GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 1 pinnacle Quark11:JOBS:BOOKS:REPRO:

1

Behavioral Ecology and the Transition from Hunting and Gathering to Agriculture

Bruce Winterhalder and Douglas J. Kennett

he volume before you is the first THE SIGNIFICANCE OF THE TRANSITION T systematic, comparative attempt to use the concepts and models of behavioral ecology There are older transformations of comparable to address the evolutionary transition from so- magnitude in hominid history; bipedalism, en- cieties relying predominantly on hunting and cephalization, early stone tool manufacture, gathering to those dependent on food produc- and the origins of language come to mind (see tion through plant cultivation, animal hus- Klein 1999). The evolution of food production bandry, and the use of domesticated species is on a par with these, and somewhat more ac- embedded in systems of agriculture. Human cessible because it occurred in near prehistory, behavioral ecology (HBE; Winterhalder and the last eight thousand to thirteen thousand Smith 2000) is not new to prehistoric analy- years; agriculture also is inescapable for its im- sis; there is a two-decade tradition of applying mense impact on the human and non-human models and concepts from HBE to research worlds (Dincauze 2000; Redman 1999). Most on prehistoric hunter-gatherer societies (Bird problems of population and environmental and O’Connell 2003). Behavioral ecology degradation are rooted in agricultural origins. models also have been applied in the study of The future of humankind depends on making adaptation among agricultural (Goland 1993b; the agricultural “revolution” sustainable by pre- Keegan 1986) and pastoral (Mace 1993a) pop- serving cultigen diversity and mitigating the ulations. We review below a small literature environmental impacts of farming. Simple pop- on the use of these models to think generally ulation densities tell much of the story. Hunter- about the transition from foraging to farming, gatherers live at roughly 0.1/km2; rice agricul- while the papers collected here expand on turists in Java at 1,000/km2, a ten-thousand-fold these efforts by taking up the theory in the difference. There were an estimated ten million context of ethnographic or archaeological case humans in the world on the eve of food produc- studies from eleven sites around the globe. tion (Price and Feinman 2001: 194); now over

1 GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 2 pinnacle Quark11:JOBS:BOOKS:REPRO:

six billion people live on this planet, an increase were transmitted broadly through preexisting of 600% in only ten millennia. Agriculture is the exchange networks (Hastorf 1999), stimulating precursor, arguably the necessary precursor, for the migration of agriculturalists into the territo- the development of widespread social stratifica- ries of hunter-gatherers, who were in turn ulti- tion, state-level societies, market economies, and mately replaced or subsumed into agricultural industrial production (Diamond 1997; Zeder economies (Cavalli-Sforza 1996; Diamond and 1991). Social theory (e.g., Trigger 1998) main- Bellwood 2003). tains that present-day notions of property, equal- Foraging peoples initiated domestication. ity and inequality, human relationships to na- They did so through the mundane and neces- ture, etc., are shaped, at least in part, by the social sary daily tasks of locating, harvesting, pro- organization, technology, or food surpluses en- cessing, and consuming foodstuffs. The Mass tailed in our dependence on agriculture. from the 1928 Book of Common Prayer Domestication today is a self-conscious en- (Protestant Episcopal Church 1945, 81) speaks terprise of advanced science and global-scale ef- eloquently of “these thy gifts and creatures of fort, an applied research endeavor comprised of bread and wine . . .” In less poetic non-ecclesi- thousands of highly trained and well-supported astical terms, but with no less awe at the high international specialists. Major research centers importance and, well, the simple gastronomic like the International Potato Center in Lima, pleasure of domesticates in our lives, this vol- Peru (www.cipotato.org/) support ongoing ef- ume attempts to advance our understanding forts to further the domestication of useful of why and how this happened. In particular, species; seed banks have been established in we hope to demonstrate the utility of a branch many countries to insure the future diversity of of evolutionary ecology, human behavioral the world’s key domesticated plants (www.nal. ecology. usda.gov/pgdic/germplasm/germplasm.html). The prehistoric beginnings of agriculture though DEFINITIONS were quite different. The modern world that funds and depends on this continuing process Clear, standardized terms for the biological and of domestication is, in fact, a creation of the first cultural processes involved in the origins of agri- early humans that pursued, consumed and, in culture worldwide remain elusive, despite con- doing so, modified the wild ancestors of the sta- siderable efforts to define them (Flannery 1973; ples that we consider to be important today— Ford 1985; Harris 1989; Harris 1996a and b; wheat, millet, sweet potato, rice, and domesti- Higgs 1972; Piperno and Pearsall 1998; Rindos cated animals such as camelids, pigs, sheep, 1984; Smith 1998; Smith 2001a; Zvelebil 1993; goats, and cows—to name a few. At present it Zvelebil 1995; Zvelebil 1996). The reasons for in- appears as if at least six independent regions of consistencies in the treatment of terminology are the world were the primary loci of domestica- several and tenacious because they are ultimately tion and emergent agriculture: the Near East; rooted in the nature of the problem itself. sub-Saharan Africa; China/Southeast Asia; These include, but are not necessarily limited Eastern North America; Mesoamerica; and to the following: (1) research on domestication South America (Smith 1998), roughly in the and agricultural origins is inherently a multi- time period from thirteen thousand to eight disciplinary activity, and as such, a wide-ranging thousand years ago (Binford 1971; Diamond set of specialists have worked on the problem, 2002; Flannery 1973; Henry 1989). The archae- each emphasizing definitions that are somewhat ological record suggests that this transforma- parochial; (2) historical change in each research tion took place in societies that look much like tradition of archaeology, botany, and genetics has modern day hunter-gatherers (Kelly 1995; Lee resulted in a range of definitions that may have and Daly 1999). Many of the early domesticates been suitable at the time they were conceived but

2 hbe and the transition to agriculture GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 3 pinnacle Quark11:JOBS:BOOKS:REPRO:

now add to the confusion; (3) rapidly expanding product of artificial selection by humans, and empirical knowledge and the characterization of whose reproduction and subsistence are man- local developmental sequences results in special- aged directly by people. For plants, such manage- ized language that does not transfer well to other ment almost always involves an investment in regions where similar transformations occurred; seed selection; clearing, systematic soil tillage, (4) agricultural origins are an inherently evolu- terracing to prepare fields, crop maintenance, tionary question and, as in any system of descent weeding, fertilization, and other crop mainte- with modification, categorical or taxonomic dis- nance; and, development of infrastructure and tinctions have fuzzy and, for different cases, facilities from irrigation canals to processing unevenly and perhaps differently demarcated facilities and storage bins. Parallel efforts are en- boundaries; and, (5) food production and agricul- tailed in animal husbandry. Even societies prac- ture have an impact on multiple features of ticing the most intensive forms of agriculture human societies—e.g., economic, political, so- may engage in incidental hunting and gathering cial, and ideological, any one of which might be of wild foods, depending upon their availability featured in definitions. or desirability (e.g., deer, blackberries). Dense Like earlier attempts, our definitions reflect populations and centralized state-level societies limitations of our knowledge and approach. like our own depend upon increasingly complex Hunting and gathering entails obtaining daily systems of agriculture (Boserup 1965; Zeder sustenance through the collection or pursuit of 1991) involving modification to soil texture, wild foods; wild foods in turn being species structure and fertility (Harris 1989) and some- whose reproduction and subsistence are not di- times resulting in severe environmental degra- rectly managed by humans. Data from around dation, one of the great challenges of our day the world indicate that prior to approximately (Stockstad and Vogel 2003). thirteen thousand years ago, all people known Our definition of agriculture emphasizes archaeologically relied upon hunting and gath- domesticated plants and animals. Domesticates ering wild foods. Hunting and gathering popu- are new plant or animal varieties or species lations expanded into a broad range of habitats created from existing wild species through inci- during the Terminal Pleistocene and Early dental or active selection by humans (Smith Holocene when foraging strategies diversified 1998). Typically selection leads to biological char- (Stiner 2001), in part due to the extinction of acteristics that are advantageous to humans; previously targeted, large-game species, but larger seeds, thinner seed coats, greater docility, also because of the broad array of resource al- smaller size animals. Because humans intervene ternatives afforded by warmer Holocene cli- in the natural lifecycle of these plants and ani- mates (Richerson et al. 2001). Hunting and mals, many domesticates loose their ability to gathering societies have persisted in various survive without human management. This out- parts of the world (Lee and Daly 1999), but come is not surprising since it is well known that starting after about 13,000 BP (before the pres- foragers alter the landscape that they inhabit ent) most foragers evolved into or were sub- by burning, transferring plants and animals be- sumed or replaced by groups practicing mixed tween habitats, and occasionally interjecting foraging and cultivation strategies and, ulti- themselves into other species’ lifecycles (Hastorf mately, agriculture (Diamond and Bellwood 1999; Smith 1998). 2003). Some plant species were better suited to do- On the other end of a mixed spectrum of mestication than others due to their ability to do subsistence strategies is agriculture. We define well in the artificial environments created by hu- agriculture as the near total reliance upon do- mans (Smith 1998). In some instances, the bio- mesticated plants or animals; domesticates be- logical changes may have begun incidentally as a ing varieties or species whose phenotype is a co-evolutionary by-product of human exploitation

hbe and the transition to agriculture 3 GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 4 pinnacle Quark11:JOBS:BOOKS:REPRO:

(Rindos 1984). In other cases domestication may mixed foraging and cultivation/husbandry strate- have occurred under conditions of repeated culti- gies were common and often stable, in the sense vation and harvest (Harlan 1992c; Harris 1989; that they were practiced by people for thousands Ford 1985; Piperno and Pearsall 1998). Cultiva- of years before they developed a full commitment tion is the tending of plants, wild or domesti- and reliance upon agriculture (Smith 2001a). cated; husbandry is the parallel term for animal species. Use of the term cultivation specifically RESEARCH TRADITIONS acknowledges the possibility that humans IN AGRICULTURAL ORIGINS tended wild plants for significant time periods before we would classify them as domesticates Speculation about the origins of food produc- based on observable genetic alterations (Keeley tion is probably as old as the first encounter be- 1995; Piperno and Pearsall 1998). We reserve the tween peoples who recognized that they differed term cultigen for domesticated plants under these appreciably in their dependence upon domes- same conditions. ticated plants or animals. Longstanding tradi- A variety of stable subsistence economies, tions in western thought have seen foragers as extant, historic, and prehistoric, draw upon scarcely removed from animal nature, thus, as elements of hunter-gatherer and agricultural societies, simple and primitive, living without modes of production. These are difficult to char- the many accoutrements and means of control acterize in existing terminologies except as over nature that we associate with agriculture “mixed” economies, engaged in what Smith and industrial culture (Darwin 1874, 643; Powell (2001a) has characterized as low-level food produc- 1885). Agriculture as an advance was instantly tion. They typically depend significantly on hunt- understandable. Hobbes’s famous sentiment ing and gathering while to varying degrees using that hunting and gathering was a life “solitary, cultigens or keeping domesticated animals. Hor- poor, nasty, brutish, and short” (Hobbes 1952, ticulture, the small-scale planting of domesti- 85) is widely cited, but his views were generally cated species in house gardens or the use of shared in the nineteenth century, for instance swidden plots, combined with routine hunting by the novelist Charles Dickens (Dickens 1853). and gathering of wild foods for a significant part We today dismiss this kind of progressive of the diet, would be considered a form of low- evolutionism as simple-minded ethnocentrism. level food production. Contemporary casual Foragers may not be the “original affluent so- farming by the Mikea hunter-gatherers of Mada- ciety” claimed by Sahlins (1972; Hawkes and gascar would be an example of this practice O’Connell 1981), but most foraging societies (Tucker 2001; Chapter 2, this volume). elude the generalizations implied in each of The boundary between low-level food produc- Hobbes’s five famous adjectives. We cannot so tion systems and agriculture is inherently fuzzy. easily dismiss questioning just what distin- We believe the term agriculture is merited when guishes foragers from food producers and how foraging recedes to an episodic, infrequent or humans evolve, in either direction, from one to recreational activity, regular provisioning using the other of these subsistence forms or maintain domesticates takes over daily subsistence, while a mixture of the two for long periods of time. agricultural work and animal husbandry come to European scholarly tradition, informed by dominate the activity schedules of adults. Al- increasingly reliable ethnography and archaeol- though numeric boundaries are somewhat arbi- ogy, has a long engagement with agricultural trary and unsatisfactory, agriculture implies that origins (see Gebauer and Price 1992b; Redding approximately 75% of foodstuffs are acquired 1988; Smith 1998). We highlight three of the from domesticated sources. Although few con- most popular forces employed by archaeologists temporary societies engage in low-level food pro- to explain the origins of agriculture: demo- duction, the archaeological record suggests that graphic pressure, environmental change, and

4 hbe and the transition to agriculture GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 5 pinnacle Quark11:JOBS:BOOKS:REPRO:

socioeconomic competition. Demographic pres- sustained without significant amounts of envi- sure and environmental change are exogenous ronmental degradation and pressure for change forces and socioeconomic competition is endoge- (Glassow 1978). There have been attempts to bet- nous. None in and of itself satisfactorily explains ter contextualize demographic change by meld- the origins of agriculture; each was probably an ing it with ecological models, usually in relation important element of the process, whatever the to variations in climate (Bar-Yosef and Meadow strength of its causal role. One of the virtues of 1995; Binford 1971; Flannery 1971; Flannery HBE is its ability to integrate multiple variables 1973; Hassan 1977; Henry 1989; Hesse 1982b). like these, with an emphasis on behavioral re- These models sometimes lack specificity about sponses to changing socio-ecological conditions. the form or degree of demographic pressure required to provoke subsistence change, and DEMOGRAPHIC PRESSURE they seldom explain why hunter-gatherer popula- Population-resource imbalance caused by de- tions grew more rapidly and stimulated domesti- mographic pressure is one of several univariate cation and agricultural development in certain explanations for the origins of agriculture parts of the world and not others (Keeley 1995). (Cohen 1977; Smith and Young 1972; Smith One response to the early overemphasis on and Young 1983). In the best known formulation demography has been to heavily discount its of this idea, Mark Cohen (Cohen 1977) argued importance in the process of domestication that worldwide population growth explained and agricultural development (Hayden 1990; why hunter-gatherers living in different loca- Hayden 1995a). This is unfortunate because for- tions independently turned to agriculture at the agers clearly have dynamic relationships with end of the Pleistocene. The argument was based their living resources and this in turn has popu- on the premise that the adoption of agriculture lation level effects (Winterhalder and Goland resulted in a net increase in workload and a 1993; Winterhalder et al. 1988). Even small decrease in food diversity and sufficiency, and hunter-gatherer populations alter the distribu- therefore an overall reduction in the quality of tion and availability of harvested plant and ani- life, a situation that any rationally minded hunter- mal species (Stiner et al. 2000). Sometimes this gatherer would not enter into freely. Cohen ar- results in decreased availability or resource gued that as hunter-gatherers exceeded envi- depression; in other instances, it may result in ronmental carrying capacity, food shortages increased resource abundance. The effects that pushed them to experiment with plants and hunter-gatherers have on the density, distribu- animals and, ultimately, with agriculture. Hunter- tion, and productivity of resources is well docu- gatherers over-filled salubrious habitats world- mented in California (e.g., Kumayeey; Shipek wide and were compelled to augment their sub- 1989) and Australia (Gidjingali; Jones and Mee- sistence with food production. han 1989). Environmental change independent Critics of this position were quick to point of humans is ubiquitous and can also affect the out that the archaeological record does not sup- distribution and availability of important species. port the idea that environments worldwide Economic decisions by prehistoric foragers to were saturated with hunter-gatherer popula- experiment with and ultimately manage certain tions on the eve of agricultural development species of plants and animals occurred within (Bronson 1977; Reed 1977; Rindos 1984). Even this dynamic context of demographic change localized populations in the primary centers of and varying plant and animal densities. early domestication appear to be relatively small (Piperno and Pearsall 1998). Others have em- ENVIRONMENTAL CHANGE phasized the difficulties of measuring population V. Gordon Childe was one of the first, and cer- levels in the archaeological record or determin- tainly the most notable, archaeologists to explic- ing the overall population levels that could be itly hypothesize that changes in climate at the

hbe and the transition to agriculture 5 GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 6 pinnacle Quark11:JOBS:BOOKS:REPRO:

end of the Pleistocene stimulated the transition Hostetler and Mix 1999; Kennett and Kennett to agriculture (e.g., his Oasis or Propinquity 2000; Rittenour et al. 2000; Whitlock 1992; Theory; Childe 1928; Childe 1951). According to Woodhouse and Overpeck 1998). Childe, agriculture developed rapidly, hence the These records show that the domestication term Neolithic Revolution, and thus was syn- of key cultigens in the Old and New Worlds oc- chronous with the onset of dry conditions that curred during an interval marked by significant climate records were suggesting in the Near fluctuations in global climate (13,000–8,000 East at the end of the Pleistocene. To survive, BP; Richerson et al. 2001; Piperno and Pearsall humans and potential domesticates concen- 1998). Environmental change at the end of the trated together in well-watered locations like Pleistocene was most pronounced at higher lat- oases and river valleys, where their close inter- itudes as ambient air temperature increased, actions naturally led to domestication and ulti- glaciers receded, sea-levels rose, and forests mately agriculture. The discovery of sickle replaced periglacial tundra (Roberts 1998). Dra- blades and grinding stones in the Carmel Caves matic fluctuations in high latitude environmen- of coastal Palestine suggested that hunter- tal conditions parallel substantial changes in gatherers collected wild cereals during the temperature and rainfall regimes at lower lati- Natufian Period (13,000 BP–10,000 BP), evi- tudes (Henry 1989). These changes instigated dence used by Childe in support of this idea regional biotic shifts in resource abundance (Henry 1989, 6). Although propinquity is overly and density. Some regions witnessed the extinc- simplistic (Redding 1988), subsequent paleoen- tion of several large animals, a likely product of vironmental and archaeological work suggests environmental change and intensified human that regionally specific climatic and biotic predation at the end of the Pleistocene (Lister changes did occur at the end of the Pleistocene. and Sher 1995; Piperno and Pearsall 1998; cf. These surely played a role in shaping spatially Grayson and Meltzer 2003). Others experienced local cultural developments, including the do- the expansion of wild plant species that were mestication of plants and animals and ulti- intensively harvested by foragers and, through mately the adoption of agricultural practices selective manipulation, became important culti- (Henry 1989; Wright, Jr. 1968; Wright, Jr. gens (e.g., barley and emmer wheat; Henry 1993). 1989, 32). It is under these dynamically chang- Unfortunately, the overly deterministic na- ing environmental conditions that foragers ture of the Oasis Theory also provoked a back- altered their subsistence regimes and made lash in the broader archaeological community dietary choices that led to plant and animal against the importance of changing environ- domestication, low-level food production, and mental conditions during the Late Pleistocene ultimately agriculture. and Early Holocene (e.g., Braidwood and Howe 1960; Wagner 1977). For many years the role of SOCIOECONOMIC COMPETITION climate change was simply ignored or deem- Endogenous social change, particularly the de- phasized relative to other mechanisms per- velopment of prestige economies via socioeco- ceived to have greater explanatory value. With nomic competition, has recently become a pop- several noteworthy exceptions (Harris 1996a; ular explanation for the transition to agriculture McCorriston and Hole 1991; Piperno and (Bender 1978; Blake et al. 1992a; Hayden 1990; Pearsall 1998; Watson 1995; Wright, Jr. 1993), Hayden 1995a; Price 1995b; Smalley and Blake this continues today, even with the development 2003). The mechanism for change in these of sophisticated paleoenvironmental tech- models is status-seeking individuals, usually niques (e.g., Piperno 1998) and the aggressive men, who encouraged and controlled the advance of earth system science and high reso- growth of potential domesticates to create sur- lution climate records (Hodell et al. 1995; pluses for social purposes such as competitive

6 hbe and the transition to agriculture GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 7 pinnacle Quark11:JOBS:BOOKS:REPRO:

feasting, alliance formation, and extortion, before conditions promoted socioeconomic rather than as primary sources of food. Hayden competition, at least in Asia, Africa, and the (1995a, 289) has been the most outspoken advo- Americas (Piperno and Pearsall 1998, 14; Smith cate of this idea as a general explanation for the 1998, 209). It appears that many domesticates transition to agriculture worldwide—from the in Mesoamerica, the Near East, and eastern earliest plant and animal domestication through North America were used by hunter-gatherers the development of more intensive forms of at a low level for thousands of years prior to food production. their intensified use (Smith 2001a, 19). This Hayden’s model is based on five testable hy- hints that socioeconomic competition is more potheses (Hayden 1990; see Keeley 1995: 244): likely to be significant in the later stages of the (1) domestication and agriculture will emerge in transition. resource-rich, not resource-poor, zones; (2) it The social significance of food is patent. will first develop in ranked societies that have That some plant species might initially have marked status inequalities; (3) individuals within been grown to brew beer is intriguing; the these societies will hold competitive feasts; (4) social aspects of drinking intoxicating liquids the first plants and animals domesticated will are difficult to refute (Blake et al. 1992a; Hayden be intoxicants, delicacies, or prestige goods 1990; Smalley and Blake 2003). However, rather than bulk or mundane food items; and plants used to brew intoxicating liquids can also (5) evidence for resource stress and malnutri- serve as valuable food items whether they are tion caused by population pressure or climate fermented or not. This means that multiple cur- change will be absent. In archaeological terms rencies must be considered when resource Hayden’s scenario implies correlation between value is assessed by archaeologists. The ability plant and animal domestication and agricultural to store surplus food must also be analyzed for development, and the emergence of socioeco- its social significance. Individuals who success- nomic complexity, marked archaeologically by a fully grow, store, and defend food items can use high degree of sedentism (typically large sites these stores to their social advantage, gaining with substantial architecture), at least two-tiered prestige and influence. Use of surplus food to settlement hierarchies, intensified production improve social advantage, at least under certain agriculturally or otherwise, storage, specialized environmental and demographic conditions, production of prestige items or status markers, should be examined by scholars employing intensified exchange, acquisition of exotic items HBE models. by elites, and differential distribution of pres- tige items in households and burials. HBE RESEARCH ON AGRICULTURAL There are several fundamental flaws with ORIGINS the socioeconomic competition model; there are also some intriguing and potentially impor- There is a small HBE literature on agricultural tant insights. As a stand-alone model for agri- origins. Keegan (1986, 92) made an early and cultural origins, socioeconomic competition prescient argument that foraging models could fails on two levels. First, it lacks a unifying be extended to the study of horticultural pro- explanation for why agriculture developed in duction. He highlighted horticulture because it several independent regions at approximately represents a mixed subsistence system, transi- the same time—other than suggesting it was a tional between the economies of hunter-gather- historical accident (Piperno and Pearsall 1998, ers and agriculturalists. Using data from the 14). Second, although there is evidence that agri- Machiguenga of Peru, Keegan argued that the culture often developed in resource-rich habitats key variables of the diet breadth and patch-use (Price and Gebauer 1995b, 8), the initial domesti- models have direct analogs in food production, cation of most plants and animals occurs well facilitating the use of these cost-benefit models

hbe and the transition to agriculture 7 GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 8 pinnacle Quark11:JOBS:BOOKS:REPRO:

in analysis of this system and the evolutionary declines in population growth rates” (Hawkes transitions that gave rise to it. His calculations and O’Connell 1992, 64). They also draw atten- showed that the Machiguenga generally were tion to HBE arguments for a gendered division stocking their gardens with optimal combina- of labor (Hawkes 1991) that might have been tions of cultigens and, with allowance for sea- important in the evolutionary processes under- sonal and nutritional constraints, making effi- lying subsistence transitions. cient trade-offs among fishing, forest hunting, In a second follow-up paper, Winterhalder and gardening. and Goland (1993) addressed the population In a 1991 paper, Layton et al. (1991) described growth prediction by Hawkes and O’Connell, a “complete break” from the standard, evolution- cited just above. They used a dynamic, popula- ary progression theories of agricultural origins. tion ecology variant of the diet breadth model to They proposed instead an approach that sees show that declining foraging efficiency associ- hunting, gathering, herding, and cultivation as ated with expanding diet breadth may result in alternative strategies of subsistence that may be a decrease or an increase in forager population taken up alone or in various, stable combina- density. The deciding factors are the density and tions, depending on socio-ecological circum- reproductive potential—together, the sustain- stances, and without any implication of irre- able yield—of the low-ranking resources that versible directionality to transitions among happen to come into the diet. them. For instance, there is nothing to prevent Subsequently, Winterhalder and Goland food producers from evolving into foragers. (1997) expanded on these arguments for using Various conceptual elements from foraging a HBE form of analysis in agricultural origins theory, such as the ranking of resources by pur- research. They cited three advantages that dis- suit and handling costs, cost-benefit analysis of tinguish HBE from other research traditions: subsistence trade-offs, boundary defense, and (1) it engages selectionist explanations (Smith risk minimization are found throughout their and Winterhalder 1992b) that are more power- argument. In support of their interpretation ful than the more commonly used functionalist they summarized numerous ethnographic ones; (2) it has tools for non-normative analysis cases in which these strategies are mixed in of unpredictable variation in environmental fea- shifting and sometimes stable balances, remi- tures and the risk-minimizing adaptive tactics niscent of Smith’s (2001a) concept of low-level they elicit; and (3) it focuses on localized and food production. immediate resource decisions and their conse- Layton et al. stimulated two follow-up quences for people “on the ground.” HBE thus papers, both of them making more explicit use engages the behaviors most likely to be causal to of foraging theory to critique or amend specific evolutionary change: “The changes we summa- predictions from their article. Hawkes and O’- rize under broad concepts such as domestication Connell (1992; cf. Layton and Foley 1992) used and the Neolithic revolution have their origin and a sharper distinction between search, and pur- form in the ecologically situated choices and ac- suit, and handling times—the central concep- tions of individuals” (Winterhalder and Goland tual distinction of the diet breadth model—to 1997, 126; italic in original). Winterhalder and argue that high-ranking resources will not drop Goland used the diet breadth model to show out of a forager’s diet in response to exploitation how foragers might initially come to exploit the and depletion. However rare, they will always be organisms that became domesticates, and to pursued when encountered. Hawkes et al. ex- speculate on the adaptive consequences of this pand discussion of the circumstances likely to co-evolutionary engagement. Among the ef- promote subsistence innovation, and argue fects examined were the consequences for re- that “increases in diet breadth result from source depletion, human population density, reduced foraging return rates and so lead to and risk management tactics, using evidence

