6 CALCULATING LATE CLASSIC LOWLAND MAYAPOPULATION FOR THE UPPER BELIZE RIVER AREA

Anabel Ford, Keith C. Clarke, and Sebastian Morlet

The problem ofestimating ancient Maya populations has vexed archaeologists since the first temples were discovered,and existing best estimates range considerably. Using a probability map of Maya settlement pa tterns derived fro mpredictive Bayesian modeling ofthe Upper Belize River Area, we have developed a means of estimating populations inthe unknown areas. Using the ancient land use patterns ofresidential units revealed in surveyed areas, we expandedour model across a greater area, creating a probability map ofall Maya sites of the area, discovered or not. Basedon a classification of the sites and demographic assumptions about the average fa mily, we derive estimates and theirranges ofpopulation fo r the Late Classic Maya that reveal an intensive land use system.

IntroductionEstimating Maya population numbers

throughout their settlement history inCentral America has been a preoccupationof Maya archaeologists for nearly a century.Contemplating the abandoned, forest­covered temples and the numerous smallerplatforms that appeared to be everywhere,gave rise to a rich variety of interpretations.Yet it was the focused attention on the majorarchitecture and the abundance of thesmaller structures that led researchers toenvision a dense and widespread system ofMaya settlement.

As first noted by Bullard, as one looksbeyond the temples and plazas variation inMaya settlement forms and configurationscan be recognized (1960; see also Fedick1995; Fedick and Ford 1990; Ford 1991;Iannone and Connell 2003; Isendahl 2002;Sabloff 1992; Smyth et al 1995; Webster2008). Still, the perception remains of vastcities surrounded by their sustaining ruralhinterland habitations (Rice and Culbert1990 following Redfield 1967). Estimatesfor Late Classic settlement densities alsoappear to reflect this division (Rice andCulbert 1990:30-31). Yet this distinction ispossibly more structured by our ownexperiences of contemporary urbanization,and not by actual evidence from the ancientMaya landscape, where the perceivedurban/rural dichotomy falls short ofdescribing the diversity of Maya settlement(Ford 1991; Levi 2002). It is hard toevaluate the efficacy of the dichotomy,however, since archaeological surveys are

.A.DziNnc:he

Figure 1. The Central Maya Lowlands with Sitesindicated.

obstructed by the density and isolation of theMaya forest itself. Centers ranging in sizefrom the large site of Tikal to the residentialcommunity of Barton Ramie (Figure I)relied on the landscape for their subsistenceneeds. In this research we sought answers toquestions of how the Maya used their land,and consequently, how they chose to settleand farm the landscape. With answers tothese questions, estimates of ancientpopulations become feasible .

The nature of settlement patterns anddensities suggests nuances that can only berevealed by a more detailed examination of

Research Reports in Belizean Archaeology, Vol. 8, 2011, pp. 75-87.Copyright © 2011 by the Institute of Archaeology, NICH, Belize.

Late Classic Lowland Maya Population

Chan b

8anana Bank. •

! I,. .1--

"FerminReyea

"CaS~n~ Lc;okout'Onlar1o~R~1e

• • "Unltedvlle

'Chorro

M eh

Figure 2. The Upper Belize River Area with Study area indicated.

Figure 3. Comparison of the Original Model (L) with the New Model (R).

76

the Maya landscape. In prior work, wecreated an assessment for Maya sites of theUpper Belize River Area with thedevelopment of a predictive model usinggeographic variables (Ford and Clarke 2006 ;Ford et al. 2009). This research involvedextensive data correlation, building a modelbased on Weight-of-Evidence (Bayesian)predictive methods , appl ication of themodel , and its validation through fieldtesting. The model proved highly successfulin predicting both the greater and lesserpresence of Maya settlements (Ford , Clarke,and Raines 2009).

Our work with Maya settlement patternsoffers a new perspective that builds on theconstellation of environmentalcharacteristics commonly associated withancient patterns of settlement (Campbell etaI. 2006). Put simply and in local terms,well-drained ridges within the tall canopyforests had high settlement densities (Ford1986:68-69, 88, 1991); poor-drainedlowlands with short forest had lowsettlement densities; and seasonallyinundated and perennial wetlands had nosettlement (Fedick and Ford 1990). Giventhat the Maya civic centers are perchedwithin well-drained areas and that thegeneral non-central surveys incorporate thevariety .of landforms, it may be that theproportions of varied environmentalcharacteristics explain the differences indensities. High densities around majorcenters such as Tikal were located in areaswith a high proportion of well-drainedterrain ; archaeological surveys that surroundsuch centers will therefore represent well­drained ridges and high settlement densities.Broader estimates based on these densities ifextended will overestimate populations.Intersite areas , such as those locatedbetween Tikal · and Yaxha, incorporatedifferent' environmental zones ranging fromridges to wetlands (Ford 1986; 2003) andyield average settlement densities lower thanthat of Tikal , for example.

