The Origins of Agriculture as a Natural Experiment in Cultural Evolution

Peter J. Richerson, Robert Boyd, and Robert L. Bettinger

Do Not Cite In Any Context Without Permission Of Authors

Peter J. Richerson, Department of Environmental Science and Policy, University of California—Davis, Davis, CA 95616,

Robert Boyd, Anthropology Department, University of California—Los Angeles, Los Angeles, CA 90024,

Robert L. Bettinger, Anthropology Department, University of California—Davis, Davis, CA 95616,

  Abstract: Several independent trajectories of subsistence intensification, often leading to agriculture, began during the Holocene. No plant rich intensifications are known from the Pleistocene, even from the late Pleistocene when human populations were otherwise quite sophisticated. Recent data from ice core climate proxies show that last glacial climates were extremely hostile to agriculture—dry, low in atmospheric CO2, and extremely variable on quite short time scales. We argue that agriculture was impossible under last-glacial conditions. The quite abrupt final amelioration of the climate was followed immediately by the beginnings of plant intensive resource use strategies in some areas, although the turn to plants was much later elsewhere. Almost all trajectories of subsistence intensification in the Holocene are progressive and eventually agriculture became the dominant strategy in all but marginal environments. In the Holocene, agriculture is, in the long run, compulsory. We use a mathematical analysis to show that the rate limiting process for intensification trajectories must almost always be the rate of innovation of subsistence technology or subsistence related social organization. At the observed rates of innovation, population growth will always be rapid enough to sustain a high level of population pressure. Several processes appear to retard rates of cultural evolution below the maxima we observe in the most favorable cases.

   Resumen: Varias trayectorias independenties de la intensificación del sustento , muchas de las cuales conducieron a la agricultura, empezaron durante el Holoceno. No conocemos ninguna intensificació que usara recursos vegetales durante el Pleistoceno, inclusive el Pleistoceno último, cuando las poblaciones humanas fueron muy sofisticadas en otros ámbitos. Datos recientes de cilindros de hielo sacados de Groenlandia muestran que la última glaciación fué extremadamente hostil a la agricultura, ya que fué—seca, baja en CO2, y extremadamente variable en el corto plazo. Proponemos que la agricultura fué imposible en estas condiciones de la última glaciación. La súbita mejora del clima al final de la glaciación fué seguida inmediatamente por la iniciación de usos intensivos de los recursos vegetales en algunos lugares, aunque mucho mas tarde en otras partes. Casi todo las trajectores de intensificación en el Holoceno eran occurrieron sin retroceso. Finalmente, la agricultura se convirtió en el modo principal de sustento en todas partes, excepto por zonas muy frías o muy secas. En el Holoceno, la agricultura se vuelve, a final de cuentas, obligatoria. Usamos un análisis matemático para mostrar que los procesos limitante de la taza de intensificación debe ser casi siempre la taza de la inovación tecnológica en las estrategias de sustento, o la taza de inovación en las formas de organazación social en relación al sustento. Con las tazas de inovación que se observan, el crecimiento de la población siempre es suficientemente rapido como para crear alto nivel de presión poblacional. Al parecer , varios processos normalmente retardan la velocidad de la evolución cultural abajo de las tazas máximas que observamos en el modelo.

While observing the barbarous inhabitants of Tierra del Fuego, it struck me that the possession of some property, a fixed abode, and the union of many families under a chief, were the indispensable requisites for civilization. Such habits almost necessitate the cultivation of the ground; and the first steps would probably result from some such accident as the seeds of a fruit tree falling on a heap of refuse, and producing some unusually fine variety. The problem, however, of the first advance of savages toward civilization is at present much too difficult to be solved.

                    Charles Darwin Descent of Man 1874


 Evolutionary thinkers have long been fascinated by the origin of agriculture. While Darwin declined to speculate on agricultural origins, Twentieth Century scholars were bolder. The Soviet agronomist Nikolai Vavilov, American geographer Carl O. Sauer, and British archaeologist V. Gordon Childe wrote influential books and papers on the origins of agriculture in the 1920s and 30s (see MacNeish 1991: 4-19, for a discussion of the intellectual history of the origins of agriculture question). These explorations were necessarily speculative and vague, but they stimulated interest. Vavilov and Sauer argued that agriculture originated in locations where the wild ancestors of later crop species attracted the attention of hunters and gatherers, leading eventually to domestication. Childe argued that climate change accompanying the end of the Pleistocene was responsible for the initial steps that led to domesticated crops.

    Immediately after World War II, the American archaeologist Robert Braidwood (Braidwood, et al. 1983) pioneered the systematic study of agricultural origins. From the known antiquity of village sites in the Near East and from the presence of wild ancestor species of many crops and animal domesticates in the same region, Braidwood inferred that this area was likely a locus of early domestication. He settled on a region in the foothills of the southern Zagros Mountains in Iraqi Kurdistan as a likely place for domestication to have begun. He then embarked on an ambitious program of excavation using a multi-disciplinary team of archaeologists, botanists, zoologists, and earth scientists to extract the maximum useful information from the excavations. The availability of 14C dating gave his team a powerful tool for determining the ages of the sites.

    The Braidwood team’s main effort focussed on the Jarmo site, but excavation at a site nearby uncovered an earlier settlement that had been occupied seasonally by hunter-gatherers who were collecting wild seeds, probably the ancestors of wheat and barley, and hunting the wild ancestors of goats and sheep about 11,000 years ago. (Ages are given here as calendar dates before present (B.P.), where present is taken to be 1950, estimated from 14C dates according to Stuiver, et al.’s (1998) calibration curves.) Near Eastern sites older than about 15,000 B.P. excavated by Braidwood (Braidwood and Howe 1960) and others were occupied by hunter-gatherers who put much more emphasis on hunting and much less on collecting and processing seeds (Henry 1989; Goring-Morris and Belfer-Cohen 1998). At Jarmo itself, the team excavated an early farming village dating to about 9,000 B. P. Using much the same seed processing technology as their immediate gathering predecessors, the Jarmo people no longer moved their residences with the seasons. Analyses of plant and animal remains suggested that the process of domestication was underway. This early agricultural village is at the base of an archaeological record of larger and increasingly sophisticated agrarian settlements that characterize the Near Eastern archaeological record leading to the first state level societies in Mesopotamia about 5,500 BP.

    Since the pioneering Jarmo excavations, a few dozen multi-disciplinary archaeological teams, have studied likely sites of agricultural transitions in the Near East, North and South America, the Far East, and Sub-Saharan Africa. At least six regions (Table 1) were autonomous centers of plant and/or animal domestication, and incomplete, controversial evidence is taken by some scholars to imply at least two more. Ehret (1998) argues on grounds of “linguistic archaeology” that Saharan and Sub-Saharan Africa had three independent centers of domestication rather than the one more commonly accepted. Historical linguists have reconstructed three sets of agricultural vocabulary uncontaminated by loan-words, implying independent, rather early, episodes of domestication. These investigations have discovered no region in which agriculture developed earlier or faster than in the Middle East, though a North Chinese center of domestication of millet may prove almost as early (see discussion below). Indeed other centers seem to have developed later, or more slowly, or with a different sequence of stages, or all three. The spread of agriculture from centers of origin to more remote areas is well documented for Europe and North America. Ethnography also gives us cases where hunters and gathers persisted to recent times in areas seemingly highly suitable for agriculture, most notably much of western North America and Australia. Attempts to account for this rather complex pattern are a major focus of archaeology.

Origin of agriculture is a macroevolutionary “experiment”

    “Macroevolution” is the evolutionary biologist’s term for the large scale and long-term pattern of evolution. The contrasting term, “microevolution,” refers to generation-to-generation changes. Because much important evolutionary change takes place at macroevolutionary time scales, it is important for evolutionary theory to explain such changes. Unfortunately, while small-scale microevolutionary changes are accessible to direct observation and critical experiment, processes that act on long time scales are not. Thus a major project for students of both organic and cultural evolution is to link theory derived from microevolutionary studies to the realm of macroevolutionary events.
    Because evolving systems are normally complex, and data that speak to the past are so sparse, testing evolutionary hypotheses with macroevolutionary data is difficult. The best cases to work with are those where large, simple, changes are imposed upon the system. Large changes ensure that the system’s evolutionary response to the change will not be lost in the welter of unknown causes of change that will ordinarily make past events exceedingly difficult to understand. Simple changes make it easier to follow the inevitably complex chains of cause and effect in history. The classical laboratory experiment, similarly, depends upon making large, simple (controlled) manipulations to disentangle complex causal processes. Subtlety comes only when basic processes are very well understood.
    The several independent origins of agriculture provide just such a case. They all occurred relatively recently and began rather abruptly. Agricultural developments occurred at least 90 millennia after the last major biological change evidenced in the fossil record, the evolution of anatomically modern humans. They occurred about 30 millennia after human cultural artifacts reach a level of stylistic and technical sophistication that convincingly demonstrates that people were cognitively modern as well. Our argument here is that a near step-function change in the earth’s environment from Pleistocene to Holocene climatic conditions about 11,600 years ago transformed the world from a place where agriculture was impossible anywhere to a place where it was possible on a large fraction of the earth’s surface. The various trajectories of agricultural origin and spread in different parts of the world thus result from a single, strong, “manipulation.” The replication of origins and spreads under different local conditions give us variation in additional factors. In particular, the great variation in rates of progress toward agriculture and in the rates of increase of sophistication of agriculture after its initial development give us insight into the processes which govern the rate of cultural evolution.

Cultural evolution is best studied using Darwinian methods

    The division of labor in the social sciences between economists, sociologists, and ethnographers on the one hand, and historians and archaeologists on the other, is very similar to the division of labor between paleontologists and students of microevolution. However, formal evolutionary theory is less well developed in the social sciences than in biology. The conceit of this paper is that thinking about social science problems, like the origins of agriculture, in the way that biologists think about analogous problems is a useful approach. Boyd and Richerson (1985) have developed this project mainly through investigating formal theoretical models of cultural evolution. Bettinger (1991; Bettinger and Eerkens 1999) has addressed the application of such theory to archaeological problems. The models are built on the idea that the transmission of culture by imitation and teaching is analogous to the transmission of genes. Cultural traditions, like genetically transmitted traits, are a population level phenomenon. The ideas we inherit culturally are a sample of those that characterize the larger population we live among. To a large extent, we are the prisoners of our cultural history, just as we are prisoners of the genes we inherit from our parents. At the same time, individuals (and more problematically groups of individuals) are the locus of the processes that are currently transforming cultural traditions into something new. If living individuals are the prisoners of the past they are also the architects of the future. Darwinian theory is, at its most formal, a calculus for analyzing change in an immortal population of mortal individuals when important state variables of the system are the result of the transmission of information from individual to individual.
    The biological theory already makes room for a rather long list of structural alternatives in the nature of inheritance systems and for a long list of “forces” that generate evolutionary dynamics. To apply the calculus to culture requires expanding both lists to take account of such things as the possibility of having more than two “parents” in cultural transmission. More fundamentally, we have to recognize that people actively and purposefully “engineer” the cultural traditions they inherit in a way that has no analogy in organic evolution. This expansion leads to many interesting differences between genetic and cultural evolution. Of the greatest significance is that cultural evolution is more rapid than genetic evolution. When the results of individual learning and choices among alternative cultural variants are transmitted to others by teaching or imitation, the potential for higher rates of change is obvious. To the extent that invention and choice are guided by adaptive rules, these processes will act to favor the same variants as selection, and rates of acquisition of new adaptations will be higher for a cultural than for a purely genetic system. This makes cultural traditions superior to organic adaptations when environments are rapidly changing or highly variable in space. We believe that the cultural transmission system arose in humans as an adaptation to the highly variable climate regime of the last few million years (Richerson and Boyd 2000).

