Volume 46, Supplement, December 2005 F S109

New World Settlement Evidencefor a Two-Stage NeolithicDemographic Transition1

matthew s. bandyDepartment of Anthropology, University ofOklahoma, 455 W. Lindsey, Dale Hall Tower 521,Norman, OK 73019-0535, U.S.A. (inti@devnull.net). 17v 05

The adoption of an agricultural village lifeway roughlycoincided with a great increase in the absolute numberof humans and in the size of human communities. Thisincrease in the growth rate of human populations hasbeen called the Neolithic demographic transition (Boc-quet-Appel 2002:637). It is one of the fundamental struc-turing processes of human history, and it is thereforehighly relevant to the comparative study of cultural andsocial evolution at a global scale.

Albert Ammerman and L. Luca Cavalli-Sforza (1971),in their original formulation of the demic diffusion hy-pothesis, provided a model for the Neolithic demo-graphic transition. In their scenario, the transition to anagricultural economy led to an increased net reproduc-tive rate, lasting for some generations until a new plateauof population density consonant with the carrying ca-pacity of the land was reached (p. 687). The transitionwas therefore, in their view, a two-stage process. In thefirst stage, reduced mortality and/or increased fertilityproduced rapid population growth. In the second stage,limitations of the natural environment produced a lowerpopulation growth rate. In subsequent publications, Am-merman and Cavalli-Sforza (1984; Ammerman, Cavalli-Sforza, and Wagener 1987) have argued that both the tim-ing of the onset of the Neolithic in Europe andcontinental-scale genetic distribution patterns are con-sistent with this model. However, evidence for a two-stage demographic transition has remained indirect. Inprinciple, the observed archaeological and genetic pat-

2005 by The Wenner-Gren Foundation for Anthropological Re-search. All rights reserved 0011-3204/2005/4605S5-0007$10.00

1. The survey of the Taraco Peninsula was carried out with thesupport of the National Science Foundation (BCS-9813395), theWenner-Gren Foundation for Anthropological Research, and a Ful-bright-Hays fellowship from the United States Department of Ed-ucation. I thank Christine Hastorf, Amanda Cohen, Robin Beck,Erika Simborth, Lee Steadman, William Whitehead, Kirk Frye, JohnJanusek, Carlos Lemuz, Richard Wilshusen, Pilar Lima, FelipeChoque, Facundo Llusco, Eduardo Pareja, Andrew Roddick, and Mi-chael Wilkes for help either in the fieldwork phase of the projector in the formulation of the ideas presented in this paper. StephenKowalewski graciously provided me with an electronic copy of theOaxaca settlement data. I acknowledge support during the prepa-ration of this article from the Cotsen Institute of Archaeology atthe University of California, Los Angeles, and from the Universityof Oklahoma Department of Anthropology.

terns could have been produced by other kinds of his-torical processes.

Bocquet-Appel (2002) has recently reconstructed pop-ulation growth rates from standardized age profiles in asample of Mesolithic and Neolithic cemeteries from Eu-rope and North Africa. Though the results are tentativeand suffer from limited sample size, especially for theMesolithic, Bocquet-Appel reasonably concludes that hisdata support a two-stage demographic transition coin-cident with the transition to an agricultural economy.After arranging his data according to chronological dis-tance from the Neolithic diffusion front in such a waythat zero represents the beginning of the Neolithic inany given region, he identifies three demographic periods(pp. 64445). In the first, prior to the transition to agri-culture, estimated population growth is stationary ornegative. This is a difficult point in the argument, asBocquet-Appel acknowledges, because of the limitednumber and distribution of the Mesolithic cemeteries inthe sample. In the second period, covering the firstroughly 800 years of the Neolithic, the populationgrowth rate increased dramatically, to an average of ap-proximately 1.24% per annum. In the third, the rate ofgrowth decreased somewhat, to an average of 1.16% perannum. The confidence intervals associated with thesegrowth rate estimates are very wide ( 1.07%), and forthis reason we should not attempt comparisons of ab-solute growth rates between world regions. However, thedata are sufficient to indicate that the Neolithic demo-graphic transition in Europe was in fact a two-stage pro-cess as the demic diffusion model suggests. The rate ofpopulation growth increased significantly in the firsteight centuries or so of the Neolithic and declinedthereafter.

