ON THE MAYACOLLAPSE

RobertS. Santley,Thomas W. Killion, and Mark T. Lycett Departmentof Anthropology,University of NewMexico, Albuquerque,NM 87131

The collapseof Late ClassicMaya civilizationinvolved more than the disintegrationof politicalstructure. It involveda totalsystem failure in whichboth centers and dependent villageswere abandoned by elitesand commonersalike. The lowlandrainforest habitat whereMaya civilizationdeveloped was not significantlyreoccupied until comparatively recenttimes. The collapse was differential,in thatcenters in coastalareas or drierregions suchas northernYucatan were not depopulatedto thesame degree. The collapse of many Mayacenters in theforested interior is hereattributed to threeinteracting sets of variables: (1) nutritionalstress, disease, and demographic instability; (2) agriculturalintensification, monocropping,and degradationof the agrarianlandscape; and (3) the relativeabsence of macroregionalresource extraction structures. These factors had littleimpact on Maya populationsliving near the coast or in lowlandareas not originallycovered by tropical rainforests.

BETWEENA.D. 300 AND800 the ancientMaya developed a complex, hierar- chicallyorganized society in an environmentwe know todayto be highlysus- ceptible to environmentaldegradation and characterizedby slow natural regeneration(R. E.W. Adams 1977; Culbert 1973; Gomez-Pompa and Vasquez- Yanez1981; Rice and Rice 1984; Wilkerson1985). In the Late Classicperiod (A.D. 600-800) substantialparts of the MayaLowlands were denselyoccupied by populationsof peasant farmers. Because most cultivablesections of the Peten and adjacentregions alreadyhad been broughtunder intensive culti- vation,these farmingpopulations made great demandson their tropicalrain- forest habitat. Classic populationscombined dense settlement and intense agriculturalproduction in physiographicregions as diverse as the Pasionand Usumacintariver drainages, the bajosand lakes of the centralPeten, slopelands in the Rio Bec region, the river valleysand wetlandsof , andthe drier regions in the northernsection of the YucatanPeninsula (Figure 1). It now seems likelythat the very success of the ClassicMaya at settlingand developing the Lowlandforest laid the foundationfor the collapseof southernLowland Mayacivilization at the end of the Late Classicperiod. The growthand rapid decline of Mayacivilization in the southernLowlands providea sharpcontrast with other areas where ancient civilizations developed. In semiaridregions such as CentralMexico, the Andes, and greater Meso- potamia,periods of relativelysustained growth and development,separated by phases of politicalfragmentation, appear to be the norm throughoutthe prehistoricsequence. The sociopoliticalsystems of these areas all faced the problem of provisioning increasingly larger populations on circumscribed ag- riculturallandscapes. The trajectory of culturaldevelopment in the Maya Low- lands, however, was different.The fluorescenceof Classicperiod in the Peten and adjacent areas was followed not by periods of balkanization

123 124 JOURNALOF ANTHROPOLOGICALRESEARCH

Dzibilchaltun

Chunchucmil

Sayil

NMayapan 0 50 100 200 Northern Lowlands KM "

Gulf of

Becan Candelaria SCalakmul Central Lowlands f/Pulltr:usdrSwamp

Palenque Uaxactn Barton Ramie SAguadasN - , Southern Lowlands PasionR. Caribbean Sea Altar de Sacraficios

Southern Highlands

0 Copan

The Maya Area - -.TWK Figure1. Mapof the MayaRegion andregrowth but by wholesalesystem collapse,with major parts of the Low- landsabandoned until comparatively recent times. In the Maya case we will arguethat the need to sustainprogressively larger populations placed severe constraintson the foodproduction base whichprecluded further development, renderedthe system inviable,and preventedrecovery. In this paperwe will focus on the problemof maintainingintensive sustained-yield agroecosystems in a rainforestenvironment. Agricultural utilization and degradation of the trop- ical forest habitatby growingpopulations of peasantfarmers have evidently been a commonfeature of landuse fromprehistoric times to the present. The ONTHE MAYA COLLAPSE 125 fallof ClassicMaya civilization in the ninthcentury, then, is seen here as the negativeoutcome of a series of interrelatedbiological, demographic, and so- cioenvironmentalproblems faced by all archaiccivilizations and is an important objectlesson for the contemporaryuse and developmentof the tropics. Argumentsconcerning causes of the Mayacollapse have includedsuch fac- tors as inherentenvironmental limitations of the rainforesthabitat, climatic perturbations,soil exhaustion,famine, disease, civilstrife, andforeign invasion (Culbert1973). As Lowe (1985) has recently pointedout, these arguments are generallyinadequate because they lack systemic character.Other more integrativemultifactor models of the collapse have been criticizedfor their inabilityto explainknown patterns of variabilityin the MayaLowlands. One of the most intriguingaspects of the Maya"experiment" in civilizationwas the initialgrowth and success of Classicpopulations at utilizingthe Lowlandtropical rainforestand transforming it intoa highlyproductive agroecosystem. Evidence for landscapemodification and sustainedagricultural intensification continues to accumulatefor the MayaLowlands from the Preclassicuntil Late Classic times. Whythen did this seeminglysuccessful and long-enduringexperiment in civilizationfinally fail at the end of the Late Classic? It is our positionthat, althoughinitially successful, Maya populations even- tuallyapproached the limitsof the southernLowland resource base at a time whenthe cumulativeeffects of malnutritionand disease madethe maintenance of an elaborateagrotechnology increasingly difficult. The combinedeffects of environmentaldegradation due to the slow regenerativecapacity of Lowland forests and soils, physiologicalstresses related to inadequatenutrition and disease infestations,and the relativeabsence of regionalresource extraction networksled to demographicinstability, the collapseof the agriculturalsystem, andthe eventualabandonment of a majorityof the southernLowland centers by elites andnonelites alike.' Another property of the Mayacollapse is that it was spatiallyrestricted. In contrastto morelandlocked portions of the southern Lowlands,populations located in coastalareas where marineresources pro- vided an importantcomponent to subsistence (e.g., Belize), in areas where soils were less susceptibleto degradationand/or had a higherregenerative capacity(e.g., highlandGuatemala), or in areas where alternativeresources such as salt were available(e.g., northernYucatan) were not disruptedto the samedegree. Becausethe structureof the resourcebase utilizedby populations inhabitingthese areas was different,a collapse phenomenonwould not be expected,although rearrangements in localsystem organizationmight be. This is preciselywhat the archaeologicalrecord indicates.

STRUCTUREOF THE LOWLANDMAYA RESOURCE BASE

Our treatment of the structure of the LowlandMaya resource base is divided into three parts. First, we present data on Maya settlement patterns which indicate that populationdensities reached peak levels throughoutthe Lowlands in the Late Classic. Next, we discuss both archaeologicallyknown and potential 126 JOURNALOF ANTHROPOLOGICALRESEARCH modesof agriculturalproduction, focusing on agrotechnologiesapplied to slope- land, riverine, lake margin,and bajo environments.Third, we look at the locationand availabilityof other resources, particularlyhunted game, which may have providedcritical dietary supplements. Inherent difficulties faced by growingfarming populations utilizing the tropicalforest are discussed.These difficultiesare related to the inabilityof the agroecosystemto sustain high productionlevels over time withoutdegradation.