Macroevolution is Interesting and Important

    Macroevolution is of particular interest when some evolutionary processes have such long time scales that they cannot not satisfactorily resolved by microevolutionary investigation. Such enigmatic, long time scale patterns are evident in both organic and cultural evolution. For example, in organic evolution, theory suggests and observations confirm, natural selection can easily lead to appreciable genetic change from one generation to the next (Endler 1986). In the case of artificial selection, huge differences can arise in a relatively few generations, as in domesticated breeds like dogs. The rate at which selection can fill up a new lake with a flock of new species is, under favorable conditions, measured in thousands of years (Johnson, et al. 1998). Yet, we know from the fossil record that long-term change is typically much, much slower. Even rather rapid changes, such as the expansion of brain size in Homo during the last 2 million years, would be undetectable by even the most sophisticated microevolutionary observations. In the case of cultural evolution, we live in a period in which change is easily appreciable from generation to generation. Invention is piled upon invention, leading to rapid technological change. Evolution of social institutions is also very fast. But the modern era is extraordinary. The Near Eastern trajectory of agricultural innovation was also comparatively rapid, but the whole sequence of increasing dependence upon plants and then upon domesticated plants and animals leading to nearly complete dependence on domesticates occupied roughly 4,000 years. At the average rate of change, it is unlikely that individuals living around Jarmo could possibly have been aware of the extraordinary evolutionary trajectory that they were on. On the other hand, perhaps there were periods of intense innovation alternating with periods of stagnation, such that, like ourselves, people were acutely aware of change during the periods of innovation. Or, perhaps, change is always rapid, but largely random in direction relative to the long-term trend. The Jarmo people, for example, might have always been aware of rapid change, but mostly alternating back and forth between more and less dependence on crops as (say) the primitive, unstable, social institutions necessary for settled life struggled to emerge, only to collapse in a few generations. We know from later examples, such as the collapse of the Mayan city-states, that rapid growth followed by rapid decline is not unusual (Tainter 1988).
    Thus, what we discover about evolution from investigating short-term processes often does not fit comfortably with what we know of the long-term trajectory of evolution (Boyd and Richerson 1992). The pioneering attempt to reconcile paleontology with the mainly microevolutionary theory of the mid-century Neo-Darwinian Synthesis was George Gaylord Simpson’s (1944) Tempo and Mode in Evolution. The difficulty of this project is illustrated by the storm of controversy that arose in evolutionary biology following Eldredge and Gould’s (1972) proposal that most evolution occurred in unusual punctuational events accompanying speciation, followed by selection among the resulting species. The claim was that the species selection process that dominates long-term evolution is decoupled from minor microevolutionary changes that evolutionary biologists can study. The debate became heated. Acerbic critics came to refer to the punctuation hypothesis as “the theory of evolution by jerks.” Test of the species selection hypothesis have not been kind to it (Carroll 1997; Levinton 1988), but the controversy has had the effect of focussing attention on the problem of explaining why evolution seems fast to “neontologists” but slow to paleontologists.
    In the social sciences the debate is equally acrimonious, and the theoretical gap even worse. At one extreme, rational choice theorists use sophisticated but completely ahistorical models explain human behavior. At the other, historians and cultural anthropologists describe historical change in compulsively detailed but atheoretical terms. Darwinian models of cultural evolution should help to address such questions. They incorporate the rational choice theorist’s decision-making effects and the historian’s transmission of traditions in the same models. The question becomes: What regulates the tempo and mode of cultural evolution?

Macroevolutionary hypotheses come in two flavors

    There are two kinds of hypotheses we can invoke to explain the tempo and mode of Darwinian evolution (now counting cultural evolution as a type of Darwinian evolution). First, rates of long-term change may be regulated by factors external to the evolutionary process itself. We know that the earth itself is an evolving system, so the environment in which evolution is occurring is always changing. Continents move, climates change, and organisms evolve in response. In a classic paper Valentine and Moores (1970) argued that the long-term pattern of evolution is at least partly regulated by the geographical and geochemical changes of the earth caused by seafloor spreading. Seafloor spreading causes slow trends in geochemistry and topography, and these slow changes often trigger abrupt switches from one stable regime of the ocean-atmosphere system to another. Natural selection is a rapid process on the geological time scale, so most of what paleontologists see is organic evolution tightly tracking environmental change caused by geological processes. The tempo of organic evolution is regulated by environmental change and the main mode of change is the ordinary adaptive evolution by natural selection that neontologists study. Since cultural evolution is an even more rapid process, the argument should be even stronger in that case.
    Second, internal processes limit the rate and direction of evolution. Even though selection is a very potent force on the geological time scale, it is by no means instantaneous. Some lag must exist between current environments and current organisms unless environmental change is vanishingly slow. For example, the poorly understood process of speciation limits the rate of evolution of new species bearing major new adaptations. Speciation is slow, many evolutionists believe, because geographical accidents must first isolate new proto-species. Otherwise, interbreeding with the mass of populations still adapted to the old niche swamps the effect of selection acting to reshape adaptations in a population that might otherwise enter a new niche. The number of species extant at any one time will be fewer than can potentially exist if environmental change is fast enough to prevent evolution from reaching the equilibrium number. Walker and Valentine (1984) estimated that on average the number of shelly marine invertebrate species is about 70% of the number that would occur if environmental change stopped long enough for equilibrium to be reached. The emergence of agriculture, and the subsequent evolution of complex societies over the last 10,000 years, are patterns that may well reflect mainly the impact of internal limitations to the rate and direction of cultural evolution, analogous to the limitations that speciation seems to impose on the rate of organic evolution. If so, understanding what these rate limiting process or processes are is fundamental to understanding how human history works.

Climate Change Started the Experiment

    How “clean” is the agricultural origins natural experiment? One possibility is that climate and other environmental factors that influence agricultural origins change in complex ways. Indeed, many of the external hypotheses invoked to explain agricultural origins appeal to detailed, local changes that imply that external influences were very complex. Childe (1951) argued that terminal Pleistocene desiccation crowded human populations around Near Eastern oases where more intensive land uses had to be developed to support the population, leading to agriculture. Wright (1977), finding that climate in the Near East became warmer and rainier around the time of plant domestication, argued that climate change brought together the suite of plants suitable for domestication in the Fertile Crescent area where the pioneering domestications took place. Since climate is always changing, at least a little bit, other climate changes could act as a trigger in the pioneering domestication sequence in other areas as well. Or other factors may be more important elsewhere even if climate was the key in the Near East. If the role of climate change is sufficiently complex, external and internal factors will be entangled in an even more complex picture. Climate change will have produced a much better natural experiment if it changed in one dramatic global episode and then stopped changing. If so, a large number of replicate human populations were set on a new evolutionary trajectory by the external step change, but all of the subsequent changes were due to internal processes or at least to other external processes. The many replicate populations faced different impediments to the evolution of agriculture and thus give us clues about what the internal rate limiting processes are. As it turns out, climate change does appear to provide the requisite conditions for such a clean experiment.

Agriculture was impossible in the Pleistocene

    The Pleistocene geological epoch is characterized by dramatic glacial advances and retreats. Using a variety of indirect measures of past temperature, rainfall, ice volume, and the like, mostly from cores of ocean sediments, lake sediments, and ice caps, paleoclimatologists have constructed a stunning picture of climate deterioration over the last 14 million years (Lamb 1977; and Crowley and North 1991; Partridge, et al. 1995; Bradley 1999). The Earth’s mean temperature has dropped several degrees and the amplitudes of fluctuations in rainfall and temperature have increased. For reasons that are as yet ill understood, glaciers wax and wane in concert with changes in ocean circulation, carbon dioxide, methane and dust content of the atmosphere, and changes in average precipitation and the distribution of precipitation. The resulting pattern of fluctuation in climate is very complex. As the deterioration has proceeded, different cyclical patterns of glacial advance and retreat involving all these variables have dominated the pattern. A 21,700 year cycle dominated the early part of the period, a 41,000 year cycle between about 3 and 1 million years ago, and a 95,800 year cycle during the last million years.
    This cyclic variation is very slow with respect to rates of cultural evolution. As the data from Jarmo indicate, major changes in subsistence take a few thousand years to complete, but not 20,000 years or more. The glacial cycle variation will not generate clean experiments because genetic and cultural evolutionary process could both be important on such long time scales. However, the variability that occurred on the time scales of the major glacial advances and retreats is also correlated with great variance at much shorter time scales. For the last 120,000 years, very high-resolution data are available from ice cores taken from the deep ice sheets of Greenland and Antarctica. Resolution of events lasting little more than a decade is possible in ice 90,000 years old, improving to monthly after 3,000 years ago. During the last glacial, the ice core data show that the climate was highly variable on time scales of centuries to millennia (GRIP 1993; Clark et al. 1999; Ditlevsen, et al. 1996).
    Figure 1 shows the data from the GRIP Greenland core. The 18O curve is a proxy for temperature; less negative values are warmer. Ca2+ is a measure of the amount of dust in the core, which in turn reflects the prevalence of dust-producing arid climates. The figure also shows histograms illustrating the obvious. The last glacial period was arid and extremely variable compared to the Holocene. Sharp excursions lasting a millennium or so occur in estimated temperatures, atmospheric dust, and greenhouse gases. The intense variability of the last glacial carries right down to the limits of the nearly 10 year resolution of the ice core data. Figure 2 shows Ditlevsen et al.’s (1996) analysis of a Greenland ice core. Not only was the last glacial much more variable on time scales of a century or more (150 yr low pass filter) but also on much shorter time scales (150 yr high pass filter). Even though diffusion within the ice core progressively erases high frequency variation in the core, the shift from full glacial conditions about 18,000 years ago to the Holocene interglacial is accompanied by a dramatic reduction in high frequency variation. The Holocene (the last relatively warm, ice free 11,600 years) has been a period of very stable climate, at least by the standards of the last glacial.

    Holocene weather extremes have significantly affected agricultural production (Lamb 1977). It is hard to imagine the impact of the qualitatively greater variation that characterized of most if not all of the Pleistocene. Devastating floods, droughts, windstorms, and the like, which we experience once a century, might have occurred once a decade. Tropical organisms did not escape the impact of this climate variation; temperature and especially rainfall were highly variable at low latitudes (Broecker 1996). Plant and animal populations responded to climatic change by dramatically shifting their ranges, but climate change was significant on the time scales shorter than those necessary for range shifts to occur. As a result, natural communities must have always been in the process chaotic reorganization as the climate varied more rapidly that they could reach equilibrium. The pollen record from the Mediterranean illustrates how much more dynamic plant communities were during the last glacial (Allen, et al. 1999).

    In addition to fluctuations in temperature and rainfall, the CO2 content of the atmosphere was about 190 ppm during the last glacial, rising to about 250 ppm at the beginning of the Holocene (Figure 3). Photosynthesis on earth is CO2 limited over this range of variation (Sage 1995). During the last glacial period, seed yields may have been something like 2/3s of Holocene yields. Nothing appears to be known about the evolutionary responses of plants to lower CO2 concentrations. If low CO2 caused plants to allocate more photosynthate to vegetative rather than reproductive tissues like seeds or storage organs like tubers, then the impact on low CO2 on the attractiveness of plant foods might be underestimated by the data reviewed by Sage.

    During the last glacial, at least, people lived under environmental conditions that almost certainly made heavy dependence on food plants unattractive. On present evidence we cannot determine whether aridity, low CO2, or climate variability is the main culprit in preventing the evolution of agriculture. Low CO2 and climate variation would handicap the evolution of dependence on plant foods everywhere. Plant food rich diets take considerable time to develop. Processing technology has to be invented and made efficient. Plant foods are generally low in protein and often high in toxins. Some time is required to work out a balanced diet of plant foods. The direct archaeological evidence suggests that people began to use extensively the technologies that underpinned agriculture only after about 15,000 B.P. (Bettinger in press). As CO2 levels rose, climate variability decreased, and rainfall increased, human populations in several parts of the world began to turn to the exploitation of locally abundant plant resources, but only during the so-called Bølling-Allerød period of near interglacial warmth and stability. One last siege of glacial climate, the Younger Dryas from 12,900 yrs B.P. until 11,600 yrs B.P., probably reversed these adaptive trends (e.g., Goring-Morris and Belfer-Cohen 1998). The Younger Dryas climate was appreciably more variable than the preceding Allerød-Bølling and the succeeding Holocene (Grafenstein, et al. 1999; Mayewski, et al. 1993). The ten abrupt, short, warm-cold cycles that punctuate the Younger Dryas ice record were probably felt as dramatic climate shifts all around the world. After 11,600, the Holocene period of relatively very warm, wet, stable, CO2 rich environments began. Subsistence intensification and eventually agriculture followed. Thus, while not a perfectly instantaneous change, the shift from glacial to Holocene climates was a very large change, and took place much more rapidly than cultural evolution could track.