Many questions remain. Perhaps the most importantof these concerns the generality of the two-stage de-mographic transition. Bocquet-Appel suggests that thevery high early Neolithic population growth rate reflectsan increase in fertility associated with the transition toagriculture. This is reasonable, considering that agricul-turalists do in fact have higher average fertility rates thanothers (Bentley, Jasienska, and Goldberg 1993). Further,direct paleodemographic evidence has recently been pub-lished demonstrating a substantial increase in fertilitywith the beginning of the Neolithic in the Levant (Eshedet al. 2004). He further hypothesizes that population den-sity increased as the Neolithic wore on and that thepromiscuity of humans and animals and the anastomoticprocess of village populations (Bocquet-Appel 2002:647)resulted in a disease-related mortality increase and a con-sequent decline in the rate of population growth. Hetherefore disagrees with Ammerman and Cavalli-Sforzas original hypothesis that the decline in popula-tion growth (the second stage of the two-stage transition)was the result of the populations having reached thecarrying capacity of the land, arguing rather that it re-sulted from the operation of density-dependent effectsin Neolithic village populations. This hypothesis haselsewhere been referred to as the first epidemiologicaltransition (Barrett et al. 1998).

S110 F current anthropology

It is important to recognize that none of the factorsentering into Bocquet-Appels hypothetical reconstruc-tion of European Neolithic demographic events are spe-cific to Europe. While every local Neolithic transition isno doubt historically unique, all are characterized by thetwo essential ingredients of Bocquet-Appels model:year-round occupation of permanent villages (sedentism)and a substantial degree of dependence on food produc-tion for subsistence (agriculture). It is therefore unim-portant for this discussion that village life preceded ag-riculture in some parts of Europe (Zvelebil and Lillie2000, Zvelebil and Rowley-Conwy 1986) or that cropdomestication preceded village life in many parts of theNew World (Flannery, Moser, and Maranca 1986, Pearsall1992, Smith and Cowan 2003). Bocquet-Appels modelshould in principle be equally applicable to the Neolithictransitions of the two world regions. In this article I usethree archaeological settlement pattern data sets to eval-uate the hypothesis that the transition to agriculturalvillage life was associated with a two-stage demographictransition in the New World.

analysis of selected settlement data sets

In the millennium following the transition to agricul-tural village life, the hypothesis of a two-stage demo-graphic transition predicts that the earlier half of thatmillennium will be characterized by relatively high pop-ulation growth rates and the later half by substantiallyslower growth. In order to test this hypothesis, I willexamine population growth rates during the first mil-lennium of the Formative period of three areas: (1) theBasin of Mexico, (2) the Valley of Oaxaca, and (3) theTiticaca Basin. Before proceeding, however, it remainsto address the question of how to monitor populationand population growth using archaeological data.

As Bocquet-Appel (2002:637) has observed, there aretwo broad classes of data that permit direct monitoringof prehistoric population growth rates: bioarchaeological(the age structure of mortuary populations) and archae-ological. In this article, I take an archaeological approachto measuring prehistoric population growth not only be-cause it is my area of specialization but also becauselarge, well-analyzed, and well-published mortuary sam-ples of the kind employed by Bocquet-Appel in his anal-ysis are a relative rarity in most of the New World andbecause settlement archaeology is relatively well-devel-oped in the New World and a number of excellent datasets are available that span the first millennium follow-ing the transition to agricultural village life.

There are many archaeological methods for estimatingthe population of sites. These include site area (Bandy2001, Feinman et al. 1985, Stanish et al. 1997), combinedsite area and surface artifact density (Parsons 1971, San-ders 1972), number of structures or rooms (Adler 1992,Hill 1970, de Montmollin 1989), roofed area (Naroll1962), and measures of refuse accumulation (Gallivan2002, Varien and Mills 1997, Varien and Potter 1997). Allof the data sets employed in the present analysis providepopulation estimates of all sites for each phase of their

occupation. For the present analysis I will use publishedsite population estimates for three regions, all of whichare derived from some combination of site area and sur-face artifact density. However, in order to estimate re-gional population growth rates it is necessary to derivea measure of regional population that can be monitoredover time. I call this measure the phase population in-dex. The phase population index utilized here will bethe simple sum of estimated population for all sites ina region during a given phase.

The most significant complication in the use of thephase population index concerns the problem of site con-temporaneity. Sanders (1972:102) expressed the problemin this way:

Most researchers deal in blocks of time involvinghundreds of years. Conceivably in the history of anarea, there could have been a period in which settle-ments were less sedentary and village sites were fre-quently shifted, followed by a period in which peo-ple lived in more stable communities. In this case asimple calculation of the total amount of habitationarea or number of ruined houses of the various sitesfrom each period would not give an accurate pictureeven of the relative population size of the twoperiods.

The percentage of sites which are occupied simulta-neously in a given phase will vary from phase to phaseand between different types of sites within a single phase.In stable settlement systems composed of small numbersof relatively large, permanent sites, the phase populationindex will closely correspond to the actual population ofa phase at its maximum. However, in a phase charac-terized by small sites occupied on a short-term basis theindex will significantly overestimate that population.The phase population index as defined here will thereforeexaggerate the contribution of small and ephemerallyoccupied components of a total settlement system rel-ative to sedentary and continuously occupied compo-nents. In other words, farmsteads will count for rela-tively more than villages, town, and cities. It will alsotend to exaggerate the population estimates for phasescharacterized by predominantly transitory settlementvis-a`-vis more settled or urban periods.

