Agricultural and Historical Ecology of the Lake Region of Peten, Guatemala

Agricultural and historical ecology of the lake region of Peten, Guatemala

Item Type Dissertation-Reproduction (electronic); text

Authors Wiseman, Frederick Matthew, 1948-

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 06/10/2021 11:19:15

Link to Item http://hdl.handle.net/10150/191049 AGRICULTURAL AND HISTORICAL ECOLOGY OF

THE LAKE REGION OF PETEN, GUATEMALA

by Frederick Matthew Wiseman

A Dissertation Submitted to the Faculty of the DEPARTMENT OF GEOSCIENCES

In Partial Fulfillment of the Requirements For the Degree of DOCTOR OF PHILOSOPHY

In the Graduate College THE UNIVERSITY OF ARIZONA

1978 THE UNIVERSITY OF ARIZONA

GRADUATE COLLEGE

I hereby recommend that this dissertation prepared under my

direction by Frederick Matthew Wiseman

entitled AGRICULTURAL AND HISTORICAL ECOLOGY OF THE LAKE REGION OF PETEN, GUATEMALA

be accepted as fulfilling the dissertation requirement for the

degree of Doctor of Philosophy

'brAit; Ix- Do-77 Dissertation Director Date

As members of the Final Examination Committee, we certify

that we have read this dissertation and agree that it may be

presented for final defense.

Teti, 1 0 /97r( ,

Final approval and acceptance of this dissertation is contingent on the candidate's adequate performance and defense thereof at the final oral examination. STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to bor- rowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his judgment the proposed use of the material is in the in- terests of scholarship. In all other instances, however, permission must be obtained from the author.

SIGNED: --;?2,„1„_0‹ ACKNOWLEDGMENTS

I wish to express my gratitude to those who have been most helpful during my research, the results of which are presented here.

Partial support for my field work has come from the National Science

Foundation through its grant (BMS 72-01859) to Dr. E. S. Deevey. Funds for laboratory analysis and computer time were kindly made available through the Department of Geosciences, University of Arizona, and from

Dr. Paul S. Martin (NSF grant DEB 75-13944).

Dr. T. Patrick Culbert has continuously offered all possible aid in the archaeological and anthropological sections of this paper, as well as translating palynological jargon into comprehensible lan- guage. Drs. V. C. LaMarche, T. J. Blasing and Ms. Linda Drew of the

Laboratory of Tree-Ring Research have made some order of the chaos of data through their aid in the realm of computers. Dr. A. M. Solomon has kindly assisted me in understanding the mechanics of modern pollen transport which is critical to a thorough analysis of pollen influx.

Dr. Raymond R. Turner, U.S. Geological Survey, has aided this research in the areas of remote sensing and vegetation analysis. Dis- cussions with David P. Adam, Wayne Wendland, Emil Haury, Jack Ewel, Ken

Petersen and others have aided in the formulation of many of the con- cepts described in this work. Betty Fink has edited and typed many sections of this and other papers. iv

Finally, I wish to thank Martha Ames, whose aid in field and final preparation has allowed me to complete this work. TABLE OF CONTENTS

Page

LIST OF TLLUSTRATIONS viii

LIST OF TABLES ix

ABSTRACT xi

1. INTRODUCTION 1 2. THE ABIOTIC BACKGROUND 4

Geological Considerations 4 Climatic Regime of the Maya Lowlands 6 Modern Climate 6 Climatic History 8

3. MODERN VEGETATION ANALYSIS 13 Field Methods 13 Numerical Classification of Sample Sites 18 Vegetation Mapping 19 Comparison of Habitat Variability -- Mayan Lowlands vs. the Sonoran Region 23

4. MODERN POLLEN ANALYSIS 28 Introduction 28 Pollen Collection and Extraction 29

5. RESULTS OF THE VEGETATION AND POLLEN ANALYSES 38 Vegetation Types 39 Type 1: Pachira apuatica-Terminalia Swamp . . . 39 Type 2: Mesic Forest 44 Type 3: Monsoon Forest 48 Type 4: Orbignya cohune Forest 53 Type 5: Aguada Forest Vegetation 56 Type 6: Encinal (Oak Woodland) 59 Type 7: Thorn Scrub 63 Type 8: Sahalal 64 Type 9: Nanzal 67 Type 10: Grasslands 69 vi

TABLE OF CONTENTS - -Continued

Page

The Independent Data Set: Peten Sediment Network . . . 72 Sediment Network Peten-1, Aguada Tikal, Peten, Guatemala 75 Sediment Network Peten-2, Lake Petenxil, Peten, Guatemala 76 Sediment Network Peten-3, Lake Eckixil, Peten, Guatemala 81 Sediment Network Peten-3, Aguada "Libertad," Peten, Guatemala 84 Sediment Network Peten-4, Lake Oquevix, Peten, Guatemala 84

6. ETHNOECOLOGY 89

Introduction 89 The Limits to Agricultural Growth 90 The Agricultural Systems 97 System 1: Milpa Agriculture 97 System 2: Dooryard Gardens 106 System 3: Fuel Procurement 108 System 4: Hunting 110 Prehistoric Maya Subsistence 112 System 5: Intensive Milpa 115 System 6: Artificial Rain Forest 115 System 7: Arborculture 123 System 8: Terracing 124 System 9: Ridged Fields 124 Agricultural Simulation 126 Scene 1: Preclassic and Postclassic Agriculture 135 Scene 2: The Chicanel Expansion 135 Scene 3: Classic Period Intensive Agriculture • • 136

7. A DESCRIPTIVE MODEL FOR POLLEN/VEGETATION CHANGE IN THE MODERN MAYA LOWLANDS 139

Derivation of the Analog Model 141 Analog 1: The Edaphic Scale 143 Analog 2: The Successional Scale 146 Comparison of Analogs 1 and 2 148 Use of the Analog Model 149

8. FOSSIL POLLEN ANALYSIS 151

Results of the Fossil Pollen Core Analyses 152 vii

TABLE OF CONTENTS--Continued

Page

Core P-17 Lake Petenxil, Peten, Guatemala . . . . 152 Core P-20 Lake Petenxil, Peten, Guatemala . . . . 153 Core E-5 Lake Eckixil, Peten, Guatemala 156 Core E-8 Lake Eckixil, Peten, Guatemala 159 The Pollen Sequence of the Lake Region of Peten • • • 159 Principal Components Analysis of Cores 162 Application of the Analogs to the Fossil Data . . . . 166

9. RECONSTRUCTION OF THE PREHISPANIC MAYA LANDSCAPE 169

A Preliminary Agricultural Sequence for the Central Peten 172 Middle and Late Preclassic, 500 B.C.-A.D. 250 (Base of Core) 172 Classic Period A.D. 250-A.D.850 174 The Terminal Classic Maize Episode 176 The Maya Collapse A.D. 800-A.D. 900 180 The Early Postclassic A.D. 900-A.D. 1300 181 The Itza Intrusion A.D. 1350-A.D. 1697 The Colonial Period 1697- 182

LIST OF REFERENCES 183 LIST OF ILLUSTRATIONS

Figure Page 1. Mean monthly precipitation, Flores, Peten 9 2. Site location map 15

3. Vegetation stand similarity index values; sites SL-1, VT-3, and VT-9 21

4. Vegetation of the central Peten 24

5. Pollen degradation rate 32

6. Pollen sample network, Lake Petenxil, Peten 78 7. Pollen sample network, Lake Oquevix, Peten 87 8. Labor input -- milpa system 99 9. Cover values for five plant classes, Site VT-5, Santa Elena, Peten 105

10. Artificial rain forest, a hypothetical reconstruction . . 121 11. Program MAYAPOL flowchart 128

12. Simulation polygon used for calculating pollen influx values 129

13. Tropical agricultural systems arrayed along two edaphic gradients 134

14. The bidimensional model of modern vegetation of the Maya Lowlands 140

15. Eigenvector amplitudes of modern pollen sample sites . . 145

16. Eigenvector amplitudes that describe disturbance pollen . 147

17. Agricultural indicators, Lakes Eckixil and Petenxil, Peten, Guatemala 173

viii LIST OF TABLES

Table Page

1. Site data, 22 LANDSAT-1 ground truth sites 16 2. Similarity index matrix for 22 LANDSAT ground-truth sites, Peten, Guatemala 20

3. Classification of sample stands derived from similarity indices 22

4. Post-collection pollen degradation in tropical soil samples 31

5. Pollen extraction method 35 6. Pollen and vegetation data, Site VT-2, Finca Eckixil, Peten 41

7. Pollen and vegetation data, Site VT-1, Petenxil, Peten • 45

8. Pollen and vegetation data, Site SL-7, La Libertad, Peten 51

9. Pollen and vegetation data, Site SL-2, Macanche, Peten . 54

10. Pollen and vegetation data, Site VT-8, Bosque el Caobal, Peten 57

11. Pollen and vegetation data, Site VT-4, Santa Marta, Peten 61

12. Pollen and vegetation data, Site SL-13, Sibun, Peten . . 65

13. Pollen and vegetation data, Site VT-6, El Guanal, Peten . 68

14. Pollen and vegetation data, Site VT-7, Santa Ana Vieja, Peten 70

15. Pollen and vegetation data, Site VT-9, Pacay, Peten . . . 73 16. Pollen percentages of 17 statistically significant taxa, Aguada Tikal, Peten 77 17. Pollen percentages of 17 statistically significant taxa, Lake Petenxil, Peten 80

ix LIST OF TABLES--Continued

Table Page

18. Pollen percentages of 17 statistically significant taxa, Lake Eckixil, Peten 83

19. Pollen percentages of 17 statistically significant taxa, Aguada Libertad, Peten 85

20. Pollen percentages of 17 statistically significant taxa, Lake Oquevix, Peten 88

21.' Important indigenous tropical cultigens, their productivity, and ecological requirements 100

22. Pollen percentages of 17 statistically significant taxa, Site VT-5, Santa Elena, Peten 107

23. Mesofauna inhibiting the Maya Lowlands 111

24. Carrying capacity estimates for the Maya Lowlands . . . . 114

25. Plants that may have been used in an "artificial rain forest" 118

26. Summary of treatments during a 100-year run of program MAYAPOL 131

27. Output of simulations: MAYAPOL program 132

28. Eigenvector weights (scalars) of 17 selected lowland pollen taxa 144

29. Pollen percentages of 17 statistically significant taxa, Core P-17, Lake Petenxil, Peten

30. Pollen percentages of 17 statistically significant taxa, Core P-20, Lake Petenxil, Peten 155

31. Pollen percentages of 17 statistically significant taxa, Core E-5, Lake Eckixil, Peten 157

32. Pollen percentages of 17 statistically significant taxa, Core E-8, Lake Eckixil, Peten 160

33. Eigenvector weights (scalars) of 17 selected lowland pollen taxa, Peten cores 164

34. Matrix of core eigenvector correlation coefficients . . . 165 ABSTRACT

The modern Maya lowlands are covered by a variety of vegetation types, ranging from freshwater swamps, through high "quasi rainforest," to open grasslands, each with its own exploitable potential and effect upon subsistence.

Limiting factors such as pests, leaching, and competition would have decreased the potential harvests of prehistoric Mayan agriculture.

Several ecologically sound methods, including increased crop diversity, mulching, and quarantine measures, reduce the impact of these limiting factors.

Modern Maya agriculture is practiced at such low levels that it evades some limits to its potential productivity. Hypothesized prehistoric systems, such as intensive milpa, ramon cultivation, raised fields, and artificial rain forest, must have reached equilibrium with their biotic, climatic and edaphic environments. Using ethnographic and crop productivity data, with certain assumptions, quantified systems models of prehistoric agriculture have been derived.

An ecologically compatible combination of intensive milpa, artificial rain forest, ridged fields, and marsh cultivation theoretically will support over 400 people per square kilometer of upland in the Peten. These data are within the limits of archaeological demographic estimates ranging from 40 to 900 people/km 2 .

xi xii

Principal components analyses of pollen from edaphic and successional gradients serve as modern analogs for statistical comparison with two cores taken in the lake district of central Peten, Guate- mala. Results indicate that agricultural activity, not climatic change, caused changes in the prehistoric vegetation.

The Maya Classic landscape was an agriculture-dominated regime, with little untouched natural vegetation. Orchards, artificial rain forest and woodlots, although not supported by pollen evidence, may have covered much of the lowlands.

The Maya collapse was followed by a general depopulation of the

Peten. The Peten-Itza recolonization of the lake district, and the modern population influx appear as two minor agricultural episodes in a largely arboreal Postclassic landscape. CHAPTER 1

INTRODUCTION

Palynology has long been used as a method to describe past changes in climate and vegetation. The pollen grain is preserved due to its chemical composition, and can be identified by its structure.

From the time pollen analysis began in Scandanavia, it remained essentially a northern temperate zone technique. Pollen analysis in the

Americas seems to be centered in the east-central United States, probably because of the local abundance of glacial lakes. Any state in that restricted region has had more study than the whole of Latin

America.

The American tropics have come under increasing study. Deevey

(1944) studied the paleolimnology of several Mexican lakes, and included pollen analysis in these studies. Sears and Clisby (1955) studied highland Mexican pollen, Martin (1964) studied the Costa Rica cordillera, followed by Cowgill and Hutchinson's(1966) paleoecological studies of the Maya lowlands, and Van der Hammen's South American research (1963, 1967). Today, much of tropical American palynology is concerned with the origin and spread of various cultigens (MacNeish

1967).

However, much of the theory developed for temperate-zone pollen analysis has yet to be applied in the lower latitudes. Tropical forests, characterized by a high, closed canopy, are essentially

1 2 impervious to wind and are easily visualized as an extreme Tauberian model (Tauber 1965). The low incidence of anemophilous forest trees and high incidence of anemophilous disturbance plants in the tropics necessitate rethinking pollen-vegetation relationships developed in temperate zones. Absolute pollen influx estimates (Davis, Brubaker and Beiswenger 1974) are unknown in the American tropics as are statistical analyses of pollen data (Webb 1974; Birks, Webb and Bert. 1975; Adam 1974), except for a cursory study by Tsukada and Deevey (1967) of highland Guatemala.

Application of these techniques in a new area is fraught with problems since there are no standards for comparison, but until such analyses are performed, the American tropics will remain a poor paly- n6logical cousin to the Great Lakes region of the United States.

A greater problem exists in the realm of human landscape modification. Archaeologists and paleoecologists in northern Europe and the eastern United States are considering the utility of pollen analysis for paleodemographic and subsistence hypothesis testing, as should the tropical paleoecologist.

The archaeology of southern Mesoamerica is probably better known than that of any other tropical region. However, the complexity of the extinct social and subsistence bases exceeds by orders of magnitude temperate-zone cultures. This fact alone should give cultural paleoecologists pause; combined with the aforementioned problems in pollen analysis, the job may seem hopeless.

A method of ordinating the chaos of variables is needed in an analysis of prehistoric cultural ecology. Systems theory breaks a 3 multifaceted system into its components, each having a quantified relation to all others. When applied to modern sociocultural-ecological systems, this method derives response and transfer functions which may then be tested against archaeological and palynological reality. The modern analog is the keystone of a systems approach. If a model can predict variables synchronically, it should be able to predict them diachronically. Modern systems must be thoroughly studied before analysis of fossil data, or subjectivity, invariably enters to bias the results. CHAPTER 2

THE ABIOTIC BACKGROUND

Geological Considerations

The central Maya lowlands lie at the southernmost end of the

Yucatan platform physiographic province (West 1964). The Peten bedrock

is essentially marine dolomitic limestone of early to mid-Tertiary age

to the north of Lake Peten, and Cretaceous carbonates to the south.

The Cretaceous and early Tertiary beds were later folded, and, to a lesser degree, faulted, during the middle Tertiary as part of the

Antillean orogeny.

Quaternary alluvium (mostly clays, with some flint, chert and rubble), derived from solution weathering of substrate, blankets much of the lowlands, providing agriculturalists with abundant but not highly fertile soil.

Classical karst topography is subdued due to the heavy vegetative cover, soil mantle, and perhaps the advanced development of landforms. Northern Peten, i.e., that region north of Lake Peten Itza, shows more karst-like development than the south. This zone is characterized by heavily weathered hills and clay-filled depressions. The

NE-SW trending bajo system of northeastern Peten is probably related to the fault system of identical orientation in southern Quintana Roo

and - northern Belice (West 1964). The bajos (or ak'alches), which will be described later in more detail, are shallow, seasonally inundated 4 5 depressions filled with clays derived from solution weathering of the

uplands and organic material from local vegetation. In general, their

altitudes lie from 50 m to about 150 m.

To the west of the faulted ridge and bajo system is a large,

little-known zone drained by the Rio San Pedro. The most prominent

geomorphic feature is a 100-m high escarpment forming a ragged boundary

between an extensive low bajo area and high forest. This feature may

best be seen from the isolated town of Paso Caballos, Peten. Much re-

connaissance needs to be done in this region to delimit its vegetation

and soil structure.

The lake region of Peten is a series of east-west interior

drainage basins, probably related to the Antillean orogeny. They range

from the many aguadas and lakes near the Rio San Pedro, through little- known Laguna Perdida and heavily-settled Lake Peten Itza, to Lakes

Yaxha and Sacnab, not far from the Belize border. To the south of Lake

Peten lie isolated basins, Lagunas Ija and Oquevix, showing the same

east-west trend and related to the major lacustrine axis. Immediately

south of Lake Peten is a heavily dissected escarpment rising from less

than 150 m at its base at the lake to over 250 m along its rugged crest.

Southern Peten is comprised of broad, poorly drained, clay

filled valleys and low, east-west trending hills of between 200 and

300 m elevation. Topography is gentler than that in northern Peten, and the soil mantle on the hills is thin. The Rio Sibun thrusts a finger of drainage into southern Peten from the west, but I suspect that subsurface drainage carries the main burden of water discharge. 6

Numerous aguadas, local water-filled depressions, dot broad, grassy,

clay plains. Most of the aguadas are dry during the spring dry season,

but a few are more or less permanent. I believe that the southern clay-

filled basins have origins similar to the bajo region north of the lake,

but local factors cause a different soil formation and resultant plant cover.

South of the Peten savannah region lies the heavily-forested

area drained by the Rio Pasion, which grades into the Guatemalan high-

lands. The southeast corner of the central Maya lowlands merges into

the Maya Mountains, a complex Tertiary igneous and metamorphic mass with elevations of 800-1000 m (3700 ft). This is the only region in

the Maya lowlands in which the underlying crystalline rocks, such as

granite, form extensive outcrops. The topography is extremely convo- luted, as is testified in Cortez' letters in which he states that the mountain trail was so difficult that two-thirds of the horses were lost and progress was less than 2 km per day (McBryde 1945). Coxcomb Peak, the highest point in the Maya lowlands, is over 1000 m altitude, enough to contain relictual highland plants such as Pinus oocarpa, Podocarpus,

Alnus and Liquidambar styraciflua (Standley et al. 1943) which are represented as minor components of the Peten pollen rain (Wiseman 1974).

Climatic Regime of the Maya Lowlands

Modern Climate

The Maya lowlands lie within the belt of easterly flowing air, the northeast trade winds. Cloud forests of the eastern scarp of the

Mesoamerican highlands receive their high precipitation via uplift of 7 the warm, moist airmass that causes adiabatic cooling and resultant moisture release. The lowlands, however, need an added convective component to release the trapped moisture. As the sun "moves north" during the spring, so does the intertropical convergence zone, the region of highest sea-surface temperatures. This belt of hot, ascend- ing air moves to approximately 12° north latitude during the summer, causing the moisture-laden trade winds to ascend for many hundreds of kilometers to the north of its calm center. The Peten, at 16° north latitude, begins to experience convective action and resultant precipitation increase during April and May, when mean monthly precipitation increases from ca. 50 mm during March to 175 mm for April and 240 mm in

May (Vivo Escoto 1964). The rainy season may last until late in the year when the oceans have lost their stored heat. As the intertropical convergence zone moves south, the Peten is left under the influence of the subtropical calms, masses of descending air, which render the trade winds adiabatically stable and generally incapable of significant precipitation.

A minor but important air mass involved in the regional climate is continental polar air from the Northern Hemisphere. From October until the beginning of the summer rains, these cold air masses produce frontal systems along the contact zone with the local tropical maritime air, bringing some rain and a drop in temperature. In the fall, tropical cyclones may also add significant amounts of precipitation to the decreasing monsoonal input.

The result of this interplay between air masses and solar inso- lation is a biseasonal regime consisting of a dry season dominated by 8

the subtropical calm, and a wet monsoonal season. Figure 1 summarizes

the rainfall regime of the Maya lowlands. The onset of the monsoons in

April and May is the main meteorological input into the human ecosystem.

As will be discussed later, the milpero plants in anticipation of the

monsoon, and if it fails to appear within a critical period, the plant-

ing is lost. Of course, the duration and intensity of the monsoon

affects the growth of the crop. March and April have the highest mean

temperatures and the lowest precipitation of the year, a xeric combi-

nation preventing establishment of classic rainforest (Lundell 1934).

Today's climate is well matched to the present agricultural

regime, but a change in either set of variables may necessitate re-

thinking of the role of climate in the subsistence base of the Maya.

Climatic History

In contrast to that of temperate zones, the climatic history of

the American tropics is relatively unknown. Since the immense in-

fluence of tropical air masses upon more northerly climate is beginning to be understood, any synthesis of global climate change cannot ignore the importance of the tropical regions.

Human influence on the natural vegetation in Latin America for over 4000 years complicates the use of pollen analysis for climatic reconstruction during the later Holocene (Wiseman 1974). Tree-ring analysis may fill in the later portions of this probable "climatic- anthropogenic" confusion zone. Since many arboreal tropical genera are seasonally deciduous (e.g., Ceiba), and many portions of the tropics have a distinctly biseasonal regime, the potential of 9

II I

P. Cd

co Cd Cd a)

G=4 'CS •n-1 4-4 •e-1

a) C) cf) X I 1— a)Ea 0 o

r—i

0 4-) cd

•r-I o 0:1) s.4

I 10

low-latitude dendrochronology cannot be ignored. A sample of Swietenia macrophylla collected in the savanna region of the Maya lowlands shows

a well-defined ring structure and sensitivity, but cross-dating has not

been established. This route must be explored further.

Past climate must therefore be deduced from paleolimnological

and glacial data collected from areas relatively remote from the Peten.

Pollen spectra from long cores taken at Gatun Lake, Panama Canal Zone,

mainly reflect sea level changes and some terrestrial vegetation (and therefore local climatic) fluctuations (Bartlett and Barghoorn 1973).

A sample dated 35,500 2500B.R is dominated by Phizophora pollen, in-

dicating an estuarine environment in what is now a fresh-water regime,

and interpreted as correlative with a northern interstadial. Para-

doxically, the local climate of this time was interpreted as being

ca. 2.5°C cooler based on the presence of Ericaceae pollen (ca. 2%) in

the sample. I believe that the last interpretation is unwarranted.

Long-distance transport, contamination, or genera occupying niches in

the past different from those of the present may account for its pres-

ence. Also, there is no replication of the Ericaceae peak by "close-

interval" resampling (Faegri and Iversen 1964). A period of non- deposition followed, indicating low sea level and, by extension, increased glacial activity to the north. This historical lacuna is

filled by a highland Costa Rican pollen profile (Martin 1963, 1964).

The Paramo, a tropical vegetation type somewhat similar to alpine tundra (Gomez-Pompa 1973) was found to be 800 m below the present tree- line during the late Wisconsian ca. 20,000 B.P. from which Martin

(1964) infers a 3.6°C drop in mean annual temperature. This is the 11

closest we have to a "full glacial" neotropical vegetation reconstruc-

tion. The terminal Wisconsian (11,000-9000 B.P.) data from Central

America show that rainforest was well established in the lowlands, but

that there may have been a depression of highland taxa 500-1000 m

below their present lower elevational limits (Bartlett and Barghoorn

1973). The very fact that deposition occurred in the Gatun Basin im-

plies higher base level and, by inference, glacial retreat to the north.

If we accept Bartlett and Barghoorn's vegetational data, it appears

that there was a lag in the Panamanian vegetation (high-elevation taxa)

response to the onset of non-glacial conditions (increased sea level).

The early Holocene of the neotropics was not significantly different

vegetationally from that of today. By 8080 t 170 B.P., the Paramo had

been succeeded by montane rainforest similar to that found today in

Costa Rica (Martin 1964).

From core data of Lake Eckixil, Peten, Vaughn (1976) infers an

early Holocene high (mesic) forest which later gave way to a drier monsoon forest during the mid-Holocene. This view is based upon data which are rendered equivocal by the modern pollen analog analyses pre-

sented in this paper. It seems to be an attempt to force the Peten

data into a global mode rather than structured comparison of modern vegetation-edaphic-climatic-pollen relationships. Lake Chichankanab

in Yucatan was apparently deep from 5500 to 1500 B.P. (Covich and

Stuiver 1974). However, Bartlett and Barghoorn (1973) infer a slightly more seasonal regime from 7300 to 4200 B.P.

There seems to be a general consensus that the fourth millennium B.P. was a time of increased precipitation. Cowgill and 12

Hutchinson (1966) reconstructed a rising lake level in Lake Petenxil before 2800 B.P., indicating higher rainfall, perhaps more than that of the present.

The paleoclimatic record is so disturbed during the last 2000 years that it is impossible at this time to make valid climatic reconstructions. It seems likely that there have been few climatic fluctuations capable of drastically affecting either plants or man in the

Peten. CHAPTER 3

MODERN VEGETATION ANALYSIS

Field Methods

Ideally, the relationship between vegetation and pollen entrap- ment will rest upon a foundation of well-described floras and plant communities. In the tropics such is not the case. There, palynology is pursued in a region of maximum ecological enrichment and minimal ecological information. The southern Yucatan Peninsula synoptic vegetation and flora study is 40 years old (Lundell 1934). Standley's

(Standley et al. 1943) monumental "Flora of Guatemala" is taxonomical, not ecological in outlook. The dearth of literature has given me the opportunity to integrate vegetation ecology and taxonomy with pollen analysis.

Essentially, the paleoecological cognitive process is visualized as a solution to the equations V -n = Pm (1) m and = V (2) f where V = vegetation, P = pollen, m = modern, f = fossil, and n = transfer function. The first equation is the analog. If a qualitative or quantitative relation can be derived between a pollen assemblage and its parent vegetation, then it may be used (with discretion!) to reconstruct prehistoric communities from fossil pollen spectra. However, relations

13 11+ which exist today may not have functioned in the past. The paleo-

ecologist must know his analogs (modern ecology), so that incongruous

reconstructions are not intuitively or statistically derived.

The analog approach used in this study began with the null

hypothesis: 0 = the Peten ecosystem is homogeneous through space and

time. This hypothesis was tested synchronically by a vegetation sample

design which emphasized diversity of communities. Numerical analysis

of sample sites derived an objective gradient to serve as a benchmark

for reconstruction of fossil plant communities.

The data set used in vegetation analyses consists of approxi-

mately 30 stands visited by the author in 1973 and 1975 (Fig. 2, Table

1). Due to the limited time available and the proposed use of non-

parametric statistics in entiation (characterization of the vegetation

type from stand data; Mueller-Dombois and Ellenberg 1974), the sample

stands were chosen subjectively, without bias (Mueller-Dombois and

Ellenberg 1974; Poore 1962). Although not initially recognized as

such, stand sampling was along an environmental gradient that was objectively determined in later statistical analysis.

The steps in sampling consisted of: (1) floral familiarization;

(2) reconnaissance; (3) entiation; and (4) qualitative and quantitative sampling. Floral familiarization was essentially a reverse of standard botanical procedure (Benson 1959), in that published floras and herbarium sheets were consulted prior to departure to familiarize myself with the vegetative character and ecological niche of important taxa in the Peten, a procedure which I heartily recommend to ecologists venturing into new ecosystems. Reconnaissance was essentially 15

SITE LOCATION MAP

o VEGETATION TRANSECT (VT)

SPECIES LIST (SL) LAKE

0 5 10 lacodoeil KM

-17°00'

Figure 2. Site location map. 16

Table 1. Site data, 22 LANDSAT-1 ground-truth sites.

No. Site Name Data Date

SL-1 Tikal Species abundance

SL-2 Macanche Species abundance 24 Feb 73

VT-1 Petenxil Vegetation transect 26 Feb 73; 5 Mar 75 VT-2 Finca Eckixil Vegetation transect 27 Feb 73; 5 Mar 75 SL-4 Eckixil isle Species abundance 9 Mar 75 SL-3 Petenxil-Eckixil Species abundance 9 Mar 75

VT-3 Santa Marta-N Vegetation transect 25 Feb 73; 3 Mar 75

VT-3 Santa Marta-S Vegetation transect 25 Feb 73; 3 Mar 75 VT-5 Santa Elena Vegetation transect 22 Feb 73 SL-5 Paxcaman Species abundance 23 Feb 73 SL-6 Porucita Species abundance 4 Mar 75

SL-7 La Libertad Species abundance 10 Mar 75

VT-6 El Guanal Vegetation transect 10 Mar 75 VT-8 Bosque el Caobal Vegetation transect I Mar 75

SL-8 Santa Rita Species abundance 10 Mar 75

SL-13 Sibun Species abundance 5 Mar 75

SL-14 Sibun Species abundance 5 Mar 75

SL-9 Sajalal Species abundance 2 Mar 75 VT-7 Santa Ana Vieja Vegetation transect 2 Mar 75

SL-10 Excocon Species abundance 4 Mar 75 SL-11 Excocon Species abundance 4 Mar 75 VT-9 Pacay Vegetation transect 2 Mar 75

SL-12 Oquevix Specie's abundance 9 Mar 75 17

familiarization with structure and composition of various stands in the lake region of Peten. At this point, it became clear that sampling of mesic forest subassociations was impractical due to the number of

sterile and unknown plants encountered in this vegetation type. Pro-

spective sample sites were selected which met the following require-

ments: (1) a site must include most species which belong to the com-

munity it represents; (2) the habitat and cover must be fairly

homogeneous; and (3) the zones traversed must be under no great stress

from human activity.

Those stands which met the criteria were sampled both qualita-

tively and quantitatively. A species-abundance list was taken at each

site, using the Daubenmire cover scale (Daubenmire 1959, 1968). At

nine sites, this estimate was checked quantitatively. No attempt was

made to assess the total flora of each site, but a constant set of 85

known taxa was used as indicators to entiate the communities.

Quantitative sampling took three distinct forms. In predomi-

nantly herbaceous communities, i.e., grassland and savanna, a cross

oriented to the cardinal directions, 10 Warm was placed at random in

the site. A 2 dm x 10 dm plotframe (Daubenmire 1968) was placed at

1 m intervals along the arms, yielding 40 estimates of herb cover,

later averaged into a site mean. Shrub and tree cover were estimated

by the line intercept method (Cain and Castro 1959). One-hundred-meter

tapes were placed at random as before and the distance along the tape

subtended by each plant entered into the logbook, along with the num-

bers of individuals of each taxon intercepted by the tape. 18

One-tenth hectare plots (Cain and Castro 1959) were used to

derive quantitative data on tree cover, abundance and size. By this 2 method, successive pairs of 50 m subplots were arrayed along a tape

oriented to a cardinal direction until the requisite 1000 m2 of plot

was reached. Within each subplot, all trees over 2.5 cm DBH were de-

termined taxonomically if possible, their bole diameters measured with

a tree tape, and their heights and crown diameters estimated. All

plants which had their growing stalk or trunk in the plot were counted.

In all cases, each stand was evaluated in terms of potential

pollen productivity and dispersal. Bole density and understory height

of canopy were considered in the field as important aspects to be tabulated.

Numerical Classification of Sample Sites

To derive an objective gradient, numerical seriation methods

were used. Sites with the requisite number of species were chosen for

similarity index analysis. The similarity index used is SSrensen's (Mueller -Dombois and Ellenberg 1974):

2C I = — x 100, s A+B (3 ) where A = total number of known taxa in site A, B = total number of known taxa in site B, and C = number of taxa common to A and B. I chose Srensen's index over Jaccard's since it is more valid statistically, in that it measures coinciding taxon occurrences against all the theoretically possible ones (Mueller-Dombois and Ellenberg 1974). I did not use the Braun-Blanquet "cover abundance coefficient of community similarity" (Mueller-Dombois and Ellenberg 1974) due to the necessarily 19 crude nature of my cover estimates. Simple presence-absence data analysis, while containing less information, is statistically more valid. Table 2 presents a matrix of similarity coefficients for 22 sample stands which contain a requisite number of taxa. I selected three stands which showed maximum dissimilarity among them in the matrix (sites SL-1, VT-3, and VT-9), and expressed all other sites as deviations from these three sites (Fig. 3). Since VT-3 can be considered as the "middle" of the proposed gradient, I transformed its unidirectional gradient into a bidirectional one: one arm toward the

forest, and one arm toward grassland. Agreement among the three gradi-

ents is strong, indicating an objective ordination of sites by their

component taxa. Table 3 illustrates the results of this ordination,

which proceeds from swamp forest to grassland, exhibiting a "quasi-

objective" grouping of communities arrayed along the gradient. Each

community will be discussed in more detail in the following section.

Vegetation Mapping

A vegetation map of the lake region was constructed by con-

sidering the stands as ground-truth sites for LANDSAT-1 remote sensing

imagery. Using a printed copy of LANDSAT multispectral scanner image

No. the sample stands were located by reference to 8157115 5005A000 ' roads, lakes and other high-visibility markers. The reflectance of the

vegetation in bands 5 and 6 was determined in the following way.