Settlementand PopulationDensity Table 1 providessettlement densities and populationestimates based on house-moundcounts for a broadsample of Lowlandsites. The evidence of highpopulation densities surrounding Classic Maya centers shows that exten- sive, shiftingcultivation probably could not have supportedmore than a fraction of the Mayapopulation. Studies of contemporaryswidden cultivators document that shiftingcultivation in the Mayaarea normallysupports less than eighty persons per squarekilometer (Cowgill 1962; Nationsand Nigh 1980). This figurefalls well below Classicperiod population estimates for the Lowlands basedon surveydata from Maya centers and their rural sustaining areas. While some scholars(e.g., Sanders 1973) have suggested that all house mounds mightnot have been occupiedsimultaneously, little actualevidence exists for noncontemporaneityor for seasonaloccupation of structuresfrom the heart of the southernMaya Lowlands. In fact, the structuralredundancy in the pattern of Mayaresidential settlement from center to more ruralperiphery suggests year-roundoccupation for most Classicresidences. Ceramicevidence from a numberof Lowlandcenters such as (Thomas1981), Tikal(Fry 1969; Culbertand Kosakowsky1985; Puleston1973), Seibal(Tourtellot 1982), and Yaxha(D. Rice 1978; Rice and Rice 1985) indicatesthat populationsreached a peak in the Late Classicwhen most, if not all, structureswere occupied. Highpopulation densities were not restrictedto the largestMaya centers and their immediateperipheries. House-mound counts from survey transectsbe- tween majorMaya centers in the Peten show thatpopulation densities in rural areaswere also high(Culbert n. d.). Settlement-patterndata from around Uay- milalso suggest continuouslyhigh mound frequencies across most of southern QuintanaRoo (Harrison1981, 1982). To the southin northernBelize, mound counts are againquite high (ca. 107 moundsper squarekilometer), even in the relativelyrural Late Classiccommunities surrounding (Harrison1983). Mayafarmers evidently occupied a densely populatedlandscape in a way moreakin to the dense settlementsystems of centralAfrica than to the pattern of nucleatedsettlement found in CentralMexico (Sanders 1976, 1981;Netting 1977). Populationincrease eventually led to a "fillingup" of the habitableland- scape, a process common to many archaic civilizations dependent on simple agriculturalproduction. According to Culbert(n. d.), the demographictrajectory at large Maya sites such as Tikalwas exponential, with populationsexperiencing curvilineargrowth throughout the Classic period. Continuous and fairly even ONTHE MAYA COLLAPSE 127

TABLE 1 Residential Structures and Population Density at Late Classic Lowland Maya Sites

StructureClusters Settlement or HouseMounds EstimatedPopulation Location (per sq. km.) Denisty(per sq. km.) Reference Site center 112 450 Puleston1974 Ricketsonand Ricketson 1937 Puleston1973 Dos Aguadas 222 Bullard1960 Tikal Center 235 713 Puleston1974 Periphery 145 - Carrand Hazard 1961 Surveystrip 119 600-700 Puleston1973 Haviland1966 MajorPeten sites Center - 600-1000 Puleston1973 Periphery - 200-300 Seibal Center 101 450-590 Tourtellot1982 Periphery 53 Tourtellot1982 Becan-RioBec region 134 503-670 Turner1983b Turner1974 Becan Site area 222 900 Thomas1981 BartonRamie (BelizeRiver) 131 735 Willeyet al. 1965 R.E.W.Adams 1981 397 2000 Kurjackand Garza T. 1981 Centralzone 309 1400 Folan,Kintz, and Fletcher 1983 Chunchucmil 400 Vlacek,Garza de Gonzales, andKurjack 1978 Copan 192 Willeyand Leventhal 1979 128 JOURNALOF ANTHROPOLOGICALRESEARCH habitationof the landscapemeant that Mayafarmers probably brought a ma- jorityof Lowlandhabitats into agriculturalproduction at the same time. That muchof the forest had been removedis supportedby palynologicalevidence fromthe Peten (Rice 1978) and northernBelize (Wiseman1978). Certaincomponents of the tropicalforest environmentwere transformed into highlyproductive and resilientagricultural systems, the raisedfield sys- tems in a numberof wetlandenvironmental settings being one example.Slope- landswere also intensivelycultivated; however, with the exceptionof kitchen gardeningnear residences, high yields cannotbe maintainedon intensively cultivatedslopeland plots, andtheir repeated cultivation may actuallyenhance degradationprocesses, thus severely limitingtheir productivepotential. The expansionof dense settlement was associatedwith an increasednumber of buildingsper unitof land,and this increasein the area coveredby structures, combinedwith agricultural intensification on shallowtropical forest soils, altered local hydrologyand acceleratedrates of erosion (Rice and Rice 1984). The extensionof systems of intensiveslopeland over a largeproportion of the landscape,then, wouldhave increasedthe susceptibilityof segments of the foodproduction system to degradationprocesses and decreased its capacity for resilience.Slopeland agricultural intensification and its adverse impacton cropyields and exacerbation of erosionalprocesses, however,cannot by them- selves explainthe timingand all-encompassingnature of the Mayacollapse.

Knownand PotentialAgrotechnologies Recent populationestimates for the Lowlandsattest to the presence of a dense populationspread over much of the availableagrarian land (Ashmore 1981). Concomitantwith the growthin our knowledgeabout Lowland Maya settlementpatterns and the rejectionof the swiddenhypothesis has been the accumulationof hardevidence that the ancientMaya utilized a complexand diverse range of agrotechnologies(Harrison and Turner1978). Throughout the variedhabitats of the Lowlands,it appearsthat Mayafarmers combined fieldmodification (where existing soil, moisture,and slope characteristicsof agriculturalplots are altered) with spatiallydiversified cropping techniques (where differentcrop mixes and fallowcycles are practicedwithin the same generalsettlement zone) in order to increasesubsistence yields and feed an increasinglydense population. Evidence of intensive agrotechnologiesin the Lowlandshas been docu- mentedat a numberof sites. These are listedin Table2. In lightof the obvious inadequacyof swiddencultivation to supportestimated population densities, potentiallyapplicable agrotechnologies are also suggested in Table2. These alternativemethods of subsistenceproduction may have providedthe Maya with increasedyields at varyingcosts in laborinvestment and environmental stability. Shifting agriculture is conducted on dryland outfieldplots which are located at some distance from the more densely settled residential areas. These fields are not irrigated, receive little or no fertilizer, and are structurally and func- ON THE MAYACOLLAPSE 129

TABLE 2 Known and Potential Agrotechnologies for Different Late Classic Lowland Maya Sites

