The Application of Pedology, Stable Carbon Isotope ...

Brigham Young University Brigham Young University

BYU ScholarsArchive BYU ScholarsArchive

Theses and Dissertations

2004-07-13

The Application of Pedology, Stable Carbon Isotope Analyses and The Application of Pedology, Stable Carbon Isotope Analyses and

Geographic Information Systems to Ancient Soil Resource Geographic Information Systems to Ancient Soil Resource

Investigations at Piedras Negras, Guatemala Investigations at Piedras Negras, Guatemala

Kristofer Dee Johnson Brigham Young University - Provo

Follow this and additional works at: https://scholarsarchive.byu.edu/etd

Part of the Animal Sciences Commons

BYU ScholarsArchive Citation BYU ScholarsArchive Citation Johnson, Kristofer Dee, "The Application of Pedology, Stable Carbon Isotope Analyses and Geographic Information Systems to Ancient Soil Resource Investigations at Piedras Negras, Guatemala" (2004). Theses and Dissertations. 549. https://scholarsarchive.byu.edu/etd/549

This Thesis is brought to you for free and open access by BYU ScholarsArchive. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of BYU ScholarsArchive. For more information, please contact [email protected], [email protected].

THE APPLICATION OF PEDOLOGY, STABLE CARBON ISOTOPE

ANALYSES AND GEOGRAPHIC INFORMATION SYSTEMS TO

ANCIENT SOIL RESOURCE INVESTIGATIONS AT PIEDRAS

NEGRAS, GUATEMALA

by

Kristofer Dee Johnson

A thesis submitted to the faculty of

Brigham Young Univeristy

In partial fulfillment of the requirements for the degree of

Master of Science

Department of Plant and Animal Science

Brigham Young Univeristy

August 2004

BRIGHAM YOUNG UNIVERSITY

GRADUATE COMMITTEE APPROVAL

of a thesis submitted by

Kristofer Dee Johnson

This thesis has been read by each member of the following graduate committee and by majority vote has been found to be satisfactory. ____________________________ ___________________________________ Date Richard E. Terry, Chair ____________________________ ___________________________________ Date Sheldon D. Nelson ____________________________ ___________________________________ Date Mark W. Jackson

BRIGHAM YOUNG UNIVERSITY

As chair of the candidate’s graduate committee, I have read the thesis of Kristofer Dee Johnson in its final form and have found that (1) its format, citations, and bibliographical style are consistent and acceptable and fulfill university and department style requirements; (2) its illustrative materials including figures, tables, and charts are in place; and (3) the final manuscript is satisfactory to the graduate committee and is ready for submission to the university library. ___________________________ ____________________________________ Date Richard E. Terry Chair, Graduate Committee Accepted for the Department ____________________________________ Sheldon D. Nelson Department Chair Accepted for the College ____________________________________ R. Kent Crookston Dean, College of Biology and Agriculture

ABSTRACT

THE APPLICATION OF PEDOLOGY, STABLE CARBON ISOTOPE ANALYSES AND GEOGRAPHIC INFORMATION SYSTEMS TO

ANCIENT SOIL RESOURCE INVESTIGATIONS AT PIEDRAS NEGRAS, GUATEMALA

Kristofer Dee Johnson

Department of Plant and Animal Sciences

Master of Science

The ancient inhabitants of the Maya Lowlands enjoyed a long and fruitful period

of growth which climaxed at around AD 800. At that time, millions of people

successfully subsisted in a challenging environment that today only supports a population

a fraction of that size. These facts, and the subsequent “Maya Collapse”, are the impetus

of many recent studies that utilize environmental data, in addition to conventional

archaeology, to investigate this Maya mystery. Pedological studies and stable carbon

isotope analysis of soil organic matter, combined with Geographic Information Systems

(GIS) are three tools that can be used to answer crucial questions as to how the Maya

managed their soil resources.

GIS maps that indicated areas of best agricultural potential based on slope and

soil type were used as a guide to opportunistically sample soils in an area south of Piedras

Negras Guatemala – an area that was densely vegetated and unexplored. Soils that

represented the different soil resources of the area were sampled with a bucket auger at

15 cm intervals. The samples were then tested in a laboratory for physical and chemical

characteristics and δ13C values were determined for soil organic matter. Soil taxonomical

descriptions indicated that overall the soil resources of the area were very good as almost

all the soils were classified as Mollisols - the most fertile of all the soil orders. The suite

of great groups found was Haprendolls, Argiudolls, Argiaquolls and Udorthents. The

characteristics which distinguish these great groups were used to further investigate

relative agricultural productivity from an ancient soil resources point of view.

Haprendolls were better drained and probably made for good agricultural soils given soil

depth and rainfall were adequate. The Argiudolls and especially the Argiaquolls were

probably less favored because of very high clay contents that made them more difficult to

work with and poor drainage.

Stable carbon isotope analyses revealed strong evidence for maize agriculture in

some environments of the study area. δ13C values as high as -16.6‰ (76% C4–Carbon)

were observed in areas of significant soil accumulation in well drained and moderately

drained soils. Minimal evidence of maize agriculture was found in more marginal

environments such as those with little soil accumulation or poorly drained areas. Also,

the pattern of the graph of δ13C values versus depth indicated that ancient agriculture

occurred continuously in some areas, but in other areas as distinguishable events.

Finally, when the strength of the C4 signal was represented graphically and overlaid with

a modified GIS agricultural potential map, a visual representation of the extent and

degree of ancient agriculture was achieved.

Our findings suggest that upland agriculture was favored by the ancient Maya of

Piedras Negras and that the region between Piedras Negras and Yaxchilan was an

agriculturally important breadbasket. The methods and results of this study provide

foundational information for the investigation of ancient Maya agriculture. In future

studies, it may be possible to more systematically map ancient agricultural fields and

estimate the carrying capacity of a region based on its soil resources.

ACKNOWLEDGEMENTS

I would like to acknowledge first and foremost my family for their love and

support during this project. Also, I would like to thank the countless friends who were

always concerned with my progress and success in this endeavor.

I would like to recognize the unique faculty at Brigham Young Univeristy for

their sincere desire to provide quality education and financial support. I am truly grateful

for the opportunities that were afforded to me to participate in conferences, laboratory

research and teaching. I am especially indebted to Dr. Richard Terry for his academic

guidance, personal advice and for just plain putting up with me so that I could achieve

some of my most important life goals. Dr. Sheldon Nelson and Dr. Mark Jackson also

deserve my most sincere thanks for their assistance in the production of this thesis. I

could never forget Dorine Jesperson and her friendship and constant help with University

formalities.

I am also most grateful for the help of the many people who worked with me

during the 2002 and 2003 field seasons including Dr. Charles Golden, René Munoz,

Andrew Scherer, Marcelo Zamora, Edwin Román Ramirez, Pánfilo Regino Hernandez,

José Luis Aldana Alvarado, Eduardo Isidoro Saquij and others of the Sierra del Lacandón

Regional Archaeological Project and Dr. Takeshi Inomata and Marcus Eberl of the

Aguateca Regional Archaeological Project. Also, I would like to thank Dr. Henry

Schwarcz and Dr. Elizabeth Webb of McMaster University for their help and

collaboration.

I would also like to express my gratitude to lab friends that helped me day after

day grinding soils, performing soil tests, and keeping good company. Most helpful were

Chris (Don) Jensen, David (El Petenero) Wright, Carmen Lopez, Suzanna Peña, Craig

Paul, Sarah Kendall, Marco Alvarez and others.

TABLE OF CONTENTS

Title Page…………………………………………………………………………………...i

Graduate Committee Approval…………………………………………………………….ii

Final Reading Approval and Acceptance………………………………………………….iii

Abstract…………………………………………………………………………………….iv

Acknowledgements………………………………………………………………………..vii

Table of Contents…………………………………………………………………………..ix

Introduction………………………………………………………………………………...xi

Manuscript 1………………………………………………………………………………..1

Ancient Soil Resources of the Usumacinta River Region, Guatemala I: Pedology

Manuscript 2………………………………………………………………………………..30

Ancient Soil Resources of the Usumacinta River Region, Guatemala I: GIS and

Stable Carbon Isotope Anaylses

Appendix…………………………………………………………………………………...54

INTRODUCTION The history of the relationship between the ancient Maya and their seemingly arduous environment is an antiquated microcosm of events that now challenge Mesoamerica’s modern inhabitants. The Maya civilization’s prolonged success in subsistence, and then their arcane and abrupt ending, forms the “Maya Mystery”. The Maya mystery is the impetus of many studies that probe at the questions of “how did they do it” and “what happened?” The intriguing history of the relationship between land and Mesoamerican peoples goes back to the 9000 to 8000 BC time period when a hunter-gatherer way of life was dominant. Then about 5000 BC maize, bean and other crops were domesticated at various smaller locations in Mexico and Central America until about 3000 BC when a widespread adoption of crop cultivation occurred. Signs of permanent settlement, and thus the roots of the Maya civilization, appeared much later around 1400-1200 BC during what is known as the Early Preclassic Period (2500 - 1000 BC). By the Late Preclassic (400 BC – AD 250), a complex society had emerged as evidenced by dramatic changes in pottery styles and the scale and design structures. Also during this time there appeared religious symbols, fortifications, stone monuments and early hieroglyphs while the overall population increased substantially. The Early Classic Period (AD 250 – AD 600) marked the beginning of the Maya King dynasties and increased warfare. Many of the elements of the Preclassic such as art, architecture, iconography, writing and agriculture were advanced or aggrandized during this time period. The height of the Maya civilization occurred between AD 600 and AD 800 known as the Late Classic period when the achievements of this society were unmatched and unprecedented. Styles of religion, writing and architecture were more similar and widespread than ever before, the artifacts of which would earn the Maya a place among the world’s greatest civilizations. It is ironic that the most impressive accomplishments of the ancient Maya were realized on the eve of their abrupt ending, known as the “Maya Collapse”. Over the next 150 – 200 years during the Terminal Classic Period (AD 800 – AD 1000) there occurred a notable decline or cessation of dated monuments and buildings which reflected a failure of politics and culture. More recent studies point out that the collapse was not uniform in time or space, but instead occurred in succession. First, the western part of the southern Maya lowlands collapsed around AD 800 whereas the northern lowlands continued to flourish and hold out until AD 1000. Most Mayanists from a variety of academic backgrounds believe that many variables were to blame for the Maya collapse including drought, catastrophic events, epidemic warfare and invasion. In addition, it is common to amend the explanation of land stewardship practices which exhausted soil nutrients and caused soil erosion. Whatever the cause, it is a subject of study that will not soon grow old. In comparison to other more visible elements of Maya society (e.g. architecture, epigraphy, religion, warfare etc.), their agricultural activities are uncelebrated and more ambiguous to study through conventional archaeology. Yet, an understanding of this crucial part of Maya life is the crux of investigations about the Maya’s relationship with their environment. The successes and failures of this civilization (and any civilization for that matter) were probably rooted in, or at least associated with, its subsistence strategies and/or land stewardship practices. Scientists and archaeologists are turning more and more to environmental data to explore these issues.

This thesis focused of the ancient soil resources available to the Maya, or lack thereof in some areas, to understand the “how’s” of the Maya agricultural methods. Our main objective was to use pedological investigations, soil chemical analyses and Geographic Information Systems (GIS) as tools to provide a more complete understanding of the agricultural landscape of the Late Classic Period. Further, our aim was to answer questions about the agricultural resources of the ancient Maya city of Piedras Negras. Specifically, we wanted to know if the Macabilero region between Piedras Negras and Yaxchilan was agriculturally more important than surrounding areas (i.e. was it a breadbasket?) and where might the people of the defensively situated city of Aguateca have cultivated crops to support their population?

The thesis contains two manuscripts, “Ancient Soil Resources of the Usumacinta River Region, Guatemala I: Pedology” and “Ancient Soil Resources of the Usumacinta River Region, Guatemala II: GIS and Stable Carbon Isotope Analyses”. The first manuscript focuses on the soil resources of the area as revealed by soil taxonomical descriptions of soil profiles investigated in the area. The second manuscript investigates the soil resources from a different angle using GIS and stable carbon isotope techniques. The two manuscripts are prepared for submission to Latin American Antiquity.

