Paleolimnology of the Maya region

PALEOLIMNOLOGY OF THE MAYA REGION

By

\MARK BRENNER

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

1983

ACKNOhLEDGEMENTS

I would like to thank the many people who have helped me complete

this project. First and foremost; .I thank my committee chairman, Dr.

E.S. Deevey, for guidance and encouragement. Dr. Deevey.introauced me

to paleolimnology, and I am indebted to him for nurturing my

fascination with the lacustrine sediment record. I appreciate the

input of Dr. Frank Nordlie, who taught me the fundamentals of

limnological research. I am grateful to Drs. Tom Crisman, Carter

Gilbert, and Doug Jones for reviewing the thesis and for teaching

courses that inspired some of the ideas in this work.

The multidisciplinary nature of this paleoecological study

demanded the collaboration of a large number of researchers. I was

reliant on assistance from many individuals, and I want to acknowledge

their aid. The success of the 1978 coring operation was largely aue to

Sid Flannery's help. He also helped produce some of the core chemistry

data. Dr. Sam Garrett-Jones also collaborated on the 1978 drilling

campaign and generated pollen data. Dr. Hague Vaughan provided me with

instruction in paleolimnological techniques and was responsible tor

producing several pollen profiles.

I have fond memories of the many months spent in Peten with Drs.

Don and Prudence Rice. I am grateful.to them for allowing me to use

both their published and unpublished archaeological data. Special

thanks are due Pru for delivering me to the hospital in Guatemala city

during my 1979 bout with hepatitis.

Dr. Mike Binford's presence in Peten in 1980 made that a most

productive field season. Mike provided granulometric data from Peten

cores and listened patiently to many of my ideas, answering nurneious

questions along the way. I am also grateful for the feedback obtained

from other members of the Paleoecology Lab: bob Snodgrass, Tom

Whitmore, and Antonia Higuera-Diaz.

Carl Miles and Andy Ogram instructed me in the use of the atomic

absorption spectrophotometer and the Leco induction furnace. Liz

Fisher provided encouragement every step of the way. her aid is

gratefully acknowledged. Rhoda Bryant typed the final manuscript, and

her aid is certainly appreciated.

The successes of the "Historical Ecology of the Maya" project have

been largely attributable to the cooperation of many Guatemalans.

Special thanks are due the following friends tor assistance far beyond

the call of duty: Rafael and Clemencia Sagastume, Jaime and Mari

Sobalvarro, Antonio and Aura Ortiz, and Robert Dorion. I thank the

many residents of northern Guatemala who have made me feel at home in

Peten.

Finally, I would like to thank the Peten. This study sheas some

light on Peten's prehistory. Many questions remain unanswered, and the

region, with its spectacular Maya ruins, tropical forests, and lakes,

continues to captivate my imagination. I look forward to visiting and

working in the area again.

This project was supported by grants to Dr. E.S. Deevey (NSF DEB

77 06629, NSF EAR 79 26330, and EAR 82 14308). A graduate research

assistantship from the University of Florida Division of Sponsored

Research is also gratefully acknowledged.

TABLE OF CONTENTS

ACKNOWLEDGEMENTS . . . . . . . . .. . .... ii

ABSTRACT . . . . . . . . ... . . . . . vii

INTRODUCTION . . . . . . . . ... . . ..... 1

Prehistoric Maya and Contemporary human Populatiuci
Densities of Peten, Guatemala . . . . . . . 1
Maya Cultural Development in the Tropical Forest Ecosystem .
The Origins of Maya Civilization . . . . . . ... 6
Maya Agricultural Practices . . . . . . . . .. .10
The Classic Maya Collapse . . . . . . . . ... 14
The Historical Ecology of the Maya . . . . . ... .17
Measuring Maya Environmental Impact Paleolimnologically . 17
Lake-Watershea Interactions . . . . . . ... .18
The Paleolimnological Perspective . . . . . . 22
The Contemporary Peten Environment . . . . . . . 26
Geology ................ ....... .. 26
Climate and Rainfall .................. 28
Soils . . . . . . . . . . . . . 29
Vegetation . .. . .. . .. .. . .. ... 32
Regional Limnology . . . . . . . ... .. .36
Maya Settlement in the Yaxha ano Sacnab Catchments . .. .. 45
Maya Demography . . . . . . . . .. . . 45
The Twin Basin Seoiment Record . . . . . . . 49
Phosphorus Loading of Lakes Yaxha and Sacnab . . .. .52

THE HISTORICAL ECOLOGY OF THE MAYA AT LAKES QUEXIL, SALPEiEN,
AND MACANCHE . . . . . . . .. . . . 55

The Archaeological Record . . . . . . . . ... 55
Soil Chemistry . . . . . . . ... ..... .65

PALEOLIMNOLOGY OF LAKE QUEXIL . . . . . . . . .. 83

Comparing Shallow-water and Deep-water Sedimentation ... .83
Proximate Chemical Composition of the Lake Quexil
Sediments ................... .... 97
Chemical Accumulation Rates in Lake Quexil Sediments . 1U2

Paleoproductivity in Lake Quexil . ... ........ .... . .. 11
The Significance of Microfossils . . .. . . . .... 11b
The Microfossils of the Quexil Cores . . . . .. 120
Microfossil Accumulation Rates and Phosphorus Loading . 141
carbon-Nitrogen Ratios in Quexil Core H . . . ... 152
Comparing Microfossil Accumulation Rates at the Two
Quexil coring Sites.. .. . . . .. ... . 154

TESTING THE PHOSPHORUS LOADING MODEL DEVELOPED AT YAXHA-SACNAB .160

Coring in Lakes Salpeten, Macanche, and Quexil, 1980 ... ... .160
Sediment Chemistry of the Macanche and Salpeten Cores . 175
Assessing the Maya Annual Per Capita Phosphorus Output ...... 176
Zoning the Cores Chemically . . . . . . ... 178
The Impact of Sediment Focusing . . . . . . .. 191
Zoning the Cores Palynologically . . . . . . . 201
Assessing the Impact of Soil Nutrient Loss . . . . .. 205
Erosion Rates for the Peten Watersheds ....... . . .. 210

SUMMARY ............................. . .224

BIBLIOGRAPHY ......... . . . . . . . .237.

BIOGRAPHICAL SKETCH .... . . . . . . .. ...... . 249

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the.
Requirements for the Degree of Doctor of Philosophy

PALEOLIMNOLOGY OF THE MAYA REGION

Mark Brenner

April .1b3

Chairman: Edward S. Deevey, Jr.
Major Department:: Zoology

Archaeologically supported estimates of riparian Maya populations

were combined with paleolimnological information to quantify the impact

of prolonged, prehistoric Maya settlement on the watersheds ano lakes

of the karsted lowlands in Peten, Guatemala. The pollen record

indicates that regional, human-induced deforestation began prior to the

Middle Preclassic period (1000 B.C.). Forest recovery commenced with

depopulation at the end of the Postclassic period (1600 A.D.).

Vegetation removal caused profound changes in sediment and

nutrient loading of the Peten lakes, ano the impact was sustained for

nearly 3 millennia. Pre-Maya organic sedimentation was replaced by

rapidly accumulating inorganic deposits, but was restored some 400

years ago when forest regrowth began. Maya exploitation in the

catchments accelerated the rate of total phosphorus delivery to the

lakes. In the absence of soil-anchoring vegetation, phosphorus reached

the lake shores as colluvium, i.e. as redepositea soil. As was shown

in 1979 at Lakes Yaxha and sacnab, phosphorus loading was

Maya-density-dependent (0.5 kg P*capita-l1yr-1). Settlement

vii

and core chemistry data from 3 other basins are consistent with the

quantitative conceptual model based on this constant.

Computed microfossil accumulation rates, though confounded by

diagenesis, indicate that productivity in the Peten lakes was not

enhanced by the anthropogenic phosphorus. Severe siltation may have

inhibited lacustrine production; moreover, most of the haya-perion

phosphorus load was probably delivered to the lakes in biologically

unavailable form.

Shallow-water and deep-water cores from 2 lakes demonstrate the

differential distribution ("focusing") of sediment and imply that

single cores in conical basins are inadequate to describe accurately

the accumulation of chemical and fossil constituents.

Soil and sediment chemistry data (46 soil profiles, 40.5 m of

analyzed Holocene lake sediments cored from 4 basins) indicate that

Peten soils lost perhaps a third of their phosphorus stock as

Maya-generated colluvium. Agricultural yields may therefore have.

declined due to soil nutrient depletion. Concomitant lacustrine

siltation could have reduced the availability of aquatic protein.

Together, nutrient sequestering and siltation may have tunctioneu as a

servomechanism, restricting haya population growth and contributing to

the 9th century Classic population collapse.

INTRODUCTION

Prehistoric Maya and Contemporary Human Popqlation Densities of
Peten, Guatemala

Guatemala's northern Peten region encompasses 35,854 km2 and

constitutes a large portion of the core area in which southern lowland

Maya civilization arose and developed (Fig. 1). Tropical, lowland, cry

forest (Holdridge 1947) covers most of the Peten (Lundell 1937), as

well as portions of Belize, Nexico's Yucatan peninsula, Tabasco, ano

Chiapas. It was in this environmental context that lowland Maya

culture originated and persisted for several millennia before

collapsing mysteriously in the 9th Century A.D. (Culbert 1973).

Perhaps as a testimony to the inhospitable nature ot the tropical

forest, the Peten remains largely depopulated today. Cowgill and

Hutchinson (1963a) reported a Peten population of 20,362 in 1957,

although very recent development has raised the figure nearly 10-fold

(Castellanos 1980). The sparse nature of contemporary human settlement

is placed in perspective by comparison with assessments of Late Classic

(550-850 A.D.) Maya population levels. Delimiting Tikal's areal extent

to just under 163 km2, Haviland's (1969, 1972) Late Classic estimates

for this site alone fall between 40,000 and 49,000. Thompson (196b)

indicates that proposed figures for pre-Columbian population levels are

highly variable. They range from a low of just over a million for the

Fig. 1. Regional map. The Maya Lohlanos incluue arLa. ot belize,
Guatemala's Peten region, and portions of Mexico's Yucatan
Peninsula, labasco, and Chiapas.

Campeche

Peten

100 km

16N

Yucatan

entire Maya region to 13,000,000 for the Yucatan peninsula alone.

Admitting that his own whole-area estimate of 3,000,UUU might be a bit

low, Thompson (1966) was prepared to revise his figure upward, but

conceded that the data base was so small as to make all assessments of

Late Classic levels mere "guesstimates." In a more recent study, Adams

et al. (1981) accept the notion that soae 14,000,000 inhabitants

occupied the Maya lowlands by A.D. 800.

Accurate appraisals of modern Peten population levels are also

difficult to obtain, partly due to the seasonal immigration into the

area associated with industries such as the harvest of "chicle"

(Simmons et al. 1959, Cowgill and Hutchinson 1963a). From the turn of

the century until the late 1960's, "chicle" was Peten's most important

commodity, and the flourishing business attracted large numbers of men

from Mexico, highland Guatemala, and other Central American countries

(Schwartz 1974). The last two decades have witnessed a profound change

in the region, as new settlers, pressured by land shortages in the

highlands to the south, have streamed into Peten in response to the

promise of abundant agricultural terrain. Formed in 1956, the

governmental agency, Empresa Nacional de Fomento y Desarrollo Economico

del Peten (FYDEP) began a land distribution program in the rlio-1960's,

finally granting titles to the assigned parcels in 1974. Under the

auspices of FYDEP, road building and maintenance were priority

projects, and by 1970 the overland route from Flores to Guatemala City

was completed, making the Peten interior more accessible than ever

before. In addition, communication systems were established, air

service was improved, and as a consequence of the subsequent human

influx, the region is beginning to lose its distinctive cultural

identity.

The attraction of available lana has converted highlanders into

"Peteneros" at an alarming rate. Schwartz (1977) reported an increase

from 25,910 Peten inhabitants in 1964 to about 100,000 only 13 years

later. Castellanos (1980) indicates that the rapid immigration that

continue through the late 1970's left the Peten with soiie 20b0,(i

people by the close of the decade. While government efforts to

relocate land-hungry highlanders appear laudable, the program is not

without its drawbacks. Native "Peteneros," while enjoying the new

prosperity, feel their patrimonies are threatened (Schwartz 1977) ana

concern exists thatthere will be insufficient employment opportunities

to accommodate the newcomers.