Known Potential Site AgriculturalFeatures Agrotechnologies References Tikal terraces(near shiftingagriculture in Puleston1973 Flores) uplands;infield gardens; Turner1974 orchards;some terraces Uaxactun terraces(near shiftingagriculture in Puleston1973 Flores) uplands;infield gardens; Turner1974 orchards;some terraces Rio Bec terraces;raised shiftingagriculture in Turner1983b (Becan) fields uplands;terraces; raised R.E.W.Adams 1981 fields;infield gardens (farmsteads);orchards Northern raisedfields shiftingagriculture in Turnerand Belize uplands;raised fields; Harrison1983 (Pulltrouser infieldgardens; orchards Swamp) Pasionregion raisedor channelized intensiveflood water R.E.W.Adams 1983 (Seibal) fields farmingon alluvium; Tourtellot1982 raisedor channelized fields;shifting agriculturein uplands; infieldgardens; orchards Belize River terraces shiftingagriculture on Willeyet al. 1965 (Barton uplands;intensive flood Ramie) waterfarming on alluvium;infield gardens; orchards QuintanaRoo raisedfields shiftingagriculture on Harrison1981, 1982 uplands;raised fields; infieldgardens; orchards irrigationcanals shiftingagriculture on Matheny1978 N.W. (reservoirs) uplands;infield gardens; Yucatan irrigatedfields; orchards (Edzna) Coba kitchengardens shiftingagriculture on Folan,Kintz and uplands;raised fields on Fletcher1983 lakes;infield gardens; orchards;pisciculture 130 JOURNALOF ANTHROPOLOGICALRESEARCH tionallyequivalent to the contemporaryMesoamerican outfield milpa. Maya farmerscould have decreasedfallow periods on outfieldmilpa plots andturned extensiveswidden zones into areasof annualcropping. The problemsof main- tainingsoil fertility on these plots,though, greatly limit the long-termproductive capacityof this agriculturalstrategy. Productionlevels on drylandoutfield plots croppedin maizedecrease mark- edly throughtime because soil nutrientscritical for plantgrowth decline as parcelsare consecutivelycultivated. Measurements on experimentalplots in the Peten, for example,demonstrate that the amountsof nitrogen,potassium, phosphorus,and magnesium increase with each fallowyear, indicatingthat the amountsavailable for plantutilization decrease steadilywith each year of suc- cessive cultivation(Urrutia 1967). The amountof nitrogenavailable is partic- ularlyimportant because it controlsseed productivityand the proportionof proteinpresent and governs the utilizationof potassiumand phosphorus, which are essentialfor starchformation, seed maturation,and chlorophylldevelop- ment in leaf tissues (Donahue,Shickluna, and Robertson1971). Soil fertility dropsrapidly in the tropicsbecause most nutrientscritical for plantgrowth are either lockedup in the forest canopyor present only in the thin layer of leaflitter decaying on the surface.With removal of the forest, leafymatter no longeraccumulates on the surface,and a rapiddrop in the nutrientsrequired for plantproductivity occurs. Prolongeddeflection of the forest to cropplants, successionalspecies, or exposedearth also leads to a breakdownin soil struc- ture, soil compaction,loss of surfacepermeability, and reduced infiltration, all of whichaccelerate erosion. Croplosses also occurbecause weeds, insectinfestations, and plant diseases becomemore significantproblems the longera field is undersuccessive cul- tivation.These problemsare particularlysevere whenfields are monocropped or plantedin onlya few species. Most plantpredators in the humidtropics are specialists.Although biological diversity is high and the densityof each plant species is low in climaxsettings (Colinvaux1973; Margalef 1968), forest clear- ancegreatly reduces the numberof differentspecies presentand concentrates plant-specificpredators in the area aroundagricultural plots. Althoughthis processgenerally takes severalyears to be felt, as predatorsreact to changes in the density and distributionof plantprey, it ultimatelycontributes to re- ductionsin yields. Workloadsalso increaseas slopelandsin the humidtropics are used more frequently(Boserup 1965; Sanders1976; Steggerda1941). This increasein laboris primarilya functionof the need for more frequentcultivation as com- petitionfrom weeds and grasses becomes more pronouncedafter several consecutiveyears of landuse. Since weedy species are not numerouson the forest floorin climaxsituations, competition generally does not pose a serious problemduring the first year after forest removal.Their numbers, however, increasemarkedly thereafter because the extraamount of solarradiation reach- ing the surfacecreates optimalconditions for plantsuccession (Daubenmire 1968). Weeds also exhibitgrowth profiles comparable to cereals and hence ONTHE MAYA COLLAPSE 131 effectivelycompete for both criticalnutrients and space. Extra labor must thereforebe allocatedfor theirclearance in orderto maintainproduction levels. The total time devoted to weeding is often considerable,not uncommonly amountingto 50 percentof alllabor expended. More labor may also be devoted to fencingand field preparation before planting. So, as the frequencyof cropping increases, the overallefficiency of the system declines. Farmerscan hedge againstdrops in productivityby cultivatingmore space; but because labor expendituresper unitarea have risen, withintensification greater and greater constraintsare placed on the total amountof land farmerscan bring under simultaneouscultivation. Complexcropping techniques, where cultigenswith varyingmineral and nutrientrequirements are cultivatedtogether, could have been used to help maintainsoil fertilityand lengthenfield productivity(Norman 1979; Sanchez 1973, 1976;Wilken 1971). Mixedspecies gardeningis less susceptibleto land degradationby erosion and compactionbecause crop mixes tend to create a tieredvegetative cover whichshields the groundsurface from the deleterious effects of rainwater andbaking by the sun. Monocroppedfields, unlikemixed fields, are normallycropped at one time leavinga blockof landdenuded and subjectto compactionor mass wastage. Complextechniques in outfieldcon- texts are normallypracticed at lower populationdensities (Nationsand Nigh 1980). However,mixed planting on outfieldplots with shortened fallow periods eventuallyfaces the same process of decliningyields due to decreasedfertility and soil exhaustionas do monocroppedplots on permanentlyand semiper- manentlycultivated fields. Mixed cropping,therefore, could not have solved the Mayafarmer's long-term problem of decliningyields and decreasedpro- ductivityin areas where forest successionhad been permanentlydeflected to shortfallow. Mulchingand the depositionof nightsoil are two means of artificialenrich- mentby whichthe Mayamay have tried to maintainsoil structureand replenish nutrientslost throughfrequent cropping on outfields(Rice and Rice 1984). Mulching,we suspect, wouldnot have solved the problemof nutrientloss, since the prevailingmode of transportwas humanpowered and slopeland sources of mulchwere probablyfew in numberby the Late Classic.Literally tons of organicmatter would have been requiredfor each hectareunder cul- tivation,and the majorsource of this materialwould have been bajos. Since mulchingwould have required transport uphill by humancarriers, any increases in productionmay well have been offset by the extra laborexpenditures in- volved.The applicationof largeamounts of nightsoil, whileenergetically fea- sible at infieldgarden locations, is also impracticalat more distantoutfield locationswhere human-poweredtransport costs are high. Becauseof allthese problems,decreases in productivitywould have rapidly favored abandonmentof outfields subject to decreased fallow unless they were located on annuallyflooded alluvialbottomlands or on terraced upland slopes where soil nutrients are partiallyreplenished by the retention of slope-washed sediments. However, annualcropping on alluvialbottomlands and along natural 132 JOURNALOF ANTHROPOLOGICAL RESEARCH river terraces is areallyrestricted in the Maya Lowlandsdue to the limited amountof landavailable in narrowriver valleys and to bajoflooding in wider river courses. The terracingof uplandslopes, then, wouldhave been one of the prime means by which the Maya could have tried to increase outfield agriculturalproduction. Terracingslopelands is one way to checkerosional processes. Terracesalso promotesoil buildupand limit nutrient loss by trappinginorganic