Ancient Soil Resources of the Usumacinta River Region,

Guatemala I: Pedology

Kristofer Dee Johnson, Richard E. Terry, Sheldon D. Nelson Deparment of Plant and Animal Sciences and Department of Geography Brigham Young Univeristy Provo, UT 84602 USA A manuscript to be submitted to Latin American Antiquity

Introduction The ancient Maya of the Guatemala Lowlands, in what is now known as the department

of Petén, managed to subsist successfully for hundreds of years in a fragile environment (Beach 1998). This is evidenced by the density of ruins that are observed today – some 100 to 200 structures per km2 in the Belize River Valley alone (Ford 1990, p. 180). Populations perhaps numbered in the millions during the Late Classic Period (550-800 AD) (Turner 1990; Whitmore et al. 1990, but see also Webster 2002, p. 264), making the demand on agricultural resources a constant and prime concern for generations. Simple slash and burn, or swidden agriculture, as it is practiced in the Petén today, was probably also practiced in the earlier years of Maya society (around 1000 B.C.; Dunning et al. 1998), yet this method cannot explain the sustained food production and prosperity that the Maya apparently enjoyed for hundreds of years. It is also interesting to note that the last estimated modern populations living in the same area number to only about a 367,000 in Petén (Instituto Nacional de Estadística 2003) and struggle to balance subsistence with sound land stewardship practices. Perhaps something can be learned from the methods of land stewardship practiced by the ancient Maya. It is clear that alternatives to simple and unsustainable slash and burn methods must have been employed by the Maya. Nations and Nigh (1980) for example found that the modern Lacandon Maya of Chiapas, Mexico managed to subsist with minimal impact on their environment and even produced greater yields than migrant farmers who used conventional slash and burn. Further, Johnston (2003) believes that larger populations such as the Late Classic Maya could have intensified agricultural production through weeding and mulching.

How did the ancient Maya interact with their environment? Or, more specifically, “how did ancient Maya farmers decide where to establish their farmsteads” (Fedick 1996; p. 107). These questions are fundamental to understanding what role Maya agriculture, and the Mayas’ interaction with the environment, played in the successes and failures of this great civilization. This issue has been at the center of many debates and its complexity is indicated by the substantial number of studies dedicated to it (Beach et al. 2003a,b; Beach and Dunning 1995; Cowgill 1962; Deevey et al. 1979; Dunning and Beach 1994, 2000, 2004; Dunning et al. 1998; Fedick 1996; Harrison 1993; Harrison and Turner 1978; Jacob 1995; Kunen et al. 2000; Pohl et al. 1996; Pope et al. 1996; Puleston 1978; Turner 1985; Turner and Harrison 1983). To investigate the question of how the Maya subsisted from, and adapted to, their environment, one must recognize that subsistence strategies and environmental factors varied geographically and temporally (Dunning 1996; Turner 1985). Indeed “…it is dangerous to extrapolate environmental circumstances, and inferred agricultural adaptations to them, from one area of the Maya Lowlands to another” and “What is needed now are more careful examinations of specific adaptations made in the suite of microenvironments presented in the various regions of the Maya Lowlands” (Dunning 1996, p. 54; see also Dunning and Beach 2000; Fedick 1995). There is a rising need for open-minded and case-by-case approaches to ancient agricultural investigations that employ a variety of tools and academic backgrounds (Dunning and Beach 2004). To address this issue, archaeological science has stretched beyond interpreting strictly artifactual evidence to pedological investigations, which shed light on ancient land resource management and the inherent challenges of cultivating tropical soils.

This paper will report on the soil resources in a much understudied region of the Guatemalan Lowlands (Dunning et al. 1998, p. 94), located between the important ancient cities of Piedras Negras and Yaxchilan. It will also propose some preliminary ideas about how the ancient Maya in this area may have used their soil resources for subsistence food production.

2

These results contribute to a foundation of natural resource information that can help answer much broader questions dealing with land resource planning, land control, farming methods, carrying capacities, and the role of ancient agriculture may have played in sociopolitical issues of the ancient Maya. Review of pedological research in the Maya Lowlands

About 40 to 60 percent of the Maya Lowlands is covered with wetlands, or bajos (Dunning et al. 2002; Jacob 1995). Soil studies that have investigated wetland agriculture focused on the eastern parts of the Yucatan as well as other bajos near Tikal and El Mirador in the southern lowlands. Wetland agriculture may have developed because the Maya were obligated by both natural and anthropogenic factors (e.g. drought, sea level rise, deforestation) to practice intensive agriculture in these marginal areas (Dunning et al. 2002). The amount of arable soil (i.e. not limited by water saturation) that exists on the fringes of these wetland systems changes according to the amount of precipitation and the time of year.

Vegetation type is an indicator of water holding capacity and drainage which in turn are indicators of agricultural potential. There can be considerable variation in bajo topography and vegetation which affects settlement and agricultural opportunities (Culbert et al. 1990). Kunen et al. (2000) and noted “islands” of higher land where settlement was found in Bajo La Justa, Guatemala (see also Sever and Irwin 2003). Kunen et al.’s study classified bajo areas according to two dominant vegetation types: “palm bajo” and “scrub bajo”. Palm bajo had a higher vegetation canopy, more open understory and was dominated by the escoba (Crysophilia argentes Bartlett), corozo (Orbignya cohune) and botan (Sabal mayarum) palms. Scrub bajo had thick low-canopy vegetation dominated by palo de tinto (Haematoxylum campechianum L.), pucte (Bucida buceras), sapamuche (Oreopanax guatemalense) and zacate de heuche (Scleria sp.). Palm bajo also seemed to indicate desirable land for cultivation according to locals because of better drainage and sufficient moisture retention into the dry season.

In a separate study of the Mirador Basin, Hansen et al. (2002) referred to the local names of bajos and civales for different areas of wetlands. Bajos were only seasonally inundated for about two months of the year and were dominated by palo tinto trees. Also, the vegetation canopy in the bajo areas was noticeably lower than in the surrounding upland rainforest. Civales were more similar to marshes because they were smaller and are located within or adjacent to bajos. Also, the civales were treeless and comprised of mostly grasses and sedges.

Ancient Maya wetland agriculture is seldom mentioned without reference to drainage modifications that would have been required for agricultural production in some areas (Bloom et al. 1983,1985; Pohl et al. 1996). Specifically, this could mean the construction of simple canals, the transfer of soil above the natural surface (raised fields) (Bloom et al. 1983; Jacob 1995; Turner 1974), modifications of natural hummocks (Pohl et al. 1996), or the construction of other features that control water movement and soil deposition (Fedick 2001). Evidences for such structures come from soil profiles where buried organic soils appear to be ditched and are associated with cultural artifacts, phytoliths and pollen (Bloom et al. 1983, 1985; Pohl et al. 1996). The physical appearance of surface soils viewed from aerial photographs, radar and satellite imagery can also indicate raised fields (Adams et al. 1981; Pope and Dahlin 1989; Turner 1974), although this is controversial since they can be easily confused with naturally forming features such as gilgai patterns (Puleston 1978). There is also ethnographic evidence of these wetland drainage ditches from eyewitness Spanish conquistador accounts of the 1500’s in Tabasco, Mexico (see Pohl 1985). The extent of wetland modifications and their significance to ancient food production is still a subject of much debate.

3

Beach et al. (2003a) reported that the karstic depressions around La Milpa and Dos Hombres, Belize contained mainly Histosols and Vertisols that graded into upland Mollisols, Inceptisols and Alfisols with vertic or lithic properties. Redoximorphic features were found including grey and red mottles and iron and manganese concretions. These soils were also high in cation exchange capacity (CEC) and base saturation, which are typical for soils of the Yucatan Peninsula. The fertile and deep Histosols may have been a valuable seasonal soil resource if they were adequately drained in the dry season. Soil pH ranged from very acidic to moderately alkaline. The clay mineralogy of the bajos mainly consisted of chlorite/smectite and to a lesser degree smectite and vermiculite (see also Jacob 1995). In addition to these characteristics of bajos, Jacob (1995) found soils massive gypsum deposits in Pulltrouser Swamp, Belize.

Several undesirable qualities for agriculture, both chemical and physical, are sometimes observed in bajo soils (see Table 1 for summary). The macronutrient phosphorus seems to be ubiquitously low throughout the Maya Lowlands. High clay content and high pH are two possible reasons for increased phosphate fixation and therefore lower Phosphorus availability (Beach 1998). Zinc was also found to be very low in the bajos of the Three Rivers Region as well as conditions for acidity in localized areas (Beach et al. 2003b). Moreover, the physical properties of these clayey soils make them hard to work with as they are sticky when wet and extremely hard when dry (Dunning et al. 2002). Sometimes these characteristics could have evolved from erosion deposition caused by the ancient Maya themselves (Beach et al. 2003a).

The stratigraphy of the bajo soils in Belize and Northern Guatemala generally include the following components: 1) surface A or histic horizons that reflect a period of little change, 2) one or more layers of clay, known as the “Maya Clay” (Deevey et al. 1979), below the surface soils and below this a 3) simple buried A horizon or a stratigraphically unconformable and contorted horizon (Beach et al. 2003a; Dunning et al. 2002; Hansen et al. 2002; Jacob 1995). The term “Maya Clay” does not necessarily apply to clay deposition at a specific time period (e.g. Late Preclassic or Late Classic), but rather is used to indicate clay deposition caused by Maya from forest clearance and associated erosion events (see Beach et al. 2003a; Dunning et al. 1998; Jacob 1995).

Beach et al. (2003a) proposed anthropogenically induced erosion and sedimentation processes to explain the complex stratigraphic patterns of soil profiles in Northwestern Belize. First, episodes of deforestation by fire, beginning in the Preclassic, resulted heavy soil erosion and Maya Clay deposition horizons. The Maya Clay overlies A horizons thought to be the ancient surface. As sediments were rapidly deposited, soil deformation of the buried soils occurred because of “differential sediment loading that induced shearing by expansive clays against the sediment overburden” (Beach et al. 2003a). This deformation may have also been influenced by periodic droughts (Hodell et al. 2000). The absence of Maya activities then led to different soil formation processes due to reforestion and moister soil conditions. The practical upshot of these interpretations is that the effects of deforestation by the ancient Maya caused perennial wetlands, or civales, to transform into only seasonally inundated bajos. This loss of water source probably had an enormous impact on settlement and adaptation strategies by the Maya. The environmental changes observed in the soil pedological record explain at least some of the reasons for desertion in some areas and construction of water management structures in others (see Hansen et al. 2002; Dunning et al. 2002).

Other pedology studies have focused on upland soil resources of the Maya. In many upland summit soils of the Petén, Guatemala, soil depth only reaches to about 20 cm or less over a lithic contact with bedrock (Beach 1998a,b; Fernandez 2002; Jensen 2003; Johnson et al. n.d.).

4

They are classified as Lithic Rendolls with simple O-A-C horizons (Beach 1998; Johnson et al. n.d.). Despite their shallow and highly erodible nature, these soils are favored by contemporary farmers as they are well drained and high in CEC. Moreover, the level of organic carbon is high to moderate (3-5 percent) except in deeper horizons where it is usually much lower, with the exception of buried A horizons. Soil pH levels are generally higher in upland soils (7.2-8.4) than in bajo soils because of the carbonate limestone parent material. High percentages of calcium carbonate (greater than 50 percent) from the parent material affect zinc and phosphorus plant availability (Beach 1998; Beach et al. 2003a; Johnson et al. n.d.).

Fedick (1995) used the USDA system of land evaluation to classify soil types in Belize based on factors that were limiting to agriculture (e.g. effective root zone, susceptibility to erosion, workability, drainage, and fertility). Soil types that were associated with valley alluvial bottoms were judged to be the best for agricultural production. The highest density of settlement occurred on limestone platforms that were also agriculturally productive, though not to the same extent as the alluvial bottoms. This indicated that the Maya in this area chose to reside close to, but not within, areas that could be intensely cultivated (see also Dunning et al. 2002; Fedick 1996; Hansen et al. 2002; Jensen 2003). Fedick (1994) used soil, parent material and slope to predict the occurrence of agricultural terraces in the Upper Belize River area. The model was field tested and 13 terrace systems were identified. All of them were associated with low slopes and consolidated limestone parent material. Different types of terrace systems such as box terraces, contour terraces and cross-channel terraces were also documented and used to refine the model.

In other upland areas such as the Petexbatún region, karstic sinkholes became areas of soil accumulation. They are locally known as rejolladas, when well drained, and aguadas when waterlogged (Dunning and Beach 1994). Rejolladas may have been politically important and controlled by local elites. Aguateca, for example, is located on the edge of an escarpment for defensive purposes – away from wetland soil resources. A rejollada zone which extends to the west of this site may have been favored for its soil resources, especially during the wet season when other areas became inundated. Furthermore, natural steps that closely resemble terraces formed as the sinkholes successively collapsed and trapped soil at midslope locations. It is possible, therefore, that they were used in ancient times as agricultural terraces (Johnson et al. n.d.). At Aguateca, Johnson et al. (n.d.) found shallow midslope soils that classified as Lithic Rendolls which were well drained but easily eroded. Rejolladas that had steeper slopes (10%) accumulated a greater amount of soil than those with more gentle slopes (3%). Soil accumulation of the more inclined rejolladas reached up to one meter depth and classified as Typic Argiudolls or Aquertic Argiudolls depending on the vertic properties of the soils. Beach (1998) observed similar soils plus the occurrence of Vertic Rendolls and Rendolls in some footslope soils. The main soil fertility limitation in the Petexbatún seems to be low phosphorus and in some cases low nitrogen or potassium (Beach 1998; Johnson et al. n.d.).