Another issue that must be addressed is how the immigrants are

faring with subsistence farming practices in the Peten lowlanas. They

are unaccustomed to the high temperatures ot the region and unfamiliar

with the soils, vegetation, and shifting agricultural techniques that

characterize the area. Whether the recent settlers will be able to

cope in the new environment remains to be seen, but the Peten will

continue to undergo drastic social and environmental change in the

future as a result of increasing population density.

While the modern "experiment" in the use of the tropical, lowlana

forest is in its incipient stage, and it would clearly be premature to

pass judgment on its success now, it can be said that the prolonged

persistence of prehistoric Maya culture in the same context constituted

a success that is without parallel. Bronson (1978) has likened the

Maya.to three Asian civilizations that developed in apparently similar

settings, but this hardly detracts from the achievements of the haya,

and the latter are, with the possible exception of the Olmec, the only

New World high civilization to arise and flourish within a tropical,

lowland forest ecosystem.

Maya Cultural Development in the Tropical Forest Ecosystem

While some anthropologists are perplexed by the apparent

incompatibility of high civilization ano the tropical forest setting,

adherents of "environmental determinism" have gone so far as to deny

that lowlano Maya culture could have developed in situ. Meggers (1954)

argued strenuously that Classic Maya culture (250-850 A.D.), with its

monumental architecture, art, hieroglyphics, concept of zero, corbeed

arches, calendrical system, and stela cult, could not have arisen

autochthonously. Believing that the degree of cultural complexity

achieved is dependent upon the agricultural potential of a region,

Meggers classified the Peten as an area of limited agricultural

potential (Type 2), and thus believed the area was not conducive to the

florescence so evident in the archaeological record. The logical

conclusion, based on the assumptions of this agriculturally

deterministic law, was that Classic Maya culture was imported into the

lowlands fully formed and was destined to decline following its

arrival. This simple theory accounted for both the mysterious origin

and inexplicable downfall of the civilization.

In the years following Meggers' (1954) paper, a number of

refutations were published, perhaps laying to rest the broco claims of

environmental determinists, but keeping alive the controversy

surrounding the enigma of the prehistoric high.civilization in the

tropical forest environment. Coe (1957) took exception to Meggers'

claims for a number of reasons. He contested her view that there is a

lack of Preclassic-classic transition in the lowland record and felt

there was no archaeological basis for the argument that classic culture

was imported from the highlands. Additionally, he pointed out that

several lowland Maya achievements, such.as the corbeled arch, the Long

Count, and the stela cult, seem to have no non-lowland origin. Coe's

disagreement with Meggers extended to the perception of the Classic

Maya collapse. While Meggers referred to a "gradual decline,"

initiate following the arrival ot Classic culture in the lowlanas, Coe

pointed out that the evidence indicated a rather widespread, rapid

termination following some 600 years of incremental growth.

In the year following Coe's reply to Meggers, Altschuler (1958)

presented Formative phase ceramic evidence that supported the claim for

autochthonous origin of lowland Classic Maya civilization. His

conviction was that political problems generate the 9th century

collapse. Specifically, he proposed that the ruling class attempted a

political structuring that was doomed to failure because it lacked the

developed techniques of exploitation.

Ferdon (1959) disagreed with Meggers' assessment of the Peten's

agricultural potential, and using his criteria of temperature, soils,

precipitation, and land form, reclassified the lowlands as a favorable

(Type 3 = improvable) site for agriculture. This analysis aid not

constitute a refutation of deterministic law as applied to the haya by

Meggers. It simply freed the civilization from having to conform to

the expectations dictated by a Type 2 (limited potential) environment.

However, Ferdon did take issue with environmental aeterminism ano

presented data supporting his contention that there is no correlation

between natural agricultural potential ana cultural develop.ernt.

Having disproved the role of determinism in the eventual Classic

collapse, he proposed the notion that grass invasion of cleared plots

and the subsequent inability to plow the tracts contributed to the

downfall of the civilization.

The Origins of Haya Civilization

Recent archaeological excavations in the Orange Walk district of

Belize have established that occupation of that area coinenced as early

as the third millennium B.C. At the site of Cuello, fragments of

partially burned wood associated with ceramic material assigned to the

Early Formative Swasey complex (Hammond et al. 1979) have produced

several radiocarbon dates indicating an age ot between 4000 and 5000

calendar years (Hammond et al. 1976, Hammond et al. 1977, Hammono 1960,

1982). These data confirm the claim that lowlana classic Maya cultural

ontogeny had its roots in the lowlands.

Within the Peten region per se, the earliest ceramic material

discovered to date comes from the sites of Seibal ano Altar de

Sacrificios (Adams and Culbert 1977). Assigned respectively to the

Real and Xe complexes, the material is date to the Middle Preclassic

(1000-250 B.C.). Middle Preclassic settlement is known at Tikal and

has also been documented in the Yaxha-Sacnab watersheds of eastern

Peten (Rice 1976). Following an extensive settlement survey and

test-pitting program during the 1979-80 field season, occupation of

this age is now recorded in four more central PeEen drainages (Rice and

Rice 1980a).

While the numerous discoveries of Preclassic settlement in the

lowlands contradict the argument that Classic culture was brought to

the region intrusively, debate continues as to what factors were the

driving forces contributing to the formation of lowland Classic Maya

civilization (Adams 1977). .Although influences from beyond the

lowlands are considered, several intrinsic processes are invoked to

explain the cultural growth and change that produced the social ranking

and stratification that ultimately characterized the Classic period.

Willey (1977) provides a multi-causal model for Classic Maya

development based on a synthesis of processual interpretations proposed

by others. In formulating his 'overarching model," he incorporates

these models and processes that are broadly assigned to three major

categories: (1) ecology-subsistence-demography, (2) warfare, (3)

trade. Be notes that none of these models advocates a monocausal

explanation for the Classic Maya rise ano cautions that the role of

ideology should not be discounted.

Invoking environmental factors, though not as a determinist,

Sheets (1979, 1981) sees the rise of Classic Maya civilization in the

lowlands as reflecting a natural disaster in the highlands of El

Salvador. Acknowledging that Classic haya culture woulo have developed

in any case, he argues that the eruption of El Salvador's Ilopango

volcano in A.D. 260 might have accelerated the process by forcing

increased political and agricultural organization to cope with the

influx of immigrants who descended to the lowlands when highland

agricultural land was rendered useless by the deposition of a thick ash

blanket. Additionally, he points out that Tikal's Early Classic

(250-550 A.D.) florescence may have been stimulated by the diversion of

trade routes with Mexico that previously had passed along the Pacific

coast. With the coastal road destroyed, Tikal became the major eastern

site on the traoe route that extended into the basin of Mexico.

Maya Agricultural Practices

During the last two decades, Maya archaeology has become

preoccupied with questions about pre-Hispanic agricultural practices.

Until that time, the prevailing dogma was that Maya subsistence was

dependent on maize-based, swidden (slash-and-burn, shifting)

agriculture. How this belief became entrenched in the literature

remains unclear, but it was not necessarily supported by the

archaeological record. Though corn is depicted in Maya art, the

contemporary reliance on the "milpa" and the ubiquity of cornfields in

the Guatemalan lowlands at the time of colonial contact ana toaay must

certainly have contributed to the acceptance of the doctrine. While

this agricultural technique may be appropriate for the sparsely

populated modern Peten, there is doubt as to whether it was a feasible

alternative during Classic haya times. Because slash-and-burn requires

land to be fallowed for long periods, much larger areas than the plots

actually under cultivation are required. In the swicaen cycle, forest

is felled during the dry season (January-May), and prior to the onset

of the rains, the dried vegetation is burned, delivering nutrients to

the soil. Following the burn, seeds are sown in holes bored with a

dibble stick, and growth commences after the first showers of the rainy

season with no additional working of the soil required.

Confronted with differing assessments of Peten's agricultural

potential (Meggers vs. Ferdon), Cowgill (1962) sought to resolve the

disparity and interviewed 40 farmers in the region of Lake Peten-ltza

in an effort to ascertain actual corn production. Summarizing the data

from the interviews, she noted that first year plot yields were

1425 lb*acre-1, second year plots averaged 1010 lb*acte-1, and

the five farmers who tried three years of successive planting averaged

417 lb'acre-1, with the decline likely due to nutrient depletion.

Stable swidden of this nature would demand a four-year fallow following

a single crop and six to eight years' rest on a plot worked two years

in succession. Cowgill concluded that swidden farming could support

about 38-77 people*km-2, and tentatively noted that classic period

population densities may have exceeded the restrictions imposed by

stable swidden strategy.

Recent surveys of ancient Maya settlement in the Peten make it

easier to evaluate the feasibility of slash-ano-burn to have supported

prehistoric populations. Rice (1978) useu demographic data gleaned

from the archaeological record together with agricultural output

information to formulate a model that demonstrates the insufficiency ot

maize-baseo agriculture to have met the subsistence neeus ot

populations in the Yaxha-Sacnab subbasins. At this locality, where

slow, steady exponential population growth occurLre from. the iioCale

Preclassic (1000-250 B.C.) until the Late Classic (550-b5U A.D.) (Rice

1978, Deevey et al. 1975), oepenuence on maize-basea swiocen

agriculture would have resulted in tooo shortages by Lat 'Preclassic

(250 B.C.-250 A.D.) times, necessitating shorter tall(w perious or the

farming of less.preferrea sites solely in order to meet subsistence

needs. The incorporation into the subsistence strategy oi root crops,

as originally hypothesized by Bronson (196b) or breadnut, "ramon"

(Brosimum alicastrum) ((Puleston 976, 1b92), is sr-o1w by tt.e nmoue to

have greatly enhanced the probability of supporting large Liassic

populations.

With the demise ot the myth that .,aya subsistence was tota

reliant on maize-based swidden came suggestions for other food sources

ano tood-prooucing systems. Lange (1971) proposed that narinie

resources might have.been an important constituent in the haya diet.

An inventory of Peten's potential iloral ana taunal traoe ittins

(Voorhies 1962) contains a varied assortment of foou types, but the

list of comestibles is by no means compete, as perishaLie goous,

incapable of being transported long distances, were excluoea from this

tabulation. According to hilkin i19/1), t.aoya tuou procurenmet was

likely dependent on an array of systems that incluaeo gardening,

arboriculture, gathering ana intensive practices such as terracing,

irrigation, and drainage.

There is now abunuant archaeological evidence that the loularno

Maya did employ intensive agricultural techniques. In Campeche,

Mexico, riogeG tielos ale known in the Rio LanoelaLiai LeioC iSlr.eien

and Puleston 1972), and hatheny (1976) reports that extensive measures

for water control were taken at Luzna. In the Rio Bec Legioin ot the

southern Yucatan peninsula, terraces ana raise fielos are reporter by

Turner (1974), ana Turner ana HiBrrisozn (1961) Laive aiscoveiea rai&.s

fields at Pulltrouser Swamp, Belize, that may have been conistructea as

early as Late Preclassic (250 b.C.-450 A.L.) times, thereby in oictlig

the great antiquity of the intensive agricultural systems.

The widespread presence of relict canals assuciateu with intensive

cultivation has been documented using raoar imagery (Aoams 1980, AGams

et al. 1981). The canals, locate near swamps, lakes, pounds, ahn

rivers, were apparently constructed for drainage purposes ano have been

veritieo on the ground in areas at hexico altu Belize. Archaeological

ground truth within the Peten is less secure, but should the grio

patterns from the raoar mapping prove to be drainage canals, it winl

indicate the tremendous scale on which intensive agriculture was

practice in the haya lowlands.