particles and leafymatter that wouldnormally be washeddownslope. Terracing may serve a varietyof otherfunctions, including moisture retention, protection from high winds,and retardation of phosphateloss by colluvialprocesses. Diverse crop- pingregimes, with field borders planted in economicallyuseful trees, is another way to inhibitsoil loss, because the species plantedform a multitieredcom- munityof vegetationand have differentroot system structures.Moreover, since the species plantedhave varyingphysical requirements, the utilization of soil nutrientsis maximized.The evidence for terracingfrom the Maya Lowlandsis substantialand suggests the extensive utilizationof this technique in the RioBec settlementarea, in the centralPeten, andalong the BelizeRiver (Donkin1979; Puleston 1973; Turner 1979, 1983a;Willey et. al. 1965). Short- ened fallowoutside of these specializedsettings wouldhave been subjectto decreasedyields per unit of laborinput over time and to acceleratedrates of erosionwhere agricultural plots were morefrequently denuded between plant- ings. The same problemswould have also occurredin areas like the Peten if polyculturalcropping was replacedby monocroppingin maizeduring the Late Classic. Erosionis most severe in situationswhere the slope angle is great, large contiguousareas of groundsurface are directlyexposed to physicalforces, and rainfallpatterns involve the depositionof large amountsof precipitation, often in torrentialdownpours. In the Basinof Mexico, for instance,archaeo- logicaldata suggest that a majorerosional episode occurredin the Guadalupe Rangeduring the MiddleFormative (Sanders and Santley 1977). This evidence of degradationis associatedwith shiftin settlementsfrom sloping localities in the piedmontto the alluvialplain and with the adoptionof floodwaterand canal irrigation(Nichols 1982; Santley 1977). The same sequence of events may have occurredin the TeotihuacanValley several centurieslater. We suspect thatthe GuadalupeRange eroded at such an earlydate becauselarge areas of slopingterrain had been denudedof vegetationfor dry farming,which accel- eratedrates of soil loss andforced villages to exploitalluvial bottomlands. Such erosionof short-fallowedslopeland agricultural plots wouldhave also had im- portantconsequences for the manipulationof intensivelycultivated settings in the MayaLowlands where interiordrainage is common. Archaeologicalevidence suggests thatflooded bajos, lake margins, and low- lying riverine settings were prime locations utilized by Maya farmers for the construction of raised and channelized fields. These are highly resilient and productive agrotechnologies that would have maintainedhigh yields on a con- tinuous or semicontinuous basis (R.E.W. Adams, Brown, and Culbert 1981; ONTHE MAYA COLLAPSE 133 Harrison1977, 1978; Siemans1978, 1982;Turner 1974; Turner and Harrison 1983). Raisedfields involve the excavationof a grid of canals to facilitate drainageand lower the watertable,with the soil piled next to the canalsto elevate the fieldlevel andcreate a relativelydry planting surface. Channelized fields, on the other hand, consist of long trenches excavated into wetland marginsto enhancedeeper water drainage and bring water to the dryerinterior soils. Incidentally,evidence for the actualcanal irrigation of drylandagricultural plots fromEdzna in Campecheis problematicaland suggests that this formof hydraulicmanipulation was rare in the Maya Lowlands(Matheny 1978). If managedproperly, raised fields can produce substantial yields each year without degradation,and it is possibleto schedulemore thanone crop per year. This situationis very muchunlike that on slopelands,where productionlevels drop each year undersuccessive use. The majordrawback of these systems is the large amountof laborrequired for yearlymaintenance: on the order of 200- 250 man-daysper hectareof cultivatedspace (Sanders,Parsons, and Santley 1979; Turner1983a). Earlierestimates of the area affectedby channelizationand field raisingin the MayaLowlands ran into the hundredsof thousandsof hectaresfor northern Belize, QuintanaRoo, andthe Peten. Subsequentground verification and his- toricalresearch has temperedthese estimates but not substantiallyreduced the probablespatial extent of these features (R.E.W. Adams, Brown, and Culbert1981; Millet Camara1984). Muck and soil rich in decayingorganic mattercan be excavatedfrom canalbottoms and appliedto field surfacesto maintainfertility. High levels of agriculturalproduction can be maintainedin- definitelyin these contexts, as is evidencedby the continuingproductivity of the Lake Chalco-Xochimilcochinampa system in the Basin of Mexico after morethan five centuriesof use (Armillas1971; Parsons 1976). Intensiveannual croppingon uplandsoils surroundingareas of wetlandcultivation would have hadlong-term negative effects on raisedand channelized agricultural systems. Slopelanderosion increases rates of downstreamsedimentation, slowly de- creasingthe area of land availablefor raised field cultivation.Sedimentation also reducesthe amountof organicmuck available for use as mulchto maintain soil fertility,which would have decreased yields on this criticaland highly productivecomponent of the Lowlandagroecosystem. Anotheragrotechnology available to the ClassicMaya was infieldgardening, a system of cultivationwhere arableland adjacent to residenceswithin settle- ments is broughtunder intensive cultivation.Infield or kitchengardening, a commonfeature of subsistencein the humidtropics, is an importantsource of food, medicine,condiments, building material, and fuel (E. Anderson1952; Kimber1966, 1973; Killion1985; Wilken1971). The ethnohistoricand eth- nographicrecord indicates that horticulturalpractices within settlements were widespread among the LowlandMaya (Landa1938; Lundell 1938; Redfield and Villa-Rojas1934; Roys 1943; Scholes and Roys 1968; Steggerda 1941; Wilken 1971). Puleston (1973), Folan, Kintz, and Fletcher (1983), and Sanders (1981) have provided convincingarguments for the importance of dooryardand infield 134 JOURNALOF ANTHROPOLOGICALRESEARCH gardeningat ClassicMaya settlements. Archaeologicalevidence for kitchen gardeningis also knownfor other partsof Mesoamericaduring the Formative period(Santley et al. 1984; Sheets 1983). The depositionof nightsoil andother household refuse acts to increasethe fertilityof arableland within settlements. According to AnnKirkby (1973:120), a familyof five individualscan produceenough manure each year to fertilize an area of aboutfour hundredsquare meters. If enrichmentwas necessary every two to three years, approximatelyone-tenth of one hectarecould have been fertilizedwith nightsoil in the immediatevicinity of the residence. The infieldareas between residences within Maya settlements therefore represent naturallymore fertile and agriculturallymore resilientsettings for cultivation thando drylandoutfield plots. Tourtellot(1982) suggests that approximately one hectareof cultivableland was availableto each residentialunit at Seibalin Late Classictimes. Table3 presentsthe meanamount of arableland available to residentialunits visible at the surfacewithin the residentialcores of a sample of Late ClassicLowland centers. Cropproduction on this unoccupiedland can be expectedto have givenhigher and more sustainableyields than did dryland settings where fertilizationwas energeticallyless feasible.While infield plots couldnever have providedall of the staple dietaryneeds of Lowlandsettle- ments, they wouldhave producedimportant food and nutritional supplements withoutsuffering the decreases in productivityto whichoutfield plots were subject. To recapitulate,Classic Maya farmingpopulations would have faced the problemof insufficientyields if they reliedsolely on drylandshifting cultivation at relativelyhigh population densities. Where feasible, terracing and raised or channelizedfield construction would have increasedthe overallproductivity of the agriculturalsystem. Infieldgardening and orchardproduction were appli- cable to virtuallyall Lowlandsettings. More intensiveforms of shiftingculti- vationwith variable rates of fallowcould have been practicedon the remaining segments of the landscapeat increasinglylower levels of productivityover time. TABLE3 Estimates of Cultivable Land per Residential Unit at Late Classic Lowland Maya Sites