Upland soils have also been investigated in the less studied and physiographically different Usumacinta River Basin region by Aliphat (1996) and Fernandez (2002). Aliphat (1996) mapped soil series of the region and reported on their potential for agricultural productivity with the land capability class scheme used by the USDA. He reported that nearly all the soil series had a land capability class of II, meaning they only had moderate limitations under hand cultivation. However, this conclusion lacks thorough investigation as Aliphat (1996) only sampled along the Mexican side of the Usumacinta River Basin. These soil types were then extrapolated to the Guatemalan side of the river based on topographic features. A sizeable area

5

near the Laguneta Lacandòn on the Guatemalan side was mapped as red soils that are well-drained and deep with weak granular structure – named the Chequique I and II series. The main limitations of the Chequique series are rock outcroppings due to the irregular terrain and low levels of fertility in the lower horizons. These are the best soils for agricultural production after considering slope, fertility and drainage. The Yaxchilan series are clayey and deep red soils on level ground with minor undulation. Their poor drainage often results in a “pseudo-gley” horizon. Menche soils are similar to the Yaxchilan except that they have good drainage. The La Pasadita soil series were the only soils with a land capability class of V. Class V soils are generally considered not suitable for hand cultivation. La Pasadita soils were dense, deep, very poorly drained and had gleyed horizons called the La Pasadita I and II series. Where the water table is high, these soils would need to be artificially drained to be farmed. Another soil worthy of mention occurs in the steep karst terrain, where terracing from the ancient Maya is evident – called the Lacandón series. These soils are very shallow (at most 50 cm), except in depression areas where they accumulate and may become fertile and well-drained. The agricultural potential of the Lacandón series is therefore moderate to good in certain locations.

Fernandez (2002) reported soil profile investigations of backslope, footslope and toeslope toposequence characteristics near Piedras Negras. Backslope soils are thin (10 to 40 cm), silt loam Mollisols with favorable bulk densities (1.26 to 1.5 g/cm3) and mainly A-C sequences. Toeslope and footslope locations contained thick clay loam Mollisols that were moderate to high in calcium, magnesium, potassium and carbonates. Footslope soils are also thicker (about 25 to 45 cm), with A1-A2-C or Cr horizons, and weak to moderate soil structure. Surface organic carbon was higher in backslope soils (9.7% to 15.1%) than footslope soils (3.8% to 11.8%). Phosphorus was limited in the soil (8.3 mg/kg), however it is suggested that the use of natural fertilizers could have remedied this problem making the land agriculturally productive. Moderate soil stickiness and severe shrink-swell hazards were also observed, although not ubiquitously, and could be a limitation to the workability of the soil.

Overall, Fernandez (2002) found that back-slope and foot-slope soils had lower agricultural potential than toeslope soils, although they could also be farmed successfully if measures were taken to control soil erosion. It was also estimated that the rate of soil formation for the areas was approximately 0.092 mm/yr. At Motul de San Jose, Guatemala, Jensen (2003) found similar soil characteristics from toposequence profiles to those observed by Fernandez (2002). Argiudolls in footslope areas and Haprendolls in backslope areas typify the landscape. In addition, important information about soil types was gained from local contemporary Maya farmers and combined with laboratory tests to produce a soils map. Fertile but shallow soils on hilltops were settled and deeper and still fertile soils were seemingly left for agricultural purposes both in ancient and modern times.

Study Area The study area is located within the Sierra del Lacandón National Park along the Mexican

border in western Guatemala (Fig. 1). It is located about 8km to the southeast of Piedras Negras and about 30 km northwest of Yaxchilan. Two lesser known sites, El Cayo and Macabilero, and the recently re-discovered site of Texcoco are located nearby (Golden et al. n.d.). The geology of the area causes the bordering Usumacinta River to wind through an anticline that is periodically faulted (Aliphat 1996). During the Lower Cretaceous large amounts of carbonate materials were deposited. Tertiary limestone of the Paleocene and Eocene was deposited after the tectonic events at the end of the Cretaceous (Aliphat 1996). The weathering of both the Cretaceous and Tertiary limestone has produced a karst landscape of highly variable soils,

6

hydrology and topography. Some of the sinkholes reach hundreds of meters in depth. However, there are some areas, such as the present study area, that are relatively more gently sloped.

The impact of humans on the study area, and therefore climate and vegetation conditions (Binford et al. 1987; Leyden et al. 1993; Mendez 1999), has remained fairly constant since the Terminal Classic abandonment by the Maya about 800 A.D. Since then, small groups of Lacandon Maya used the land for hunting and fishing, but mostly remained on the Mexican side (Nations and Nigh 1980). Later, small paramilitary groups known as the Comunidades Populares en Resistencia en el Petén (CPR-P) had occupied the area at different times and in different areas for about 40 years until a recent peace agreement with the Guatemalan government. A few small and intermittent fields that were cleared for crops and an old airstrip are still visible but now covered with dense secondary vegetation. Although logging occurred in the early part of the twentieth century, the occupation by the paramilitaries, in addition to the park’s remoteness, has discouraged other anthropogenic influences, such as slash and burn agriculture, that affect much of the Petén. As a result, nearly the entire park is covered with primary forest.

The Parque Nacional Sierra Lacandona is interesting both archaeologically and environmentally. Archaeologically, it is home to a large density of ancient Maya ruins. Piedras Negras is the most famous and was a major political center of its day. Its impressive stelae and panel collection depict a rich dynastic history and frequent wars with its neighbor, Yaxchilan (Houston et al. 1999, 2000). Piedras Negras is also well known because of its association with Tatiana Proskouriakoff’s (1960) work that helped decipher the Maya code. Two possible population peaks have been documented at Piedras Negras, one during the Preclassic around 200 BC, followed by an apparent decline, and another during the Late Classic about 800 AD. The Late Classic population peak coincides with the height of Maya population observed throughout much of the Petèn (Houston et al. 2000).

After initial excavation performed in the 1930’s under the University of Pennsylvania, Piedras Negras has received recent attention with excavations performed by the Piedras Negras Archaeological Project under Brigham Young University (BYU) and La Universidad del Valle (1997-2001). Before this project, many peripheral sites in the area remained undocumented until a reconnaissance in 1998 of La Pasadita (Golden et al. 1998) and in 2000 of Macabilero (Golden et al. 2001). This was followed by the Sierra del Lacandón Regional Archaeological Project and reconnaissance of the areas surrounding the Laguneta Lacandon in 2003. The 1998, 2000, and 2003 expeditions confirm the density of settlement in the Sierra del Lacandon and the need for further exploration of terra incognita. During the most recent expedition, three archaeological sites that were previously unknown to archaeologists were identified (Esmeralda, Fajardo, and Tecolote) and one known site was located (Texcoco)(Golden et al. 2003). The area was probably politically important because of its location at the proposed border of the Yaxchilan and Piedras Negras polities (Hernandez 2003). Moreover, the area sits near significant water and agricultural resources. Golden et al. (2003) hypothesized that the bajos, and large areas of gently-sloped and well drained uplands adjacent to the bajos, not only provided food for the local population but also could have been an important breadbasket for the larger political centers.

Another reason that this area draws a lot of interest is because of the environmental issues it faces. As in other remote places of Guatemala, there is intense pressure to open the land up for logging and slash and burn agriculture. This type of farming scars the landscape and leads to heavy losses of soil and a decrease in soil fertility. Many squatter farmers from both Mexico and

7

Guatemala are constantly threatening the border areas of the park. Most are forced to look for new areas to cultivate because of decreasing land availability due to land stewardship practices.

Furthermore, a 46-m high dam has been proposed at Boca del Cerro, Mexico (Comision Federal de Electricidad 2003). The completion of this dam would have a number of negative environmental impacts. An obvious direct effect would be the destruction of huge tracts of wildlife habitat. Indirect effects would include easier access to remote areas which would lead to more slash and burn and logging activities. The proposed dam also makes the current cultural resource situation a matter of rescue archaeology. An unknown number of ruins would be completely or partially covered by water including the recently discovered and unexcavated sites of Esmeralda and Fajardo. Piedras Negras, the park’s most important cultural resource, would also suffer serious damage.

Methods Twenty three soil profiles across the study area were sampled with a bucket auger at 15

cm depth intervals. Our main focus was to investigate the deep soils found in the toeslope and footslope positions, although a pair of backslope profiles were also sampled. “Toeslope” refers to the hillslope position that forms a gently inclined surface at the base of the slope. “Footslope” refers to the hillslope position that form the inner, gently inclined surface at the base of a slope (Soil Science Society of America). An additional two profiles were investigated near the archaeological site of Porvenir, located downriver to the northwest, and were included in the analyses. Aliphat’s (1996) soil series map was used as a rough guide for soil profile sampling site decisions. Five of the six soil series that were mapped in the study area were sampled. The Lacandon series was not sampled because adequate data was already available from Fernandez (2002).

Site information was recorded in the field and included UTM coordinates, parent material, drainage, hillslope position and vegetation (e.g. old growth forest vs. secondary forest). Careful attention was paid to avoid contamination soils from different depths. The samples were then transported to the BYU Soils Laboratory, Provo, Utah. Horizon designations were assigned by laying out each sample to simulate a vertical profile to study the physical characteristics of the soils. These profiles were described by horizon, depth, structure, consistency, wet color, dry color and redoximorphic features. Incremental horizon samples were combined to make a composite 50 g sample from which the following were determined. Soil texture was determined by the hydrometer method (Gee and Bauder 1982). Soil pH measurements were determined from a saturated soil paste with a glass electrode. The linear extensibility (LE) was found by drying the saturated paste and measuring the amount of shrinkage in centimeters. LE is not a measure of the smectite or montmorillonite clay content, but rather an indicator of the shrink-swell potential of a soil. Total C and total N were analyzed with an elemental analyzer. Total organic C and N were analyzed by dry combustion on an elemental analyzer after acidification to remove calcium carbonate (CaCO3) and were expressed as percent organic nitrogen (%O.N.) and percent organic carbon (%O.C.) (Nelson and Sommers 1996, pp. 973-974).

The soils were classified according to USDA system where taxonomy names are organized by orders, suborders, great groups, and subgroups (Soil Survey Staff 1999). The USDA system of soil taxonomy is a useful tool to understand ancient agricultural potential because it is based on both qualitative and quantitative soil properties rather than soil formation processes. If we use this system to determine ancient agricultural potential we must make two assumptions about our study area within the past 1,000 to 2,000 years: 1) landscape topology has not changed significantly so that we can infer that the limiting soil types that determined

8

agricultural potential in ancient times existed in the same geographic space as they do today and 2) the effects of soil genesis that has occurred since ancient Maya occupation do not confound our results and discussion about agricultural potential. Statistical analysis was carried out with non-parametric procedures (Wilcoxon Signed-Rank Test) because of outliers and variability in the data.

Results It is important to note that agricultural potential is not necessarily the same as the issue of

agricultural preference, or “where did the Maya choose to farm”. The discussion about the demand of soil resources could encompass number of likely scenarios depending on population changes, technological advances, sociopolitical issues as well as environmental factors such as the time of year, catastrophic events, and decreased soil fertility (see Dunning and Beach 2002). Since we cannot realistically make assumptions about these factors, our discussion about the ancient agricultural landscape must focus largely on what our results suggest about ancient agricultural potential among soil types.

Physical and chemical data for the soils are listed in Tables 2, 3, and 4. Of the 25 profiles that were classified, 23 of them were of the Mollisols order and of the Haprendolls, Argiudolls and Argiaquolls great groups. Mollisols are considered the best for sustainable agriculture because they are well-drained, fertile and have adequate depth. The other two profiles classified as Entisols (soil order) and Udorthents (great group).

The desirability of one soil type above another was indicated by differences of certain soil properties. The adequate sample populations of the Haprendolls and Argiudolls allowed us to compare the two and infer statistical significance for drainage, LE, pH and depth. Other characteristics such as ppm P, %O.N., %O.C. and %clay showed no statistically significant difference at the 0.05 level. Differences between all the soil types were compared with descriptive statistics because of the variability and low sample size populations for the Argiaquolls and Udorthents. Haprendolls

The Haprendolls are a great group of the Mollisols and are characterized by a mollic epipedon that is less than 50 cm deep, a CaCO3 equivalent of more than 40 percent and the absence of an argillic horizon (Table 2). An argillic horizon is a layer in the soil that is significantly higher in phyllosilicate clays than the overlying soil material (Soil Survey Staff 1999). In other words, for Haprendolls the A horizon is sufficiently deep for root growth, generally not as affected by acidity, and less clayey (i.e. more workable) than the other Mollisols. Ten Haprendolls were investigated in the area and further classified as Lithic Haprendolls, Vertic Haprendolls, and Typic Haprendolls. The “Lithic” modifier was assigned when the depth of the soil was not only shallow but also limited by underlying bedrock. Shallow soils were probably limiting to agricultural production because of soil erosion risks that depended on their erodibility and the slope. The “Vertic” modifier meant that, despite being less clayey, the amounts of shrink-swell clays (e.g. smectite and monmorillonite) were greater than in other Haprendolls, which resulted in cracking when dried. Soils with vertic properties would have been difficult to work with under hand cultivation because they are sticky when wet and very hard when dry. The “Typic” modifier was used when the Haprendolls did not fit into any other categories, i.e. no defining characteristics such as an argillic horizon or vertic properties were observable. Therefore, the workability and greater depth of Typic Haprendolls would make them more desirable soils for cultivation than the Lithic and Vertic Haprendolls in flat to gently sloped areas.