The classic Maya Collapse

The recent revelation that classic haya subsistence was at least

in part dependent on intensive agricultural practices helps explain how

the populations of the densely settled region were supported. Unless

it can be claimed that these subsistence strategies caused

environmental detriment, their discovery contributes little to the

search for a cause of the Classic Maya collapse. What makes the Maya

downfall such a perplexing event in New World Prehistory is that the

disintegration of the framework that characterized the socially

stratified Classic culture was accompanied by an extreme demographic

change involving the relatively rapid depopulation of ceremonial

centers as well as the countryside. With a shift in archaeological

focus, beginning around 1950, from an exclusive preoccupation with

ceremonial architecture to an interest in settlement surveys, the

magnitude of the population decline was more fully appreciated (Willey

1982). Willey (1956) expressed the need for more settlement work, but

the lack of evidence for Postclassic (1000-1600 A.D.) occupation of

housemounds excavated during his Belize valley surveys led him to the

conclusion that commoners disappeared with the demise of elite

society. Stated simply,

If collapse occurred--and, indeed, something did
occur--Maya priest and peasant collapsed and
vanished together. Willey (1956: 781)

Adams (1973) has summarized the "collapse" problem and provides a

concise recapitulation of the hypotheses that have been proposed to

account for the Maya downfall. Adams details the archaeological

evidence for the disappearance of the elite class, and his summary view

of the collapse encompasses not only the cessation of elite activity,

but the attendant population decline, both of which occurred throughout

the southern Maya lowlands within a period of 50-100 years.

Noting that there has been a recent tendency to reject single-

factor reasons for the collapse in favor of multiple-factor models,

Adams nevertheless classifies the explanations into broao categories.

Ecological models invoke various environmental disasters, such as soil

exhaustion, water loss and erosion, and savanna grass competition.

Under this heading, Sabloff (1973) might ado insect infestation and

climatic change. Catastrophic events such as earthquakes (Mackie 1561)

and hurricanes are proposed, but serious earthquakes are not

characteristic of the lowlands, and events stemming from disasters such

as these are hard to document using the archaeological record.

Meggers' (1954) 'environmental determinism" is considered an

evolutionary model, but is now discredited because it is known that

lowland culture arose autochthonously, ana the agricultural potential

of the region has been reevaluated. A demographic model for the

population decline was set forth by Cowgill ana hutchinson (1963a), but

receives little attention today. Studying the Indian populations

around Lake Peten-Itza, they discovered that by the fifteenth year of

life, the sex ratio in the population was 1.8cT:1.02. Ihis they

attributed to the poor care given female children between one and four

years of age, by which time the skewed ratio is established. Neglect

and consequent high mortality of female children (1-5 years) is

documented for many non-industrial countries (Cowyill and Hutchinson

1963b), but is likely compensated for by slightly higher male death

rates in the ensuing years. It is pointed out that the extreme case

discovered in the Indian populations of the Peten could have disastrous

consequences. An explanation for Classic haya depopulation was sought

using this scenario, with the realization that it would have to be

applied in a long-term situation.

The destruction of the social structure as a consequence ot

internal revolution has been posited as a factor influencing the

collapse, but while it may account for the disintegration of the social

hierarchy, this alone would not necessarily have led to depopulation.

Invasion from outside the area has also been proposed, and in view of

the post-collapse Toltec takeover in the northern Yucatan, such a

scenario is not out of the question. There is some archaeological

basis for the claim that intrusive elements were present at Seibal and

Altar de Sacrificios prior to the decline.

Finally, disease has been repeatedly implicated as a causal factor

in the collapse, though debate surrounds claims about the pre-Columbian

presence in the New World of illnesses like malaria, yellow fever, and

syphilis. Recent paleopathology work with skeletal material from Altar

de Sacrificios has revealed the occurrence of health problems in the

Maya population that occupied the site. Evidence for physical injury

in the bone sample is minimal, but vitamin C deficiency and anemia,

either diet-related or parasite-inouced, are amply documented (Saul

1973). Additionally, the presence of bone lesions indicative of

syphilis or yaws is noted.

It is clear that the various explanations for the collapse are not

mutually exclusive, and several of the proposed causal factors may have

worked in concert to produce the resultant downfall. Unfortunately,

some of the proposed hypotheses are difficult to reject using the

archaeological record alone, but a systematic program ot testing the

different single-factor theories may one day lead to a synthetic mocel

that reasonably explains the classic Maya disappearance. Such an

approach will doubtless be dependent on evidence provided by

ecologists, ethno-historians, soil scientists, and others outbioe the

realm of archaeology per se.

The Historical Ecology of the Maya

Measuring Maya Environmental Impact Paleolimnologically

This.study does not directly address the question of the

mysterious Maya collapse, though the data collected do in fact suggest

that ecological factors played a role in the event. Instead, the

design of the experiment was chosen with the objective of shedding

light on the impact that long-term Maya agro-engineering practices haa

on the watersheds and lakes of the Peten. In a sense, the question has

been approached conversely in the anthropological literature, as social

scientists have sought to determine the influence the natural

environment had on settlement patterns, foou-producing systems,

socio-political organization, etc. Cognizant of the fact that the

interaction between humans'ana the ecosystem constitutes a feedback

loop, this study takes a decidedly "ecocentric" viewpoint, exploring

the effects of prolonged, dense human occupation on terrestrial ana

aquatic systems of the tropical lowlands. The Peten lake district

(Fig. 2) provides a unique opportunity to examine human-environment

interaction, for, in fact, the "experiment," exploitation of the

lowland tropical forests and lakes, has already been conducted by a

civilization now long.gone. The results of that experiment need.only

be elucidated, and with this in minor, the "Historical Ecology ot the

Maya" project was conceived in 1971. Employing a multidisciplinary

approach, the program involved the use of archaeological and

paleolimnological techniques to examine changes in.the aquatic and

terrestrial systems that resulted from extended human interference.

Lake-Watershed Interactions

Proper study of lacustrine systems in general ana the

paleolimnological record in particular demands a view of lake basins as

integral parts of a larger landscape. Though often considered as.

distinct entities, aquatic and terrestrial ecosystems are .inextricably

linked by meteorologic, geologic, ana biologic processes that transfer

nutrients and energy from one system to the other (Likens and Bormann

1974a). Lakes can be considered "downhill" with respect to their

terrestrial surroundings and the meteorologic, geologic, and biologic

Figure 2. The Peten Lake District.

TIkal

Nokum
x

* C -- --
L. p.e. t ..

FLOtS

/-;i '"

MI- *.-

ha

- *

S0 5 10 20 km

L. Oq uvi

L. P.rdida

1J7<

L. Sacpuy

vectors that join the systems ultimately carry nutrients from upland

sites of accumulation to the waters below. The kincrs of nutrients and

their rates of supply have a profound effect on the lake, exerting a

controlling force on the physical, chemical and biological processes

that occur in the aquatic realm. Mature, intact, terrestrial

ecosystems tend to maintain tight nutrient cycles, with loss to the

lacustrine sector minimized by the presence of the soil-anchoring,

standing vegetation. When the biological component of the land system

is disturbed, through forest clearance, nutrient cycles are disrupted,

accelerating the delivery of'dissolveo and particulate matter to the

lake.

Inadvertent enrichment of lake waters can occur as a result of

forest clearance, but often, lacustrine pollution occurs as if by

design. The "downhill" nature of lakes makes them convenient disposal

systems for unwanted, accumulated wastes like domestic sewage and

industrial by-products. These practices are not without their

consequences and our contemporary cultural eutrophication problems stem

from the casual manner in which many lakes have been used as recipients

of sewage and agricultural run-off rich in plant nutrients (aEmondson

1968, 1970, 1972; Vallentyne 1974). The realization that the source

area for some of these unwanted.nutrient inputs lies some distance from

the water's edge demands that investigators look beyond the lake itself

in the study of lacustrine processes. Furthermore, awareness of acid

rain problems (Likens and Bormann 1974b) demonstrates the neeo for

consideration of the regional airshed in assessing lake dynamics.

Bormann and Likens (1967) suggested that small watersheds make

ideal sites for examining nutrient cycle problems ano proposed that the

entire watershed be considered the basic unit for ecosystem-level lake

study. Modern studies at Hubbard Brook, New Hampshire, have

accumulated biogeochemical output data from undisturbed, natural

ecosystems (drainage basins) (Likens et al. 1977), ano these baseline

values can be compared to nutrient losses from clear-cut forests

(Bormann et al. 1968). Contemporary investigations of this type

measure altered nutrient outputs as they occur. That is, data on the

rate of nutrient transfer between the terrestrial-aquatip interface can

be collected immediately following a "treatment" like deforestation.

In attempting to assess the effect of past events on lakes, it is

necessary to rely on the paleolimnological record.

The Paleolimnological Perspective

Paleolimnological study can document past changes in a lake ana

its drainage, because shifting conditions in the watershed had an

impact on the lake; and a record of the alterations, though perhaps

somewhat distorted, is preserved within the sediments on the lake

bottom. Frey (1974) has said,

The task of the paleolimnologist is to "read" the
history of the lake-watershed-atmosphere systems
from the record "written" in the sediments.
(1974:95)

In dealing with the origin and developmental history of basins,

paleolimnology can address questions about the ontogeny of lakes that

were free from human disturbance. Climatic change can be inferred from

the palynological record and theoretical questions about ecosystem

development can be approached using sedimented microfossil assemblages

(Deevey 1969). With the greater awareness that human activity can

radically alter watershed nutrient cycles, paleolimnological techniques

are being used increasingly as a tool to assess the iagnituoe ot

human-induced changes. Sometimes historic data on demographics and

waste dumping are known for a lake-watershed system, but the

limnological record is restricted to the postdisturbance period. In

Lake Washington, east of Seattle, the sediments yielded information

concerning baseline lacustrine conditions, prior to the eutrophication

that resulted from excessive sewage input to the lake (Eamondson 1974).

The paleolimnological record, in conjunction with early historical

records or archaeological data, has been used to establish the impact

of human activity on a number of basins, sedimentary changes having

been correlated with density of occupation or shifts in lana use

(Cowgill and Hutchinson 1964). Like the vast majority of contemporary

limnological investigations, most paleolimnological projects have been

undertaken in the temperate area lakes of now-industrialized countries

(Mikulski 1976, Penninyton 1978, Vuorinen 1978, Warwick 19bU). lnere

is a paucity of literature concerning tropical paleolimnology,

especially with regard to the impact of human disturbance on tropical

ecosystems. Though regrettable, there are several factors that likely

account for the restricted development of tropical paleoliinology.

First, well-established limnological research centers are generally

confined to temperate regions, often in close proximity to lake

districts. Therefore, mounting drilling campaigns in tropical areas

can be quite costly, necessitating a large initial outlay for travel.

Once in the tropics, one may encounter additional ditticulties. Poor

road conditions or a complete lack thereof can render potential coring

sites inaccessible. Also, such projects often require the permission

of foreign government officials, and even when permits are forthcoming,

the local political or scientific community may lack the infrastructure

to be of assistance. Political instability is a hazard to be

considered and makes many potentially exciting study sites

'off-limits." Finally, little is known of the regional limnology in

most tropical areas, making interpretation of the paleolimnological

record somewhat less secure.

Despite the many drawbacks and logistical difficulties associated

with paleolimnological work in the tropics, there are significant

arguments that convincingly speak to the need for more study in these

regions. As the human populations in tropical countries continue to

increase, deforestation and resource exploitation will accompany the

demographic change. What impact the forest felling and farming

practices will have on freshwater sources is not known. Limnological

monitoring of newly settled drainages must begin, and sediment studies

can be used to gather baseline information from basins with a long

history of occupation. With continued demophoric growth in the

tropics, management schemes for freshwater resources will have to be

instituted and cannot be formulated using the temperate data base.

Despite the differences between temperate and tropical systems,

paleolimnological techniques should be applicable in both settings for

documenting human intrusion in watersheds. Nutrient cycles in

undisturbed, tropical watersheds are very tight and maintained by the

standing vegetation. Any disruption in the drainage basin should have

a noticeable, if not profound, effect on the lake and consequently the

sediments (Oldfield 1977).

Several characteristics of the Peten lakes recommend them as study

sites. First, the basins are closed, and because the lakes lack

outlets, the sediments are the ultimate sink for much of the dissolved

and particulate matter washing into the lake as well as biogenic

material formed autochthonously. Secondly, a long history of haya

settlement in the Peten watersheds should be expected because of the

scarcity of surface water in the lowlands. Initial settlement in the

region might be supposed to have clustered around readily available

sources of water. That access to water was a problem for the haya is

evident in the archaeological record at some sites in the interior.