A crude but effective densitometer was used to analyze the

image density at each ground-truth site. A standard photomicroscope

with attached Minolta camera served as the densitometer. Seventy

▪

\ I-rn H irn 4.n 00 0 0 0 a\ 0 4-1 0 H cc CY\ c0 0 0 Cs- CT\ -1- -1- cc _1- 00 n O N- 0 c0 \0 N ONHN t•rn H -1- -1- N N 111 trn N -1- Nrn N

N Lf1 Ir\ K1-1- N- -1- rcn co L.N 111111 K1 00 U\ '.0 0 on N 0 0 0 0 0 0 H f--1000 r.rn H 0 re\ Hr-1 111111

n0 Vs\ r11 N- CO 0 CO H 00 0\ N N N N- 0 -7 00 H \SD n.0 00 N 141 K1 re\ -1- N rr\ -1- rr\ r•r\ r4-n r4-n

HOU1CT nDa\NNU1H N1H1110 0 0 L11 LI1 H N N Lrn L.Cn 0 0 N- -1- N re\ N N nC r*.(\ r+1 r+-\ irn Lrl

c0 -1- t+1 00 N-Nr-1 n0NHK\OCO n.0 as 0 -1- n.0 N- 1:7\ reN a\ 0\ 0\ v0 ON 0 N • N H re1 N N N re1

is-nn0 Ern n.0 %.0 01 CO CN N H N r4-n c0 H c0 H N H N CO K\ LC\ 0 K1 NNNN L11-4- 141 -4- reN t.(\ r41-1-

0 rel reN H Cs- 0 141 n.0 CO 0\ 0 ifn N N 00 r-I111 n0 01H N-1' 0 re1 N1-1- ren irn _d- Lf1 1.(1 X.) U)

4-) N- Q' II\ Lf1-1- 0 -4- N- \.0 re1 0 ON rI N- n.0 H K1 IS1 H ONHNHI--4L-1-tNrelN-1- .0 4-3 Cs-- re\ r1 0 C\ Lr\ r-en 0\ reN co H 00 K1 rel O -I' N 0 0 L.rn n.0 _1- IA L.1-n re1 . • . -P O k.0 n.0 c0 1 csn rr N r-I -1- 0N-N\-0OHNOR re1 Lf1C0 N \.0 00 H re\ \ 0 LI1 EN- N .-4" V.\ --t N II II II II 0 0 154N u) tO N c0 CO -4- NO LA N N n.0 n.0 re1 0 0 01 E-I - 4- 4- c.c-N ren Lrl n.0 N

0 \-0 0 NHUQ' N N- U1\0 00 te\ 0 H C\J N U"\ 1.4 N- 00 N - O \.0 U1 0 N0 '.0\..0 0 HQ' 0 N 4- O r41 c0 o$.1 n.0 o 1'4 -1- 0 c0 ccc Lr1 01 0 Ill N Lr1 te\ N F-1 • C`-- Q' t41-4- cd U\0 U1 U1 Cs- \.0 LC1 N.- Cs- N

a) o N- 0 Lf N- O CO 0 N

-P •r-I

H 4-4

c r) re1 te1

te1 OH N • N H N re1 111 ti1 \.0 N \.0 CO 00 Cs- H H 01 H It II It II 11111111111111111111111 1-4 E-1 1-) E-1 E-4 I-4 1-1 )-4 1.4 E-1 n-) I--4 > En En En En En erl ti) Cr) CI) 21

puyissv

4saaoj 4 - 7S 'eTTa4san- 4seaoj 9-7S

4saaoj Lap L - 1S qmaos qnsog-vuuvAs cinaLis >rims 13T - 7S 4sea0j epyrdv V- IA ones euquosaSs 5-LA vdTTw, Nyo P00r t - IA 4saao; ypeniv puv•sst 6-LA 4saaoj xso - 1/1 ;saao; slim -as zsaao; )ivo T - 7S vdrau .uopueq 4 - IA spulirmsnaq +7-'1S

cinaos xuaq T - '1S

4S0.10j BpSnOlf II - 7S 4SZJOj )1E0 4-ZA - ;saaoj 3.1e C - LA ;saao; T-IS

vITaZvan 9- IA sdrp 4saa0j -7S 4saaoj apTinSy 4saaoj T.TT • - IA saaoj ti2T4 4-7S vurposaSia (pa&

durems vaTIqp-ed

.aoj epyriy UT - 7S ;saaoj auntloo 4-7S aoj La 9-7S vdrpli •uopuycly 4 - IA

• AVS VUTUOSJSE 4 - 1A ;seaoj Sap 1 - '1s dumms vaTliovd - A qn.rqs 3fszvo FT-1S sdiTw-4saao; 2-7S ;saaoj aurapo -7S

cinaos-VUUVAES +7T - 7S ;saaoj Lap L-7S q.seaoj 9-'IS .aoi ypyn2 g-1A viTa;van

ydrp ;saaoj 2-7S

4saa0s oTsaA ;saaoi otsaN

4saaoj xvo T

puvipoom sTe +T- 111

› 22

Table 3. Classification of sample stands derived from similarity indices.

Site No. Site Name Vegetation Type

VT-2 Finca Eckixil Pachira aquatica Terminalia swamp SL-1 Tikal VT-1 Petenxil Mesic forest SL-3 Petenxil-Eckixil SL-6 Porucita VT-5A Santa Elena Monsoon forest SL-7 La Libertad SL-2 Macanche — Corozo forest SL-5 Paxcaman VT-8 Bosque el Caobal Aguada forest SL-11 Excocon • VT-4 Santa Mar ta-S VT-3 Santa Mar ta-N Encinal

SL-10 Excocon .1.1111n1 SL-13 Sibun ---1-_ Thorn scrub SL-14 Sibun VT-6 El Guanal Sahalal VT-7 Santa Ana Vie ta Nanzal VT-9 Pecay Grassland

SL-4 (Eckixil Isle), SL-12 (Oquevix) and VT-5C, D, E do not fit this gradient. 23 millimeter positive prints of the image bands 5 and 6 were carefully

dusted using a lens brush and placed on the microscope from which the

mechanical stage had been removed. The ground-truth site was located

under low power and then the 20X objective was swung into place. This

covered a very small portion of the transparency, the pixel of the

ground-truth site and its immediate environs. With the microscope lamp

turned on high power, the camera light meter reading was recorded.

When all ground-truth sites had been analyzed in this manner, random

sites were chosen and classed according to similarity to various

ground-truth site light-density standards.

A problem arose in the course of analysis in that the emulsion

became scratched, passing more light than it should have. This error

was minimized by simply avoiding the scratches. Once I became familiar

with the values of the different vegetation types, a map was drawn on

acetate covering the Peter photo print.

Grassland and bajo could be quickly discriminated, as could dry

and moist forest, but oak woodland, milpa, and sections of the dry

forest were indistinguishable. The resulting map is presented as

Figure 4.

Comparison of Habitat Variability -- Mayan Lowlands vs. the Sonoran Region

The similarity index matrix analysis and the resultant gradient

describe the variability of habitats within the Peten. As a conclusion

to this chapter, I will illustrate the magnitude of this variability, by a simple comparison with another biome, that of the Sonoran Desert

region of northwestern Mexico, which has documented environmental 21+

• -.1- 0

••-i rz• 25

variability (Shreve and Wiggins 1964) similar to that of the Tehuacgn

Valley of Mexico (MacNeish 1964) or the Shoshonean Region of the Great Basin (Thomas 1973).

Rathje (1971) bases his trade models, in part, on the concept

of environmental homogeneity of the Maya Lowlands. Homogeneity is an

implicit assumption that has long been held by Mayanists. This concept

is valid in the realm of economic geology. The Cretaceous and early

Tertiary limestone of the Peten indeed lacks hard stone for grinding

implements and obsidian for cutting tools. The "resource redundant model" of Rathje implies little variability of botanical resources or habitats. However, the results presented above indicate that variability does exist, that one does not find the same flora covering the entire lowlands.

MacNeish, on the other hand, bases his developmental models on highland habitat variability, requiring scheduling of activity to harvest the resources of the Tehuacgn Valley (MacNeish 1964). The dichotomy of an ecologically variable highlands and ecologically homogeneous lowlands has therefore become essentially a paradigm for modern Meso- american "ecological archaeology." While the highland variability model is valid, the lowland "resource redundance" model is untested.

Since there are no data amenable to comparison from highland Central

Mexico, a data set from the tropical portions of the Sonoran Desert, the adjacent Short Tree Forest, and Thorn Forest was used. These data are in the form of species lists taken by me during field research from

1973 to 1977 (unpublished manuscripts, Laboratory of Paleoenvironmental

Studies, Tucson). Twenty-two sites with representative species lists 26

were taken from the independent data set in such a way as to maximize

variability. Differing geographic zones, substrate types, elevations,

and slopes were used. Maximization of variability of the Sonoran data

set increases the validity of the comparison, since vegetation is simi-

lar in Sonora within small geographic and edaphic limits. If the

Sonoran sites are specifically chosen to be diverse, and if the Peten

data set is shown to be as diverse, then the Peten sites will be shown to be ecologically heterogeneous.

The similarity index matrices for the two regions were compared

by: (1) testing for independence and normality of the 242 points with-

in each matrix; and (2) then calculating a mean similarity index for

each which, in essence, gives an average amount of plant taxa shared

by sites in the region (Mueller-Dombois and Ellenberg 1974). The mean

value for the Peten matrix is 36.7 ± 18.1, and for Sonora it is 32.0 ±

24.6, which indicates that there is no statistically significant dif-

ference between the two matrices.

The implication of this test is that the Peten vegetation/ pollen collection sites are as diverse as those in Sonora. If we assume that Sonoran vegetation types are as diverse as the Great Basin

(Thomas 1973) or the Tehuacan Valley (MacNeish 1964), then the "habitat variability" dichotomy between highlands and lowlands does not hold.

The lowland resource redundant model must be rejected on an ecological basis, at least for the Peten, and possibly for other portions of Mesoamerica. Acceptance of this may be difficult for

Mesoamericanists but will eventually result in new models of forest 27 exploitation by Preclassic lowlanders. For the remainder of the discussion to follow, habitat variability will now be assumed. CHAPTER 4

MODERN POLLEN ANALYSIS

Introduction

In the previous chapter, I described the spatial heterogeneity

of the modern Peten vegetation through the use of simple statistical

analyses of species lists and abundance tables. This was essentially

the modern vegetation analog or "Vm" in formula (1), page 13. Since the final aim of this research is to derive "V f" (past vegetation), a transfer function is necessary. This will be provided by pollen analy-

sis. In this chapter, I discuss the methods of collecting and chemi- cally treating modern soil and mud samples, and arraying the data in a form that can be used as the first part of the transfer function.

If the entiation of vegetation types as performed in Chapter 3 is a valid reflection of reality, then changes among types must be a result of changes in their component taxa. Since many of these taxa produce distinctive and preservable pollen types, it should be theoretically possible to entiate communities by their pollen productivity.

With the objective gradient already defined in Chapter 3, it is a simple matter to array pollen sites along it. Several pollen collection methods were used in an attempt to derive a feeling for the diversity of tropical pollen.

28 29

Pollen Collection and Extraction

The modern pollection collection strategy is twofold -- a regional and local network -- that provides modern analogs for the fossil core data.

The sample sites were chosen (for those I collected) to represent a "typical" expression of a given vegetation type, i.e., one that is fairly homogeneous in structure, cover and species composition.

Except for riparian and milpa samples, each site was located away from obvious ecotones. The represented vegetation type is usually as described in the appropriate local flora. Biases in site selection method are alleviated somewhat by reference to a published work, and thus are held more constant for other researchers who may study the same formation.

Soil pollen samples were collected in the designated zone by either single scoop, "random pinch" (Martin 1963), or transect methods

(Wiseman 1974). The simplest collection method that I used was to take a single sample with a clean stainless-steel spoon or other utensil, then place it in a newly opened Whirlpak. The Whirlpak is the best sampling container for storage of soil samples since it is only open when putting the sample in it, and is sterilized and factory-sealed before and resealed after collection. The random-pinch sampling method is simply to collect about 20 pinches of soil surface and place them 2 in the sample container. The collection area was under 500 m . The more complex transect method was used in conjunction with vegetation analyses (Wiseman 1974). A line transect was placed on the ground at random, along which ten independent soil samples were taken. 30

Quantitative coverage, diversity, and density estimates were also recorded. This method resulted in a one-to-one relationship between pollen and vegetation.

After the samples were taken, two precautions were observed.

Since destruction by moisture and soil organisms render paper labels illegible in time, Dymo plastic labels were used in the field. The slight inconvenience was outweighed by the clarity of the resulting labels. 10 A drop of phenol (C 6H 0) was added to each soil sample upon collection and the sample was agitated to spread the toxin which killed soil dwelling organisms that degrade pollen. The advantages of this method are illustrated by an experiment performed upon a sample collected in a tropical forest environment near Lake Petenxil, Peten, on

February 27, 1973. Upon collection, one part was treated with phenol, and one part was left untouched. Subsamples of the untouched section were taken at specified intervals after arrival in the U.S. and "killed" to determine pollen degradation rates. Table 4 and Figure 5 summarize the results of this experiment. Absolute pollen frequencies (APP) were determined for the control and subsamples using the exotic pollen addition method. The results indicate that untreated tropical soil samples may have lost much of their pollen before extraction. Heating the sample to 180 0 in an oven upon arrival at the laboratory saved some pollen, but not enough to produce a countable sample. Indications from these data are that the pollen found in moist tropical soils may be recent. However, paucity of pollen in tropical soils does not necessarily mean that such data are useless. A comparison between soil 31

Table 4 • Post-collection pollen degradation in tropical soil samples.

Grains/gram Sample No. Type Treatment Date Count 1 Count 2

VT-5a-1 Soil Phenol 3/27/72 1240 1320 1280 VT-5a-2 Soil 3/27/73 890 1210 1050 VT-5a-3 Soil 3/29/73 1320 1220 1270 VT-5a-4 Soil 4/2/73 670 890 780 VT-5a-5 Soil 4/3/73 720 420 570 VT-5a-6 Soil 4/8/73 570 270 420 VT-5a-7 Soil 4/25/73 120 330 225 VT-5a-8 Soil 5/20/73 20 130 75 VT-5a-9 Soil 6/12/73 5 lo 8 VT-5a-10 Soil 8/15/73 0 10 5 VT-5a-11 Soil 3/21/74 90 30 55 VT-5a-12 Soil 4/2/74 20 0 10 VT-5a-13 Soil 5/5/74 10 10 10 VT-5a-14 Soil 4/6/76 0 10 5 32

VT-la-1 1000 Tropical FrirPst nnii

• VT-la-

' VT-la -3 500 • VT-1 a-4

• VT -1a -5

100

VT-la-6

VT4a-7

1.0 1 2 3 5 Months from Collection

Figure 5. Pollen degradation rate. 33 samples and lake sediment samples (to be discussed below) indicates

little significant bias in soil pollen vis-â-vis lacustrine samples.

Either soil-dwelling organisms degrade pollen in a pattern identical

to lacustrine organisms, or degradation is not differential. There-

fore, I feel that soil data are a valuable tool for analog-building,

if used in conjunction with independent lacustrine "checks."

Phenol should be handled with care in the field due to its

extreme toxicity and absorptivity by the human skin. I carried it in an unbreakable Nalgene bottle (with dropper) inside a wide-mouth Nal-

gene bottle with polyurethane foam between them, making a fairly safe, portable "killing bottle."

Filled sample bags were washed with water to rinse away dust adhering to the outside, dried, and packed in groups of five in another

Whirlpak shipped to the Department of Geosciences, University of Arizona.

The local network mud-water interface samples were collected from a Cedrela mexicana dugout canoe and a small fiberglass dinghy with a Davis-type and a modified piston corer. The samples consisted of the top two cm of lake sediment in consolidated sediments (Davis 1974), or more if the sediment was too "soupy" to adhere to the core tube walls with only 2 cm of bearing surface. This interval approximates, by dry weight, a single cm of sediment at depth. The samples were placed in labeled Whirlpaks and treated in a manner similar to the soil samples.

An attempt was made to sample pollen from tank- and poster- type epiphytic bromeliads, ideal modern pollen receptacles. The tank-type bromeliads store water in a "cup" formed by interlocked

linear leaves (Padilla 1973). This cup would catch and store pollen.

Sludge from the bottom of an Aechmaea bracteata in the encinal

near Lake Petenxil was collected and examined for pollen but, unfor-

tunately, less than 30 grains were found in the sample. Possibly the

abundant microfauna living in the cup digested the pollen, or collec-

tion of the sample resulted in losing much of the sludge. The polster-

type Tillandsia show more promise, since peltate scales on their leaves

catch and hold passing pollen in a manner similar to a sieve. When

vigorously agitated in distilled water, Tillandsia scheideana yielded

over 300 pollen grains. Dissolving the entire plant during pollen

extraction doubtlessly would result in recovering more pollen.

Upon arrival in the laboratory, each sample was assigned an

alphanumeric code. Throughout extraction and counting, the code was

used, preventing me from knowing the origin of the sample being counted.

This method eliminated unconscious bias toward the expected values.

For example, if I expected a large amount of pollen type X, any uni-

dentified or heavily-corroded pollen grain might be assigned to that

taxon. After counting the sample, the original site data were re-

stored.

Removal of the matrix from the pollen sample was accomplished

by dissolving successive fractions in a series of acids and caustic

sodas (Faegri and Iversen 1964). Table 5 gives a summary of the chemical treatment employed. Exotic tracer pollen from the Australian tree,

Eucalyptus was added in the form of pills. Lyconodium pills were also added as a check on the estimates derived from Eucalyptus. The 35

Table 5. Pollen extraction method.

I. Preparation of dry sample

A. Unbag sample carefully in still air B. Screen through 4o mesh screen (if soil sample) C. Weigh fraction; record D. Record color of samples from soil color chart E. Place fraction into labeled phial for storage

II. Chemical extraction

A. Decant sample and tracer tablet(s) into 250 ml beaker; wet down with distilled water B. Add concentrated HC1 slowly until fizzing stops; note reaction C. Decant beaker into centrifuge tube D. Centrifuge and decant E. Add concentrated HC1; stir; centriguge; decant F. Water wash G. Add 50% HF; let stand 25 hours H. Centrifuge; decant I. Add 70% HF; let stand 24 hours J. Hot H20 bath; 20-30 minutes K. Two hot water washes L. Add 3/4 tube HNO3; let stand 10 minutes M. Hot water wash N. Boil in HC1 1-2 minutes O. Two boiling H20 washes P. Boil in 5% KOH 3 minutes Q. Hot H20 wash until decants clear

III. Sample storage

A. Wash centrifuge tube into labeled vial B. Centrifuge; decant H20 C. Add glycerol and one drop phenol

IV. Slide preparation

A. Pipette subsample onto slide; add one drop stain; cover B. Label slide; place in storage; ready to count 36 presence of epiphytic Lyconodium linifolium in the local flora renders interpretation of these results slightly misleading. However, the local Lycopodium pollen is infrequent, probably little influencing estimates derived from the addition of large amounts of tracer pollen.

After extraction, pollen samples were placed in glycerin and heated for two days to expel any residual moisture. Phenol was read- ministered to the sample to prevent fungal degradation of the pollen distillate, and basic fuchsin was added as a stain. The samples were then stored in 10 cc phials until mounted on a slide for counting.

Pollen slide counting was done with a Zeiss binocular microscope using a 40X objective for routine counting, and the 100 x oil immersion objective for determination of sculptural characters of unknown pollen types.

The reference collection of the Laboratory of Paleoenviron- mental Studies of The University of Arizona was used to aid in identification, as well as several published taxonomic keys (Tsukada 1964;SalFado-

Labouriau 1973). Depending upon ease of finding and recognizing pollen, counts of from 200 to 1500+ pollen grains were made per sample. The soil samples took from three to five times longer to count than the lake samples due to bad preservation, low absolute numbers, and excessive matrix on the slide. Slide-scan technique was designed to reduce bias in counting from differential movement of pollen grains under the coverslip. Both the center and the edges of the cover slip were scanned in approximately equal amounts. The running tally of counted grains was kept with a Veeder Vary Tally, speeding this aspect of 37 pollen analysis. Minor types were tallied on paper, and unknown types were photographed with a photomicroscope. The resulting counts were arrayed as both relative (RPF) and absolute pollen frequencies (ARP) for use in later analyses. CHAPTER 5

RESULTS OF THE VEGETATION AND POLLEN ANALYSES

Due to the complex nature and distribution of the Peten vegetation, it was impossible to find lake sampling sites that would characterize a given vegetation type. I therefore used soil samples collected in the manner previously described. With all the biases inherent in soil samples, they nonetheless can be collected within a specific vegetation type, making a stronger plant community-pollen assemblage bond.

As described in Chapter 3, I derived a gradient of vegetation types through analysis of a similarity index matrix. This gradient will function as the modern vegetation analog. However, transforming a descriptive analysis into a link in a deductive chain necessitates a critical review of the assumptions made in the method of transforma- tion. The first assumption is that the sample array includes all important modern vegetation types. A sample size of 22 sites will not cover all possible formations throughout the Peten. There are many minor types mentioned by Lundell (1934) and Bartlett (1935) which are not covered by my analysis. I doubt that they would lie outside the derived gradient, except possibly for bajo vegetation (tintai, escobal) and pinelands (pinar). Due to logistic problems, these vegetation types were not sampled adequately.

38 39

I believe, however, that the formations included in the present

analysis (except for the three types mentioned) comprise the important

types to be found in the Peten, and constitute a sample large enough

for statistical analysis. More research will fill in the voids in the gradient.

The next section follows the derived gradient from Pachira

swamp through high forest, to savannah and grassland. Pollen counts

are presented and a summary of the results is given here. The Peten

soil sample network results are followed by the results of the mud-

water interface sample network which represents the local modern pollen

analogue for the fossil pollen core samples.

Vegetation Types

Type 1: Pachira aquatica-Terminalia Swamp

Where the Peten forest abuts a water's edge, a distinctive vegetation type, dominated by Pachira aquatica, can be found. Arising

from slim, buttressed roots, Pachira forms an open canopy (87% cover) of palmately compound leaves, 6 m high at lakeside, rising to over 16 m high in the swamp's interior. The subdominant Terminalia (or

Buchenavia-Combretaceae) occupies a habitat slightly less moist than

Pachira. Its contorted branches, with simple leaves at growing tips, seem to seek increased light at the water's edge, covering but 23% of a transect taken in the swamp. Terminalia is a conspicuous lacustrine and riverine bank dweller in the tropics from the Peten to Argentina

(V. Markgraf, personal communication 1976). Various species of figs

(Ficus spp.) are locally common in the lacustrine swamp, and although 40 unintercepted in the plot, seem to assume a 75% coverage in certain portions of the Pachira-Terminalia swamp.

The sparse, herbaceous understory includes young Pachira, ferns,

Pistia stratioides, isolated clumps of Typha dominguensis, and

Hymenocallis cf. americana, found flowering in March 1975. The epiphytic component is large, composed of an assortment of scaly lichens, many epiphytic ferns, Tillandsia of various types (T. scheideana, T. festucoides, T. balbisiana, T. fasciculata), Catopsis cf. abides, an epiphytic triangular Cereus, and an epiphytic Araceae.

As tree cover decreases toward the lake, Typha dominguensis and the sedge, Cladium jamaicense, assume dominance, eventually giving way to Nymphaea ampla in scattered colonies as water depth increases.

Toward the uplands, however, Pachira rapidly is replaced by taxa characteristic of mesic high forest.

Species characterizing the Pachira swamp and its ecotones are given in Table 6.

The major affinity of the Pachira swamp is to the mesic forest

(mean similarity index: 65.2), and may be considered as an edaphic variant of the generalized, highly diverse Peten forest.

Five soil and mud samples were collected during the dry seasons of 1973 and 1975 by single scoop, and with a Davis corer. Two samples had 200-grain counts and three samples had 500-grain counts. Pollen preservation was good, but counting was hampered by excessive matrix residues and organic(?) colloid that could not be completely eliminated by re-treatment. The mean pollen concentration for three of the five sub-samples was ca. 12,000 grains per gram. Table 6. Pollen and vegetation data, Site VT-2, Finca Eckixil, Peten.

Midpoint Cover Pollen Percent (N=250 grains) Taxon Class 1 2 3 If 5 X

Acacia spadicigera 2.5 0 0 1.0 0 0.5 0.3 Acanthaceae 0.5 0 0 0 0 0 0.0 Aechmea bracteata 0.5 0 0 0 0 0 0.0 Alnus sp. 0 0.5 0 0 0.5 1.0 0.4 Bucida buceras 2.5 0 0 0 0 0 0.0 Bursera simaruba 1.5 2.0 3.5 6.5 1.0 0 .5 2.7 Caryophyllaceae 0 2.0 4.o o 2.5 o 1.7 Catopsis abides 0.5 0 0 0 0 0 0.0 Cecropia peltata 2.5 0 0 1.0 2.5 0 0.7 Celtis spp. 0 0 2.0 0 0 4.0 1.2 Cereus sp. 0.5 0 0 0 1.0 0 0.2 Chaetoptelea mexicana 0 0 0.5 1.5 0 0 0.4 Cheno-am 2.5 2.0 0.0 1.5 0 2.5 1.2 Chrysophyllum sp. 2.5 0 0 0 0 0 0.0 Coccoloba belizensis 15.0 0 0 1.0 0 0.5 0.3 Compositae, high-spine 2.5 5.0 0 2.5 0 10.0 3.5 Compositae, low-spine 0.5 10.0 5. 0 7.5 7.0 12.5 8.4 Conostegia sp. 2.5 0 0 0 0 0 0.0 Cyperaceae 15.0 10.0 17.0 12.5 13.5 15.5 13.7 Eugenia sp. 2.5 2.0 2.0 5.0 0.5 1.0 2.1 Ficus sp. 2.5 0 0 0 0 0 0.0 Graminae 2.5 12.0 10.0 6.0 14.0 4.5 9.3 Guazuma ulmifolia 2.5 0 0 0 o o 0.0 Hymenocallis americana 2.3 0 0 1.0 0 1.0 0 .4 Inga sp. 2.5 0 0 0 0 0 0.0 Ipomoea sp. 0.5 0 0 0 0.5 0 0.1 Liquidambar styraciflua 0 0 0.5 0 1.5 0.5 0.5 Lysiloma bahamanenis 2.5 0 0 0 0 0 0.0 Malpighiaceae 0 5.0 0 2.0 1.5 4.5 2.7 Manilkara achras 0 0 2.0 0 1.5 0 0.7 Monstera spp. 2.5 0 0 0 o o 0.0 Moraceae 2.5 10.0 12.0 5.0 7.5 12.5 9.4 Myrtaceae 0 2.0 0 1.5 0 0 0.7 Nymphaea aquatila 2.5 1.0 4.0 2.5 0 3.5 2.2 Pachina aquatica 62.5 0 0 2.5 0 3.0 1.1 Pimenta officinalis 2.5 0 0 0 0 0 0.0 Phoradendron sp. 0.5 0 0 0 0 0 0.0 Pinus spp. 0 3.5 10.0 12.0 14.5 6.0 9.2 Podocarpus spp. 0 1.0 0 0 1.0 0 0.4 Pseudobombax ellipticum 0.5 0 1.0 2.5 1 .5 0 1.0 Quercus oleoides 0 7.5 1.0 2.5 6.5 8. 0 5.1 Sabal morrisiana 0.5 0 0 0 0 0 0.0 Sapotaceae 0 4.0 1.0 1.0 6.0 0.0 2.4 Simaruba glauca 2.5 0 0 0 0 0 0.0 42 Table 6--Continued.

Midpoint Cover Pollen Percent (N=250 grains) Taxon Class 1 2 3 4 5 X

Smilax sp. 0.5 o o o 0 0 0.0 Solanaceae 2.5 2.5 1.0 1.0 3.5 2.5 2.1 Spondias sp. 2.5 0 0 o 0 0 0.0 Terminalia-buchenavia 15.0 7.0 17.0 12.0 6.5 2.0 8.9 Tillandsia balbisiana 0.5 0 0 0 0 0 0.0 Tillandsia fasciculata 0.5 0 0 0 o 0 0.0 Tillandsia festucoides 0.5 o 0 0 0 0 0.0 Typha dominguensis 15.0 6.0 5.0 2.0 10.0 8.5 6.3 Vismia ferruginea 0.5 0 o o 0 0 0.0 Vitis sp. 0 1.0 0 0 2.0 0 0.0 Zanthoxylum spp. 0 1.0 1.5 0.5 0 0.5 0.7 Zea Mays 0.5 1.5 0 0 1.0 0.5 0.6 143

The two dominant pollen types found at Site VT-2 were

Terminalia-type, varying between 7 and 17 percent, and Moraceae, varying between 10 and 12 percent, for the five samples. The dominance of

Terminalia-type is not surprising since Terminalia is a common canopy tree in the Pachira swamp (62.5 percent coverage). From analysis of reference material and published sources(TsulEcaida1964; Salgado-Labouriau

1973), I feel confident that the genus Terminalia produces pollen falling within this type. However, pollen taxonomy of lowland Central

America is still little known, and it is wise to keep an open mind when interpreting such equivocal data. Moraceae pollen, a thin-walled, diporate type, is an established pollen taxon. Ficus, the dominant member of the mulberry family in the Pachira swamp, does not release its pollen into the atmosphere. The inflorescences of Ficus line the interior of a specialized, fleshy branch end, and the pollen cannot escape the receptacle. Therefore, I assume that the Moraceae pollen does not come from the local Ficus, but from other members of the

Moraceae. The pollen of Chlophora and Brosimum seems to be more anemophilous than Ficus from evidence of their staminate catkins as preserved on herbarium vouchers. Typha dominguensis pollen, ranging from 5 to 6 percent, Nymphaea

(1-5 percent), and sedge pollen (10-17 percent), are indicators of the rank, weedy growth surrounding lakes Eckixil and Petenxil.

Other pollen types represented are Graminae, low-spine Composi-

tae and high-spine Compositae from weed-covered old fields surrounding

the lake, exotic pine, oak from the encinales 1.5 km to the south of 44 the lakes, and Bursera, Sapotaceae and Eugenia from the mesic and deciduous forests surrounding the lakes.

Type 2: Mesic Forest

The mesic Peten forest has been described elsewhere (Lundell

1934) in such detail that I can add little, due to limited time and experience with these complex ecosystems. Sites SL-1, VT-1 and SL-3 were grouped as being mesic forest. Strictly speaking, these cannot be called rain forest in sensu Richards (1952), since the mesic forest is neither as high nor as stratified as true rain forest, but it is essentially evergreen.

The mesic forest does not have any dominant arboreal species in sites VT-1 and SL-3, but seems to have a suite of about 12 species which are co-dominants. These include Brosimum alicastrum, Ficus spp.,

Coccoloba belizensis, Lysiloma bahamensis, Cedrela mexicana, Swietenia macrophylla, Simaruba glauca, Bursera simaruba, Pseudobombax ellipticum,

Ceiba pentandra, Manihkara achras and Chrysophyllum sp., which vary in their degree of codominance from locality to locality.

Ten soil surface samples from beneath a closed canopy and two mud-water interface samples were collected at Site VT-1 to describe pollen productivity of the Peten mesic forest (Table 7). Absolute soil pollen influx was very low at this site, with a mean value of about 150 grains per gram of soil (dry weight). The cause is a combination of rapid degradation of pollen by soil-dwelling microorganisms, mechanical breakage by larger soil biota, and low local productivity of pollen by the mesic forest taxa. The pollen that was preserved was often

C•- r(n (\.1 n rrN N tr1 C) CY• co N CO Es- -D N0 (Y\ • • . • • • S • . • O H 0 0 0 N 0 n.0

n 0 o u- 0 0 000 000000 1 i 000000000n • • • • • . . 01 N ifn I IN NH -a- C- i I I i 1 1 i I ai 0 0 0 l 00000 0000 0 I 10000000000 • . . • . . rrn Cr\ H II n.0 n..0 N H I I 0 0001 i.fn 0 ir1 0 0 0 0 0 0 0 0 1 I 0 0 0 0 0 0 0 Ll' 0 0 11 • • • • • 0 0 r;\ • 1 I 0 0 0 r-1 r-1 1 I I 4-) 0001 0000 0 0 0 0 0 0 I 0 0 0 0 0 0 0 0 0 0 • a) . .1 • . o n..0 H Cd CO ,..0 H ....1- 4-) ;.4 CL) CI) H a) ci Cul c.) E cd 0 0 0 H 0000 0nn000.-14-)0000000000 g 111 • • • . • cd 0 • • • • H 0 0 r--1 r-I p., P, te\ 1(1 N co -ri cd tr) O I c24 o ti-n 0000 0 0 0 0 0 0 00000110000 • • • • • I I • • • . a) 00 N r(\ r-i -a- C,-- H N i i 1 i 0 0 0 1rn 0 0 0 0 0 0 0 0 0 i.rn 0 0 0 0 0 0 0 0 0 1 1 • 1 1 • • • • . • N 0 Hoe n..0 CTN k..0 o. I 1 1 1 00 0 01 0 1-fl 0 0 or-lincoo 000u-,000000 -H • • . • • 1 I • • • • a • k..0 H 0 IT \ N H n.0 N -.1" l'eN ',D II H ai 1 1 4-4 000 01 0 0 0 0 0 0 0 0 0 1!".\ 001-C\0000000 ai • • • I 1 • • • 5 • rd H r-1 0 N HO 0 0 H H H O -H 4-4 ai 4-3 a) tiO e•—••n ai CL) 7:1 al g cd Q) •H P-n ai o -I-) CO P-1 ai -H •r-1 CO ail-;-' Cd cul cul a) p- Q) $-d ai Q) a) g Cd C1) CO t•1 6.0 3 r-1 a)I a) cd cul cul F-r F-1 •s-1 0 4-) C-) L.4 O Q) .1.4 C.) ai E • .0 H o -ri c.) -ri (Ti C.) 0 cd g a) • ct g ai • H al Cd r-I a) ai • (1) cd HI X) tID P. a) ;-1 P4 cd o H • ai g, • Hi a) a) • a) p-O • 0 ,P • H P- fai C1) >.D cd cd ai p- ai 1.0 • c.) P, E ta CO Cul CO CO -1-) E 4 4 .0 4-4 o p.., a) Cs- (a (Q g ai •1-1 rl ai P, -4 P., 0 H -ri -H c.) czi ai cd cd a) E L-4 g Cli P, o 'o 0H H CO CO ai cd -H g C1) E a) -ri 0 • 0 a) ai -P tO LO 0 a) 0 0 -H 0 (O -H 0 o 4-> o -0 -P a) g E P P TI a) a) P E cd c.) C.) o 4-, c.) HL ai o L-I 0 E E E p, tc P, u ai Cd o a) cul $.1 c2L Cul a) a) a) o eC .Z 0 O O 0 0 P P -r-1 o 0 o 000 0 U 0 C..3 000WWgr-iC

1+6

4 c0 Lr•O• O H 4- O• N Orr•OON o •..0 r.rn H • • • • • • • • • • • O H 00 OH 0 0 N H H H

n00000Ul000000 0000000 000000 I • • le O 0 \..0 1H H 1H

00n0000000000 100000001000000 • • • • • IN0\.0 re)I rel

00000000n000o 100n00011•1000000 • • S • IN 0

0000000000000 100000001000000 • • • • • • Lt• N tr1 tr• O 0 d H

00n0000000000 •z 0000000r-In00000 • • . ct • • 0 P n.o a.0 H

00 0Lr• 000 00 00 00 00 00000 000000 • t • • t 0 N N HI

0000000000000 ' Oc O 0000 O 00 000 • • 1 • . 1 HN Lr•

00000U10000000 00 O 0 0 0 O O 00 0 00 • • • • • t • • • • OH NHH co

000000000000 00 0 0 00 O 00000 • • I • • rrn cc\ 0‘.0 t.4-n o

d -H H 4-) ZH 40.0 • O fa,Z o o OH E P • • m Sa,• 4-1E WP, d U) r4 ai d Ho 0 O 0 bD o H 41) 0.3 d WW 1+7

u" \ CV -d- L1\00000 N -.1- • • • I--1 rcn 0 N

O 0 00 0 0 \ 0 0 00 • s • • O r—I

O 0 0 0 0 0 0 0 0 0 co n.0

O 0 0 0 0 0 0 0 0 0 0 • • co

O 0 0 0 0 0 0 0 0 0 0 • • • N N g-1

O• 0 0 0 0 0 0 0 0 0 0 co ren

Lc\0000000000 • • rc\

O 0 0 0 0 0 0 0 0 0 0 • • H rc te\ 0• 0 0 0 0 0 0 0 0 0• 0• C"-- r-i

\ 0 00 000 0000 • •

g ai 48 corroded or broken, especially the Moraceae-type. However low the in-

flux and degraded the pollen, the mesic forest data are extremely important, as they represent the most abundant lowland vegetation types.