When the environmental characteristicsare taken into account, there is a widevariation. These variations may be difficult

77

Ana bel Ford et. al.

to appreciate from the perspective offeredvia a survey transect, but GeographicInformation Systems (GIS) offers a means toidentify patterns and produce maps based onsamples of identified settlements andpatterns. This is how we constructed apredictive model of Maya settlement for theUpper Belize River Area (Ford and Clarke2006 , Ford et al 2009) . The results of theGIS research forms the basis of this furthereffort to calculate Maya population densityand distribution.

We begin with an assessment of thepredictive model of Maya settlements as abasis for the extrapolation of settlement andresidential patterns. Refining our predictivemodel to build the population estimates, weused the transect surveys for the BelizeRiver Archaeological Settlement Survey(BRASS), along with additional surveys ofBarton Ramie , to characterize residentialconfigurations and settlement patterns acrossthe wider Upper Belize River Area. Therecognized patterns are then propagatedfrom the actual surveys to the broader map,based on the predictive model using the GIS.

Populations were estimated on the basisof defined residenti al units plotted with theGIS to create the broader settlement map.To calculate population, we develop astrategy to first determine primary andsecondary residence derived fromethnohistoric and ethnographic cases. Theresults provide both a view of howsettlement and population vary across thelandscape and also present a picture of highpopulation levels and intense land use forthe Late Classic Maya .

Predictive Model Map of Ancient MayaSettlement Patterns

Over the past several years, we havebeen working with the results of a predictivemodel of Maya sites drawing on datagathered in the GIS (Ford and Clarke 2006 ,Ford et aI. 2009; Merlet 2009, 2010).Focused on the data gathered by the BRASSresearch , where three basic transects in thewestern portion and targeted quadrants tothe east were designed to gain an

Late Classic Lowland Maya Population

;- T~~i - - __n _ --- - - ---- rP-;-~am;te-;'---- -- .---.- - -- -- ----- r Re~~it -- -- - -- -

Tra ining sites for II Training point s - I R~~do~ Sal~p li ng-~fTr~ i ni ngPOints ,------ - ----- ----- - - - - ---- - -f- ----- - - - -- --- - - - ------------ ----- - - - -- - - -- --- -:.-----

~. -;~::::: :;b~;SI ;t ;" -- t::~;~;-::;;,~~mts-- -J~;:~::~~:;~;:~:;:':':i;::i~S I,---- -.. --- - - -- ------- - -.......-- - -------- .---- -- ----._-.-----.---------L--- --- .. ,! ~~g.i~~ i c ~egE~~}<?!1 _ . .___ _R__a~.~~~__<?f.~_e!ght~ . .____ _ 1 .~~!£~Et~~d_. J3.egre.~~<?~ Statist ic . iL~eigh ts of EYigence (~ofE) Raster.of Weight.s . _ I Calcu !ated.~eigh!_s_of Ev ide!1.~e -J: f\g~e..~le..~g :...C heng__Ies~_ __ _ ~~!e!!<?!_~_obab l !I ty S~~nd~~~ ~_ Te3~ _~o_n~!.~!.O'!1 ~!_l~dep.e..~.d ef.lc ~_ j:_A.~~!.!::~~gue'!9 _Tab 1~2.. nos~e..!io. !:..~robab i 1_~~x_I!..~~!l_~y~ints_Tabl~_o..f ~~.e..__~red i ct iye ~~~~J _ ___

Table L Geogra phic Information System Database for the Predictive Mode l of Maya Settlement (Merlet 2009).

appreciation for the variation of sett lementsin the alluvial vall ey, the rolling marlfoothills and the well-drained ridges to thenorth . Before the BRASS study, Mayastudies concentrated in the vall ey suggestedthat sett lement was a ribbon line patternonly along the river (Coe and Coe 1956;Thompson 1942; Willey et al. 1965). Thisview is now altered and we have a morecomplex understanding of settlem entpatterns today. Thu s we can compensate forthe sampling bias of prior sites selected tooclose together.

Based on the BRASS research, we havedeveloped a predi cted map of the c. 1300 sqkm of the Upper Belize River Area thatpredicts sites and their density at a 96%confide nce level (Figure 2 and 3). Based onthe known sites and their relationship to thegeographic variables of soil fertili ty, so ildrainage, distance from rivers, andtopographic slope, we constructed apredictive map based on the steps outl inedin Table I.

The original tests (Fo rd et al. 2009)applied a model training mask that boundedall the transect areas (Ford et al 2009:8) forthe Upper Belize River Area and includedrivers as a geographic theme. In our presentmode l, we have refined the focus byrecreating the train ing mask coverage tobound precisely around only those areaswith survey coverage, thus we control forlocations with s ites as we ll as for those areaswitho ut sites (Merlet 2009: 16-17). Theresults prov ide a more powerful model withthe quality of predicting not only the

presence of sites but also the absence ofsites .