The long dry season necessitated the construction of reservoirs at

Tikal, and evidently some "chultuns," hollowed-out, underground

caverns, were employed for water storage (Matheny 1982). To the north,

in the drier Yucatan, the situation was even more critical, and it has

been pointed out that the Maya of that region developed a civilization,

in a sense based on groundwater, with population centers located near

water supplies in the form of natural cenotess," caves, ano "aguadas,"

or man-made wells (Back 1981).

One scenario for the initial Maya invasion of the Peten interior

envisions the pioneers entering on the river systems ano later

expanding into the drier regions of the central core area (Puleston ana

Puleston 1971). It has been suggested that the lack ot available water

and necessity to cope with the problem may have been the driving force

that led to substantial social organization, ihe oldest Peten sites of

Seibal and Altar de Sacrificios are situated on rivers, and it is

conceivable that the earliest emigrants from these communities, or

other, as yet undiscovered river villages, traded the benefits of

riverine settlement for the advantages of riparian occupation on the

lake shores. Both localities would have been favorable settlement

areas, providing water as well as sources of aquatic protein.

The Contemporary Peten Environment

Geology

With the exception of the mountainous Lacanoon area in the

northwest and the extension of the Belizean Maya Mountains in the

extreme southeast, the Peten is characterized by low-lying karstea

terrain varying in elevation from about 100 to 300 m above sea level.

As is typical of limestone regions, the countryside is irregularly

pocked with caverns and sinkholes. The haystack hill topography

developed on limestones of Cretaceous and Tertiary age (West 19b4).

The Peten lake district (Fig. 2), with its center at 170N,

8940'W, lies within the Santa Amelia Formation, a deposit of early

Eocene age (Vinson 1962). North of the lake region, the Santa Amelia

is overlain by the slightly younger limestones of the Buena Vista

Formation, the basal portion of which contains a 200-m thick zone of

gypsum. Both formations are locally interbedaed with dolomite and

gypsum. During middle Tertiary times, compressional folding and

concomitant emergence resulted in a mid-Eocene to Oligocene

depositional hiatus, though locally there are deposits in the Peten of

Oligocene to Pliocene age. By late Pliocene, uplitt, folding, ano

faulting put an end to Tertiary sedimentation in the region.

The lake chain at 17N (Fig. 2) is aligned along a series oi

east-west trending en echelon faults, the basins occupying depressions

below steep north shore scarps (Tamayo and West 1964). The principal

lakes in the fault zone chain extend some 80 km from westernmost Sacpuy

eastward to the twin basins of Yaxha and sacnab, only 30 km tron the

Belize border. Farther to the west, but outside the main graben, lies

the relatively large and limnologically unexplored Lake Perdica. In

addition to the localized standing bodies of water are seasonally

inundated depressions interspersed between the limestone hills,

features that are not uncommon over a large portion of the Peten

landscape. These "bajos" or "akalches" are characterize by thick clay

soils that give rise to swamp-thicket vegetation. It has been

suggested that the "bajos" were once shallow lakes, providing water,

lacustrine resources and a mode of transportation for the Maya who

inhabited the shores. The silting-in of these shallow basins has been

invoked as a contributing factor for the Classic collapse (Cooke 1931,

Harrison 1977). A 5-m pit dug in the Bajo ae Santa Fe, near Tikal,

revealed that indeed the clays that lined the floor to the.aepression

resulted from the solution of:upland limestone, but there was no

evidence for lacustrine deposition having occurred during Holocene

times (Cowgill and Hutchinson 1973).

The calcareous bedrock of the Peten provided the resource base for

Maya architectural endeavors, as building stone was easily quarried.

In addition, limestone was burned and mixed with calcareous sano

("sascab") to make construction mortar or plaster. The Maya also

exploited the localized flint and chert beas, using the siliceous rock

for making points and cutting tools (hest 1964).

Climate and Rainfall

Lying at low altitude, within the tropic, the Peten is

characterized by year-round high temperatures, the mean annual value in

excess of 25C (Vivo Escoto 1964). Mean monthly temperatures for the

region range between 22*C and 26C, but as expected in tropical areas,

daily fluctuations in temperature often exceed the limits of the

monthly extremes.

Within the Peten, precipitation is highly variable from station to

station and varies on an annual basis at any given site. Rainfall

records indicate annual precipitation values ranging from ca 90b to ca

2500 mm. A regional, yearly mean of 1601 mm is reported based on 54

station-years of data collected at 10 sites (Deevey 197b). Within the

tropics, as a rule, the distribution of rainfall throughout the year is

highly seasonal (Richards 1979), and Peten is no exception. There is a

long dry season from January to May with a secondary period of reduced

rainfall, "canicula," interrupting the wet season in July or August.

The most pronounced aridity occurs from January to March during which

time the rainfall amounts to less than 10% of the total annual income

(Deevey 1978).

Soils

The soils of the Peten catchments represent a large potential

source for lacustrine nutrients. Under conditions of deforestation,

enhanced delivery of dissolved and particulate matter to the lakes is

expected, and the tremendous erosive potential of intense tropical

rains is a major contributing factor in the transfer of nutrients from.

the land to water.. The extreme seasonality and heavy downpours

characteristic of the tropics make rainfall at those latitudes more

erosive than equivalent annual precipitation in temperate areas where

the rains are distributed more evenly throughout the year (Stevens

1964).

Roughly 0.4% of the Peten landscape is covered by the major lakes,

and the balance of the region is overlain by soils assigned to 26

series by Simmons et al. (1959). They relegated the department soils

to two major groups: savanna soils that cover some 9.8% of Peten ano

forest soils that blanket 89.8% of the.region. These major categories

were further divided, savanna soils characterized as deep well-drained,

deep poorly or deficiently-drained, and shallow deficiently-drained.

Within these subdivisions, the soils were assigned to a particular

series based on a number of characteristics, including parent material,

relief, color, texture and consistency, ano profile thickness. With

the exception of some localized soils that overlie clay-rich schists

and some alluvial deposits, Peten soils are derived from the underlying

limestones.

Zonal soils develop under the primary influences of regional

climate and vegetation, their distribution being highly correlated with

patterns reflected in the climatic regime and plant associations.

Within the central portion of the Peten, soil genesis is influenced

tremendously by the calcareous bedrock as well as drainage factors, to

the extent that zonal soil development is precluded. In the lake

district, the local geology ana hydrology have generated primarily

intrazonal soils assigned to the Renazina and Bydromorphic great soil

groups (Stevens 1964). Azonal Lithosols, black calcareous soils

resembling Rendzina are abundant, and Stevens (1964) speculates that

these youthful soils are now regenerating following a long period of

erosion and depletion induced by Maya farming practices.

Though 26 soil series are described for the Peten, only three

surround the lakes examined in this study. The Yaxa series covers some

15.57% of the Peten and consists of shallow well-drained forest soils

that often cap flat expanses as well as hilly slopes. These black

calcareous Lithosols are highly fertile, and cultivation of these soils

is only restricted by their high erosivity and presence on steep

slopes. Yaxa soils surround Lake Macanche and nearly encompass Lake

Salpeten, the southwest shore of which is contacted by soils or the

Macanche series. Macanche series soils blanket 5.11% of the department

landscape and are shallow soils with deficient drainage. Confined to

primarily level topography, these Rendzinas are highly fertile and not

particularly subject to erosion. Poor drainage and the adhesive nature

of the soils are the only drawbacks for agriculture. Indeed, the black

calcareous Lithosols and Rendzinas were certainly exploited by the haya

and.numerous Classic Maya ceremonial centers are found in association

with soil series of these groups (Stevens 1964).

According to Simmons et al. (1959), the western edge of Lake

Quexil is contacted by Yaxa soils, but the balance of the drainage is

occupied by soils of the Exkixil series. A deep poorly drained savanna

soil, the Exkixil series is restricted to only 0.23% of the Peten and

is found in flat areas. These soils can be assigned to the

Hydromorphic great soils group ana are typified by high clay and silt

content and poor fertility. They are not easily eroded and tooay

support grasses and open oak woodland.

It is noteworthy that the soils map developed by Simmons et al.

(1959) is rather crude with respect to accurately delimiting the areas

covered by the various soil types. Within the Peten, great variation

in topography and perhaps bedrock geology can be encountered over short

distances. This in turn leads to great heterogeneity and patchiness of

soil types within limited areas. ihe incongruity of mappeo soil zones

(Simmons et al. 1959, and personal observations) is most clearly

demonstrated at Lake Quexil. While the map shows the Quexil watershed

to be dominated by Exkixil series soils, the true drainage is primarily

covered by forest soils, probably of the Yaxa series.

Vegetation

Though commonly referred to as "tropical rain toiest," the

vegetation of Peten grows in a region that is too dry for the

development of true rain forest. Employing the climatic data criteria

established by Holdridge (1947), the Peten falls within the tropical,

lowland, dry forest life zone. Lundell (1937) applied the tenm "quasi-

rain-forest" and though the vegetation is principally evergreen, some

species lose their leaves periodically, the degree ot deciduousness

dependent on the annual distribution and amount of rainfall.

Simple description of the Peten vegetation is impossible due to

the variability of vegetation types that reflect topographic and

edaphic differences. Wagner (1964) reports that some 75% of the uplana

forest is covered by the "zapotal" association, named for the

prevalence of "chico zapote" (Manilkara) in the miaole tier of the

forest. Characterizing this dominant association, he notes that the

major floristic components in the top story are Calophyllum, Swietenia,

Rheedia, Lucuma, Sideroxylon, as well as several species of Ficus.

Below the uppermost tier lies a middle story of Manilkara, Vitex,

Ficus, Cecropia, Bursera, Spondias, Aspidosperma, Brosimum,

Pseudolmedia, and members of the Leguminosae ano Lauraceae. Averaging

10 m, the lower story is typified by Trichilia, Siderokylon, Sapium,

Sebastiania, Misanteca, Parmentiera, Myriocarpa, Lucuma, Louteridium,

Laetia, Deherainia, Annona, Sabal, Pimenta, Protium, Ocotea,

Zanthoxylon, ana species of Pithecolobium, aalisia, Cordia, and

Croton. The underwood plants are piper, Psychotria, Ruellia, Justicia,

and various palms. Lianas are common, as are orchids, bromeliads, ano

ferns.

Wagner's (1964) enumeration of the genera that typify tie.Peten

forest provides an impression of the floristic composition of at least

one major association. The forest can also be described base on its

physiognomy.. During a 1974, week-long reconnaissance and vegetation

sampling trip near Lake Yaxha, Ewel ana Myers (1974) identities four

vegetation types, the physiognomy of which reflected the underlying

topography. Three of the four distinct vegetation classes were

sampled, including (1) the forests of steep slopes and ridges, located

on well-orained soils and possessing an irregular canopy, the tourlaea

crowns of the tallest individuals often separated by large openings;

(2) Gentle slope forests occupying deeper soils, with standing water in

localized depressions, and typified by a smooth canopy with only

occasional emergents; and (3) Seasonally dry "bajo" vegetation of

short stature (< 20 m), with a smooth canopy and a high incidence of.

windfall. The unexamined wet "ba]o" vegetation appears on aerial

photos and evidently consists of stunted vegetation with closely packed

trees. As part of their sampling procedure, Ewel and hyers established

six 0.1 ha plots, two in each of the three investigated vegetation

types. Identifying all trees with a stem diameter of more than 10 cm

at breast height, the investigators tallied 57 taxa. Strangely, many

of the upland site species were shared by the "bajo" localities, though

trees .on the latter sites were much smaller.

In a similar study, I established five 10 m x 100 m plots in three

vegetation types in.the central Peten. One plot was placed in high

forest near Lake Macanche. The remaining 0.1 ha units were located in

forested areas of the primarily savanna region lying south and

southwest of .Lake Peten-Itza and in the area close to the

archaeological sites of Chakantun and Fango (Rice and Rice 1979). At

each of these sites, a sampling transect was designated in a

substantial, forested area bordered by savanna and a second plot was

established in a "sukche," an island of forest surrounded by savanna.

All trees with a diameter ot more than 5 cm at breast height were

recorded, revealing a diverse flora of 77 taxa on the five transects.