    Might we not expect agriculture to have emerged in the last interglacial 130,000 years ago or even during one of the even older interglacials? No archaeological evidence has come to light suggesting the presence of complex technologies that might be expected to accompany forays into intensive plant collecting or agriculture at this time. The human populations of the last interglacial were still archaic in anatomy and behavior, though the first anatomically (but not yet culturally) modern humans occur in the archaeological record just after 130,000 years ago (Klein 1999: Ch.7). Very likely, then, the humans of the last interglacial were neither cognitively nor culturally preadapted for the evolution of agricultural subsistence. We should point out, however, that an external explanation might also explain the lack of agriculture during previous interglacials. Ice core data from the thick Antarctic ice cap at Vostok show that each of the last four interglacials over the last 420,000 years was characterized by a short, sharp peak of warmth, rather than the 11,500 year long stable plateau of the Holocene (Petit, et al. 1999). Further, the GRIP ice core suggests the last interglacial (130-80,000 B.P.) was more variable than the Holocene although its lack of agreement with a nearby replicate core for this time period makes this interpretation tenuous. On the other hand, the atmospheric concentration of CO2 was higher than during the Holocene in the three previous interglacials, and was stable at high levels for about 20,000 years following the warm peak during the last interglacial. The highly continental Vostok site unfortunately does not record the same high frequency variation in the climate as most other proxy climate records, even those in the Southern Hemisphere (Steig, et al. 1998). Some Northern Hemisphere marine and terrestrial records suggest that the Last Interglacial was highly variable while other data from suggest a Holocene length period of stable climates ca. 117,000-127,000 B.P. (Frogley, et al. 1999). In sum, better data on the high frequency part of the Pleistocene beyond the reach of the Greenland ice cores is needed to test hypotheses about human macroevolutionary events antedating the latest Pleistocene. That aside, the comparatively primitive nature of human adaptation during the last interglacial seems sufficient by itself to account for the observed lack of agricultural experimentation then.

In the Holocene, agriculture is compulsory in the long run

    Once a more intensive subsistence system is possible, it will, over the long run, replace the less intensive subsistence system that preceded it. The reason is simple: all else equal, any group that can use a tract of land more efficiently will be able to evict residents that use it less efficiently. More intensive uses support higher population densities, or wealthier societies per capita, or both. An agricultural frontier will tend to expand at the expense of hunters and gatherers as rising population densities on the farming side of the frontier motivate pioneers to invest in acquiring land from less efficient users. Whether the competition for land is economic, military, or for social prestige, the hunter-gatherer will be offered an attractive purchase price, dismal choices between flight, submission, or military defense at long odds against a more numerous foe, or an attractive idea about how to become richer through farming. Subsistence improvement generates both literal and metaphorical arms races. The archaeology supports this argument (Bettinger in press). Societies in all regions of the world undergo a very similar pattern of subsistence intensification in the Holocene, albeit at very different rates. Since ever more intensive subsistence systems have continued to evolve right up to the present, we do not have to worry about any very significant relaxation in the selection pressure for more efficient subsistence during the Holocene. A set of sustainable equilibrium adaptations to the Holocene may exist, but we have not yet discovered them. Thus, the experiment is rather clean. The Pleistocene-Holocene transition was a massive environmental change that was followed by more or less unchanged environmental conditions for the last 11,500 years. In response, human cultural evolution has generated a macroevolutionary trend towards more efficient production per unit land area.
    Given that competitive arms races drive the evolution of food production, the problem is to discover what the rate-limiting steps are in the cultural evolutionary processes leading to more intensive subsistence. The climate change “experiment” considerably simplifies our search for causal explanations by imposing a large, sudden, change and thereafter interfering minimally with responses dictated by the competitive ratchet, rate limited by internal processes. The mode of acquisition of agriculture (rarely indigenous development, more commonly diffusion), rate of progress, and exact sequence of forms of subsistence will depend upon local ecological and social conditions, regional setting, historical happenstance, and the like. Thus, each example of independent evolution of agriculture and each case of spread by conquest or diffusion is a case that can be examined for clues as to the relative importance of different evolutionary processes.
    The prospects for getting data adequate to test hypotheses are good. Archaeologists, in their attempts to explain particular transitions to plant rich and eventually agricultural subsistence in particular locations typically offer scenarios that are implicitly, at least, competitive ratchet models. Pristine origins of agriculture require plants that are susceptible to improvement via domestication. There have to be ways of incorporating proto-domesticates into the subsistence systems of hunter-gatherers. Seasonality may be important in establishing a premium on large scale planting for storage for the low productivity season. If the local ecological conditions for intensification are favorable, it occurs irreversibly. Workers such as MacNeish (1973), Flannery (1973), and Harris (1977), inspired by a particular sequence of intensification leading to agriculture, attempt to draw generalizations that should apply to other sequences. As information has improved about the particular sequences, a complex pattern of similarities and differences between cases has appeared. MacNeish (1991) attempts to cover the diversity of patterns involved in his “trilinear” theory of agricultural origins. Similarly, the spread of agriculture from its original centers to more distant regions is susceptible to archaeological, linguistic and biological investigations (Cavalli-Sforza, et al. 1994). Studies of hunter-gatherer sequences of intensification that stop far short of agriculture similarly often give us ecological and socially detailed accounts of the intensification processes including some with a quite clear account of the way the competitive ratchet probably worked before agriculture proper (e.g. Bettinger and Baumhoff 1982).

Alternative hypotheses are weak

    Aside from other forms of the climate change hypotheses described above, archaeologists have proposed two prominent internal hypotheses, population growth and cultural evolution, to explain the timing of the origin event. They were formulated before the nature of the Pleistocene-Holocene transition was understood, but are still the hypotheses most widely entertained by archaeologists (MacNeish 1991). Neither hypothesis provides a close fit with the empirical evidence.

Population Growth Has Wrong Time Scale

    Cohen’s (1977) influential book argued that population pressure was responsible for the origins of agriculture beginning at the 11,600 B.P. time horizon. He imagines that subsistence intensification is driven by increases in population density, and that a long, slow buildup of population gradually drove people to intensify subsistence systems to relieve shortages caused by population growth, eventually triggering a move to domesticates. Looked at one way, this idea is just the population growth part of the competitive ratchet. However, this argument fails to explain why pre-agricultural hunter-gatherer intensification and the transition to agriculture began in numerous locations after 11,600 years ago (Hayden 1995). Assuming that humans were essentially modern by the Upper Paleolithic, they would have had 30,000 years to build up a population necessary to generate pressures for intensification. Given any reasonable estimate of the human intrinsic rate of natural increase under hunting and gathering conditions (a large fraction of 1% yr-1 to 3% yr-1) populations substantially below carrying capacity will double in a century or less. Even much smaller rates would be sufficient to generate population pressure in far less than 30,000 years. The natural time scale of demographic processes is far too short to explain the long period of low population density in the Pleistocene followed by a rather sudden, widespread interest in intensification of subsistence in a narrow, rather recent, time horizon. It is also too rapid to explain the rather gradual increase in the sophistication of agriculture and other production systems over the last ten millennia.
Since the population explanation for agriculture and other adaptive changes connected with intensification remains very popular among archaeologists, it is worth taking the time here to examine it formally. The logistic equation is one simple, widely used model of the population growth. The rate of change of population density, N, is given by:

where r is the “intrinsic rate of natural increase”—the rate of growth of population density when there is no scarcity—and K is the “carrying capacity,” the equilibrium population density when population growth is halted by Malthusian checks. In the logistic equation, the level of population pressure is given by the ratio N/K. When this ratio is equal to zero the population grows at its maximum rate; there is no population pressure. When the ratio is one, Malthusian checks prevent any population growth at all. It is easy to solve this equation and calculate the length of time necessary to achieve any level of population pressure, p = N/K.

where p0 is the initial level of population pressure. Let us very conservatively assume that the initial population density is only 1% of what could be sustained with the use of simple agriculture, and that the maximum rate of increase of human populations unconstrained by resource limitation is 1% per year. Then it will take only about 920 years to reach 99% of the maximum population pressure (i.e. p = 0.99). Serendipitous inventions (e.g., the bow and arrow) that increase carrying capacity do not fundamentally alter this result. For example, only the rare single invention is likely to so much as double carrying capacity. If such an invention spreads within a population that is near its previous carrying capacity, it will still face half the maximum population pressure. At an r of 1% such an innovating population will again reach 99% of the maximum population pressure in 459 years.
One might think that this result is an artifact of the very simple model of population growth. However, it easy to add much realism to the model without any change of the basic result. Here we consider three such extensions: more realistic population dynamics, a model with dispersal in space, and a model in which people respond to population pressure by intensifying subsistence.

More realistic population dynamics. The logistic equation assumes that an increment to population density has the same effect on population pressure at low densities as at high densities. We know that this assumption is not correct for all species. For example, hunters pursuing herd animals that are easily spooked may generate much population pressure at low human population densities because killing only a small fraction of the herd make the many survivors difficult (but not impossible) to hunt. On the other hand, subsistence farmers spreading into a uniform fertile plain may feel little population pressure until all of plots of farmland of conveniently workable size are taken. If returns to additional labor on shrinking farms then drop steeply, most population pressure will felt at densities near K. To allow for such effects, ecologists often utilize a generalized logistic equation 

Population pressure is now given by the term (N/K)q. If q > 1, population pressure does not increase until densities approach carrying capacity, as is usually the case for species like humans that have flexible behavior and considerable mobility and thus can mitigate the effects of increasing population density over some range of densities. It seems intuitive that this would increase the length of time necessary to reach a given level of population pressure. However, this intuition is wrong. The generalized logistic can be used to derive a differential equation for p = (N/K)q

Thus, the differential equation for population pressure is always the ordinary logistic equation in which K = 1 and r is multiplied by q. This means that when q > 1, it takes less time to reach a given amount of population pressure than would be the case if q = 1. Reduced population pressure at low densities leads to more rapid initial population growth, and since population growth is exponential this more than compensates for the fact that higher densities have to be reached to achieve the same level of population pressure.
Allowing for dispersal: Once, after listening to one of us propound this argument, a skeptical archaeological colleague replied, “But you’ve got to fill up all of Asia, first.” This natural intuitive response betrays a deep misunderstanding of the time scales of exponential growth. Suppose that the initial population of anatomically modern humans was only about 104 and that the carrying capacity for hunter-gatherers is very optimistically 1 person per square kilometer. Given that the land area of the old world is roughly 108 km2, p0 = 104/108 = 10–4. Then using equation 2 and again assuming r = 0.01, Eurasia will be filled to 99% of carrying capacity in about 1400 years. The difference between increasing population pressure by a factor of 100 and by a factor of 10,000 is only about 500 years!

Moreover, this calculation seriously over estimates the amount of time that will pass before any segment of an expanding Eurasian population will experience population pressure because populations will approach carrying capacity locally long before the entire continent is filled with people. R. A. Fisher analyzed the following partial differential equation that captures the interaction between population growth and dispersal in space: 

Here N(x) is the population density at a point x in a one dimensional environment. Equation (5) says that the rate of change of population density in a particular place is equal to the population growth there plus the net effect of random, density-independent dispersal into and out of the region. The parameter d measures the rate of dispersal, and is equal to the standard deviation of the distribution of individual dispersal distances. In an environment that is large compared to d, a small population rapidly grows to near carrying capacity at its initial location, and then, as shown in Figure 4 (Redrawn from Ammerman and Cavalli-Sforza 1984), begins to spread in a wave-like fashion across the environment at a constant rate. Thus at any given point in space, populations move from the absence of population pressure to high population pressure as the wave passes over that point. Figure 4 shows the pattern of spread for r = .01 and d ~ 30. With these quite conservative values, it takes less than 200 years for the wave front to pass from low population pressure to high population pressure. More realistic models that allow for density dependent migration also yield a constant, wave-like advance of population (Murray 1989), and although the rates vary, we believe that the same qualitative conclusion will hold.
Intensification: The models so far assume that the carrying capacity of the environment is fixed (save where it is increased by fortuitous inventions). However, we know that people respond to scarcity caused by population pressure by intensifying production, for example by shifting from less intensive to more intensive foraging. Since intensification increases carrying capacity, intuition suggests that it might therefore delay the onset of population pressure. However, as the following model shows, this intuition, too, is faulty. Intensification allows greater population increases over the long run, but it does not change the timescales on which population pressure occurs.