In the fina l model refinement s,experiment results were compared with andwithout the coverage of rivers. We foundthat the interact ion of the river cove rage wasclouding patterns of land use, espec iallywhere there was a low probabili ty ofsettlement (Ford et al. 2009: 12-13). Therelat ion between rivers and sites was foundto be compl ex, positive at some d istancesand negat ive at others. Eva luat ing therelia bility of the models with and withoutthe river coverage, we determined that wehad better predictive results without therivers.

The validation of the predictive modelused successive fie ld tests based on theinitial mode ls (Monthus 2004; Ford andClarke 2006; Ford et al. 2009). Unsurveyedareas were field visited and site locationswere plotted with a Global Pos itioningSystem (GPS) rece iver and mobil e GIS.These site validat ions were used to refine afinal model and provide the new probabilitymap (F igure 3). Our curre nt mode l resu ltsuse the detailed mask of the actual surveyareas and only three geograph ic varia bles:soil fertility, soi l drainage, and topographicslope. With these, our predictive map isstatist ically both highly explanatory androb ust.

Domestic Architecture and ResidentialUnits in the Upper Belize River Area

From the outset , anc ient structures havebeen used as a proxy for houses and the ir

78

LV l.J . .... ......_& ..t-'. _ . ~ , r - -

' M WoI..,,

fundamental starting point , and are bothvalid and comparable (Relay et al. 2007).

At present , we do not concern ourselveswith the invisible remains for precisely thatreason, they are invisible. This does notdiscount the many issues raised about such aproblem (see Johnston 1992, 2002), but doesemphasize that we are basing the estimateson visible remains only. Indeed, we see thevariety of visible residential configurationsand their spatial distribution as important ,considering groups as distinct from solitarystructures a critical point and considered inour definition ofprimary residential units

79

q" '" \\\

\ ,.4' ,' 1

\\ ,\ ,

\".4' ,

: LegendI PtoblbAlty

H1gIl

, Lowi:,:«a0'O·

\\

Anabel Ford et. al.

inhabitants, linking to the obviousrelationship of residential structures andfamilies . The move from domesticarchitecture to homes and people has vexedarcheologists and caused much discussion inthe archaeological literature (see Culbert andRice for a bibliography). Any estimationsystem . is fraught with problems andassumptions that influence the results(Turner 1990). Our attempt here is toincorporate the standard system ofcalculation based on a new strategy ofdefining permanent residential units. Wefeel that the traditional survey data are a

Figure 5. Results of the Predictive Model for High Priority Zones on the BRAa Transect: Valley and Ridges (BacabNa (L) and El Pilar Transect (R)).

Figure 4. Results of the Predictive Model for Low Priority Zones on the BRASS River (Pilar (L) and Yaxox Transect(R)).

Late Classic Lowland Maya Population

------------------- --------,

MesoAmerican Research Center, 2009

Maya' Sites

_ Secondary Resklence

_ Rank1

_ Rank2

Rank 3_ Rank4

_ Rank5

Figure 6. The Residential Unit Distribution for the Upper Belize River Study area.

(Willey 1956; Ashmore 1981; Levi 2002,2003).

We set up an approach to classify thenature of the domestic architecture based onthe requisites of a basic house of the Mayaarea (see Wachope 1938 as an example). Wetake the position that the recorded surfaceremains provide a foundation for estimationsof land use intensity and population densityin and of themselves. Weare comparingsimilar features when we discuss visiblearchitectural remains, as surface remainsprovide a common comparative base .

We use the recorded remains ofdomestic architecture from thearchaeological surveys of the BRASSprogram (Ford 1985, 1990; Ford and Fedick1992), along with the Barton Ramie map(Willey et al. 1965) to generalize over thewhole Upper Belize River Area. Thisrequired the identification of residentialsites, the consideration of SIze, the

80

understanding of composition, and thelocation based on the predictive model. Forour map, we consider the site locationsbased on the BRASS surveys of the 1980s,including the three transects as well as theancillary surveys conducted in peripheralareas (Fedick 1988) and the Barton Ramiemap (Willey et al. 1965). Based on thesesite locations and their association with thepredictive values generated by our GISmodel, we have been able to propagatepredicted patterns of settlement across thelandscape . These patterns have beenvalidated at the 96% confidence level (Fordet al. 2009: 13).

