments because there were more crania used as scrapers at some camps which may have broken into more pieces because they were used as a tool or because people brought in more meat?). I do not think that the meaning of cranium frequencies can be evaluated at this point in time. MNI was calculated by dividing the number of each specific skeletal element by the number of that element in the animal per taxon. Ail iden- tifiable elements, except tools, were used in computing this figure, in- cluding fragmentary pieces. A consideration of animal age (based on tooth