Two pertinent observations can therefore be made re-garding the phase population index. First, as long as thedegree of sedentism and site permanence remains aboutthe same between phases the phase population indexshould vary directly with actual population; it will al-ways be some constant multiple of maximum momen-tary population for the phase. For the calculation of pop-ulation growth rates, the accuracy of the phasepopulation index is not as important as its precision. Itmay yield an entirely incorrect population estimate foreach phase, but as long as it does so consistently it isstill possible to calculate the population growth rate.Since I am considering here only the first millennium ofeach sequence, it is possible to say that no dramatic set-tlement transformations (such as, for example, typically

Volume 46, Supplement, December 2005 F S111

table 1Phase Names and Absolute Date Ranges (BCE) for Selected Settlement Data Sets

Basin of Mexico Valley of Oaxaca Titicaca Basin

Early Horizon 1300800 Tierras Largas 15001150 Early Chiripa 15001000First Intermediate 1 800500 San Jose 1150850 Middle Chiripa 1000800First Intermediate 2 500200 Guadalupe 850700 Late Chiripa 800250

Rosario 700500

table 2Demographic Information by Phase, Basin of Mexico

Phase Sites Hectares PopulationGrowthRate (%)

Early Horizon 7 34.4 685 First Intermediate 40 297.5 6,314 0.74Second Intermediate 103 1,152.2 30,790 0.53

accompany episodes of state formation) took place in anyof the three areas. Rates of occupation continuity arequite high in all three areas, and the majority of thepopulation resided in large, permanent villages that wereoften occupied for many centuries and even millennia.Under these conditions the phase population indexshould yield useful results.

Second, assuming continual settlement expansion andpopulation growth, the phase population index measurespopulation at the end of the phase, the time of maximummomentary population. This assumption holds for allthree of the areas to be considered for the first millen-nium following the transition to agricultural village life.In each case, then, the end of each phase may be con-sidered a point in time for which we have an acceptableregional population estimate.

Considering all of the above, the annual growth rateduring a given phase may therefore be calculated usingthe formula (Hassan 1981:139)

1 Pfr p ln ,

T Pi

where r p annual rate of population change, T p numberof years in phase, Pi p phase population index value ofprevious phase, and Pf p phase population index valueof phase. Bocquet-Appels hypothesis makes two predic-tions: (1) the transition to agriculture will coincide witha dramatic increase in the rate of population growth and(2) the rate of population growth will decrease signifi-cantly sometime in the second half of the first millen-nium following the transition. A limitation shared by allthree areas to be examined is a lack of reliable data onpreagricultural population levels or growth rates. At pre-sent, it is not possible to calculate a population growthrate for the first phase in each sequence. I will thereforebe able to test only the second prediction in each area.

The Basin of Mexico. The Basin of Mexico was the siteof perhaps the first really ambitious archaeological set-tlement-pattern project. Between 1967 and 1973 JeffreyParsons and Richard Blanton conducted a pedestrian sur-vey of some 3,500 square kilometers, recording thous-ands of sites from all time periods of the regions pre-history (Blanton 1972; Parsons 1971, 1976; Parsons et al.1982; Parsons, Kintigh, and Gregg 1983; Sanders, Par-sons, and Santley 1979). A settled agricultural villagesystem is well-documented for the southern Basin ofMexico by the Early Horizon, around 1300 BCE (Blantonet al. 1981, Parsons 1974). In the earlier part of the Pre-

classic population was very heavily concentrated in thesouthern parts of the basin, with no substantial occu-pation of the north until after 500 BCE (Parsons 1974:9193).

The early settlement history of the Basin of Mexico isproblematic for several reasons (Parsons 1974:91). Theprimary reason is the recent expansion of Mexico City.Urban sprawl has resulted in the destruction of manysites, among them Tlatilco, probably a major Early Ho-rizon center. In addition, the area surrounding the siteof Cuicuilco, also probably a major early center, was cov-ered by a lava flow sometime in the first or second cen-tury BCE that obscured the bulk of its associated habi-tation. It is problematic to estimate population andtherefore population growth in the absence of reliableinformation on the size or density of what were possiblythe two most populous settlements in the Basin of Mex-ico. The Preclassic settlement record of the Basin of Mex-ico is fundamentally and irremediably incomplete. Nev-ertheless, the majority of habitation sites do appear tohave been preserved and recorded. If a two-stage demo-graphic transition took place, it may be evident in thesettlement record.