With the BRASS residential database ,we used recorded data on structure size(diagonal length in m), residential unitcomposition (number of structures andevidence of plaza) , labor investment(calculated based on Arnold and Ford 1980),and the location to compute the weights of

- iT

Anabel Ford et. ·al. III,

SS Transects

(km") Population Density % Population % Area

.01 0 0 0% 22%

.51 64 124 3% 06%

.20 154 129 8% 13% I

.57 298 190 15% 17%

.72 1452 390 74% 41% I

.01 1968 218 100% 100% ,Transects.

II

aII

(km-) Population Density % Population % Area

85 0 0 0% 38%

43 30255 124 17% 19%

76 9818 129 5% 6%

25 42800 190 23% 18%

56 99727 390 55% 20%

284 182600 142 100% 100%

tudy area.

'I'al unit evoke varied strategies that typically Iate the involved several residences (Farris 1984;

Fedick 1992, 1996; Zetina and Faust In I

culation Press; Netting 1977; Redfield and Villaal map Rojasl962; Steggerda 1941; Villa Rojas

I

ranked 1945; Zetina 2007). In fact, recentat takes ethnographies of the Maya suggest anber of average of three residences per family basedce of a on the seasonal round of agricultural

sis for activities (Zetina and Faust In Press; Zetinag their 2007), thus creating a pattern of multiple

residence based on agricultural demands ines have the field.e house Based on the ethnographic studies, we "

as been consider that the very small structures woulds of have served secondary domestic services

nstrated and should be excluded from the estimational units of population. Thus, solitary structures with

II81

..

.~

.- "X IX ,- , '< _I~ _, n V J U_. ...--

2

2

2

4

9

2

3

o

Area

o11

28

53

352

259

Late ClassicResidences

0

2 5403

3 1753

4 7643

17808

32607

Table 3. Total estimated Population for the S

evidence statistic . These resident!variables were then used to propagpatterns on the map.

Using the labor investment cal(LI) originally developed for the Tik(Arnold and Ford 1980), sites werefrom 1-5 based on a calculation thinto account the size and numstructures and the presence or absenplaza. This provides the badistributing sites and considerinoverall impact on the landscape.

While the presumed house sitbeen accepted as such, how to use thsites as a proxy for population hfraught with problems. Patternsubsistence agriculture have demothat families that occupy residenti

WofEClass Residential AreaUnits

Total Population Distribution of BRA

Table 2. Population Distribution of BRASS

Total Population for the Study Are

Late Classic Lowland Maya Population

an LI less than 500 with diagonals averagingc. 10m, were designated as secondaryresidences. These secondary residential unitscomposed 41% of the BRASS sample, butonly c. 26% of mapped structures . Theresidential units with LI greater than 500and an average unit diagonal of c. 24 m weredesignated as primary residential units andprovided the basis for our populationcalculation. The primary residential unitscomposed 59% of the domestic architecture .of the BRASS surveys.

Propagating Maya Settlements Across theUpper Belize River Area

The identification of primary andsecondary residential units in the BRASSand Barton Ramie surveys were used topropagate patterns of residential settlementsproportionally across the entire UpperBelize River Area. The propagation usedthe probabilities of our predictive model asthe foundation. We used the residential siteclassification by labor investment rank (LI;Ford and Arnold 1980) as they patterned inthe surveys. Figures 4 and 5 provide asample of the relationship of the predictivemodel and the Maya settlement in the UpperBelize River area. The first pair of images(Figure 4) shows the low priority areas andthe nature of settlement in those areas . Thesecond pair of images (Figure 5) shows thehigh priority areas and the patterns of thoseareas. The variations in the densities, thecomposition of the residential units , andtheir size and configurations, werecategorized and schematically coded for thepropagation to the larger mapped area.Based on the nature of residential settlementin these surveyed areas , we propagatedresidential sites to the larger map of theBelize River Area using random simulation.

Based on the predictive map , we talliedresidential sites that qualified as primaryresidential units and then used the standardproportion of occupation in the Late Classicperiod of 95% to build the map. This isconsistent with the BRASS data (Ford 1985;Ford et al. 2009 :14). From this we wereable to associate the residential units withthe predictive zones (classed from the WofE

82

or Weights of Evidence model predictiveprobabilities) to create a table of thepopulation distribution (Table 2). Togenerate our population estimation for theUpper Belize River Area, we use thegenerally accepted standard of 5.6 personsper residential unit (Narroll 1962; Puleston1973; Turner 1990; Healy et al. 2007).Turner (1990) argued that this is aconservative estimation for "paleotechnicagrarian economies" (Turner 1990:305).

The result of this process produced amap that provides a picture of thedistribution and a view of the concentrationsof settlements and populations in the BelizeRiver Area (Figure 6). Distributions arediverse and we can see how areas ofconcentrated settlements in the east, south ,and the northwest demonstrate where elitecontrols should be manifest.