While high forest of great complexity is a most striking feature

of the Peten landscape, nearly 10% of the region is covered by

savanna. ihe most expansive grasslands lie to the south of Lake

Peten-Itza and in many areas interdigitate with stands of forest.

While the savanna is relatively devoid of woody growth in some places,

other localities like the area south of Lake Quexil support numerous

oaks (Quercus). At other sites, "nanze," Byrsonima is the.predominant

tree and is distributed rather evenly over the grassland.

It has been suggested by Lundell (1937) that the savannas are in

fact vegetational artifacts of human disturbance, created by Maya land

clearance and repeated burning. Reinvasion of the deforested areas by

trees may be prevented by human-induced edaphic changes, but tire

frequency is certainly a factor maintaining the grassland (Vaughan

1979). Palynological investigations of the sediments from savanna

Lakes Oquevix and Ija as well as studies of grassland soils will be

necessary to resolve questions about the genesis ana maintenance of

this unique vegetation.

Lundell (1937) believed that the high forest of mooern Peten was

absent during the Maya florescence, the land having been cleared for

agricultural purposes. This contention is now amply supported by

palynological evidence from a number of Peten lake sediment cores

(Tsukada 1966, Vaughan 1979, Deevey et al. 198bc). Lundell (1937) also

felt that the modern standing vegetation represents a climatic climax

forest, there having been sufficient time for its development following

the Maya decline. This view is somewhat contradicted by his belief

that the prevalence of many useful tree species on Maya ruins

constitutes evidence for the claim that the ancient Maya practiced

arboriculture. Fruit-bearing trees, such as Brosimum, Talisia, and

Manilkara, are common on sites as are other economically useful

species, such as incense producers like Protium. While it has been

suggested that these trees were selectively spared during Maya forest

clearance or cultivated to some extent, Puleston (197b, 1982) argues

strenuously that at least one species, "ramon" (Brosimum alicastrum)

was actively planted and that its starch-rich seeds comprise a major

portion of the Maya diet. Another possibility that may account for the

presence of economically useful trees on previously occupied sites is

that the topographic and agronomic factors that may have been

attractive to Maya settlers, like upland, level, well-drained fertile

areas (Rice and Rice 1960b), may simply coincide with the ecological

requirements of the tree species. Additionally, it is possible that

the Maya modified the edaphic conditions, inadvertently creating

optimal chemical or drainage microenvironments for the growth of the

trees, thereby permitting them to flourish after the Classic collapse

(Lambert ana Arnason 1982).

Regional Limnology

The principal basins of the Peten lake district were former when

water filled the troughs of the east-west graben that lies at 17*N.

Most of the lakes are small (< 5 km2) with the exception of the two

largest, Peten-Itza (99.6 km2) and Yaxha (7.4 km2). today Lake

Peten-Itza supports a substantial human population on its shores, the

major riparian settlement occupying areas in contact with the southern

arm of the lake. The bulk of the lakeside residents inhabit three

towns, including the mainland "pueblos," San Benito, ano Santa Elena,

as well as Peten's political hub, Flores, formerly an island community

but now connected to the southern shore by a causeway. San Andres and

San Jose are the principal towns on the north shore of the lake and

overlook the deep, main basin. Lakes Macanche ana Sacpuy also support

small, but rapidly growing "aldeas," while the remainder of the lakes

are largely undisturbed, though isolated houses ana farn.ing activity in

the watersheds have been noted.

While relatively small in surface area, the lakes are rather deep,

many in excess of 30 m. The maximum depth is often associated with

sinks or trenches that are responsible for the conical morphometry of

some basins (Deevey et al. 1980a). Typically, the basins possess deep

trenches in proximity to the north shore, where steep scarps descend to

the lake edge. Southern shores are generally flatter, ana water depth

increases gradually with distance from land.

Despite the fact that exchange with groundwater in the lakes was

shown to be one-way downward and slow, based on studies at Yaxha ano

Sacnab (Deevey et al. 1980a), the lakes are prone to rapid fluctuations

in level that are probably not solely dependent on direct precipitation

and drainage income. A rise of about 3 m was detected in all the lakes

between 1979 and 1980 and was likely associated with groundwater

intrusion. The notion that changes in the regional water table did

occur is supported by the formation of a new lake in the savannas, a

basin that local residents claim filled when groundwater broke through

the country rock. The lakes continued to rise above the 1980 level (P.

Rice, pers. comm.), but this is not the first time they have aovancea.

Older residents of the Peten report that Lake Peten-Itza rises every

40 years, maintaining a high stand for five years before retreating

S(Anonymous 1980a). The last high water was recorded in 1938, when the

level was several meters above the 1980 mark, and much of Flores was

inundated. Other lakes in the district have also experienced high

stands in the past. Supporting evidence comes from the presence of

aquatic snail shells in soils (old lake muds?) well above modern lake

surfaces. H.K. Brooks (pers. comm.) reports snail shells some 13 m

above the 1973 Yaxha surface. Fred Thompson (pers. conu.) ioentifieo

three species of aquatic gastropods, Pyrgophorus exiguus, cochliopina

infundibulum, and Aroapyrgus cf. petenensis, founo in Salpeten south

shore soil samples collected from pits dug several meters higher than

the 1980 peak water mark.

Limnological reconnaissance of the Peten lakes was first

undertaken by Brezonik and Fox in the summer of 1969. They reported

the clinograde oxygen profiles and thermal stratification that

characterize the basins (Brezonik and Fox 1974). Surface waters in the

Peten often exceed 30C, and while thermal stratification aoes seem to

be persistent, hypolimnetic water temperatures sometimes differ from

epilimnetic values by only 3-4C. Encountering anoxic hypolinnia in

most Peten lakes as well as benthic fauna indicative of oxygen stress,

Brezonik and Fox (1974) concluded that thermal stratification of the

lakes was stable. Additionally, they found evidence of meromixis in

Lake Quexil and two small, but deep sinkhole basins, Paxcaman ano

Juleque. The maintenance of the thermocline was attributed to three

factors: the lakes are well protected from strong winds by the

forested limestone bluffs that surround them; the basins are typically

small, but very deep, with contours that inhibit mixing; ana finally,

they note that the density difference per degree change in temperature

is much greater in warm waters than in cold waters.

Over the past decade, the Peten lakes have been sampled

extensively and intensively, supplementing and contradicting some oi

the original findings. Lake Quexil has been studied during several

field seasons, and neither chemical data nor conductivity readings

point to modern meromixis, though an episode of early Holocene

meromixis was detected in the sediment record (Deevey et al., in

press). If contemporary meromixis is to be documented in Peten waters,

it will probably be encountered in the small deep sinks like Juleque

and Paxcaman or in sulfate-rich Monifata I (Leevey-et al. 1980a).

Numerous oxygen profiles from several of the basins show complete

oxygen depletion in hypolimnetic waters to be rare and indicate that

Peten lakes are perhaps polymictic, their frequent circulation

associated with nocturnal cooling. However, predawn data from Lakes

Quexil and Macanche taken in May 1977 demonstrate the persistence ot

the thermal gradient through the night, surface temperatures exceeding

bottom values in the two basins by 5.9C ano 4.00C respectively.

Repeated night sampling, between May and July 1980 of a deep sink

(43 m) located near the northern shore and within Peten-Itza's southern

basin, failed to reveal a breakdown of the thermal gradient.

Oligomixis, or irregular ano uncommon turnover, may better define the

mixing regime of the Peten.lakes, the frequency nevertheless sufficient

to prevent prolonged hypolimnetic anoxia. If episodes of mixis are to

be detected, the most probable period for their occurrence is between

August and December when cooler air temperatures, strong winos, ana

heavy rains might induce turnover. Unfortunately, these are the months

for which limnological data are scanty.

Calcium and bicarbonate are the major cation and anion of Peten

waters, but exceptions are tound in magnesium and sulfate-rich lakes

that may overlie dolomite or gypsum beds. That geological variation

occurs over short distances is quite evident when nearby lake waters

are compared chemically. Lakes Macanche and Salpeten are only

2 km apart but differ appreciably. hacanche is a magnesium sulfate

lake and contains about twice as many magnesium as calcium ions, while

the sulfate:bicarbonate ionic ratio is 1.3:1. Salpeten is a calcium

sulfate lake in which magnesium ions are half as prevalent as calcium

ions, and there are nearly 60 sulfate ions for each bicarbonate ion.

The aptly named Salpeten, also seen in the literature as Peten-suc and

Sucpeten, is highly saline, and the water contains 4.76 g1l-1 total

dissolved solids. Peten waters are alkaline, their ph ranging from

neutrality to 6.6, with surface waters generally producing daytime

readings around 8.0.

Nutrient concentrations of Peten lakes are surprisingly low for

basins in limestone terrain. In a survey of eight lakes, Brezonik ana

Fox (1974) reported total phosphorus concentrations from < 10-33 ug

Pl1-1, relegating these water bodies to the oligo-mesotrophic

(5-10 ug P*1-1) or meso-eutrophic (10-30 ug P-1-1) categories,

as defined by Wetzel's (1975) modification of Vollenweider's (1968)

phosphorus-dependent trophic classification scheme. In another series

of total phosphorus determinations, nine examined lakes displayed a

range from Ib to 54 ug P'1-1, the low value being from savanna Lake

Oquevix (Deevey et al. 1980a). The generally higher phosphorus levels

discovered in the latter study are perhaps attributable to more

complete digestions or.higher sestonic content in the samples, but

demand placement of the lakes in the meso-eutrophic (10-30 ug

P'1-1) or eutrophic (30-100 ug P-1-i) categories. Assignment

to the trophic classes is tentative as a large fraction (40-80%) of the

seston, measured in fresh sediment from traps or cores, is inorganic

(Deevey et al. 1977, Deevey et al. 1980b). Much of this resuspended

silt may contain mineralogic phosphorus that is biologically

unavailable. In any case, high N:P ratios (>25) in Peten waters and

surficial sediments suggest that productivity is phosphorus limited.

Secchi disc transparency, an indirect measure of productivity,

measures turbidity in Peten lakes, as organic color is low (brezonik

and Fox 1974). However, measured transparencies may not be a very good

indication of algal standing crops if much suspended material is

inorganic seston. Secchi disc readings range from 1 to 5 m, the

clearest water generally found in the sulfate-rich lakes. Lake Quexil,

dominated by calcium and bicarbonate ions, displays low transparency,

less than 2 m at times, ano because seston of the lake shows relatively

high loss on ignition (50-60%), the rapid disappearance of the disc

with depth may indicate a.substantial algal biomass. It is noteworthy

that Lake Quexil (also seen in the literature and on maps as Lckixil,

Ekichil, or Exiquil) derives its name from the Maya word Ekexiil

(another variant) that means "agua oscura con hierbas," or water

darkened by weeds (Anonymous 1980b).

Direct measurement of productivity was accomplished by light-aark

bottle experiments, conducted by H.H. vaughan at twin Lakes Yaxha ana

Sacnab in 1973 and 1974. Five sets of experiments at Yaxha and three

at Sacnab yielded nearly identical mean values, 251.6 + 121.6 mg

C-m-2*day-I (Yaxha) and 251.7 + 11U0. mg Ccm-2*day-1

(Sacnab). A single experiment was undertaken at Lake Quexil in

mid-March 1978 and gave a gross production figure of 196 mg

C.m-2.day-1 (Deevey et al. 1960a). The values obtained suggest

that the lakes are at the low end of the mesotrophic (250-1000 mg

C.m-2.day-1) or upper end of the oligotrophic (50-300 mg

C*m-2-day-1) scale (Wetzel 1975), but this assignment is

tentative as production undoubtedly fluctuates throughout the year, and

the data set is small. Nevertheless, the range of production values

encountered in the twin lakes, 157.6-449.3 mg C'm-2.day-1

(Yaxha) and 135.3-354.5 mg Cmm-2-day-1 (Sacnab) is indicative

of oligotrophy or mesotrophy using temperate standards.

The Peten lakes that are free from human impact generally support

a small algal biomass, at times apparently dominated by the cyanophyte,

Lyngbya (Brezonik and Fox 1974). Peten waters are noticeably deficient

in diatoms, though Melosira ambigua and M. granulata are common in

Lakes Yaxha and Sacnab. Members of the Bacillariophyceae are rare in

the lake sediments, probably due to dissolution at high pH. While

Yezdani's checklist of Peten phytoplankton (Leevey et al. 198Ua) is

long, the implied diversity is misleading, as several dominants

probably constitute a large fraction of the standing crop year-rouna.