The phases we are concerned with in this analysis arethe Early Horizon and First Intermediate 12 phases (Par-sons 1974). The absolute date ranges for these phases aregiven in table 1. Table 2 displays some basic informationregarding the settlement system in each phase. I havecalculated a phase population estimate for each of thethree phases in question and calculated the average an-nual growth rate during each phase, assuming that thephase population estimate represents the terminal pop-ulation of the phase. When these data were published(Parsons, Kintigh, and Gregg 1983), high and low popu-lation estimates were provided for each site for eachphase. I have simply averaged these two numbers to ar-rive at a population estimate for the occupation. A pop-ulation index value for each phase (table 2) was calcu-

S112 F current anthropology

table 3Demographic Information by Phase, Valley of Oaxaca

Phase Sites Hectares PopulationGrowthRate (%)

Tierras Largas 24 14.8 325 San Jose 39 104.2 1,937 0.60Guadalupe 44 95.8 1,782 0.06Rosario 83 92.1 1,828 0.01

table 4Demographic Information by Phase, Titicaca Basin

Phase Sites Hectares PopulationGrowthRate (%)

Early Chiripa 9 14.2 693 Middle Chiripa 23 35.2 1,752 0.46Late Chiripa 31 65.8 3,507 0.13

lated by summing the population estimates for alloccupations pertaining to that phase.

Indeed, the Basin of Mexico data are consistent withthe two-stage model (table 2). In the First Intermediate1 population grew at approximately 0.74% per annum,a very high rate for an agricultural village society withoutaccess to antibiotics or modern medicine. In the follow-ing First Intermediate 2, after about 500 BCE, the rateof population growth fell to 0.53% annually, still quitea high rate but a considerable drop from the previousphase. Thus, the rate of population growth declined ap-proximately 800 years after the establishment of per-manent agricultural villages. This falls precisely into thetime frame predicted by Bocquet-Appels hypothesis.

The Valley of Oaxaca. The Valley of Oaxaca is anotherNew World area that has benefited from archaeologicalsettlement research on a massive scale. By 1985 almostall of this large highland valley, an area of some 2,150square kilometers, had been archaeologically surveyed(Feinman et al. 1985:333), and these data have been pub-lished in their entirety (Blanton et al. 1982, Kowalewskiet al. 1989). The present analysis is based on the datapublished by Kowalewski et al. (1989:appendix 1). Thesedata are somewhat more robust for our present interpre-tive purposes than the Basin of Mexico data. The Valleyof Oaxaca has suffered little from the effects of recenturbanization. The settlement record of this region re-mains largely intact, and it is probable that no significantpre-Hispanic population concentrations remain unre-corded. A two-stage demographic transition shouldtherefore be readily apparent. As with the Basin of Mex-ico data, this data set provided high and low populationestimates for each occupation of each recorded site. Forthe present analysis I have simply averaged the two inorder to arrive at a population estimate for each site. Myphase population estimates (table 3) are the simple sumof the average population estimates for all occupationsdating to each phase.

Agricultural village life was well-established in theValley of Oaxaca by the beginning of the Tierras Largasphase, at 1400 BCE (Flannery 1983, Marcus and Flannery1996). The following millennium included the TierrasLargas, San Jose, Guadalupe, and Early Rosario ceramicphases (table 1). Before the Rosario phase, settlement wasconcentrated in the Etla arm of the valley; the remainderwas relatively vacant (Feinman et al. 1985). The Valleyof Oaxaca data are consistent with the model of a two-stage demographic transition. Population grew duringthe San Jose phase at the rapid rate of 0.60% annually.After about 850 BCE, in the following Guadalupe andRosario phases population growth fell to almost zero orwas even slightly negative.2 This drop in the growth ratetook place approximately 650 years after the transition

2. These growth rates differ slightly from the ones published byFeinman and colleagues (1985:346), who report growth rates of0.59%, 0.03%, and 0.02%. This is because I use a slightly differentmethod of calculating phase population and because I calculategrowth from the end of one phase to the end of another while theycalculate it between the temporal midpoints of each phase (p.346).

to agriculture at the beginning of the Tierras Largas phase(fig. 1), exactly as predicted by the model.

The Titicaca Basin. The Lake Titicaca Basin of Peruand Bolivia is one of the nuclear areas for village andlater state formation South America (Bermann 1990, Jan-usek 1994, Kolata 1993, Ponce Sangines 1981, Stanish2003). Permanent agricultural villages appeared in thebasin around 1500 BCE, about the same time as in Oa-xaca. Though a significant amount of settlement re-search has taken place in this region in the past 20 years(Albarracn-Jordan 1992, 1996; Albarracn-Jordan andMathews 1990; Janusek and Kolata 2003; Mathews 1992;Stanish et al. 1997), problems of ceramic chronology havelimited the chronological resolution of Formative-periodsettlement data sets. Lee Steadmans studies (Hastorf etal. 2001, Steadman 1999) of Early and Middle Formativeceramics have only recently made possible the identi-fication and distinction of these early ceramic phases insurface assemblages (see table 1). In 199899 I conducteda survey of the Taraco Peninsula, the first survey to em-ploy Steadmans improved chronology (Bandy 2001,2004). The resulting settlement data set is therefore theonly Titicaca Basin settlement data set useful for thepresent analysis.