8 hbe and the transition to agriculture GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 9 pinnacle Quark11:JOBS:BOOKS:REPRO:

from eastern North America to exemplify their incorporated into specialized agricultural pro- arguments. duction systems, in seasonal, low elevation Working on the prehistoric development of forested habitats of the neotropics. agriculture in eastern North America, Gremillion Piperno and Pearsall focus their analysis on (1996a) used diet-breadth and risk-minimization the climate and vegetation changes occurring at models along with opportunity-cost arguments 11,000 to 10,000 radiocarbon years BP and to generate and evaluate predictions about the their likely effects on Neotropical foragers. The circumstances in which new cultigens will be first inhabitants of the neotropics encountered adopted by groups already practicing some agri- a salubrious, open-grassland foraging environ- culture, and whether they will replace existing ment that persisted for only a short time. At plant resources, as did maize following a signifi- around 10,500 BP, the transition to a wetter cant delay from its first appearance, or become a Holocene climate began to produce a seasonal, supplement, as in the case of peaches. In a sec- deciduous forest cover in the lowland tropics. ond study, Gremillion (1998) analyzed macrobo- Piperno and Pearsall hypothesize that due to tanical data from the Cold Oak rock shelter in this habitat shift, and perhaps also to human eastern Kentucky to show that increased de- exploitation (1998, 181), the abundance of the pendence on cultivation of seed crops around high ranking, “open habitat,” plant and animals 1000 BC was accompanied by greater anthro- species decreased, along with foraging effi- pogenic disturbance of habitats and a shift in ciency. While the new seasonal forests remained mast resources from acorns to hickory nuts. She relatively hospitable to mobile hunter-gatherers developed several HBE hypotheses to address at low population density (1998, 71), the diets of this situation, finding greatest credence for the early Holocene foragers expanded to encom- idea that an increase in the overall abundance of pass a broader array of dry-forest plants, species mast resources led to specialization on the most that previously had been ignored. For instance, profitable species, in this instance hickory, at the comparative studies of the efficiency of harvest- expense of the less highly ranked oak. Alterna- ing tubers suggest they likely were outside of tively, increases in the ranking of profitability of the optimal diet in the late Pleistocene (1998: seed crops such as maygrass, chenopod, and 85), but moved into that diet as a low-ranked but knotweed may have displaced acorns from the critical resource once early Holocene habitats diet due to their high processing costs. In each of became more forested. these applications Gremillion argued that HBE The low-ranking, newly important species is a fertile source of new and archaeologically found in seasonally dry forests were subject to testable hypotheses about the subsistence and human interest and manipulation, either inten- economic changes associated with the origins tional or inadvertent, routed into cultivation and of agriculture. eventually domesticated (1998; 27, 82). Because The most thorough existing application of they were sparsely distributed over the land- HBE to the question of agricultural origins is scape, hence relatively unattractive to human Piperno and Peasall’s (1998) monograph, The foragers, there arose an immediate advantage Origins of Agriculture in the Lowland Neotropics. for those who manipulated through burning or Over half the crop plants domesticated in the harvested species from these habitats so as to Americas are thought to have wild progenitors increase their density and yield of useful energy native to neotropical lowland habitats. Among or materials. them are New World staples such as manioc, Piperno and Pearsall cite three rationales for yams, achira, sweet potato, peanut, gourds, using the diet breadth model in this analysis squashes, beans, and perhaps maize. These (1998, 236): (1) the archaeological evidence plants likely were first used by foragers, who shows that early hunter-gatherer/horticultural cultivated, domesticated, and subsequently residents of the neotropics had an expanding

hbe and the transition to agriculture 9 GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 10 pinnacle Quark11:JOBS:BOOKS:REPRO:

diet breadth followed by increasing subsistence effort of modeling and hypothesis testing commitment to low-ranked species; (2) the pre- within an explicitly selectionist, neo-Darwinian historic changes of concern are evident enough theoretical framework (Smith and Winterhalder that short-term precision in the use of the 1992b; Winterhalder and Smith 1992). In both model isn’t necessary (cf. Smith, this volume); respects, HBE provides tools that complement and finally (3) evidence from ethnographic tests or make other traditions more complete. At the shows that this model and an energy currency very least, HBE provides a theoretically well- are commonly successful in predicting the eco- grounded set of tools to begin exploring the nomic response of foragers to changing envi- transition to agriculture in a variety of environ- ronmental circumstances. They conclude, mental and social contexts. “[b]ehavioral ecology seems to us to be the most For instance, although Hayden (Hayden appropriate way to explain the transition from 1990; Hayden 2001) presents his competitive human foraging to food production” (1998, 16). feasting model as a sufficient social explanation Many of the dozen or so early HBE papers for the origins of agriculture, in effect as an al- on domestication and agricultural origins are ternative to models drawing on materialist or fairly general and conjectural. They ask, without ecological explanations, we would prefer a more too much attention to specific cases or the em- cooperative form of analytic engagement. We pirical record of prehistoric findings on this might assume that social stratification and com- topic, how might the ideas of HBE be used to petitive feasting increase the demand for re- address the question of agricultural origins? By sources and then ask how this source of ecolog- and large, their authors are ethnographers ical change would be represented in terms of whose experience is with extant hunter-gatherer foraging models—those extant, adapted, or de- societies. And, they generally have been written veloped specifically for this purpose—and with by people who already placed themselves within what consequences for predictions about subsis- the research tradition of HBE. By contrast, most tence choices and the co-evolution of humans of the papers in this volume are based on em- and their resources. Taking this a step further, pirical case studies, and they are written largely HBE might help us to identify the socio-ecolog- by archaeologists. Most are authored by individ- ical circumstances and evolutionary processes uals for whom behavioral ecology is a new ana- that combine to generate a competitive social lytic tool. hierarchy like that expressed in feasting (Boone We do not claim that the HBE research tra- 1992). A signal strength of HBE is its ability to dition is a complete replacement for the other carry into hypothesis generation a wide variety approaches that we have identified and briefly of postulated sources of causation—global cli- described. We view it rather as a sometimes mate change to the aggrandizement of domi- complementary and sometimes competing nant individuals. form of explanation. It is complementary in two Nonetheless, to the extent that HBE is suc- respects: (1) HBE takes up issues rarely or never cessful in addressing the question of agricul- addressed in these approaches; search and pur- tural origins, it will raise doubts about or contra- suit trade-offs in the harvest of low-ranking dict elements of other research traditions. In the resource species; risk-sensitive adaptive tactics; process it will help us sort out, appraise and dis- and, (2) it frames these issues in quite a different card faulty elements of these approaches. Thus, manner than other, sometimes older, anthropo- for reasons of parsimony as well as theory, those logical and archaeological research traditions by working in the HBE tradition are skeptical of the focusing on the costs and benefits associated adequacy of explanations couched at the level of with individual-level subsistence decisions in global prime movers such as climate change. localized ecological settings. This framing dif- Likewise we doubt the efficacy of explanations ference is determined largely by the analytical made in terms of universal, directional pressures,

10 hbe and the transition to agriculture GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 11 pinnacle Quark11:JOBS:BOOKS:REPRO:

such as Childe’s postulated trend of increasing generate theoretically robust predictions (Elster energy capture (Childe 1965) or ecosystem ap- 1986). Constrained optimization is an analyti- proaches premised on cybernetic properties cally powerful starting point that does not entail such as homeostasis (Flannery 1968). the belief that behavior is routinely optimal, only that there be a tendency towards optimal forms of behavior. HUMAN BEHAVIORAL ECOLOGY