The most abundant taxon in the soil samples is Terminalia,

followed by the extralocal Pinus. Other forest taxa represented in

the pollen rain include Moraceae, Zanthoxylum, Sapotaceae, Eugenia,

Bursera, and others in trace amounts (Table 7).

The evidence for underrepresentation of forest types is the

relative abundance of disturbance and encinal pollen in the forest

sample suite. Although insignificant in absolute influx compared to

grassland sites, these taxa may obscure patterns in the arboreal data

when arrayed in percentage form. Although grass was absent in the

transect data, it contributed 6.94 percent of the pollen, as did

Quercus, with 7.44 percent of the pollen rain.

The mud-water interface samples collected in 1975 are similar

to the soil sample suite, with several exceptions. Moraceae, which

were much less common than Terminalia in the soil series, is equally

abundant in the two core samples. Sapotaceae and Burseraceae become

more abundant in the lacustrine series and Pinus becomes less important,

in seeming contradiction to the belief that lakes accumulate more

exotic pollen than soil samples do.

Type 3: Monsoon Forest South of the Lake Peten basin lie karstic knolls and clay-

filled valleys mantled by forests of pronounced seasonal character.

During the height of the dry season in February and March, many canopy 49 species are leafless and some are in flower. Their stature is similar to that of the mesic forest (Type 2) 1 but less stratification is evident below the jumbled canopy.

This seasonal formation is hard for me to characterize due to

the prevalence of unknown and/or undeterminable dominants (due to their

sterile and leafless condition at the season of my study there), and

the extensive cutting of the forest by highland Maya settlers in such

new towns as Porucilo. Two species-abundance sites were classified

into the monsoon forest type by the similarity index-gradient analysis.

Sites SL-6 and SL-7, well defined by the analysis and with a high similarity index of 72.5 between them, were nonetheless a surprise, since their aspects were different in the field. This indicates the usefulness of the method in objectively grouping formations vis-à-vis personal observational bias.

The monsoon forests are mainly arboreal; fewer shrubs, epiphytes and lianas are found in this type. Only along brightly lit road cuts do we find leguminoseous shrubs in any quantity. Lianas grow in areas that receive much light and give an impression of being common, but in undisturbed portions of the forest, an occasional Smilax, Monstera, and

Philodendron will be seen, with other unknown lianas. The epiphytes are mainly members of the Bromeliaceae; orchids, cacti and ferns are rare. The giant epiphytic, Aechmea bracteata, is often seen, as are the xeromorphic species of Tillandsia.

Diverse trees support the epiphytes, from the towering Ceiba pentandra, through intermediate-sized trees such as the "sausage pod," 50

Cassia grandis, to the hemi-epiphytic Ficus and Clusia. A list of

plants identified from the monsoon forest is presented in Table 8.

Since the monsoon forest is found on well drained sites, such

as the karstic knolls south of Lake Peten, or in clay filled valleys,

water stress probably is an important determinant of its distribution.

It is similar in structure and composition to the deciduous tropical

forest found in Yucatan and on the Pacific slopes of Mexico (Rzedowsii

and McVaugh 1966), zones which today have as little as 500 mm of annual

rainfall, and a lengthy dry season. In such areas, deciduous forest

is a climatic climax, not edaphic as in the central Peten. However,

it is important to consider that if precipitation had been less in the

past, the monsoon forest may have become a much expanded climatic cli-

max vegetation type in the past. The palynological implications of

this possibility will be analyzed in the concluding chapters.

Four surface samples taken north of La Libertad, Peten (Site

SL-7), serve as the pollen data point for the monsoon forest vegetation

type. The nearby aguada had been heavily used by the inhabitants of La

Libertad and local cattle, creating a local disturbance flora in a ring about the aguada.

The pollen, which was well preserved and abundant (ca. 2500 grains/gram), is virtually indistinguishable from that of the mesic forest, except for a somewhat smaller representation of Terminalia,

Zanthoxylum, low-spine Compositae, and higher representation of Pinus and Bursera (Table 8). The distinction between monsoon and mesic forest is critical for climatic reconstruction, since the tao will 51

Table 8. Pollen and vegetation data, Site SL-7, La Libertad, Peten.

Midpoint Cover Pollen Percent (N=250 grains) Taxon Class SL7-A SL7-B SL7-C SL7-D

Acacia sp. 2.5 1.5 0 2.0 0.5 1.0 Acrocomia mexicana 2.5 0 0 0 0 0 Aechmaea bracteata 0.5 0 0 0 0 0 Alnus sp. 0 1.0 0 0 0 0.25 Andropogon sp. 2.5 0 0 0 0 0 Arthrostylidium sp. 0 0 0 0 0 0 Bromelia karatas 0.5 0 0 0 0 0 Brosimum alicastrum 2.5 0 0 0 0 0 Bursera simaruba 15.0 0 0 0 0 0 Carpinus caroliniana 0 1.5 0 0 0 0.38 Caryophyllaceae 2.5 0 2.5 0 1.0 0 .88 Cassia grandis 15.0 0 0 0 0 0 Catopsis abides 0.5 0 0 0 0 0 Cecropia peltata 15.0 0 2.5 0 0 0.63 Ceiba pentandra 15.0 0 1.0 0 0 0.25 Celtis sp. 0 2.5 0 1.5 0.5 1.12 Chaetoptelea mexicana 0 0 0 0 0.5 0.13 Cheno-am 2.5 2.5 1.5 0.5 1.5 1.50 Chrysophyllum sp. 2.5 0 0 0 0 0 Coccoloba belizensis 2.5 0 0 1.0 0 0.25 Cochlospermum vitifolium 15.0 0 0 0 0 0 Commelina sp. 0.5 0 0 0 0 0 Compositae, hi-spine 0.5 2.5 0 5.5 1.0 2.25 Compositae, lo-spine 2.5 0 7 • 4 10.2 6.5 6.13 Crysophila argentea 0.5 0 0 0 0 0 Cyperaceae 2.5 2.0 1.5 3.5 o 1.75 Eugenia sp. 2.5 1.0 0.5 1.5 2.5 1.38 Euphorbiaceae 2.5 0 0 0 0 0 Ficus spp. 2.5 0 0 0 0 Graminae 0.5 Guazuma ulmifolia 2.5 0 0 0 0 0 Haematoxylon sp. 0 0.5 0 0 0 0.13 Inga spp. 2.5 0 0 0 0 0 0.5 0 0 0 0 0 2.22112ta 'EPP° Liquidambar styraciflors 0 0.5 1.0 0 0 0.38 Lysiloma bahamanensis 2.5 0 0 0 0 0 Machaerium setulosum 0.5 0 0 0 0 0 Malpighiaceae 0.5 0 0.5 0 2.5 0.75 Manilkara achras 0 1.0 0 0.5 2.0 0 .88 Monstera sp. 0.5 0 0 0 0 0 Moraceae 15.0 0 0 0 0 0 Myrica sp. 0 0 0 0 0 0 Myrtaceae 0 0 0 2.0 0 0.50 52

Table 8--Continued.

Midpoint Cover Pollen Percent (14.250 grains) Taxon Class SL7-A SL7-B SL7-C SL7-D X

Orbignya cohune 2.5 0 0 0 0 0 Palmae 15.0 0 0 0 0 0 Philodendron smithii 0.5 0 0 0 0 0 Phoradendron spp. 0.5 0 0 0 0 0 Pinus sp. 0 10.0 2.5 8.5 12.0 8.25 Pinar sp. 2.5 0 0 0 0 0 Podocarpus sp. 0 0 0.5 0 0 0.13 Psychotria sp. 0.5 0 0 0 0 0 Quercus oleoides 0 2.5 7.0 1.0 9.5 5.0 Randia sp. 0.5 0 0 0 0 0 Sabal morrisiana 2.5 0 0 0 0 0 Sapotaceae 4.0 1.0 0.5 0.0 1.38 Schizolobium sp. 0.5 0 0 0 0 0 Simaruba glauca 2.5 0 0 0 0 0 Smilax sr. 0.5 0 0 0 0 0 Solanaceae 0.5 1.0 3.5 2.5 0.5 1.88 Spondias sp. 2.5 0 0 0 0 0 Tabebuia sp. 2.5 0 0 0 0 0 Terminalia-Buchenauia 2.5 Tillandsia spp. 0.5 0 0 0 0 0 Trema micrantha 2.5 Zanthoxylum spp. 0 0 0 0 0.5 0.13 53 alternate in response to climatic change, but in both floral composi-

tion and pollen productivity they are very similar.

Type 4: Orbignva cohune Forest

Occasionally on gentle slopes with deep soils, a palm-dominated

association can be found. , The dominant perennial is the corozo,

Orbignya (Attalea) cohune, a tall "feather palm" sometimes reaching

over 50 feet (ca. 16 m) in height. The palm groves, called "corozales"

in Belice, may be anthropogenic in origin, since the tree is sometimes

spared in clearing new milpas. Also occurring with Orbignya may be

Scheelia lundellii, a vegetatively similar-appearing palm, but easily

differentiated on inflorescence characters. Portions of the corozales

lack other dominant tree species and all arboreal cover is of Orbignya.

However, shrubby perennials are common (see Table 9).

Site SL-2, located just to the north of the Flores-Melchor de

Mencos Road, was on a north-facing slope of 15° with deep alluvial

soils showing slight oxidation. The upland sites were covered with

Orbignya, with isolated other taxa including the taller fan palm, Sabal

morrisiana, and the shorter Acrocomia mexicana on more weathered soils.

The dicot trees on the uplands were Simaruba glauca, Inga sp., Guazuma

ulmifolia, and one Quercus oleoides. Valley bottoms contained a more broad-leafed vegetation type,

similar in composition to the mesic forest found elsewhere in the Peten.

In addition to the taxa mentioned above, Bursera simaruba, Brosimum

alicastrum, Ficus spp., Chrysophyllum sp. and Spondias sp., are found

here on well-drained sites. 54

Table 9. Pollen and vegetation data, Site SL-2, Macanche, Peten.

Midpoint Cover Pollen Percent (N=250 grains) Taxon Class SL2-A SL2-B

Acacia sp. 2.5 2.5 0.5 1.5 Acrocomia mexicana 2.5+ 0 0 Aechmea bracteata 0.5 0 0 Alnus sp. 0 0.5 0 0.25 Andropogon sp. 15.0 Arthrostylidium sp. 0.5 0 0 Bromelia karatas 0.5 0 0 Brosimum alicastrum 2.5 0 0 Bursera simaruba 2.5 3.5 2.0 2.75 Caryophyllaceae 2.5 1.5 2.0 1.75 Cecropia yeltata 2.5 1.5 3.0 2.25 Ceiba pentandra 2.5 0.5 0 0.25 Celtis sp. 0 2.0 1.5 1.75 Chaetoptelea mexicana 0 0.0 1.0 0.50 Cheno-am 2.5 1.5 2.0 1.75 Chrysophyllum sp. 2.5 0 0 Cochlospermum vitifolium 0.5 0 0 Compositae, high-spine 0 5.0 4 • 5 4.75 Compositae, low-spine 2.5 4.5 3.5 4.0 Conostygia sp. 2.5 0 0 Cordia sp. 0.5 0 0 Crysophila argentea 0.5 0 0 Cyperaceae 2.5 0.5 1.0 0.73 Eugenia sp. 2.5 0.5 0 0.25 Euphorbiaceae 0.5 3.0 2.5 2.75 Ficus sp. 0.5 0 0 Graminae 15.0 25.5 46.5 36.0 Guanzuma ulmifolia 15.0 0 0 Haematoxylon sp. 0.5 1.0 0 0.50 Inga sp. 0.5 0 0 Ipomoea spp. 0.5 0.5 0 0.25 Liquidambar styraciflua 0.0 0.5 0 Machaerium setucosum 0.5 0 0 0.25 Malpighiaceae 0.5 2.5 1.0 Melastomaceae 2.5 0 0 1.75 Monstera sp. 0.5 0 0 Moraceae 2.5 10.5 8.5 9.5 Mucuna andreana 2.5 o o Myrica sp. 0 2.0 0.0 1.0 Myrtaceae 0 1.5 0.5 1.0 Orbignya cohune 75.0 0 0 Palmae 75.0 2.5 4.5 3.5 Pinus spp. 0 6.5 2.5 4.5 Lips" spp. 2.5 0 0 55

Table 9--Continued

Midpoint Cover Pollen Percent (N=250 grains) Taxon Class SL2-A SL2-B

Podocarpus spp. 0 0.5 0 0.25 Psychotria sp. 2.5 0 0 Quercus oleoides 0.5 6.5 2.5 4 • 5 Randia sp. 0.5 0 0 Sabal morrisiana 2.5 0 0 Sapotaceae 2.5 1.0 0.5 0.75 Schizolobium sp. 0.5 0 0 Simaruba glauca 0.5 0 0 Smilax sp. 0.5 0 0 Solanaceae 2.5 1.0 0 0.50 Spondias sp. 0.5 0 0 Tabebuia sp. 0.5 0 0 Terminalia sp. 2.5 6.5 10.5 8.5 Trema micrantha 2.5 4. 0 0 .5 2.25 Vismia ferruginea 0.5 0 0 Zanthoxylum sp. 0.0 0.5 0 0.25 56

Disturbance of the Orbignya forest at Site SL-2 was not pronounced, although many trails and roads were evident from the Flores-

Melchor highway. Cecropia peltata, Trema micrantha var. floridiana, and Acacia spadicigera were found in disturbed zones near trails and roadcuts.

I cannot visualize the position of the Orbignya forest in a climatic-edaphic gradient, although the numerical classification allies it to the monsoon forest. This may be a very specialized edaphic community, or an anthropogenic one. The palm forest, like the encinal, must be studied in much more detail before a synoptic view of its place in the history of Peten may be understood.

Type 5: Aguada Forest Vegetation

Situated within the grassy plains south of Lake Peten are numerous small sinkholes or aguadas. The aguadas are usually depressions of varying sizes that are capable of holding water for part of the year. During the dry season of 1975, most savanna-region aguadas had dried to mud-cracked sediments or contained a rapidly diminishing puddle of cattle-muddied water. Usually the clay sediment forms an impervious "cup," allowing no water to flow out. Some aguadas seem to have underground discharge. Surrounding the aguada is an arboreal vegetation vastly dif-

ferent from the broad, grass-covered plains. This vegetation type, which I call aguada forest, is a high forest (trees from 5-20 m) consisting of a jumbled, closed canopy and numerous emergents. It is most closely related to monsoon forest and encinal (Table 10). 57

Table 10. Pollen and vegetation data, Site VT-8, Bosque el Caobal, Peten.

Midpoint Cover Pollen Percent (N=250 grains) Taxon Class 1JT-8-a VT-8-b

Acacia sp. 2.5 0 0.5 0.25 Acanthaceae 0.5 0 0 Acrocomia mexicana 2.5 Andropogon sp. 2.5 Arthrostylidium sp. 0.5 Bromelia karatas 0.5 Brosimum alicastrum 2.5 Bursera simaruba 2.5 1.5 0 0.75 Byrsonima crassifolia 0.5 0.5 1.0 0.75 Cecropia peltata 2.5 Cheno-am 0 0.5 3. 0 1.75 Clusia sp. 0.5 0 0 Coccoloba belizensis 2.5 0 0.5 0.25 Cochlospermum vitifolium 2.5 0 0 Commelina sp. 0.5 0 0 Compositae, high-spine 0.5 4.5 2.0 3.25 Compositae, low-spine 0.5 1.0 2.0 1.50 Conostygia sp. 2.5 0 Crescentia cujete 0.5 0 Cyperaceae 0 3.5 4.o 3.75 Graminae 2.5 24.0 14.5 19.25 Guazuma ulmifolia 2.5 Ipomoea spp. 0.5 Jacquinia sp. 0.5 Liquidambar styraciflua 0 1.0 0 0.50 Machaerium setucosum 2.5 Malpighiaceae 0.5 2.5 4. 0 3.25 Melastomaceae 2.5 Miconia argentea 2.5 Mimosa spp. 0.5 Moraceae 2.5 20.5 24. 0 22.25 Mucuna andreana 2.5 Myrtaceae 0 Orbignya cohune 2.5 Palmae 2.5 1.0 2.5 1.75 Philodendron smithii 0.5 Phoradendron sp. 0.5 Pinus sp. 0 8.5 12.0 10.25 Piper sp. 2.5 0.5 0 0.25 Podocarpus spp. 0 1.5 1.0 1.25 Pseudobombax ellipticum 2.5 2.5 3.0 2.75 Psychotria sp. 0.5 58

Table 10 - -Continued.

Midpoint Cover Pollen Percent (N=250 grains) Taxon Class 1JT-8-a VT-8-b

Quercus oleoides 2.5 10.0 14.0 12.0 Randia sp. 2.5 Rhynchosia pyramioaus 0 .5 Salvia sp. 0.5 1.5 0.75 Simaruba glauca 0.5 Syngonium sp. 0.5 Terminalia sp. 0 16.0 4.5 10.25 Trema micrantha 2.5 3.5 o 1.75 Tillandsia spp. 0.5 Vismia ferruginea 2.5 Zea mays 0 59

The form of the "ideal" aguada forest is in the shape of a doughnut, with the mud-water depression as the doughnut hole. The outer border of the forest is well-defined; a vegetation transect taken from the grassland into the forest indicates a transition from herb to tree within two meters. This transition zone is similar to marginal forest, but is laterally compressed.

There are no visible dominants in this vegetation type, so the species lists taken in these sites must suffice (Site VT-8). However,

Leguminosae and Melastomaceae dominate the outside edge of the "doughnut," rarely penetrating the central portion of the aguada forest.

The clambering, scrambly palm, Bactris, is found toward the central depression, sometimes becoming a sub-dominant.

Whether these forests are relicts or vanguards of high forest will probably never be known. Today, however, they are edaphically controlled formations which are adapted to a marginal existence in the broad, treeless plains of southern Peten.

Type 6: Encinal (Oak Woodland)

The tropical oak woodland is similar in structure to temperate oak woodlands. It is dominated by the encino, Quercus oleoides var. australis, an 18-m high tree with widespread crown and small entire, or three-toothed coriaceous leaves. The trunk is massive, with a large swelling at ground level, almost a platform. Density of these trees may be as little as ten or as many as 60 trees per hectare. The branches of the encino support ephiphytes, especially the bromeliads

Tillandsia Aechmea, Catopsis, and lichens. 60

An understory of typically savannah trees such as Brysonima form a sparse undercanopy in more open situations (<5% cover), but not within the shade of the encinos. The coyol palm Acrocomia mexicana seems to be a minor constituent of the encinal, but it is nowhere common. Acacia cornigera (ca. 3% cover),a common understory tree, also grows as a disturbance indicator in the drier parts of encinal. In two of the sites I visited, the acacias had been cut, scattering their hard, bullhorn spines upon the ground and rendering travel unpleasant for the unwary. Jacquinia sp., a subshrub found in the encinal, has the habit of deciduating during the wet season and photosynthesizing during the winter dry season. Other minor encinal dwellers are Miconia (<5% cover) and several other Melastomaceae, the terrestrial bromelaid, Bromelia karatas, Eugenia, and many unknown species (Table 11).

The herb component is dominated by grasses and sedges, with a tiny Mimosa being common. Except for an occasional Rhynchosia, Ipomoea, or Philodendron, vines and lianas are virtually absent in the encinal.

The encinal borders are generally well defined. Oaks grow on alluvial clay soils, and quickly give way to a dry, partly deciduous forest (Type 3) on slopes or a mesic high forest (Type 2) along some streamways. Where edaphic changes are not abrupt, the encinal border grades imperceptibly into mixed forest with a slow enrichment by new species, usually beginning as stunted, scraggly individuals. The encinal may also occur as isolated pockets within savannahs occupying slightly undulating terrain.

The palynological results of the encinal sample suite were the most satisfying in that they nearly fit the vegetational data. 61

0 tfn 0 0 00 0 Ltn rc-N N. rrn H H 1/4D N- Cs• 0 • • • . • • • • 0 H 0 0 OH Lt\ HI itn

0 0 0 0 0 • • • 0 \*1 Cu

0 0 14- \ 0 00 0 0 0 • • S • • 14" c\J Cs- n.1) n -1 0 t11 0 0 00 0 Lr\ 0 Cri • • • • • Ill C\JI H CO

o o 00 U o • 0 ;21 ( CO

O 0 0 0 00 crn 0 0 • • 6 H \ 00 L.C1

0 0 L.r.\ 0 00 0 u1 0 • • • 0

o 0 0 0 00 Lrn 0 • o. rcN r‘jS

O 0 0 0 00 11 \ 0 O

r-1S C \I tn-• VD

o 0 0 00 0 0 0 • • • • v:)

o 0 0 00 0 If\ 0 • S S • tc1

Lnnnno••••••••••••• inno n u-\ o ntnLnLnino • • • • nnLnno• • • . • Ln• • inn• o. . NOONTn NOONN000 000000N00000 00

-r-1 -ri COP, I Ul r]W ,g I -P 111C 0 •I-1 0 • • I) H w 0

0 0 0 0 CC 0 ai . 1-1 • ri • -P -p -4-) G) g CO Q) Q) 000 E P-RO M E EEZ o o 0 0 $.4 u 0

U1 U1 0 0 0 LI1 0U\ 0 H CO ri 1-N H -.I- 1.11 CC N • s s s s s • s • 0 Cc\ 0 0 0 c0 HO 0 H H

O 0 0 0 0 Nr\ 00 0 • • g 4- trN H Lrn

0 0 0 0 0 Lf1 00 0 e e s (*Z Ns H N r-I

O 0 0 0 0 0 00 0 • • • •

O TN 0 0 0 0 00 0 • • 0 H

O \ 0 0 0 0 00 0 • C-.!) cC n.C) rcn H

O 0 0 0 0 0 00 0 • • • • • H

O 0 0 0 O O 0 LN 0 co

0 1-N 0 0 0 0 00 0 • •

O 0 0 0 0 L11 00 • • - • 121 0 H

if\ 0 0 0 0 0 0 00 0 • s • • • N 0 H !sr\

n Lr\ 0 IN 0 LN tr\ LA 0 1-N 0 0 1-No LN Lr\Lr Lro usn t.N if\ 0 • • • • • • • • • • • • • • . • • • • • • • . • • • • ONNNONOON00000 ool.rn 000 000 0000

Ui G) cri • rcs P-n CO 0 • ct 0 P, CO M U) 0 cd o rI

cd 0' 63

Absolute pollen concentration was high, ca. 2000 grains/gram, and the pollen was well preserved. Grass and oak pollen were the dominant types, as they were in actual coverage. Extralocal pine, local

Moraceae, Cyperaceae, Bursera, long-spine Compositae and Terminalia also were present in amounts above traces. I suspect that the wind- pollinated grass and oak from such sites as these send a cloud of pollen out over vast areas of tropical forest and obscure the true pattern of the tropical pollen taxa.

Type 7: Thorn Scrub

A rarely encountered vegetation type is the Peten Thorn scrub, a low, xeric woodland, with an infusion of leaf and stem succulents, recalling Sinaloan Thorn forest (Shreve 1937). This vegetation formation, between two and five meters high, is mostly deciduous during the dry season. It occupies well-drained, karstic outcrops in the Peten savannah country, and can be seen from the Flores-Sayaxche Road near

Sibun where two sample sites are located (SL-12, SL-13).

The dominant tree is the bullhorn acacia (Acacia spadicigera) with an estimated coverage of 5. Subdominant Leguminosae are Acacia angustissima, and a 2 ni -high Mimosa (4. cf. albida), and others unknown to me. The siiktree, Cochlospermum vitifolium, is found as scattered individuals on the rubble-covered slopes, highly conspicuous due to its yellow flowers. A rosette-leaf succulent member of the Agaveaceae found here remains a puzzle. I am unaware of any lowland Agave or

Furcraea described from this region, nonetheless this plant appears to belong to one of these genera. The leaves compare closely with 64 herbarium specimens of Furcraea bedinghausii, but the paniculate inflorescence resembles Agave

A semi-epiphytic Clusia on rocks and road cuts in the thorn scrub, and the stem succulent, Plumeria acutifolia, is rare to abundant.

Surprisingly, melastomes were not encountered. Herb cover is sparse, ca. 20%, but a bright red Salvia was common, as was the semi-woody

Rhynchosia pyramidalis. On clay alluvial flats, the thorn scrub is taller and approaches the height of the monsoon forest. Savannas are small and tree-filled.

The Peten thorn scrub may be related to marginal forest (Lun- dell 1934), but it differs from it in edaphic habitat, presence of microphyllous legumes, and absence of Melastomaceae and Byrsonima.

There is always a desire to infer from these scrublands an anthropogenic or climatic cause for their presence as is the vogue among savannah enthusiasts. However, they are maintained today by edaphic factors and probably fire (charcoal found at Site SL-13). The component species are not endemics, but can be found in diverse habitats throughout the Peten. Table 12 is a species list taken in thorn scrub as found in the Peten.

Type 8: Sahalal

The first savanna formation encountered along the gradient is the sahalal, a jumbled, weedy orchard-like formation. The dominant tree is the saha Curatella americana, a 3-6.5 m tall tree with large, sandpaper-like leaves borne on twisted, upright branches. Measurements taken at Site VT-6 near La Libertad, Peten, indicate that the mean 65 Table 12. Pollen and vegetation data, Site SL-13 Sibun, Peten.

Midpoint Cover Pollen Percent Taxon Class 1 2 Acacia sp. 15.0 3.5 2.0 2.75 Acanthaceae 2.5 0 0 0 Aechmea bracteata 0.5 0 0 0 Agave S. 0.5 0.5 0 0.25 Bursera simaruba 0.0 2.5 0.5 1.50 Byrsonima crassifolia 2.5 0 0 0 Cassia sp. 0.5 0 0 0 Cecropia peltata 0.5 0.5 2.0 1.25 Cheno-am 0.0 6.5 10.5 8.50 Chrysophyllum 2.5 0 0 0 Clusia sp. 0.5 0 0 0 Cochlospermum vitifolium 2.5 0 0 0 Compositae, high-spine 0.5 5.0 2.5 3.75 Compositae, low-spine 0.0 2.5 1.0 1.75 Crescentia cu'ete 2.5 0 0 0 Curatella americana 2.5 0 0 0 Cyperaceae 0.0 3.5 0.5 2.00 Erythrina americana 2.5 0 0 0 Eugenia sp. 0.0 1.0 0 0.50 Graminae 25.0 27.5 35.2 31.50 Guazuma ulmifolia 2.5 0 0 0 Liquidambar styraciflua 0.0 0.5 0 0.25 Machaerium setolosum 0.5 0 0 0 Malphigiaceae 0.5 2.5 1.0 1.75 Mimosa sp. 2.5 0 0 0 Moraceae 0.0 18.5 24.0 21.25 Myrtaceae 0.0 1.0 4.5 2.75 Pinus sp. 0.0 7.5 2.0 4.75 Plumeria acutifolia 0.5 0 0 0 Podocarpus sp. 0.0 0.5 0 0.25 Pseudobombax ellipticum 0.0 0.5 1.0 0.75 Quercus oleoides 0.5 2.5 4.o 3.25 Randia sp. 0.5 0 0 0 Rhynchosia sp. 0.5 0 0 0 Sabal morrisiana 0.5 0 0 0 Salvia sp. 0.5 0 0 0 Terminalia sp. 0.5 11.0 5.0 8.00 Trema sp. 0.5 2.5 4.0 3.25 66 density of Curatella is about 160 trees/hectare. The saha supports a few epiphytes, such as Tillandsia spp., a tiny orchid, and many lichens.

Growing in the open shade under each saha is a jumbled mass of shrub- bery of the Melastome family. I recognized Miconia sp. and two species of Conostegia in the tangled eyots.

The problem of the distribution of Melastomaceae is interesting.

The Melastome family is a moist climate-loving group of plants. Her- barium vouchers, Standley's "Trees and Shrubs of Mexico" (Standley et al. 1943) and "Melastomaceae of the Yucatan Peninsula" (Gleason

1940), indicate only one species, Conostegia xalapensis, growing in areas having high water stress (Sinaloan thorn forest).

In the more arid Yucatan Peninsula, the Melastomaceae are much less common than in the Peten. Indeed, the Melastomes seen in open sites were usually wilted, while those seen in forest situations were turgid. Somehow, the combination of high light intensity and micro- edaphic conditions in the immediate vicinity of savanna "nurse plants" is an optimal habitat for Melastomaceae. This paradox between mesic and xeric habitat preference invites study.

The grasslands surrounding the Curatella-;Melastomaceae are characterized by numerous forbs, of which Salvia and Brickellia were recognized. Andropogon spp. are less common than in more grass- dominated sites.

Byrsonima crassifolia, usually found growing in the Curatella in tropical savannas (Wagner 1964), occurs only as a rare straggler in sahalals in the Peten. It forms its own association, the nanzal, to be discussed next, indicating a divergent habitat preference in the Peten. 67

A list of species found in the sahalal is given in Table 13.

Sahalal Pollen. Pollen deposition in sahalal vegetation was investigated at Site VT-6, El Guanal, Peten. The data are from a single soil sample taken from a large grassy zone within the arboreal savanna.

The palynological results were unexpected in that grass pollen, from a taxon which had a coverage value of 74.5, had a 96 percent pollen deposition onto the soil, the highest found within any tropical vegetation type, including full grassland. The only other pollen taxon represented was the anamolous Terminalia-type, with 2 percent representation. All other taxa were either absent, or were present only in

trace amounts (Table 13).

Type 9: Nanzal

The most abundant arborescent savanna in the Peten is the

nanzal, an open community dominated by Byrsonima crassifolia and vari-

ous grasses. The nanzal appears as an arm of a larger grassland, a

transition to marginal forest, or more rarely, as a glade in marginal

forest or encinal. The dominant Byrsonima averages 90 trees/hectare

(Site VT-7) or less. It is a small, gnarled tree ranging from 1.25 to

3.5 m in height. Epiphytes on Byrsonima are restricted to lichens;

only one tiny Tillandsia was observed. Byrsonima has leathery, usually bug-infested, simple leaves which turn bright red before falling, and

a tartly delicious, highly prized fruit. The subdominant family is the Melastomaceae (>1% cover), represented by Miconia argentea, another

species of Miconia, and Conostegia sp., which may nestle under

Byrsonima, or grow in solitary clumps. 68

Table 13. Pollen and vegetation data, Site VT-6, El Guanal, Peten.

Midpoint Cover Pollen Percent (N=250 grains) Taxon Class 1

Acacia sp. 0.5 0 Acrocomia mexicana 0.5 0 Andropogon sp. 25.0 0 Bursera simaruba 0.0 0 Byrsonima crassifolia 0.5 0.0 Cecropia peltata 0.5 0 Cheno-am 0.5 0.0 Compositae, high spine 2.5 1.5 Compositae, low spine 0.0 Conostegia sp. 15.0 0 Curatella americana 15.0 0 Cyperaceae 2.5 0 Graminae 14.5 96.0 Ipomoea sp. 0.5 0 Machaerium setolosum 2.5 0 Malphigiaceae 0.5 0 Melastomaceae 2.5 0 Moraceae 0.0 0 Myrtaceae 0.0 0 Pinus sp. 0.0 0 Podocarpus sp. 0.0 0 Pseudobombax ellipticum 0.0 0 Quercus oleoides 0.0 0.5 Terminalia sp. 0.0 2.0 Tillandsia sp. 0.5 0 Trema micrantha 0.0 0 69

The herb component is dominated by Graminae (some Andropogcon), which has an average cover of 70.25% (Site VT-7). The dry grass of

February was less than 50 cm in height. Herbs are rare in the nanzal for some reason, except for Mimosa pudica, and Rhynchosia sp. which nonetheless have coverage values of less than 1% (Table 14).

The borders of the nanzal are well defined. Toward open grassland, there is rarely a transition; the nanzes terminate in a ragged line, and toward the forest, the usual composition of nanzes is rapidly supplemented by marginal forest elements.

The similarity-index analysis classifies the nanzal formation as being the last vegetation type that has an arborescent element, and this agrees with field observation.