Consider a population of size N in which the per capita income of the population is given by: 

where ym is the maximum per capita income, and I is a parameter that represents the degree of intensification. Thus per capita income declines as population size increases, but for a given population size, greater intensification raises per capita income. As in the previous models, we assume that as population pressure, now measured as falling per capita income, increases, population growth decreases. In particular, assume:

where ys is the per capita income necessary for subsistence. If per capita incomes are above this value, population increases; if per capita income falls below ys, population shrinks. If I is fixed, this equation is another generalization of the logistic equation. In an initially empty environment, population initially grows at a rate

but then slows and reaches an equilibrium population size

To allow for intensification we assume that people intensify whenever their per capita income falls below a threshold value yi. Thus
When per capita income is less than the threshold value yi, people intensify increasing the carrying capacity and therefore decreasing population pressure. When per capita income is greater than the threshold, they “de-intensify.” This may seem odd at first, but such de-intensification has been observed occasionally, for example, when horticultural Polynesian populations returned substantially to foraging on reaching the previously uninhabited archipelagos of Hawaii and New Zealand (Kirch 1984). The rate at which intensification changes is governed by the parameter a.
    If such a population begins in an empty habitat, it experiences two distinct phases of expansion. (Figure 5). Initially, per capita income is near the maximum, and population grows at the maximum rate. As population density increases, per capita income drops below yi, and the population begins to intensify, eventually reaching a steady state value
The steady state per capita income is above the minimum for subsistence but below the threshold at which people experience population pressure and intensify their production. At this steady state population growth continues at a constant rate,

that is proportional to the rate of growth in intensification. Thus, there is an initial phase in which the population grows rapidly until population growth is slowed by population pressure followed by a steady state in which population pressure is constant, and just enough intensification occurs to compensate for population growth. For plausible parameter values, the second phase of population growth steady state is reached in less than a thousand years. Interestingly, increasing the intrinsic rate of intensification, a, or the intensification threshold, yi, reduces the waiting time until population pressure is important.
    This picture of the interaction of demography and intensification leads to predictions quite different from those of scholars like Cohen (1977). For example, we do not expect to see any systematic evidence of increased population pressure prior to major innovations (something that apparently did not occur in the case of agricultural origins, Hayden 1995). If people are motivated to intensify whenever population pressure causes per capita income to fall below yi, and if, in the absence of intensification, populations adjust relatively quickly to changes in K by growth or contraction, then evidence of extraordinary stress, for example skeletal evidence of malnutrition, is likely only when rapid environmental deterioration exceeds a population’s capacity to respond via a combination of downward population adjustment and intensification or other forms of innovation. Some human populations might have curtailed birth rates in order to preserve higher incomes at any given level of I. In a sense, such populations have just redefined K to be a lower value that permits higher incomes by employing what Malthus called the “preventative checks” on population growth. The rest of the above analysis then applies with K measured in suitably emic terms. Cultural differences in the value of ys or K (Coale, 1986) will make evidence of stress more likely in populations where the effective carrying capacity is close to the subsistence carrying capacity compared to populations that reduce population growth rates some ways from absolute subsistence limits set by preventative checks. Similarly, populations that begin to intensify at a relatively high value of yi, will be less likely to suffer in environmental crunches. In other words, population pressure will tend to stay constant to the extent that rates of population growth and intensification are successful in adjusting subsistence to current conditions. Normally population growth and decline are quite rapid processes relative to rates of innovation and will keep average population size quite close to K. Short-term departures from K caused by short-term environmental shocks and windfalls should be the commonest reasons to see especially stressed or unstressed populations.

    Thus, for parameter values that seem anywhere near reasonable to us, population growth on millennial time scales will be rate limited by rates of improvement in subsistence efficiency not by the potential of populations to grow, just as Malthus argued. Populations can behave in non-Malthusian ways only under extreme assumptions about population dynamics. If the rate of intensification is more rapid than exponential population growth for any significant time period, then per capita incomes can rise under a regime of very rapid population growth, as in the last few centuries. This regime, if it had occurred in the past, should be quite visible in the historical and archaeological record because it so rapidly leads to large populations. Alternatively, population growth may have been limited in past populations by the analog of the modern demographic transition. Thus, hunter-gatherers might have resisted the adoption of plant based intensification because they viewed the life style associated with plant collecting or planting as a decrement to their incomes. However, resisting intensifications that increase human densities makes such groups vulnerable to competitive displacement by the intensifiers unless the greater wealth of the population limiters allows them to successfully defend their resource rich territories. On the evidence of the fairly rapid rate of spread of intensified strategies once invented, such defense is seldom successful (e.g. Ammerman and Cavalli-Sforza 1971; Bettinger and Baumhoff 1982).

    Of course, in a time as variable as the Pleistocene, populations may well have spent considerable time both far above and far below instantaneous carrying capacity. If agricultural technology were quick and easy to develop, the population pressure argument would lead us to expect Pleistocene populations to shift in and out of agriculture and other intensive strategies as they find themselves in subsistence crises due to environmental deterioration or in periods of plenty due to amelioration. Most likely, minor intensifications and de-intensifications were standard operating procedure in Pleistocene times. However, the time needed to progress much toward plant rich strategies was greater than the fluctuating climate allowed, especially given CO2 limited plant production.

Cultural Evolution Has Wrong Time Scale

    Robert Braidwood (1960) argued a “settling in” hypothesis to the effect that once humans acquired enough familiarity with plant resources, they naturally turned to them as a more efficient source of calories. This proposal is also quite consistent with the competitive ratchet. The issue is when the settling in process began. If our argument is correct, settling in could only begin in the latest Pleistocene and at the beginning of the Holocene. On this interpretation, our argument is an elaboration of Braidwood’s insight that cultural evolution is a rather slow process. However, if we interpret his argument to be that the settling in process began with the evolution of behaviorally modern humans, the time scale is wrong again. There is no evidence that people were making any progress at all toward agriculture for 30,000 years, and Braidwood’s excavations at Jarmo show that more like 4,000 years was enough to go from a plant-light hunting and gathering subsistence system to settled village agriculture in a fast case. 10,000 years in the Holocene was sufficient for even the slowest cases to develop plant-intensive gathering technologies.

Strong Similarities and Differences in Different Replicates of the Experiment

    The first test of the general hypothesis outlined here is whether the great mass of human societies have indeed been on an out-of-equilibrium trajectory toward more intensive subsistence techniques for the last 15,000 years (especially for the last 11,600 years). Most evolutionary scenarios imply quite different patterns. For example, if the Pleistocene-Holocene transition played a small role in creating environmental conditions favoring agriculture, then subsistence innovations should not be correlated with its appearance. If there has been no recent external change of global proportions, there is no reason to expect any coherent progressive trends among all the world’s societies.

Agriculture was independently invented ~ 10 times in the Holocene

    Table 1 gives a rough time line for the origin agriculture in seven fairly well understood centers of domestication, two more controversial centers, two areas that acquired agriculture by diffusion, and two areas that were without agriculture until European conquest. The list of independent centers is complete as far as current evidence goes, and while new centers are not unexpected it is hard to believe that the present list will be doubled by further investigation. The areas that acquired agriculture by diffusion are very numerous, so the three areas in Table 1 are but a small sample. The number of non-arctic areas without agriculture at European contact is small and the two listed, Western North America and Australia, are the largest and best known.
    Archaeologists are convinced that the seven centers of domestication are indeed independent on several grounds. First, the domesticates taken up in each center are distinctive and no evidence of domesticates from other centers turns up early in the sequence. For example, the Eastern North American center took into cultivation sunflower, a goosefoot, marsh elder, a squash, and other local plants. Long after these plants were taken into cultivation, maize was traded into the region and was grown in small quantities beginning around 2,000 B.P. However, it remained a minor domesticate until around 1,100 B.P., when it suddenly became the dominant domesticate, crowding out several traditional cultivars (Smith 1989). The Eastern squash is closely related to the Mexican squash, but genetic and morphological evidence indicates that two subspecies were independently taken into cultivation. Second, archaeology suggests that none of the centers had agricultural neighbors at the time that their initial domestications were undertaken. The two problematic centers, New Guinea and Lowland South America, present difficult archaeological problems (Smith 1995). Sites are hard to find and organic remains are rarely preserved. The New Guinea evidence consists of apparently human constructed ditches that might have been used in controlling water for taro cultivation. The absence of documented living sites associated with these features makes their interpretation quite difficult. The Lowland South American evidence consists of starch grains embedded in pottery fragments and phytoliths, microscopic silicious structural constituents of plant cell walls. The large size of some early starch grains and phytoliths convinces some archaeologists that root crops were brought under cultivation in the Amazon Basin at very early dates. Given the recency with which the Eastern North American and North and South China centers were firmly established, the discovery of a few more centers is likely. Recall Ehret’s (1998) linguistic evidence for multiple centers in Africa.
    Note that the date of the initiation of agriculture varies quite widely. In the Near East, Natufian peoples lived in settled villages and exploited the wild ancestors of wheat, barley beginning in the Allerød-Bølling warm period (14,500 -12,900 B.P.) (Henry 1989), and then reverted to mobile hunting and gathering during the sharp, short Younger Dryas (12,500-11,500 B.P.), the last of the high-amplitude fluctuations that were characteristic of the last glacial (Goring-Morris and Belfer-Cohen 1997; Bar-Yosef and Meadow 1995). Ice core evidence suggests that the Younger Dryas was significantly colder and more variable than the Allerød-Bølling and the Holocene (von Grafenstein, et al. 1999; Mayewski, et al. 1993). Post-Natufian cultures began to domesticate the same species virtually the moment warm and stable conditions returned after the Younger Dryas, around 11,600 B.P. Unfortunately, a flat spot in the 14C/calendar year calibration curve makes precise dating difficult for the most critical several hundred years centered on 11,600 B.P. (e.g., Fiedel 1999). In North, and possibly South, China, however, agriculture probably followed within a thousand years even though the earliest clearly agricultural complexes are considerably later (Crawford 1992; An 1991). Agriculture may prove to be as early in North China as in the Near East, since the earliest dated sites, which extend back to 8000 B.P., represent advanced agricultural systems that must have taken some time to develop. Excavations in North China north of the earliest dated agricultural sites document a technological change around 11,600 B.P. signaling a shift toward intensive plant and animal procurement that may have set this process in motion (Elston, et al. 1997).

The dates in Table 1 reflect considerable recent revision stemming from accelerator mass spectrometry 14C dating, which permits the use of very small carbon samples and can be applied directly to carbonized seeds and other plant parts showing morphological changes associated with domestication. Isolated seeds tend to work their way deep into archaeological deposits, and dates based on associated large carbon samples (usually charcoal) often gave anomalously early dates.

    The exact sequence of events also varies quite widely. For example, in the Near East, sedentarism preceded agriculture, at least in the Levantine Natufian sequence, but in Mesoamerica crops seem to have been added to a hunting and gathering system that was dispersed and rather mobile (MacNeish 1991: 27-29). For example, squash seems to have been cultivated around by 10,000 B.P. in Mesoamerica, some 4,000 years before corn and bean domestication began to lead to the origin of a fully agricultural subsistence system (Smith 1997). Some mainly hunting and gathering societies seem to have incorporated small amounts of domesticated plant foods into their subsistence system without this leading to full scale agriculture for a very long time. According to MacNeish, the path forward through the whole intensification sequence varied considerably from case to case.