Using the tools of the GIS and thepredictive model , we were able to use thedistribution map (Figure 6) to develop atable of residential distribution for the entirestudy area of the Upper Belize River area(Table 3). The distributions of the BRASSsurveys are the basis of the new map that isthe foundation of the population estimations.This new map provides the probabilitydistribution of Maya sites at the 96%confidence level. The landforms andpredictive zones are of different proportionsin the whole study area when comparee}withthe BRASS transects. . The predictive zones ,however, are represented in the BRASStransects and the validation of the modelincludes the Upper Belize River Area (seeFord et al. 2009 :14-15) .

The population estimates for the UpperBelize River Area are substantial. The rangeis extreme: 0 persons per sq km in areas oflowest probability (38% of the area) to 390persons per sq km in the highest probabilityzones (55% ofthe area) . The overall averageis 142 persons per sq km, a density that issignificantly greater than the early estimates(Turner 1990:317) but within the range thathas been estimated for the Late Classicperiod (Culbert and Rice 1990). This isrelatively dense by Boserup 's (1981:9-11)reckoning. The density estimated for Ming

1 / T T.

Dynasty China at 1500 AD? is estimated atonly 64 persons per sq km and pre-modernJapan at 1750 AD is roughl y 128 personsper sq km. Rosenberg (2010) summarizesour world 's current popu lation density bycontinent:

North America - 83 peop le per square kilometerSouth America - 189 people per square kilomet erEurope - 347 peopl e per square kilomet erAsia - 525 people per square kilometerAfr ica - 168 peopl e per square kilometerAustra lia - 17 peop le per square kilometer

Boserup (1981 :9), in her synthetictreatise on population and technology,evaluated population dens ities worldwideand determined that density greater than 64are dense and over 256 very dense. Shepresented a tab le of continental densitiesbased on the conditions in 1975. The denseand very dense categories are only found inEurope and Asia (Boserup 1981: 11).

Reviewing our population est imates forthe Late Classic Maya of the Upper BelizeRiver Area suggests a very intensive use ofthe landscape. In the c. 1300 sq km that wasstud ied, ] 82, 600 people must have livedand farmed. Yet while areas of the greatestdensity reach 390 per sq km, very dense byBoserup's calcu lations (1981 :9), thisencompasses only 20% of the area .Consequently the most intense land usetakes in only one-fifth of the terrain. Thissma ll percentage would be a complexmosaic of houses, varied open sunlit area s ofmaize field s and home orchard gardens, asdescribed in the ethnographic literature(Redfield and Villa Rojas 1962; Zetina2007). These dense sett lement zones foundthroughout the study area are both near andfar from major and minor centers. Theycontrast with the virtually unoccupied areasof low sett lement priority making up 38% ofthestudy area. These extensive unoccupiedzones , encom pass ing nearly two -fifths of thestudy area , wou ld have been wood lands ,thus supporting the natural reso urce needs ofthe Maya.

83

Anabel Ford et. at.

SummaryThere have been many strategies used to

develop popu lation estimates for prehistoricsocieties . The Maya have presented one ofthe more difficult cases as the tropical forestcover limits surveys . New satellite andgeospatial tech nologies have soug ht toovercome these obstacles (e .g. Garrison etal. 2008), and in the future we may have anopportunity to test our land use mode l wit hthese more sophisticated capabi lities. Ourpred ictive mode l prov ides one way toaddress land use based on statisticalprobabi lities (Ford and Clarke 2006 ; Ford etal. 2009).

With the new population densi ty mapbased on our predictive mode l, wedemonstrate in th is paper a means to create asett lement map using archaeological surveysand their pattern s against a predictive modelmap that may have used beyond our conte xt.Whi le it has been recognized thatsett lements are not evenly distributed, ourmap is based on geographic var iables onwhich the settlements are depe ndent. Byenumerating the proportions of residentia lunits and propagating their distributionacross the landscape from the known areasto the predicted areas , we have derived thepatterns of primary and seco ndaryresidentia l sites. Our total estimates project142 persons per sq km for the whole studyarea, with a range of 0-390 persons per sqkm. The est imates we have come up withare approximately ten times the populationdensity of Belize today.

These popu lation estimates are foundedon a series of assumptions. We recognizethe subsistence farmer ' s strategy of usingprimary and secondary residences, and havemade our calc ulations based on the primaryresidential units alone. We use 95%occupation commonly used for the Mayaarea in the Late Classic period , comparableto Late Classic occupation in the area. Withall these considerations taken into accou nt,we have discovered that the density ofoccupation of the Upper Belize River Areawas potentially very high . We believe theresults of our research provide anunderstanding of how the landscape was

Late Classic Lowland Maya Population

used . We present a mosaic of intensi tybased on the geographic variables of theUpper Belize River Area. We show areas ofintense use and areas that are largelyunoccupied. This variation is acknowledgedbut had not been we ll appreciated. Ourmodel demonstrates that high settlement andpopulation densitie s are not somethingconfined to the urban civic centers , butoccur where the constellation of geographicvariables coincide: we ll drained fertile soilareas with moderate slope (Ford et al. 2009).Access to natural water sources was only aminor factor (Merlet 2009). Sett lementswith high predicti ve values would be goodcandidates for civic center presence. Thisco uld be tested by further fie ld surveys.