9

Argiudolls The Argiudolls represent soils that have an argillic horizon somewhere in the profile

(Table 3) (Soil Survey Staff 1999). The argillic horizon is usually, but not always, located below the zone of root growth and therefore does not affect the productivity of the soil. Of the 10 Arigiudolls that were investigated, the most agriculturally desirable were the Typic Argiudolls. These soils didn’t have any major physically limiting characteristics such as vertic properties or water saturation. “Aquertic” Argiudolls, like the Argiaquolls, had redoximorphic features which indicated aquic conditions for some time during the year. The term “aquic conditions” is used to describe soils that experience continuous or periodic water saturation and reduction (Soil Survey Staff 1999). However, unlike the Argiaquolls, these saturated conditions were not present in the mollic epipedon but directly below it. Therefore, Aquertic Argiudolls could have been used year round under normal precipitation years because the depth of water saturation would not likely affect root growth.

One profile classified as an Oxyaquic Vertic Argiudolls. These soils experienced aquic conditions like the Aquertic Argiudolls except that they didn’t have any redoximorphic features because of better drainage. This profile was located about 50 m from the Rio Macabilero and had somewhat less clay content than the other soils, possibly due to alluvial parent material. Although soils like this were subject to flooding and vertic restrictions, their better aeration (when not flooded) and their courser texture would make them a good resource for dry season agriculture.

Two other subgroups of the Argiudolls were observed in the study area, the Vertic Argiudolls and the Pachic Vertic Argiudolls. These differed from the Aquertic Argiudolls in that they didn’t have any indications of water saturation for long periods during the year. Therefore they were similar to other Argiudolls in their capability to grow crops during most of the year, and even better during excessively rainy seasons that would affect other soils with more shallow water tables. The “Pachic” modifier was added to indicate a thicker surface A horizon (greater than 50 cm) and would be more desirable for crop production. The main physical limitations to these soils were their vertic properties. Argiaquolls

The Argiaquolls are another great group of the Mollisols that are characterized by an argillic (Bt) horizon, a very dark mollic epipedon (chromas of 1 or 2) and horizon(s) directly underneath the mollic epipedon that have reddish hues (Table 4) (Soil Survey Staff 1999). Redoximorphic features are commonly found in the mollic epipedon of Argiaquolls, but this was not true in the two soils investigated in our study area. This reflected adequate drainage in the zone of root development even after water saturating events. All of the Argiaquolls were associated with the “Vertic” modifier because of the high amounts of shrink-swell clays found, which were apparent from cracks as wide as 2 cm. Some of the agricultural limitations of these soils are a high water table during the rainy season that restricts root growth and poor workability of the soils because of heavy clays (mean %clay = 66; std. dev. = 1.88). Udorthents

The Udorthents are Entisols that lack the general defining characteristics of the other soil orders (Table 4). They are different from the Mollisols because, they have no mollic epipedon even though the profile can be quite deep. The two Udorthents in the study area were further classified as Oxyaquic Udorthents. Because of their location on the banks of the Usumacinta River, they are saturated during the rainy season within 150 cm of the surface. The loamy

10

texture of these soils was coarser and desirable for almost any crop, assuming adequate moisture and nutrient could be provided. Comparison of great groups

The Haprendolls had better drainage than the Argiudolls (p < 0.02). However, under very dry conditions they may not hold adequate water long enough to sustain the crops during the dry season. Kunen et al. (2000) found insufficient soil moisture in the “high forest” soils surrounding Yaxha, Guatemala. The question of available soil moisture among soil types has yet to be investigated in this area. The two Udorthents were both excessively drained as would be expected by their coarse textures. Most of the A horizons (80%) for both the Haprendolls and Argiudolls had an LE of 9 or greater which placed them in the “very high” shrink-swell class, and indicated possible damage to crop roots and difficulty working with the soil. However, the Argiudolls were significantly lower in LE than the Haprendolls (p < 0.02), which suggests that the former soils had better tilth, making them easier to work with under dry conditions. The low LE values for the A horizon of the Argiudolls were also evidence for the eluviation of clays from the A horizons to the Bt horizons of the Argiudolls. The Udorthents had the lowest LE of all the great groups (mean = 1.2; std. dev. = 0.58).

For most plants, maximum nutrient availability occurs at pH ranges of 6.5 to 7.5. The average pH for the Haprendolls was 6.9 (std. dev. = 0.45) and was significantly higher than the pH for the Argiudolls which was 6.3 (p < 0.07 level). The two Argiaquolls had an average pH of 5.5 (std. dev. = 0.60) which is low and suggested that some plant nutrients such as nitrates and potassium could be limited in these soils. Depth of A horizon was significantly greater (p < 0.06) for the Argiudolls (mean = 46.5; std. dev. = 24.95) than for the Haprendolls (mean = 24.5; std. dev. = 7.62). This difference would have been more drastic if the area were subject to deforestation and increased erosion. The Argiaquolls A horizons were moderately deep (mean = 36.5; std. dev. = 10.60) and the Udorthents were the most shallow (mean = 15; std. dev. = 0).

Discussion A scoring scheme was devised to compare the agricultural potential of all the soil types in

the study area (Table 5). Under this system, higher scores reflect more agriculturally limited soils. It was patterned after the USDA system for land capability classification but omits some limiting factors that were less applicable to the current study. For example, there was no significant difference between organic matter and texture in the soils, so more meaningful parameters were chosen for relative comparisons such as pH and LE. The scores reflect only hand cultivation techniques. The method follows Fedick (1996), except that here we look at individual soil types and not the entire soil series and do not break them into a classes. Although the scoring scheme serves to make comparisons, each factor was weighted the same. Therefore, the reality that some factors would more heavily limit agriculture was not reflected in the overall score. Also, the actual area of soil types was not determined in this study and so this component of soil availability could not be considered. From the distribution of scores, it is evident that many of the soil types have few limitations to hand cultivation. The soils that scored 7 or 8 have the least limitations and may have been the most preferred for agriculture in ancient times. The Haprendolls and Oxyaquic Vertic Argiudolls scored the about the same for each factor. For year round cultivation in flat areas they would be considered the most suitable for maize production because of the adequate fertility and better drainage. Drainage would become especially critical under abnormally wet

11

years. Contemporary Lacandon Maya farmers on the Mexican side of the river, generally plant a crop around May or June when the rains begin and again in the fall (Nations and Nigh 1980). These soils would sustainable if certain measures were taken to control nutrient leaching and soil erosion such as planting certain fast growing crops (Nations and Nigh 1980) and construction of earthworks or terraces. If no measures were taken, these soils would be more susceptible to soil erosion and under deforestation and intense land pressure might be less favored. As a result, ancient farmers might have moved on to deeper and more workable soils as populations increased.

When obligated to look for more sustainable soils, albeit more marginal, a good candidate would have been the Vertic Argiudolls. Despite their poor drainage compared to the Haprendolls, the Pachic Vertic and Vertic Argiudolls have the advantage of generally better workability. They also show no limitations due to pH. During the dry season they would also hold more soil moisture. On the other hand, during the rainy season they are often flooded and can’t be used without special drainage modifications. No such modifications were immediately observed which suggests that this level of labor-intensive agriculture was not necessary in this part of the Maya Lowlands as in others (Bloom et al. 1983, 1985; Pohl et al. 1996; Turner and Harrison 1983).

The Aquertic Argiudolls and Argiaqoulls had the greatest agricultural limitation. In addition to poor drainage and workability, the low pH due to aquic conditions could cause nutrient deficiencies. Nevertheless, the poorly drained soils would have been important when extreme environmental variability became an issue. Under drought conditions, when there was not enough rain and sufficient soil moisture, perhaps farmers would be obligated to farm the poorly drained soils. The question stands whether or not these marginal soils were used extensively and intensively for maize cultivation. Stable carbon isotope analysis is one new method that may help resolve this issue (Johnson et al. n.d. this volume).

The Udorthents scored favorably and may have been used by ancient farmers because of low relief, accessibility from the river, and workability. However, Nations and Nigh (1980) found that the Lacandon Maya in Chiapas, Mexico rejected sandy soils, such as the Udorthents, as possible locations to farm. They preferred instead loosely packed red and black soils. This suggests that our score is inaccurate for this soil type as it does not consider other important factors such as geographic extent. Indeed, these soils are uncommon and could not support large populations because of their limited area along the banks of the Usumacinta River.

Conclusions Pedology is a useful tool in the study of ancient Maya agriculture because an

understanding of soils and soil resources are fundamental to answering broader land resource questions. This study demonstrated that soil profile investigations were useful in the qualitative comparison of the soil resources in between the Piedras Negras and Yaxchilan polities. It is also a beginning point for the examination of adaptations made by ancient farmers in their struggle for subsistence in this understudied region of the Maya Lowlands.

Investigations of soil properties indicate that overall, soils are desirable in the study area for agriculture and they may have provided for large-scale crop production. Further investigations that involve stable carbon isotope analysis and geographic information science also provide clues for the ancient agricultural potential of this interesting region (see Johnson et. al. this volume). The Haprendolls were common in the study area and would have had minimal challenges to cultivation compared to the Argiudolls. The desirable characteristics associated with the Haprendolls were: good drainage, gentle slopes, overall fertility, absence of shrink

12

swell clay and clay textures (i.e. adequate, but not excessive amounts of clay). Therefore, these soils, along with the Udorthents, would have been more preferred than the Argiudolls until a certain carrying capacity of the soil was reached because of a growing population. If population continued to grow, then prime agricultural land would have become less available and steeply sloped soils would have become more eroded. At this point, more marginal soils such as the common Argiudolls, and then the Argiaquolls may have been cultivated seasonally under after considering water saturation and methods of agricultural intensification.

No physical evidence of intensive agriculture was discovered in the form of agricultural terraces, wetland manipulation or household gardens. However, we cannot rule out that intensive forms of agriculture were practiced, especially when considering the study area’s location between the large population centers of the Yaxchilan and Piedras Negras kingdoms. Furthermore, the limited extent of the area of reconnaissance and the inherent difficulties of detecting cultural artifacts in the dense subtropical rainforest may have left many features undiscovered. It is hoped that future studies that involve highly resolved remote sensing data will improve the capability to detect such ancient agricultural artifacts.

Acknowledgements Funds for this research were provided in part by Brigham Young University in collaboration and the Sierra del Lacandón Archaeological Project, funded by the Foundation for the Advancement of Mesoamerican Studies, Inc. (FAMSI). Additional support was provided to the Parque Nacional Sierra del Lacandón by the World Monuments Fund. Project participants included: Americo Ixcayau, Chico León, Pánfilo Regino Hernandez, Eduardo Martínez, Ismael Mijangos, José Luis Aldana Alvarado, Elder Urbelino Alvarado Rodríguez, Leonardo de Jesús Damián Méndez, Huber Edilberto Heredia Rodríguez, Graciano Saquij Barahona, Eduardo Isidoro Saquij Choc. We thank the Defensores de la Naturaleza, particularly Marie-Claire Paíz and Rosa Maria Chan, whose collaboration has made this work possible. The authors would also like to thank the many hours dedicated by BYU undergraduate mentorees who helped with the soil analyses including: David Wright, Sarah Kendall, Susanna Peña and Marcos Alvarez. Statistical analyses were aided by The Center for Collaborative Research and Statistical Consulting at BYU. The project took place within a regional concession graciously granted by the Instituto de Antropología e Historia (IDAEH). The Consejo Nacional de Areas Protegidas (CONAP) granted permission for this work.

13

References Cited

Adams, R. E. W., Brown Jr., W. E., and Culbert T.Patrick 1981 Radar mapping, archaeology and ancient Maya land use Science . 1457-1463.

Aliphat, M. M. 1996 Ecology and settlement in the Usumacinta River Basin. University of Calgary.

Beach, T. 1998 Soil contraints on northwest Yucatan Mexico: Pedoarchaeology and Maya subsistence at Chunchucmil. Geoarchaeology: An International Journal 13:759-791.

Beach, T., Dunning, N., Luzzadder-Beach, S., and Scarborough, V. 2003 Chapter 8 Depression soils in the lowland tropics of northwestern Belize: Anthropogenic and natural origins. In The Lowland Maya Area: Three Millenia at the Human-Wildland Interface, edited by Gomez-Pompa, A., Allen, M. F., Fedick, S. L., and Jimenez-Osornio, J. J., pp. 139-174. Food Products Press, New York.

Beach, T. and Dunning, N. P. 1995 Ancient Maya terracing and modern conservation in the Peten rainforest in Guatemala. Journal of Soil and Water Conservation 50:138-145.

Beach, T., Luzzadder-Beach, S., Dunning, N., Hageman, J., and Lohse, J. 2003 Upland agriculture in the Maya Lowlands: Ancient Maya soil conservation in Northwestern Belize. The Geographical Review 92(3):372-397.

Binford, M. W., Brenner, M., Whitmore, T. J., Higuera-Gundy, A., and Deevey, E. S. 1987 Ecosystems, paleoecology and human disturbance in subtropical and tropical America. Quaternary Science Review 6:115-128.