Intensification of Plant Gathering Precedes Agriculture

    In all known cases, the independent centers of domestication show a sequence beginning with a shift from a hunter-gatherer subsistence system based disproportionately upon the capture of large game to a strategy based upon small game and especially plant seeds or other labor-intensive plant resources (Hayden 1995). The reasons for this shift are the subject of much work among archaeologists (Bettinger in press). These shifts invariably occur in the latest Pleistocene or later. Driven by the competitive ratchet, hunter-gathers who subsidize the hunting population with a large measure of plant-derived calories will tend to deplete the most desirable big game to levels that cannot sustain hunting specialists. Once better climates made the shift possible, the first escalation began in what would become the agricultural subsistence race. Braidwood’s reasoning that pioneering agriculturalists would have gained their intimate familiarity with proto-domesticates first as gatherers is logical and supported by the archaeology.
    The cases where intensification of plant gathering did not lead to agriculture are in some ways as interesting as the cases in which it did. The Jomon of Japan represents one extreme (Imamura 1996). Widespread use of simple pottery, a sure marker of well developed agricultural subsistence in Western Asia, was very early in the Jomon, contemporary with the latest Pleistocene Natufian in the Near East. By 11,000 yrs B.P., the Jomon people lived in settled villages, depended substantially upon plant foods, and used massive amounts of pottery. However, the Jomon domesticated no plants, although they cultivated a few minor crops like bottle gourds. Agriculture came to Japan with imported rice from the mainland only about 2,500 B.P. Interestingly, acorns were a major item of Jomon subsistence. The people of California were another group of sedentary hunter-gatherers that depended heavily on acorns. However, in California the transition to high plant dependence began much later than in the Jomon (Wohlgemuth 1996). Millingstones for grinding small seeds became important after 4,500 B.P., although seeds were of relatively minor importance. After 2,800 B.P. acorns processed with mortars and pestles became an important subsistence component and small seeds faded in comparative importance. In the latest period, after 1200 B.P., quantities of small seeds were increasingly added back into the subsistence mix alongside acorns in a plant dominated diet. Other peoples with a late onset of intensification include the Australians. The totality of cases tell us that any stage of the intensification sequence can be stretched or compressed by several thousand years even though reversals are rare (Harris 1996; Price and Gebauer 1995). Farming gave way to hunting and gathering in the southern and eastern Great Basin of North America after a brief extension of farming into the region around 1,000 B.P.(Lindsay 1986). A similar reversal occurred in southern Sweden between 2,400 and 1,800 B.P., Zvelebil 1996), and we have already noted the case of Polynesian de-intensification on newly settled islands.

More Intensive Technologies Tend to Spread

    Once well-established agricultural systems exist, they expand at the expense of hunting and gathering neighbors (Bellwood 1996). Ammerman and Cavalli-Sforza (1971) summarize the movement of agriculture from the Near East to Europe, North Africa and Asia. The spread into Europe is best documented. Agriculture reached the Atlantic seaboard about 6,000 B.P. or about 4,000 years after its origins in the Near East. The regularity of the spread, and the degree to which it was largely a cultural diffusion process as opposed to a population dispersion as well, are matters of debate. Cavalli-Sforza et al. (1994: 296-299) argue that demic expansion by Western Asians was an important process with the front of genes moving at about half the rate of agriculture. They imagine that pioneering agricultural populations moved into territories occupied by hunter-gatherers, intermarried with the pre-existing population. The then mixed population in turn sent agricultural pioneers still deeper into Europe. They also suppose that the rate of spread was fairly steady, though clearly frontiers between hunter-gathers and agriculturalists stabilized in some places (Denmark, Spain) for relatively prolonged periods. Zvelebil (1996) stresses the durability of frontiers between farmers and hunter-gatherers and the likelihood that in many places the diffusion of both genes and ideas about cultivation was a prolonged process of exchange across a comparatively stable ethnic and economic frontier. Further archaeological and paleo-genetic investigations will no doubt gradually resolve these debates. Clearly, the spread process is at least somewhat heterogeneous.
    Other examples of the diffusion of agriculture are relatively well documented. For example, maize domestication is dated to about 6200 B.P. in Central Mexico and to about 4000 B.P. in , in the southwestern U.S. (New Mexico; Smith 1995; Matson 1999). In this case, the frontier of maize agriculture stabilized for a long time, only reaching the Eastern U.S. at a comparatively late date as noted above. Maize failed entirely to diffuse westward into the Mediterranean parts of California even though peoples growing it in the more arid parts of its range in the Southwest used irrigation techniques that would have worked well there. As with the origin process, the rate of spread of agriculture exhibits an interesting degree of variation.

Agriculture a “Natural” Outcome of More Intensive Technologies

    A number of plant and animal biologists have taken an interest in the process of the evolution of domesticates (Zohary and Hopf 1993, Rindos 1984, Bretting 1990, Smith, 1995). The first changes in seed crop plants generally include larger seeds, the loss of natural seed dispersal mechanisms, and the loss of seed dormancy. These changes do not necessarily require any conscious selection by cultivators. Once humans began to intensively exploit wild seed plants, especially once they prepared seedbeds and deliberately planted seeds, these changes will tend to be automatic. Larger seeds yield more competitive seedlings in the artificial seedbed. Those seeds that do not drop off the stalk are more likely to be harvested, favoring genotypes with non-shattering seed heads. With humans controlling the time of planting and storing seed in cool, dry locations, seed dormancy, rather than producing adaptively facultative germination, merely increases the chance that a seed will fail to germinate in time to be harvested. As Flannery (1973) observed, it is not clear which species is the domesticated and which the domesticator in agriculture! Some plants responded to human harvest in such a way that they induced humans to plant more of that species in competition with others, raise larger numbers of children to plant still more fields, and eventually to carry the plant across continents.
    Of course, every human population includes close observers and inveterate experimenters. No doubt deliberate ingenuity and sustained artificial selection also played roles in the evolution of domesticates, as they do in the development and conservation of landraces today (Brush 1995).

What Regulates Tempo and Mode of Cultural Evolution?

    The overall pattern of subsistence intensification is clearly consistent with the hypothesis that agriculture was impossible in the Pleistocene but mandatory in the Holocene, but the real test is whether or not we can give a satisfactory account of the variation in the rate and sequence of intensification. Work on this project is in its infancy, and only some rather preliminary and speculative answers are possible.

External Processes Play a Role

In general, external processes of macroevolution will tend to operate at longer time scales and internal processes at shorter time scales. Evolution has to be flexible and rapid enough to roughly track environmental variation, and environmental variation is greater at longer than shorter time scales. Those species that fail to track longer-term variation will go extinct, and will not contribute to the evolutionary processes we see in action. At the same time, no evolutionary process responds instantaneously. There is always stickiness at short enough time scales. If the world were simple, the time scales of internal and external processes would be entirely separate. The Pleistocene/Holocene transition, we argue, is a good natural experiment because the separation of scales is very good. To a first approximation, the lack of agriculture before the Holocene is due to the external effect of the Pleistocene climate; after that the rate of progress toward ever more intensive subsistence is largely regulated by internal processes. However, external causes did not entirely disappear in the Holocene. To completely isolate the effects of internal processes we have to control for residual external effects.
    Climate change may play a small role. The Holocene climate is only invariant relative to the wild oscillations of the last glacial (Lamb, 1977). For example, seasonality (difference between summer and winter insolation) was at a maximum at the beginning of the Holocene and has fallen since. The so-called “Climatic Optimum,” a broad period of warmer temperatures during the middle Holocene, caused a wetter Sahara, and the expansion of early pastoralism into what is now forbidding desert. The late medieval onset of the Little Ice Age caused the extinction of the Greenland Norse colony. Agriculture at marginal altitudes in places like the Andes seems to respond to Holocene climatic fluctuation (Kent 1987). While the effect of Holocene climate fluctuations on regional sequences must always be kept in mind, the dominance of the nearly always monotonic tendency to increased subsistence efficiency per unit land seems likely to be driven by other processes.
    Geography may play a big role. Diamond (1997) argues that Eurasia has had the fastest rates of cultural evolution in the Holocene because of its size and to a lesser extent its orientation. Plausibly, the number of innovations that occur in a population is, to a first approximation, a function of total population size and the flow of ideas between sub-populations. Societies acquire most innovations by diffusion from other societies, and isolated societies will be handicapped in their rate of innovation by a slower rate of diffusion. Eurasia is the largest continent and is especially large in its east-west dimension. Innovations occurring at one end of the continent will eventually diffuse to the other, spreading along lines of latitude with relatively similar environments. Small land-masses like Australia and New Guinea are substantially isolated from the larger world and have a much smaller base of innovators. The Americas, though quite respectable in size, are oriented with their major axis north-south. Consequently innovations have to mainly spread across lines of latitude from the homeland environment to quite different ones. Wheat varieties originating in the Near East could spread to Spain and to Western China without substantial modification. The maize moving north from Mexico would have to be selected to respond appropriately to longer day length and a shorter growing season. Maize spreading to South America would have to move through a belt of hot, wet, tropics with little day length variation before reaching the cool, semiarid, subtropical highlands of South America so similar to its original homeland.

    Since we know that the original centers of agricultural innovation were relatively small compared to the areas to which agriculture later spread, it is clear that complex adaptations arise infrequently, but are relatively easy to acquire by imitation. The most isolated agricultural region in the world, Highland New Guinea, underwent an economic revolution in the last few centuries with the advent of American sweet potatoes, a crop that thrives in the cooler highlands above the malaria belt of lowland New Guinea (Wiessner and Tumu 1998). The agriculture of both the Old and New Worlds was impacted substantially by the Columbian exchanges beginning 500 years ago. Eurasian agriculture found good uses for American maize, potatoes, squash, beans, peppers, and tomatoes, despite the already deep list of ancient Eurasian cultivars. The marginal benefit of increased numbers, hence diversity, of domesticates and inventors seems to have been substantial even on the largest continent on earth.

Coevolutionary Processes Play a Big Role

    In the case of domestication, the distinction between external and internal regulation of process rates is ambiguous without further definition. The genetic changes in the proto-domesticate, as it responds to cultural practices, are external to the cultural evolutionary system in one sense. In another, the coevolutionary system as a whole is, perhaps, more naturally seen as having intertwined the evolution of human culture and plant and animal genomes as to make such a distinction artificial. Human genomes may also be involved in the coevolutionary system.
Agriculture requires pre-adapted plants and animals. In each center of domestication, people domesticated only a handful of the wild plants that they formerly collected, and of this handful even fewer are widely adopted outside those centers. The same is true for domesticated livestock. Wheat, rice, and maize make disproportionate contributions to total world crop production. Cattle, sheep, goats, hogs, and chickens make an outsized contribution to total world livestock production. A tiny fraction of the world’s plant and animal diversity ended up as significant domesticates. Zohary and Hopf (1989) have listed some of the desirable features in plant domesticates. Aside from obvious things like large seed size, most Near Eastern domesticates had high rates of self-fertilization. This means that farmers can select desirable varieties and propagate them with little danger of gene flow from other varieties or from weedy relatives. Maize, by contrast, outcrosses at high rates. Perhaps the later and slower evolution of maize compared to Near Eastern domesticates is due to the difficulty of generating responses to selection in the face of gene flow from unselected populations. Smith (1995) discusses the many constraints on potential animal domesticates. For example, many large ungulates like deer depend upon rapid flight to avoid predators. They are thus very skittish and adapt poorly to human handling. Larger herd ungulates that are less susceptible to large predators, like cattle, have shorter flight distances and more stolid temperaments. Even in the most favorable cases, the evolution of new domesticates is not an instantaneous process. In at least some times and some places, the rate of evolution of domesticates was likely the rate-limiting step in the agricultural intensification process.
    Diamond (1997), drawing on the work of Blumler (1992), notes that the Near Eastern region has a flora that is unusually rich in large-seeded grasses (Table 2). California, by contrast, is quite depauperate in large seeded grasses, having not a single species that passed Blumler’s criterion. California has so many climatic, topographic, and ecological parallels with the precocious Fertile Crescent that its very tardy development of plant-intensive subsistence systems is a considerable puzzle. The presence of a large suite of agriculturally preadapted plants in the Near East but not in California provides a plausible hypothesis to explain the difference in agriculture but not plant intensification itself.

Humans have to adapt biologically to agricultural environments. While the transition from hunting and gathering to agriculture resulted in no genetic revolution in humans, a number of modest new biological adaptations were likely involved in becoming farmers. Perhaps the best-documented case is the evolution of adult lactose absorption in human populations with long histories of dairying (Durham, 1991). Most human populations, as in mammals more generally, adults lose the ability to digest the milk sugar lactose. However, the frequency of adult absorption reaches near fixation in Western European and African populations with a long tradition of drinking fluid milk. It is present in intermediate frequencies in populations in the circum-Mediterranean and other subtropical regions that more typically consume milk as cheese or fermented products low in lactose. Other cases suggest human genetic coevolution with domestication. For example, agricultural populations metabolize alcohol more rapidly than hunter-gatherers, either to improve the nutritional yield from such products or to avoid toxic effects. To some extent the relatively slow rate of biological adaptation will act as a drag on the rate of cultural innovation.