To support an estimated population of142 persons per sq km, a 'successfulsubsistence economy was required . Theinvestments that began to take shape in thePreclassic before 1000 BC and grew overtwo millennia must be able to account forthis intense land use pattern. We proposethat the fo rest garden milpa cycle (Ford andNigh 2009) was the subsistence base thatcould provide a long-term sustainablesubsistence base for the Upper Belize RiverArea Maya. Our next step will be to build amodel of Maya land use based on the Fores tGarden Milpa cycle and the food andresources it could supply.

Acknowledgements We thank UCSBsResearch Across Disciplines program forfunding aspects of our research and ourco llaboration with spatia l engineering intersform the Ecole Super ieure des Geometres etTopographes in Le Mans.

References Cited

Arnold, Jeanne E. and Anabel Ford1980 A statistica l exami natio n of sett lemen t

patte rns at Tika l, Guatemala. AmericanAntiquity 45(4):7 13-726.

Ashmore, W.198 1 Some Issues of Method and Theory in

Low land Maya Sett lement Archaeo logy. InLowland Maya Settlement Patterns, edi tedby W. Ashmo re, pp. 37-69 . University ofNew Mexico Press, Albuquerque.

84

Bill ie Lee Turner, II1976 Population Density in the Class ic Maya

Low lands: New Evide nce for OldApproaches. The Geographical Review66( 1):73-82.

Boserup, Ester1981 Population and Technological Change:

A Study 0/Long-Term Trends. Universi ty ofChicago Press, Chicago .

Brow n, Clifford T. and Walter R.T. Witschey2003 The fracta l geometry of ancient Maya

sett lement30 :1619-1 632.

Chase , Ar len F. and Diane Z. Chase1998 Scale and Intensity in Classic Period

Maya Agriculture: Terracing and Sett lementat the "Garden City" of Caraco l, Belize.Culture & Agriculture 20(2 /3) .

Cowgi ll, Ursula M.1960 Soil fert ility, popul ation, and the

ancie nt Maya. Proceedings 0/ the NationalAcademy ofSciences 46 :1009 -1011.

Culbert, T. Patrick and Don S. Rice (ed itors)1990 Precolumbian Population History in

the Maya Lowlands. University of NewMex ico Press, Albuque rque.

Fedick, Scott L.1988 Prehis toric Maya Settlement and Land

Use Patterns in the Upper Belize RiverArea , Belize, Central America. Ph. D.Dissertation, Arizo na State University.

Fedick, Scott L1989 The Econo mics of Agricultural Land

Use and Sett lement in the Upper BelizeRiver Valley. In Prehistor ic MayaEconomies 0/ Belize, edi ted by P. A.McAnany and B. L. Isaac, pp. 2 15-254 .Research in Economic Anthropology,Supp lement no. 4. JAI Press, Greenwich.

Fedick, Scott L.1992 An Agricu ltura l Perspective on

Prehistor ic Maya Hou sehold Locat ion andSett lement Density. ln Memorias DelPrimer Congreso fnternacional deMayistas; Mesas Redondas, Arquelogia,Epigrafia , pp. 87-108 . Universidad NacionalAutonoma de Mex ico, Mexico D.F.

Fedick, Scott L1995 Land Evaluation and Ancient Maya

Land Use in the Upper Bel ize River Area,Belize, Centra l Amer ica. Latin AmericanAnt iquity 6( 1):16-34.

- _ L4 .... _ ..... ' A't:3 nrnnl )SC::

Fed ick, Scott L. (edi tor)1996 The Managed Mosaic: Ancient Maya

Agriculture and Resource Use. University ofUtah Press, Sa lt Lake City.

Fed ick, Scott L. and Anabe l Ford1990 The prehistoric agricultural landscape

of the central Maya lowland s: Anexa mination of local variability in a regionalcontext. World Archaeology 22( 1):18-33.

Ford, Anabel1985 Maya Settlement Pattern Chrono logy in

the Belize River Area and the Implicationsfor the Deve lopment of the Cen tral MayaLow lands. Belcast Jour nal of BelizeanAffairs 2: 13-32.

1990 Maya Sett lement in the Bel ize RiverArea: Variat ions in Residence Patterns ofthe Central Maya Lowland s. In PrehistoricPopulation History in the Maya Lowlands ,edited by T. P. Cu lbert and D. S. Rice, pp.167-181. Univers ity of New Mexico Press,Albuquerque .