Bloom, P. R., Pohl, M., Buttleman, C., Wiseman, F., Covich, A., Miksicek, C., Ball, J., and Stein, J. 1983 Prehistoric Maya wetland agriculture and the alluvial soils near San Antonio Rio Hondo, Belize Nature. 417-419.

Comision Federal de Electricidad, Direccion de Proyectos de Invasión Financiada, Subdirección de Construccion, y Coordinación de Proyectos Hidroelectricos 2003 Proyecto Boca del Cerro. Mexico DF.

Cowgill, U. M. 1962 An agricultural study of the southern Maya Lowlands. American Anthropologist 64(2):273-286.

14

Culbert, T. P., Levi, L. F., and Cruz, L. 1990 Lowland Maya wetland agriculture: The Rio Azul Agronomy Program. In Visoin and revision in Maya studies, 1st ed. Series edited by Flora S. Clancy and Peter D. Harrison. University of New Mexico Press.

Deevey, E. S., Rice, D. S., Rice, P. M., Vaughan, H. H., Brenner, M., and Flannery, M. S. 1979 Mayan urbanism: impact on a tropical karst environment. Science 206:298-306.

Dunning, N. P., D. J. Rue, T. Beach, A. Covich, and A. Traverse 1998 Human-environment interactions in a tropical watershed: the paleoecology of Laguna Tamarandito, El Petén, Guatemala. Journal of Field Archaeology 25:139-151.

Dunning, N. P., Luzzander-Beach, S., Beach, T., Jones, J. G., Scarborough, V., and Culbert, T. P. 2002 Arising from the bajos: The evolution of a neotropical landscape and the rise of Maya civilization. Annals of the American Association of Geographers 92(2):267-283.

Dunning, N. and Beach, T. 2000 Stability and Instability in Prehispanic Maya Landscapes. In Imperfect Balance: Landscape Transformations in the Precolumbian Americas, edited by Lentz, D. L., pp. 179-202. University Press, New York.

Dunning, N., Rue, D. J., Beach, T., Covich, A., and Traverse, A. 1998 Human-Environment Interactions in a tropical watershed: The paleoecology of laguna Tamarindito, el Petén, Guatemala. Journal of Field Archeology 25:139-151.

Dunning, N. P. 1996 A Reexamination of Regional Variability in the Pre-Hispanic Agricultural Landscape. In The Manage Mosaic: Ancient Maya Agriculture and Resource Use, edited by Fedick, S. L., pp. 53-68. University of Utah Press, Salt Lake City.

Dunning, N. P. and Beach, T. 2004 Noxious or nurturing nature? Maya civilization in environmental context. In Golden, Charles W.// Borgstede, Greg, pp. 125-141. Routledge, New York.

Dunning, N. P. and Beach, T. 1994 Soil Erosion, slope management, and ancient terracing in the Maya lowlands. Latin American Antiquity 5:51-69.

Dunning, N. P., Luzzadder-Beach, S., Beach, T., Jones, J. G., Scarborough, V., and Culbert, T. P. 2002 Arising from the Bajos: The Evolution of a neotropical landscape and the rise of the Maya Civilization. Annals of the Association of American Geographers 92(2):267-283.

15

Fedick, S. 1994 Ancient Maya agricultural terracing in the Upper Belize River area. Ancient Mesoamerica 5:107-127.

Fedick, S. L. 2001 Ancient manipulation of wetlands in the Northern Maya lowlands: Report on archaeological investigations of 1996-97. ed. FAMSI, FAMSI website.

Fedick, S. L. 1996 An Interpretive Kaleidoscope: Alternative Perspectives on Ancient Agricultural Landscapes of the Maya Lowlands. In The Manage Mosaic: Ancient Maya Agriculture and Resource Use, edited by Fedick, S. L., pp. 107-131. University of Utah Press, Salt Lake City.

Fedick, S. L. 1995 Land evaluation and ancient Maya land use in the upper Belize River area, Belize, Central America. Latin American Antiquity 6:16-34.

Fernandez, F. G. 2002 Chemical and Physical Properties of Anthrosols in Rural Areas Near the Ancient Maya City of Piedras Negras, Guatemala. Brigham Young University.

Ford, A. 1990 Maya settlement in the Belize River area: variations in residence patterns of the central Maya lowlands. In Precolumbian Population History in the Maya Lowlands, edited by Culbert, T. P. and Rice, D. S., pp. 167-182. New Mexico Press, Albuquerque.

Gee, G. W. and Bauder, J. W. 1982 Particle Size Analysis. In Methods of Soil Analysis: Part 1 - Physical and Mineralogical Methods, edited by Klute, A., pp. 383-411. 2 ed. American Society of Agronomy Inc. and Soil Science Society of America Inc., Madison, WI.

Golden, C., Munoz, A. R., Scherer, A., Zamora, M., Ramirez, E. R., and Johnson, K. n.d. Of frontiers and floods: Recent investigations in the Sierra del Lacandon National Park, Peten, Guatemala.

Golden, C., Vassaux, A. G., and Parnell, J. J. 2001 Reconocimiento en Macabilero, Capitulo 30. In Proyecto Arqueológico Piedras Negras: Informe Preliminar No.4, Cuarta Temporada, 2000, edited by Houston, S. D. and Escobedo, H. L., pp. 515-527.

Hansen, R. D., Bozarth, S., Jacob, J., Wahl, D., and Schreiner, T. 2002 Climatic and environmental variability in the rise of Maya Civilization. Ancient Mesoamerica 13:273-295.

16

Harrison, P. D. 1993 Aspects of water management in the southern Maya lowlands . In Economic Aspects of Water Management in the Prehispanic New World. Research in Economic Anthropology Supplement 7, edited by Scarborough, V. and Isaac, B., pp. 71-122. JAJ Press, Greenwich, CT.

Harrison, P. D. and Turner II, B. L. 1978 Prehispanic Maya Agriculture. New Mexico Press, Albuquerque.

Hernandez, A. A., Guenter, S. P., and Zender, M. U. 2003 Sak Tz'i, a Classic Maya Center: A location model based on GIS and epigraphy. Latin American Antiquity 14(2):179-191.

Hodell, D. A., Brenner, M., Crutis, J. H., and Guilderson, T. 2001 Solar forcing of drought frequency in the Maya lowlands. Science 292:1367-1370.

Houston, S., Escobedo, H., Child, M., Golden, C., Terry, R., and Webster, D. 2000 In the land of the Turtle Lords, archaeological investigations at Piedras Negras, Guatemala, 2000. Mexicon 22:97-110.

Houston, S. D., Escobedo, H., Hardin, P. J., Terry, R. E., Webster, D., Child, M., Golden, C., Emery, K., and Stuart, D. 1999 Between mountains and sea: investigations at Piedras Negras, Guatemala, 1998. Mexicon XXI:10-17.

Jacob, J. S. 1995 Archaeological pedology in the Maya lowlands. Pedological Perspectives in Archaeological Research 44:51-80.

Jensen, C. 2003 Soil Chemical Investigations of Agricultural Resource Control and Distribution in Chunchucmil, Mexico and Motul de San Jose, Guatemala. Brigham Young University.

Johnson , K. D., Wright, D. R., and Terry, R. E. n.d. Application of Carbon Isotope Analysis to Ancient Maize Agriculture in the Peten Region of Guatemala.

Johnston, K. 2003 The intensification of pre-industrial cereal agriculture in the tropics: Boserup, cultivation lengthening, and the Classic Maya. Journal of Anthropological Archaeology 22:126-161.

Kunen, J. L., Culbert, T. P., Fialko, V., McKee, B. R., and Grazioso, L. 2000 Bajo communities: A case study from the Central Peten. Culture and Agriculture 22(3):15-31.

17

Leyden, B. W., Brenner, M., Hodell, D. A., and Curtis, J. H. 1993 In Late Pleistocene Climate in the Central American Lowlands, edited by Leyden, B. W., Brenner, M., Hodell, D. A., and Curtis, J. H., American Geophysical Union.

Mendez, C. 1999 How Old is the Peten Tropical Forest? In Thirteen Ways of Looking at a Tropical Rainforest, edited by Nations, J. D., pp. 31-34. Conservation International, Washington, D.C.

Nations, J. D. and Nigh, R. B. 1980 The evolutionary potential of Lancandon Maya sustained-yield tropical forest agriculture. Journal of Anthropological Research 36(1):1-30.

Nelson, D. W. and Sommers, L. E. 1996 Total Carbon, Organic Carbon and Organic Matter. In Methods of Soil Analysis. Part 3. Chemical Methods, pp. 961-1010. vol. SSSA Book Series no. 5, Soil Science Society of America and American Society of Agronomy, Madison, Wisconsin, USA.

Pohl, M. 1985 An Ethnohistorical Perspective on Ancient Maya Wetland Fields and Other Cultivation Systems in the Lowlands. In Prehistoric Lowland Maya Environment and Subsistence Economy, edited by Pohl, M., pp. 35-45. Harvard University Press, Cambridge, Massachusetts.

Pohl, M. and Bloom, P. 1996 Prehistoric Maya Farming in the Wetlands of Northern Belize: More Data from Albion Island and Beyond. In The Manage Mosaic: Ancient Maya Agriculture and Resource Use, edited by Fedick, S. L., pp. 145-164. University of Utah Press, Salt Lake City.

Pohl, M. D., Pope, K. O., Jones, J. G., Jacob, J. S., Piperno, D. R., deFrance, S. D., Lentz, D. L., Gifford, J. A., Danforth, M. E., and Josserand, K. J. 1996 Early agriculture in the Maya lowlands. Latin American Antiquity 7(4):355-372.

Pope, K. A., Pohl, MD., and Jacob, J. S. 1996 Formation of ancient Maya wetland fields: Natural and anthropogenic processes. In The Managed Mosaic: Ancient Maya Agriculture and Resource Use, edited by Fedick, S. L., pp. 165-176. University of Utah Press, Salt Lake City.

Pope, K. O. and Dahlin, B. H. 1989 Ancient Maya Wetland Agriculture: New Insights from Ecological and Remote Sensing Research. Journal of Field Archaeology 16:87-106.

Proskouriakoff, T. 1960 Historical implications of a pattern of dates at Piedras Negras, Guatemala. American Antiquity 25:454-475.

18

Puleston, D. E. 1978 Terracing, raised fields, and tree cropping in the Maya lowlands: A new perspective on the geography of power. In Pre-hispanic Maya agriculture. Series edited by Peter D. Harrison and B.L. Turner II. University of New Mexico, Albequerque.

Sever, T. L. and Irwin, D. E. 2003 Landscape archaeology: Remote-sensing invetigations of the ancient Maya in the Peten rainforest of northern Guatemala. Ancient Mesoamerica 14:113-122.

Soil Science Society of America 2002 Online Soil Glossary.

Soil Survey Staff 1999 Keys to soil taxonomy. 8 ed. Pocahontas Press Inc., Blacksburg, VA.

Turner II, B. L. 1985 Issues Related to Subsistence and Environment among the Ancient Maya. In Prehistoric Lowland Maya Environment and Subsistence Economy, edited by Pohl, M., pp. 195-209. Harvard University Press, Cambridge, Massachusetts.

Turner II, B. L. 1990 Population Reconstruction of the Central Maya Lowlands: 1000 B.C. to A.D. 1500. In Precolumbian Population History in the Maya Lowlands, edited by Culbert, T. P. and Rice, D. S., pp. 301-324. University of New Mexico Press , Albuquerque, NM.

Turner II, B. L. 1974 Prehistoric Intensive Agriculture in the Mayan Lowlands. Science 195:118-124.

Turner II, B. L. and Harrison, P. D. 1983 Pulltrouser Swamp: Ancient Maya Habitat, Agriculture, and Settlement in Norther Belize. University of Texas Press, Austin.

Webster, D. 2002 The Fall of the Ancient Maya. ed. Thames and Hudson Ltd., London.

Whitmore, T. M., Turner II, B. L., Johnson, D. L., Kates, R. W., and Gottschange, T. R. 1990 Long-term population change. In The Earth as Transformed by Human Action, edited by Turner II, B. L., Clark, W. C. , Kates, R. W., Richards, J. F., Mathews, J. T., and Meyer, W. B., pp. 25-39. Cambridge University Press, Cambridge.

19

TABLES

Table 1. Summary of limitations to agriculture in some Guatemalan soils.

Table 2. Physical and Chemical Properties of Haprendolls.

Table 3. Physical and Chemical Properties of Argiudolls.

Table 4. Physical and Chemical Properties of Argiaquolls and Udorthents

Table 5: Scoring scheme for agricultural potential among soil types based on limiting soil characteristics.

FIGURES

Figure 1. Soil profile locations and taxonomic great groups for the study area.

20

21

Table 1. Summary of limitations to agriculture in some Guatemalan soils.