The chlorophyte Botryococcus braunii and the blue-green Microcystis

aeruginosa are nearly ubiquitous ano are encountered in large numbers.

B. braunii constantly oominates net tows anu its conspicuous absence

from centrifuged water samples collected by Brezonik ano Fox probably

resulted from resuspension following centritugation. As a Qonmlhiat, b.

braunii is indicative of oligotrophic conditions in temperate regions,

but its value as an indicator species in the Ieten likec is

questionable, as it is founa unoer a variety of conditions in tropical

ana subtropical basins (hutchlnson 19Sb7.

The open water zooplankton is numerically aominateo by copepoas,

the four most abunoant species being Liaptomu acr sails, hesocyciop

inversus, M. eaax, and Iropocyclops prasinus mexicanus. Another

important member of the zooplankton is the enuemic, peiagic ostracoo,

Cypria petenensis that distinguishes itself by being rare or absent in

sulfate waters. Juveniles are present in the water column at all

times, while adults are migratory, leaving the benthic habitat anc

entering the pelagic environment only at night, a strategy that Likely

helps them avoid predation by fish. Eubosmina tubicen (Leevey ana

Leevey 1971) is the only common pianktornic cl.oocerara, with

representatives of the Sididae, Daphniidae, hacrothricidae, ano

Chyaoriaae notably rarer in open water net tows.

Thirty-two species of entomostracans are reported from plankton

tows in 10 Peten lakes, the most diverse tauna (23 spp.) truno in the

largest lake, Peten-Itza (Deevey et al. 1980b). The lakes average 11.4

species each, the least diverse tauna encountered in sultate-rich Lakes

Salpeten (4 spp.) and Monifata I (5 spp.). hhile the planktonic

component of the entomostracan fauna is notably universe, systematic

sampling of a littoral zone study site in Lake Peten-Itza has expar.aea

the list of entomostracan taxa, with 31 species of chydorids (29

identified, 2 unknown) alone, having been enumerated (M.h. Bintord,

pers. comm.). Only six chydorid species were taken in the Lake

Petenxil plankton (Deevey et al. 1980b), but some four tires as many

species were identified from remains in the Petenxil sediments (Goulden

1966). Chydorid remains are integrated into the sedimentaLy matrix

over space and time so that more taxa are frequently encountered in mud

samples than can be collected in a systematic lake sampling program

(Frey 1960).

Twenty-two species of fish have been identified fror. collections

made in five lakes (R.M. Bailey, pers. comm.), and three of these, the

clupeid Dorosoma petenense, the characin Astyanax fasciatus, and an

atherinid Melaniris sp., may be responsible for the low diversity and

small body size that characterize the Peten zooplankton. Other

potential consumers are the second characin Hyphessobrycon compressus,

as well as several poeciliids and juvenile cichlids. Some members of

the latter family attain significant size (1-2 kg), and some Cichlasoma

species and the "blanco," Petenia splendid, are exploited by modern

"Peteneros," as they undoubtedly were during Maya times.

Maya Settlement in the Yaxha and Sacnab catchments

Maya Demography

Twin Lakes Yaxha and Sacnab were originally selected as stuoy

sites because early archaeological reports indicated differential

degrees of Maya settlement in the two watersheds (Bullaro 1960). It

was surmised that the dissimilar levels of environmental impact woulo

be reflected in the paleolimnological record. laxha supported an urban

center on its north shore, and the island site of Topoxte, lying within

Lake Yaxha, was densely inhabited during Postclassic times (1000-1600

A.D.). Urbanization never developed in the Sacnab watershed, and the

drainage remained devoid of habitation following the classic collapse.

In order to assess changing ancient Maya riparian population

densities, Don S. and Prudence M. Rice established transects radiating

from the lake shores that were searched for housemounds. The

transects, 2 km long and 0.5 km wide, were located on the lake edges at

randomly selected points and ran in a north-south direction,

perpendicular to the fault line, thereby permitting the sampling of all

microtopographic zones. Six sampling units were staked out at Yaxha,

and four were established at Sacnab (Fig. 3), the total area of the

transects in each case equivalent to about 25% of the subbasin

drainages. At Yaxha, the island of lopoxte was also designated as a

sampling area. The plots were systematically searched for mounds, ano

a randomly selected 25% of the mounds were test-pitted so that periods

of occupation could be established using the ceramic sherds extracted.

Figure 3. Twin Lakes Yaxha and Sacnab, showing the position of the archaeological sampling transects
and soil pits. Op = operation.

SCALE

2Cnnou,. in n.)

These data were then employed to calculate population densities for the

ceramically defined periods using the assumptions previously applied at

Tikal,.that 84% of the mounds discovered were residences and that the

average household consisted of 5.6 persons (Havilano 1970). by

convention, the calculated densities represented levels of occupation

at the end of the ceramically defined periods.

Middle Preclassic (1000-250 B.C.) population densities were

slightly higher at Sacnab (34 persons*km-t) than at Yaxha (22

persons*km-2), but by the close of the Late Preclassic (25U

B.C.-250 A.D.), Yaxha, having grown more quickly, had 70

persons*km-2 to Sacnab's 51 persons*km-2. A lag in growth at

Yaxha permitted Sacnab to catch up by the end of the Early Classic

(250-550 A.D.), the two subbasins hosting 101 persons*km-2 ano 102

persons'km-2 respectively. The centripetal draw of the Yaxha urLan

center ultimately left that subbasin with a dense (256

persons*km-2) human population by the ena of the Late Classic

(550-850 A.D.), at which time Sacnab supported 168 persons-km-.'

Considered as a single population, the prehistoric Maya of the combined

twin basins increased their numbers slowly, doubling about every four

centuries, the logarithmic phase of population growth lasting about

1700 years (Deevey et al. 1979).

The Twin Basin Sediment Recora

Using a Livingstone piston corer (Deevey 1965), H.K. Brooks

obtained a sediment core from Lake Yaxha in 1973, and a year later H.H.

Vaughan and D.S. Rice collaborated to raise a shallow-water long core

from Lake Sacnab (Fig. 4). Palynological study of the 7.4-m Yaxha core

revealed a stratigraphy identical to that discovered by isukana (9tt)

in the sediments of Lake Petenxil, 50 km to the west. Seaiments rich

in pollen of grassland and grassland arboreal species were overlain by

postdisturbance mud dominated by high forest pollen indicators.

Evidently, cores from both the lakes had failed to penetrate Maya-

period muds, dense urban settlement at Yaxha resulting in the

deposition of more than 6 m of clay-rich sediment. However, the 6.3-m

Sacnab core demonstrated that the savanna pollen zone was underlain by

organic-rich sediments with a high proportion or horaceae pollen. The

basal, high forest pollen section showed some indications of human

disturbance and was assigned an Early Preclassic age, a time period for

which archaeological evidence is lacking in Peten, but that is coeval

with the Early Formative at Cuello, in Belize. With supplementary

evidence for a predisturbance high forest zone coming from Lake Quexil

(Vaughan 1979, Deevey et al. 1979), Tsukada's (1966) conclusion that

the Maya had transformed savanna into high forest was reevaluated. In

fact, the Maya had converted high forest into grassland, the forests

later recovering after the abandonment of the region.

Figure 4. Bathymetric maps of Lakes Yaxha and Sacnab, showing the locations where long cores were
retrieved for paleolimnological study.

0 LONG CORE SITES
------. *3 METER SHORE, 1962
- SHORE LINE. 1973
6 DEPTH IN METERS
A TRAP
CORE

0 1 2
km

Phosphorus Loading of Lakes Yaxha and Sacnab

Efforts to assess past changes in trophic state ot the lakes were

attempted through an evaluation of ancient phosphorus budgets.

Phosphorus has been shown to be the limiting nutrient in many aquatic

systems (Schindler 1974, Vallentyne 1974) with productivity highly

dependent on total phosphorus loading rates (Vollenheiaer 19tb, Lillon

and Rigler 1974, Oglesby and Schaffner 1978). Phosphorus, unlike

carbon and nitrogen, lacks an atmospheric compartment in its

biogeochemical cycle. Thus, nearly all phosphorus transported from the

watershed to the lake is ultimately sequestered in the basin

sediments. Phosphorus movement in the watershed is essentially

unidirectional, removal from a lake by.mechanisms such as insect

emergence representing a small proportion of the phosphorus originally

delivered to the lake (Vallentyne 1952).

Past rates of nutrient delivery to the basin can be calculated

when phosphorus concentrations in the seaiment are known ana several

levels in a core are accurately dated. Unfortunately, radiocarbon

dates run on bulk sediments from the Peten lakes, even those pretreated

for carbonate removal, proved to be unreliable due to the hard-water

lake effect. Ancient carbonates, lacking the radioisotope, are

solubilized, and their carbon is incorporated into autochthonous

organic material, making dates appear too old (Deevey ana Stuiver

1964). Ultimately, all bulk sediment radiocarbon dates were

disregarded and archaeologically correlated dates were assiSnea to

levels in the cores based on the identification of discrete pollen

assemblages (Vaughan and Deevey, 1961). The assumption used for zoning

the cores was that the degree of deforestation expressed in the pollen

profiles reflected changing population densities as derived from the

archaeological program.

When phosphorus loading rates to the twin basins weLe calculated,

they tracked the slow, steady exponential rise in population aensity

that occurred between the Middle Preclassic and Late Classic. Roughly

delimiting the area circumscribed by the two drainages, export of

phosphorus from the watersheds, already seen to be haya density

dependent, was shown to occur at a rate of about 0.5 kg*person-1

yrt-1. The calculated per capital rate of delivery is, perhaps

coincidentally, almost equal to the physiological output of phosphorus

(0.55 kg) from human bodies that are in equiliLrium with respect to

their intake and output of the nutrient (Vollenweider 1968). Watershed

soils were the principal source of disturbance-zone phosphorus,

nutrient-rich surface soils moving rapidly downhill by colluviation

following Maya-induced deforestation (Brenner 1976, Leevey et al.

1979). Because bulk soil transport was the primary mechanism carrying

phosphorus from the land to the lake, it was impossible to determine

what proportion of the sedimented nutrient actually cycled through

human bodies.

Disturbance-zone sediments of the twin lakes are highly inorganic

and do not indicate that the episode of deforestation was accompanied

by high lacustrine productivity. Large amounts of silica and detrital

carbonate were transported with the phosphorus as soil was eroded into

the lakes. The siliceous, Maya-zone sediments are aominateo by

montmorillonite clay, the residue of the down-wasted country rock.

High phosphorus delivery rates might be expected to correlate with high

autochthonous organic production. That high rates of nutrient input

were associated with inorganic mud suggested that much seaimented

phosphorus was biologically unavailable and bypassed the biota before

deposition on the lake bottom. Corroborating evidence for this

inference came from microfossil analyses that failed to demonstrate

that high primary or secondary production was associated with the

enhanced phosphorus delivery (Brenner 1978). Unfortunately, enumerated

microfossil remains from Peten sediments provide unreliable estimates

of past productivity. Though amounts and accumulation rates of

microfossils decreased with increasing phosphorus input to the twin

basins, strong diagenesis (postdepositional destruction) makes the

thanatocoenosis a poor estimator of the biocoenosis trom which it was

derived.

IhE HISTORICAL ECCLOGY OF THE MAYA AT LAKES QUEXiL,

SALPETEN, AND MACANCHE

The Archaeological Recora

The goal of this stuoy is to expand the geographic scope of the

"Historical Ecology of the Maya" project westwaLo to exazaite tthet

unexplored lake basins. With soil, archaeological, ano

paleolimnological oata trom the new watersheds, seolnmentatioi, processes

in the Peten lakes can be evaluate better, and the phosphorus loading

model developed at Yaxha ano Sacnab can be testee. Ihe new lakes

considered, Quexil, Macanche, ana Salpeten, hitter trom the twin basins

in failing to display the slow, steaay exponential human population

growth from the Middle Preclassic until the Late Classic peak.