    Diseases limit population expansions, protect inter-regional diversity. McNeill (1979) and Crosby (1986) draw our attention to the coevolution of people and diseases. The increases in population density that resulted from the intensification of subsistence invited the evolution of epidemic diseases that could not spread at lower population densities. One result of this process was likely to slow population growth and hence slow the rate at which growing populations fuel the competitive ratchet. In more disease prone environments, such as malarial tropical lowlands, this restraint on the growth process may have been quite severe. A suite of hemoglobins have arisen in different parts of the world that confer partial protection against malarial parasitism (Cavalli-Sforza, et al. 1994). Some authors have suggested that land clearance associated with agriculture increased malarial infection rates. Cavalli-Sforza et al. estimate that it would take about 2,000 years for a new mutant hemoglobin variant to reach equilibrium in a population of 50,000 or so individuals. The relatively late and slow rate of the evolution of agricultural systems in Africa might, in part be attributable to the high rates of disease infection of people and livestock slowing the rate of intensification induced population growth to the macroevolutionary time scale (see also Gifford-Gonzales in press).

    As both McNeill and Crosby note, regional populations tend to have their own diseases, to which they are more or less immune, while they tend to be susceptible to the diseases of people from other regions. This pattern of susceptibility will tend to act against communication between regions. Travelling strangers who might bring new ideas and stronger competition will tend to fall sick and to bring diseases that cause their hosts to fall sick (at least temporarily reducing competition for land). Disease barriers will tend to isolate regions and hence slow rates of diffusion of ideas. This effect was probably historically most severe for Africa and other Old World tropical areas that were quite unhealthy for people from temperate Eurasia. On the other hand, in cases like the European invasion of the New World and smaller insular areas with low loads of infectious disease, European conquest was greatly aided if not entirely made possible by a disease gradient operating at the expense of native populations. To the extent that populations sustaining higher densities sustain more endemic diseases, they will have a competitive edge against less dense populations without a history of exposure to so many diseases.

Purely Internal Processes Play a Role

    The macroevolutionary patterns in which we are most interested are the processes strictly internal to cultural evolution that might limit the rate of intensification. We (Boyd and Richerson 1985; Bettinger 1991) view cultural evolution as a Darwinian process of descent with modification. Evidence about the macroevolutionary tempo and mode of cultural evolution bear directly on our picture of the micro scale processes that form the body of the theory. For many characters human decisions have relatively weak effects in the short run although they can be powerful when integrated over many people and appreciable spans of time. If the rational-choice element in our models is sufficiently strong, then the rate of cultural evolution itself will never be rate limiting and cultural evolution will lack the Darwinian element. The tempo of subsistence intensification will then be entirely governed by external or coevolutionary processes.
    Selective processes can also cause quite rapid evolution compared to the time scales of external processes. Darwinians use the metaphor of an “adaptive topography” to picture the action of natural selection. Selection acts to drive populations up slopes of fitness toward peaks in an adaptive topography. If fitness topographies are smooth, the populations climb more or less directly and quickly to global optima. On the other hand, if fitness topographies are rough, much more complex trajectories will ensue. Populations will tend to get stuck on local peaks in the foothills of a fitness topography. Chance events may eventually carry small populations across fitness valleys. For example, minor climate change may shift the topography allowing one population to escape a local peak and run up to a higher one. A population of populations (a metapopulation in the jargon) will, by various historically contingent routes, gradually filter to higher and higher fitness peaks. Historically minded scholars have always believed that such complexities were important (e.g. Vayda 1995), but only recently have models of biological, cultural, and economic change become sophisticated enough to investigate them formally. Typically the structural changes in such models are straightforward extensions of classical models that behave ahistorically. We think that such models should lead to closer and more productive interactions between theorists and historians (Boyd and Richerson 1992). Several internal processes may lead to rough fitness topographies and thus act to limit the rate of cultural evolution of intensification.
    New technological complexes evolve with difficulty. One problem that will tend to slow the rate of cultural (and organic) evolution is the sheer complexity of adaptive design problems. As engineers have discovered when studying the design of complex functional systems, discovering optimal designs is quite difficult. Blind search algorithms often get stuck on local optima. Piecemeal improvements at the margin are not guaranteed to find globally optimal adaptations by myopic search. Yet, myopic searches are what Darwinian processes do.

    Parallel problems are probably rife in human subsistence systems. The shift to plant-rich diets is complicated because plant foods are typically deficient in essential amino acids, and vitamins, have toxic compounds to protect them from herbivore attack, and are labor intensive to prepare. The diet of Pleistocene hunters and gathers probably focused on high rates of meat intake supplemented by high quality plant foods such as ripe fruit and nuts. High quality plant resources are scarce and the inefficiency of natural herbivore populations means that meat offtake rates are usually quite limited. Intensification requires a focus on seeds low in essential amino acids (maize), tubers with poisonous protection (bitter manioc), and the like. Even the best plant resources like wheat require protein supplementation with animal products or legumes. Skeletal material suggests that early agricultural peoples were often less well nourished than their hunter-gather ancestors (Cohen and Armelagos 1984). Finding a mix of plant and animal foods that provides adequate diet is not a trivial problem. Many different mixes may work more or less well, but which one(s) are globally or nearly globally optimal?

    New World farmers eventually discovered that boiling maize in wood ashes improved its nutritional value. The hot alkaline solution breaks down an otherwise indigestible seed coat protein that contains some lysine, an amino acid that is low in maize relative to human requirements (Katz, et al. 1974). Hominy and masa harina, the corn flour used to make tortillas, are forms of alkali treated maize. The value of this practice could not have been obvious to its inventors or later adopters, yet all American populations that made heavy use of maize employed it. The dates of origin and spread of alkali cooking are not known. It has not been reinvented in Africa even though many African populations have used maize for centuries.

    The prehistory of subsistence intensification consists of a long sequence of inventions that gradually increase the sophistication of food producing systems. Arboriculture, irrigation, animal traction, dairying, wool use, and the like were added to toolkits thousands of years after people were committed to agricultural production. Plausibly, each of these inventions was the product of prolonged experimentation and development in a center of origin, followed by a slow spread to adjacent areas. Once again, the large and size east-west axis of the Eurasian land mass likely accelerated the pace of development relative to other areas. The New World, among other problems, was poor in animal species suitable for domestication due to the terminal Pleistocene megafaunal extinction event, arguably caused by the initial wave of human settlers (Martin and Klein 1984). To make matters worse, the most important New World domesticates, the camelids of the temperate Andes, did not spread across the wet tropical belt into Central and North America. As intensification proceeded, each innovation adopted had to be fitted into an existing system of land use, labor allocation, and diet. In rapidly developing Eurasia, the rate of subsistence intensification could well have been limited partially or even mainly by the technical complexity of the innovation process.

    Some of the difficulties in evolving new subsistence systems may stem from dynamic problems in the economy. Lee (1986) imagines that the rate of innovation is correlated with the rate of profit for invested capital. In a Ricardian world, profits will be low when population density is low because wages will be too high. Conversely, when population density is too high, profits will be squeezed by rents to land. With profits low, entrepreneurs will lack surplus capital to invest in innovation. Entrepreneurs will thus innovate only in a narrow window between these two constraints. If we imagine a population fluctuating due to epidemic disease, famine, and the like, little bursts of innovation will occur as population fluctuates through the key window. Day and Walter (1989) investigate a similar model the exhibits a chaotic growth path.

    New social institutions evolve with difficulty. As anthropologists and sociologists such as Julian Steward (1955) have long emphasized, human economies are social economies. Even in the simplest human societies, hunting and gathering is never a solitary occupation. At the minimum, such societies have division of labor between men and women. Hunting is typically a cooperative venture. The unpredictable nature of hunting returns typically favors risk sharing at the level of bands composed of a few cooperating families because most hunters are successful only every week or so (Winterhalder 1986). Portions of kills are distributed widely, sometimes exactly equally, among band members.

    The deployment of new technology requires changes in social institutions to make best use of innovations. The increasing scale of social institutions associated with rising population densities during the Holocene have dramatically reshaped human social life. Even the first steps of intensification required significant social changes. Gathering is generally the province of women and hunting of men. Male prestige systems are generally tied up with hunting success. A shift to plant resources requires scheduling activities around women’s work rather than men’s. Using more plants will conflict with men’s preferences as driven by a desire for hunting success; it will require a certain degree of women’s liberation to intensify subsistence. Since men generally dominate women in group decision-making, male chauvinism will tend to limit intensification. Bettinger and Baumhoff (1982) argue that the spread of Numic speakers across the Great Basin a few hundred years ago was the result of the development of a plant-intensive subsistence system in the Owens Valley. Once developed, the spread eastward to the Rockies because the Numics could outcompete the previous inhabitants who neglected plant resources in favor of the hunt. The evidence suggests a demic expansion rather than a cultural diffusion. Apparently, the groups specialized in the hunt would not or could not shift to the more productive economy to defend themselves, perhaps because males clung to the outmoded, plant poor, subsistence. Those with a vested interest in current social institutions are naturally conservative.

    Classic hunting and gathering societies are relatively small in scale, and are quite egalitarian (Woodburn 1980). Hence they have very informal leadership. The evolution of formal coercive leadership, hierarchical command and control systems, and the rise of marked social inequality are resented and resisted, most likely because they violate our social “instincts,” for example our sense of what is and is not fair (Richerson and Boyd 1998, 1999). As people began to collect in villages, ethnographic analogies suggest that formal political leaders likely became important. As the agricultural toolkit expanded, larger markets for sophisticated products came to favor an ever-finer division of labor. Various sorts of institutions arose to manage the division of labor—caste systems, command economies, market systems, international trade systems and the like. The creation of storable and portable wealth exacerbates defense and security problems, leading to the rise of armies and police forces.

    The evolution of complex social institutions is arguably as or more difficult than the evolution of technological innovations. If so, social institutions will tend to regulate technological progress. For example, North and Thomas (1973) argue that new and better systems of property rights set off the modern industrial revolution, much as Bettinger and Baumhoff argue that new social institutions were required to favor innovations in plant collection strategies at the inception of the intensification trajectory. A major revolution in property rights is likely also necessary for intensive hunting and gathering and agriculture to occur (Bettinger 1999). The sharing ethic that characterizes many simple hunter-gatherer groups discourages individual efforts to intensify production since freeloaders enjoy the extra resources generated by hard working individuals. Thus, accumulating resource surpluses to store for seasons of shortfall is heavily contingent on an ideology in which collected resources are regarded as private property. The shift from resources as public goods to resources as private goods, however, is likely to be difficult since any form of innovation along these lines will, at least initially, be regarded as antisocial. The comparative history of the social institutions of intensifying societies exhibits many examples of societies getting far ahead of others in one dimension or another. For example, the Chinese merit based bureaucratic system of government was established at the expense of the landed aristocracy, beginning in the Han dynasty (2,200 B.P.) and completed in the Tang (1,400 B.P.) (Fairbank 1992). In the West similar institutions arose only in the last few centuries.

    Progress aside, functionally analogous institutions are notoriously different in different culture areas. The Indian caste system is a unique, or at least uniquely hypertrophied, method of organizing a complex division of labor. The Turkish system of raising armies (Janissaries) and even ruling classes (Mamlukes) by enslaving Christian boys is similarly unique. Social institutions seem generally to diffuse much less readily than technology. In the late mediaeval and early modern period, the West acquired a considerable number of important technical innovations from China, but certainly not the idea of a Confucian bureaucracy. Social institutions violate four of the conditions that tend to facilitate diffusion (Rogers 1983). Foreign social institutions are often (i) not compatible with existing institutions, (ii) complex, (iii) difficult to observe, and (iv) difficult to try out on a small scale. Technical innovations like the compass and the cannon are thus much more easily diffused than social institutions. Hence historical differences in social organization, many of them less nearly optimal than those already traditional elsewhere, are quite persistent.