Location Environment Physical Limitations Chemical Limitations Source Three Rivers Region, northwestern Belize

Bajo

Excessive clay content (58-94%)

Shrink-swell clays

Soil pH (4.5 – 8.3) Acidic conditions in surface

horizons Low Zinc (0mg/kg) Low Phosphorus (47 ppm

average for three fractions)

Beach et al. (2003a)

Mirador Basin, northern Guatemala

Bajo Poor drainage Excessive clays Shrink-swell clays Other challenges

associated with limited water availability and vegetation that is difficult to remove

Hansen et al. (2002)

Succotz, Belize

Floodplain Low water retentions because of sandy loam soils

Easily eroded soils

Low Phosphorus (11.7 ppm average extractable P; Bicarbonate method)

Arnason et al. (1982)

Chunchucmil, Mexico

Coastal High water tables Low water retention

because of coarse textures (sand to fine sandy loams)

Low Phosphorus (1-18 ppm extractable P; Bicarbonate method)

High salinity (14 dS-1) High pH (about 8.0)

Beach (1998)

Chunchucmil, Mexico

Karst Plain Insufficient soil depth Poor drainage

Low Phosphorus (94 ppm average extractable P; Bicarbonate method)

Low Potassium (90 ppm) Low Zinc (0.5 ppm)

Beach (1998)

Piedras Negras, Guatemala

Upland Karst

Easily eroded soils Low Phosphorus (8.3 ppm average extractable P; Mehlich II method)

Fernandez et al. (2003)

Motul de San José, Guatemala

Upland Karst

Excessive clay content (34-67%)

Easily eroded soils

Low Phosphorus (5.6-16.6 ppm extractable P; Mehlich II method)

Jensen et al. (2004)

Aguateca, Guatemala

Upland Karst

Easily eroded soils Low Phosphorus (5.82 ppm average extractable P; Bicarbonate Method)

Johnson et al. (2004)

Table 2. Physical and Chemical Properties of Haprendolls.

Description Hillslope Draina Hor. Depth LEPb Clay Textc Strd Cone Color Color R.F.f pH P Total Total Notes Name cm % % dry dry wet ppm O.N O.C Great Group: HAPRENDOLLS Subgroup

Vertic Haprendolls Footslope WD A 0-5 23 78 c 2,gr VH 10YR 2/1 10YR 2/1 7.1 3.1 0.86 7.65 Old growth CUEVA A2 5-30 21 83 c 2,gr VH 10YR 2/1 10YR 2/1 7.2 3.7 0.37 3.58 forest B 30-60 6 32 sicl 1,m HA 10YR 5/6 10YR 4/6 7.5 2.9 0.13 1.32

Cr 60+ Vertic Haprendolls Backslope SE A 0-15 19 67 c 2,gr HA 10YR 2/2 10YR 2/2 7.2 4.6 1.12 10.25 Secondary FAJARDO2 Bt 15-30 21 77 c 2,gr HA 10YR 3/3 10YR 3/3 7.4 4.1 0.36 3.13 forest BC 30-60 14 57 c 2,gr VH 10YR 6/3 10YR 6/3 7.8 3.0 0.07 0.67

Cr 60+ Vertic Haprendolls Footslope MW A 0-10 12 63 c 1,gr HA 10YR 2/2 10YR 3/3 6.9 4.5 0.64 5.36 Old growth TEP3 A2 10-30 17 73 c 2,gr HA 7.5YR 5/3 10YR 3/3 6.3 2.9 0.12 1.15 forest Bt 30-60 21 93 c 2,gr HA 10YR 3/6 7.5YR 4/6 5.0 4.1 0.09 0.79

Btc 60-90 20 78 c 2,m VH 10YR5/6 7.5YR 5/6 n 6.7 3.5 0.08 0.69 BCc 90-105 6 42 sc 2,m VH 10YR 7/6 10YR 7/6 7.8 4.7 0.06 0.46

Cr 105+ Vertic Haprendolls Footslope WD A 0-20 21 57 c 2,gr HA 10YR 2/2 10YR 2/2 7.4 30.8 0.68 6.59 Secondary LAPISTA B 20-75 9 62 c 2,gr HA 10YR 3/6 10YR 3/6 7.4 22.2 0.22 2.12 forest Ab 75-105 13 52 c 2,m HA 10YR 3/3 10YR 2/2 7.5 10.2 0.16 1.90 Near mounds.

Bb 105-120 0 32 cl 2,gr HA 10YR 3/6 10 YR 3/4 7.6 16.8 0.13 1.42 Cr 120+

Avg. for A horizons 23.75 19 70 7.0 8.3 0.63 5.76

Lithic Haprendolls Footslope MW A 0-15 20 93 c 1,gr SH 10YR 2/2 10YR 2/2 6.0 4.2 0.70 6.25 Old growth ARO2 A2 15-30 20 100 c 2,m SH 10YR 3/3 10YR 3/3 6.0 4.3 0.32 2.72 forest

Cr 30+ Lithic Haprendolls Footslope SE A 0-15 11 51 c 2,gr HA 10YR 4/2 10YR 2/2 7.0 0.33 2.89 Secondary PAPAYO2 A2 15-30 9 66 c 2,gr HA 10YR 4/2 10YR 3/3 7.5 0.17 1.57 forest AC 30-45 9 71 c 2,gr HA 10YR 4/3 10YR 4/3 7.9 0.11 1.02

Cr 45+ Lithic Haprendolls Backslope WD A 0-10 17 73 c 1,gr SH 10YR 2/2 10YR 2/2 7.2 5.2 1.16 8.79 Secondary

22

ESMERALDA MILPA AC 10-20 23 93 c 2,m HA 10YR 3/3 10YR 3/3 7.3 6.6 0.55 3.94 Close to.

C 20-30+ 0 18 sil 1,m EH 10YR 8/2 10YR 8/2 7.1 4.7 0.10 7.99 mounds Lithic Haprendolls Footslope WD A 0-20 23 68 c 2,gr HA 10YR 2/2 10YR 2/2 6.9 6.4 0.53 4.27 Old growth GROUP44 AC 20-30 2,gr HA 10YR5/2 10YR 3/3 0.21 2.17 Near mounds

Cr 30+ Lithic Haprendolls Footslope WD A 0-10 18 88 c 3,gr HA 10YR 2/1 10YR 2/2 6.3 4.0 0.45 3.91 Old growth RANDOM2 A2 10-30 22 103 c 2,m VH 10YR 2/1 10YR 2/2 6.7 3.1 0.33 3.43 forest AC 30-45+ 23 88 c 2,m VH 10YR 4/2 10YR 3/3 7.2 0.26 2.78

Avg. for A horizons 26.00 17 80 6.7 4.5 0.50 4.23

Typic Haprendolls Footslope WD A 0-30 21 67 c 3,gr VH 10YR 2/1 10YR 2/1 m 7.0 3.1 0.57 5.87 Old growth EL CINE Bc 30-90 4 27 l 2,gr SH 7.5YR 7/4 7.5YR 5/4 m 7.5 3.6 0.07 0.72 Near small.

Bc2 90-195+ 6 27 l 2,gr SH 7.5YR 7/4 7.5YR 5/4 n 7.6 3.8 0.03 0.37 spring

aDrainage: PD = Poorly Drained; SP = Somewhat Poorly Drained; MW = Moderately Well Drained; WD = Well Drained; SE = Somewhat Excessively Drained; E = Excessively Drained bLinear Extensibility Percent (LEP) is the percentage volume change to the dry length of a soil clod; LEP = ((moist length - dry length)/dry length)*100 cTexture: sl = sandy loam; l = loam; sil = silt loam; scl = sandy clay loam; cl = clay loam; sicl = silty clay loam; sc = sandy clay; sic = silty clay; c = clay dStructure: 1 = weak; 2 = moderate; 3 = strong; gr = granular; m = massive eConsistency = slightly hard; HA = hard; VH = very hard; EH = extremely hard fR.F. = Redoximorphic Features; n = nodules; m = mottles; d = depletions

23

Table 3. Physical and Chemical Properties of Argiudolls.

Description Hillslope Draina Hor. Depth LEPb Clay Textc Strd Cone Color Color R.F.f pH P Total Total Vegetation/

Name cm % % dry dry wet ppm O.N O.C. Other notes

Great Group: ARGIUDOLLS

Subgroup

Typic Argiudolls Footslope SP A 0-15 16 67 c 2,gr HA 10YR 2/1 10YR 2/1 6.3 0.58 5.65 Old growth forest

FAJARDO CAMP Bt 15-60 16 72 c 2,m HA 10YR 2/2 10YR 2/2 6.6 0.11 1.42 Near small arroyo.

Btc 60-105 13 52 c 2,m VH 10YR 4/3 10YR 4/3 8.0 0.02 0.48

Btgc 105-120+ 16 43 c 2,m VH 10YR 5/2 5B 6/1 7.9 0.02 0.30

Typic Argiudolls Footslope MW AB 0-60 5 53 c 2,gr VH 10YR 5/4 10YR 3/4 5.6 3.7 0.17 0.92 No canopy

PORVENIR Bt 60-150 10 68 c 2,gr HA 10YR 4/3 10YR 3/4 6.1 4.0 0.07 0.26 Near park station.

Bw 150-180 14 53 c 3,gr SH 10YR 5/3 10YR 3/4 7.2 2.9 0.08 0.38 Modern artificially

Cr 180+ leveled surface

Typic Argiudolls Toeslope SP A 0-10 2,gr MH 10YR 2/1 10YR 2/1 0.93 8.60 No canopy

CIVAL ABt 10-60 20 72 c 2,m MH 10YR 4/1 7.5YR 2.5/1 7.3 9.7 0.21 2.86 On the edge of large cival.

Bt 60-150 18 73 c 2,m HA 10YR 5/4 10YR 5/4 d 7.5 7.7 0.03 0.63 Sedge vegetation

Avg. for A horizons 28.33 18 70 6.8 0.76 7.12

Aquertic Argiudolls Toeslope MW A 0-15 8 87 c 3,gr SH 10YR 6/3 10YR 4/3 5.9 4.3 0.36 3.25 Old growth forest

PAPAYO1 B 15-30 13 92 c 3,m SH 10YR 5/2 7.5YR 5/2 5.7 3.5 0.15 1.02

Bg 30-45 15 92 c 3,m HA 7.5YR 5/2 7.5YR 5/2 m 5.5 3.5 0.10 0.75

Bg2 45-90 15 97 c 3,m HA 7.5YR 5/2 7.5YR 5/2 m 5.5 3.3 0.08 0.59

Btg 90-150 17 92 c 3,m HA 7.5YR 7/1 7.5YR 7/1 m 5.5 3.8 0.03 0.20

Btg2 150-165 20 87 c 3,m HA 7.5YR 8/1 7.5YR 8/1 m 4.7 6.2 0.01 0.07

Btg3 165-180 19 87 c 3,m HA 7.5YR 8/1 7.5YR 8/1 m/n 6.4 3.4 0.03 0.12

Btg4 180-210+ 3,m HA 7.5YR 8/1 7.5YR 8/1 m/n 3.2

Aquertic Argiudolls Toeslope WD A 0-30 6 36 cl 1,gr SH 10YR 2/2 10YR 2/2 5.8 4.8 0.36 3.65 Secondary forest

AGUACATE 1 ABt 30-60 9 56 c 2,sbk HA 10YR 3/3 10YR 3/3 5.5 4.3 0.10 0.90

Bt 60-135 12 61 c 3,m VH 10YR 5/3 10YR 5/3 5.6 3.9

BCc 135-165 10 51 c 3,m VH 10YR 5/3 10YR 5/3 n 6.0 3.8 0.03 0.15

BCc2 165-210+ 7 36 cl 3,m HA 10YR 5/6 10YR 5/6 d 6.3 3.3

Aquertic Argiudolls Toeslope SP A 0-15 16 51 c 2,gr VH 10YR 2/2 10YR 2/2 5.8 3.5 0.20 1.82 Old growth forest

24

ARO 1 A2 15-45 17 76 c 2,m HA 7.5YR 3/4 7.5YR 3/4 5.2 3.3 0.16 1.20

Btc 45-60 17 76 c 2,m VH 5YR 3/4 5YR 3/4 n 5.4 4.1 0.14 0.92

Btc2 60-75 15 76 c 2,sbk HA 10YR 4/4 10YR 4/4 n 5.4 3.7 0.12 0.80

Btc3 75-135 16 87 c 2,sbk HA 10YR 4/4 10YR 4/4 m 5.4 4.9 0.10 0.54

Bt 135-150 17 97 c 2,m VH 10YR 3/4 10YR 3/4 5.7 3.5 0.11 0.58

Cr 150+

Avg. for A horizons 40.00 11 61 5.7 4.0 0.23 2.16

Vertic Argiudolls Footslope MW A 0-15 19 98 c 2,gr HA 10YR 2/1 10YR 2/1 6.7 3.0 0.77 6.94 Old growth forest

TEP2 Bt 15-30 21 103 c 2,m HA 10YR 3/3 10YR 3/3 6.7 3.4 0.26 2.54

Bt2 30-60 2,m HA 10YR 4/4 10YR 3/4 3.1 0.16 3.65

Cr 60+

Vertic Argiudolls Toeslope WD A 0-15 13 42 sic 2,gr MH 10YR 4/2 10YR 2/1 7.0 4.9 1.79 19.31 Old growth forest

TEXCOCO A2 15-45 6 42 sic 1,gr HA 10YR 4/3 10YR 3/3 7.7 8.5 0.88 8.82 Lake Plain-Terrace

B 45-90 0 32 sicl 2,gr SH 10YR 8/2 10YR 5/3 7.7 3.2 0.49 5.83

B2 90-135 6 48 sic 2,gr HA 10YR 3/3 10YR 3/3 d 7.7 7.6 0.25 3.39

Bt 135-150 17 73 c 2,m VH 10YR 6/4 10YR 4/4 7.7 2.9 0.10 1.27

Abt 150-180+ 21 97 c 2,m VH 10YR 3/3 10YR 3/3 7.6 5.1 0.05 0.51

Avg. for A horizons 30.00 13 61 7.1 5.4 1.15 11.69

Pachic Vertic Footslope SE A 0-45 10 38 cl 2,gr HA 10YR 4/2 10YR 2/2 7.6 5.1 0.86 8.70 Secondary forest

Argiudolls RANDOM A2 45-90 7 38 cl 2,gr HA 10YR 4/2 10YR 2/2 8.0 4.2 0.35 3.74 Small arroyo nearby.