Nanzal Pollen. Nanzal, investigated at Site VT-7, Santa Ana

Vieja, Peten, was dominated by grass pollen (77 percent). Other taxa were well represented, but in small amounts. Unexpectedly, Terminalia was the second most common, at 8 percent. This may be evidence for

Vaughan's contention that the Terminalia-type includes the Melas- tomaceae (Vaughn 1976). Other taxa well represented are oak (7 percent), Leguminosae (4 percent), high-spine Compositae (4 percent), and other minor taxa. Pollen was abundant (ca. 5000 grains/gram) and was well preserved.

Type 10: Grasslands Thousands of hectares south of Lake Peten are devoid of trees, forming small prairies. These grasslands occupy broad, level plains, bounded by low, limestone forest-covered hills, sinuous forrested 70

Table 14. Pollen and vegetation data, Site VT-7, Santa Ana Vieja, Peten.

Midpoint Cover Pollen Percent (N=250 grains) Taxon Class 1

Acacia sp. 0.5 0 Andropogon spp. 39.5* o Bursera simaruba 0.0 0.5 Byrsonima crassifolia 15.0 0 Comnositae, high-snine 0.5 4.0 Compositae, low-spine 0.0 0 Conostegia sp. 0.5* 0 Cyperaceae 0.5 0 Graminae 70.5* 77.0 Malpighiaceae 0.0 0.5 Miconia argentea 0.5* 0 Mimosa pudica 0.5* 0 Melastomaceae 0.5* 0 Moraceae 0.0 0.5 Myrtaceae 0.0 1.0 Pinus sp. 0.0 1.0 Piper sp. 0.5 0 Ouercus oleoides 0.0 7.0 Rhynchosia sp. 0.5* 0 Terminalia sp. 0.0 8. 0 Tillandsia spp. 0.5 0 Trema micrantha 0.5 0.5

*Measured, not estimated. 71 streamways, and aguadas. The soils are acidic clays, a dark brownish- red surface layer, grading into a yellow-brown to brick-red horizon.

Sparse organic material may mask the red tint of the classic oxidized savanna soils.

Grass coverage from a site near Laguna Oquevix (Fig. 2) is

67.75 percent. This figure is slightly less than that of a site nearby which has sparse tree cover, which may indicate that open sites are absent, soil nutrients or structure are the limiting factors to the grasses. The most common grassland genus is Andropogon, represented by several species including the taller A. cf. condensatus, and the coarse-leaved A. bicornis. Salvia, possibly S. coccinea, is the most common forb, followed by Brickellia cf.oliganthes, Erigeron sp. and several other members of the Compositae, Leguminosae, and Acanthaceae families. An occasional scrambling Mimosa is the only semi-woody plant

I found in the grasslands. A second open grassland station near La

Libertad, Peten, had similar vegetation, with a mean grass coverage of

65.25 percent, but had less Andropogon. This was replaced by three other grasses which were sterile in February. The forbs were similar, with the addition of a species of Rhynchosia and one of Ipomoea,.

The border of the grassland is sharp, grading into marginal scrub forest, aguada forest or encinal. At the site near Libertad, I investigated the transition from treeless plain to aguada forest. A transect was run from full grassland, changing within 3 m (from m 14 to 17) to a forest regime transect characterized by low herb coverage values. Forbs, especially Compositae, Leguminosae, and Melastomaceae, 72 favor this transition zone, obtaining significant light and increased fertility of forest soils.

The grasslands may grade into marginal scrub forest. The most notable species in February and March is Cochlospermum vitifolium because of its conspicuous yellow, buttercup-like flowers and silvery trunk. A marginal scrub forest species list taken north of Lake

Oquevix includes Vismia cf. ferruginea, Miconia argentea, Termina-lia excelsa, Chrysophyllum sp., Inga punctat, Brysonima crassifolia,

Eugenia sp., and many other small trees. Agave sp. occurs as scattered clones at the forest edge. The aspect is very xeric, as are many members of its flora.

Fires often sweep the grasslands, retarded by roads or marginal scrub which act as a firebreak. Apparently unaffected by fire, individuals of the genera Byrsonima, Crescentia, Curatella, and Acrocomia stand like silent Ents, vanguard or survivors of the distant forest.

Grassland Pollen. The grassland vegetation was investigated at

Site VT-9, near Pacay, Peten, and 1 km north of Lake Oquevix. As expected, grass pollen was by far the dominant type, comprising 58 percent of the pollen. Oak pollen (12 percent), pine pollen from the pinelands 30 km to the southeast (11 percent), low-spine Compositae (4 percent), and numerous other types complete the pollen spectrum of the

Peten tropical grassland (Table 15).

The Independent Data Set: Peten Sediment Network

The relationship of soil surface samples to lake sediment pollen samples is unknown in the American tropics (except for one 73

Table 15. Pollen and vegetation data, Site VT-9, Pacay, Peten.

Midpoint Cover Pollen Percent (N=250 grains) Taxon Class

Acanthaceae 0.5 0 Acrocomia mexicana 0.5 0 Agave sp. 2.5 0 Andropogon spp. 4. 0 o Brickellia sp. 0.05* 0 Bursera simaruba 0.0 1.0 Bvrsonima crassifolia 0.5 0 Cheno-am 0.5 2.5 Compositae, low spine 0.0 4.o Crescentia cujete 0.5 0 Curatella americana 0.5 0 Cyperaceae 2.5 3.5 Erigeron sp. (hispine) 0.12* 0 Graminae 67.75* 58.5 Ipomoea sp. 0.5 0 Mimosa pudica 0.5 0 Moraceae 0.0 3.0 Myrtaceae 0.0 0.5 Pinus sp. 0.0 11.0 Popocarpus sp. 0.0 1.0 Pseudobombax ellipticum 0.0 0.5 Quercus oleoides 0.0 12.5 Rhynchosia sp. 0.5 0 Salvia sp. 0.24* 0 Terminalia sp. 0.0 2.5 Trema micrantha 0.0 0

*Measured, not estimated. 74

small study, Wiseman 1974). Different pollen transport, deposition,

circulation, and degradation systems probably characterize terrestrial

and lacustrine environments, adding unknown biases in the analog

method. Nonetheless, such data have long been used in temperate

regions. Soil and moss-polster sample data, when collected in an area

palynologically well known, may be discarded if anomalous. In a rela-

tively unknown area, such as the Maya Lowlands, no criterion exists

for recognition of anomalous samples. An independent data set is necessary to function as a touchstone for the analog.

The touchstone is a series of mud-water interface sediment

samples taken in three lakes, two ponds and two aguadas in the Central

Maya Lowlands. The sediment network, similar in depositional environment to the fossil samples, is easily comparable to soil samples taken around the basins. The sediment network is unfortunately too small for rigorous statistical comparisons, and represents too few vegetation types to serve as an analog.

Each lake sample network is tied to the vegetation gradient discussed in Chapter 3 by a contiguous vegetation plot and pollen sampling station. The applicability of the gradient analog is assessed by comparison of the mean values for each lake's pollen variables with the mean values of gradient pollen variables. If the gradient is a valid analog for fossil lacustrine samples, then there must be significant correlation between pairs of sediment and soil samples taken within a given vegetation type, but no correlation between those from different vegetation. A basic assumption in this analysis is that different vegetation types produce different pollen spectra. 75 In addition to its function as a criterion for assessing the analog, the sediment network tests the model of differential pollen dispersal and deposition within lacustrine environments that has been

derived from temperate zone pollen analyses (Davis et al. 1971). The use of a mean value negated, in the computations that follow, the bias

of differential flotation of pollen and sediment focusing.

The sediment network roughly parallels the gradient and will

be ordinated, for purposes of discussion, by the contiguous plot vege-

tation type.

Sediment Network Peten-1, Aguada Tikal, Peten, Guatemala

The Tikal aguada was built, or rebuilt, for the use of archae-

ologists during excavation of the site in the 1950's. It is now

slightly overgrown with Typha and Pistia (a floating aquatic plant of

the Arum family). The area actually holding water was much less than

a hectare and surrounded by regenerating brush and forest, periodically

cleared by Tikal gardeners.

Tikal Aguada is surrounded on all sides by high mesic forest

dominated by Brosimum alicastrum, Swietenia macrophylla, Manilkara

achras and many other trees. In bajos surrounding Tikal is found a stunted arboreal vegetation type called Tintai, dominated completely

by Haematoxylon campechianum.

Three samples were collected from the mud-water interface of

the aguada with a Davis corer tool attached to a series of Lichtwardt

rods so as to scrape the mud from the bottom into the tube. While

crude, this system was the only feasible one at the time. 76

Pollen from Aguada Tikal was partially degraded by organisms in the water or oxidation. As a consequence, rare or easily confused pollen types were almost absent from the counts, a condition not encountered in other sediment samples except Lake Oquevix. A colloid engendered during extraction hindered identification of some pollen grains.

The resulting pollen spectrum is different from other vegetation sites in many ways (Table 16). Pine pollen was the major type, representing 32.5 percent of the total. The overabundance of pine is explained in part by the low airborne pollen productivity of the local flora, biasing the data toward anemophilous pollen, even if it is not local. Also, there is an enigmatic "island" of pine in the bajo region to the east of Tikal (Dahlin, personal communication 1976), and is shown on published Guatemalan vegetation maps. The pine is probably

Pinus caribaea, the lowland pine of Belize.

Terminalia and Moraceae pollen were less common here, but

Bursera, Manilkara, Celtis and Haematoxylon were more common. Of the non-arboreal pollen types, grass was most important, with composites less so. Unexpectedly Cheno-Am pollen was fairly common. This site is the closest to what may be the "climax" forest pollen assemblage, but still contains a weed component.

Sediment Network Peten-2, Lake Petenxil, Peten, Guatemala

Lake Petenxil is a small (0.7 x 1.3 km) lake of generally north- south orientation (Fig. 6). The outlet, at the eastern end, drains into Lake Peten Itzâ (Cowgill and Hutchinson 1966). North and 77 Table 16. Pollen percentages of 17 statistically significant taxa, Aguada Tikal, Peten.

Taxon 1 2 3 AP

Acacia spp. 0.5 1.0 1.0 0.83 Bursera simaruba 4.5 1.5 6. 0 4.00 Celtis spp. 5.5 4.o 7.5 5.67 Haematoxylon spp. 0.5 1.0 4.5 2.00 Manilkara achras 5.5 1.5 3.5 3.50 Moraceae 25.5 44.0 50.5 40.00 Ouercus oleoides 12.0 5.0 3.5 6.83 Terminalia-type 10.5 12.0 13.0 11.83 Zanthoxylum 10.0 4.5 6. 0 6.83

NAP Cheno-Am 4.0 2.0 2.0 2.67 Compositae, high spine 6.5 7.0 0.5 4.67 Compositae, low spine 2.5 0.5 1.5 1.50 Cyperaceae 4.5 4.5 0 3.00 Euphorbiaceae 4.o 3.5 o 2.50 Graminae 4.5 12.0 1.0 5.83 Senecio-type 0 1.5 1.0 0.83 Zea Mays 0 1.0 o 0.33 78

Figure 6. Pollen sample network, Lake Petenxil, Peten. 79 northwest of the lake is mesic and monsoon forest, degraded by swidden agriculture. It is shallow and gently sloping on the southern end, between three and four meters deep. The northern end of the lake has a sharp drop from shore to about five meters.

Sedges (Eleocharis sp.) and Typha are common along the banks, forming extensive areas in the shallow southern portions of the basin.

Two minor basins adjoin Lake Petenxil. A karst sinkhole 150 m in diameter lies to the northwest, and a separated southeastern arm probably is joined to the major basin during periods of high water.

Lake Petenxil is palynologically the best known site in the

Maya Lowlands (Cowgill and Hutchinson 1966). A series of 22 sediment samples were collected in 1975 from the mud water interface with a

Davis corer. The sample collection was in the form of two lake transects at approximately right angles to each other. The top two centimeters of sediment, representing about 30 years of deposition (data 14 from C dated sequence, Cowgill and Hutchinson 1966) were collected.

Pollen spectra from the two small basins are similar to that of Lake

Petenxil, indicating that, at least in this case, lake sixe is not a critical factor when small basins are used. Terminalia pollen and

Moraceae are the common local taxa, while extralocal pine pollen is the most common type found in the samples. Other local trees represented are Celtis, Bursera Haematoxylon among others (Table 17). Non- arboreal pollen is represented by grasses, composites, Cheno-Ams and others. The aspect of the spectrum is less weedy than that of the larger lake to the east, which may indicate a true correlation between 80 Table 17. Pollen percentages of 17 statistically significant taxa, Lake Petenxil, Peten.

Taxon P-01 P-02 P-03 P-04 P-05 P-06 P-07 P-08 P-09 P-10 P-11 P-12 P-13 P-14 P-15 P-16 P-17 P-18 P-19 P-20 P-21 P-22 AP

Acacia sp. 4.0 1.0 0.5 1. 0 o 0.5 0.5 1.0 0.5 1. 0 o o 5.0 0.5 4. 0 3.5 1.5 1.5 2.5 1.5 0.5 6.5 Bursera simaruba o 0.5 4.o 2.0 1.6 o o 4.0 4.5 6. 0 1. 0 0.5 2.5 0 2.0 2.5 1.5 0.5 3.5 1.0 0.5 o Celtis spp. 6.0 3.0 3.5 1.0 o 1. 0 o 4 •5 5.5 7.5 4.5 o o 6.5 3.0 10.5 3.5 0.5 10.5 4. 0 5.5 2.0 Haematoxylon spp. 1.0 0.5 0 0.5 0.5 0 0 6.0 0.5 4 • 5 0.5 0.5 2.5 0 0.5 6.0 0.5 0.5 3.5 3. 0 o 8.5 Manilkara achras 2.0 0.5 5.0 1.5 0 0 0.5 1.0 4.5 3.5 2.0 0 3.5 0.5 1.0 o o 0.5 1.5 2.5 0.5 0.5 Moraceae 18.5 24.0 41.5 32.5 61.0 27.5 31.5 32.5 21.5 40.5 29.5 26.0 24.6 41.0 27.0 40.0 51.0 19.5 35.0 28.0 31.0 37.0 Quercus oleoides 10.5 12.0 5.5 2.5 8.5 4.5 3. 0 2.0 5.5 3 •5 7.5 7.0 10.5 4.5 6. 0 4. 0 5.5 12.5 7.5 13.5 10.0 8.5 Terminalia - type 21.5 26.0 19.7 29.5 16.5 36.0 29.5 24.5 20.0 23.0 14.5 21.0 21.5 16.0 21.5 12.0 13.1 40.0 10.5 14.5 16.5 12.0 Zanthoxylum o 4. 0 0.5 o 1.5 0.5 o 1.0 10.5 6. 0 o 0.5 0.5 6.5 o o o o 7. 0 1.0 4.5 1.0

NAP

Cheno -Am 4.0 2.5 0.5 1.5 o o o 3.5 4. 0 o o o 4. 0 o 5.5 3.5 2.5 0.5 2.5 2.5 0.5 Compositae, high spine 3.0 4.5 3.0 6.0 1.5 10.5 4.5 2.5 6.5 0.5 1.5 6.5 10.5 8.5 4.0 4.0 6.0 2.5 0.5 6. 0 7.5 2.5 Compositae, low spine 1.5 3.0 6.0 3.5 1.5 7. 0 6.5 3.5 2.5 1.5 2.0 5.5 3.5 6.5 3.5 3.5 4 .5 6. 0 1.5 2.5 1.5 6.5 Cyperaceae 24.6 4.0 7.0 4.5 1.0 2.0 10.0 5.5 4,5 o 24.2 21.0 9.5 4.5 6.5 o 0.5 4. 0 4.5 10.5 10.5 4.5 Euphorbiaceae o 3.5 o 1. 0 o o 0.5 2.0 4.0 0 0 0.5 0 0 1.5 3.0 6.0 5.0 3.5 1.5 4.0 1. 0 Graminae 3.5 7.5 2.5 9.5 4.5 10.5 13.0 16.0 4. 0 3. 0 12.0 10.5 0 10.5 2.5 3.5 0.5 6.5 2.5 5. 0 3.0 4.0 Senecio-type 0 2.5 1.5 3.5 2.0 o o 0.5 4.5 1. 0 0.5 0 2.0 0 3.5 4.0 2.0 0 3.5 2.0 3.5 4.5 Zea Mays 0 1.0 0 + + 0.5 0.5 0.5 0.5 0.5 + 0.5 0 1.0 2.0 0 1.0 0 1.0 0.5 0.5 1. 0 81 disturbance in a basin's watershed and percentage of weed pollen in a lake core.

Sediment Network Peten-3, Lake Eckixil, Peten, Guatemala

Lake Eckixil (sometimes called Quexil) is a large, 4 x 1 kilometer basin in a limestone substrate. Its east-west trend parallels the axis of lakes Peten Itza and Petenxil, all of which are determined by the orientation of folding in the central Peten. The eastern and

western shores of Eckixil meet gentle slopes, while to the north and

south lie rolling hills, some of which have 200 meters relief.

The hills surrounding Lake Eckixil are covered with a mosaic of regenerating vegetation, with smaller patches of natural vegetation.

The predominant climax type is monsoon forest with Bursera, Orbignya, and Coccoloba among the common trees. On steep slopes close to the water and in arroyos I found extensive zones covered with mesic forest.

Encinales are common 1.5 km to the south of Eckixil, but were not seen by me near the water's edge. These oak woodlands are extremely important to interpretation of the fossil record since they are close to the long core site (Vaughn 1976) and contain several savanna or grassland elements, but are different floristically from them. True savanna or grassland is uncommon, but a few small patches of Byrsonima or recently made pastures were encountered.

An arboreal swamp vegetation type, usually dominated by the bombacaceous tree Pachira aquatica, forms the western and southwestern borders of Lake Eckixil, while the rest of the shores are covered with sedges (Eleocharis) and cattail (Typha). Finca Eckixil, a large 82 landholding, lies to the northeast, while small huts cluster on the surrounding hillslopes. 14 The longest C dated paleoecological sequence in the Maya Low-

lands has come from cores taken in this lake, stretching into the sixth

millennium B.C. (Vaughn 1976). A series of sediment surface samples

were collected in Eckixil to serve as a modern analog for long cores.

A Davis corer was used to collect the samples at this site, and re-

sulted in more than the standard 2 cm of sediment being collected, due

to its construction. The size of Eckixil, a dinghy, and bad weather

precluded intensive sampling of the basin so two transects were taken

in the western quadrant of the basin. In addition, three short cores

were collected in the deepest (8.0 meter) portion of the transects.

The results of the pollen transects are similar in many re-

spects to the Lake Petenxil network. The impact of recent slash and

burn agriculture is greater on Eckixil, as shown by higher relative

frequencies of disturbance vegetation pollen (Table 18). Also, oak

pollen from the Encinal vegetation south of the lakes is more common

in Eckixil. Terminalia-type pollen is less important in Eckixil, replaced by the diporate Moraceae type. There is little difference in pollen spectra from deep sections of Eckixil and those taken in shallow, shore-side environments. The overall aspect of the pollen deposition in Lake Eckixil well reflects the lake's environs of forest, old-field, and encinal.

▪ •

0 ren N co LN- tr. oN -1- O N N N N IQ 0 N r•rn N 0 N VD 0 fc\ CO N \ I>< • • • • • • • • • • • • • • • • • H Pr\ -1- N N 0 N N reN Cs- 1-4 0 \ H

H 0 0 LA LA 0 00 LA 0 0 LA LA LA lA \ \ 0 • • • s • • • • • • • • • • • H1 1/40 N ‘,,0 0 CO re \- 0 H H

0 1.11O isn LC\ 0 0 0 LA LA 00 LA 0 0 0 0 0 H • • • • • • • • • • tA N ren \JD ren Ett- • N O N --1-

CT\ O 000 LA u^‘ LA LA 0 0 0 LA LA 0 0 LA 0 • • • • • . • • • • • • • • • N H n.0 0 Co LN- H tr1 H

CO O 0 0 LA LA LA LA o o tr o 111 0 \ • S • • • • • • • • • • • • • • C'•• rel Pr\ 0 N N Pr\ -I' n.0 LA N • Pr\

IN 0 0 0 0 LC\ LA LA 0 LA 0 LA LA LA 0 0 LA • • • • • • • • . • • . • • . • • H H- rteN N 03 re\ --I- H ND NO H LA N 0 ri tteN H H

O 0 LA 0 OtAtA tAtA 0000 0 LAS 0 0 • • • • • • • • • HO H V.\ • N ren n.0 H H

0 0 LA 0 Ifs\ LA IA 0 u-n 0 LA LA LA 0 0 0 LA • • • • • • IS •IA . • • • • • • r(1ern --1- re\ 1`41 r-4 LN N Co r-I 0 \

O LA 0 LA LA LA 0 LA LA 0 0 0 \ LA LA LA 4- • • • • • • • • • • • • • • • • H n0 N co N 0 ern IA 0 n..o CTn 0

IN IN 0 0 0 0 0 0 IA LA II\ 0 LA LA 0 0 11-n • • • • • • • • • • • • • • • • 0 00 N.\ 0 H LI\ N 00 0 H 0 k.0 N t•c\ r--1 H

O tA 0 0 If 1 IA Lrn 0 LA 0 0 LA LA LA 0 0 LA • • • • • • • IS • • • • • • • N 0 LAN ern LA 0 1/4.0 N H re\ N H H

Lrn LA LA LA LA u**n LA 0 V\ 0 LA trn o tA LA 0 0 • • • • • • • • • • • • • • • • n O tr \ 0 0 rs*— 0 %.S) er lf \ H rml ('J H H

ce 1:11 tO ti) an R.1 DS a) 0 0 CO TS 4 .r-1 t 0 0 04' • a • o cis a) I P., - ,-.1 R, r-I r-1 ai S P-1 CO S:2-n P-.-. CIS 0 -r-I r-I CO CO >4 5-4 el 0 CZ ai CO rg O $4 CO 4-) .. 0 0 0 •1-1 0 -ri 0 r-I 0 -ri 4 C.) CO -P E 4-1 E -P ce H Q) 0 42) f-i 0 O g C) g CIS Z a) ai <4 Pl 0 = Z 0? E-1 el ‘<414 I <444 1 84 Sediment Network Peten-3, Aguada "Libertad," Peten, Guatemala

Aguada Libertad is a small 0.5 hectare pond 1.5 km north of La

Libertad, Peten. Since it contained about one half of its water

capacity during the height of the dry season of 1975, the aguada is

probably permanent. This aguada was chosen because of its permanence,

proximity to extensive savanna, and dry monsoon forest, and the lack

of local encinales. This was to attempt to factor out the different

inputs of oak woodland and savanna.

The soil surrounding the aguada was red oxidized clay of a

savanna type, and patches of grassland were common. Cattle and humans

had trampled the muds of the aguada on the south and west sides and

these areas were avoided in sampling. The water was not deep, 0.5-1.0

meters was the average, but probably fills to 3.0 meters during the

rainy season. These factors reduce its comparability to larger basins,

but its environs outweigh problems of basin size.

The results of pollen analysis of three samples from Libertad

are similar to those of Lake Eckixil and contain almost as much oak

pollen as samples from Petenxil and Eckixil (Table 19). There is less

disturbance pollen in Aguada Libertad, except for grass pollen, which

probably is of grassland origin.

Sediment Network Peten-4, Lake Oquevix, Peten, Guatemala

Lake Oquevix is one of two lake basins which lie to the south

of the main lake axis (Fig. 2). The lake is of essentially east-west

orientation and is slightly over 4.0 meters deep on the western end of 85 Table 19. Pollen percentages, of 17 statistically significant taxa, Aguada Libertad, Peten.

Taxon 1 2 3

_AP

Acacia spp. 1.0 0.5 2.0 1.17 Bursera simaruba 4. 0 2.0 1.0 2.33 Celtis spp. 5.0 4. 0 6. 0 5.00 Haematoxylon spp. 6.5 1.0 1.0 2.83 Manilkara achras 1.5 1.5 4.5 2.50 Moraceae 41.5 44.5 30.5 38-83 quercus oleoides 2.0 4.5 8.5 5.00 Terminalia-type 13.5 12.0 7.5 11.00 Zanthoxylum 2.0 4.5 0 2.16

NAP Cheno-Am 3.5 2.0 0 1.83 Compositae, high spine 1.5 7.5 7.5 5.50 Compositae, low spine 5. 0 0.5 6.5 4. 00 Cyperaceae 5. 0 4.0 4.5 4.5o Euphorbiaceae 2.5 2.5 7.0 4.00 Graminae 16.0 15.0 11.0 14.00 Senecio-type 0.5 1.5 3.5 1.83 Zea Mays o 0 0.5 0.17 86 the basin (Fig. 7). Twenty-one samples were taken with a Davis corer from a fiberglass dinghy. Unfortunately, pollen was poorly preserved, only four samples having countable pollen. Upon arrival in the laboratory, many sample bags were distended from gas formation, indicating the presence of degrading bacteria due to non-treatment with phenol.

The pollen of the four samples Oq 4, 5, 10, and 20 is in poor condition, comparable to that found in southwestern archaeological sites.

The results of the pollen transects are similar to the encinal and grassland soil pollen sample sites. Grass pollen dominates the mud-water interface samples, ranging from 55-70%. Pine, oak, and

Terminalia-type follow in descending amounts. High spine Compositae,

Acacia, and Haematoxylon were present in small amounts (Table 20).

There is a good match between Lake Oquevix pollen and the adjacent grassland soil pollen samples, indicating that savanna and grasslands are heralded in modern pollen samples by grass pollen. 87

OQUEVIX 1 PLOT- 500M

gUAZUMA VLMIFOLIA

MIXED SCRUB Lli=1.1tA° .J 14 SCALE (M) DEPTH CONTOUR INTERVAL- 1/2 METER

Oquevix, Peten. Figure 7. Pollen sample network, Lake 88 Table 20. Pollen percentages, of 17 statistically significant taxa, Lake Oquevix, Peten.

Taxon 0q-4 0q-5 Oq-10 0q-20 -AP Acacia spp. 3.0 4.o 6.5 2.5 4•o Bursera simaruba 0 0.5 1.0 0 0.37 Celtis spp. o o 1.0 o 0.25 Haematoxylon spp. 0.5 2.0 1.0 0 0.87 Manilkara achras 0 0 1.5 2.5 1.00 Moraceae 0 0.5 1.0 4.0 1.25 Quercus oleoides 8.5 6.5 1.0 2.5 6.87

Terminalia - type 5.5 12.5 4.5 6.5 7.25 Zanthoxylum 0

NAP

Cheno-Am 0 0 0.5 0 0.13 Compositae, high spine 3.5 10.5 2.5 6.5 5.75 Compositae, low spine 0 0 3.5 4.5 2.00 Cyperaceae 3.5 0.5 4.5 2.0 2.63 Euphorbiaceae 2.5 0.5 3.5 0 .5 1.75 Graminae 70.5 62.5 55.0 67.0 65.75 Senecio-type 2.5 0 4.5 1.0 2.00 Zea Mays 0 0 0 0.5 0.13 CHAPTER 6

ETHNOECOLOGY

Introduction

Understanding man's role in tropical landscape modification is

needed before an attempt can be made to describe past vegetation change.

Although climatic variance was probably the main cause of vegetation

change during the Pleistocene and early Holocene, the spread of Meso-

american agriculture into the lowlands may have overshadowed it during

the later Holocene. Since the fossil pollen record reflects only the

change in vegetation composition, we cannot be certain if climate,

plant competition, man, or a combination of the three accounts for most

of the variance in the pollen profile. There are several ways of

"factoring out" the human element from a pollen diagram, some dealing

with internal patterning in the pollen data, some with analogy, and

some with cross-dating. These will be described in more detail in the

final section of this paper. However, I shall discuss in some detail

the possible methods which the Maya may have used to alter their

natural surroundings. These may serve as a complement to the natural

landscapes described in Chapter 3.

The natural landscape is coming under increasing alteration by modern Maya colonists. The milpa, or slash-and-burn agricultural system, is claiming thousands of hectares of climax moist forest. In

this chapter, I outline the subsistence base of the modern and ancient

89 90 Maya as ethnographers visualize it. The interplay of the natural and

artificial systems as reflected in the pollen rain is exceedingly com-

plex, so a preliminary familiarity with the systems is desirable.

The modern Maya agriculture of the Lake Peten region has been

studied twice in recent years, giving two somewhat divergent views of

the subsistence base of the modern Maya. The diet is largely carbo-

hydrate, Indian maize being the stable crop. Protein is obtained from

beans, game, and some local fish during late summer.

Subsistence is based upon the milpa system, which is admirably

adapted to tropical environments, avoiding many factors which limit

more intensive systems. As an introduction, I describe those limiting

factors in the natural environment which affect crop yield in the Peten.

I then describe the milpa and other subsistence options which function within those limits. A discussion of hypothetical prehistoric systems follows. In conclusion, I present a quantified systems model of prehistoric agriculture using as variables the vegetation and subsistence data.

The Limits to Agricultural Growth

Many publications describing prehistoric Mayan agriculture mention geographic limiting factors (Wilken 1971; Puleston and Puleston

1971; Turner 1974). The specific biological and edaphic mechanisms which affect agricultural productivity remain enigmatic. To modern tropical horticulturalists, these factors may mean the difference between success and failure. There is considerable literature concerning 91 tropical pests, soils, and crops which should be studied by those de- siring to reconstruct prehistoric Mayan subsistence.

The first factor limiting agricultural success in tropical

America is biological: organisms which compete with man for his crops,

a constant, low-level destructive agent with occasional dramatic popu-

lation surges. Although tropical forests have the highest diversity

of herbivarious insects and plant diseases of any terrestrial vegeta-

tion type, their potential damage to crops has been little studied.

Since pests are adapted to many different niches, there is a high

potential for damage to cultigens from local or exotic species "pre-

adapted" to a given cultivated host plant. Tropical woody plants have

come into equilibrium with such pests through long association, but

defense systems for imported seed crops, as planted today, are severely limited (Janzen 1970).

To protect itself from pests, a plant has two general defenses.

A plant may manufacture substances and structures harmful to larger

herbivores, insects and parasites (internal defense), outside factors

reduce pest populations (external defense). The first internal defense

is a series of secondary compounds, such as oxalic acid crystals, latex,

and strychnine that are toxic to herbivores, manufactured by the plant

and dispersed throughout its tissues. The second internal defense is

the generation of thorns, spikes, spines, and hooked trichomes --

adaptations against large herbivores, now a minor component of the

tropical forest biome. The third is the use of a symbiotic animal to repel herbivores, as is done by the biting ants (Pseudomyrex spp.) inhabiting the bullhorn acacia (Acacia cornigera) (Janzen 1966). 92

External plant defenses consist of cyclic climatic variation, distance between individuals of a species, and predation upon herbivore populations. Rain forests have the least amount of climatic variation of all vegetation types, so winterkill and droughtkill of pests are

almost absent. Only in northern Peten, toward the more arid Yucatan

Peninsula, is there a dry season extensive enough to limit herbivore populations under a closed canopy. The lack of limiting climatic cycles which allows herbivores to rapidly adapt to culturally caused variations in their food supply, results in a heavier density of parasites, herbivores and diseases at critical times. The artificial establishment of regular cycles in pest populations may restrict predators.

Carnivorous animals are specifically adapted to the habits and life cycles of their prey. For example, temperate zone predators are adapted to winterkill of their prey by either seeking alternate food sources, or surviving lean times as egg or pupae, but tropical predators have no such adaptation to cyclic variations. If short-lived pests become attuned to crop cycles, predators cannot cope with the population explosions since they must stay at a level that will survive low prey frequency. The net result is little biological check on pests, making this factor potentially dangerous to continuous cropping.

Natural internal and external defenses are limited or nonexistent for most seed crop species, such as maize or beans. The best ecological protection available for such cultigens in tropical regions is diffuse distribution of individuals throughout the plant community.

This tactic is clearly impossible with cereal cultivation since harvest of seed from widespread individual plants is unproductive. The modern 93 milpero uses this general principle by separating his plots in space and time so that localized outbreaks of pests such as corn borers or

leaf cutting ants (Cowgill 1962) will not become a pestilence. The low

modern population density allows the milpero to choose secondary forest

for clearing. Advanced secondary forests have lost those pests adapted

to early successional plants, such as grasses and forbs. However,

during past peak population periods, fields may have had short or no

fallow cycles, and thus have been highly susceptible to damage due to

clearing of areas still having early successional pests and to pest

innoculum from adjacent croplands. Any type of intensive cultivation

using seed crops would have been prone to damage from rapidly expanding

insect and disease populations that could not be handled by the secular technology available to the ancient Maya.

Another factor limiting prehistoric agriculture is the loss of

plant nutrients through solution weathering. A combination of high

diurnal temperatures, heavy rainfall and calcareous parent material

characteristic of the Maya Lowlands allows intense leaching of soil

which forms an impoverished upper soil. However, the great biomass and

species diversity of tropical forest communities indicate most neces-

sary minerals are concentrated in the vegetation itself instead of the

soil (Billings 1970). In undisturbed forest, leaf litter and dead vegetation return stored nutrients to the upper soil horizons where

they may be released for building new plant tissue by decomposers in the soil.

In modern milpa agriculture, cutting, burning, and weeding release stored minerals to the soil as soluble ash, highly susceptible 94 to solution transport. For a short time, the upper soil of the milpa is saturated with organics and minerals, until heavy rains, unrestricted by canopy or leaf litter, carry them away through percolation. During this short, fertile period, shallow-rooted seed crop plants such as maize, beans and squash may be grown effectively. Decreasing fertility then necessitates a fallow period to increase the nutrients necessary for successful agriculture through reaccumulation by successional plants. This is a major reason that a short period of cropping, followed by a long fallow period, must have been used by the ancient Maya as well as their modern descendents, when raising seed crops in unal- tered upland plots. Any hypotheses describing extensive or intensive agricultural systems must take leaching into account, since tropical seed crop cultivation will eventually fail without providing from above ground these necessary nutrients, either as fertilizer or ash.