The area surveyed was 98 square kilometers, a verysmall area compared with the Basin of Mexico and Valleyof Oaxaca surveys. The density and size of sites pertain-ing to the earliest ceramic phases were however, veryhigh. An inspection of table 4 will reveal that the TaracoPeninsula, though covering less than 5% of the surfacearea of the Valley of Oaxaca survey, nevertheless hassimilar overall population estimates for these earlyphases. It is important to remember that in both theBasin of Mexico and the Valley of Oaxaca population washighly localized in the earliest ceramic phases. Thisseems to be a general pattern in early agricultural villagesocieties in many parts of the world. Therefore, havingsurveyed the Taraco Peninsula is similar to having sur-veyed the Etla subvalley of Oaxaca or the Chalco/Xoch-

Volume 46, Supplement, December 2005 F S113

Fig. 1. Population growth rate in relation to the transition to agriculture. 0, the beginning of the Formativeperiod in each area.

imilco region of the Basin of Mexico; most of the earlyagricultural village population was located in this rela-tively small area. This is confirmed by the much lowerdensity of early occupation in the adjacent surveyed ar-eas of the Tiwanaku Valley and the Katari Basin (Albar-racn-Jordan 1992, Janusek and Kolata 2003, Mathews1992).

The Taraco Peninsula data, like those from the Basinof Mexico and the Valley of Oaxaca, are consistent withthe two-stage model. Population grew at a relativelyrapid rate of 0.46% in the Middle Chiripa phase anddropped to 0.13% during the Late Chiripa phase. Thisdrop in the rate of population growth took place at 800BCE, approximately 700 years after the transition to ag-riculture in the southern Titicaca Basin, precisely as pre-dicted by the model.

conclusion

All three of the New World archaeological settlementdata sets examined are consistent with Bocquet-Appelshypothesized two-stage demographic transition. Figure1 displays the estimated population growth rates of allthree areas in reference to the date of the transition toagriculture and village life. In all three cases, an initialvery high rate of population growth is followed by a dropin the growth rate that occurs 600800 years after what

in the Old World would be called the Neolithic transi-tion. The profiles for the two better-documented areas(Oaxaca and the Titicaca Basin) are very similar even intheir absolute rates of population growth. These resultsare exactly as predicted by Bocquet-Appels model, andall of the growth rates in question fall well within theconfidence intervals of those reconstructed by Bocquet-Appel (2002:fig. 5) for the European Neolithic.

It is true that the settlement data sets examined hererepresent relatively small and widely scattered samplesand cannot necessarily be considered representative ofall New World agricultural village traditions. However,this study is significant for two reasons: first, it has pro-vided positive evidence that Bocquet-Appels two-stageNeolithic demographic transition took place in the NewWorld Formative as well as the European Neolithic; sec-ond, it has done so using an independent class of data.Both European bioarchaeological data and Latin Amer-ican settlement data suggest that a two-stage demo-graphic transition may be associated with transitions toagricultural village life worldwide.

References Cited

a d l e r , m . a . 1992. The upland survey, in The SandCanyon Archaeological Project: A progress report. Edited by W.

S114 F current anthropology

D. Lipe, pp. 1123. Crow Canyon Archaeological Center Occa-sional Papers 2.

a l b a r a c n - j o r d a n , j . 1992. Prehispanic and early colonialsettlement patterns in the Lower Tiwanaku Valley, Bolivia.Ph.D. diss., Southern Methodist University, Dallas, Tex.

. 1996. Tiwanaku: Arqueologa regional y dinamica seg-mentaria. La Paz: Editores Plural.

a l b a r r a c n - j o r d a n , j . v. , a n d j . e . m a t h e w s .1990. Asentamientos prehispanicos del Valle de Tiwanaku.Vol. 1. La Paz: Producciones CIMA.

a m m e r m a n , a . j . , a n d l . l . c a v a l l i - s f o r z a . 1971.Measuring the rate of spread of early farming in Europe. Man6:67488.