HBE has been used to analyze hunter-gatherer FUNDAMENTAL CONCEPTS economies with favorable results for over two Behavioral ecology likewise is grounded in the decades. This work is both ethnographic (Hill observation, now well confirmed by non-human and Hurtado 1996; Smith 1991) and archaeo- as well as anthropological studies, that some logical (Bettinger 1991b). Because the basics of fundamental economic concepts transcend their this approach are well-described elsewhere scholarly origins in microeconomic attempts (Smith and Winterhalder 1992a), we offer here to explain the functioning of market-oriented only minimal coverage of assumptions, funda- economies. They are useful for studying adap- mental concepts, analytic tools, and models and tive decision making whether the questions hypotheses, with an emphasis on the models concern the behavior of capitalists and workers, employed by the contributors to this volume or the subsistence choices of hunter-gatherers, and concepts and tools that may be of future use horticulturalists, and agriculturalists, not to say to scholars interested in exploring the problem juncos (Caraco et al. 1980) and bats (Wilkinson of agricultural origins. 1990). At a minimum this list would include marginal valuation, opportunity costs, discount- THE OPTIMIZATION ASSUMPTION ing, and risk sensitivity. Behavioral ecology begins with an optimization marginal value. For most tasks we pur- premise. As a result of natural and cultural evo- sue and things we consume, immediate value lutionary processes, behavior will tend toward changes with quantity, be it duration of the constrained optimization (Foley 1985). This as- activity or the amount of a good obtained or in- sumption makes operational the long-standing gested. The first breakfast sausage is more sat- view of anthropologists that hunter-gatherers isfying than the sixth or seventh; an hour-long tend to be skilled and effective in the food quest bath is a delight, but four hours in the tub (Winterhalder 2001). Efficiency, say in captur- makes even insipid alternatives attractive. This ing food energy, is important even if food is would be trivial except for the additional obser- not in short supply because it affords hunter- vation that the decision to suspend consuming gatherers the time and resources to engage fully something like sausage or doing something like in other essential or fitness-enhancing activities taking a bath is based on its marginal rather (Smith 1979). We state this premise as con- than initial, average or total value. Because of strained optimization because we do not expect marginal valuation we move from doing one behavior to be fully optimal. Temporal lags in thing to another even though the intrinsic qual- adaptation and compromises among conflict- ities of the options themselves may be un- ing adaptive goals impede this outcome. Opti- changing. The formulation of marginal analysis mization likewise must be determined within was fundamental to microeconomics (Rhoads the cognitive capacities, beliefs and goals of the 2002), and the careful reader will find marginal organism under study. We adopt the assump- trade-offs in each of the foraging models we dis- tion of constrained optimization rather than cuss below. “satisficing” because the latter—while it may opportunity costs. The idea of opportu- lead to superficially similar predictions—is an nity costs is closely related: the decision to switch empirical concept and is therefore not able to from one behavior—a kind of consumption; a

hbe and the transition to agriculture 11 GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 12 pinnacle Quark11:JOBS:BOOKS:REPRO:

work activity—to another depends not only on and with certainty. For instance, we would pay its marginal value, but on the return to be less at planting time for a corn crop which gained from the available alternatives. Thus, one might after all fail, than for that same crop at ceases to consume sausage when it becomes harvest time when the yield is certain. We dis- more attractive to sip orange juice; one stops count in this manner when the cost of an activ- bathing when preparing a ceremony is more ity such as planting occurs immediately but the compelling. More to the point of our subject, reward, the harvest, is delayed and, perhaps be- one ceases to forage for mussels when the op- cause of that delay, uncertain. Delay alone can portunity and benefits of doing something else be important because the opportunity to bene- take precedence. In each case we assess the fit, even from a completely assured harvest in current activity, be it consumption or purchase the most extreme case might diminish or pass, against what we might be doing instead. In tech- were the cultivator to die in the meantime. nical terms, the opportunity cost of an activity Delay also offers opportunities for hailstorms, refers to the value of the opportunity that is fore- locust plagues and other unforeseen events to gone or displaced by continuing it. For instance, reduce the value of the reward itself. For both the diet breadth model (see below) sets the deci- reasons, effective behavior will hedge, finding it sion to pursue a particular resource against the economical to discount delayed rewards. Use of opportunity cost of ignoring it in favor of search- this concept is fairly recent in behavioral ecol- ing for a more profitable resource to pursue. ogy theory (Tucker 2001). Because the shift Much of microeconomics is a logical and from hunting and gathering to agriculture rep- mathematical elaboration on the workings of resents a shift from immediate- to delayed- marginal valuation and opportunity costs, as reward activities (the original terms are those of they are manifested in the environment of a Woodburn 1982) this basic concept likely will market economy. Using these ideas, econo- be quite important in economic analyses of the mists ask how a wage earner’s consumption transition from foraging to farming. patterns change in response to an increase in risk-sensitive behavior. Basic (or deter- her income. By contrast, the behavioral ecolo- ministic) behavioral ecology models assume gist analyzes how these two concepts play out as that all environmental variables are constants an organism interacts with a natural environ- and that a forager pursuing an optimal set of ment of physical processes and other organ- resources gets the expected (average) reward at isms in the roles of predators, competitors, food all times. By contrast, risk-sensitive models aim resources, potential mates, and offspring. She to be more realistic by introducing a stochastic asks, how might the resource choices of a for- element to the relevant environmental vari- ager shift as a consequence of a decline in ables. All hunters recognize the large role of the density of a highly valued resource, or an chance in the discovery and successful capture improvement in the technology used to harvest of game. In a risk-sensitive model the acquisi- a particular species? tion rate experienced by the forager is expressed Marginal value and opportunity costs and by a statistical distribution; outcomes can be as- benefits are at the heart of behavioral ecology signed probabilities but the actual rate at any models. The most basic claim of the papers in time is unpredictable. Therefore, the optimiza- this volume is that these same ideas can be tion problem must take into account both the adapted to an understanding of decisions faced long-term average and the inevitable periods of by humans during the evolutionary transition shortfall. Risk-sensitive models do this. They between foraging and agriculture. are generally more realistic and more compli- discounting. Discounting refers to the cated than deterministic models, sometimes situation in which we assign a future reward generate like predictions and, given the heuristic less value than if it were available immediately nature of the modeling effort, may not always

12 hbe and the transition to agriculture GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 13 pinnacle Quark11:JOBS:BOOKS:REPRO:

be the preferred option for analysis (Winter- tion that are taken for granted (more formally, halder 1986). treated with a ceteris paribus assumption; see There is a well-developed literature regarding Boyer 1995), in order to focus analysis on one the risk-sensitive behavior of foragers and food- set of effects. producers, taken separately (Cashdan 1990; The measure we use to assess costs and ben- Halstead and O’Shea 1989; Winterhalder et al. efits is known as the currency. While the cur- 1999), but little has been written about risk- rency might be any feature of a resource that sensitive adaptation during the transition from gives it value, foraging theorists typically as- one of these subsistence systems to the other sume that food energy is the most important (Winterhalder and Goland 1997). attribute. After oxygen and water, mammals re- quire metabolic energy in large amounts on a MODEL FEATURES nearly continuous basis. The omnivorous diet The concepts just reviewed—marginal valua- of most hunter-gatherers makes it likely that tion, opportunity cost, discounting, and risk- meeting one’s need for energy entails meeting sensitive analysis—signal that behavioral ecology the needs for other nutrients. This may be more is an attempt to assess the costs and benefits of problematic with agriculturalists. The kcal cur- alternative courses of action under a range of rency is expressed as an efficiency, the net acqui- environmental conditions. In operational terms, sition rate (NAR) of energy. Where energy is not we accomplish this task with models that have limiting or is less limiting than some other in common four features: an alternative set, factor—e.g., protein—then that can be used as constraints, some form of currency, and a goal. the currency. For instance, we know that some Within a particular model, the range of pos- forms of energy, especially those from large or sible behavioral actions is known as the alterna- dangerous game animals, are more prestigious tive set. For instance, the diet breadth model than others (tubers, for instance; Hawkes and specifies an alternative set of ranked combina- Bliege Bird 2002), suggesting that not all kilo- tions of potential resources (see below). In the calories are equal. Prestige might enter into the marginal value theorem, the alternative set currency in some cases. Behavioral ecologists refers to patch residence times. The alternative generally emphasize secondary currencies like set is the dependent variable in the analysis; a kcals or mating success because they are more particular socioenvironmental factor constitutes tractable than the primary neo-Darwinian the independent variable. The model itself does measure of reproductive fitness (Shennan not specify what might cause the independent 2002, 108–11). variable to take on a certain value, or to change. The final feature of models is the goal. A de- It thus leaves open the opportunity for exploring terministic foraging model likely would have how diverse influences such as habitat or cli- the goal of maximizing energy capture while mate change, seasonal variations in population foraging. A risk-sensitive model would empha- density, over exploitation, competition from an- size the goal of avoiding harmful shortfalls of other predator or pressure to extract a surplus energy. Behavioral ecology models of food might affect a behavior like resource selection. transfers in a social group might stress the evo- The specifics of the organism’s capabilities lutionarily stable equilibrium of distribution and the environmental features that structure tactics. The polygyny threshold model for mat- resource selection opportunities are constraints. ing tactics would emphasize the goal of repro- In the diet breadth model constraints include ductive success. Different goals usually imply things like the size of the forager, the hunting different methods: simple optimization analy- and gathering technology used, and the distribu- sis for energy maximization; stochastic models tion and caloric value of the targeted resources. for risk minimization; game theory for fre- Constraints are all of the elements of the situa- quency dependent behaviors, like intragroup