The only alternative to fallowing is the use of fertilizer which would necessitate composting, storage, and distribution of organic waste. Allowing loss through incomplete retrieval, leaching, and runoff, fertilizer exploitation would be unsuitable for stabilizing large tracts even without insect problems and erosion. However, this concentration of nutrients would be of value in establishing relatively small, highly fertile gardens that could contribute valuable crops. An important edaphic factor hindering effective total utiliza- tion of the rain forest for agriculture is erosion. Modern milpa and intensive systems of agriculture are extremely destructive of natural vegetation cover and leaf litter. Since rain forests have little humus, destruction of available cover results in the loss of water retention. 95 A cleared plot yields more water as runoff than an equal plot of

forest, and this water flows erratically as sheetwash with erosive

potential that may render the area useless for agriculture through

gullying and removal of fragile topsoil. Since soil is the result of

climatic and biological alteration of the substrate for thousands of

years, it cannot be quickly renewed. Topsoil loss would have had

disastrous and long-term effects upon Maya civilization. Prehistoric

erosion is easily detected in lake basin sediments.

Other limiting factors, working on a smaller scale than those

previously discussed, are root competition, shading, drought and me-

chanical damage due to wind and hail. Those models which stress tree

and shrub crops must deal with the effect of large plants which cover much space with roots and photosynthetic surfaces. Many tropical tree

genera have large, shallow root networks; others have incredibly

small, fibrous root systems; and several have roots which seek great depths for nutrients. Those trees with shallow root systems may compete with herbaceous crops, stunting those which need copious amounts of water and nutrients. A peripheral zone, varying in size according to the tree, is impractical for raising many crops. Plants such as

Manihot esculenta, Xanthosoma violaceum, or Vanilla fragrans may be used in this zone since they make smaller inroads on the nutrient balance than do seed crops. Root competition from secondary weeds, such as the herbaceous members of the grass, sunflower, legume, and amaranth families, rapidly becomes limiting to crop plants similar in niche, while little affecting woody crops. Cutting of weeds with a machete, as is done in the Peten today, does not destroy the rhizomes 96 and scatters the seed. Simple manual weeding or companion planting

will alleviate the problem, but will require a great outlay of time.

Substitution of useful crops into the weed niche will reduce weed in-

festation, as well as reduce soil temperature, allowing survival of much-needed soil organisms.

Shade from large trees or shrubs may affect crops planted be-

neath them. Many seed crops are intolerant of shade and suffer at

light intensities under 5000 foot-candles. Other crops seem to flour-

ish in shade and exist at less than 1000 foot-candles for a twelve-

hour day (Graf 1957). Examples of this type are the large-leaved root

crops and vanilla. An equilibrium between overstory and understory

crops needs to be established to derive maximum productivity from a mixed system.

Physical damage to crops by wind and hail are fairly uncommon

in the Peten and cannot be counteracted by the agricultural system used.

Windbreaks may help, but meteorological catastrophy may have been a

deciding factor in those parts of the Maya Lowlands frequented by hurri-

canes and violent thunderstorms.

Failure of the summer monsoon to arrive on time or insufficient

cient quantity has been dealt with by others (Morley 1950; Thompson

1964; etc.) and will be mentioned only in passing. Drought will affect

those crops which are most dependent upon sufficient precipitation, such as the larger, leafy seed crops -- maize, squash, and beans -- while tree and root crops will have a reserve to draw upon in times of water deficit. All will fail in periods of protracted drought 97 (Anonymous 1975). The agriculturalist had little recourse but to irrigation (as practiced in the Motagua Valley) and his gods.

The Agricultural Systems

Now that those general biotic, edaphic, and climatic factors limiting agriculture have been discussed, I would like to describe

Mayan subsistence, both ethnographic and hypothetical. Each has, or may have had, its effects upon the carrying capacity of the land and its potential productivity within the limits previously discussed.

Modern Maya agriculture of the Lake Peten region has been studied twice in recent years, giving two somewhat divergent views of

the subsistence base of the modern Maya (Cowgill 1962; Reina 1967).

The modern Maya diet is largely carbohydrate, Indian maize being the

staple crop. Protein is obtained from beans and game, with occasional

local fish added during the late summer (Reina 1967).

System 1: Milpa Agriculture

The slash-and-burn agricultural system practiced by the modern

Maya has been much discussed (Morley 1950; Thompson 1964; Cowgill 1962;

Coe 1966; Reina 1967; Culbert 1974). I shall present a quantified sum-

mary of the inputs and products of the Maya agricultural system derived

mainly from local data collected by Cowgill (1962) and Reina (1967)

supplemented by the data from northern Yucatan (Emerson and Kempton

1935) and other portions of the lowland tropics.

The milpero's family averages 5.78 persons. It may include

sons-in-law and aged relatives in addition to his nuclear family (Cow-

gill 1962). Either his son or son-in-law may help him in the milpa 98 throughout its cycle, aided at need by the rest of the household, usually during weeding and harvest (Reina 1967). The amount of land

cared for by the milpero and his helper(s) averages 5.28 hectares

(Cowgill 1962), which may be divided into several plots in slightly

different habitats (Reina 1967). The use of multiple plots increases agricultural stability due to past quarantine effects and insulation

from yearly climatic variation. If one field is affected aversely,

there is a chance that the other fields will provide the family with sufficient food for the year (Reina 1967). Labor input into the system has yet to be quantified in the Peten, so figures for nearby Yucatan will be used (Emerson and Kempton 1935) (Fig. 8). Locating a milpa for clearing usually takes but a single day and may occur at any time of the year. Cutting of the forest is the longest chore associated with milpa agriculture, taking 75 days' time, usually in late fall and spring. Burning, lasting but one day, occurs in April or May, in anticipation of the summer rains, and is immediately followed by planting the crops with the aid of a dibble stick, which takes about 14 days.

Weeding the fields during the summer rains, using 14 days' time, is followed by harvest of the maize crop in late fall or early winter, which takes 36 days' labor. The result is 147 days of activity for the milpero and his helper, leaving 218 days not devoted to milpa activity each year.

Table 21 lists the productivity of many of the known cultivated indigenous crops. The data may be little more than crude estimates of reality, but as far as I know, they are all we have. Cowgill (1961) 99

▪ • 3 ▪▪ • •

100

•.to g 0 .4_, 00000 000 0 0 0 0 0 0 00000 00 0 0 g .1,) ;J 00000 0 0 0 0 0 G) 000 0 0 0 0 cp 0 0 0 00K100,41t41 00003 oc) ts,n oo co co oo cd • E 1--10$ 11111 III O 0 1 IIII1 H 0 0 000 0§0 0 -t-1 C'- P -P 00000 000 0 -ri .0-P 8 8 4It 8 8 cd U1 0000V1 000 0 il\ 0 v 00 O hip0 --i—ice\r—I -d-(-4-d- —t- V -ri 0 Cd r.-1 O rIcri P .....- ri O 0 O -P â O CO 0 -ri a) .2' ti)4ggg E*4-1 0 4M1-DI-D1-D 1-1 4-) nD -1-- • O W 0.• rel 0 t ct t 0 - O" 11\ 0

0 O H 0 Id 6.0 F-4 r-1 -0 0 Z O 0 CO . 44 44 0 ct0ZOM O "0 0 .0 4= ===Z C., 0 0 0 1 0 0 I 1111A. 4-) gg >- 0 p ...i 00Z1-1 NN12,7 c, "4rii,ifttt a. -H ."0• E 0 44 •ri 0 0 cri P 0 -r1 0 H ri ri r-11-1 -rf •F1 •r4 •,-1 -1-1 G) G) 4•-e 00 0 O 0 000 0 0 -p 0 0 0 -.1 H ' cd cd cd A A cd rCI 0 0• "0 1:1 4-) .4 O 0 O cd rl:i .0 0 CD 0 CO N- -..i -.I Z 0 0 cd 0 0 0 qd O -0 0 • 0.• ...4 H H 0 a) rl ri -1-1 4-1 .1-1 0 O 0 r--1 .0 cd O 000 0 01 cdad g; ;. r-lr-i 0/ 000 -p -0 P c001 O 0 0 0 14 -.4 0 0 . 45 45 45 L 0 bD -P 0 0 • 0 0 113 -ri V•00 • 4) rb '0 0 -P O .0 O -P004-) 4) 4-) 4-> 0 2 g /-4 4-) 4., .1-, 4-) ri .0 U) a) gl- O 0-.1,40 O 00 0000000• OC 0 -1-14-4H .nH-r4H-ri.r-i.ri.,r1-r-1-, bD 1)-1 ,-1 qd P1.0P1-.f0 O 00FidSra,1.4000000i-t 4-) 171 =PG:1Z XZZAMAZXXZZ=P H 4-> g 0 -ri "CS 1-n rQ N N N O tO g ‹ N Q Cr, uh Ch rch c0 N- U1 0 4- c0 Ch ▪ $.4 'd N o n.0 Q 0 r-i H Lth 40) 0 tch c0 n ri H D 4- o Oh 0 N- Lch o v0 H 4- c\J c15 .ri o t, ,0 HNNO0 --1- N- OhWHH 4)4.) -ri Nr4h4- N Cd 010 0 H H CU H 0 0 Cd 0 • • Cd 0 4-) a) ta 0 0 0 0 4 o • 0 E 100 g Cd 4-) -P H 0 O P. (0 Cd O ri O OH 0 O 00 cd Cd 0 Cd C-) 0 g ril E • ;., Cd cd cd cd cd cd o -ri = -p 4-) 1-1 • -ri 4-) ;-n 4 ,--1-.1 cd o Cd 0.n o o o ,0.0 cd cis 0 al cd 0 ai 0 cd ri Cd 0 EEEEEEEE01-1 -0 .c:41 99999999° O .0 Cd cd cd ”-1 4-) 4 § -ri o O o 0 4400 C) O cd Cd 01 cd 0 0 gigigiqgigigigig g .0 ri 1:14 000 (-QNCILNINNNN o 6 >4 bD n.0 -ri J.J (T\ cI4 H g P 0 0 O 0 00 .... cd fa., cd 4-e Cd Cd (OH 0 OD 0 H H N tr1 0 0 3 0 ri 0 0 I., -P -ri 0 -P H g .1:2 0 00 r-i •rii-D 00 .., 0 ri 1>z (.) 00 Cd bL 04 0 od 0 cd 1-, Cd n 10 cd 8 O g 0 e c..) r—i .e.,-1 0 rti 1.:)) 741 - -1 -e2I :1 -1 g :',4 ;1'is 10 • 5., 0 P cd0 cd cd cd 0 5-1 H 4 -,401010101 .0 a) r-1 0 Cd C13' g) g 1 0 fX ›"I C.) = X ›"I r•LI P.I C..) 4:4N Z Z Z .-1C/2 0,-irtl tn p., <4 0 A, N 101 lists the nutritional makeup of several indigenous cultigens, including the interesting statement that 285 kilograms (628 pounds) of maize or

658 kilograms (1450 pounds) of manioc will support a human being for one year. Obviously, these intake figures imply additional nutritional input from other crops. These data are of prime importance to agricultural reconstructions since they determine the number of people that may be supported by a given plot of land.

After harvest, two options are available. The plot may be left fallow, or it may be planted for a second year, but usually no more.

The mean fallow time for one year of cropping is 2.5 years, and for two years' cropping, five years (Cowgill 1962). I believe that a conserva- tive estimate of stable, long-term agriculture would be four or more years of fallow for every year of cultivation. A milpa under cultivation normally produces tremendous amounts of Zea pollen, little of which, however, gets far from the milpa due to the rapid fall velocity of the large Zea grains (Raynor, Ogden and Hayes 1972).

Pollen productivity of modern Maya agriculture is presently unknown and is needed to derive response functions in the modern Peten pollen rain. To obtain data relevant to this problem, I collected pollen from both newly planted and abandoned milpas.

The collection sites (VT-5A-E) were located 1.5 km southeast of

Santa Elena, Peten, originally of monsoon and marginal forest; they now are a mosaic of regenerating brush lands in various stages of succession. Several pollen collections were made along transects in this locality which crossed a planted cornfield of the "yaxkin" type 102

(Reina 1967), abandoned milpas with herbaceous weeds and root sprouts, a newly burned plot and monsoon forest.

Site VT-5A, a patch of forest that stands uncleared: It

appeared to be old secondary growth, due to the general absence of

epiphytes, small bole diameters, and presence of transgressive trees.

It was similar to monsoon forest (Sites SL-6 and SL-7), but not so tall

and more open. A 100-m line transect run in this formation was of

little use since most of the plants were unknown to me at the time the

transecwas taken. Total tree cover was 70%, with Graminae covering

6.5%, Ipomoea, a large scrambler, covering 2.5%, and miscellaneous herbs covering ca. 21%.

Site VT-53 an area of forest burned a few days before the

study was innaugurated: Three soil samples were taken here to deter-

mine the effectiveness of the milpa system in obliterating pollen on

soil during the burn stage. Virtually no pollen was found in the

samples from this site. Therefore, pollen found on the soil surface

of a milpa has been deposited since the field was cleared and burned.

Site VT-5C, a planted yAxkin cornfield (Reina 1967) with

stunted, chlorotic, but mature maize which was planted in November

1972: The field occupied a small knoll of calcareous rubble, an out-

lier of the weathered hills south of Lake Peten (see Chapter 1). The

soil was black, sticky and clay-filled but drainage was rapid. No

forest trees were spared in making the clearing, and the small, dead

.stumps in the field indicated that the original forest was probably

secondary. Burned "bullhorns" of Acacia spadicigera found in the

rubble support this view. 103

I obtained permission, to sample the field from the local family.

Accompanied by dogs and youngsters, I collected my vegetation and pollen data. Zea, the only crop in the field, had a coverage value of

22% (line transect; Cain and Castro 1959), an unexpectedly dense stand for milpa. The field appeared not to have been weeded, for young

Graminae (12% cover), Compositae (high-spine type, 7%), Ipomoea (ca.

), Mimosa (2%), and manyunknown herbs (29% cover) made up much of the ground cover.

Ten pollen samples were taken in the field by the transect method. Since I knew the age of the field (4 months) from informants' 2 data, I collected pollen from a series of 25.2 cm rectangles placed upon the ground to determine the absolute pollen influx rates for the field (Davis 1974). Clearing and burning a milpa destroys most of the soil pollen on the milpa, in effect providing a "clean slate" for later accumulation of pollen which can then be expressed in terms of grains/ 2 cm /month.

Site VT-5D, a standard milpa, cleared in the spring of 1970, planted 1970, 1971: The field, which occupies rolling alluvial terrain, is densely covered with tall, herbaceous growth varying from

1.5-3 m high. Due to the difficulty of penetrating this brush, and the complexity of the jumbled growth, only one line transect was taken.

Grass, a visible dominant, covered but 18% of the line transect. clambering Ipomoea and other Convolvulaceae had approximately 7% cover.

The rest were unidentified members of the Solanaceae (Solanum), Com- positae (ca. 12%), Passifloraceae, and Cucurbitaceae (?). Only five 104

pollen samples were taken at this site due to the difficulty in obtaining accurately measured soil debris.

VT-5E, a site abandoned for 2.5 years consisting of lush, early

secondary growth: This site has been described elsewhere (Wiseman

1974), as I believed that it was the most representative site for

modern pollen analysis of milpa agriculture. Except along pathways,

herb cover was fairly low, arount 15% total. These consisted of large,

semi-woody grasses (ca. 6% cover), a few Compositae, most along path-

ways (7%) and many unknowns. The canopy cover was quite heavy (ca.

75%), and was composed of secondary woody perennials with a strong in-

fusion of Leguminosae. The canopy, a jumbled mass of vines and trees, varied from two to four meters high.

Succession in the Milpa System. Figure 9 summarizes my cover

estimates for pollen-producing herbs and total woody plants. When the

five agricultural sites are arrayed by the amount of time elapsed since

clearing, a secondary succession pattern becomes apparent. The culti-

gen, Zea, has a strong initial surge due to planting, care, weeding(?),

and ash nutrients, then rapidly subsides after harvest. Weeds increase at a slower rate, but persist longer, slowly to decrease in the second and third years. Secondary woody plants rise slowly, then with great rapidity until high coverage values are obtained, shading out the more shade-intolerant weeds.

Most of the producers of disturbance pollen (e.g., weed pollen) gain prominance one year after the field has been cleared, a short

"lag effect" in pollen indicators of agriculture. 105

35' -1A

ac - IA •

99 - 1A 0

-o V9-1A .0N IMS

o o o N. ,c) -zr

renop 4uswed . 106

Pollen Deposition in Milpa Zones. Results of pollen analyses of Site VT-5 nonetheless indicate a weed-dominated regime. Site VT-5-A and VT-5-E, which were covered by woody plants showed high percentages

(up to 75%) of herb pollen blown in from adjacent croplands, as well as a strong arboreal component (Table 22). These sites characterized by a herbaceous cover (VT-5-C, VT-5-D) were almost entirely characterized by herb pollen of grass, low spine Compositae, high spine Compositae, and Chenopodiaceae + Amaranthus.

Herb pollen is over-represented in the central Maya Lowlands, since it penetrates far into forests, while arboreal pollen is almost nonexistent in weedy zones. This fact is of critical importance to paleoecological reconstructions. Small plots of successional plants will palynologically overshadow large tracts of tropical forest.

The agriculturalist is not heralded by the pollen of his crops, but rather by the weeds that infest his fields and fallow plots.

System 2: Dooryard Gardens

The modern dooryard garden (classified as a semi-intensive system), as seen in Flores and San Benito, Peten, is a small, fenced enclosure, approximately 0.2 hectare in size. Plants raised in the gardens are flowers, cash and subsistance crops ranging from medicinal herbs such as apozote (Chenopodium ambrosioides), to crops such as chile (Capsicum annum, semi-arborescent plants (Yucca elephantipes), and fruit and shade trees (Cocos nucifera). Much household activity is carried out in the shade of the dooryard garden, and refuse is often used for mulching and fertilizer. Chickens and dogs seem to be allowed 107

Table 22. Pollen percentages of 17 statistically significant taxa, Site VT-5, Santa Elena, Peten.

Taxon VT VT VT VT VT VT 5-A 5-B 5-C 5-D 5-E 5-F AP

Acacia spp. 1.0 + 2.5 0 0 0.5 Bursera simaruba 2.0 + 0.0 0 0 1.5 Celtis spp. 1.0 0 1.0 0 0 1.0 Haematoxylon spp. 0.5 0 0.5 0 o 2.0 Manilkara achras 0.5 0 0.5 0 0.5 1.5 Moraceae 21.0 + 3.5 2.0 14.0 30.5 Quercus oleoides 0.5 0 11.0 6. 0 10.1 4. 0 Terminalia-type 17.0 + 6.0 4.0 10.5 21.5 Zanthoxylum 1.0 o o o o 1.5

NAP Cheno-Am 6.5 + 6. 0 7.5 2.0 3.5 Compositae, high spine 4.5 + 10.5 15.7 16.5 11.5 Compositae, low spine 10.5 + 24.5 25.0 12.0 9.5 Cyperaceae 2.5 + 6.5 4.5 2.0 1.5 Euphorbiaceae 1.0 + 2.5 1.5 0 0.5 Graminae 27.5 + 21.5 25.0 25.2 7.0 Senecio-type 3.5 + 4.5 5.5 6.5 2.5 Zea Mays 0 0 5.5 5.5 1.0 0.5 108

within the confines of the garden in February, probably since there are

no delicate young plants that could be molested. One house on the

Flores-Melchor road had a raised garden, like an isolated windowbox,

in a hollowed log patched with scrap lumber which was planted with

flowers. This device may have protected the plants from the ravages

of local domestic fauna. Dooryard gardens were also observed in the

savanna region of Peten, and the crops raised within them did not seem

to suffer because of the dense clay soil. Perhaps the intensive care

which garden plants receive negates, to a certain degree, the supposed poor edaphic conditions of red savanna soils.

System 3: Fuel Procurement

A little-studied facet of modern Maya subsistence concerns

energy. Although tiendas in Flores, San Benito, and even some of the

more remote villages, use gasoline, bottled gas and kerosene for cook-

ing and illumination, the fuel of the Peten is hardwood. Wood cut

from the forest is hauled to towns and hamlets on ancient pickup trucks

and burros. The types of woods used are unknown, but it is said that

the wood of the Brasil tree (Haematoxylon brasiletto?) is highly valued.

Diameter of the logs is smaller than those used in North American fire- places. A stack of firewood delivered to San Benito, Peten, had small logs from three to eight cm in diameter, and very few over 10 cm, although logs seen on a flatbed truck were of larger diameter. The small tienda and restaurante near Laguna Macanche had logs seven to 12 cm in diameter piled near it, doubtless for use in the more "commercial" kitchen. The small diameter and blade marks on log ends indicate that 109

one to three machete strokes were used to sever the log into fagots

varying from 0.6 to 2 m long. Probably this size is most efficient to

harvest and transport using modern tools.

The firewood use rate per capita is fairly rapid, since the

pile of wood at San Benito was reduced to approximately one-fourth its

original size (.375 cord) in three days, or a use rate of 0.1 cord/day.

If but two households used the 0.1 cord/day, as seemed to be the case,

then each family would use between 10 and 20 cords of hardwood per

year (or ,v3 cords/capita/yr). This implies a considerable selective

pressure on dense wood species with 2.5-12.5 cm boles. A field aban-

doned two and one-half years near Santa Elena, Peten, had softwood

trees such as Cecropia mexicana up to 12.5 cm in diameter, but no hard-

wood species had a DBH of over 2.5 cm (Wiseman 1974). Therefore,

abandoned milpas (acahuales) probably cannot support efficient wood

harvest for well over five years.

In fairly undisturbed situations, the density of trees with

bole diameters of 2.5-12.5 cm ranges from less than five plants/hectare

in open savanna, to 300 trees/hectare in oak woodlands, to well over

600 trees/hectare in old successional forest.

If the mean firewood log diameter is 7.5 cm, and the length of

usable wood from each tree is a little over 3 m, then it takes about

120 trees to yield a cord of firewood. In favorable situations, this results in five cords of cut logs per hectare of forest. Allowing five years as a minimum forest regeneration time, a sustained yield rate from the Peten forests would be around one cord/hectare/year. In an equilibrium model, three hectares of sustained yield land could supply 110

the energy requirements of each person. If an expansionist model were

used, in which regeneration is not assumed, 0.6 hectare could support a person's fuel requirements.

Although these estimates are crude, the implications are that

if the prehistoric Maya used grass or bush fallow systems (Sanders

1973), extensive terrace systems, or ridged field systems, the energy

resource base may have been depleted. Since I infer that the prehis-

toric Maya had a population density many times greater than at present,

the energy demands upon the tropical forest probably necessitated a

well-programmed fuel production and allotment system. This could have

been accomplished through several ways: wood-lot plantations, milpas

with a wood harvest during fallow times, or isolated wood crop plants grown within other agricultural systems.

System 4: Hunting

The larger animal population of the Maya lowlands is quite

diverse (Table 23). With the encroachment of a cash economy, hunting

is now a relatively minor component of Peten subsistence, but the milpero still utilizes game to supplement the products of his field.

Today hunting is also used to acquire cash by the trapping of parrots, macaws, and iguanas for export to the United States. An uninjured scarlet macaw brings 50 Quetzales (= $50) in the Peten.

The .22 calibre rifle seems to be the universal hunting arm, usually a rusty bolt-action with rudimentary iron sights. I did not see any larger calibre rifles, shotguns, or pistols, possibly due to restrictive gun control laws. Hunters usually seem to be young boys; 111

Table 23. Mesofauna inhabiting the Maya Lowlands. -- After Hall and Kelson (1959).

Order Common Name Scientific Name Animals Marsupial Opossum Diadelphis marsupialis Four-eyed opossum Philander opossum Primate Howler monkey Alouatta villosa Spider monkey Ateles geofroyi Edentate Tamandua Tamandua tetradactyla Armadillo Dasypus novemcinctus Lagomorph Forest rabbit Sylvilagus brasiliensis Rodent Yucatan squirrel Sciurus yucataensis Yucatan squirrel Sciurus deppeii Mexican porcupine Coendon mexicanus Spotted cavy Agouti paca Carnivore Gray fox ifrocyon, cinereo argenteus Cacomistle Bassariscus sumichrasti Racoon Procyon lotor Coati Nasua narica Kinkajou Potos flauus Jaguar Fens onca Mountain lion Felis concolor Ocelot Felis pardalis, Mar gay Felis wiedii Jaguarundi Felis yagouraroundi Pinneped Manatee Trichechus manatus Perissodactyl Tapir Tapirus baindii Collared peccary Tayassu tajacu White-lipped peccary Tayassu, pecani Artiodactyl White tail deer Odocoileus virginianus Brocket Mazama americana

Birds Cracidae Plain chachalaca Ortalis vetula Crested guan Penelope purpurascens Great curassow Crax rubra Meleagrinidae Ocellated turkey Agriocharis ocellata Psitacidae Scarlet macaw Ara macao 112 only one man was seen with a rifle. A secondary, but very effective weapon is the slingshot, wielded with deadly accuracy by small boys against the smaller fauna.

Reina's (1967) informants considered the milpa itself as a major game habitat, and regularly still hunted from its margins. The succulent, weedy growth, as well as the crop plants themselves, attract

the herbivores and their predators that find less sustenance in the climax forest. Hunting in the milpa decreases mesofaunal pest invasion through harvest and learned fear. Unfortunately, game harvest data are unavailable, but it would seem that an increase in milpa and acahual would increase the "edge-loving fauna," providing hunters, either ancient or modern, with more protein per unit land.

The pre-contact Maya lacked the firearm, the elastic slingshot, and the bow and arrow, but undoubtedly compensated for them by meticu- lous woodsmanship, game drive techniques and the proper ritual. Hunt- ing may be practiced at any time of year in almost any habitat, and this has not changed under the influence of the Spaniard.

Prehistoric Maya Subsistence

The prehistoric systems are much more difficult to understand than the modern, since observational data are lacking and secondary evidence is limited to ethnographic analogy, relict earthworks, housemound density and palynology. They are divided into two main groups, intensive and extensive. Extensive systems use much land, but little labor input; examples are hunting, gathering, and milpa farming, techniques used by the modern Maya. 113

Intensive agricultural systems may be considered as either biointensive or geointensive. A biointensive system relies upon increased efficiency of energy flow through a modified vegetationcompo- sition and structure to increase usable productivity. Smaller attention is given to the physical environment of the cultigens than to the biotic. Examples of biointensive systems are milpa, ramon cultivation

(Puleston 1969) and artificial rain forest (Wiseman 1973).

Geointensive systems alter the geomorphology of the agricultural plot to increase available moisture or to improve physical soil characteristics. Examples are irrigation and raised field networks.

Since they leave obvious remains, archaeologists have stressed geointensive agricultural methods in reconstructions of Maya subsistence.

The disparity between high past population density and low present agricultural productivity leads me to hypothesize that adoption of some unknown intensive methods.

Various authors have estimated carrying capacities for the Maya

Lowlands based on either archaeological or agricultural data. They 2 range from a minimum of 28 persons per km for the entire Lowlands, to 2 site residential estimates of 900 persons per km . Table 24 presents a cross-section of such estimates. The data bases vary widely, as do the estimates. Therefore, I will select Willey et al. (1965) as a fair non-agricultural estimate for population density over the entire Maya

Lowlands, and Puleston's (1974) estimates from an atypically dense zone in northeastern Peten as a potential maximum for use in model- building, to be described later.

Table 24. Carrying capacity estimates for the Maya Lowlands.

2 People per km max min Remarks

200 40 Arch. rural Willey et al. 1965

700 600 Arch. Tikal Haviland 1970

900 300 Arch. Tikal site-intersite Puleston 1974 .39 Arch. Tikal Post-Classic Puleston 1974 85 28 Agri. Vogeler 1974

65 28 Agri. recal. Vogeler Turner 1974

77 38 Agri. Cowgill 1962 150 Agri. intense, fertilized Turner 1974

310 Agri. continuous manioc Cowgill 1961

827 1561 Agri. chinampa Palerm and Wolf 1957

Arch. = Archaeological demographic estimate Agri. = Agricultural demographic estimate (based upon crop productivity) 11 5 System 5: Intensive Milpa

The intensive milpa, a term to describe a more productive swidden system, is a biointensive model. Instead of the extensive,

" ecologically wasteful" modern milpa, this hypothetical construct avoids the limiting factors imposed upon it to produce up to three crops a year with reduced fallowing. The dry season limiting factor may be avoided by simply finding an area with few rainless months in which to plant a milpa, such as is found arount Poptun, Guatemala

(Culbert, personal communication 1976). Crop rotation (Wilken 1971) and companion planting have been hypothesized as fallow-reducing mechanisms. Perennial species, such as manioc planted with annuals, produce cash or subsistence crops during the fallow period, materially increasing the milpa's overall productivity. Fertilizing, mulching, hand-weeding, and other techniques may also intensify milpa technology to qualify it as a biointensive technique greatly increasing productivity. Since increased diversity and coverage insulate against pest and sheetwash problems,a hypothetical intensive milpa could have been more stable than the modern slash-and-burn system.

System 6: Artificial Rain Forest

The most ecologically efficient hypothesized biointensive agricultural system is essentially a proxy for the "quasi-rain forest" which occupies the majority of the Maya Lowlands (Lundell 1934). The artificial rain forest model is an array of tree crops, vine crops, root crops, and standard seed crops combined in a way so as to preserve 116 the primeval energy (and nutrient) cycles of the parent forest (Wiseman 1973).

Establishment of the system could have occurred as a result of selective clearing as practiced by the modern Maya. Cowgill (1962),

Reina (1967), and Lundell (1934) point out that the modern Peten milpero does not clear-cut the forest, but spares certain culturally useful species, while eliminating those plants not considered valuable to him or to the society at large.

Large useful trees are considered communal property, and their harvest is accompanied by a local tax (Reina 1967). The selective advantage given to useful plants increases their relative numbers in the cleared zone. Favored species suddenly benefit from more light, space, and nutrients. Simple care in burning protects such trees and shrubs from smoke damage. When the field is fallowed, the nonselected plants sprout again from still-living root systems and seed from surrounding forest, but suffer from competition with more established useful plants and secondary growth. If prehistoric population pressure caused the fallow period to be shortened to less than the maturation period of climax forest plants (about 6-10 years), the non-useful plants would have again been cut before reaching maturity and young useful plants spared. Continued over time, useless climax species would be repeatedly slain before reproducing and replaced by successful secondary and useful species. The eventual result may have been regional extinction of non-useful plants that have a maturation period longer than the fallow period. Spared species would assume dominance and only weeds would remain. These would be short-lived plants similar 117

in niche to the cultigens, producing a pollen rain similar to that of modern milpas (Wiseman 1974).

The establishment of an artificially selected woody plant component (Table 25) brings up the question of agricultural efficiency below the canopy. Would seed and root crops have been successful as an understory in view of the restricted amount and changed composition of light penetrating the layers of preserved rain forest? Light intensity of the forest canopy is 10,500 foot-candles, reduced to 100 foot-candles on the densely shaded forest floor (Allee 1962, data from

Barro Colorado Island, Panama). Small herbaceous crops which cannot reach canopy layers, invest their more modest metabolic expenditure for smaller, above-ground structure and in fruits, seeds and underground starch bodies. Simple maize agriculture requiring 4000-8000 foot-candles per 12-hour day seems unlikely; with partial tree cover,

1000-3000 foot-candles. Selection for shade-tolerant varieties was essential, as well as extensive use of shade-tolerant root crops such as manioc (Manihot esculenta) and malanga (Xanthosoma violaceum), which produce starch bodies at 1000 foot-candles and make fewer demands upon the local nutrients than seed crops such as cereals and beans (Graf

1957). Of course, the density of larger woody plants would have had to be adjusted to lower-story cultivation -- a closed canopy would not be compatible with cultivation of an understory.

Once the artificial overstory was established, the understory cultigens would gain many advantages. First, leaf litter and unused fruit would fall from the overstory and recycle some nutrients to the soil. Decomposers would be less disturbed by tillage, hastening decay u8 Table 25. Plants that may have been used in an "artificial rain forest."

Trees Vines Sapodilla (Achras zapota) Pitahaya (Cereus sp.) Sapote (Casimiroa edulis) Vainilla (Vanilla fragrans) Chirimoya (Annona caeriabola) Oopchi (A. reticulata) Herbs Zaramuya (A. squabose) Maiz (Zea mays) Ramon (Brosimum alicastrum) Frijol (Phaseolus sp.) Aguacate (Persea americana) Tomate (Lycopersicum esculentum) Sabal (Saba]. sp.) Squash (Cucurbita spp.) Coyol (Acrocomia mexicana) Pimienta (Capsicum annuum) Cacao (Theobroma cacao) Root Perennials Mamey (Calocarpum mammosum) Yam (Dioscorea trifida) "Cherry" (Pseudolmedia spuria) Sweet potato (Ipomoea batatas) Corozo (Orbignya cohune) Malanga (Xanthosoma violaceum) Porn (Protium copal) Yuca (Manihot esculenta) Caoba (Swietenia macrophylla) Yam bean (Pachyrrhizus tuberosus) Chacah (Bursera simaruba) Ayal (Crescentia cujete) 129 and release of nutrients to the cultigens, while other soil organisms would have improved the physical character of the soil. This would have had the effect of low-level tillage and fertilization that would yield more reliable harvests and reduce the fallow time. The combination of canopy cover, leaf litter and herb layer and oil organisms would increase the water retention capacity of the soil. Slowing rainfall runoff has three favorable consequences: (1) the elimination of destructive raindrop impact and solution erosion, thus preserving the little humus present on slopes up to 15° (Greenland 1975); (2) the

'Preservation of nutrients in the local soil, by inhibiting nutrient solution in surface runoff; and (3) slowing of the leaching process.

Alternatives to maize cultivation, such as ramon or manioc, would have meant less dependence upon strictly cereal cultivation, increasing the resource base and resulting in stable, productive system.