    Ideology may play a role. Forces other than strictly utilitarian ones like natural selection govern the evolution of fads, fashion, and belief systems. They are susceptible feedback and runaway dynamics that defy common sense (Boyd and Richerson 1985: Chap. 8). The arbitrariness of the runaway dynamic produces path dependent historical change, often in defiance of economic or fitness rationality. The links between belief systems and subsistence are nevertheless incontestably strong. To build a cathedral requires an economy that produces surpluses that can be devoted to grand gestures on the part of the faithful. The moral precepts inculcated by the clergy in the cathedral underpin the institutions that in turn regulate the economy. Arguably, ideological innovations often drive economic change. Recall Max Weber’s classical argument about the role of Calvinism in the rise of capitalism. The Twentieth Century’s largely failed experiments with nationalist and socialist command economies is an example of an ideology driven exploration in the space of all possible social formations, but not one that was very well controlled to make progress.    Contemporary Russia illustrates the great difficulty of making the transition from a failed set of institutions to a new set. So, arguably, does the inability of the United States to take risks with economic growth in pursuit of policies that may be necessary to prevent global warming and similar threats to economic sustainability. Contemporary consumerism is arguably the result of institutionalizing the pathological tendency of status competition to escalate without limit (Frank and Cook 1995, Easterlin 1995).

We Have a Shadowy Outline of the Tempo and Mode of Cultural Evolution

    The large step change in environment at the Pleistocene-Holocene transition set off the trend of subsistence intensification of which modern industrial innovations are the latest examples. The diversity of trajectories taken by the various regional human sub-populations since 11,600 B.P. are natural experiments that reveal the internal limitations on the rate of cultural evolution. We believe that the discussion above is not only a basic review of the main cultural macroevolutionary hypotheses currently entertained by scholars but also a crude picture of how cultural macroevolution has actually happened. There is a rather long list of processes that probably interacted to regulate the nearly unidirectional trajectory of subsistence intensification, population growth, and institutional change that the world’s societies have followed in the Holocene. Social scientists are in the habit of treating these processes as mutually exclusive hypotheses. They seem to us to be competing but certainly not mutually exclusive. At the level of qualitative empiricism, tossing any one out entirely leaves puzzles that are hard to account for and produces an obvious caricature of the actual record of change. If this conclusion is correct, the task for historically minded social scientists is to refine estimates of the rates of change that are possible due to the various evolutionary processes and to estimate of how those rates change as a function of natural and socio-cultural circumstances. The origins of agriculture and ensuing events are an excellent macroevolutionary “laboratory” for understanding the internal processes of cultural evolution.

The End of Agriculture?

    Those who are familiar with the Pleistocene often remark that the Holocene is just the “present interglacial.” The return of climate variation on the scale that characterized the last glacial is quite likely if current ideas about the Milankovich driving forces of the Pleistocene are correct. Sustaining agriculture under conditions of much higher high frequency environmental variation than farmers currently cope with would be a very considerable technical challenge. At the very best, lower CO2 concentrations and lower world average precipitation suggest that world average agricultural output would fall considerably.
    In one sense, though, the Holocene is not just another interglacial. Recall that Petit et al. (1999) show it to be uniquely long, although decidedly cooler than the maximum temperatures of the previous four interglacials, at least in continental Antarctica. Current anthropogenic global warming via greenhouse gasses threatens to elevate world temperatures to levels that in past interglacials apparently triggered a large feedback effect producing a relatively rapid decline toward glacial conditions. The Arctic ocean ice pack is currently thinning very rapidly (Kerr 1999). A dark, open Arctic Ocean would dramatically increase the heat income at high northern latitudes, and have large, difficult to guess impacts on the Earth’s climate system. No one can yet estimate the risks we are taking of a rapid return to colder, drier, more variable environment with less CO2, nor evaluate exactly the threat such conditions imply for the continuation of agricultural production. Nevertheless, the intrinsic instability of the Pleistocene climate system, and the degree to which agriculture is dependent upon the unusually long Holocene stable period, should give one pause (Broecker 1997).



Allen, J. R., U. Brandt, A. Brauer, H.-W. Hubberten, B. Huntley, J. Keller, M. Kraml, A. Mackensen, J. Mingram, J. F. Negendank, N. R. Nowaczyk, H. Oberhansli, W. A. Watts, S. Wulf and B. Zolitschka
    1999 Rapid Environmental Changes in Southern Europe During the Last Glacial Period. Nature 400:740-743.
Ammerman, A.J. and L.L. Cavalli-Sforza
    1971 Measuring the Rate of Spread of Early Farming in Europe. Man 6:674-688.
An, Z.
    1991 Radiocarbon Dating and the Prehistoric Archaeology of China. World Archaeology 23(2):193-100.
Barnola, J.M., D. Raynaud, Y.S. Korotkevich and C. Loris
    1987 Vostok Ice Core Provides 160,000-Year Record of Atmospheric CO2. Nature 329: 408-414.
Bar-Yosef, O, and R.H. Meadow
    1995 The Origins of Agriculture in the Near East. In Last Hunters, First Farmers: New Perspectives on the Prehistoric Transition to Agriculture, edited by T.D. Price. and B. Gebauer, pp.39-94. School of American Research Press, Santa Fe.
Bellwood, P.
    1996 The Origins and Spread of Agriculture in the Indo-Pacific Region: Gradualism and Diffusion or Revolution and Colonization. In The Origins and Spread of Agriculture and Pastoralism in Eurasia, edited by D. R. Harris, pp. 465-498. Smithsonian Institution Press, Washington, D. C.
Bettinger, R.L.
    1991 Hunter-Gatherers: Archaeological and Evolutionary Theory. Plenum, New York.
Bettinger, R. L.
    1999 From Traveler to Processor: Regional Trajectories of Hunter-Gatherer Sedentism in the Inyo-Mono Region, California. In Settlement Pattern Studies in the Americas: Fifty Years Since Viru, edited by B. R. Billman and G. M. Feinman, pp. 39-55. Smithsonian Institution Press, Washington, D.C.
Bettinger, R.L.
    2000 Holocene Hunter-gatherers. In Archaeology at the Millennium: A Sourcebook, edited by G.M. Feinman and T.D. Price. Plenum, New York, in press.
Bettinger, R.L., and M.A. Baumhoff
    1982 The Numic Spread: Great Basin Cultures in Competition. American Antiquity 47: 485-503.
Bettinger, R. L. and J. Eerkens
    1999 Point Typologies, Cultural Transmission, and the Spread of Bow and Arrow Technology in the Prehistoric Great Basin. American Antiquity 64:231-242.
Blumler, M.
    1992 Seed Weight and Environment in Mediterranean-type Grasslands in California and Israel. Ph.D. Thesis, University of California, Berkeley. UMI, Ann Arbor.
Boyd, R. and P.J. Richerson
    1985 Culture and the Evolutionary Process. University of Chicago Press, Chicago.
Boyd, R. and P.J. Richerson
    1992 How Microevolutionary Processes Give Rise to History. In History and Evolution, edited by M.H. and D.V. Nitecki, pp.179-209. State University Press of New York, Albany.
Bradley, R.S.
    1999 Paleoclimatology: Reconstructing Climates of the Quaternary, Second Edition. Academic Press, San Diego.
Braidwood, R.J.
    1960 The Agricultural Revolution. Scientific American 203(3):130-148.
Braidwood, R,J. and B. Howe
    1960 Prehistoric Investigations in Iraqi Kurdistan. Studies in Ancient Oriental Civilization 31. University of Chicago Oriental Institute, Chicago.
Braidwood, L., R. Braidwood, B. Howe, C. Reed, and P.J. Watson (editors)
    1983 Prehistoric Archaeology Along the Zagros Flanks. Studies in Ancient Oriental Civilization 105. University of Chicago Oriental Institute, Chicago.
Bretting P.K. (editor)
    1990 New Perspectives on the Origin and Evolution of New World Domesticated Plants. Economic Botany 44 (Supplement).
Broecker, W.S.
    1997 Thermohaline Circulation, the Achilles Heel of Our Climate System: Will Man-Made CO2 Upset the Current Balance? Science 178:1582-1588.
Broecker, W.S.
    1996 Glacial Climate in the Tropics. Science 272: 1902-1903.
Brush, S.
    1995 In Situ Conservation of Landraces in Centers of Crop Diversity. Crop Science 35:346-353.
Carroll, R.L.
    1997 Patterns and Processes in Vertebrate Evolution. Cambridge University Press, Cambridge.
Cavalli-Sforza, L.L., P. Menozzi, and A. Piazza
    1994 The History and Geography of Human Genes. Princeton University Press, Princeton.
Childe, V.G.
    1951 Man Makes Himself. Watts, London.
Clark, P.U., R.B. Alley, and D. Pollard
    1999 Northern Hemisphere Ice-Sheet Influences on Global Climate Change. Science 286:1104-1111.
Cohen, M.N.
    1977 The Food Crisis in Prehistory: Overpopulation and the Origins of Agriculture. Yale University Press, New Haven.
Cohen, M.N. and G.J. Armelagos
    1984 Paleopathology at the Origins of Agriculture. Academic Press, Orlando.
Crawford, G. W.
    1992 Prehistoric Plant Domestication in East Asia. In The Origins of Agriculture, edited by C. W. Cowan and P. J. Watson, pp. 7-38. Smithsonian Institution Press, Washington, D.C.
Crosby, A.W.
    1986 Ecological Imperialism: The Biological Expansion of Europe, 900-1900. Cambridge University Press, Cambridge.
Crowley, T.J. and G.R. North
    1991 Paleoclimatology. Oxford University Press, New York.
Darwin, C.
    1902 [1874] The Descent of Man and Selection and Selection in Relation to Sex. American Home Library, New York.
Day, R.H. and J-L. Walter
    1989 Economic Growth in the Very Long Run: On the Multiple Phase Interaction of Population, Technology, and Social Infrastructure. In Economic Complexity: Chaos, Sunspots, Bubbles, and Nonlinearity, edited by W.A. Barnett, J. Geweke, and K. Schell, pp. 253-289. Cambridge University Press, Cambridge.
Diamond, J.
    1997 Guns, Germs, and Steel. Norton, New York.
Ditlevsen, P.D., H. Svensmark, and S. Johnsen
    1996 Contrasting Atmospheric and Climate Dynamics of the Last-glacial and Holocene Periods. Nature 379: 810-812.
Durham, W.H.
    1991 Coevolution: Genes, Culture, and Human Diversity. Stanford University Press, Stanford.
Easterlin, R.A.
    1995 Will Raising the Incomes of All Increase the Happiness of All? Journal of Economic Behavior and Organization 27:35-47.
Ehret, C.
    1998 An African Classical Age: Eastern and Southern Africa in World History, 1,000 B.C. to A.D. 400. University of Virginia Press, Charlottesville.
Eldredge, N. and S.J. Gould
    1972 Punctuated Equilibria: An Alternative to Phyletic Gradualism. In Models in Paleobiology, edited by T.J.M. Schopf, pp. 82-115. San Francisco: Freeman.
Elston, R. G., C. Xu, D. B. Madsen, K. Zhong, R. L. Bettinger, J. Li, P. J. Brantingham, H. Wang and J. Yu
    1997 New Dates For the North China Mesolithic. Antiquity 71:985-993.
Endler, J.A.
    1986 Natural Selection In the Wild. Princeton University Press, Princeton.
Fairbank, J.K.
1992 China: A New History. Harvard University Press, Cambridge MA.

Fiedel, S.J.