Instead, the archaeological record shows that haya population it, the

western watersheds experience a decline curing the Terminal Freclassic

(100 B.C.-250 A.D.) ano Early Classic (25b-550 A.D.) before risitr. once

again in the Late Classic (550-850 A.D.) and finally collapsing about

650 A.D.

Archaeological study of the three watersheds was oirecteo Lu

Don S. and Prudence M. Rice during the 1979 and 1980 field seasons. to

sample about k5% of the watershed area, three transects raOiating itcm

the lake edge were established in each basin, ano some 25% of the

mounds surveyed were test-pitted to determine time of occupation. At

Quexil (Fig. 5), a single north shore transect was lasi out ana two

south shore sampling plots were staked. Transects were run in a north-

south direction, and as at Yaxha were U.b km wloe. In keeping with

procedures used at the twin basins, the north shore transect was 2 km

long, but the south shore surveys were run about .2i km, to the Lase

of the karst uplift that encloses the basin. This permitted the

sampling of a large savanna south of the lake. Aaoitionaill, Quexil's

two islands were searched, revealing b structures on the western island

and 20 on the eastern island.

At Salpeten, standard size (2 km x U.5 km) transects uere

established with north-south orientation, two on the north shore ano a

single plot on the south shore (Fig. 6). A large, oensely settler site

on salpeten's peninsula was also investigated. The site, called

Zacpeten, is primarily of Postclassic age. At hacanche (Fig. 7), the

single north shore transect and two south shore plots were oriented

northwest-southeast. Mounds discovered within a walled site (Op 2),

Muralla de Leon, were also surveyed, as were some structures between

transects. In this basin, time period of occupation was determined by

test-pitting about 25% of the mounds both on and off transects. Using

the proportion of mounas showing occupation outing the various serious,

population levels were figure based on the density of mounds founa on

the aelimitec transects. In all three drainages, changing, haya

population densities were calculated employing the assumptions used at

Tikal (Haviland 1970) and at Yaxha ane SacnaL, that 64% of the icunus

were residences ana each household consisted of 5.6 persons. Figure b

Figure 5. Lake Quexil, showing the position of the archaeological
sampling transects and soil pits. Op = operation.

LAKE QUEXIL
0 km 1
SArchaeological Transects
Soil Pits

~^/
I
-iu

Op2

Figure 6. Lake Salpeten, showing the position oi the archaeological
sampling transects ana soil pits. Op = operation.

LAKE SALPETEN

0 km 1
I I

SArchaeological Transects
SSoil Pits

Op 3

Figure 7. Lake Macanche, showing the position ot the archaeological
sampling transects and soil pits. Op = operation.

LAKE MACANCHE

0 km 1

[ | Archaeological Transects
SSoil Pits
l'Op 2

7
Op 1

1N-

Op 3

Figure 8. Cruce population densities, calculated trom housemoura
densities, in the tive lake basins for which both
archaeological and paleolimnological histories have teen
studied. Population reconstructions are based on
radiocarbon dated ceramic chronologies.

300-

200-
YAXHA
100-

0 1000

200
SACNAB
100

0 1000

200
SQOUEXIL
S100-

0 1000

300-

200o MACANCHE

100-

0 1000

300-

200- SALPETEN

100-

0 1000

I I
1000 0 1000 2000
B.C. A.D.

Date

shows the changing population patterns for five Peten watersheds base

on mainland transect mouno densities.

Soil Chemistry

Translocatec soils were shown to have been the principai source or

phosphorus for Lakes Yaxha and Sacnab curing the period of haya forest

disturbance (Deevey et al. 1975). In extening the stuuy to three new

watersheds, soil samples were collected from several localities in each

basin to assess total phosphorus ana other chemical concentrations and

distributions in riparian soil profiles. Because bulk movement of

soils is the mechanism by which phosphorus reaches the lakes, totat

phosphorus was analyzed rather than the plant-available fraction.

During the 19b0 fiela season, 21 soil pits were aug in the ,uexil

drainage (Fig. 5), while 11 trenches were dug in both the Nacanche and

Salpeten basins (Figs. 6 anc 7). In each case, soil pit locations were

determined subjectively in order to sample several microtopographic

zones as well as areas associated with haya construction. IIn most

instances, samples were collected along archaeological transects so

that distances from shore coulQ Le determined by rereience to

established, staked positions in the archaeological sampling plot.

Soil pits were dug to bedrock, sterile "sascab," or as oeep as ha%

feasible. Maximum depth from site to site was variable, but none of

the sample depths exceeded 100 cm. Samples were remove rrom the

exposed profile wall at 10 cm intervals, each sample representing a

composite of soil collected over the full 10 cm. About 100-200 g of

soil were taken at each level ano stored in inoivioual plastic bags.

In the lab, four levels from each soil pit (0-10 cm, 10-20 cm, an

intermediate level between 20 cm and bottom, ana bottom) were removed

and air-cried, though only top ano bottom samples were considered in

Quexil's shallow pit #4. A subsample of each air-orieu level was

extracted and ground with a mortar ana pestle. A portion of the

powered soil (1-6 g) was removed, weighed, ano place in a 'ihermoiylte

Type 1500 furnace for two hours at 5500C to assess organic matter

content by loss on ignition. Following combustion ana reweighing, a

0.3-1.5 g sample of ash was remove, weighed, ano digested in lb ml of

a heated 2:1 nitric-perchloric acio mixture. hhen aense, white hbloU

fumes appeared, the 250 ml beakers containing the mixture were covered

with watchglasses and the oxidation process was continued toL an

additional hour, with distilled water added periodically to prevent

total drying. After digestion, samples were tilterer atnc the nitrate

was brought to a known volume. The filtrate was delivered to the

University of Florida Institute of Food ano Agricultural sciences Soils

Laboratory, where cation concentrations were determined by atomic

absorption ano phosphorus analyses were run on a lechnicc.n Auto

Analyzer. I measured sulfur content in filtrate samples from surface

ana bottom profile levels using the turbioimetric procedure in StanoaLo

Methods (APHA 1971). Sulfate turbidity was read on a Coleman Mooel 14

Universal Spectrophotometer. Prior to running sultate analyses on the

digested, ashed soils, a series of paired ash-whole soil determinations

were made to assess sulfur loss due to ignition. Measured sulfur

content, as concentration per grani whole soil was shown not to aitfte

statistically when the two methods were compared. All soil chemical

concentrations are expressed as amount per gram of air-orieo wiole

soil.

Total phosphorus analyses run on soil profiles from Quexil,

Macanche, and Salpeten reveal that in all three watersheos a strong

gradient is maintained with respect to the nutrient, surface soils

clearly enriched as compared to levels deeper in the profiles (Figs.

5-12). Though exceptions oo occur, increasing depth in the profile is

generally accompanied by decreasing phosphorus concentration. This

trend is evident when the mean surface soil (0-10 cm) phosphorus

concentration in a basin is compared to the mean value obtained for the

basal (variable depth) levels of the pits. At Quexil, where the 21

soil pits were dug to an average of 74.3 cm, mean phosphorus

concentration in surface soils was 178 ug P'g-1, or 2.46 times the

average concentration (112 ug Pg9-1) calculated tor the deepest

strata in the pits. The average depth attained at Macanche was 73.6

cm, ano an even stronger phosphorus graoient was revealed by analyses

of the 11 profiles from the watershed. The mean surface soil

phosphorus level (594 ug P*g-1) exceeds the bottom concentration

(181 ug P*g-1) by 3.28 times. At Salpeten, where 11 pits

bottomed-out at a mean depth of 62.7 cm, the average surficial soil

nutrient concentration (598 ug P*g-1) was 2.21 times the mean founa

in the basal profile levels (271 ug P.g-i).

Figure 9. Total phosphorus concentrations at selected levels in soil
pits 1-12 at Quexil.

QUEXIL

0
20

40
60"
Bo PIT 1 PIT2 PIT3

100

E 40
60
S80 PIT4 PIT 5 PIT 6
C 100

0

.. 20.
0
4 40
60
so PIT 7 PIT a PIT 9
100 i .

0
20
40
60

80 PIT 10 PIT 11 PIT 12

100, 1 I

0 200 400 600 800 1000 0 200 400 600 800 1000 0 200 400 600 800 1000

,g P-g-i

Figure 10. Total phosphorus concentrations at selected levels in soil
pits 13-21 at Quexil.

QUEXIL

0
1/
20
40
60
so PIT 13 PIT 14
100

PIT 15

0
20
40
60
so PIT 16 PIT 17 PIT 18
100 I I I

0
20-
40-
60.
80 PIT 19
100 ,
I0 I0 I 8I 1
0 200 400 600 800 1000

( PIT 20 PIT 21

200 600 800 1000 0 200 600 1000
0 200 400 600 800 1000 0 200 400 600 800 1000

pg P-g9-

Figure 11. Total phosphorus concentrations at selected levels in 11
soil pits at Salpeten.

SALPETEN

PIT 3

0
20
40
60.
0o PIT 4 PIT 5 PIT 6
too

PIT 8

PIT9

0
20
40
60
so PIT 10
100

0 200 400 600 800 1000

PIT 11

0 200 400 600 800 1000 0 200 400 600 800 1000

pug P-g-'

Figure 12. Total phosphorus concentrations at selected levels in ii
soil pits at Macanche.

MACANCHE

0

20

60

100

0

20
E 40
U 60
. so PIT4 PIT5 PIT6

0.

20
40

60

'o PIT7 PIT8 PIT 9
100 .
os

20

60"

8o PIT 10 PIT 1
100
I II i I I I I I I I I I I
0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200 0 200 400 600 800 1000

ug P-g-'

To assess how organic matter and various soil chemical

concentrations change with oepth in the profile, mean conLcentrations

for all of the combined top two levels (0-1U cm and i0-20 cm) in the

profiles of a basin are compare to the average concentrations

calculated for the summed bottom two levels (intermediate between 20 cm

ana bottom, plus bottom) of the basin pits liable i). Quexil's shallow

pit #4 was excluded from this tabulation because only two levels were

analyzed, ano assessments ot changing sultur concentrations witi aepth

are reliant on surface and bottom analyses only. Aaoitionally, whole

profile mean concentrations tor the various soil paraneteir are given,

thereby permitting a rough interbasin comparison of soil

characteristics.

Organic matter distribution in the soil profiles displays a trent

similar to that seen for phosphorus, % loss on ignition generally

decreasing with greater depth. Surface soils are also enriched in

sulfur, deeper profile levels noticeably deficient with respect to the

nutrient. Similar trends in organic matter 'anu sulfur concentrations

suggest that much soil sulfur is present in organic form. However,

ignition of the.samples at b50C caused no apparent loss ot sultur,

perhaps indicating that the bulk of sulfur is present in mineral form,

or that the organic sulfur fraction is not volatilized on1 turning.

High inorganic sulfur content would not be unexpected, particularly at

Salpeten where a gypsum outcrop overlooks the northwest shore ot the

lake. Were this the case, sulfur distribution woula be expected to

track magnesium and calcium in the profiles, but it aoes not.

Table 1. Summarized chemical concentrations from soil pits ouy in the
Quexil, Macanche, and Salpeten watersheds. Concentrations in
the two uppermost levels of the profile were aveiageu as were
the two bottommost samples, thus giving a rough idea of
chemical graaients in the soil profiles. Whole profile mean

concentrations
concentrations
only.

are also given for each chemical type. Sulfur
were determined on surface ana bottom levels

Profile Profile Loss on Ca Mg Fe
Levels Depth n Ignition % ag*gm-1 mygm-i1 si .gs-I

QUEX1L

Top 2 1-10 cm
Levels 10-20 cm

Bottom 2
Levels

Variable

Whole All
Profile Samples

Top 2
Levels

Bottom 2
Levels

14.1

0-10 cm
10-20 cm

Variable

3.4 2b.3

MALANCHE

Whole All
Profile Samples

SALPETEN

Top 2 0-10 cm
Levels 10-20 cm

44 14.3 2b4

31.2 5.7

Bottom 2
Levels

Whole
Profile

Variable

All
Samples

Table 1--extended.