The woody trees, shrubs and vines, already in equilibrium with rain forest pests, utilize internal defenses, making them steady, dependable sources for food and materials. Insects could still be a problem for stabilizing understory species which generally lack internal defense systems. Alternation of maize, or other seed cropping, with manioc or yam (Dioscorea trifida) cultivation in the understory would decrease chances for attack by the same pest, since their life forms and edible parts have different constitutions, environments, and parasites. Crop rotation in the annual component of the artificial rain forest would limit incursions of species-specific pests to minor outbreaks. 120 Weed infestation could be handled by hand weeding or substitution of

useful species in the "weedy" niche. This could be accomplished by

companion planting, such as manioc and cucurbits (Cowgill 1961). The

cucurbits, shading out the intolerant weeds, would yield even more

edible tissue resulting in maximum productivity. Root competition be-

tween tree and herbaceous crops is the main objection that I can see

for such a system. If tree crop species are characterized by extensive,

shallow root networks, the advantage that the herbaceous crops gain by

litter and shade may be offset by depletion of local nutrients. Root

crops and the semi-epiphytic Vanilla fragrans may produce under such

conditions, since they use fewer nutrients than seed crops. There may

have to be a zone around each tree that is either weed covered or sup-

porting slowly growing root crops. These areas may be used for storage or apiaries.

However, since competition is between crops, rather than be-

tween crop and weed or pest, the net gain in usable productivity makes

this model attractive from an environmental, as well as agricultural

viewpoint.

Figure 10 illustrates a greatly simplified version of the arti-

ficial rain forest hypothesis. Brosimum alicastrum, the ramon tree,

occupies 28% of the simulated plot, approximating its density on large

Maya sites in the Peten (Puleston 1969). For our purposes, the 28% of

the plot surface will be used by the tree alone, since root competition and shading decrease productivity of the herb component within this

zone (Wiseman 1973). The trees may be cropped continuously, providing an average minimum of 1122 kg/ha/yr (2470 pounds) of fruit (data from 121 122 Puleston and Puleston 1971). . The remaining 72% of the plot receiving

full or partial sunlight may produce (using Cowgill's 1961 figures) 1871 kg (4118 pounds) of Manihot, or 1151 kg (2534 pounds) of maize, or 17289 kg (38,036 pounds) of beans, or 4773 kg (10,505 pounds) of

sweet potatoes, or 5239 kg (11,526 pounds) of squash per hectare of "artificial rain forest."

However, it is probable that some fallow time was required, or that the system was not totally efficient. Therefore, I will assume one year of rest for each year of understory cropping (the ramon, being perennial, produces each year). Fallowing also insulates the plot from pest invasion (Janzen 1970). Today, a family of 5.78 persons can tend a milpa of 5.28 hectares (Cowgill 1962). Since an artificial rain forest plot would require hand weeding, mulching and more care in general than a modern cornfield, I assume approximately 3.0 hectares is about what a family could have managed with efficiency if the homesite were near the plot.

In a maximum stability model, instead of planting the whole plot with one herbaceous cultigen, a combination of manioc, maize, sweet potato, squash, and beans were planted to increase diversity, effectively stopping pest invasions, runoff, etc. (Wiseman 1973). The crops may have been sown in groups or companion planted. Using Figure

10 as a standard, approximately 32% of the plot had fairly full sun, as required by the shade-intolerant Zea, Cucurbita and Phaseolus, and 40% in the semi-shade and root competition zone allowed by the more tolerant Manihot, Xanthosoma, Vanilla and Ipomoea. 123 System 7: Arborculture

The ramon tree (Brosimum alicastrum) has been proposed as a

species used in a monocultural orchard system, as it is capable of

sustaining high yield and therefore high population density (Puleston

and Puleston 1971). There is a demonstrated high correlation of the

ramonai occurring with archaeological sites noted by Lundell (1934)

and Puleston ( 1969), possibly remains of former orchards. This

curious distribution may be evidence of ramon groves surrounding the

sites when they were active, but edaphic factors such as better drain-

age, increased available phosphorous (Cowgill 1962), or different soil

type may also account for this phenomenon. Ramon, however, is not the

only species exhibiting this curious distribution; Lundell (1954) lists

guayo, aguacate, mamey, and other fruit trees as occupying archaeolo-

gical sites, indicating a more complex orchard system than ramon mono-

culture. Table 21 indicates that the annual yield of ramon nuts is

1245 kg (2740 pounds) /hectare. Unfortunately, there are no data on

the quantity of ramon to support an individual per year. Therefore,

the potential carrying capacity of ramon orchards must at this time go

uncalculated. However, Puleston and Puleston (1971) indicate ramon

nuts are storable in chultuns (underground chambers) and may have

served admirably as a reserve in times of necessity, somewhat insulat- ing the ancient Maya against yearly climatic variation. The presence of secondary compounds in the ramon probably reduces its susceptibility to disease, providing a potential stable food source for the ancient

Maya. 124 System 8: Terracing

Turner (1974) gives an excellent summary of the evidence for upland geointensive methods in the Maya Lowlands, as does Wilken (1971).

However, implicit in such "intensive" models is the use of fertilizer

to allow these terraced zones to be highly productive. Without mulching or fertilizing, the terraces are little more efficient than the abundant flat lands (less than 15° slope). The use of check dams increases the amount of land available for agriculture in a way that avoids limiting factors such as erosion and chemical weathering. Ter- races are good evidence for almost total land use, since all arable lowland would be under cultivation before the expense of terrace construction would be undertaken. Since there is no evidence, either archaeological or ethnographic, to indicate such use, I do not treat productivity of terraced slopes differently from that of the flatlands.

Silt traps may be considered localized "high productivity" sites, but they are more than offset by the xeric nature of the rest of the slope

(which would best be planted with drought-hardy cash crops such as

Bursera, Agave or Yucca). In the quantitative reconstruction to follow the 5% of the Peten with 15° or more slope (Greenland 1975) will therefore be included as part of the general upland zones, to be treated with techniques only slightly differing from those used on flat or rolling terrain.

System 9: Ridged Fields

The most difficult system to evaluate is the ridged field system, or chinampa (Wilken 1971; Siemens and Puleston 1972), a 125 geointensive system best suited to swamps and river floodplains, which comprise approximately 21% of the Peten. Although much has appeared in print concerning its supposed efficiency (Siemens and Puleston 1972), there is little hard data amenable to a simulation scheme, either in amount of work necessary to construct drained field systems, or their potential productivity. The figures of supporting capacity of highland

Mexican chinampas (Palerm and Wolf 1957) are almost certainly too high to be applicable to a Maya agricultural situation.

There is scanty evidence that corn borers significantly reduce maize yield when planted in bajo situations, perhaps since there are plants in the bajo which are alternate hosts for the borers (Reina

1967). Any intensive chinampa-like method of bajo exploitation might be limited by pest incursions if maize were the principal crop used.

However, in parts of northeastern Peten, with high site density, over

60% of the terrain is covered with bajo-type vegetation (Bullard 1960;

Lundell 1934). Ridged field remains have been found in the Bajo de

Santa Fe (Harrison 1975), and more may exist in the Bajo Juventud west

of Tikal (Dahlin, personal communication 1976). Therefore, circum- stantial evidence is good that they contributed food or cash crops to sites. If, the economies of Tikal, Ifaxactun, and other bajo-oriented crops, then the as I suspect, bajos were used for cash (or trading) am building a mini- subsistence contribution would be slight. Since I population densities mal model, if I support archaeologically estimated model's purpose will have been without adding bajo productivity, the

served. 126 Agricultural Simulation

The data, quantitative and qualitative, presented in the first

two sections are bits and pieces of an organic system that has func-

tioned in various forms since man arrived in the Maya Lowlands, prob-

ably over 4000 years ago (Cowgill and Hutchinson 1966; Hammond 1976;

Puleston 1976). Using a systems approach, these data may be linked in

such a way as to simulate the system described in the ethnographies

and hypothesized by archaeologists. While this method is fraught with

problems, it points out those aspects of each system that yet have no

data; and the researcher, by rigidly structuring his data, may obtain

a new and perhaps different view of the organism under study.

The major problem with a simulation is that it reflects the

researcher's bias in its initial stages more than a hypothesis and test

approach. The second problem is that, since it is a model, it may not

reflect reality at all, but rather some chance patterning or correla-

tion. The third problem in a simulation is the number of assumptions

that must be made in lieu of actual data when they are lacking.

The program (originally written for a WANG 700, now adapted to

a CDC5400 computer) simulates as closely as possible the actual rela-

tions of one sub-system to another. Program MAYAPOL handles variables

of pollen emission, farm personnel, time (man-day) input, and produc-

tivity of crop plants in such a way that variables of one type may be expressed in terms of other variables. In this way, any variable may be manipulated to see how other variables of the system change in response. Resultant estimates may then be tested against archaeological and paleoecological data. 127 The program has several subroutines which are important to describe in a general way. The program MAYAPOL itself is merely a

quantified system of labor and time, transferring one data set into another via ethnographic data derived from Cowgill (1961), Reina (1967)

and Emerson and Kempton (1935). Figure 11 provides a summary of aspects of the program.

The estimated pollen production output from the main program

is then put into subroutine A, which is essentially a semi-log regres-

sion of pollen deposition against distance from source area. Data are

derived from Raynor, Ogden and Hayes (1972). However, the spatial

distribution of the pollen-producing plots are of prime importance

since a large field at distance will produce the same pollen deposition

at a given locus as a small plot up close. To avoid this rroblem, I

have constructed a "simulation polygon" which is no more than a randomly

generated landscape that approximates that of the modern Peten. Its 2 dimensions are 10 km a side (100 km ) , in the center of which is a

hypothetical lake 0.4 km in diameter. LANDSAT 1 imagery indicates that

approximately 10% of the land surface of the Peten is covered by tropi-

cal savanna (grasslands), 20% by bajo (swamps) and 70% by upland forest

(Fig. 4). These are expressed in the polygon as randomly placed 1-km2

blocks of that vegetation type added until the requisite coverage is

obtained (Fig. 12). Agricultural fields of any type may be placed at

random within the polygon in a simulation of reality.

Each randomly placed field within the polygon is a given dis-

tance from the lake, and therefore has a given potential downwind pollen input. However, only those fields directly upwind of the lake will 128

u- E

O o_

o 1 29

BAJO 0 MILPA 0 !

K M. ....::::- SAVANNA \5 WIND VECTOR

pollen influx Figure 12. Simulation polygon used for calculating values. 130 contribute pollen; those to the side will not. This factor is handled

by randomly generated wind vectors expressed as ten lines extending from the core site. All fields intercepted along each line are recorded as to their distance, which is converted to pollen influx via the regression formula, then totaled for that vector. This is intended

to simulate pollen from each field blowing along the line to the depositional basin. The 10 wind vector-pollen influx sums are averaged and this value is used as the estimate of pollen influx for that agricultural system dispersed randomly within the simulation polygon.

The results of the MAYAPOL program, as described above, are within one yearly cycle, and may be considered as an optimum model.

In reality, other factors prevent expression of such an optimum efficiency. Therefore, this program is run for a 100-year period, with randomly selected years having adverse effects upon the productivity.

Reina (1967) indicates that pests (corn borers), climate (insufficient rainfall), and miscellaneous cultural causes adversely affect the maize crop. Table 26 illustrates the type, magnitude and year of factors reducing the potential yield of maize pollen and crops. Each

MAYAPOL simulation is run for 100 consecutive years and the mean productivity calculated, since a pollen sample (or housemound) would integrate many years as collected in unvarved sediments.

Table 27 gives results of several runs of program MAYAPOL, representing 100-year periods, and allowing but one type of agricultural practice. Each agricultural system presented in Table 27 is a discrete entity. I assume the prehistoric Maya did not use any one method, but probably selected those which best fit the diverse habitats 131

Table 26. Summary of tr eatments during a 100-year run of program MAYAPOL. Assumptions: Climate decreases yield 7% in 20 randomly chosen years (1 in 5). Climate decreases yield 15% in 5 randomly chosen years (1 in 20). Climate decreases yield 50% in 1 randomly chosen year (1 in 100). Corn borers decrease yield 10% in 18 randomly chosen years (1 in 6). Corn borers decrease yield 15% in 5 randomly chosen years (1 in 20). Corn borers decrease yield 25% in 1 randomly chosen year (1 in 100). Miscellaneous causes decrease yield 10% in 5 randomly chosen years (1 in 20).

Decrease Decrease Percent in Pollen Percent in Pollen Year Decrease in Yield Prod. (%) Year Decrease in Yield Prod. (%) 2 7C 7 53 7 7 4 10 B 58 10 5 15 C + 10 m 15 59 15 15 6 7C 7 62 10 8 15C 15 63 7 7 11 10 3 68 7 7 13 10 B 69 15 15 14 7C 7 71 10 1? 7C 7 73 10 19 7C 7 75 10 20 C + 10 B 7 78 10 + 7 C 23 10 B 81 7 7 24 7 C + 10 B 7 82 7 7 27 15B 84 7 7 31 10 B 85 10 35 15 B 86 10 37 15 B 88 7 + 10 B 7 38 C + 10 B 7 89 15 + 10 B 15 40 7 c + 10 B 7 93 10 49 15 B + 10 M 96 7 7 50 7C 7 99 10 52 50C 50 100 7 7

C = Climatic input; B = Corn borer attack; M = Miscellaneous causes of yield decline. Mean percent of potential crop yield allowed by climatic and other inputs: Mean X . 94.55%, standard deviation s = 7.85. Mean percent of potential pollen production allowed by climatic and other inputs: ïc = 97.41%, s = 7.92.

• •-• '

132

CO 0 0 -41- N 0 40 N N Cs- \ H N OS' hrN

.4., (I) O CD rti \ %.0 0 •d O . H N t(1 0 bl 1 4-) 4.rn \.0 O H CO •ri TJ e et 0 H H • ai ai 4-) 4.3 H Si CO 0 gl 4-) 0 O g g O . •1-1 4 H 1-1 7. O cIS O TI cr-i E A g 4-1 i--1Ui ai G) .0 Il le P.-4 4-' ‘..0 N- O 0 *••, 0 0 0 0 O N a) b10 o E H 4.)rg r0 r0 n P., ai N H r-4 H4-' • ,, 0 • • • ro O al tO <1) N \ r-1 fai 0 g pi, (0 •H g aS (H N N crn CUD 0 III( -1-4 ai • t 0000 O -1-, P4 g F4 0 H 0 g 0 r0 r0 H •H bD 0 4 ‘..0 H CO v o O E er. H e • • • • E r-I H Z OC) H CNI (Ni r1 'CO0 0 0 (0 ai P. Pi .0 a) ••• n .0.n ..--. ,---, a) a) 01 .:40_, 4 CV 4a) -P 4-, N Cs- .V.• Cs- Cs,- l'cl 0 H ...., (,) ...., ..... g 0 cH 4:0 -P 0") E N •ri f.i 0 ai CO .. TI ....., li TI TS 4 (O r-4 0 P‘ 4 g g -P ••••N rd i-1 • S-i (rn al 0 ai (0 ri 0 •r-I d) E a) 0 >. cH -P 4444 Or-4 .f -D H ri H 3 .g .- (0 cc CO Pi CO Pi Pi 0 •ri • rai n 0 A X (OHO CU ;•- i). '0 N- $-1 •H HG) 0, r0 4-) a) 4 (-1 X • 0 0 •H -P o 0 g -P 0 1 cH 0 c., r0 crn r-I H r-i H H 0 bD 0 0 44 0 40> -H >- P-, ;-n ts- H 01, Pi O <0 0 g .-4 0 °CO 1 O Si H •H (11 0 c4-4 g g f-i -p • b. g 0 0 0 a) Pi H 0 O0 4-1 •H ..-4 4-) E 0 -0 c.) 4-) 4-, H 0 0 0 4.) 0 g CIS CI5 0 • Pi P4 P. 0 00)0o • $-4 '1-1 .0 CU H HO f--1 H .4-4 0 E -P -P 4 0 H Z Z -r4 -ri .1-1 .4 o a) CO (-1 GO 4-) •H H E P, E-1_ = Z 0 = •-1 1:4 ‹ •ri -1-1 4 0 -r-I 0 ou A 3 3 P., (0 1:4 CO H O H N (41 L(\ 'O C.- • • r42 Pr\ 133 surrounding a given site. Figure 13 illustrates the agricultural systems previously discussed, arrayed along two edaphic gradients, to determine the potential "maximal habitat" of each system (Wiseman 1976).

By combining the data produced by seven runs of the MAYAPOL program with the gradient analysis, we may discover which methods are potentially competitive and which are complementary in the Maya Lowlands.

These methods which ecologically compete must be assessed as to their stability (in terms of agricultural limiting factors) and productivity.

More stable and/or productive systems should be assigned greater weight in a multi-system reconstruction, to approximate the apparent high productivity of past Mayan agriculture.

I wish now to present a series of fanciful episodes in the history of agriculture in the Maya Lowlands, by combining the derived figures presented in Table 27, the gradient analysis, and the simulated landscape shown in Figure 12. Although the values for productivity, pollen influx, and personnel are expressed in yearly terms, they are mean values for a one hundred year run (Table 26). Population figures are expressed as: (1) producers (farmer + family); (2) persons supported (the mean maximum population that may be supported by agricultural productivity; and (3) persons per square kilometer (persons

supported 4- 100). Due to the number of assumptions which must be made, the precision of this quantified reconstruction may be no greater than an order of magnitude.

134

90 -

70 -

ZONE OF MARGINAL 50 AGRICULTURAL UTILITY

30 - EROSION CONTROL

NO TERRACES 20 - AVAILABLE HABITATS

SILT TRAPS

MIL PA ARTIFICIAL RAINFOREST RAMON CULTURE

DOORYARD GARDEN INTENSIVE MI LPA RAISED ARTIFICIAL — FIELDS RAINFOREST LEVEE RAMON CULTIVATION CULTURE

INUNDATED SEASONALLY WELL DRAINED OCCASIONALLY DRY INUNDATED MOIST MOISTURE CONTENT

Figure 13. Tropical agricultural systems arrayed along two edaphic gradients. 135 Scene 1: Preclassic and Postclassic Agriculture

A small hamlet of less than twenty structures was situated upon

the shores of the central lake (of the simulation polygon). The in-

habitants (approximately 15 producers) cultivated the 100-meter wide

band of marsh surrounding the lake, raising mostly manioc, which pro-

duced an average annual harvest of 16,400 kilograms of tubers, and some

maize (500 kilograms harvest). Their neighbors raised crops in scat-

tered upland milpas within 0.5 km of the lake. They had approximately

twelve hectares of planted fields each year, which were then let fallow

for four years. The thirteen (milpa) producers raised 19,000 kg of

maize per year. The combined milpas and drained fields contributed 2 0.0109 Zea pollen grains/cm /year to the lake sediments. Surpluses

were the rule, since the supportable population was 66 persons, while

there were only 31 producers. The maximum population density for the 2 polygon as but 0.67 persons/km . Remains of from seven to thirteen

house mounds date to this hypothetical period (data from Puleston 1974

were used to convert population to housemounds).

Scene 2: The Chicanel Expansion

The small Preclassic hamlet may have undergone a population

surge during the centuries bracketing the birth of Christ. Milpa

technology now overshadowed marsh cultivation, and had spread through-

out the upland forest zone of the simulation polygon at this early

date. Fourteen percent of the total land surface was covered with productive field and 56% was covered with bush fallow. Only culturally spared trees, small patches of forest, and the unusable bajos remained 136 untouched. Fifteen hundred producers, from numerous small hamlets, worked in the milpas, obtaining a mean annual harvest of well over two-

million kilograms of maize. The supportable population was 7700 per- 2 sons, or 77 persons/km , resulting in seventeen structures per square kilometer dating from this episode.

The fields had a mean nearest neighbor distance of 425 meters

(s = 137), separated by early secondary vegetation (less than four

years old). I suspect that pest populations and leaching may have

quickly become limiting to such a system. Unstable agriculture of this

type probably caused increased competition for richer soils, or a beginning of intensive agricultural systems.

Scene 3: Classic Period Intensive Agriculture

The first possible Maya biointensive method to receive wide

attention was the Ramon orchard model (Puleston 1969). If one half of

the upland forest zone was covered with ramon groves, the total pro-

ductivity of the orchards would be almost four-million kilograms.

Unfortunately, we have no data linking productivity to a supportable

population, so these aspects must go uncalculated. No Zea pollen was produced in these groves.

Ridged field systems occupied the bajo soils which cover 20$ of the simulation polygon. They provided the Classic Maya with a stable, highly productive agricultural option. The fields were continuously cropped, but probably needed more labor input per hectare than a milpa plot of similar size. Twenty thousand hectares of ridged fields had a potential productivity of over three million kilograms of 137 maize, if that were the sole crop raised. The supportable population of the ridged fields was 11228 persons, or 112 persons per square kilometer. Twenty-five hundred house mounds were left by the ridged field cultivators and those they supported.

The intensive milpa system occupied one half of the upland forest zone (3500 hectares), on those soils to which it was best adapted. Terraces, erected on hillsides to retard erosion, were in- corporated into this intensive system. Maximum yearly harvest of maize from the intensive milpas was slightly less than two million kilograms, since it was raised on but one plot in three due to crop rotation and/ or fallow. The labor input of 3600 producers supported over 6500 persons (6500 persons/km2 ). These inhabitants left behind 1400 house 2 mounds for later archaeologists to discover, and 0.0018 Zea gr/cm /yr for the palynologist.

Half of the upland forest zone was occupied by artificial rain forest (Wiseman 1973) which provided, in addition to maize, beans and squash, a host of minor cash crops such as Vanilla, cacao, and other crops. Producers, numbering a little over 2500 persons, tended the immensely stable and productive forests. Yearly harvest was four million kilograms of ramon, 1,800,000 kg of maize, 3,640,000 kg of manioc, and numerous smaller cash and subsistence crops. The population supported by just the maize (6315 persons) and manioc (5523 persons) productivity is approximately 12,000. However, the addition of the ramon and other products probably makes the figure closer to

18,000 persons (180 persons/km 2 ). Pollen influx into the central lake from the artificial rain forest is small, about 0.00004 Zea 138 2 grains/cm /yr. The inhabitants lived in and around four-thousand houses which now are but mounds covered by the forest they sought to control.

Using the maximal habitat concept, I wish to describe the total productivity and supportable population by the combined output of the artificial rain forest, ridged fields, lakeside cultivation, and intensive milpa, which are ecologically complementary in the simulation polygon. Ten percent of the land was set aside for urban construction 2. or for scrublands. The combined harvest for the 100 km simulation polygon was four-million kg of ramon, 6,800,000 kg of maize, and

3,656,400 kg of manioc. These combined fields produced a pollen influx 2 of 0.015 Zea grains/cm /yr in the central lake. This stable, diverse 2 system supported a population of 35,758 persons (358 persons/km ) who left a total of 8000 scattered rural and suburban house mounds in addition to their art and monumental architecture. CHAPTER 7

A DESCRIPTIVE MODEL FOR POLLEN/VEGETATION CHANGE IN THE MODERN MAYA LOWLANDS

From the field data collected in 1973 and 1975, a strong case

for environmental heterogeneity in the Peten emerges. While the causes of this variance are still unclear in some cases (such as savanna vegetation), the direction and magnitude of vegetation change are becoming well documented. I would like to summarize the modern data in the form of a descriptive model incorporating two independent gradients that will allow assignation of any type of upland Peten vegetation to a locus in the bidimensional space (Fig. 14). The model may then be used to compare and contrast different habitats by their component pollen taxa. This model is amenable to use as a paleoecological tool, if used carefully as an analog with the uniformitarian assumption.

The basic assumption of the model is that there is significant change in Peten habitats. The species-abundance tables given in Chap- ter 5 prove this assumption is valid and true, and may therefore be used as a fact. The second assumption is that vegetation varies along two gradients: one a "natural" forest-savanna vector, and the other a successional (forest-old field) vector, if only upland, non-montane vegetation is considered. This assumption seems to be valid in light of the field work done in the Peten and reference to texts on tropical vegetation (Beard 1953; Lundell 1934; Breedlove 1973, Graham 1973).

139

140

cy\ 1 E-1 I Q) u) i—i to ,cs la, ct g E t-I al CD r--1 06 C,-, ta 1 O E-+ H al O N .-1 — 0 4-) Cd Ct 4-) Z 0 no Q) I .--i Q) El Ct 't5 +) › H .T1 + ''' ai g to ..g at g (1) trl O H .8-1 Pi 4) tcl -../. O 5 H H 0) VI I I g 4 I-I i-4 I-1 $.4 Z H g Cf) •Crl O 0 . .4 0 0 -I-I E-4 to 4-) O CL) Ct CD 4-) 4-) 0 H• CD reN -I' H H rn L12 II 1 O a) E-I H I-1 1 A. > > > to 0 + + - -r-I + e C.) g r4

U1 N I I 4-4 0-7 to ti) . +

c.) r4 II I Ir.\ u-N 1,( E-i E-1 e-I + + H

trn 1

S• :24 E ai k o 3 d 0 0 o ca CC H aONVEMLISICE DNIIIEEIHM sus so Itraoaad ita Since the Peten contains no estuarine or highland ecosystems, the assumption probably holds here.

However, wide dispersal in pollen renders the model equivocal

in a palynological sense unless the long-range inputs into the system

can be eliminated. The third assumption is that all potential habitats

in the Maya Lowlands can be validly placed within the bidimensional

space of the model. This assumption does not hold true for the Pinar

or pine forests found in southeastern Peten and in British Honduras.

They are related to savanna vegetation but do not fit the "natural"

edaphic gradient, and probably add a third dimension to the model.

Bajo or swamp vegetation, while not included in the Upland Model, is

also related to the savanna and thorn scrub vegetation types.

Derivation of the Analog Model

Pollen percentage figures were used in analyses. These pollen

data, arrayed along the derived environmental gradients, were evaluated

through principal component analysis to distill the main patterns from

the data (Adam 1974; Webb 1974). The pollen information, arrayed as

an M by N matrix, where M = number of pollen taxa (variables), and N = number of sites in the gradient (observations), were punched onto computer cards and processed through program PRINCI„ an R mode principal component analysis program which standardizes the data before calcu- lation of the correlation matrix. The resulting set of M eigenvectors are patterned representations of the original variables. Important eigenvectors (principal components) account for large amounts of variance in the original data and have potential paleoecological 142 significance, while the less important eigenvectors describe residual noise, error, or variation caused by a closed percentage sum in the pollen counts. Each eigenvector consists of M weights associated with

the original pollen taxa, so that the weight is proportional to the im-

portance of a given pollen type to that eigenvector. Further manipu-

lation of the eigenvector matrix results in a matrix of amplitudes of

the principal components, arrayed as a function of sample (observation)

rather than variable. The amplitudes can be a matrix through space, as

in the gradient, or a time series (Adam 1974) expressed as a "goodness

of fit" coefficient of each site data set to that empirically derived

principal component (eigenvector). As an example, if eigenvector 1 has

heavy weights for pollen types Pinus, Quercus and Graminae, and nega-

tive weights for Alnus and Compositae, and sample A has high percent-

ages of the first three taxa, then its amplitude will be high. There

is the same amount of information in the principal components matrix

as in the original data matrix, but it is statistically distilled into

a few independent patterns empirically describing most of the variance.

Due to statistical problems inherent in principal components analysis,

the number of observations (samples) must be equal to or greater than

the number of variables. Many pollen types were eliminated from analy-

sis for this reason (Webb and Bryson 1972). Taxa rejected include

extralocal pollen of Pinus, Carpinus and Myrica, the purely lakeside

genus Typha, and those taxa which do not have significant variance and EuFenia. along the gradient (or low percentages) Brosimum-Cecropia,

The remaining taxa, containing most of the information in the gradient 143 but less noise, were counted as a "second fixed sum" of 1500 grains, and this was used in the analysis.

Analog 1: The Edaphic Scale

Pollen variance along the vegetation gradient discussed in

Chapter 3 is presented as the first principal component of the gradient

factor analysis (Table 28, Fig. 15). This is the main pattern expressed in the pollen variable/gradient matrix, accounting for 22.43% of the total variability, a strong and statistically meaningful signal.

However, the ecological validity of the pattern must be assessed.

Those pollen types with high positive weights (Table 28) are, without exception, trees of the mesic and monsoon forests, and grass pollen, with negative weight, is characteristic of savanna and grasslands. The low values for other NAP indicate that there is little input of weed pollen into this pattern.

High forest (mesic and monsoon varieties) produces a pollen flora that is dominated by the following types (in order of descending importance) Moraceae, Terminalia type, Manilkara achras, Haematoxylon campechianum, and Bursera. Savanna and grassland are heralded by grass and Acacia pollen. A review of the plant species abundances along the gradient indicates that these seven pollen types reflect their producer plant's position along the gradient. Therefore, the first principal component will function as a quantitative analog for detecting the presence or absence of grassland/savanna and high forest.

Eigenvector 1 has insignificant amplitude on sites VT-4 and

SL-10, both of which are Encinal sites (Fig. 14). Encinal pollen can Table 28. Eigenvector weights (scalars) of 17 selected lowland pollen taxa.

Vegetation Successional Gradient Gradient Taxon (E ) 1 (E2 ) Habitat Non-arboreal Pollen Zea mays -0.05 0.87 Agriculture High-spine Compositae -0.28 0.71 Disturbance Low-spine Compositae -0.12 0.52 Disturbance Graminae -0.82 0.72 Savanna Chenopodiaceae 0.11 0.41 Disturbance Senecio-type 0.05 0.41 Disturbance Euphorbiaceae 0.08 0.36 Forest disturbance, savanna Cyperaceae -0.67 -0.11 Banks, savanna

Arboreal Pollen Quercus 0.11 0.11 Encinal Haematoxylon 0.59 -0.02 Bajo, savanna Zanthoxylum 0.09 -0.20 Forest Bursera 0.55 -0.19 Forest Achras 0.67 -0.08 Forest, disturbance Acacia -o.69 0.19 Forest, encinal, scrub Celtis 0.23 -0.42 Forest Terminalia 0.62 -0.70 Forest Moraceae 0.70 -0 .66 Forest 145

FIRST PRINCIPAL COMPONENT, VEGETATION GRADIENT Eigenvector Amplitudes That Describe Forest Pollen

SECOND PRINCIPAL COMPONENT, VEGETATION GRADIENT Eigenvector Amplitudes That Describe Encinal Pollen

2 .0 1.0 2.0 10

•

20 •

FIRST PRINCIPAL COMPONENT, VEGETATION GRADIENT Eigenvector Amplitudes That Describe Savanna Pollen

0 Pachira Mesic Monsoon Co ozo Aguado Swamp Forest Forest Fo est Fo est

Figure 15. Eigenvector amplitudes of modern pollen sample sites. -- Arrayed as a function of the edaphic and successional gradients. 146

be detected by eigenvector 2 which does have a high amplitude at these

loci on the gradient. Pollen types that have a high factor loading are

encinal dwellers, Quercus, Acacia, Terminalia type, and insignificant

loading on other pollen taxa. Encinal is easily distinguished by oak

pollen, while the other less important loadings characterize plants that grow elsewhere than encinal.

If these multivariate data are to function as an analog, they

must be comparable to fossil data. The method will be discussed at the end of this chapter.

Analog 2: The Successional Scale

The effects of artificial disturbance on the pollen rain were

described in Chapter 6. The data from Site VT-5 were analyzed by

principal components analysis using the same 17 taxa as the edaphic

gradient in an attempt to "factor out" the patterns in pollen variance

caused by human disturbance, and separate them from patterns caused by

natural habitat variability.

Unfortunately, the abundance of weed pollen at Site VT-5 over-

shadowed the tree pollen in such a way as to cause the main pattern to

express variance in weed pollen alone. This pattern which explained

314 of the variance was uninterpretable in ecological terms. The second principal component did fit the gradient and made ecological sense. In this principal component, there is a change in amplitude along the gradient that corresponds to the cover values of plants at Site VT-5. High positive loadings occur on Zea mays,

Graminae, high-spine Compositae, and low-spine Compositae, which are plants occurring in the milpas and new fallow plots (Fig. 16). The

• •

1 1+7

0 F-1 II-4 0

• 0 0 O .0 H H • • O 0 4 P4 E tO E.) cli

$.4 4-) 0 • -1-4 • '0 CO O 0 vs\ 0 u"\ • • • • O (0 N.\ C H 8 0 4-) -r-4H 1:1 F-t 0S 0.) c..) (Oat to 0 Cl) .Zt 'CI ta

4-) 0 Cli C) .0 0 O +' 4-) 0 N O G) (0E C-1 0 0 O 0 21 44

• f—I • •rf O• -'-1 Et' O W CIS 0 COG) 00 0P C.) a) 0 7 0 0 4-) 0 P0 • (0 b.° Ct •ri H CO P.1 F.4 zits 4) 0 FL4 • n.0 r—I ciS •r-I E .0 05 bD c.) 3 H aS 114

SONVEHRISIG DNILIEIHXE ElIS 30 ammad 148

importance of grass pollen in the successional gradient will cause some

confusion with savanna grass pollen, but the other weed pollen can be

used to factor out the effects of man on the natural communities of the

Maya Lowlands. High negative loadings occur on Terminalia-type and Moraceae, trees of the local forests.

Comparison of Analogs 1 and 2

Both sets of factor analyses were performed on the same sta-

tistical set of pollen variables, with the same pollen sum so that the

results are directly comparable. Factor analysis discriminated in an

elegant manner the difference between the gradients, and described

those pollen types that reflect the different natural and artificial communities of the Maya Lowlands.

In both gradients, forest was described by high factor loadings

on Moraceae, Terminalia-type, and to a lesser extent, Manilkara,

Haematoxylon, and Bursera. These pollen types are the marker taxa

for identifying Peten high forest in pollen samples. Not surprisingly,

since they share much of their flora (Tables 7, 8), mesic and monsoon

forest are palynologically indistinguishable.

Oak forest or encinal is distinguished by the presence o

Quercus pollen alone. Additional pollen taxa that described oak forest

in the second principal component of the edaphic gradient, had signifi-

cant loadings in different vegetation types in other principal components. Acacia pollen was important in savanna sites and Terminalia pollen was important in forest sites. These data indicate the tran- sitional nature of the encinal. 149

Savanna and grassland may be detected through an abundance of grass pollen, without large amounts of other herbaceous pollen (such as Compositae) in modern or fossil pollen samples.

Secondary herbaceous vegetation is characterized by Zea pollen

(where maize agriculture is practiced), and that of the Compositae family, with smaller increases in other herb pollen. If these other

types are present, then high loadings on grass pollen indicate the presence of weedy grasses, not savanna species.