Site Hectaresper ResidentialUnit Reference CentralTikal .16-.2 Puleston1983 PeripheralTikal .5-1.5 Puleston1983 Seibal .98 Tourtellot1982 Dos Aguadas .6 Bullard1960 Chunchucmil .25 Vlacek,Garza de Gonzales,and Kurjack 1978 .69 Sabloffet. al. 1985 Coba .81 Folan,Kintz, and Fletcher 1983 .25 Smith1962 (LateClassic) ON THE MAYACOLLAPSE 135 SecondaryResource Utilization Huntedgame, gatheredplants, and aquaticresources are importantcom- ponents in many food productionsystems because they providecritical nu- trientslacking in agriculturalcrops. Of all the secondaryresources utilizedin prehispanicMesoamerica, only three are discussedhere: deer, fish, and the proteinaceousalgae, Spirulina geitleri. In manylandlocked parts of Mexicothe white-taileddeer was by far the most importantspecies of terrestrialfauna hunted.Fishing, of course, wouldhave been restrictedto riverine,lacustrine, and marinesettings, while Spirulinaprocurement would have been confined to relativelysaline lake situations. White-taileddeer are browsersand thrivebest in woodlandand deciduous forest biomes (Flannery1968; Leopold1959). The deer biomasspotentially availableis primarilya functionof the amountof browsinghabitat present and the distributionof that habitatrelative to humanpredators. In manydensely populatedareas, agricultural land use modifieslocal plant synecology, producing a habitatoptimal for white-taileddeer. The degree to whichagricultural inten- sificationand modification of the tropicalrainforest environment affected Low- landdeer populationscan be demonstratedby comparingthe spatialdistribution of croplandand potentialgame preserves in the MayaLowlands with that in centralhighland Mexico. In CentralMexico the primaryzone underagricultural exploitation,the alluvialplain and the lower slopes of the adjacentpiedmont, was not a habitatin whichwhite-tailed deer thrivenaturally in the wild(Sanders, Parsons, and Santley 1979). Agriculturalintensification thus wouldhave had little negativeimpact on the density of woodland/forestbrowsers once canal irrigationbecame the dominantmode of agriculturalsubsistence in Classic times. In fact, intensificationmay have actuallyincreased the numberof deer availablelocally by creatingbrowsing habitats on fallowagricultural land on the alluvialplain, while maintaining reserves in uplandforests. In the MayaLowlands, on the other hand, slopelandagriculture gradually resultedin the removalof the forest andthe destructionof the habitatof many species of terrestrialfauna. Although intensification may have increasedthe densityof certainanimal species as moreand more of the forest was converted intovarious stages of regrowth,eventually white-tailed deer musthave become less plentifulas huntinglevels exceeded rates of replacement.It is likely, therefore,that by Late Classic times many species of terrestrialfauna had been almosttotally decimated throughout large portions of the southernLow- lands. Hence, while spatialsegregation of agriculturalzones and terrestrial faunalpreserves may have acted to maintainat least some game stocks in CentralMexico, congruity in the distributionof agricultureand game territories in degradedLowland forest habitatseventually led to the virtualextinction of criticalgame reserves. Althoughrodents and other smallmammals may have providedsome meat, the Late ClassicMaya, we believe, would have been hard pressed to find the necessary supplements of animal protein in their agriculturallydominated landscapes. Fish and aquaticmammals are another source of protein. The estuarine and 136 JOURNALOF ANTHROPOLOGICALRESEARCH marinehabitats of the Gulfof Mexicoand CaribbeanSea supportsome of the richestsaltwater fisheries in the New World.Fish are especiallyrich in protein, andthe kindof proteinpresent is of highbiological value (Church and Church 1975). Althoughmarine fauna are consumedtoday throughoutMexico, their widespreadutilization is primarilydue to recentimprovements in transportation and refrigerationtechnology. Archaeological evidence indicatesthat aquatic faunawere a majorcomponent only in the diet of coastalpopulations (Hamblin 1984; Healy 1983; Wing1978). Althoughnot manyarchaeofaunas have been describedfor sites frominland , those few thathave been reported containvery few marineresources (Drennan 1976; Pohl 1983;Starbuck 1975; Whalen1981). Whilemarine fish could have been filleted, dried, and then shippedinland, there is no evidencethat the Mayadeveloped such a production- distributionsystem. Somefish were probablyobtained from perennially flooded , rivers, and inlandlakes, but given the high populationdensities Mayanistshave suggestedfor the Late Classic,the totalamount of freshwater faunaconsumed per capitawas undoubtedlyquite small. Besides fish, lakes and marshesmay also provideother resources rich in criticalnutrients. The proteinaceousalgae, Spirulina geitleri, was used by the Aztecs as an importantadditive to many sauces (Furst 1978). By weight Spirulinacontains 65-70 percentraw protein, of which84 percentis digestible; andtryptophan, the essentialamino acid lacking in maize-based diets, is present in quantitiesthree times thatof deer andtwenty times thatof corn(FAO 1970; Pirie 1975a; Santley and Rose 1979). Also, Spirulinahas a very favorable leucine/isoleucineratio and is rich in salt and in manyvitamins. Consumption of onlyminor amounts of this resourcetherefore would have allayedmany of the nutritionalproblems associated with a maize-baseddiet. However, Spi- rulina only grows in highlyalkaline lakes (pH 9-10) and consequentlywould not have been present in the MayaLowlands. Chlorella and other species of blue-greenalgae rich in protein,essential amino acids, andother nutrientsdo thrivein the freshwaterlakes and marshes which abound in the MayaLowlands. Experimentalresearch, however, indicates that they producesevere diarrhea (Pirie1975b). The physiologicalutility of such species is thus quitelimited, so it is doubtfulthat they were ever utilizedas a dietarysupplement by the ancient Maya. Throughtime, the juxtapositionof decliningyields on intensivelycultivated agriculturallandscapes and the increasinglack of terrestrialgame, fish, and aquaticresources posed insurmountableproblems to Mayafarmers unable to open up new segments of the landscapefor agriculturalproduction. The high Lowlandpopulation densities suggest that virtually all segments of the Lowlands suitablefor settlementhad been filledby Late Classictimes. Hence, expansion of the system was impossiblebecause most of the surroundinglandscape was already occupied by similar groups facing the same problems. Late Classic Maya populationswere consequently faced with the difficultproblem of solving their food productionneeds with the resource base at hand. ON THE MAYACOLLAPSE 137 THE CLASSICMAYA COLLAPSE