Al Na K P S
mggm-g1 ug*gm-1 uggm-1 ug gm-1 9 -gi1

41.6

47.2

44.4

8.6

8.0

8.3

13.1

16.0

14.6

848

673

761

609

453

531

1240

1226

1233

Cation distributions throughout the soil profiles are more even,

without the pronounced top-to-Lottom gradieit seen tot suitut,

phosphorus and organic matter. Potassium concentrations are slightly

higher in upper soils of the profiles, while sodium content is a bit

richer in the deeper levels. In all three watersheds, calcium is a bit

more concentrated in deeper soils, a trena that is anticipated in

limestone terrain. Magnesium displays rather uniform concentration

throughout the soil profiles, and though at Quexil upper level strata

contain slightly higher amounts of the cation than deep soils, the

trend is reversed at Macanche and Salpeten.

Intra-basin variation in the chemical profiles is most apparent at

Quexil, where forest soils, presumably of the Yaxa series, were sampled

along with Exkixil, savanna soils. Low fertility, Exkixil soils are

depleted in phosphorus as compared to forest soils. Surface soils from

the six savanna profiles (#1, #5, #8, #9, #10, #11) have a mean

concentration of 164 ug P*g-1, while the 15 forest pits display an

average upper level concentration about twice as high (324 ug

P.g-1). Disregarding shallow pit #4, mean whole profile

concentrations for various chemical constituents in the six savanna

pits can be compared to values obtained on the 14 forest trenches.

Aluminum is highly concentrated in the clay-rich savanna soils (105.4

mg Al'g-1), whereas the forest profiles contain only 18.4 mg

Al-g-1 High iron content in the savanna soils evidently is

responsible for the rich red color of the earth south of Quexil, and

savanna iron levels (b7.15 mg Fe.g-1) exceed forest soil levels

(7.0 mg Feg9-1) by nearly an order of magnitude. The lack ot

calcium in grassland soils is striking, as they possess only 2.4 mg

Ca*g-1. Forest soils, differing by more than two orders or

magnitude, contain 256.1 mg Ca*g-1. Purest soils are also richer

in magnesium than savanna soils, the two series containing 4.1 nmg

Mg*g-l and 1.7 mg hg'g-1 respectively.

When mean, whole profile chemical concentrations are compared on

an inter-orainage basis, several differences are apparent. Average

iron and aluminum concentrations are higher at Quexil than at Macanche

and Salpeten because of the high metal content of the grassianb soils

at Quexil. Likewise, calcium deficiency in the Quexil savannas is

largely responsible for giving that basin an overall mean whole protlie

calcium value that is relatively low. Magnesium content of the

Macanche and Salpeten soils exceeds that in Quexil by.7.1 anoa 9. times

respectively. The difference is not accounted for solely by the low

magnesium content of Quexil's savanna profiles, because even forest

soils at Quexil possess considerably less magnesium than encountered in

samples from Salpeten anr hacanche. Lolomiitizatior. in the hacanche-

Salpeten-district, also reflected in.the water chemistry of the two

lakes, is the probable cause of high magnesium levels in the basin

soils.

Though it is not certain, high sulfur concentrations eetermir.ed

for the hacanche and Salpeten watershed soils probably point to the

presence of gypsum in the underlying bedrock of these uasins.

Evaporites are evidently less common at Quexil, ana soil sulfur values

from the three basins generally reflect chemical levels in the lake

waters (Deevey et al. 1980a).

Phosphorus is highly concentrated in the upper levels of Peten

soil profiles, and under conditions of deforestation, erosional

processes would be expected to carry large amounts of the nutrient to

the lakes. Rapid mobilization of the available phosphorus traction is

possible, but not likely, as the homogeneous distribution in the

profiles of highly soluble sodium and potassium argues against

leaching.

Maya agro-engineering activities enhance delivery rates ot

phosphorus to the lakes not only by accelerating downhill bulk

transport of soils, but perhaps by first concentrating the nutrient in

surface soils through physiological cycling, interment, ano refuse

disposal. Though the oata are meager, three soil profiles Qug in

housemounds as well as several other pits located near construction

contain extremely high levels of phosphorus suggesting an anthropogenic

source for the nutrient in enriched surface soils. At Saipeten, south

shore soil pits #6 ano #7 were piaceo in housemounos sibb ano jitj,

last occupied during Late Classic times. Surface soils from these pits

contain 821 ug P*g-1 ano 9i2 ug P*g-1 respectively, much more

than the overall mean value for surficial soils in the basin (558 ug

P.g-1). On the north shore of the lake, soil pit tl was locate at

the crest of a steep slope, in close proximity to rubble from collapse

construction. Ceramic sheros were encountered in the excavated soil

pits, and its surface soils contained 995 ug P*g-1. Dowunlope tron

this site, topsoil from pit #2 yielded 895 ug P-g-l. The four

Salpeten soil pits within or near Maya structures had an average

surface soil nutrient concentration of 906 ug pg-l1, while top

level soils of the remaining seven pits had a mean of only 42C ug

p.g-l.

At Quexil, soil pit #21 was located in south shore housemouno

#823, a structure that was occupied during middle and Late Preclassic

times ana again in the Late classic period. Surficial soils from the

pit contain 467 ug P-g-1, significantly more than the overall

surface soil mean of 278 ug P-g-1 for the waterbhea. At hacanche,

pits #1 and #2 lie on the north shore slope just below construction ano

their highly enriched surface soils contain 1074 ug Pg9-l ano 1275

ug p9g-1 respectively, exceeding the mean surface concentration

(465 ug P*g-1) found in the remaining 5 pits.

PALEOLIMNOLOGY OF LAKE QUEXIL

Comparing Shallow-hater ana Deep-hater Sedimentation

The primary goal of the paleolimnological research undertaken in

the three new watersheos was to assess how the proximate composition ot

the sediments and accumulation rates of various chemical constituents

of the muo varied as a function of shifting Maya population levels.

The first basin considered was Lake Quexil, lying some b km east of

Flores and only 1 km from Lake Petenxil (Fig. 2), the basin where Maya

impact on the Peten sedimentary record was first revealed (Cowgill et

al. 1966).

In 1972, H.h. Vaughan ano G.H. Yezoani useu a Livingstone piston

corer (Deevey 1965) to get a 6.5-m section in 7 m of water. Ihe core

was taken south of the lake's western island (Fig. 13). The sfali lake

(area = 2.101 km2, 2max = 32 m, = 7.2 rt: Deevey et al. 18ba)

was cored again in 1978. In March of that year, M.S. Flannery, 5.L.

Garrett-Jones, ano 1 raised a 9.2-m core from 2b ru ot watel usin a

modified, gravity-driven, Kullenberg apparatus (Fig. 13). This core,

designated Quexil H, was one of several long cores coiiecteu in the

lake's deep, central basin in our effort to secure sediments of

Pleistocene age.

Figure 13. Bathymetric map of Lake Quexil, showing the locations of several coring sites in the
basin.

LAKE QUEXIL S

Site Core
S Shallow-
C C
H H
F F
X 80-1
X 80-2

Year
1972
1978
1978
1978
1980
1980

Water death
7m
29.8 m
27.7 m
19.8 m
29.0 m
29.0 m

CONTOUR INTERVAL 2m
0 500 IW m

SCALE

Depth below M.W. Interface
0-6.5 m
0-7.9 m
0-9.2 m
0-9.8 m
7.49-19.88 m
4.57-9.65 m

Lake
Quexil
Quexil
Quexil
Quexll
Quexll
Quexil

The shallow-water core was returned to Florida in the aluminum.

coring tubes, ano the ,uexil H core was transported to the Florioa

State Museum in the plastic tubing that lined the iron Kullenberg

coring pipe. Cores were refrigerated at 4C prior to ano following

extrusion. Sediment chemistry ana palynology of the shallow-water

section were reported elsewhere (Brenner 1978, Deevey et ai. 1975,

Vaughan 1979), but without accompanying aata on Maya population

densities in the basin. This consideration not only correlates

sedimentary changes with shifting population levels, but by comparing

two sediment columns, demonstrates the profound influence that core

location and basin morphometry haa on sediment chemistry ana measure

sediment accumulation rates.

Samples were removed from the extruaeo cores at 5-20 cm intervals,

and water content was evaluated on weighed volumetric samples by drying

at 1100c. A second set of samples was dried for total carbon ano

nitrogen analyses that were run on a Perkin-Elmer Model 240 Elemental

Analyzer. A third series of volumetric samples was weighle anr

digested in 15 ml of 2:1 nitric-perchloric acid. After oxidation, the

samples were filtered, ana the filtrate was brought to a known volume.

Cation analyses were run on the filtrate by atomic absorption at the

University of Florida Institute of Food ana Agricultural Sciences Soils

Laboratory. Filtered digestate from the deep core samples was analyze

for phosphorus content on a lechnicon Auto Analyzer at the Soils

Laboratory. Aliquots of digestate from the shallow-water core were

retained for phosphorus analyses, which were run colcrimetrically on a

Coleman Model 14 Universal Spectrophotometer following blue color

development by the ascorbic acid-ammonium molyboate method in Standaro

Methods (APHA 1971). Sulfur content in Quexil core H was measured by

assessing the quantity in the filtered digestate using the

turbidimetric technique in Standard Methods (APHA 1971).

When all chemical analyses were complete, chemical concentrations

in dry sediment were calculated (Figs. 14 and 15), and level by level

proximate composition of the sediment was figured (Figs. 16 ano 17). To

compute the chemical make-up of the mud at each level, the carbonate

equivalents of magnesium and calcium were first calculated, thereby

permitting the assessment of inorganic (carbonate) carbon content.

Next, the inorganic carbon quantity was subtracted from the total

carbon value to yield an organic carbon figure. Then, as at Yaxha ana

Sacnab (Deevey and Rice 1980), the organic carbon value was multiplied

by 2.5 to produce a figure for organic matter content. Iron is

reported as the oxide, Fe203, and Si02, likely an

alumino-silicate, is the residue following subtraction of organic

matter, CaCO3, MgCO3, and Fe203.

Dating the cores was once again dependent on changes in the

relative pollen diagrams. An exception was provided by numerous wood

fragments that were encountered at 623-624 cm in the shallow-water

core. Age determination on these allochthonous plant remains is free

from the confounding effects of haro-water-lake error, ano a dated

sample (DAL 198) gave a 14C age of 8410 + 160 years (Ogaen and Hart

1977). Corrected to about 9400 siaereal years (Deevey et al. 1979),

Figure 14. The chemical stratigraphy of the Quexil shallow-water core.

QUEXIL
Shallow-Water Core

Ctot NIot Ptot Ca Mg Fe *SiO"
mg-g-' mg-g-' pg.g-' mgg-i' mg"g-' mg-gg1' mBgg'

Corg
mgg-'

Figure 15. The chemical stratigraphy of Quexil core H.

QUEXIL
Core H

CaCO3&
Corg MgCO3 Ctot Ntot C/N Ptot Ca Mg Fe Mn K S SiO
mg*g-I mg.g-' mg.g-' mg.g-' Ai1g-1 mg.g-' mg-g-' mg.g-' gg-' i pg.g' mgg'-1 mg.-g-
100 200 300 40 0 100 300 5 15 10 20 400 00 I 5 15 20 40 200 600 00 00 10 20 400 800

DRY
WEIGHT
%

MILLISECOND	CLASS.METHOD	MESSAGE
0	sobekcm_page_globals.constructor
0	sobekcm_page_globals.constructor	Application State validated or built
0	sobekcm_database.verify_item_lookup_object
0	sobekcm_page_globals.constructor	Navigation Object created from URI query string
0	sobekcm_database.verify_item_lookup_object
0	sobekcm_page_globals.display_item	Retrieving item or group information
0	sobekcm_page_globals.get_entire_collection_hierarchy	Retrieving hierarchy information
0	sobekcm_assistant.get_entire_collection_hierarchy
0	cached_data_manager.retrieve_item_aggregation
0	cached_data_manager.retrieve_item_aggregation	Found item aggregation on local cache
0	item_aggregation_builder.get_item_aggregation	Found 'all' item aggregation in cache
0	system.web.ui.page.page_load (ufdc.page_load)
0	sobekcm_page_globals.constructor.on_page_load
0	html_echo_mainwriter.add_style_references	Adding style references to HTML
0	html_echo_mainwriter.add_text_to_page	Reading the text from the file and echoing back to the output stream
2	html_echo_mainwriter.add_text_to_page	Finished reading and writing the file