Although the conclusions derived from principal components

analysis are similar to those that could be made by observation of the

spectra, the data are amenable to rigorous statistical comparison with

fossil pollen spectra.

Use of the Analog Model

Pollen samples from long or short cores can be statistically

compared to the analog if they meet several criteria. The pollen

varieties must be the same seventeen taxa used in the analogs. The

pollen sum should be the same throughout the analyses, so that statistical error caused by differing sample size is eliminated (Faegri and

Iversen 1964). The pollen samples must be from the same geographic

area as the analogs.

If these criteria are met, then the use of eigenvector compari-

son is valid. However, the core samples must show temporal change in

the component taxa, so that when principal component analysis is per- no change formed on the data, strong patterns will emerge. If there is

in pollen percentages through time, then factor analysis is useless. 150

Assuming that there is change in the pollen variables, and statisti-

cally significant principal components result, then they may be com-

pared to the principal components of the analogs.

The method of comparison is Pearson's Product Correlation

Coefficient (r) between eigenvector weights of the analogs and cores:

r - 1.(x — 7c)(y — —y) (N-1) SxSy

where x = eigenvector scaler (weight) on taxon A in the gradient, y

scaler on taxon A in the core, N . number of pairs of eigenvector

scalers, and Sx, Sy standard deviations of the scalers. The princi-

pal components are independent of each other and therefore are amenable

to parametric statistical analysis. The results will tell us which

gradient the fossil data most resembles, and the amount of confidence

we may place in this resemblance. This is critical for reconstruction

of Mayan agricultural history, since we need to know whether the herb

pollen that is found in cores had its origins in savanna or fallow

field.

It is then a simple process to see how close the various individual pollen taxa weights compare in order to narrow down the choice of analog to the most ecologically and statistically sensible locus along the appropriate gradient. These data may then be used to reconstruct the agricultural and ecological history of the Maya Lowlands. CHAPTER 8

FOSSIL POLLEN ANALYSIS

Long and short cores were collected in the lake region of Peten

during 1975 to supply historical perspective to the completed modern

pollen analyses. The lake cores were collected with Davis, Hiller and

Livingston instruments (Faegri and Iversen 1964). Core slugs were

carefully extruded into special airtight plastic bags with the upper

end toward the mouth of the bag. Each subsection of the core was

labeled as to depth, site, date with a Dymo labelmaker, to prevent data

loss due to destruction of paper labels. The cores were transported

to the United States in a sturdy metal box.

In the laboratory, the subsections were cleaned to eliminate

surface contamination and oxidized portions, and then subsampled every

10 centimeters for preliminary pollen analysis. Each subsample was

treated identically to the mud-water interface samples to eliminate possible extraction method bias (see Table 5).

Counting procedure was the same as that discussed in Chapter

4, with the exception that 5,000 grain counts were sometimes used to decrease the statistical error in Zea pollen percentages. After the final 1500 grain count, 3500 more grains were scanned and tallied as either Zea or other pollen. Since maize pollen is easily recognized, this procedure did not take long.

151 152 Results of the Fossil Pollen Core Analyses

Several cores were taken in Lakes Petenxil, Eckixil, Oquevix, and Aguada Libertad. Due to limited time and equipment, cores longer

than 1 meter were not taken. I had hoped that the cores would inter-

cept enough of the Classic Period pollen deposition that multiple

analyses could be performed on Mayan agricultural pollen. Unfortu-

nately, only two cores from Lake Petenxil and two cores from Lake

Eckixil penetrated to levels shown to be of Classic Age (Cowgill and

Hutchinson 1966; Deevey 1976). The other cores were not used for paleoecological or agricultural reconstruction. The recovery of cores 11+ from these two lakes is important, since each has a published C chronology that can be cross-dated to the four cores considered here.

Core P-17 Lake Petenxil, Peten, Guatemala

Core P-17 was collected in four meters of water almost on the site of Cowgill's core II (Cowgill and Hutchinson 1966) in about the geographic center of the lake (Fig. 6). The core was taken with a modified Livingston piston corer with a core length of less than one meter. About ten centimeters at the base of the core corresponded to

Cowgill's Classic zone G (Cowgill and Hutchinson 1966), which termi- 2 nates at about 110 cm. The apparent discrepancy between two lake cores is probably accounted for by differing collection methods and sedimentation. The pollen spectra are almost identical to those of Tsukada

(Cowgill and Hutchinson 1966). Since only 10 cm were of Classic Period, this zone was analyzed intensely, so as to give enough weight to this 153 period for principal components analyses, while the Postclassic Period zone was sampled less.

Non-arboreal pollen (NAP) of (from highest to lowest mean percent) Gramineae, low-spine Compositae, Cyperaceae, high-spine Compositae,

Euphorbiaceae, Senecio-type, Cheno-Am, and arboreal pollen (AP) of

Quercus dominate the samples below ca. 75 cm. (Table 29).

Above the NAP zone is a rather abrupt transition to arboreal

pollen (AP) which occurs within 6 cm. This transition is followed by

a period dominated by Terminalia-type, Moraceae, and Pinus pollen. Of

other forest trees only Celtis, Manilkara, and Zanthoxylum show an in-

crease during this period. At about 50 cm depth there is a short-lived

increase of NAP, then a return to forest pollen until the surface of

the sediment.

Core P-20 Lake Petenxil, Peten, Guatemala Core P-20 was collected 25 meters from the north bank of Lake

Petenxil in 4.5 meters of water. The pollen sequence from P-20 is similar to P-17 with some striking differences (Table 30). First, the

section commences at about 60 cm depth, and there are Classic Period G 2 some indications that the sedimentation has been interrupted at least

once since the end of the Classic Period.

Below 60 cm NAP is dominant as in P-17, but low-spine Composi- is less im- tae pollen is more common than grass pollen. Oak pollen

portant, averaging about 10-15% in this sequence.

The transition is less abrupt in core P-20, with several taxa insignificant following one another from a position of dominance to an

154

n n 00 ++ + + Lr 0 + tr + o u-\ o trn trn skew eaz • • • • • • • • • IO CD NHOOOH

adS4 00 trn n Lf1 Lrn n 0 ISn 000 0 n • • •••••• • •• ••• • -oToeuas NHO CONOHN N-1-\ONK100

- \ 0 n Li• • U\0 • • Lf •-\ Lr• Li"• irN• 0• LI •.n 0• 0• If• \ Ir\• lr•n t.r• kr•n o• 0 _.1.- H oo I•r1 -t 0 H CC tr \ -1- H N 0 k-r) 0 -1- N H H N H H r--4 N H H n 00 Ir\ if c) u- L.rn tx1 0 0 n 0 0 ir\ Lrn I-r\ 0 IA a-eaos-pqamidna • • • • • • • G • • •• • 0H Ote\NtelHO Pr\ N 0 0

LC1 0 UO 0 uo 1.1 Lf-\ V\ o u-n Lrn avao -eaadko • • • • ••••• •• a •••••• O 1-1N-1- 0HON-1- HOcCOVD-t 1-1 t\ 0 H H n n auTds-ntoT Ir\ 0 o u- LI\ u- u i.r tr\ o Lr\ u-n trn 0 • • ••••••••••••••• • 9 -e4Ts0dw0o N N H N 0 K1 N (\ n12) N N C H -.I- 1-1 H H H H H N r--I n n n auTcis-T4STTA O• Lr•••••••••••••••••n 111 0 Lr\o if Ir\ tr, LI" \ 0 irn Lr \ t.r Lr Ir\ ae4Tsodwoo ren H re\ N trl CC n.0 tr\ CC O n.0 N N 00 H H 1-1 H H H N

Lr\ Lrn 0 Ill 0 Lf1 111 U-1 if\ lfn 00 1.11 0 LC\ u-‘\ 111 ••• OU 914 D • fa • •••••• • • •••• N0K1H 0 ir\HNO N V.") trl H 0 • cis O n u-\ o tro Lrn 00nn0Onn ••• • •• • • •• •• • wnifocoti4u -ez O N C Ir\ vo 0 1.1-n H 0 \

n n n adk4 no 000 Lr t.r tr ts-N o 000 Lr 0 0 Lrn ii. • • • • • • • • • • • e •11•0•6 n n --eTTTwew.raz, N N -...t .0 Ir cS -1- -1- U\ GN. - -1- r-1 N N N LI- \ --.1" 0 H 141 N K \ 4' K1 H N N N H H H

sapToaTo Lrl 0 U\ LI \ 0, if \ LC\ 0000 111 111 Lf1 U\ Li- \ Ill LC \ • • • • • • • • • s • • • • • • • • snoaanb tr\r-INN-1- NLC1 n.ONH-1- 0-.1- 0 nZHCC 0 1-1 H N H H H N H N H H

0 0 0 0 0 0 Lt1 0 n 0 0 Ir \ U1 0 LI-1 lf \ Ir \ Lr1 • • • • •• 11••• •••• •••• a-eao .e.low n Hr-trn Hoircr, -.1- V) N H N CC 0 -1- N D co tr\ N K1 N -I' N H N N r.11 N H H H H

0 Lr1 0 0 0 C 0 0 0is- 0 0 0 0 0 0 IA IA • . • • . • • • • • NH 00 -e-re)ITTizew 00 r-4 -.1- 0 H n • dds n01111-r• • •n Oir\o• • nnou• . • •-NO onot.r.ir\Lr. • • xoIixowaH o OK\HN H N H K1 n n n n n Lrn u-n o 00 Ir\ tr cr o Lr Lr tr • ds sTq.TeD • • • •• •• ••• • •• •• r`r“ \J H rr“) 400 CHO ON

irn Ir\ 0 Lr. L.N 0 Lf1 Lf1 LC1 0 0 -ecinzewTs 00 00 • • • • • •••• •• •• Eaasang • N H N HOc0 N H

lf1 LC\ 1.1-1 \ Lf1 0 Ir\ 0 Lr\ 0 0 IA IAO 00 • ds eTovov • • ••• • • a •• se • H nt0 K1 0 -I' H H N 00 H

• 000000LN0000000000 0 ••• ••• • Il s • ••••• II• Nonou-Nnt--- 414Q00cOON-tir\Lr\N-co N N Ls--n CO CO CO CC CO CC CO

00 a) H CV PC1 'Lro.r) N co cs, 0 r-i 000000000HHHH HHHHH N NN NNNNNNNNNNNNNNN g H HHHHHHHH H HHHHHHHHH H (14 P-4 P4 (:14 P4 Pr Pr a. Pi4 PA Pr Pr Pr Pr Pr Pr Pr 155

nn+ + + +n +nnno no nnn =lx saz • • • • • • • • • • • • 00 0 ONH4-HH000

adSq. o• no• no nnonnnnnno• • • • • • • • • no• • -oToauas NH 0 ON 04- ON rn-1- tnON

000 nninn0n000 00 n n n auuTweap • • • • Il • • • • • • • • • • • • In H CO ...1- -.I" r4. l'4- -..1" H CT. N r-4 (:) 00 4- n0 HHHN HNHH n00n0000n0n• • • • • o n n n 0 n avaps -pciaoLidna • • • • • . H .c) HO N H N d n4- 0 N nnnnonnnnnnnoonno asa3r3aadS0 • • • • • • • • • • • • • • • • • 0 0 n..t HnOHHNO0 CO C\I H auTds-moT nnnoonnnnnoonnnnn • • • • • • • • • • • • • • • • • n n n aul. -psoclizioo N NNNH re -0 N N vD Ch cc 0 H N HHH HNH O 000 nO nO nnn nnnnnn • • • • • I • • • • • • • • • • • auTds-1,14 -pti U\ re\ tn ,.ON400%.0%.000•%.0N4' ON as;Tsodiun HN HHHHH nn00 00 nnnOn0 0 0 nno tuv-ouatio • • • • • • • • • • • • • • NrnH 4- 0 NH nZO 4-W rnH'.0

• ds 00n• • ononoonnonoono• • • • • • • • • • • • -.1.-Ho onnH-d-Horntnren unITS.xoq4usz H adSq. nnoonnoononn000no • S • • I • • • • • • • • • S • • -slisuTwaaj, 4- %.o 00 W N N 0 H N H n N HHren -t H4- NNH HH H H no o00ono onnonnnnn sePT 0 aTo • • • • • • • • • • • • • • • • • snoaanb r4-\ N N N trn -t %.0 n 00 co C• H cc %.0 H HH HHHH NHH On nononoo nnnnnnnn • • • • • Il • • • • I • II • • • • as9 013.10 14 n CO O.\ f-i nHONH cc cO N N .0 N4- NNN4- reNNN HNHH SEJT4OV n000n0n000n0000• • • • • • • • • • \ •0 • NH4-4 n OH OH4-M OH Xe3iiTtre o noonoonnonoonono -dds • • • • • • • • 0 • • • • rnHH OHHNO tnNHH uoiSxo4 -ewsvH o nnnnonn0000nonnn• • • • • • • • • • • • • • .ds sTqTao • N N.\ n co H H n H 00 winavulTs noon00000nnnnono o• • • • • • • • • • • • • • saas.mil Hre\--1- N-1- NO n00KNHNNH H nonnnnonn0000000n • ds sTosoy • • • • • • • • • • a • • H 14-14- 00 tnNH H NH HO

0000 00 00 00 000000 0 • • • • • • • • • • • • • • S • • Nnon00000N-tvDonon 0 HHNtn4- n\.0%.0%.0',ONNC0 00

HNK14- n nONCOCNOHNH.1 4- n ,Z, N 000000000HHHHHHHH 0• 00000000000000 0 0 NNNNNNNNNNNNNNNNN P4P4P4O4O.O4O4O4p4O4P4O4O.O4P4O4P4 156

amount. Compositae pollen declines first, followed by grass, sedge, and Quercus pollen.

The forest section (Tsukada's G 3 unit) is unbroken and dominated by Moraceae and Terminalia-type. Other forest taxa are more well

represented in this core. Zanthoxylum, Celtis, and Bursera are more common than in core P-17.

The complete absence of the Postclassic NAP episode from the

P-20 core sequence leads me to the conclusion that something has

happened to upset the deposition in the north end of Lake Petenxil.

This is supported by the fact that there is almost a 50-cm discrepancy

between the "transition zone" in P-20 and Cowgill's core II (Cowgill

and Hutchinson 1966). This means that either sediment has been removed

or that it has slipped down slope toward the deepest portion of

Petenxil. The steep incline to the south at the core site tends to support the second hypothesis.

Core E-5 Lake Eckixil, Peten, Guatemala

Core E-5 was collected in 8.2 meters of water on the western

side of the lake basin, the deepest portion of the transect.

The sequence is similar to the two Lake Petenxil cores, since

they are from the same general area. The lower section of core E-5 is

a herb-dominated pollen sequence. Grasses, Compositaes, Cheno-Ams and other herbs predominate. Oak pollen is not as prevalent in this core as in the Petenxil cores (Table 31).

The transition zone is almost identical to that of core P-20, with a successive diminution of herb pollen types. The slower nature 157

H N 0 CO N N VID 0 te 1 H 0 0 %.0 r-I N Nt\ V) ON NOON ao1.3eAuaSTe N C\ CO 00 0 n N rc\NOcOir\c000-1-cOir\NO • • • • ••• • • • • •441441 • • • • • ••• • -apn4T -Eduly NK\NHHH T-I 0 rel NHNHNHHH Nr\ Ne\ N N \ hel 111111111111 \ Lf 00nno 1.1 n n - +LACDOon 11\ cr onnonoo Zw vez • • •• • • • •• • • • 4,041 • • 5 00OO• 0 0 0 r-4 HNOHO OH 141-1- HNNK\NNNH

ed.K4 O 00 \ Llsn 0 00 n if"\ U.1 0 L(LrO u- non000 Ir\ 11\ 0 • a a a 410•0004100001,00 0 4400 -otoeuas Ht(1 H \ K\ H te\ 0 te\ N \ -1- V) 0 Ncl H 0 tel -1- n • Lr n 0 LI1 crn o o non000 0 o noonoo 000•0000 • 00 • 0 n 0040000004100 GETI TUre .1 D CO 1- K\ O N 0 H \-0 H V) H L11 tel CO U n N ICT\ N 1-4 r-I CO O o O'N N H -1 H H H r-INNHr-INNN 0000 \ 0 LI\ 0 U \ Cr\ LI1 LI 1 V\ 11-\ Onoonnnn eveosTqaoRdna • • • 110000005000s 11000000$ IA 0 \ Nt\ N N N N C\J Lf\ H H VD 1.11

nnonnnnn Ir\ o cr..n o o IA IA nnnonn assovaadED • • • • • ••• 4 • • • ••• • ••• • •00• N te\-.1" t(1 t.11 Cr\ ....1-nreNo-o- \ N O N N N H H Lf1 euTds-MOT IA IA 111 IA O nnnon LS-N nnnnnnnn000nn • 00• 5 00•0•0• • • • • • • 0 4 • • • • ael. - 0 N r\ O n psodtuo K Nr\ .0 N N H CO 0 CY ()", ',D H N N N \ Ne\ H H HNNH r-I H H N au -pds-t13 -pq Lf`N \ LI 1 Ir\ 0 \n000 IA O If Lr\ 0 Lr\ 0 n • • • ••• • ••• • • • • • • •000 ••••••• L1- 0 Ir\ •0 as; Tsoduio ‘..0 0 NNNNLI\NNI--1 n.00r-trn Lrn r-4 VO CC N C -t H H H NNNH HNNN O 00 o o IAL'O Lr\ u"\ Lrl 0 IA 111 0 L\0 0 0 0 0 0 0 -ouatm • • • • • • • 0 • 11041 • 4, 041 • • 4 • • • WV n r•r -t O N te1 r.r\ re1 H N N vD H -I- N N H -1- rr\ C`-- H

ede • LI1 0 0 V\ 0 0 0 0 0 0 0 0 0 0 LI\ 0 0 0 0 0 0 0 Ir\ • • •• ••• • sa • • • • • • • • • • wriTILxotn.t•ez tr1 N H -1- VD H N f-i H 0 H N H H 0

acUq O 000 0 nnnnnnnn 00 00000 00 non O 000000600005090400040400 -sTreu -piaaL Cf\ 0 1"r1 O N Lr\ Cs- N-1- H-1- 0VDNLI\NNNK1Nr\-.1- Pr\ N \-0 H H r-I H H H H sapToaTo nnnnnnno Lr\ 0 L.1-\ IA Lrl not:Nnnnnou-Nonn O 400490040000040•011•09000 snoaan6 IA IA 1.4-\ oo N \.0 tr\ N CS\ N 0 00 c0 re\ N \LI N 0 N H H nnnnno 00nn 0 o ono IA nonn0000 Lr\ alaeo -e.tow O 01108000s 4,0011040•0110110000 lf\HOC\--tH CC IA O N 0 LI Nr\ tel H N 00 Crn H N 00 %..0 H Ne\ N IA -1- -1- tel te\ K1-1- Ce \ \NHNNN

seatio-e tr\ 1.11 Ir\ o IAIA IAO Lrn IA0 00 IA0 Lr\Lr0nonoo O 000000004000000000 • n v-reNITuvii • tr Nr\ N t-I H 0 t-i NNK\HO 0 0 H 0

• dds o IAIAIAO o Ir\ n 0 Ir\ IS1 0 00 V\ 0 0000 if\ trn n • • • • 5 000•0• ••• • s •••• • r\ O -1- te\ H n.0 N H V) tel rel K\ N N H c--1 Ne\ 0 H

n 1.1-N crn nnonoonon 1-rn o Ir\ n 11-n n irn 0 00 • de sT4TaD • ••• • 110• •00•0• 4 • 41414100 • • tA N. ....1- L.r\ trl -I' 0 -I- O \S) N Nr\ 11\ HOHN H H n n sqn.remTe 0 Cr\ 0 LI\ 0 111 \ 0 le \ o tr n o u- n000 n 11.\ 0 • • •0011 •S• •04 • • • • • • • 0 • • H H H sassa-ma N N H H rel N 0 0 H 0

O IA0IA00IA00 0 0000IA0 00 IA 0 0 0 0 • 0 • • ds stosolir • • • •414/00• •0040 • te \ 0 HO H NHN NHNHNH HNH HOHO

0 0 0001-110tr\Vn 00000 0000000000 • • • • • • • • • • • • •0•00•0• • • • • 0 N v0 CO N IA CC N H rel N V0 CO CO 0 H N Nr1 V) CO O N V) 00 H H H NNNK\ LC1 %.0 VO V) V) ‘40 %.0 NNNNN

N rc.\ IA '.0 N cc C\ o H N IA '.0 N CO CY\OHNK1-1- O 00000000H H H r-4 f-I r-I r-I HNNNNNN n n g Ir\ LC\ Lr\ IS\ cr\ n LC\ d\ if\ if\ cr nnnnnn Lr LC\ LC\ Us\ II\ rzlfArzlrilWWWWWWWWWWWWWWWWWWWWW 158

teN k..0 .10409ATIOSTe C•--cc • • • • • --aptil.Tiduly re\ N N N Pe\ 11111 O 0 irn o s.Estu saz • • • • • tc\ N 14-\ N N

edit; uor c • • • • t\ '..O N L.rn o irn evutare.ip • • • II • •• N-c0 N N H H N \ 0 trn trn eve os -Fq..totldna • • • • C7N N

0 LA 0 Lr\ a -saosaadiço • • • • • N N H

auTds-inoi O 0 l.f\ • • • • • as4Tsodtuo KV.° 0 H N r-I H N

au - d-742 TT{ LC1 1.1.1 tr.\ o • • • • • aE.4 -Fsoduloo N n N N N

Lr\ 0 it\ Lt.\ 0 uiv-oustip • I • • • K \ H N • ds o o trn trn Lrn • • • • cantExotRusz H H H ad.k4 n o Lrn • • • 0• 0• --E -FtsuTaaaa, o N H

s apToaTo Ir\ n o o U. \ • • • • to snosen.b H N H 0 Lrn Lrn tr\ avao -e.zow • • • • 0 0 r'rs\ O.\ H H r-4 SEXCIOV 0 V\ 0 0 0 • • • va-exTprew HOHH

• dds u-n n n n o • • • • uoTS.xo4 -eulasH ooHN Lr\ o It\ o • • • *ds sT4TeD O r-I N n n000 eqn.retuTs L.r • • • • saasang 0 r-I H

O 0 0 0 0 • ds io3y • • • r-4 H

• 0 0 0 0 0 • • • • 0 LC1 N %.1:) • CO CT\ Cr, a) co ON 0 P-1 NNNNK\ E as irn L.C1 Us\ Lrn 159 of the transition zone in this core may allow a palynological view of plant succession.

Above 60 cm the presence of high percentages of Moraceae pollen,

and that of Terminalia-type indicate a surrounding vegetation similar

in composition to that surrounding Postclassic Lake Petenxil. Pollen

of the taxa Zanthoxylum, Bursera, Manilkara, and others (Table 31) also

show an increase.

The tree dominated flora of the Postclassic is interrupted at

about 23 cm by an increase in herb pollen, similar to that in Core

P-17.

Core E-8 Lake Eckixil, Peten, Guatemala

Core E-8 was raised from 8.3 meters of water in the western

section of the Eckixil basin, approximately 60 meters to the east of

core E-5.

The sequence is virtually identical to the previous core,

except the sedimentation rate is apparently slower at this site (Table

32). For some reason this core had the smallest input of Quercus

pollen of any taken in the lake region of the Peten.

The sequence for core E-8 is so similar to E-5 that it will not

be discussed here.

The Pollen Sequence of the Lake Region of Peten 14 By cross-dating to the C dated lake cores (Cowgill and earliest pollen data Hutchinson 1966; Deevey 1976), it appears that the Preclassic, in the four cores that I collected are from the Middle 160

U1 ill if\ Lr\ 0 0 U-N 8 o tr\ o n is\ o n o n n Lr1 0 ••••••• •••••••••••• 0 0 0 H H H N H H N H H K1 N N (1.1 0 N H

O 00 0 n n o 000 n n o Lr'\ Lr\ Lrn 0 o n o • • • • •• ••••••• 0 • N H ri N K1 -1- \D 0 re1 H N CC 0 n-.0 -d-

11\ 11\ 0 U1 111 Ir\ 0 U\0 d\ 0 Ir\ LI-\ Ir\ o 4'\ If O 0 a -eu -pire-1D •••••••••••••••••••• lf\ K1-1- N O -1- CO -I- \D K1 0 K1 N CO 01 H ri CO U1 H HHHH H NrirINNNNNNN

- n n O 0 Ir\ L(' \ Lr\ O Lr' 00 tr t.r\ Lrn O Lr\ Ir\ LI-n Ir\ Q Lr\ Ir\ a -sa o -e -Fgaorgdna • • 11••••••••••••• •• n.0 te) N -I- N r‘n \.0 \.c) n -t

1Ç\0 0 U\0 U1 0 111 U1 0 00 111 0 0 111 U1 0 If\ Irn aeaveaadS0 • . • • • • • • • • ••••• • •• \D -1- -1- K1 H V) r-‘n ren co N N F U1 \D 0 N-

auTds-moT n O Lr\ 0 Lr Lr\ Us\ Ir\ If\ nonn00 Lrn Lr\ 0 111 Lr\ • • I ••••••••••••••••• a -eq Tsodwo v::) O rr\ 0 N 00 H N N Cr\ U N N HHHNNNNN out ds-1.12 Ttl O 0 0 0 0 U\0 tr\ o u-n nonoonnn0 • • • • I/ •••••••••••••• au;Tsodwoo N \-10 0 141 N Ls-n \-0 H -t K1 H CO \D 141 N H riNrINNNNNNNN

LI1 0 111 0 U1 0 U-1 0 tr\ o o tr) Ltn u•-n Lrn 0 00 ILIV-OUaT.10 • • • ••••••••••••••••• \ -I- N 141 H -I- H K1-1- N -I- CO N OONNH

n • cis Lr'N UU0 u- o 00 Ir\ 000000 11\ • • • • • • • • • • e • • • amTS.x0t1;usz O n-c) H ri %.0 kID H ri 0 H ecik; L.r\ n 00 n trn 0 Lrn 0000 nnoo tr\ • • ••,•••••••••••••••• - -eTteuTul.zaj, • 144. 0 \ 01 N O -1- \.0 N N N H K1 N Pr\ N H C•-• ri H ri H ri rl

sap -paTo LI1 tr\ u-N 111 0 LI\ LI"\ LN Ir\ 111 n 111 111 ir\ Lrn o U1 U1 • • • • • • ••••••••••• •• snoaanb NNLI1U1-tNONNN-1- -1- K1NrirIN n u-N u- \ L.r\ n Ill U1 1.11 0 V\ 0 U1 Lf1 u 111 u n a-eaovaoK •••••••••• •••••••••• N- H ri K1 H N 01 N 14-1 Ch 00 -I- CC N U\ K1 U1 N -7- -1- rr1 K1-1- N

svatpu U1 0 U1 isn L.1-n noono tr\ o 1.11 n Ir\ irn 000 • • • • ••• ••••••••• .E.IV3r[711-8 • trn N H N ‘.0 Kl -I- 000 ri

n • dds LrN Lr n - • • • 0 tf•Lr\ ••••• 0 U\0 0 0 0 0 0 0 0 0 0 0 Li \ uoiSxoq. -eulavH 0 0 --.1" trl N

n O lx\ onono u-\ n Lr\ Lr Lr\ U\0 is\ 0 • ds s -F;Tao • •••••••••••••• • • re\ es- 1.11 LS1 Lf1 n0 N O'N n0 H ifn -1- H 0 0 0

n n -eqnseurps 0 0 V\ 0 tr 0 tr trn tr\ 0 tr\ 0 0 Lrn 0 0 0 Ir\ 0 •• • •• • •• • •••• ••• vaasang Cs- Lr\ NHHN N H ri -1- H ri 0 rl

Lrn L.rn 0 0 trN 0 if\ 0 000 trn If\ 0 LR 0 0 0 Lr\ o ••• •• • cis et o -e y • • • • • • ••• HONNN 1/1 0 H H N 0 Hs-4

O 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 • 111116 ••••••••00 ••••• • N O Il\ 0 Lf\0 NO noN \D 0 Lrl 111 0 Tl —1- n n HHNN 141 r(1-1- tA Lc-\ L11 U1 \-0 .0 N N 00 CO O' n o•-n n-• •- n 1--1 K1 N -I- \D CT N- CO 1/4.0 t•r n•-•• H H H rIHNNN ...vs..., N., 0 H N C`-- cc c• o H \I r4.\ u-\ 1/4.0 LN- 00 000000000HHHH HHHHHH N cc c0 cC cc c0 cC c0 cC c0 cc c0 00 00 cC c0 cC cc cc cc r4 rx1rilrilr4WWWWWWWWWWWWW41 161 about 500-600 B.C. (core E-5). Core E-8 has a probable base date of

around the birth of Christ, as does core P-20. Core P-17 probably commences during the beginning of the Early Classic, around A.D. 300. Therefore, the information from the Preclassic is less than that for

the Late Classic. The high internal consistency of the four Prehis-

panic cores increases my confidence in using but one core for the Pre- classic.

There is a high amount of herb pollen input into the lakes

during the Middle Preclassic period, mainly grass and Compositae pollen

with some Cheno-Am and other herb pollen. The most common herb is the

low-spine Compositae, and the most common tree pollen is oak.

The pollen of the Late Preclassic is an herb-dominated regime,

with grass pollen as common as that of low-spine Compositaes.

The Early Classic period in the Peten has little change in the

deposited pollen flora, while there is an increase in human occupation of the lake region (Rice 1976).

The Late Classic period witnessed rapid changes in the per-

centages of several herb taxa. These include Gramineae and high-spine

Compositae types, but the tree pollen shows no such fluctuation. The overall herb pollen input remained fairly stable, then during the second half of the Late Classic, began to decline. Just below the transition zone there is a slight but across-the-board increase in herb pollen.

The transition zone is a quick (in palynological terms) decrease in herb pollen and a rise of tree pollen. High-spine Compositae is the first pollen type to decline, followed by low-spine Compositae. 162 Graminae and Cyperaceae pollen persist longer than other herbs but then rapidly collapse.

The Postclassic„ from A.D. 900 until today, is dominated by

pollen of Pinus, Moraceae, and Terminalia-type. Many insect pollinated

tree taxa show an increase at this time as well (Tables 29-32).

Perhaps the most interesting short-term event occurs during the

Late Postclassic, about A.D. 1400. This event is a distinct rise in

herb pollen that occurs in three of the four cores discussed here, and

both published Petenxil cores (Cowgill and Hutchinson 1966). This

event will be discussed in more detail in Chapter 9.

Principal Components Analysis of Cores

Principal components analysis of the fossil pollen data was

performed using the same seventeen taxa and pollen sum as the analog

data. The difference was that the samples were arrayed by depth (or time), instead of a derived gradient.

Two criteria for the use of core sample data were discovered

during analyses. There must be a significant change through time in

pollen percentages in the core if statistically significant and eco-

logically meaningful eigenvectors are to result. As an example,

Petenxil core P-12, of wholly Postclassic age, was subjected to prin-

cipal components analysis, and the lack of change was apparent in the results. The eigenvector weights assigned to pollen types made no sense in terms of either the gradients or habitat of the plants producing each pollen type. The pollen taxa showed no significant co- variance, and the computer could derive no significant pattern. In 163 addition to the criterion of necessary palynological transition, each distinctive zone of the sequence should have approximately the same weight, or numbers of samples in the analyses. This criterion increases the ecological clarity and explained variance of the resulting eigenvectors.

Cores P-17, P-20, P-12, E-5 and E-8 all produced first principal components that were similar in amplitude and eigenvector weights

(Table 33). The weights are statistically similar (Table 34). The correlation matrix is evidence that all cores are manifesting essentially one pattern, a strong, coherent palynological signal that ex- presses the different worlds of the Classic and Postclassic Maya Low- lands.

The pattern is a change in the first principal component from high negative amplitudes in the lower sections of the cores (Classic

Period) to high positive amplitudes in the more recent sections.

Although each important eigenvector has slightly different weights on the 17 pollen taxa used in the analyses, the data are quite similar. The pollen types that have high negative weights (and describe Classic Period pollen productivity) are as follows, proceeding from most important to least: Low-spine Compositae, Zea mays, Graminae, high-spine Compositae, Chenopodiaceae + Amaranthus, and Senecio-type.

All of these taxa are herbaceous types.

The pollen types that have high positive weights and express

Postclassic pollen deposition are as follows: Moraceae, Terminalia- type, and to a lesser extent, Manilkara, and Zanthoxylum. These pollen

• •

161+

.0 b co Q) o h CO 0 N h S. 0 O c.) 4 h 4--) h .. -0 • 5. Ca .4-) b co +) C) O u) 1.4 h .r. o O h r0 Cl) o 4-1 h O h h . O h 0 h ni o p• ,0 cd h ca th h h b h a) -P bip • > h IO Cl) (1) h O CO CO 4-) Cl.) CI) CD CO CL) . 0 b 0 JO O 0 0 0 . 5-1 0• 0 O N N N g, O b ts1 O E 4-) U) 0 a) a) (ii cO o h o o c.) 3 •ri C.) O 4 4 h E P h cd h h .., .., ,0 h - .0.0 -5.4 5-, ;-n -1,-) S-( C) 4 P b •ri (1) b 0 •T--1 4-) 4-> -P ,h 0) 4) 44 U) CO c0 cO c.) rt) (1) •r4 • H 9-1 c6 O -ri -H Z 1=1 1=1 =1 (1, G7.4

H 0 n.0 00 0 LI\ N- CN O C• H C• 0 ON 0'. NI N• re1• H• 0• NO• • Ir\• H• HO• • Lr•n • • • • rr\• O 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 I I 1 1 1 I 1

N H C• -d- N N HO N00 r.rn %,..0 H OH 14" \ U\ 0 rtn N H %...0 H Ll-sn r=1 • • • • • • • • • • • • • • • • • 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C• N r4'1 N Cr% N r+1 0 H -I' N 0 -I" 0 H -I' r-4 4+1 n-0 H N 1.41 N n 0 -d- • • • • • • • • • • • • • • • • • O 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 I I I I I I

-I- 0", r-4 N- \-0 Cr H \ \ H H 0 O• te\ H H N H H teN 0 N ren te\ Cs-. 0 -I- N- H H ‘.0 ‘.0 • • • • • • • • • • • • • • • • • O 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1

n -d- N H N N 00 00 r41 0 N c\J 1/4.0 N C• O• H n -.0 n O N --t• H -d- L) %.L) co Lr \ 0 VD 0 N N • • • • • • • • I S • • • 41 • S • 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1

• h P4 h CO .0 PLI h 0 0 • z co Me a .0 •rt h b c.) ct 4-, E o as a) t E • 9-1 • ,-.1 H h 0 • P4 a) a. P^? cd 0 •rf H te\ U1 Co x • E rt• h 0 cO rg X 4:4 h h 4-> ,. 0 h 0 1 a) • .4 h H O 0 -r-I .4 0 o 0 CO .4-> E .1-1 b Pi E 4-, b h HP 1-1 a) h C) 5. h a) C.) C) h h b 0 h ..h E4 .:o Pcn 0 Z 0 E N C.) 165

Table 34. Matrix of core eigenvector correlation coefficients.