1991 Problem s with the Eva luatio n ofPopulation from Settlement Data:Examination of Ancient Maya ResidentialPatterns in the Tikal-Yaxha Intersite Area.Estudios de Cultura Maya (18) :157- I86.

2003 Crecimiento de Poblacion yComplejidad Social: Asentamiento y MedioAmbiente en las Tierras Bajas Mayas.Translat ed by E. H. Gaytan Monografica 14.Plumsock Mesoamerican Studies /C IRMA ,Miami.

Ford, Anabe l and Keith C. Clarke2006 Predicting late class ic May a sett lement

pattern s. In Archaeological Investigations inthe Eastern Maya Lowlands: Papers of the2005 Belize Archaeology Symposium, pp.193-212. Research Report s in BelizeanArc haeo logy. vo l. 3. Institute ofArchaeology, Belmopan.

Ford, Ana bel, Keith C. Clarke and Gary Ra ines2009 Modeling sett lement patterns of the

Late Classic Maya with Bayesian methodsand Geographic In formation Systems.Anna ls of the Association of AmericanGeographers 99:496-520.

Ford, Ana bel and Scott L. Fedick1992 Prehistoric Maya Settlement Patterns in

the Upper Belize River Area: Initial Result sof the Be lize River ArchaeologicalSett lement Survey. Journal of FieldArchaeology 19:35-49.

85

Anabel Ford et. at.

Ford, Anabel and Ronald Nigh2009 Origins of the Maya fore st gar den:

Maya resource management after theHolocene thermal maxi mum. Journal ofEthno biology 29:213-236.

Garrison, Thomas G., Stephen D. Houston, Char lesGolden , Takeshi Inomata, Zachary Nel son and JessicaMunso n

2008 Eva luating the use of IKONOS satell iteimage ry in lowland Maya sett lemen tarchaeo logy . Journal of ArchaeologicalScience 35:2770-2777.

Hav iland, William Arthur1969 A New Po pulation Est imate for Tikal,

Guate mala. American Antiquity 34(4):429­433.

Hav iland, William A.1972 Family Size , Prehisto ric Population

Estimates, and the Ancient Maya . AmericanAntiquity 37(1):135-139.

Healy, Paul F., Ch ristophe G. B. Helmke, l aime J.Awe and Kay S. Sunahara

2007 Survey, Sett lement, and Popul ationHistory at the Ancie nt May a Site ofPacbitun, Belize . Journal of FieldArchaeology 32(1): 17-39.

Hendon, lul ia A.1992 The Interpr etation Of Survey Data :

Two Case Stud ies From The Maya Area.Latin American Antiquity, 3(1):22-42.

Iannone, Gyles and Samue l V. Conne ll (editors)2003 Perspectives on Ancient Maya Rural

Complexity . The Cotsen Institute ofArchaeo logy, University of Ca lifornia, LosAngeles, Los Ange les.

Isendah l, Christ ian2002 Common Knowledge: lowland Maya

urban farming at Xuch. thesis, UppsalaUniversity.

Joh nston, Kevin, Fernando Moscoso Moller andStefan Schmitt

1992 Casas No-Visibles De Los MayasClasicos: Estructuras Residenciales SinPlataformas Basales En Itzan, Peten. In VSimposio de Investigaciones Arqueo logicasen Guatemala, edited by J.P . Laporte.Museo Nac ional de Arqueo logia yEtnologia, Guatemala.

l ohnston, Kevin J.2002 Protrusion, Bioturbation, and

Settlement Detectio n during Surface Survey :The Low land Maya Case. Journal of

Late Classic Lowland Maya Population

Archaeological Method and Theory 9( 1):1­67.

2003 The intensification of pre-industrialcereal agriculture in the tropics: Boserup,cultivation lengthening, and the ClassicMaya. Journal of Anthropo logica lArchaeology 22(2): 126-161.

2004 The "Invisible" Maya: MinimallyMounded Residential Settlement at Itzan,Peten, Guatemala. Latin American Antiquity15:145-179.

Levi, Laura J.2002 An Institutional Perspective on

Prehispanic Maya ResidentialVariation:Settlement and Community at SanEstevan, Belize. Journal ofAnthropologicalArchaeology, 2 1:120-1 41.

2003 Space and the limits to community. InPerspectives on Ancient Maya RuralComplexity, pp. 89-93. vol. 49.

Merlet, Sebastien2009 A study on the use of Geographic

Information Systems (GIS)in a multi-scale archaeological project. MA,Ecole Superieure des Geometres etTopographes.

20 I0 Determination des besoins d'unepopulation Maya durant la periode duClassique Tardif au moyen d'un modelepredictif Revue XYZ 123:17-24.