hbe and the transition to agriculture 13 GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 14 pinnacle Quark11:JOBS:BOOKS:REPRO:

transfers, that result in evolutionarily stable expectation of locating more valuable resources strategies. The optimization assumption ties to- to pursue. If the net return to (b) is greater than gether constraints, currency, goal, and the costs (a), even after allowing for additional search and benefits of the alternative set. For instance, time, then the optimizing forager will elect to given constraints of resource densities and val- pass by the encountered resource, and will con- ues, and their associated costs and benefits, we tinue to do so no matter how frequently this predict that organisms will select the alternative type of resource is encountered. that provides them the highest available net ac- The general solution to this trade-off is de- quisition rate of energy. As noted earlier, even vised as follows: each of k potential resources is when there is no particular shortage of food- ranked in descending order by its net return stuffs, efficient foraging frees time for alterna- rate for the post-encounter work to obtain it. tive activities and lessens exposure to risks as- This represents a resource’s net profitability sociated with foraging. While we don’t expect with respect to pursuit, harvest, and handling the organism always to engage in the optimal costs. The alternative set then is made up of diet behavior, models based on this assumption breadths from 1 to k, in the form db ϭ 1, db ϭ have proven to be robust when compared to 1 ϩ 2, db ϭ 1 ϩ 2 ϩ 3, up to db ϭ 1 ϩ . . . ϩ k). ethnographic and archaeological datasets The derivation of the best-choice diet begins (Broughton 1999; Smith 1991). with the most profitable resource (1), and, step- wise, adds resource types, continuing until the first resource (n ϩ 1) with a profitability less FORAGING MODELS than the overall foraging efficiency of the diet Foraging models typically come with a long list that does not include it (diet breadth ϭ n). of assumptions, awareness of which is critical Resources ranked (n ϩ 1 . . . k) are excluded be- to their successful use. The models are most cause to pursue them would impose an unac- often expressed precisely in mathematical for- ceptable opportunity cost: a lower return rate mulas or graphs (Stephens and Krebs 1986). In for time spent pursuing them relative to the ex- this chapter we provide qualitative and verbal pected benefits from ignoring them in favor of summaries only; explications of greater detail both searching for and pursuing more prof- can be found in individual chapters. We trust itable types. Think of picking up change in tall that the reader wishing to apply the models and grass: if there are enough silver dollars and understand them more thoroughly and criti- quarters the income-minded gleaner will ignore cally will study the references we give for each the dimes, nickles, and pennies, no matter how model. frequently they are encountered. Notice that the DBM also entails a marginal decision: It asks, is DIET BREADTH (RESOURCE SELECTION) the profitability of the next ranked item above or The diet breadth or resource selection model below the marginal value of foraging for all (DBM) is one of the oldest and most commonly resources ranked above it? used (MacArthur and Pianka 1966; Schoener Creative use of this or any foraging model 1974; Winterhalder 1987), particularly by ar- entails thought experiments of the form: how chaeologists (e.g., Broughton 1999; Butler will an optimizing forager respond to a change 2000). It is sometimes called the encounter- in independent variable x. Predicted responses contingent model because it focuses on the are confined to options with the alternative set, decision to pursue or not to pursue, to harvest but the independent variable x might be any or not harvest, a resource once it is encountered. change in the environment or the behavioral The decision entails an immediate opportunity capacities of the forager that affects the primary cost comparison: (a) pursue the encountered model variables: resource encounter rates and resource, or (b) continue searching with the profitability. For instance, resource depression,

14 hbe and the transition to agriculture GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 15 pinnacle Quark11:JOBS:BOOKS:REPRO:

environmental change, and other factors which a dispersing flock of grouse. Patches can be diminish encounter rates with highly ranked re- ranked like resources, by their profitability sources will increase search costs, lower overall upon encounter. As a first approximation, the foraging efficiency, and as a result, may cause same predictions apply. However, predictions the diet breadth of a forager to expand to in- are somewhat less clear for the selection of clude items of lower rank. One or more items patches than for resources, because a definitive that previously ranked below that boundary prediction about patch choice is interdependent may now lie above it, making these resources with a decision about patch residence time, the worth pursing when encountered. The converse focus of the next model. is also true. Sufficiently large increases in the density of highly ranked resources should lead PATCH RESIDENCE TIME (THE MARGINAL to exclusion from the diet of low ranked items. VALUE THEOREM) A seasonal elevation of fat content, or adoption If a resource patch—which we envision as a of a technology that makes its pursuit, harvest small area of relatively homogeneous resource or processing more efficient or any factor that opportunities, separated by some travel dis- raises the profitability of a particular resource tance from other such locales—is harvested at a will elevate its ranking, perhaps enough to diminishing rate of return, it is obvious to ask move into the best-choice diet. It may, in fact, when the forager should abandon its efforts and displace resource items previously consumed. attempt to find a fresh opportunity. By moving Winterhalder and Goland (1997, Fig. 7.4) pro- on, he or she will incur the cost of finding a new vide an extended list of factors that might oper- patch, but upon locating it, will be rewarded ate through encounter rate and pursuit and with a higher rate of return, at least for a while. handling costs to change resource selectivity. The optimizing solution to this foraging deci- The diet breadth model also implies that, sion is given by the marginal value theorem under a given set of conditions, resources (Charnov 1976; Charnov et al. 1976; Stephens within the optimal diet are always pursued and Krebs 1986). The marginal value theorem when encountered; those outside the optimal postulates a decline in return rates for time diet will always be ignored. There are no “partial spent in the patch, usually approximated by a preferences,” such as “take this organism 50% negative exponential curve. The optimizing so- of the time it is encountered.” Likewise, the de- lution specifies that the forager will leave the cision to include a lower-ranked item is not present patch when the rate of return there has based on its abundance, but on the abundances dropped to the average foraging rate. The aver- of resources of higher rank. Think of the small age foraging rate encompasses the full set of change mentioned earlier. patches being harvested and the travel costs as- sociated with movement among them. To stay PATCH CHOICE longer incurs unfavorable opportunity costs be- In the diet breadth model, we envision a re- cause higher returns were available elsewhere. source that is harvested as a unit with a fixed To stay a shorter duration is also sub-optimal, value (e.g., a steenbok). By contrast, a patch is a because rates of return are, on average, higher resource or set of resources which is harvested when compared to the costs of moving on to an- at a diminishing rate, either because it is de- other resource patch. pleted in such a way that makes continued har- In this model, short travel times are associ- vesting more difficult; the densest and ripest ated with short patch residence (take the highest berries are picked first, or because the continu- return opportunities and move on quickly); long ing presence of the forager disperses or in- travel times with longer residence times. The creases the wariness of remaining resource forager optimizing his or her patch residence opportunities as in the second or third shot at time rarely will completely deplete a patch;

hbe and the transition to agriculture 15 GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 16 pinnacle Quark11:JOBS:BOOKS:REPRO:

the resources left behind are significant for the that habitats will be occupied in their rank recovery of the patch. Finally, the value of har- order, that human densities at equilibrium will vested patches, upon departure, is the same. be proportional to the natural quality of their re- The inter-dependence between the two patch- sources, and that the suitability of all occupied related models should now be more apparent. habitats will be the same at equilibrium. Predictions about patch residence time depend In the IFD the creative element resides in on patch choice; reciprocally, predictions about imaging how various socioenvironmental set- patch choice depend on residence time. Use of tings might affect the shape of the curves repre- one of these models must assume the other; senting the impact of settlement density on Stephens and Krebs (1986, 32–34) give a more habitat quality. For instance, it is possible that detailed discussion of this model. settlement at low densities actually increases the suitability of a habitat. Forest clearing by the HABITAT SELECTION (THE IDEAL newcomers leading to secondary growth might FREE DISTRIBUTION) increase the density of game available to them The ideal free distribution is a model of habitat and to emigrants. This is known as the Allee choice (Fretwell 1970; Sutherland 1996). The effect. Likewise, some habitats (e.g., small distinction between patches and habitats is one islands; see Kennett et al., this volume) may be of scale: patches are isolated areas of homoge- quickly affected by settlement, generating a nous resource opportunities on a scale such that sharply declining curve of suitability as popula- a forager may encounter several to several dozen tion densities increase, whereas others may be in a daily foraging expedition. Habitats are simi- much more resilient. If immigrants to a habitat larly defined by their aggregate resource base, successfully defend a territory there, then newly but at a regional scale. As suggested by their arriving individuals will more quickly be dis- greater relative size, habitats also invoke some- placed to lower ranked habitats, a variant what different questions, such as where to es- known as the ideal despotic distribution (IDD; tablish and when to move settlements, and see Sutherland 1996). when to relocate by migration. Generally, we ask how populations will distribute themselves with CENTRAL PLACE FORAGING respect to major landscape features like habitats. Many foragers, human and nonhuman, locate In the ideal free distribution, the quality of a at a dry rock shelter, potable water, or a valuable habitat depends on resource abundance and the or dense food source or other particularly criti- density of the population inhabiting and using cal resource—e.g., an attractive habitation site, it. The model assumes that the initial settlers or perhaps at a location central to a dispersed ar- pick the best habitat, say “A.” Further immigra- ray of required resources—and then forage in a tion and population growth in habitat A reduce radial pattern from that site. Central place for- the availability of resources and the quality of aging models (Orians and Pearson 1979) ad- the habitat drops for everyone. Crowding, de- dress this circumstance. They assume that a pletion of resources, and competition are possi- forager leaving such a home base must travel a ble reasons for this. The marginal quality of certain distance through unproductive habitat habitat A eventually will drop to that of the sec- to reach productive foraging zones. The goal— ond-ranked, but yet unsettled, habitat B. If each optimizing delivery of foodstuffs or other valu- individual in the population seeks the best habi- ables to the central place—must take account of tat opportunity, further growth or immigration the round trip travel costs between the central will be apportioned between habitats A and B place and the foraging site, in addition to the such that their marginal value to residents is standard considerations about resource selec- equalized. Lower ranked habitats will be occu- tion. The basic prediction of this model is the pied in a similar manner. This model predicts following: as travel costs out and back increase

16 hbe and the transition to agriculture GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 17 pinnacle Quark11:JOBS:BOOKS:REPRO:

with a load on the return trip, the forager should pinyon nuts (B), are found in geographically become more and more selective about what is separated habitats. They also change in their harvested. At long travel distances only the relative seasonal importance. We would expect most valuable loads justify the effort. the forager to locate adjacent to the more domi- This model has been adapted in an intrigu- nant of the two food sources (say, A), especially ing way by archaeologists who have used it to if the resource targeted is difficult to transport, address the question of field processing (Bet- and to harvest the less dominant (B) or easier to tinger et al. 1997; Metcalfe and Barlow 1992). transport item through logistic foraging expedi- Field processing entails removing parts of a re- tions. Zeanah’s model specifies in quantitative source with little or no value, in order to carry terms what shifts in yield and transport costs more of the valued portions back to the central will lead to the decision to switch the pattern of place. Shelling marine bivalves or removing settlement and logistic procurement, residing pinyon nuts from their cones are examples. adjacent to B, while harvesting A logistically. With data on parameters such as distance, fea- Although settlement models have not, to our sible bulk and weight of loads, and the costs and knowledge, been applied in studies of domesti- benefits of field processing a particular resource cation and agricultural origins, the likelihood (e.g., Barlow et al. 1993), it is possible to predict that the better foraging and farming sites have rather precisely the travel distance at which the non-overlapping distributions, and the implied forager will process in the field rather than carry changes in mobility and sedentism during a the unprocessed resource back to the central transition from foraging, to a mixed foraging & place. Field processing of course improves the farming, to farming, or back to foraging, offers efficiency of transportation, but it also commits fertile ground for exploration. to processing field time that could have been used to locate, harvest and transport more of the CURRENT DEVELOPMENTS unprocessed resource. This model predicts that IN FORAGING THEORY field processing will become more likely as travel distance increases. A list of established models might give the We cite this adaptation of the central place sense that behavioral ecology, however useful to foraging model in part because it makes the im- interpretation, is a static or completed field. In portant point that foraging theory is not a closed, fact, it is in a rapid state of expansion and devel- off-the-shelf set of tools (Kelly 2000). Rather, it opment both in ethnography and archaeology. must be, and it has considerable potential to In this section we note several of the more im- be, adapted to the particular circumstances of portant developments. The trends described human subsistence, whether foragers, farmers, here also make it evident why the more encom- or populations that mix these sources of pro- passing term, behavioral ecology, often is more duction. apt than foraging theory.

SETTLEMENT (RE)LOCATION BEYOND KCALS Settlement models attempt to predict when for- Early applications of foraging models, espe- agers will relocate their central places, due to cially the diet breadth model, adopted a straight- localized depletion of resources (Kelly 1992), forward energy currency to measure the costs seasonal or other shifts in the relative values and benefits of options in the alternative set. and availability of local and distant resource op- The value of a moose was the weight of its edi- portunities (Zeanah 2000). Zeanah’s model, ble tissue represented as kcals. This is consis- for instance, imagines a foraging group whose tent with a prime methodological predilection two most important resources, say lake margin of behavioral ecologists (Winterhalder 2002a): lacustrine species (A) and mountain sheep or begin simply. Once you understand how the

hbe and the transition to agriculture 17 GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 18 pinnacle Quark11:JOBS:BOOKS:REPRO:

simple model works and have appraised its rel- narrow question of resource selection to broader evance to the empirical problem, it is appropri- anthropological issues—the roles of gender, ate to relax restrictive and sometimes unrealistic prestige, and power in structuring economic assumptions. Thus, in foraging theory, studies activity (e.g., Broughton and Bayham 2003; of resource selection led naturally to examina- Elston and Zeanah 2002; Hildebrandt and tion of intra-group resource transfers. This McGuire 2002; Hildebrandt and McGuire move from issues of economic production to 2003). those of distribution drew attention to a differ- ent metric: marginal value. After a filling meal BEYOND DETERMINISTIC APPROACHES or two, the marginal value of the balance of a Risk-sensitive and discounting models are an- moose carcass to the forager who obtained it other set of variations on early foraging theory may drop rapidly relative to the kcals it repre- efforts. In the original models for diet breadth, sents. This observation—that medium to large patch choice, and patch residence time, all in- food packets are subject to marginal valuation— put values were taken to be averages unaffected is at the heart of behavioral ecology models of by stochastic variation. Thus the average search food transfers through tolerated theft and reci- time to locate the next resource was treated as a procity-based sharing (Blurton Jones 1987; constant, making foraging a more predictable Gurven 2004; Winterhalder 1996; Winter- enterprise than is the case. These models fo- halder 1997). Of equal importance, there may cused on a goal of maximizing acquisition rate be some cases in which the marginal value of a during foraging. Risk-sensitive models allow resource is the appropriate valuation of its prof- for stochastic variation in the factors influenc- itability for purposes of the original diet breadth ing foraging decisions, such as encounter rate model. or pursuit time. They assume that the forager A more radical variation on currency is evi- has the goal of risk minimization (Winterhalder dent in models devised to help explain an et al. 1999). For instance, Stephens and Charnov anomaly in early field studies of foraging behav- (1982) modified the marginal value theorem to ior: although each sex often could do better by show that a risk-minimizing forager, in positive harvesting the same set of resources, men some- energy balance and facing a normal distribution times specialize on large game and women on of unpredictable inter-patch travel times, would plants and small animals (Hill et al. 1987), each stay somewhat longer in a patch than a rate at a cost to their potential foraging efficiency. maximizing forager whose travel times were a The show-off (Hawkes and Bliege Bird 2002) constant. and costly signaling (Smith et al. 2003) models In general, risk-sensitive models predict that assume that resource values—and hence their optimizing foragers who are not meeting their patterns of acquisition and distribution—will average requirements will be risk prone. They sometimes be predicated on the prestige associ- will elect the higher variance options from the ated with their use or on the information their alternative set because those offer their greatest capture conveys about the prowess of the chance of a survival-enhancing windfall. For- hunter. With these models foraging theory has agers in positive energy balance will be risk carried us beyond “the gastric” (Zeanah and averse, electing the low variance options that Simms 1999) and into the realm of social the- minimize the chance of a threatening shortfall. ory (Bliege Bird and Smith 2005), making plau- The implications of these generalities for spe- sible our earlier claim that foraging theory of- cific types of decisions must be worked out in- fers broad grounds for complementing other dividually. research traditions in the field of agricultural A variation on risk-sensitive models is dis- origins (e.g. Hayden 1995a). Social valuation counting (Benson and Stephens 1996). If the moves the modeling effort of HBE from the forager has reason to discount, and faces a