Core Matrices r p Core Matrices

P-17 x P-20 0.95 0.01 P-20 x E-5 0.85 • 0.01 P-17 x P-12 0.85 0.01 P-20 x E-8 0.83 0.05 P-17 x E-5 0.96 0.01 P-12 x E-5 0.51 0.05 P-17 x E-8 0.77 0.01 P-12 x E-8 0.54 0.01 P-20 x P-12 0.32 E-5 x E-8 0 .84 0.01 166 types are all from trees. The other seven taxa of the statistical set were, in general, insignificant in most principal components.

The pattern, which accounts for 28-44% of the total pollen variance, is a change from a berb dominated pollen flora during the

Maya Classic period, to a tree dominated Postclassic. The precise

nature of the two floras can best be described by comparison to the analog gradients described in Chapter 7.

Application of the Analogs to the Fossil Data

The method used for applying analog data to fossil core data

may be called the eigenvector regression method, and as far as is known,

this is the first use in paleoecology (Fritts, personal communication

1977). As discussed in Chapter 7, principal components are independent

variables in a sense, and therefore are amenable to statistical analy-

ses. This method is similar to Canonical Correlation analysis (Webb and Bryson 1972), in that is assesses the similarity between two sets

of patterns and allows the researcher to convert one set of patterns into another via statistical transformations. However, eigenvector regression allows the formulation of hypotheses, a rriori, from either modern or fossil pollen data, then to test the hypothesis in a mode similar to the standard scientific method. Patterns can be shown to be similar or dissimilar, and an index of certainty can be assigned this decision. The researcher is able, therefore, to be closer to the flow of data through his analyses than with canonical correlation analysis. The results of the two methods are different and the data from eigenvector regression are not as elegant as canonical 167 correlation, but for an unknown area like the Peten, the more robust method is best.

For purposes of comparison, the longest core from Lake Petenxil,

core P-20, and Lake Eckixil, core E-5 will serve as the test for the analogs. The null hypothesis for the test states that there is no significant difference between the two populations of eigenvector

weights from which the samples (Table 33) are drawn (Simpson, Roe and Lewontin 1960).

H = o There is no statistically significant difference between core eigenvector 1 in P-20 or E-5 and eigenvector 1 of ANALOG 1.

Pearson's product moment correlation coefficient was used for

the testing of the null hypothesis, as stated above, since its use is

standard (Simpson, Roe and Lewontin 1960). The correlation coefficient

for ANALOG 1 and core P-20 is -0.34 with an associated probability of

greater than 0.1. These data indicate that for this pair of eigen-

vectors the null hypothesis must be rejected if the standard 0.05 level is used (Simpson, Roe and Lewontin 1960).

The correlation coefficient for ANALOG 1 and core E-5 is 0.28

with a probability of greater than 0.1. The null hypothesis must be

rejected for this pair as well.

The meaning of these statistical manipulations is that the pattern of change in fossil pollen from the Maya Lowlands does not reflect vegetation change that is similar to the edaphic gradient.

Therefore there was no significant alteration of grasslands or savanna with forest since the Middle Preclassic in the Maya Lowlands. However, 168 this fact must be used with caution since variance caused by natural vegetation community change may be masked by the influence of the factor causing eigenvector 1 in the cores.

Since the null hypothesis has been rejected, a second hypothesis is proposed.

H = 1 There is no statistically significant difference between core eigenvector 1 in P-20 or E-5 and eigenvector 2 of ANALOG 2.

The correlation coefficient for ANALOG 2 and core P-20 is 0.68 with a probability of between 0.01 and 0.001. This means that the second hypothesis is accepted at the 0.01 level.

The correlation coefficient for ANALOG 2 and Core E-5 is 0.916 with a probability less than 0.001. The second hypothesis is accepted at the 0.001 level.

The similarity of ANALOG 2 to the fossil pollen data has now been established. The major pattern expressed by the fossil pollen core spectra is a change from a weedy agricultural regime to a forest- dominated Maya Lowlands. The statistical demonstration of this fact is of prime importance since it allows a much more meaningful reconstruction of the Prehispanic Maya landscape than has been made in the past. CHAPrER 9

RECONSTRUCTION OF THE PREHISPANIC MAYA LANDSCAPE

In this final chapter, I wish to combine the palynological data

described in Chapters 5 and 6 with the ecological and agricultural in-

formation in order to describe the vegetation history of the Maya Low-

lands. Pollen analysis, while yielding much insight into past ecosys-

tems, presents, at best an incomplete picture of prehistoric man-land

relations. When combined with modern vegetation ecology, pollen pro-

ductivity, and agricultural ethnography, palynology is a strong tool

for reconstruction. From modern analogy, we know the probable makeup

of the prehistoric forests and fields. We can guess at the probable

interconnections through systems models, and we can assess the validity

of the reconstruction through archaeological and palynological remains.

The presence of Zea pollen in the cores indicates that Indian

maize was raised in the environs of Lakes Petenxil and Eckixil. The

absolute pollen influx, however, is much greater even in the Preclassic

than that predicted by any run of the simulation, including total maize

monoculture. Pollen transport mechanisms other than wind must there-

fore account for the discrepancy. Slopewash from fields surrounding

the lake is the probable cause, and is evidence for maize agriculture within the watershed of the twin lakes. The nature of this maize agriculture is less clear. Significant correlation of Zea pollen with

169 170 disturbance weed pollen is evidence that the fields were not cleared of weeds in the manner of modern agriculture in the United States. Tre-

mendous overrepresentation of herb pollen may, however, render the weediness less real than apparent. If a short fallow period was used,

say 1:1 cycle, then the fields may have been intensively fertilized and weeded, but overshadowed palynologically by neighboring fallow plots.

Another scenario that fits the palynological evidence is that well-

tended, intensive fields existed, but that weed-filled zones occurred

in field borders or in areas of marginal usefulness.

Fragments of carbonized plant material are preserved in lake sediments and presumably have a relation to fire frequency in the vegetation surrounding a lake. In the lake cores, there is no significant correlation between eigenvectors that describe agriculture and the relative frequency of carbonized fragments. This test of significance is evidence that traditional swidden agriculture may have been less likely than an intensive system as the actual form of prehispanic Mayan agriculture. Most fragments are probably grasses and sedges (Wiseman

1974). A distinct possibility is that the nature of the linear vascular tissue in monocot leaves renders them more easily preserved as charcoal than the more irregular dicot leaves. This may result in a form of overrepresentation of grassland fires over woodland (or swidden) fires, masking somewhat the presence of slash-and-burn agriculture.

Although arboreal pollen is suppressed throughout the lower section of the Eckixil core, overrepresentation of herb pollen may mask considerable expanses of forest or orchard. Comparison of the prehispanic palynological evidence with modern data (such as the Pseudobombax 171

overrepresentation example cited above) indicates that up to 50 percent of the lake region may have been covered by forest, although alter- nating with open, herb-dominated vegetation. These data are compatible with models of random cultivation, artificial rain forest and the wood lots necessary for fuel. The output of agricultural simulations,when compared to the palynological evidence, allows critical assessment of models of the late Holocene landscape (500 B.C.-A.D. 1975). The following hypotheses were rejected as monocultural techniques:

1. Artificial rain forest: weed pollen from the cores is much in excess of that from the simulation output, while arboreal pollen is below that predicted.

2. Intensive "weedless" agriculture: simulation of intensively weeded cornfields produced a ratio of Zea/weedpollen much greater than that found in the core pollen spectra.

3. Swidden agriculture: no significant correlation between carbonized fragments and other agricultural indicators in the core sample matrix.

4. Ramon cultivation: weed pollen from core samples in excess of that predicted by simulation; no statistically significant change in the frequency of the Brosimum-Cecropia pollen type through time.

Rejection of these hypotheses mean only that they were not used as the sole subsistence base of the Maya. Such analysis, however, leaves several options for reconstruction of an agricultural sequence for the lake region of Peten. 172 A Preliminary Agricultural Sequence for the Central Peten

Middle and Late Preclassic, 500 B.C.- A.D. 250 (Base of Core)

According to the simulation/palynological comparisons discussed

above, maize agriculture covered the watershed of Lake Eckixil as early

as the Middle Preclassic (Fig. 17). Such evidence supports an early

riverine-lacustrine adaptation such as that postulated by Puleston and

Puleston (1971), since the lake's immediate watershed was then as fully

utilized as during the Late Classic. The presence of abundant weed

pollen during the Preclassic and the general lack of tree pollen indi-

cate abundant secondary successional herbs and shrubs of the sunflower,

pigweed, grass and potato families. These Preclassic data palynologi-

cally resemble old field communities of the modern Peten, which are

tangled, thorny, economically useless shrublands from one to four

meters high with an occasional tree spared from the forest. Whether

these reconstructed secondary scrublands existed as part of a fallow cycle, as border communities, or independent from, was sporadic during the Preclassic, but the presence of generally higher percentages of carbonized dicot fragments hints of significant woodland fires. It is tempting to combine the maize and weed pollen data and that of the carbon fragments into a swidden-type reconstruction. Indeed, uncriti- cal comparison with the modern pollen data, in which swidden agriculture is a component, yields a Preclassic dominated by slash-and-burn agriculture. Although an early extensive system, slowly developing into intensive methods, is close to the heart of many researchers, 173

,!ssopisod /AssolD e4o1 ,!ssolD Apog D!..P...c1 04 0 1 9 D.cl .IPP1 I IIIII 1/11111 IIIII II O 0 0 O 0 0 0 0 0 0 o o O 0 0 O 0 o 0 o o co ,r N ry .":.' cnI oc .- v.- 2 ,.., co JLva31V1NIX0ôddV 174 diachronic analyses of the agricultural eigenvectors and charcoal frequencies render such an interpretation equivocal. In summary, I con- clude that the Preclassic landscape in the immediate environs of Lakes Petenxil (original Preclassic data from Cowgill and Hutchinson 1966)

and Eckixil was a mosaic of cornfields and scrublands, with occasional houses or hamlets, with a population of between 25 and 60 persons per square kilometer, based on data from the neighboring Lake Yaxha region

(Rice 1976). This reconstruction is not unlike the scenario generated by the simulation for the Preclassic.

Classic Period A.D. 250-A.D. 850

A combination of techniques (at least some of which were charac-

terized by weedy undergrowth), or extensive short fallow maize agricul-

ture, without significant burning, are two options that seem to fit the Classic period data.

The human population outputs from the simulations and pest or nutrient problems negate, to a certain extent, the short fallow maize agriculture model. The population able to be supported by a 1:2 and

1:1 cycle may approach the archaeological population estimates if all upland forest is eliminated. The resulting nature and distribution of fields, however, is incompatible with modern plant pest and nutrient loss data as described in the first section of this paper.

Secondly, since this maximal model requires the virtual elimination of upland forest, the predicted loss of the characteristically underrepresented arboreal pollen does not occur. I therefore reject 175 the short fallow model as the sole agricultural method used by the Classic period Maya.

In summary, it appears that no single agricultural technique, either actual or hypothetical, will satisfy the requirements of the demographic, ecological and palynological data. The remaining option is the most likely when the derived simulation results are compared

with real data. That is, a suite of agricultural techniques, each

fitting into its own environmental niche, with differing "intensifi-

cation" and productivity, is the valid model of Classic period sub-

sistence. The maximal habitat model for the Classic period central Maya lowlands is supported by the following data:

1. The simulation of the "maximal habitat model" (Wiseman 1976) supports populations similar to those derived from settlement surveys (Puleston 1974).

2. Pest incursions are limited by differing crops, harvest times, and land use.

3. Erosion and nutrient loss are limited by terracing (Turner

1974) and retention of tree or shrub cover.

4. The combination of weed and Zea pollen indicates maize agricul-

ture, probably accompanied by either fallowing or other weedy plots,

is one agriculture technique of the "maximal habitat" system.

5. The presence of terraces in the Peten as described by Turner

(1974) is indication of the use of a second technique.

6. The correlation of Brosimum alicastrum (and other tree crops) and some archaeological sites may be evidence of a third technique. 176 7. Raised fields, which may exist in the environs of Tikal (Dahlin 1976) are possibly a fourth technique. 8. No agricultural technique known will meet the criteria of the simulations and ecological evidence.

The "maximal habitat" hypothesis has been evaluated through its

ecological, archaeological, mathematical and palynological implications.

It has met or exceeded the requirements of the modern and prehispanic data.

The Classic period reconstruction presented above fits, in general, the modern consensus of archaeologists as presented at the Prehispanic Maya Agriculture symposium in Paris, 1976 (International

Congress of Americanists in press). These data and reconstructions support a polycultural model of Mayan Classic period subsistence.

The Terminal Classic Maize Episode

The results of palynological analyses from the central Maya

Lowlands indicate an episode in the agricultural history of the Terminal

Classic (A.D. 750-850) that may have a bearing on the Maya collapse.

Pollen diagrams from Peten elegantly illustrate the change from a

Classic period disclimax (weed-dominated) regime, to a Postclassic forest (Cowgill and Hutchinson 1966; Deevey 1976; Vaughn 1976).

There is a distinct minor peak of Zea pollen in strata immediately below the change. These data are suggestive, but due to relative scarcity of maize pollen in the profiles, such change may be attribu- table to statistical error (Faegri and Iversen 1964). The error is amplified by counts generally less than 1000 pollen grains. 177

Validation of the maize peak rests on a foundation of fine scale sampling, pollen counts of 5000 grains in the critical interval, and replication of the peak in several cores. Sampling technique consisted of a series of "primary" samples 10 cm apart, using 500 grain counts, to isolate the Classic-Postclassic interface within each core.

The interface was defined as a change from herb to tree pollen dominance. When the interface was approximately located, intermediate samples were removed from the 10 cm interval in such a manner that the transition was found within 1-2 cm. If necessary, close-interval sampling extended well below the transition to prospect for the predicted maize peak.

Pollen was counted from these samples in the manner described in Chapter 4 except that a third pollen sum was added. The third fixed sum consisted of the 1500 count, plus the total "other" pollen excluded from the sum (which varied from 500-2000 grains) and the necessary the grains to complete the 5000 grain count. Pollen types completing third sum were Zea and "other." Since Zea was included in all three was derived with a sum sums, a percentage of this economic annual plant of 5000 grains. Since statistical error decreased with greater sample fluctuations in the size, such counts were necessary to validate small percentages of Zea pollen. validated The results of rigorous analyses of pollen data have in previous profiles. the Terminal Classic maize event, that appeared the Classic period, the Of the seven sediment cores that intercepted distinct in five, indistinct in one, and absent in one. maize episode is 178

Dating of the maize event is by cross-dating the rapid change 14 from old-field to forest pollen regimes, to identical C dated se- quences. Those profiles show the change to correspond (within the 14 accuracy of C dating) to the general abandonment of the Maya Low- lands in the 9th century A.D. Such an event of high intensity and short duration functions as a dependable "horizon marker" or index fossil for pollen profile dating. Since it is this very event that is the focus of these analyses, the date of ca. A.D. 850 is good.

The magnitude of the Maize Event is more difficult to describe in a quantitative manner, due to the nature of the pollen evidence.

Comparison to the modern mud-water interface pollen samples indicates that the event is approximately an order of magnitude greater than modern maize agriculture. This figure is the deviation between the maximum percentage of the Maize Event and the modern sample of each core, summed and divided by the number of cores.

9.74

P = Maximum Zea pollen during Maize Event e = Maximum Zea pollen in modern sample Pm n = Number of cores. extensive swidden Since the environs of the cored lakes are today under cultivation, especially in the region east of Flores, their watersheds to 3/4 saturated with agriculture are conservatively estimated as 2/3 tenfold increase in maize agriculture, and regenerating brushlands. A pollen profiles, exceeds the swidden as hinted by the Terminal Classic 6.7 to 7.5, an impossible capacity of the watershed by a factor of 179 figure. These data are evidence of more intensive agriculture, and perhaps total use of the watershed for maize.

The meaning of the Maize Event is less clear. The immediate environs of lowland lakes experienced a marked increase in maize agriculture during the Terminal Classic, reversing a general trend to less

Zea pollen throughout the Early and Late Classic periods. There is a less distinct increase in weed pollen in several cores during the

Terminal Classic as well.

This interpretation of the palynological data argues for a distinct change in the subsistence base of the Maya dwelling within lowland lake watersheds during the period prior to the collapse.

Whether the Maize Event is related to the cause of the collapse, or but a reflection of social upheavel, is unclear, but there is a significant correlation.

Hypotheses that have been suggested to explain the timing and magnitude of the Maize Event include the following:

1. Rapid population growth during the Late Classic led the Maya elite to change from an overburdened diversified subsistence base monoculture. (Ramon cultivation, chinampas, terraces, etc.) to maize solution, but would have Maize would have seemed a good short-term been eventually destroyed by nutrient loss, erosion, or pest incursions. in the social chaos attend- 2. Trade in important foodstuffs ceased have ordered the conversion of local ing the collapse. The elite would as cacao or copal, to maize which would fields from trade goods such directly support the local population. 180

Whatever the reason, there was a change in agriculture attend- ing the Maya collapse, the first evidence of this type bearing on theories of the decline of the Maya city-states in the 9th century A.D.

The Maya Collapse A.D.800-A.D. 900

The collapse of the Classic Maya civilization was heralded in the archaeological record by a cessation of temple construction and a dramatic decrease in the number of dwelling units. The pollen record indicates that farming intensity likewise decreased at the close of the

Classic period. Therefore, the collapse was not a hierarchical phenomenon, but a catastrophy that affected the lower levels of the Maya social pyramid.

Since people need food to live, and food production produces a distinctive palynological "marker," a quick drop in weeds must be tied to depopulation of the Peten. The lake environment, which may have been the early center of population during the Preclassic, and is certainly the most productive zone in the Peten, nonetheless, had few, if any, agriculturalists within 100 years after the close of the Classic period. The fact that such zones were neglected, even under the tur- few remaining Peten moil of social disintegration, is a mystery. The plots to produce Maya could not have cultivated more than a few small such a scant disturbance pollen influx. The magnitude of the change that occurred when the weed domi- Classic period was replaced by Post- nated pollen regime of the Maya classic forest is probably comparable to the Pleistocene/Holocene boundary in the north temperate zone. 181 The cause of the collapse cannot be described from paleoecological research alone, but the evidence from the Maize Event and the lack of early Postclassic agriculture may help the archaeologist to

find the social pressures or ecological disasters that contributed to the fall of the Maya.

The Early Postclassic A.D. 900-A.D. 1300

The Peten during the first half of the Postclassic was a forest

dominated landscape with but a sparse population of agriculturalists.

Maize pollen was sampled throughout this period in trace amounts only.

There was a continual change in the tree dominants in the palynologi-

cal record, indicating perhaps a slow secondary succession toward the potential climatic climax of the Peten.

Neither in the archaeological nor in the paleoecological

records do we find great evidence for a continuum of Maya occupancy.

The Itza Intrusion A.D. 1350-A.D. 1697

From the later portions of the Postclassic record comes a short

lived increase in agricultural indicators (Fig. 17). Maize pollen and

disturbance plant pollen increase to levels approaching the low values

for Classic period pollen productivity. This episode is probably cor- related with the recolonization of the Peten by the Itza from Yucatan.

The Itza were a lacustrine people, as their sites are often found on islands or peninsulas. Therefore their fields were probably near Lakes

Petenxil and Eckixil and produced the distinctive weedy aspects of the

Postclassic pollen spectra. There is nothing in the palynological 182 evidence to determine the agricultural method used since the event is so short, but it resembles the Classic spectra. After 1697, the indigenous agriculturalist became reluctant subjects of the Spanish crown.

The Colonial Period 1697-

After the conquest there was another rapid decrease in agricultural indicators, similar to that of the Classic/Postclassic transition, after which the forest became ascendent. There is little evidence for extensive agriculture until the uppermost samples in the lake cores.

Apparently the Colonial Peten was a sparcely inhabited forest until the recent influx of Highland Maya have begun, once again, to transform the Maya Lowlands into an agriculture-dominated regime. LIST OF REFERENCES

Adam, D. P. 1974. Palynological applications of principal component and cluster analyses. Jour. Res. USGS 2(6):727-742.

Allee, N. C. 1962. Measurement of environmental factors in tropical rainforest of Panama. Ecology 7:273-302.

Anonymous. 1975. Underexploited tropical plants with promising economic value. National Academy of Sciences, Washington, D.C.

Bartlett, A. S. and E. S. Barghoorn. 1973. Phytogeographic history of the Isthmus of Panama. In Vegetation and Vegetation History of Northern Latin America, Alan Graham, Ed. Elsevier, Amsterdam 203-299.

Bartlett, H. H. 1935. A method of procedure for field work in tropical American phytogeography. Carnegie Inst. Pub. No. 461: 1-25.

Beard, J. S. 1953. The savanna vegetation of northern tropical America. Ecological Monographs 23(2):149-215.

Benson, L. 1959. Plant Classification. Lexington, Mass.: D. C. Heath & Co. 688 pp.

Billings, W. D. 1970. Plants, Man and the Ecosystem. Blemont, Cal.: Wadsworth. 160 pp.

Birks, H. J. B., T. Webb III and A. A. Berti. 1975. Numerical analysis of pollen samples from Central Canada: A comparison of methods. Rev. of Paleobotany and Palynology 20:133-169.

Breedlove, D. E. 1973. The phytogeography and vegetation of Chiapas In Vegetation History of Northern Latin America, Alan Graham, Ed. Elsevier, Amsterdam, 147-169.

Bullard, W. R. 1960. Maya settlement pattern in northeastern Peten, Guatemala. Amer. Antiquity 25(3):355-372.

Cain, S. A. and G. M. Castro. 1959. Manual of Vegetation Analysis. New York: Harper. 325 pp.

Coe, M. D. 1966. The Maya. New York: Praeger. 252 pp. 183 184 Coursey, D. G. 1967. Yams. London: Longmans, Green. 236 pp. Covich, A. and M. Stuiver. 1974. Changes in Oxygen 18 as a of long-term measure fluctuations in tropical lake levels and molluscan populations. Limnology and Oceanography 19(4):682-691. Cowgill, U. M. 1961. Soil fertility and the ancient Maya. Trans. Conn. Acad. Arts and Sci. 42:1-56. • 1962. An agricultural study of the Maya Lowlands. Amer. Anth. 64 :273-284. Cowgill, U. M. and G. E. Hutchinson. 1966. The history of Petenxil, Laguna de a small lake in northern Guatemala. Conn. Acad. Arts and Sci. Mem. 14:126. Culbert, T. Patrick. 1974. The Lost Civilization: The Story of the Classic Maya. New York: Harper and Row.

. 1976. Personal communication, Department of Anthropology, University of Arizona, Tucson.

Dahlin, B. H. 1976. Preliminary investigations of agronomic potential in bajos adjacent to Tikal, Peten, Guatemala. Paper presented to XLII Int. Congr. Americanists, Paris, Sept. 2-9.

. 1976. Personal communication, Catholic University, Washington, D.C.

Daubenmire, R. F. 1959. Canopy coverage method of vegetation analysis. Northwest Sci. 33:43-64.

. 1968. Plant communities, A Textbook of Plant Synecology. New York: Harper & Row. 300 pp.

Davis, M. B. 1974. Holocene migrations of ecotones. Abstr. American Quaternary Assoc. Abstracts of Third Biennial Meeting, 18-22.

Davis, M. B., L. B. Brubaker and J. M. Beiswenger. 1971. Pollen grains in lake sediments: Pollen percentages in surface sediments from southern Michigan. Quat. Res. 1:450-467.

Deevey, E, S. 1944. Pollen analysis and Mexican archaeology: An attempt to apply the method. Amer. Antiq. 10:135-149.

. 1976. Impact of Man on Aquatic Processes in Central America. Intl. Symp. Paleolimnology. Mikolaiki, Poland.

Emerson, R. A. and J. H. Kempton. 1935. 1935 agronomic investigations in Yucatan. Carnegie Inst. Yearbook 34:138-142. 185

Faegri, K. and Johs Iversen. 1964. Textbook of Pollen Analysis. New York: Hafner, 238 pp.

Fritts, H. C. 1977. Personal communication, Tree Ring Laboratory, University of Arizona, Tucson.

Gleason, H. A. 1940. The Melastomtaceae of the Yucatan Peninsula. In Botany of the Maya Area. Carnegie Inst. Pub. 522, Misc. Papers XX:325-373.

Gomez-Pompa, A. 1973. Ecology of the vegetation of Veracruz. In Vegetation and Vegetation History of Northern Latin America, A. Graham, ed. Elsevier, Amsterdam. pp. 73-148.

Graf, A. B. 1957. Exotica, Pictoral Cyclopedia of Indoor Plants. Rutherford, N. J.: Roehrs, 464 pp.

Graham, A. (Ed.). 1973. Vegetation and Vegatation History of Northern Latin America. Elsevier, Amsterdam, 393 pp.

Greenland, D. J. 1975. Bringing the green revolution to the shifting cultivator. Science 190(4217):841-844.

Hall, E. R. and K. R. Kelson. 1959. The Mammals of North America. New York: Ronald Press, 1083 pp.

Hammond, N. 1976. The Early PreclasSic in the Maya Lowlands. Paper delivered to Soc. Amer. Arch. Meeting. St. Louis.

Harrison, P. D. 1975. Intensive agriculture in Southern Quintana Roo, Mexico. Paper presented to 40th ann. meeting, Soc. Amer. Arch. Dallas.

Haviland, W. A. 1970. Tikal, Guatemala, and Mesoamerican urbanism. World Arch. 2:186-198.

International Contress of Americanists. In Press. Prehispanic Maya Agriculture Symposium XLII. Paris 1976. between ants and Janzen, D. H. 1966. Coevolution of mutualism acacias in Central America. Evolution 20:249-275. Soc. Amer. . 1970. The unexploited tropics. Bull. Ecol. 51(3): 4-7. 1973-1977. Unpublished Laboratory of Paleoenvironmental Studies. manuscripts. Tucson: The University of Arizona. sketch of the phytogeography of the Lundell, C. L. 1934. Preliminary Yucatan Peninsula. Carnegie Inst. Pub. 436:257-321. 186

MacNeish, R. S. 1964. Ancient Mesoamerican civilization. Science 143:531-537.

• 1967. An interdisciplinary approach to an archaeological problem. In The Prehistory of the Tehuacan Valley, Vol. 1, Environment and Subsistence, Douglas Byers, Ed. Austin: University of Texas Press, 14-24.

Markgraf, Vera. 1976. Personal communication, Laboratory of Paleo- environmental Studies, University of Arizona, Tucson.

Martin, Paul S. 1963. The Last 10,000 Years. University of Arizona Press, 87 pp.

1964. Paleoclimatology and a tropical pollen profile. Report VI, Intl. Cong. on Quaternary, Vol. 2. Warsaw. pp. 319-323.

McBryde, Felix W. 1945. Cultural and historical geography of south- west Guatemala. Smithsonian Inst. of Soc. Anthro. Pub. #4. Washington, D.C.

Morley, S. G. 1950. The Ancient Maya. Stanford University Press, 507 PP• Mueller-Dombois, D. and H. Ellenberg. 1974. Aims and Methods of Vegetation Ecology. New York: Wiley, 547 pp.

Padilla, V. 1973. Bromeliads. New York: Crown, 134 pp.

Palerm, A. and E. R. Wolf. 1957. Ecological potential and cultural development in Mesoamerica. Pan Amer. Un. Soc. Sci. Monograph 3:1-37.

Poore, M. E. D. 1962. The method of successive approximation in descriptive ecology. Adv. Ecol. Res. 1:35-68.

Puleston, D. E. 1969. Settlement patterns and tree crops. Paper delivered to 136th meeting, American Assoc. Advance. Sci.

. 1974. Intersite areas in the vicinity ot Tikal and Uaxactun. In Mesoamerican Archaeology, N. Hammond (Ed.). University of Texas Press, 303-312. complex prehis- • 1976. Experimenting with the ecology of a toric agricultural system. Paper presented to Soc. for Amer. Arch. meeting, St. Louis.

and 0. S. Puleston. 1971. An ecological approach to Puleston, D. E. 24(1):33-333. the origins of Maya civilization. Arch. 187 Rathje, W. L. 1971. The origin and development of Lowland Classic Maya civilization. Amer. Antiq. 36(3):275-285.

Raynor, G. S., E. C. Ogden and J. V. Hayes. 1972. Dispersion and deposition of corn pollen from experimental sources. Agron. Jour. 64:420-427. Reina, R. E. 1967. Milpas and milperos: implications for prehistoric time. Amer. Anthro. 69:1-20.

Rice, Don S. 1976. Population growth and subsistence decision-making in the Yaxha-Sacnab region, Peten, Guatemala. Paper presented to XLII Intl. Congress of Americanists, Paris.

Richards, P. W. 1952. The Tripical Rain Forest. Cambridge, England. The University Press.

Rzedowsii, J. and Roger McVaugh. 1966. Vegetation de Nueva Galicia. Contr. Univ. Mich. Herb. 9(1):1-123.

Salgado-Labouriau, M. L. 1973. Contribucao a palinologia 1973 dos cerrados. Academia Brasileira de Ciencias, Rio de Janeiro, 291 pp.

Sanders, W. T. 1973. The cultural ecology of the Lowland Maya: A reevaluation. In The Classic Maya Collapse, T. P. Culbert (ed.). Albuquerque, University of New Mexico Press, 325-365. Sears, P. B. and K. H. Clisby. 1955. Palynology in southern North America. 4. Pleistocene Climate in Mexico. Bull. Geol. Soc. Amer. 66:521-530.

Shreve, Forrest. 1937. Lowland vegetation of Sinaloa. Bull. Torrey Bot. Club 64:605-613.

Shreve, Forrest and I. Wiggins. 1964. Vegetation and Flora of the Sonoran Desert. Stanford: Stanford University Press, 1740 pp. Siemens, A. H. and D. E. Puleston. 1972. Ridged fields and associated features in southern Campeche. Amer. Antiq. 37:288. Simpson, G. G., A. Roe and R. C. Lewontin. 1960. Quantitative Zoology. New York: Harcourt, Brace and World, 440 pp. Standley, R. C. and others. 1943. Flora of Guatemala. Fieldiana: Botany 24:pts. Tauber, Henrik. 1965. Differential pollen dispersion and the interpretation of pollen diagrams. Geol. Survey of Denmark II, Series' No. 89, 66 pp. Thomas, D. H. 1973. An empirical test of Steward's model of great basin settlement patterns. Amer. Antiq. 38:155-176. 188

Thompson, J. E. S. 1964. The Rise and Fall of Maya Civilization. University of Oklahoma Press, 328 pp.

Tsukada, Matsuo. 1964. Pollen morphology and identification III. Pollen et Spores 6:393-462.

Tsukada, Matsuo and E. S. Deevey. 1967. Pollen analyses from four lakes in the southern Maya area of Guatemala and El Salvador. In Quaternary Paleoecology, E. J. Cushing and H. E. Wright Tfds.). New Haven: Yale University Press, 303-331.

Turner, B. L. 1974. Prehistoric intensive agriculture in the Maya Lowlands. Science 185:118-124.

Van der Hammen, T. 1963. Problems of quaternary botany in the tropics (with special reference to South America). Ber. Geobot. Inst. Rubel (Zurich) 34:62.

. 1967. Palynologia Neotropica. Brill, Leiden, 27 pp.

Vaughn, Hague. 1976. Prehistoric disturbance in the area of Flores, Peten, Guatemala. Paper presented to American Inst. Biol. Sci., New Orleans.

Vivo Escoto, Jorge A. 1964. Weather and climate of Mexico and Central America. In Handbook of Middle American Indians, R. C. West (Ed.). Austin: University of Texas Press, 187-215.

Vogeler, I. 1974. Data cited in Prehistoric intensive agriculture in the Maya Lowlands, by B. L. Turner. Science 185:118-124.

Wagner, P. L. 1964. Natural vegetation of Middle America. In Handbook of Middle American Indians, Vol. 1. Natural Environment and Early Cultures, R. C. West (Ed.). Austin: University of Texas Press, 570 pp.

Webb, T., III. 1974. Corresponding patterns of pollen and vegetation in lower Michigan: A comparison of quantitative data. Ecol. 55:17-28.

Webb, T., III and R. A. Bryson. 1972. Late- and Postglacial climatic change in the northern Midwest U.S.A.: Quantitative estimates derived from fossil pollen spectra by multivariate statistical analysis. Quaternary Research 2:70-115. American Indians, Vol. 1. West, R. C. 1964. Handbook of Middle Natural Environment and Early Cultures. Austin: University of Texas Press, 570 pp. 189

Wilken, G. C. 1971. Food producing systems available to the ancient Maya. Amer. Antiquity 56(4):432-448.

Willey, G. R., W. R. Bullard, J. B. Glass and J. C. Gifford. 1965. Papers of the Peabody Museum of Archaeology and Ethnology, Harvard University, Vol. 54:1.

Wiseman, F. M. 1973. The artificial rainforest. Paper delivered to Soc. Amer. Arch. meetings, San Francisco.

. 1974. Paleoecology and the prehistoric Maya. Master's Thesis, University of Arizona, Tucson, 97 pp.

• 1976. The maximal habitat model of Mayan agriculture. Paper delivered to Soc. Amer. Arch. meetings, St. Louis.