Bc 90-120 11 43 c 2,m HA 10YR 6/1 10YR 5/1 8.1 4.7 0.22 2.85

Bt 120-165+ 13 53 c 2,m HA 5YR 5/2 10YR 5/1 m 8.0 5.4 0.04 0.57

Pachic Vertic Footslope SP A 0-30 11 77 c 2,gr SH 10YR 2/2 10YR 2/2 6.2 4.2 0.52 4.29 Old growth forest

Argiudolls PORVENIR2 A2 30-60 12 72 c 2,m HA 10YR 3/6 10YR3/6 6.0 4.0 0.41 3.22 Near the archaeological

Bt 60-150 18 92 c 2,m VH 10YR 3/6 10YR3/6 6.0 3.3 0.06 0.33 site of Porvenir

Btc 150-270+ 13 87 c 2,m VH 10YR 5/8 10YR 5/8 n 7.1 2.9 0.06 0.81

Avg. for A horizons 75.00 10 56 6.9 4.4 0.54 4.99

Oxyaquic Vertic Toeslope MW A 0-15 23 97 c 1,gr MH 10YR 2/1 10YR 2/1 7.0 2.9 0.34 4.47 Secondary forest

Argiudolls CAMP ABt 15-30 22 107 c 2,m MH 2.5Y 3/6 2.5Y 3/6 6.9 2.9 0.14 2.22 Near the Rio Macabilero.

B 30-45 13 67 c 1,gr SH 10YR 7/4 10YR 7/4 7.4 4.6 0.14 1.91

25

BC1 45-150 2 47 c 1,gr SH 10YR 7/4 10YR 7/4 7.4 3.4 0.07 1.08

BC2 150-195 0 17 sl 1,gr SH 10YR 5/8 10YR 5/8 7.5 3.4 0.09 1.47

Cr 195+

aDrainage: PD = Poorly Drained; SP = Somewhat Poorly Drained; MW = Moderately Well Drained; WD = Well Drained; SE = Somewhat Excessively Drained; E = Excessively Drained bLinear Extensibility Percent (LEP) is the percentage volume change to the dry length of a soil clod; LEP = ((moist length - dry length)/dry length)*100 cTexture: sl = sandy loam; l = loam; sil = silt loam; scl = sandy clay loam; cl = clay loam; sicl = silty clay loam; sc = sandy clay; sic = silty clay; c = clay dConsistency: 1 = weak; 2 = moderate; 3 = strong; gr = granular; m = massive eSH = slightly hard; HA = hard; VH = very hard; EH = extremely hard fR.F. = Redoximorphic Features; n = nodules; m = mottles; d = depletions

26

Table 4. Physical and Chemical Properties of Argiaquolls and Udorthents

Description Hillslope Draina Hor. Depth LEPb Clay Textc Strd Cone Color Color R.F.f pH P Total Total Notes Name cm % % dry dry wet ppm O.N. Great Group: ARGIAQUOLLS Subgroup

Vertic Argiaquolls Footslope MW A 0-30 6 68 c 2,gr HA 10YR 2/2 10YR 2/1 5.9 3.4 0.55 5.71 Old growth TEP1 B 30-90 12 68 c 2,m HA 10YR 3/4 10YR 3/4 4.8 2.9 0.05 0.38 Near rocky Btc 90-180 19 88 c 2,m HA 10YR 3/4 10YR 5/4 n 4.7 2.9 0.11 0.38 outcroppings

Btc2 180-195 13 93 c 1,gr MH 7.5YR 7/6 7.5YR 5/6 n 7.2 3.1 0.06 0.44 Near mounds. Bt 195-210+ 21 98 c 2,m MH 7.5YR 3/4 7.5YR 3/4 7.2 3.9 0.04 0.30

Vertic Argiaquolls Toeslope SP A 0-15 21 68 c 3,gr HA 10YR 2/1 10YR 2/2 4.9 6.7 0.56 5.91 Old growth ESMERALDA A2 15-45 21 63 c 3,m VH 10YR 2/1 10YR 2/2 5.1 4.1 0.13 1.64 PLANADA BAc 45-75 17 62 c 3,m VH 10YR 2/1 10YR 2/2 n 5.5 3.7 0.06 0.82 Btc 75-165 24 58 c 3,m VH 10YR 4/2 7.5YR 4/6 n 6.4 5.2 0.03 0.28 Btc2 165-180 3,m VH 10YR 5/1 7.5YR 4/4 n 0.02 0.19 Cr 180+

Average for A horizons 37.50 16 66 5.3 4.7 0.42 4.42

Great Group: UDORTHENTS Subgroup

Oxyaquic Udorthents Toeslope E A 0-15 1 28 sil 1,gr S 10YR 4/1 10YR 4/1 6.1 6.2 0.35 3.44 Secondary ARO3 BC 15-30 scl 1,m SH 10YR 5/3 10YR 5/3 8.4 0.10 0.77 Near mounds. C1 30-120 scl 1,sg S 10YR 5/4 10YR 5/4 0.03 0.16 Near the banks of C2 120-165 scl 1,sg L 10YR 5/3 10YR 5/3 Usumacinta River C3 165-210+ scl 1,sg S 10YR 5/4 10YR 5/4 Oxyaquic Udorthents Toeslope SE A 0-15 2 22 l 2,gr SH 10YR 4/1 10YR 2/1 6.1 4.3 0.27 2.59 Secondary PAPAYO3 C 15-30 1 27 scl 2,gr MH 10YR 4/3 10YR 3/3 6.4 4.5 0.08 0.74 Near the banks of C2 30-210+ 0 27 scl 2,gr SH 10YR 5/4 10YR 3/4 6.2 6.2 0.02 0.21 Usumacinta River

Average for A horizons 15.00 1 25 6.1 5.2 0.31 3.01

aDrainage: PD = Poorly Drained; SP = Somewhat Poorly Drained; MW = Moderately Well Drained; WD = Well Drained; SE = Somewhat Excessively Drained; E = Excessively Drained

27

bLinear Extensibility Percent (LEP) is the percentage volume change to the dry length of a soil clod; LEP = ((moist length - dry length)/dry length)*100 cTexture: sl = sandy loam; l = loam; sil = silt loam; scl = sandy clay loam; cl = clay loam; sicl = silty clay loam; sc = sandy clay; sic = silty clay; c = clay dConsistency: 1 = weak; 2 = moderate; 3 = strong; gr = granular; m = massive

eSH = slightly hard; HA = hard; VH = very hard; EH = extremely hard fR.F. = Redoximorphic Features; n = nodules; m = mottles; d = depletions

28

Table 5: Scoring scheme for agricultural potential among soil types based on limiting soil characteristics.

Taxonomic Description

Drainage Depth Workability(LE)

Fertility (pH)

Score

Vertic Haprendolls 1 3 3 1 8 Lithic Haprendolls 1 3 3 1 8 Typic Haprendolls 1 2 4 1 8 Typic Argiudolls 3 2 3 1 9

Aquertic Argiudolls 3 2 2 3 10 Vertic Argiudolls 2 2 2 1 7

Pachic Vertic Argiudolls

3 1 2 1 7

Oxyaquic Vertic Argiudolls

1 2 4 1 8

Vertic Argiaquolls 3 2 3 4 12 Oxyaquic

Udorthents 1 3 1 2 7

Scores: Drainage: 1 = well drained; 2 = moderately well drained; 3 = poorly drained Depth (cm): 1 = 50-75; 2 = 25-50; 3 = 0-25 Workability (LE): 1 = 0-9; 2 = 10-15; 3 = 15-20; 4 = 20-25 Fertility (pH): 1 = 6.5-7.0; 2 = 6.0-6.5; 3 = 5.5-6.0; 4 = 5.0-5.5

29

Figure 1. Soil profile locations and taxonomic great groups for the study area.

30

Ancient Soil Resources of the Usumacinta River Region,

Guatemala II: GIS and Stable Carbon Isotope Analyses

Kristofer Dee Johnson, Richard E. Terry, Mark W. Jackson Deparment of Plant and Animal Sciences and Department of Geography Brigham Young Univeristy Provo, UT 84602 USA Charles Golden Anthropology Department Brandeis University Waltham, MA 02454-9110 USA A manuscript to be submitted to Latin American Antiquity

Introduction

For three millennia, complex cultures managed to survive and even thrive in the seemingly limited environment of the Maya lowlands. Turner et al. (2003) cites several environmental conditions that would limit agricultural success in this region. Soil degradation in the form of erosion, soil nutrient losses and the loss of wetland soil resources due to sedimentation would have been formidable challenges. Crops likely suffered from droughts and other natural disasters that periodically occurred, not to mention weeds and pests (Turner et al. 2003; see also Nations and Nigh 1980). As their survival was threatened by natural or anthropogenic induced changes in environmental conditions, the ancient Maya consistently adapted with technologies and strategies. They were therefore successful land stewards for hundreds of years until their eventual collapse circa 800 to 900 A.D. Modern residents of the same area number a fraction of the ancient population and inefficiently subsist at the expense of their own natural resources (Turner et al. 2003). This leads to the question, “How did the Maya do it?” The ability of the ancient Maya to succeed, their methods of food production and subsistence strategies, have yet to be explained satisfactorily. Discovery of their methods may hold keys that help shape future long term subsistence in the Maya lowlands.

Environmental and archaeological studies of the ancient Maya suggest that the ancient agricultural landscape of the Guatemalan Lowlands was more complex than previously thought. It has become increasingly apparent that no single model is adequate to describe their interactions with the environment. Methods of food production varied spatially and temporally according to the stability or instability of environmental, political and economic conditions (Dunning 1996; Dunning et al. 1998; Dunning and Beach 2004; Fedick 1996). Therefore, a method to identify the locations and extent of ancient agriculture is much needed and fundamental to resolving questions about ancient Maya farming strategies. This process must involve a case by case approach where each independent study can be pieced together to form a more complete picture of the ancient agricultural landscape in the Maya Lowlands.

Part one of this soil resources study (Johnson et. al. this volume) used inductive means based on actual pedological data to determine the agricultural potential of the soil resources in the area located between the Piedras Negras and Yaxchilan polities in the Guatemalan Lowlands. In this study we combined the pedological interpretations with stable carbon isotope data from the same soil profiles and geographic information systems (GIS) methods to determine agricultural potential. The results from both approaches proved especially useful in the study area because of its remoteness and, like many areas of the Maya Lowlands, dense vegetation.

Background GIS and predictive modeling The utility of GIS as an archaeological tool are apparently becoming more recognized (Westcott and Brandon 2000). However, in many areas within the Maya Lowlands it is difficult to obtain adequate data materials to perform detailed analysis. This may change as more remote sensing data becomes available (e.g. NASA’s AIRSAR program). Nonetheless, in Belize where reliable soil maps are available, Fedick has been able to predict terrace occurrence (1994) and analyze ancient land use patterns (1995; 1996). More recently, in a study that did not depend on environmental data, Hernandez et al. (2003) was able to apply the Gravity Model and epigraphic data to predict the unknown geographic locations and territorial extents of three Maya capitals of the upper Usumacinta region.

32

Both deductive and inductive approaches have been used in GIS to produce rules that predict “positive” or “negative” outcomes. In an archaeological context, the deductive approach draws on an archaeologist’s experience to arbitrarily decide which environmental data (i.e. slope, aspect, distance to water, soil type etc.) and parameters best indicate areas of ancient settlement. In contrast, the inductive approach draws from previously collected field data, which are the environmental predictors, to produce the rules that can then be used for the prediction of ancient settlement locations (Wheatley and Gillings 2002, p. 166).