In our opinionthree variablesplayed major causal roles in precipitatingthe Mayacollapse: (1) demographicinstability, (2) agriculturalfailure, and (3) the absenceof macroregionalresource extraction structures. The generalpicture of LateClassic Maya subsistence outlined above involves a multiple-technology agriculturalsystem greatly relianton terrace or short-fallowagriculture and raisedfields but probably augmented by kitchengardening, orchard cultivation, andperhaps pisciculture in freshwateraquatic habitats. Large portions of the cultivablelandscape were probablyexploited at high levels of intensification, andagricultural production was undoubtedlygeared to a few staples,with other plantresources, meat, and fish consumedonly in small amountsas dietary supplements.

NutritionalStress, Disease, and DemographicInstability Accordingto Don Rice (1978:55), "By the end of the Late Classicperiod the productionof allfoodstuffs, including, to some degree, ramon,was probably strained."The staples whichwere probablygrown were rich in caloriesbut notoriouslypoor in other nutrients.A maize-baseddiet, supplementedwith beans, squash,and other plants, does not containall of the nutrientsand amino acidsnecessary for a well-balanceddiet (Santleyand Rose 1979). Whilebeans makeup for the inadequateamount of the essentialamino acid lysine present in corn, both resources are deficientin tryptophan.The metabolismof tryp- tophanis closely linkedto the endogenousproduction of niacin(Aguirre et al. 1953). If the tryptophanin maize is not supplementedfrom other sources, severe niacindeficiencies will develop. In addition,the levels of methionine and phenylalaninepresent in maize are barelyadequate, and the leucine/iso- leucineratio is distinctlyunfavorable, which interferes with the metabolismof what tryptophanis present. The proteinin maize is stored primarilyin two classes: zein, whichis readilyavailable in a dilutesaline solution, and glutelin, whichis largelyunextractable by the bodyunless the maizeis processedwith alkali(Katz, Hediger,and Valleroy1975). The zein portionis extremelyde- ficientin lysineand tryptophan, while the glutelinfraction is richin high-quality protein. Processingmaize in lime liberatesthe glutelinfraction, as well as improvingthe leucine/isoleucineratio. Although maize has moderateamounts of iron, it also containsphytic acid (phytate)which inhibitsiron absorption (Ashworthet al. 1973; El Najjaret al. 1976). Successfulhemoglobin synthesis willtherefore be significantlyimpaired unless iron is suppliedfrom other sources. Becauseof the limitingeffects of phyticacid, maize-based diets are alsodeficient in zinc (Lloyd,McDonald, and Crampton1978). Root crops and ramonhave also been suggested as staples for the Maya (Bronson1966; Puleston 1968), but whether these foodstuffswere ever utilized in significantamounts is undetermined.However, even if ramonand root crops comprisedthe mainstayof Mayasubsistence, they wouldnot have allayedthe 138 JOURNALOF ANTHROPOLOGICAL RESEARCH nutritionaldeficiencies just discussed.Root crops, whilerich in carbohydrates, provideonly very limitedamounts of other nutrients,especially protein (Wu Leung 1961). Ramon,contrary to Puleston'sassertions, is not significantly richerthan maize in most nutrientsexcept calcium.Ramon seeds contain11.4- 13.4 percent proteinby weight comparedto 8-10 percent for maize on the average (Puleston 1968; Wu Leung 1961). Increasedreliance on beans is anotherpossibility, but beans are not particularlyhigh in tryptophanand me- thionine(Church and Church1975). These essential aminoacids are amply representedin fishand meat. But fish, it wouldappear, were consumedmainly by Mayaliving in coastalhabitats, one partof the Mayalandscape which never experienceda demographiccrash as severe as that whichhit landlockedsec- tions of the southernLowlands. And deer are not particularlyabundant in forests in the humidtropics (Leopold 1959). Whileforest clearancemay have increaseddeer densitiesand deer can withstandhigh levels of predation,it is unlikely,given the highpopulation densities have suggestedfor the southernLowlands (Culbert n. d.), thatthey accountedfor more thana fraction of one percentof all the foodstuffsconsumed in the Late Classic.In the Basin of Mexico, for example, the dietarycontribution of white-taileddeer would have amountedto only 1.65 grams/person/day(or about 0.3 percent of the diet), assuminga humanpopulation of 250,000, optimaldeer densities, no overkills,no importedgame, complementarypredation by wolves and large cats, andincreases in deer numberson fallowagricultural land (Sanders, Par- sons, andSantley 1979; Santleyand Rose 1979). Levels of intakein the Maya Lowlandswere probablyeven less since humanpopulation densities were probablygreater there thanin the Basinof Mexico. The diet of the Late ClassicMaya thus involveda heavy focus on domes- ticatedgrains. Maize was the principalfood, supplementedby smallamounts of beans, squash, ramon, tropicalfruits, and perhapsroot crops. Although adequateamounts of raw proteinwere probablyingested, the biologicalvalue of that proteinwas undoubtedlyquite low. It seems likelythat levels of tryp- tophanand methionine consumption were grosslyinadequate. These deficien- cies, together with the unfavorableleucine/isoleucine ratio and the intakeof largequantities of phytate,produced niacin, zinc, andiron deficiencies. Nutritionalproblems of this sort havegreat impact on the structureof human populations.The relationshipbetween nutritionand human reproductive per- formanceis dual-directional.On the one hand,a female'snutritional status has an effect on her chances of conceiving,upon the outcomeof pregnancy,on the probabilityof infantsurvival, on birthinterval, on age of menarcheand menopause,and on her life expectancy.Rates of fetal wastage and infant mortalityand the probabilityof a femaledying upon reaching child-bearing age are muchhigher in populationswhich are malnourished(Delgado et al. 1978; Bongaartsand Delgado 1979; Thomson and Hytten 1973). Malnourished women also exhibit later age at menarche, earlier age at menopause, and longer birth intervals than do females who are better nourished (Carael 1978; Frisch 1975, 1976; Stein and Susser 1975). On the other hand, births that are spaced too ON THE MAYACOLLAPSE 139 closelyhave a negativeeffect on the mother'snutritional status. This syndrome, known as "maternaldepletion," is cumulativewith each reproductivecycle (Jelliffe1973). Dietarydeficiencies in iron and zinc may play a majorrole in impairingsuccessful reproductive performance and enhancing rates of miscar- riages, stillbirths,and infantmortality (Brasel 1978; Lloyd, McDonald,and Crampton1978). Zincalso plays an importantrole in cell-mediatedimmunity, and both parasiticinfection and malnutrition,common problems in the humid tropics,lead to increasedzinc depletion, which affects the body'simmunological response (Mata1984; Solomonsand Shripton1984). Dietaryproteins are perhapsthe most importantnutrients ingested because they supplythe aminoacids andnitrogen necessary for the synthesis of body proteins and other nitrogen-containingcompounds (Lloyd, McDonald,and Crampton1978). Amino acids are essential for growth, development, gestation, lactation,and the productionof enzymes, antibodies,and serum proteins.Of the twenty-twoamino acids of biologicalimportance, eight are essential;i.e., they mustbe ingestedin adequateamounts and in properproportion or protein deficiencyresults. As indicatedabove, a majorproblem with whichthe Maya hadto dealwas the limitedamount of the essentialamino acid tryptophan they ingested, combinedwith the low level of their methionineintake and the un- favorableleucine/isoleucine ratio prevalentin their maize-richdiets. Protein deprivationalso aggravatesrates of infantmortality and often results in lower weightbabies and sometimesin infantswith "nutritionalbrain damage" (Bon- gaartsand Delgado1979; Jelliffe 1973). In addition,"age at weaningmay be earlierin populationssubsisting largely on grainsand utilizing boiling techniques for preparingfoods thanin populationswhose diet containsless carbohydrates and a largerproportion of meat"(Santley and Rose 1979:190).Thus, early weaning,while returning the motherto the reproductivepool, may also enhance rates of infantmorbidity and mortality, since the weaningdiet is often lacking in proteinand a variety of essential vitaminsand minerals.Diets lackingin meat, fish, fowl, anddairy products are also deficientin vitaminB12 which leads to perniciousanemia, neural disorders, psychotic behavior, and premenopausal infertility(Brasel 1978; Harris1985; Horrobin1971). The disease burdenof a populationis anothermajor factor which impacts demographicstructure. While there is no evidenceto suggest that the density dependantinfectious pathogens which devastated Native American populations followingEuropean contact were presentin the New Worldprior to A.D.1492 (Crosby1972; Cockburn1971; Black1975; Newman 1976), it wouldbe highly misleadingto describethe pre-ColumbianAmericas as a disease-freeparadise (Dobyns1983). Evidencederived from human coprolites (Horne 1985; Rein- hard,Ambler, and McGuffie 1985), mummifiedsoft tissue (Allison1984), and skeletalremains (Buikstra 1981; Cohen and Armelagos1984) demonstrates that prehistoric populations of the New Worldwere subject to parasitic infec- tions, malnutrition,and disease-related pathologies. However, since the num- ber of material indicators which implicate specific disease organisms is quite limited (Goodmanet al. 1984; Horne 1985), retrodictions of the specific disease 140 JOURNALOF ANTHROPOLOGICALRESEARCH organismspresent in any given regionare difficultto evaluate.Nevertheless, it is possibleto gainsome appreciationof the relativeimportance of infectious disease in the MayaLowlands and the potentialimpact of that disease burden on Mayademography. Dunn(1968) suggests thatcomplex ecosystems, suchas tropicalrainforests, supportmore species of parasiticand infectious organisms and more potential vector species and intermediatehosts than the simplerecosystems of more temperateregions. The burdenof disease on humanpopulations, then, is greatest in the tropics (Neel 1982; Nutels 1968; Dunn 1968; Walsh1984; Binfordand Chasko 1976). Infectiouspathogens, consequently, were probably animportant element of the Mayaecosystem and a potentialfactor contributing to the collapseof Mayacivilization (Shimkin 1973; Saul 1973). Shimkin(1973) has suggested that the Mayamay have faced a numberof diseases, specificallyendemic infections of Americantrypanosomiasis, Ascaris, and the enteropathogenswhich produce acute diarrhealillness. A numberof humanactivities and environmentalvariables could have intensifiedthe expo- sure of Mayapopulations to these diseaseorganisms (Anderson and May 1982; R. Anderson1982; Lycett 1985). Becausemany of the insectvectors of human pathogensspecialize in the exploitationof disturbedenvironments, environ- mentalchanges resultingfrom agriculturalintensification and settlement ex- pansionwould have increasedthe prevalenceof pathogensand thereby the probabilityof infection.Furthermore, contact between vector species andhu- manpopulations is maximizedwhen other potential hosts arereduced in number (Speilmanand Rossignol 1984). Thus, as the abundanceof Lowlandterrestrial faunadeclined through habitat destruction and human predation, the probability of humaninfection by vector-bornediseases wouldhave increased. Whenparasites are endemicin an environment,as they are in the Maya Lowlands,infection is persistent.Newborn infants are partiallyand temporarily protectedfrom disease by antibodiestransferred from the motherbefore birth and via breast milk (Carpenter1984). Subsequentdevelopment of immunity dependsupon exposure to the pathogenspresent in the child'senvironment. Infectionskill some children,possibly those with genetic, nutritional,immu- nological,or other deficiencies(Burnet and White 1972; Mata 1984); but by age three, most survivingchildren have acquiredimmunity to those pathogens whichare commonin their environment.Development of immunityis condi- tionedby boththe relativeabundance of a parasiteand host populationdensity. Thus, childrenin crowded,disease-ridden environments develop more anti- bodies at an earlierage (Fenner 1970; Burnet and White 1972; Carpenter 1984), and mortalityis concentratedin the very youngand the aged. Those age groups with the highest fertilityrates are also those with the lowest mortalityrates (Barclay1958). If, however,childhood mortality is sufficiently high, disease may well regulatethe size and growthrate of the population (Anderson and May 1979; May and Anderson 1979; May 1984; Kunstadter 1972; Chagnon and Melancon 1983). Under these conditions, nutritionalstress often becomes a majordeterminant ONTHE MAYA COLLAPSE 141