. 1984. The Neolithic transition and the genetics of popu-lations in Europe. Princeton: Princeton University Press.

a m m e r m a n , a . j . , l . l . c a v a l l i - s f o r z a , a n d d . k .w a g e n e r . 1987. Toward the estimation of populationgrowth in Old World prehistory, in Demographic anthropol-ogy: Quantitative approaches. Edited by E. Zubrow, pp. 2761.Albuquerque: University of New Mexico Press.

b a n d y, m . 2001. Population and history in the ancient Titi-caca Basin. Ph.D. diss., University of California, Berkeley, Ca-lif. http://andean.kulture.org/bandy/bandy-dissertation.pdf.

. 2004. Fissioning, scalar stress, and social evolution inearly village societies. American Anthropologist 106:32233.

b a r r e t t , r . , c . w. k u z a w a , t . m c d a d e , a n d g . j .a r m e l a g o s . 1998. Emerging and re-emerging infectious dis-eases: The third epidemiological transition. Annual Review ofAnthropology 27:24771.

b e n t l e y, g . r . , g . j a s i e n s k a , a n d t . g o l d b e r g .1993. Is the fertility of agriculturalists higher than that of non-agriculturalists? current anthropology 34:77885.

b e r m a n n , m . 1990. Prehispanic household and empire at Lu-kurmata, Bolivia. Ph.D. diss., University of Michigan, Ann Ar-bor, Mich.

b l a n t o n , r . e . 1972. Prehispanic settlement patterns of theIxtapalapa Peninsula region, Mexico. Department of Anthro-pology, The Pennsylvania State University, Occasional Papersin Anthropology 6.

b l a n t o n , r . , s . k o w a l e w s k i , g . f e i n m a n , a n d j .a p p e l . 1981. Ancient Mesoamerica: A comparison of changein three regions. Cambridge: Cambridge University Press.

. 1982. Monte Albans hinterland. Pt. 1. The prehispanicsettlement patterns of the central and southern parts of theValley of Oaxaca, Mexico. Memoirs of the Museum of Anthro-pology. University of Michigan, 15.

b o c q u e t - a p p e l , j . - p . 2002. Paleoanthropological traces of aNeolithic demographic transition. current anthropology 43:63750.

d e m o n t m o l l i n , o . 1989. The archaeology of politicalstructure: Settlement analysis in a Classic Maya polity. Cam-bridge: Cambridge University Press.

e s h e d , v. , a . g o p h e r , t . b . g a g e , a n d i . h e r s h k o -v i t z . 2004. Has the transition to agriculture reshaped the de-mographic structure of prehistoric populations? New evidencefrom the Levant. American Journal of Physical Anthropology124:31529.

f e i n m a n , g . m . , s . a . k o w a l e w s k i , l . f i n s t e n , r .e . b l a n t o n , a n d l . n i c h o l a s . 1985. Long-term demo-graphic change: A perspective from the Valley of Oaxaca, Mex-ico. Journal of Field Archaeology 12:33362.

f l a n n e ry, k . 1983. The Tierras Largas phase and the analyt-ical units of the Early Oaxacan village, in The cloud people:Divergent evolution of the Zapotec and Mixtec civilizations.Edited by K. V. Flannery and J. Marcus, pp. 4345. New York:Academic Press.

f l a n n e ry. k . , c . m o s e r , a n d s . m a r a n c a . 1986.The excavation of Guila Naquitz, in Guila Naquitz: Archaicforaging and early agriculture in Oaxaca, Mexico. Edited by K.Flannery, pp. 6595. NewYork: Academic Press.

g a l l i v a n , m . d . 2002. Measuring sedentariness and settle-ment population: Accumulations research in the Middle Atlan-tic region. American Antiquity 67:53557.

h a s s a n , f . 1981. Demographic archaeology. New York: Aca-demic Press.

h a s t o r f , c . a . , m . s . b a n d y, w. t . w h i t e h e a d ,a n d l . h . s t e a d m a n . 2001. El perodo formativo en Chi-ripa, Bolivia. Textos Antropologicos 13(12):1792.

h i l l , j . n . 1970. Broken K Pueblo: Prehistoric social organiza-tion in the American Southwest. University of Arizona An-thropological Papers 18.

j a n u s e k , j . 1994. State and local power in a prehispanic An-dean polity: Changing patterns of urban residence in Tiwanakuand Lukurmata. Ph.D. diss., University of Chicago, Chicago,Ill.

j a n u s e k , j . , a n d a . k o l a t a . 2003. Prehispanic rural his-tory in the Katari Valley, in Tiwanaku and its hinterland,vol. 2, Urban and rural archaeology. Edited by A. Kolata, pp.12974. Washington, D.C.: Smithsonian Institution Press.

k o l a t a , a . 1993. The Tiwanaku. Cambridge: Blackwell.k o w a l e w s k i , s . a . , g . m . f e i n m a n , l . f i n s t e n , r .