18 hbe and the transition to agriculture GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 19 pinnacle Quark11:JOBS:BOOKS:REPRO:

choice between a small reward at present or a among agents rather than relationships among larger one at some point in the future, he or she variables” (Kohler 2000, 2). Agent-based mod- may do best by taking the less valuable but im- els have the added advantage that they can mediate option. Tucker (Ch. 2, below; see also incorporate basic processes of learning or evo- Alvard and Kuznar 2001) argues that discount- lution, for instance by allowing the agent to ing is likely to be especially important in the adjust its behavior according to its monitoring transition from foraging to food production. of performance criteria. Because of this prop- erty, they are thereby especially useful for simu- BEYOND DERIVED AND GRAPHICAL lating adaptive or co-evolutionary processes (see SOLUTIONS examples in Brantingham 2003; Kohler and The basic foraging models described above are Gumerman 2000). Although there are at pres- products of mathematical derivation, often rep- ent no agent-based models of domestication or resented graphically. A desire for more realistic agricultural origins, behavioral ecology adapta- variants is associated with new analytic method- tions of the agent-based approach appear an ologies, such as simulation and agent-based especially promising avenue for research. modeling. For instance, Winterhalder and stu- dents (Winterhalder et al. 1988) simulated the BEYOND ETHNOGRAPHY population ecology of a foraging population in- The specific claim of this volume—that behav- teracting with multiple resource species. In this ioral ecology theory is an essential tool in the dynamic model, the human population grows analysis of the transition from hunting-and- or contracts in density as a function of foraging gathering to agriculture—is set within a broader efficiency. It harvests species identified by the assertion: that behavioral ecology can be used to diet breadth model, in amounts required to understand prehistory in general (Bird and O’- meet its food needs. And, to complete the Connell 2003; O’Connell 1995). Although ar- dynamic circuit, the densities of the resource chaeologists have been enthusiastic consumers species themselves expand or contract accord- and occasionally developers of foraging theory, ing to their degree of exploitation and their the models themselves and the bulk of their logistic potential to recover from being har- testing are the province of biologists and an- vested. The result is a more realistic application thropologists working with living species and of the diet breadth model: exploitation actually peoples. As a consequence, the models typically changes prey densities and thus encounter make predictions at the level of individual be- rates in a plausible manner, generating new hy- havior over very short time scales—minutes to potheses relevant both to agricultural origins days or perhaps weeks. In contrast, archaeolog- (Winterhalder et al. 1988) and conservation bi- ical data on subsistence production, food dis- ology (Winterhalder and Lu 1997). tribution, mobility, settlement, and the other Agent-based modeling is another new tech- topics of behavioral ecology represents the ag- nique of great promise. Agent-based analyses gregate consequences of many individual ac- rely on computer simulations to represent a tions over decades, centuries or longer. Much population of agents interacting with an envi- archaeological data conflate individual, tempo- ronment and among themselves. These models ral and perhaps spatial variability. This disparity iterate a cycle in which the agent collects infor- of scale and resolution raises thorny problems mation from the environment, and then acts regarding how HBE models are to be verified, in some fashion that changes the agent and en- applied and interpreted in archaeological con- vironment. The agent-based approach, “em- texts (e.g., Smith, this volume). How do we get phasizes dynamics rather than equilibria, dis- from a chronological sequence of faunal sam- tributed processes rather than systems-level ples, each of which represents perhaps dozens phenomena, and patterns of relationships of foraging expeditions by different individuals

hbe and the transition to agriculture 19 GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 20 pinnacle Quark11:JOBS:BOOKS:REPRO:

over decades or centuries of time, to the season- tactics, and life history evolution, while—in the ally and habitat specific foraging choices of a anthropological case—simultaneously moving particular hunter? into the analyses of societies engaged in other This problem is serious but may not be as modes of production (reviews in Borgerhoff daunting as appears on first consideration. For Mulder 1991; Cronk 1991; Smith 1992a; Smith instance, careful investigation does occasionally 1992b; Smith and Winterhalder 1992a; Winter- reveal the individual and momentary in prehis- halder and Smith 2000). The impetus for this tory. Enloe and Davis (1992) and Waguespack expansion has at least four sources: (1) the early (2002) both have shown that by analyzing the empirical success of field studies using the ap- “refitting” of bones scattered among the differ- proach; (2) the generality of the neo-Darwinian ent hearths of a campsite it is possible to reli- theory that inspired it; (3) the generality of the ably infer patterns of prehistoric food sharing. underlying concepts of marginal valuation, In broader terms, Grayson and Delpech (1998; opportunity cost appraisal, risk-sensitivity, dis- Grayson et al. 2001), Lyman (2003), Broughton counting; and, (4) the flexibility of individual (2002) and Gremillion (2002) are exploring models, which often have been readily adapted how well and under what circumstances various to problems or settings not foreseen by their archaeological measures of floral and faunal original authors. The present volume continues residues are able to capture foraging behavior this trend by carrying behavioral ecology theory changes in diet breadth. A series of reports ana- and models into analyses of domestication and lyzing broad spectrum type diet breadth agricultural origins. changes in late prehistory have made creative The transfer and extension of ideas and con- use of changing ratios of large, presumably, cepts in order to bring new topics under the highly ranked, to small prey (Broughton 1999; compass of existing theory has obvious scien- Broughton 2001; Butler 2000; Lindström tific merit (Kuhn 1977; McMullin 1983). It also 1996; Nagaoka 2001, 2002) in order to docu- has pitfalls. The failings of early “evolutionist” ment declining foraging efficiencies and ex- models of social evolution and their archaeolog- panding diets. Through a combination of ical adaptations, as well as social Darwinist in- archaeological investigation and population ecol- terpretations, are well-rehearsed subjects in an- ogy simulation, Stiner and colleagues (Stiner et thropology. Contemporary anxieties about the al. 2000; Stiner 2001; Stiner and Munro 2002) use of neo-Darwinian theory in anthropology have shown that small prey may be especially are more narrowly and analytically focused, and sensitive indicators of human resource selection sometimes not so easy to set aside. A recent ex- and the impacts of exploitation. We expect these ample would be debate over the claim by Rindos efforts to find archaeologically viable means of (1984) that his co-evolutionary account of plant using foraging theory to continue. domestication had successfully banished hu- man intent from an explanatory role in this BEYOND HUNTER-GATHERERS process (Rindos 1985). AND FORAGING In the present volume we take for granted The research tradition represented in this vol- the relevance to agricultural origins of neo- ume originated as foraging theory focused on Darwinian and behavioral ecology theory. We the study of food production in hunter-gatherer reject without explicit argument the substan- populations (Winterhalder and Smith 1981). In tivist claim of economic anthropology that none both biology and anthropology the approach of the tools of formalist, microeconomics has since has adopted the broader name—human any purchase outside of modern capitalist behavioral ecology—as it has expanded its topical economies (e.g., Sahlins 1972). To the contrary, focus to encompass resource distribution, group we believe it evident that the basic concepts size and structure, mating and reproductive of HBE (see above) are fundamental to the

20 hbe and the transition to agriculture GRBQ084-2272G-C01[01-21]. qxd 11/30/05 7:43 PM Page 21 pinnacle Quark11:JOBS:BOOKS:REPRO:

analysis of any economy. Close attention to their We envision three levels where HBE might use in HBE we believe will stimulate new appli- be applied to the question of agricultural ori- cations and models specifically designed to ana- gins. (1) Extant models, although designed for lyze mixed economies and food production. the analysis of foraging, might be applied in the We are more receptive to the argument that analysis of agricultural origins with little or no specific foraging models, developed as they alteration in their structure and assumptions. were for foragers, may be only partially appro- This is the procedure of most authors in this priate to the analysis of emergent food produc- volume. (2) Extant models might be modified so ers. For instance, the diet breadth model as- to more directly address questions or situations sumes random encounter with resources, a specific to non-foraging aspects of economy, condition increasingly likely to be violated as including cultivation and agricultural produc- foragers become involved in the manipulation tion. The modification of central place foraging of individual species. In as much as all models models to analyze the question of field processing simplify reality and thus violate at least some is an especially good example of this. (3) Finally, conditions of their application, the unavoidable entirely new models, inspired directly by the judgment is this: does the failure to fit this par- problem of explaining human subsistence tran- ticular assumption completely vitiate the heuris- sitions, might be devised using fundamental tic or analytical value of the model? With the behavioral ecology concepts such as opportu- specific cautions cited in individual papers, we nity cost or discounting. We think of these op- believe the combined weight of the case studies tions as adopt, adapt, or invent, respectively. developed in this volume add up to a strong pre- While options (2) and (3) hold great potential for sumption in favor of the utility of foraging the- novel and perhaps quite interesting analyses, it ory, even as the foragers being analyzed direct appears from the papers assembled here that more and more of their effort toward agricultural there is much to be accomplished with the sim- activities. ple adoption of existing models.

hbe and the transition to agriculture 21