of the rate of infantmortality. The relationshipbetween nutritionand infection is synergisticin that malnutritionexacerbates the severity of infectionswhile infectionaggravates nutritional deficiencies (Scrimshaw, Taylor, and Gordon 1968;Mata 1984). These conditionsare morefrequent and severe in pregnant women, infants,and young children.Because disease may severely reduce energy and proteinabsorption, infection-related malnutrition may occureven when the food supply is adequate. Chronicmalnourishment diminishes im- munologicalcompetence (R. Anderson1981), and even mildlymalnourished populationsexperience increases in chronicinfections, the severityof disease, andmortality due to disease (Torunand Viteri 1984; Mata 1984; Keusch 1984). Our reconstructionof Late Classic Maya food productionand procurement systems suggests a diet deficientin iron, zinc, niacin,and several amino acids necessary to proteinsynthesis. Immunedeficiencies appear to be more fre- quent, diverse, and severe in individualssuffering from proteindeprivation (Keusch1984), but iron and zinc deficienciesalso producealterations in im- munologicalfunction. Both parasiticinfections and malnutrition may lead to an increase in nutrientdepletion and, thus, intensifyiron and zinc deficiencies (Bakerand Jacobs 1984; Mata 1984; Solomonsand Shripton1984). Late ClassicMaya populations would have been subjectto infectionfrom a large numberof parasiticorganisms, the most devastatingbeing a series of entericpathogens which cause acute diarrhealillness. Suchillnesses are cur- rentlythe leadingcause of childhoodmortality in the developingworld. Although Escherichiacoli and rotavirusesare the most commonagents, infectionwith a widevariety of bacterial,protazoan, and helminth species mayproduce acute diarrhea.High population density and poor sanitationconditions may increase exposureand, thus, the probabilityof infection.Under such conditions, children will developmultiple episodes of diarrheain the first few years of life. While those who survivedevelop immunity to the majorityof enteropathogenspres- ent in theirenvironment, episodes of diarrhealillness contribute negatively to the nutritionalstatus, growth,and development of the child(Carpenter 1984; Guerrant1984). In additionto enteric pathogens,populations of the MayaLowlands would havebeen exposedto a varietyof vector-bornemicroparasites, macroparasitic nematodes,and other infectiousdisease organisms.These organismsand the infectionswhich they producestrongly influence and are, in turn, influenced by the nutritionalstatus of the host. The results of this synergisticrelationship includemalnutrition, retardation in growthand development, and lowered im- munologicalcompetence. Specific nutritional deficiencies of the Late Classic Mayadiet wouldhave exacerbatedthese problems.Mortality and morbidity levels amongpregnant women and those in their child-bearingyears, infants, andyoung children were probablyvery high. Furthermore,it is likelythat the health status of the population as a whole was very low. As the interaction between malnutritionand infectious disease intensified, physiological stress would have increased, and rates of infant morbidity and mortality would have become progressively higher. 142 JOURNALOF ANTHROPOLOGICALRESEARCH Whiledisease and malnutritionwould not have been responsiblefor the removalof the vast majorityof the Maya, escalationof the trends discussed abovemay have resultedin a loss of reproductivepotential sufficient to impair the capacityof Mayapopulations to replacethemselves. High levels of juvenile mortalitymay have producedcyclical or even chaoticfluctuations in population size. This demographicinstability would have been exacerbatedas depleted cohortsattained reproductive maturity. Recovery from these short-termfluc- tuationswould be expected, but only severalgenerations after the initialper- turbation. Variabilityin demography,pathology, and nutritional status has consequences whichmay be detected in skeletalseries. For example,increases in age- and sex-specificmortality should be reflectedby dramaticrises in the proportion of all deaths that occurredin each cohort, and these changes in mortality schedulesought to covarywith a reductionin the maximumstature attained by adultsand with an increasein the frequencyof nutritionallylinked paleo- pathologiessuch as porotichyperostosis and dental antimere asymmetry (Gil- bert andMielke 1985; Goodmanet al. 1984; Perzigian1977). Moreover,the stresses incurredought to be chronic(rather than episodic),which should retardoverall physiological development and limit the maximumstature attained (Nickens1976). This is in fact what the skeletal data from several LowlandMaya sites indicate.At Tikala markedreduction in noneliteadult male statue is evident in the Late Classic (Haviland1967), a patternthat also occurs at Altarde Sacrificios(Saul 1973), BartonRamie (Willey et al. 1965), andperhaps Copan (Longyear1952). At Tikalmortality was particularlyhigh for infants and females enteringchild-bearing age (Haviland1972), whichalso seems to have been the case at Copanin the Late Classic(Sanders, personal communication). At (Saul 1973) and Copan(Story, personal communication) the in- cidence of porotichyperostosis increases markedlyby Late Classic times. Porotichyperostosis, most authoritiesnow agree, occursin populationssuf- feringfrom chronicanemias, which are often caused by inadequateamounts of iron intake, althoughprolonged breast-feeding, diarrheal infections, and parasiticdisease infestationsmay also play a role (Walker1985). As we sug- gested above, irondeficiency was probablya majornutritional stress facedby the ClassicMaya. Irondeficiencies would have become particularlysevere if the Mayadiet becameincreasingly maize dependent as populationlevels rose throughoutthe Classic.In addition,the incidenceof lineardental enamel hy- poplasiais fairlyhigh for all time periodsat Altarand Seibal, indicatingthat short-termdevelopmental arrests from disease and/orinadequate nutrition were also a commonproblem (Saul 1973). The dataat handtherefore suggest that the Late ClassicMaya may have had considerable difficulty maintainingdemographic stability. It seems likely that this instabilitywas largely the result of a collection of nutritionaldeficiencies synergistically linked to a variety of parasites which collectively placed pro- gressively greater constraints on the number of individualssurviving to re- productive age. These problems, we believe, were mainly a function of the ON THE MAYACOLLAPSE 143 high levels of basic staples consumedcombined with the exceptionallylow levels of meat intake. Access to animalprotein was a factorwhich also affectedthe demographic historyof other parts of Mesoamerica.The populationgrowth curve of the Basinof Mexico, for example,is logistic-like;growth was exponentialduring the Formativeperiod, but populationlevels tapered off duringthe Classic (Santleyand Rose 1979). Coincidentwith the declineof ,however, there was a rearrangementin population,not wholesale demographicloss (Sanders,Parsons, and Santley1979). The demographichistory of the Basin of Mexicois thereforequite differentfrom that of the MayaLowlands. This variability,we suggest, is related to differencesin the respective resource bases whichhad a differentialimpact on the magnitudeof the stresses involved. In the MiddleClassic period, the frontierof Mesoamericancivilization lay directlyto the northof the Basin. One propertyof frontierlocation is access to both local resources and those from areas beyond the frontier.This is especiallythe case when the frontierboundary is permeable,a characteristic whichoften defines the boundariesbetween states andunstratified sociocultural systems (Greenand Perlman1985). The regionfrom which Teotihuacan and its dependenciesderived faunal resources was thus probablyconsiderably larger thanthe area occupiedby the core Teotihuacanpolity. Maya polities, in con- trast, circumscribedone another,a situationwhich must have greatly con- strainedhunting ranges except in instanceswhere one system was able to momentarilyexercise hegemonyover another.The spatialoverlap of agricul- turalland and natural game preserves discussed above further acted to decrease game reserves and limitexpansion of the LowlandMaya resource net. The periodof continuousgrowth during the Formativeperiod in the Basin in Mexicois associatedwith comparativelylonger life expectancyat birth(ca. 25 years) and a diverse diet relativelyrich in meat (Sanders,Parsons, and Santley1979). Although the paleopathologicalstatus of the inhabitantsof Teo- tihuacanhas yet to be definedin detail, there is reason to believe that the slowdownin growthrates in the Classicwas in parta functionof disease and nutritionalstress (Santleyand Rose 1979).According to Goodmanet al (1984:17), "ageat deathstands as perhapsthe most importantsingle indicator of stress." At the Tlajinga33 apartmentcomplex at Teotihuacan,life expectancyat birth was 13.5-16.5 years (Sanderset al. 1982; Storey 1983, 1985). This contrasts markedlywith the pooled estimate for El Arbolillo,Ticoman, and Zacatenco, whichsuggests significantlylonger lifespans(Santley 1977). The Tlajinga33 compoundwas occupiedby persons who had a relativelylow socioeconomic position,which is reflectedby their relianceon adobe architectureand by a very impoverishedfaunal assemblage (Widmer, personal communication). While Storey (1983, 1985)has suggestedthat immigrationfrom the countrysidewas the primarymechanism which kept the city demographicallyalive, the estab- lishment of many rural sites throughout the Basin of Mexico in the Middle Classic indicates that a substantial amount of migration in the other direction also occurred during Tlamimilolpatimes. The migration of young adults from compounds like Tlajinga33 and La Ventilla would result in their removal from 144 JOURNALOF ANTHROPOLOGICALRESEARCH skeletalpopulations interred in the city, makinglife expectancyat birthseem very low. Thus, whiledisease andmalnutrition may have hadgreat impacton the demographicstability of certaincomponents of the Teotihuacansettlement system, the total amountof such stress affectingthe polity as a whole was probablymuch more moderate, at leastin comparisonwith the MayaLowlands. Althoughgrowth curves vary from regionto regionin the southernMaya Lowlands,population levels generallypeak in the Late Classic(Figure 2), with manyareas subsequently undergoing a catastrophicdemographic crash (Culbert n.d.; Culbertand Kosakowsky1985; Fry 1985; Rice and Rice 1985). As we havepointed out, the landscapein whichthese demographiccollapses occurred was one in whichmeat and alternativesources of high-qualityprotein were almost entirelylacking. Although elite nutritionwas undoubtedlysomewhat better (as indicatedby taller adultstature), all componentsof were probablyhard pressed by the end of the Late Classic.The reductionin mean stature in the Classic stronglysuggests that Maya populationswere exposedto increasinglymore severe stress. We suspectthat the diet available to the averageMaya peasant throughout most sections of the Lowlandswas worse thanthat of the averageurban dweller at Teotihuacan.Life expectancy at birth,then, shouldhave droppedmarkedly as the diet becameprogressively poorer.Ultimately, age-specific mortality rates mayhave increased to the point whereparents would have hadgreat difficultyraising children to maturity,and because all parts of the Mayalandscape were experiencingthe same set of