e . b l a n t o n , a n d l . m . n i c h o l a s . 1989. Monte Al-bans hinterland. Pt. 2. Prehispanic settlement patterns in Tla-colula, Etla, and Ocotlan, the Valley of Oaxaca, Mexico.Memoirs of the Museum of Anthropology, University of Mich-igan, 23.

m a r c u s , j . , a n d k . v. fl a n n e ry. 1996. Zapotec civiliza-tion: How urban society evolved in Mexicos Oaxaca Valley.London: Thames and Hudson.

m a t h e w s , j . 1992. Prehispanic settlement and agriculture inthe Middle Tiwanaku Valley, Bolivia. Ph.D. thesis, Universityof Chicago, Chicago, Ill.

n a ro l l , r . 1962. Floor area and settlement population. Ameri-can Antiquity 27:58789.

p a r s o n s , j . 1971. Prehistoric settlement patterns in the Tex-coco Region, Mexico. Memoirs of the Museum of Anthropol-ogy, University of Michigan, 3.

. 1974. The development of prehistoric complex society: Aregional perspective from the Valley of Mexico. Journal ofField Archaeology 1:81108.

. 1976. Settlement and population history of the Basin ofMexico, in The Valley of Mexico: Studies in prehispanic ecol-ogy and society. Edited by E. Wolf, pp. 69100. Albuquerque:University of New Mexico Press.

p a r s o n s , j . r . , e . b ru m fi e l , m . h . p a r s o n s , a n dd . j . w i l s o n . 1982. Prehispanic settlement patterns in thesouthern Valley of Mexico: The Chalco-Xochimilco region.Memoirs of the Museum of Anthropology, University of Mich-igan, 14.

p a r s o n s , j . r . , k . w. k i n t i g h , a n d s . a . g r e g g .1983. Archaeological settlement pattern data from the Chalco,Xochimilco, Ixtapalapa, Texcoco, and Zumpango Regions,Mexico. University of Michigan Museum of Anthropology,Technical Reports 14.

p e a r s a l l , d . 1992. The origins of plant cultivation in SouthAmerica, in The origins of agriculture. Edited by C. Cowenand P. Watson, pp. 173205. Washington, D.C.: Smithsonian In-stitution Press.

p o n c e s a n g i n e s , c . 1981. Tiwanaku: Espacio, tiempo y cul-turaEnsayo de sntesis arqueologica. La Paz: Editorial LosAmigos del Libro.

s a n d e r s , w. 1972. Population, agricultural history, and soci-etal evolution in Mesoamerica, in Population growth: Anthro-pological implications. Edited by B. Spooner, pp. 10153. Cam-bridge: MIT Press.

s a n d e r s , w. , j . p a r s o n s , a n d r . s a n t l e y. 1979. TheBasin of Mexico: Ecological processes in the evolution of a civ-ilization. New York: Academic Press.

s m i t h , b . d . , a n d c . w. c o w a n . 2003. Domesticatedcrop plants and the evolution of food production economies inEastern North America, in People and plants in ancient East-ern North America. Edited by P. E. Minnis, pp. 10525. Wash-ington, D.C.: Smithsonian Institution Press.

s t a n i s h , c . 2003. Ancient Titicaca: The evolution of complexsociety in southern Peru and northern Bolivia. Berkeley: Uni-versity of California Press.

Volume 46, Supplement, December 2005 F S115

s t a n i s h , c . , e . d e l a v e g a . , l . s t e a d m a n , c .c h a v e z j u s t o , k . f ry e , l . o n o f r e m a m a n i , m .s e d d o n , a n d p . c a l i s a y a c h u q u i m i a . 1997. Ar-chaeological survey in the Juli-Desaguadero region of Lake Ti-ticaca Basin, Southern Peru. Chicago: Field Museum of Natu-ral History.

s t e a d m a n , l . 1999. The ceramics, in Early settlement atChiripa, Bolivia. Edited by C. Hastorf, pp. 6172. Berkeley:University of California Archaeological Research Facility.

v a r i e n , m . d . , a n d b . j . m i l l s . 1997. Accumulations re-search: Problems and prospects for estimating site occupationspan. Journal of Archaeological Method and Theory 4:14191.

v a r i e n , m . d . , a n d j . n . p o t t e r . 1997. Unpacking thediscard equation: Simulating the accumulation of artifacts inthe archaeological record. American Antiquity 62:194213.

z v e l e b i l , m . , a n d m . l i l l i e . 2000. Transition to agri-culture in Eastern Europe, in Europes first farmers. Edited byT. D. Price, pp. 5792. Cambridge: Cambridge University Press.

z v e l e b i l , m . , a n d p . ro w l e y - c o n w y. 1986. Foragersand farmers in Atlantic Europe, in Hunters in transition:Mesolithic societies of temperate Eurasia and their transitionto farming. Edited by M. Zvelebil, pp. 6793. Cambridge: Cam-bridge University Press.