    1999 Older Than We Thought: Implications of Corrected Dates for Paleoindians. American Antiquity 64:95-115.
Flannery, K.V.
    1973 The Origins of Agriculture. Annual Review of Anthropology 2:271-310.
Frank, R.H. and P.J. Cook
    1995 The Winner-Take-All Society. Free Press, New York.
Frogley, M.R., Tzedakis, P.C., and T.H.E. Heaton
    1999 Climate Variability in Northwestern Greece During the Last Interglacial. Science 285: 886-1889.
Gifford-Gonzales, D.
    In press. Animal Disease Challenges to the Emergence of Pastoralism in Sub-Saharan Africa. African Archaeological Review.
Goring-Morris, N., and A. Belfer-Cohen
    1998 The Articulation of Cultural Processes and Late Quaternary Environmental Change In Cisjordan. Paléorient 23:71-93.
GRIP (Greenland Ice-core Project Members)
    1993 Climate Instability During the Last Interglacial Period Recorded In the GRIP Ice Core. Nature 364:203-207.
Harris, D.R. (editor)
    1996 The Origins and Spread of Agriculture and Pastoralism In Eurasia. University College London Press, London.
Harris, D.R.
    1977 Alternative Pathways Toward Agriculture. In Origins of Agriculture, edited by C.A. Reed, pp. 179-243. Mouton, The Hague.
Hayden, B.
    1995 A New Overview of Domestication. In Last Hunters, First Farmers: New Perspectives on the Prehistoric Transition to Agriculture, edited by T.D. Price and B. Gebauer, pp.273-299. School of American Research Press, Santa Fe.
Henry, D.O.
    1989 From Foraging to Agriculture: The Levant At the End of the Ice Age. University of Pennsylvania Press, Philadelphia.
Imamura, K.
    1996 Jomon and Yoyoi: The Transition to Agriculture in Japanese Prehistory. In The Origins and Spread of Agriculture and Pastoralism in Eurasia, edited by D.R. Harris, pp. 442-464. University College London Press, London.
Johnson, T.C., C.A. Scholz, M.R. Talbot, K. Ketts, R.D. Ricketts, G. Ngobi, K. Buening, I. Ssemmanda and J.W. McGill
    1996 Late Pleistocene Dessication of Lake Victoria and Rapid Evolution of Cichlid Fishes. Science 273: 1091-93.
Lamb, H.H.
    1977 Climatic History and the Future. Princeton University Press, Princeton.
Lee, R.D.
    1986 Malthus and Boserup: A Dynamic Synthesis. In The State of Population Theory: Forward From Malthus, edited by D. Coleman and R. Schofield, pp. 96-130. Basil Blackwell, Oxford.
Levinton, J.
    1988 Genetics, Paleontology, and Macroevolution. Cambridge University Press, Cambridge.
Lindert, P.H.
    1985 English Population, Wages, and Prices: 1541-1913. Journal of Interdisciplinary History 15: 609-34.
Katz, S.H., M.L. Hediger, and L.A. Valleroy
    1974 Traditional Maize Processing Techniques in the New World. Science 184: 765-773.
Kent J.D.
    1987 Periodic Aridity and Prehisanic Titicaca Basin Settlement Patterns. In Arid Land Use Strategies and Risk Management in the Andes, edited by D.L. Browman, pp 297-314. Westview, Boulder. .
Kerr, R.A.
    1999 Will the Arctic Ocean Lose All Its Ice? Nature 286: 128.
Klein, R. G.
    1999 The Human Career: Human Biological and Cultural Origins. University of Chicago Press, Chicago.
Kirch, P.V.
    1984 The Evolution of Polynesian Chiefdoms. Cambridge University Press, Cambridge.
Lindsay, L. M.
    1986 Fremont Fragmentation. In Anthropology of the Desert West: Essays in Honor of Jesse D. Jennings, edited by C. J. Condie and D. D. Fowler, pp. 229-252. Anthropological Papers No. 110. University of Utah, Salt Lake City.
MacNeish, R.S.
    1973 The Evolution of Community Patterns in the Tehuacán Valley of Mexico and Speculations About the Cultural Process. In Ecology and Agricultural Settlements: An Ethnographic and Archaeological Perspective, edited by R. Tringham, pp. 43-55. Andover MA: Warner Modular Publications, Andover MA.
MacNeish, R.S.
    1991 The Origins of Agriculture and Settled Life. University of Oklahoma Press: Norman.
Matson, R. G.
    1999 The Spread of Maize to the Colorado Plateau. Archaeology Southwest 13:10-11.
McNeill, W.H.
    1976 Plagues and Peoples. Anchor, Garden City NY.
Martin, P.S., and R.G. Klein (editors)
    1984 Quaternary Extinctions: A Prehistoric Revolution. University of Arizona Press, Tucson.
Mayewski, P. A., M. L. D., W. S, M. S. Twickler, M. C. Morrison, R. B. Alley, P. Bloomfield and K. Taylor
    1993 The Atmosphere During the Younger Dryas. Science 261:195-100.
Murray, J. D.
    1989 Mathematical Biology. Springer-Verlag, Berlin.
North, D.C. and R.P. Thomas
    1973 The Rise of the Western World: A New Economic History. Cambridge University Press, Cambridge.
Partridge, T.C., G.C. Bond, C.J.H. Hartnady, P.B. deMenocal, and W.F. Ruddiman
    1995 Climatic Effects of Late Neogene Tectonism and Vulcanism. In Paleoclimate and Evolution With Emphasis on Human Origins, edited by E. S. Vrba, G. H. Denton, T. C. Partridge, L. H. Burckle, pp. 8-23. Yale University Press, New Haven.
Petit, J.R., J. Jouzel, D. Raynaud, N.I. Barkov, J.-M. Barnola, J. Basile, M. Bender, J. Cappellaz, M. Davis, G. Delaygue, M. Delmotte, V.M. Kotlyakov, M. Legrand, V.Y. Lipenkov, C. Lorius, L. Pépin, C. Ritz. E. Saltzman, and M. Stievenard
    1999 Climate and Atmospheric History of the Past 420,000 Years From the Vostok Ice Core, Antarctica. Nature 399:429-436.
Price, T.D. and B. Gebauer
    1995 Last Hunters, First Farmers: New Perspectives on the Prehistoric Transition to Agriculture. School of American Research Press, Santa Fe.
Richerson, P.J. and R. Boyd
    1998 The Evolution of Human Ultrasociality. In Indoctrinability, Ideology, and Warfare: Evolutionary Perspectives, edited by I. Eibl-Eibesfeldt and F.K. Salter, pp. 71-95. Berghahn, New York.
Richerson, P.J. and R. Boyd
    1999 Complex Societies: The Evolutionary Origins of a Crude Superorganism. Human Nature 10:253-289.
Richerson, P.J. and R. Boyd.
    2000 Climate, Culture, and the Evolution of Cognition. In Evolution of Cognition, edited by C. Heyes and L. Huber. MIT Press, Cambridge MA, in press.
Rindos, D.
    1984 The Origins of Agriculture: An Evolutionary Perspective. Academic Press, London.
Rogers, E.M.
    1983 Diffusion of Innovations, 3rd Edition. Free Press, New York.
Sage, R.F.
    1995 Was Low Atmospheric CO2 During the Pleistocene a Limiting Factor For the Origin of Agriculture? Global Change Biology 1:93-106.
Simpson, G.G.
    1944 Tempo and Mode in Evolution. Columbia University Press, New York.
Smith, B.D.
    1989 Origins of Agriculture in Eastern North America. Science 246: 1566-1571.
Smith, B.D.
    1995 The Emergence of Agriculture. Scientific American Library, New York.
Smith, B. D.
    1997 Initial Domestication of Curcurbita pepo in the Americas 10,000 Years Ago. Science 276: 932-934.
Steward, J.H.
    1955 Theory of Culture Change: The Methodology of Multilinear Evolution. University of Illinois Press, Urbana.
Steig, E.J., E.J. Brook, J.W.C. White, C.M. Sucher, M.L. Bender, S.J. Lehman, D.L. Morse, E.D. Waddington, and G.D. Glow
    1998 Synchronous Climate Changes in Antarctica and the North Atlantic. Science 282: 92-95.
Stuiver, M., P. J. Reimer, E. Bard, J. W. Beck, G. S. Burr, K. A. Hughen, B. Kromer, G. McCormack, J. V. D. Plicht and M. Spurk
    1998 INTCAL98 Radiocarbon Age Calibration, 24,000 - 0 cal BP. Radiocarbon 40:1041-1083.
Tainter, J. A.
    1988 The Collapse of Complex Societies. Cambridge University Press, Cambridge.
Valentine, J.W. and E.M. Moores
    1970 Plate-Tectonic Regulation of Faunal Diversity and Sea Level: A Model. Nature 228: 657-659.
Vayda, A.P.
    1995 Failures of Explanation in Darwinian Ecological Anthropology: Part II. Philosopy of the Social Sciences 25: 360-375.
Von Grafenstein, U., H. Erlenkeuser, A. Brauer, J. Jouzel, and S.J. Johnsen
    1999 A Mid-European Decadal Isotope-Climate Record From 15,500 to 5,000 years B.P. Science 284: 1654-1657.
Wohlgemuth, E.
    1996 Resource Intensification In Prehistoric Central California: Evidence From Archaeobotanical Data. Journal of California and Great Basin Archaeology 18: 81-103.
Walker, T.D. and J.W. Valentine
    1984 Equilibrium Models of Evolutionary Species Diversity and the Number of Empty Niches. American Naturalist 124: 887-899.
Wiessner, P., and A. Tumu
    1998 Historical Vines: Enga Networks of Exchange, Ritual, and Warfare in Papua New Guinea. Smithsonian Institution Press, Washington D.C.
Winterhalder, B.
    1986 Diet Choice, Risk, and Food Sharing in a Stochastic Environment. Journal of Anthropological Archaeology 5:369-392.
Woodburn, J.
    1980 Hunter-Gatherers Today and Reconstruction of the Past. In Soviet and Western Anthropology, edited by E. Gellner, pp. 95-117. Columbia University Press, New York.
Wright, H.E., Jr.
    1977 Environmental Change and the Origin of Agriculture In the Old and New Worlds. In Origins of Agriculture, edited by C.A. Reed, pp. 281-318. Mouton, The Hague.
Zohary, D., and M. Hopf
    1993 Domestication of Plants in the Old World: The origin and Spread of Cultivated Plants in West Asia, Europe, and the Nile Valley, 2nd Edition. Oxford University Press, Oxford.
Zvelebil, M.
    1996 The Agricultural Frontier and the Transition to Farming in the Circum-Baltic Region. In The Origins and Spread of Agriculture and Pastoralism in Eurasia, edited by D. R. Harris, pp.323-345. Smithsonian Institution Press, Washington, D.C.

TABLE 1. Dates Before Present in calendar years of achievement of plant intensive hunting and gathering and agriculture in different regions. Mainly after Smith (1995). 
Intensive Foraging
Centers of Domestication
 Near East
 North China
> 9,000
 South China
 Sub-Saharan Africa
 South Central Andes
 Central Mexico
 Eastern United States
Controversial Centers
 Highland New Guinea
Acquisition by Diffusion
 North Western Europe
 South Western U.S.
Never Acquired Agriculture

TABLE 2. World Distribution of Large-Seeded Grass Species (From Diamond, 1997, after Blumler, 1992)




 Number of species

West Asia, Europe, North Africa
 Mediterranean zone
East Asia
Sub-Saharan Africa
 North America
 South America
Northern Austalia

Figure 1. Profiles of a temperature index, d18O, and a index of dust content, Ca2+, from the GRIP Greenland ice core. 200 year means are plotted. Histograms show how much noisier the last glacial and perhaps last interglacial were compared to the Holocene. The parts of the GRIP profile representing the last interglacial (MIS 5e) may have been affected by ice flow so their interpretation is uncertain. (GRIP, 1993, copyright © 1993 Macmillan Magazines Limited.)

Figure 2. High resolution analysis of the GRIP ice core d18O data. The low pass filtered data shows that the Holocene is much less variable than the Pleistocene on time scales of 150 years and longer. The high pass filtered data shows that the Pleistocene was also much more variable on time scales less that 150 years. Since diffusion increasingly affects deeper parts of the core by averaging variation on the smallest scales, the high pass variance is reduced in the older parts of the core. In spite of this effect, the Pleistocene/Holocene transition is very strongly marked. (Ditlevsen, et al., 1996, copyright © 1993 Macmillan Magazines Limited.)

Figure 3. Panel a shows the curve of atmospheric CO2 as estimated from gas bubbles trapped in Antarctic glacial ice. Data from Barnola et al. (1987). Panel b summarizes responses of several plant species to experimental atmospheres containing various levels of CO2.  Based on data summarized by Sage (1995).

Figure 4. A numerical simulation of Fisher’s equation showing that after an initial period, population spreads at a constant rate so that at any point in space population pressure increases to its maximum in less than 500 years for reasonable parameter values.(Redrawn from Ammerman and Cavalli-Sforza, 1984).

Figure 5 plots the logarithm of population size as a function of time for the model described in the text. Initially, when there is little population pressure, population grows at a high rate. As the population grows, per capita income decreases, and people intensify. Eventually the population growth rate approaches a constant value at which the growth of intensification balances growth in population. For reasonable parameters (a = 0.005, r = 0.02, ym = 1, ys = 0.1, yi = 0.2, initial population size 1% of initial carrying capacity), it takes less than 500 years to shift from the initial low population pressure mode of growth to the final high population pressure mode of growth.