Monthus, Florent2004 Weights of Evidence and Maya

Settlement . MA Ecole Superieure desGeometres et Topographes

Redfield, Robert1967 A Village that Chose Progress : Chan

Kom Revisited. The University of ChicagoPublications in Anthropology. SocialAnthropological Series. University ofChicago Press, Chicago.

Redfield, Robert and Alfonso Villa Rojas1962 Chan Kom, A Maya Village . University

of Chicago Press, Chicago.

Rice, Don S.1978 Population Growth and Subsistence

Alternatives in a Tropical LacustrineEnvironment. In Pre-Hispanic MayaAgriculture, edited by P. D. Harrison and B.L. Turner, pp. 35-6 1. University of NewMexico Press, Albuquerque. .

Rosenberg, Matt20 I0 Population Density. 20I O. About.com

Guide. Access 6 December 20 10http://gcography.abolll .colll /od/popliia l iOl1gc!}graDhV/U/DODdcl1siI\'.hIIII

Sabloff, Jeremy A. and Gair Tourtellot1992 Beyond Temples and Palaces: Recent

Settlement Pattern Research at the AncientMaya City of Sayil (1983 - 1985). NewTheories on the Ancient Maya 77 :77.

1992 Beyond Temples and Palaces: RecentSettlement Pattern Research at the AncientMaya City of Sayil ( 1983-1985). NewTheori es on the Ancient Maya 77:77 .

Smyth, Michael P., Christopher D. Dore and NicholasP. Dunning

1995 Interpreting Prehistoric SettlementPatterns: Lessons from the Maya Center ofSayil,Yucatan. Journal ofField Archaeology22(3) :321-347.

Moriarty, Mathew D.2004 Settlement Archaeology at Moltul de

San Jose, Peten, Guatemala: Preliminaryresults from the 1998-2003 Seasons. Mayab17:2 1-44.

Thompson, 1. Eric S.1971 Estimates of

Deranging Factors.36(2):2 14-2 16.

Maya Population:American Antiquity

Netting, Robert McC.1977 Maya subsistence: Mythologies,

analogies, possibilities In The Origins ofMaya Civilization, edited by R. E. W.Adams, pp. 299-333. University of NewMexico Press, Albuquerque.

Puleston, Dennis Edward1973 Anc ient Maya Settlement and

Environment at Tikal, Gua temala:Implications fo r Subsistence Models. PhDdissertation, University ofPennsylvannia.

86

To urte llot, Gair, Jeremy A. Sabloff and Michael P.Smyth

1990 Room Counts and PopulationEstimation for Terminal Classic Sayil in thePULIC Region, Yucatan, Mexico. InPrecolumbian Population History in theMaya Lowlands, edited by T. P. Culbert andD. S. Rice, pp. 245-26 1. University of NewMexico Press, Albuquerque.

Turner, Bill LeeJ990 Population Reconstruction of the

Central Maya Lowlands: 1000 B.C. to A.D.1500. In Precolumbian Population History

" - . ~ .l .f _ . . ..... '7'7 .7 7

in the Maya Lowlands, edited by T. P.Culbert and D. S. Rice, pp. 30 1-324. l st ed.Unive rsity of New Mexico Press,

Albuquerque.

Vill a Rojas, Alfonso1945 The Maya of East Central Quintana

Roo 559. Ca rneg ie Institution, Washington ,

D.C.

Wauchope, Rob ert1938 Modern Maya houses: A study of their

archaeo logical significance. Ca rnegieInstit ution of Washington , Washington DC.

Webster, David2008 Three Maya sett lement projects and

thei r impli cations. In 6th Palenque MesaRedonda, pp. 28, Palenq ue.

Webster, David , Timothy Murtha, Kirk D. Straight,Jay Silverste in, Horacia Martinez, Richard E. Te rry

and Richard Burnett2007 The Gre at Tika l Earthworks Revi sited.

Journal ofField Arc haeology 32:41-64.

Will ey, Gordon R.1956 The Structure of Anci ent Maya

Society: Evidence from the SouthernLowlands. American Anthropologist 58:777-

782. .

Will ey, Gordon R., Jr. William R. Bull ard , John B.

Glass and James C. Gifford1965 Prehistoric Maya Settlements in the

Belize Valley . Peabody Museum,

Cambridge.

Zetina Gutierrez, Maria de Guadalupe2007 Eco logia humane de las rancherias de

Pich, Campeche: un analisis diacron ic. MS ,Centro de Investigaci6n Y EstudiosAva nzados del Instituto Politecni co

Naciona.

Zetina-Gutierrez, M . G. and B. Faust.InPre ss i,Cuantas casas cada famili a? Un reto

para la arqueologia maya de la agriculturan6mada. Estudios de la Cultural MayaUniversidad Nacional Autonoma de Mexico.

(Ok .

87

Anabel Ford et. al.