New Light on the EarliestHominid Occupation in East Asia1

xing gao, qi wei , chen shen, and susankeatesInstitute of Vertebrate Paleontology andPaleoanthropology, Chinese Academy of Sciences, P.O.Box 643, Beijing 100044, China (gao.xing@ivpp.ac.cn)(Gao and Wei)/Department of World Cultures, RoyalOntario Museum, Toronto, Ontario M5S 2C6, Canada(chens@rom.on.ca) (Shen)/University Village 701H, 601S. Providence Rd., Columbia, MO 065203, U.S.A.(keates20039@yahoo.co.uk) (Keates). 18 ii 05

The Nihewan Basin in northern China is a major areaof archaeological research in East Asia and a prime focusfor the search for early hominids. It is located in HebeiProvince, about 150 km northwest of the ZhoukoudianLocality 1 Homo erectus site near Beijing (fig. 1). TheNihewan is a large extinct lacustrine basin with exten-sive Pliocene and Pleistocene deposits containing hom-inid activity sites with associated fossil fauna and stoneartifacts. These sites are indicative of the earliest hom-inid occupation in East Asia at 40 north latitude (Schicket al. 1991, Zhu et al. 2001, Wei 1991, Keates 2000, Shen

2005 by The Wenner-Gren Foundation for Anthropological Re-search. All rights reserved 0011-3204/2005/4605S5-0008$10.00

1. This work was supported by Special Fund for Basic Research,Ministry of Science and Technology of China (2001CCA01700), andby the Hundreds Talent Plan and the Specific PalaeontologicalGrant of the Chinese Academy of Sciences, as well as by the RoyalOntario Museum Foundations DMV Acquisition and ResearchFund, Canada.

and Chen 2003). The newly identified site of Goudi, inthe eastern Nihewan Basin, where in situ lithic artifactsand fossil fauna have been found associated in EarlyPleistocene sediments, has been dated paleomagneticallyto about 1.66 million years (Zhu et al. 2004) and thereforerepresents the earliest hominid occupation in East Asia.

The search for early hominid presence in the NihewanBasin began in the 1920s, when Pierre Teilhard de Char-din and Emile Licent discovered and studied Pleistocenefaunal fossils there (e.g., Barbour, Licent, and Teilhard1927). In 1935 the Abbe Henri Breuil announced his dis-covery of a stone tool, but this find was dismissed asnonartifactual (Breuil 1935, Teilhard de Chardin 1935).The breakthrough in archaeological research in the Ni-hewan was the identification in 1978 of the Xiaochang-liang site, where an assemblage of lithic artifacts in as-sociation with faunal remains was excavated from EarlyPleistocene sediments (You, Tang, and Li 1978). At theend of the last century, a series of international collab-orations between scholars from China, the U.S.A., theU.K., and Canada conducted excavations of Early Pleis-tocene sites in the Nihewan, including Donggutuo, Fei-liang, and Xiaochangliang. These collaborative investi-gations introduced new approaches to the study of earlyhominid occupation in northern China, among them therefinement of the paleomagnetic stratigraphic frame-work in the Nihewan Basin and new methods of lithicanalysis including refitting and use-wear analysis(Schick et al. 1991; Keates 2000; Shen and Chen 1999,2000, 2003; Shen and Wei 2004).

Since the discovery of Xiaochangliang, a number ofother sites, including Banshan, Putaoyuan, Guangliang,and Xiantai, have been identified in the same sedimen-tary layer. In addition, the layers above the Xiaochang-liang sediment have yielded Shanshenmiaozui, Dong-gutuo, Madigou, Huojiadi, and Xujiapo, and one site,Majuangou, has been identified below it (Wei 1997).Goudi was found in the sediments directly below Ma-juangou (fig. 2). Chinese and American scientists haveconfirmed that the Xiaochangliang artifact-bearing layerresides in the reversed-polarity magnetozone bounded bythe Olduvai and Jaramillo subchrons (Li and Wang 1982,Wei 1991, Yuan et al. 1996), indicating an age rangingbetween 1.07 and 1.77 million years. The paleomagneticdating of Goudi by Zhu and colleagues (2001) placedXiachangliang at 1.36 million years ago. Therefore, onthe basis of comparative stratigraphic and paleomagneticdata, we believe that the date of Goudi lies somewherebetween 1.36 and 1.66 million years or possibly earlier,given the fact that the Goudi layer is close to the onsetof the Olduvai subchron.

discovery of the site

The new site was first identified in spring 2001 duringan archaeological survey in the Nihewan Basin as partof a research project of the Institute of Vertebrate Pale-ontology and Paleoanthropology (IVPP) of the ChineseAcademy of Sciences. The site was named Goudi andexcavated during August and September 2001 under the