CENTRAL RURAL P'TROWSER TIKAL TIKAL BECAN SWAMP

RELATIVEPOPULATION (% OF MAXIMUM) 120

100-

80- .

60

40

TIME PERIOD

Figure2. Populationthrough Time at selected LowlandMaya Sites ONTHE MAYA COLLAPSE 145 problems,the size of the populationas a whole beganto decline.We suggest thatthis processbegan sometime during the eighthcentury A.D., if not slightly earlier,and was well underway by the beginningof the ninthcentury.

Degradationof theAgricultural Landscape We believe that Late Classic Maya populationsmade criticalshort-term decisionsto reorganizeland use strategiesin the face of demographicloss and decliningyields fromslopeland cultivation. These decisionshad long-termef- fects on the viabilityof the agriculturalresource base. In order to modelthe system failure,first let us assume that, in response to rises in population density,most partsof the Mayaenvironment were intensivelyexploited. This intensificationresulted in the almost wholesaleremoval of the climaxforest and many successionalspecies from vast sections of the interiorLowlands. Kitchengardens, planted in a variety of economicallyuseful species and en- richedwith nightsoil and organichousehold refuse, were placednext to res- idences. Slopelandswere broughtunder short-fallow agriculture and/or annual cropping,were plantedmainly in staples (probablymaize), and were terraced to retarderosion and other soil degradationprocesses. Whereverpossible, lakeshoresand bajo margins we