In most cases, the theory-driven deductive approach and the data-driven inductive approach are not independent. The two are usually combined to produce a more effective model (Wheatley and Gillings 2002, p. 166). For areas that are understudied or unexplored, or difficult to work in because of vegetation and terrain, the deductive approach is often applied first. The result is a preliminary model that can be used as a guide for reconnaissance purposes and sampling strategy even when data sources are limited or poorly resolved spatially. After initial field reconnaissance and data collection, the inductive approach can then be used to refine the model for more accurate interpretation. For example, Fedick (1994) used only two variables, slope and soil type, to predict the distribution of terraces in the Upper Belize River Area. Theoretically, soil series that were fertile, well-drained and shallow, and that form over limestone parent material, were chosen as predictors. A slope map derived from a digital elevation model was selected for two classes (4-10º and >10-47º). The study area was then surveyed for terraces and a total of thirteen previously unidentified terrace systems were found. Soil and slope data were collected for locations that had identified terraces as well as locations that lacked terraces. A new model based on the new empirical data was then created with three slope classes (0-3º, >3-9º, and >9-16º).

The GIS methods that were applied in the present study somewhat follow those of Fedick’s (1994, 1995, 1996) except that our predicted outcome was agricultural potential instead of settlement or terraces. The variables for the model were soil type and slope. Our model assumed that areas with gentle slopes and well drained soils were conducive to agriculture and thus would have been preferred by the ancient Maya. The topography of the Usumacinta River Basin typically contains karst lanforms (e.g. steep sloped rounded hills, narrow valleys, sinkholes, disappearing streams) and some large gently sloped and relatively homogenous areas. We designated such areas as “breadbaskets” of the region, assuming that the Maya farmed them more extensively. In this way, GIS modeling was used as a tool to “prospect” for potential ancient breadbasket areas and was very effective on a regional scale (soils and topographic data were too poorly resolved to be used practically on a local scale). Stable carbon isotope analysis was then used to “ground truth” our modeled predictions and answer the archaeological question of whether or not this region was used for the large scale cultivation of maize. Stable carbon isotope analysis

Soil chemical analyses have proven effective for the identification of remnant traces of human household activities in the Maya Lowlands (Parnell et al. 2002). Maya agricultural activities, on the other hand, have been interpreted with only conventional methods such as artifact collection (cultivation tools, Turner 1983), mapping of agricultural structures (terraces, Beach and Dunning 1995), paleobotany (Pope et al. 2001) and soil studies (Fernandez 2002). In the absence of artifacts and structures, a soil chemical approach would be a useful tool (or a helpful addition in other cases) to indicate areas of ancient agriculture. Stable carbon isotope analysis of soil organic matter, which identifies remnant traces of maize organic matter, is one potential tool (Johnson et al. n.d.; Webb et al. 2004).

33

A very small fraction (1.11%) of the total carbon in nature exists as the stable 13C isotope and nearly all the remaining carbon (98.89%) is of the isotope 12C (Boutton 1991). Most plants, including the forest species of the Maya Lowlands, posses a C3 photosynthesis system and are known as C3 plants. C4 plants, on the other hand, include tropical grasses and maize and are different from C3 plants in their photosynthetic pathway. In plant-soil relationships, there is a strong correlation between the 13C to 12C ratios (expressed as δ13C values) in plant communities and the δ13C values in soil organic matter (SOM) (Balesdent et al. 1993). C3 plants discriminate more heavily against the 13C isotope than do C4 plants. This discrimination results in higher δ13C values of C4 plant tissue and therefore higher SOM δ13C values for soils influenced by C4 plants (see Boutton 1996 and Webb et al. 2004 for discussion). The SOM δ13C value of C3 plants ranges from -32 to -22‰ with an average value of -26‰. The values for C4 plants range from -17 to -9‰ with an average value of -12‰ (Bird and Pousai 1997, Boutton et al. 1998). Maize (Zea mays) has a δ13C value of -12‰. If a crop or forest is long-standing, then the relative amounts of carbon isotope concentrations retained in the SOM are commonly preserved for hundreds to thousands of years and reflect the ancient composition of the plant species at that location (Boutton 1996).

Leyden et al. (1993) used a core sample taken from Lake Quexil to investigate the climate of the Guatemalan Lowlands of the late Pleistocene. Stable carbon and oxygen isotope analyses indicated that near the end of the last ice age (about 36,000 to 24,000 years ago) conditions in the region were extremely arid. Also, pollen analysis indicated that at that time the vegetation was mostly composed of pine and oak species, and a colder climate existed. Then, about 12,000 years ago, temperatures gradually rose and δ13C increased, which indicated a replacement of C3 plants by C4 plants (Leyden et al. 1993). Between about 9,000 and 8,000 years ago, still warmer and humid conditions prevailed along with the re-establishment of C3 arboreal taxa. This was followed by the gradual establishment of essentially the same temperature and moisture gradients that are observed today in the Maya Lowlands (Leyden 2002). Therefore, if we extrapolate Leyden’s (1993) findings to the current study area, the SOM of the last 9,000 years was formed under strictly C3 vegetation, except when broken by some anthropogenic change that would result in C4 vegetation.

Rosenmeier et al. (2002) investigated the pollen record of a core taken from Lake Salpeten, Guatemala in the central part of the Maya lowlands about 200 km from the current study area. They noted a shift from C3 arboreal species to C4 maize (Zea mays) beginning in the sixth century A.D. and lasting until the ninth century A.D. Widespread maize cultivation had already been established at a much earlier date, but was intensified during this time period (Webster 2002, p. 44). Rosenmeier et al. suggested that large forested near the lake were cleared for maize cultivation during the Late Classic. Not surprisingly, this observation coincides with ancient erosion episodes and indicates that a severely eroded landscape existed at some locations in the Maya Lowlands circa 800 A.D. (Beach 1998; Beach and Dunning 1995; Dunning and Beach 1994, 1997, 2000; Dunning et al. 1998; Fernandez 2002; Jacob 1995). Pope et al. (2001) found pollen from domesticated maize (Zea mays) in the Grijalva River delta of Mexico as early as 5,000 B.C. which would soon become the staple crop of the Maya (Webster 2002 pp. 171-172; White and Schwarcz 1989).

While the Maya no doubt diversified their diet with various fruits and vegetables, maize was certainly the most important crop (Turner and Mikisicek 1984; Webster 2002, pp. 171-172). The Maize God was the father of the important Hero Twins of Classic Maya Mythology (Taube 2001). Colonial Maya documents reveal that the First Father and the First Mother created

34

humans out of maize dough and thus the reason that contemporary Maya identify themselves as “maize people” (Wagner 2001). Therefore, much time, effort and land resources would have been dedicated to producing this life-sustaining and religious food source. Maya populations numbered into the millions at its apex in the Late Classic Period about 450 to 950 A.D. (Whitmore et al. 1990), which suggests that much of the landscape was dedicated to large scale maize production. Indeed, Scherer et al. (2002) found that stable carbon isotopes of bone collagen from the peoples living at Piedras Negras during the Classic Period indicated a maize rich diet (see also White and Schwarcz 1989).

Few studies have recognized the enrichment of SOM δ13C values in the Maya Lowlands. Jacob (1995) working at Nakbe, Guatemala and Dunning et al. (2002) at La Milpa, Belize both commented on values that ranged from -18 to -23‰ in the buried soils of these areas. They suggested that the values indicated a mixed vegetation of woody C3 and herbaceous C4 plants such as tropical grasses since the soils they investigated were located in bajo environments that were probably wetter in ancient times. In a separate paper, Hansen et al. (2002) also proposed that for the same soils at Nakbe, the enriched values were the result of environmental change or possibly maize.

Stable carbon isotope analysis of soil organic matter was first used as a tool to identify remnant traces of maize by Webb et al. (2004) in their study of terrace soils at Caracol, Belize. Carbon isotope values in the soil organic matter increased up to 4.0‰ from the surface to an 80 cm depth. In this same profile, the phytolith record showed a transition at the 40 cm depth from panicoid grassland (the maize family of grasses) to overlying forest vegetation (Pearsall 1992). Webb et al. (2004) also provided an explanation for the pattern of SOM δ13C changes they noted with depth. In the time during and after 13C-enriched SOM was established due to maize cultivation, bioturbation and percolation caused a downward movement of the C4 signature. Apparently, this was the reason that the most enriched SOM was detected right above the limestone bedrock, no more than 80 cm depth. Later, additions of 13C-depleted SOM due to the return of C3 vegetation and perhaps sediment accumulation from upper slopes also moved downward by the same processes and diluted the original C4 signature. This explained the modest increases of SOM δ13C values with depth. Webb et al. (2004) concluded that analyses of SOM δ13C revealed changes between C3 vegetation and maize for the terraces at Caracol, especially in the humin fraction of the soil. Further, they concluded that the method could be used for interpretations of the size of Maya settlements as well as the extent of cultivation.

The method proposed by Webb et al. (2004) was subsequently applied to soils at the archaeological sites of Piedras Negras (Fernandez 2002) and Aguateca (Johnson et al. n.d.), both in Guatemala. SOM δ13C values increased from the surface soils to subsurface horizons by more than 5.0‰ at both sites. These differences were too great to be attributed to microbial fractionation or carbon mixing and therefore reflected the ancient growth of C4 plants (Martinelli et al. 1996). This “C4 Signature” is likely the result of long-term maize cultivation and compliments the Webb et al. (2004) study. The Webb et al. (2004) method is further validated by control profiles at both Caracol and Aguateca where SOM was 13C-depleted.

There is substantial evidence that maize is responsible for the high SOM δ13C values that were observed in the Maya Lowlands. As previously discussed, C3 vegetation dominated this area for many thousands of years before Maya occupation and about 1,200 years after the Maya collapse. During Maya occupation lands were cleared for crops. This may have allowed the invasion of C4 tropical grasses into the open fields. However, these tropical grass species were probably not sufficiently established (both spatially and temporally) to contribute significantly to

35

the C4 signature. It is noteworthy that the SOM δ13C values of the current bajo areas reflect C3 vegetation (Dunning et al. 2002; Jacob 1995). Also, two unidentified wetland plants were collected during the current study and both were found to be C3 by stable carbon isotope analysis. In addition, Powers and Schlesinger (2002) in Costa Rica (an area outside the Maya Lowlands), found that differences between surface and subsurface SOM δ13C values in these tropical soils only averaged 2.5‰ (maximum value of 4‰) which suggested continuous C3 vegetation.

Methods The prediction model was created with ESRI ARC/INFO software. Two different

models of ancient agriculture were created, one before the 2003 field season and a second, more refined model after the season. In the first model, a vector approach was used. Soils data were obtained from digitizing a soil series map of the Usumacinta River Basin at a scale of 1:200,000 (Aliphat 1996). The soils were then classified into two classes and scored: inadequate drainage (La Pasadita I and II series) and adequate drainage (all other series). A digital elevation model (DEM) of the study area was created by digitizing 20 meter contours from 1:50,000 topographic maps (Instituto Geográfico Militar 1987). This DEM was used to derive a raster slope map, which was then classified in to three classes: low (0 to 3%), moderate (3 to 9%), high (over 9%). The three class raster slope map was converted to a vector coverage, which was combined with the soil data to determine agricultural potential. Three levels of agricultural potential were created:

• poor to fair (poor drainage and slope > 9%) • good (adequate drainage and slope between 3% and 9%) • very good (adequate drainage and slope < 3%).

Figure 1 shows the refined agricultural potential map used for the “breadbasket prospection”. This map was used to plan an opportunistic sampling strategy for the study area. Despite its generality, the map was an invaluable tool due to remote and inhospitable nature of the area coupled with the lack of previous archaeological exploration.

The same 25 soil profiles used in the Johnson et al. (this volume) pedology study were used for isotope analysis. Of the 25 profiles, 23 were in the Macabilero area and two were sampled near the archaeological site of Porvenir. The same sampling methods apply as outlined in Johnson et al. (this volume), except that each 15 cm increment soil sample was kept separate for isotope analysis. In preparation for isotope analysis, each sample was air-dried, ground and passed through a 250 micrometer sieve (60 mesh). Carbonates were removed from soil with 1M

HCl in a 70 ºC water bath until effervescence ceased to remove both Ca and Mg carbonates. Next, the soils were washed, centrifuged, dried and ground again. The samples were then analyzed on a Finnigan MAT isotope-ratio mass spectrometer coupled with a Costech elemental analyzer. Each sample was analyzed twice and the average of the two readings was used for interpretation. The average standard deviation for the δ13C values was ±0.3‰ for the laboratory standards and ±0.3‰ for the repeated samples.

A “C4 strength” was assigned to each profile. This represented the difference between the surface SOM δ13C value and the most 13C-enriched subsurface sample (weak: <4.0‰, moderate: 4.0-6.0‰, strong: >6.0‰) (Table 1). The purpose of this was 1) to simplify the comparisons of 13C-enrichment between all profiles and 2) to overlay the strengths on the agricultural potential model for graphical representation of the extent and likelihood of maize agriculture in the study area. The percent composition of SOM derived from C4 plants was calculated using an equation similar to one presented in Webb et al. (2004) (Table 1).

36

Results and Discussion Many of the graphs of our SOM δ13C values with depth for each profile are similar in

pattern to the profiles described by Webb et. al. (2004). Also, many of the other patterns expressed in our soils are probably best explained by maize rhizo-deposition at depth, the accumulation of 10-15 cm of soil in the past 1200 years since Maya occupation, and the movement of SOM by bioturbation and water percolation processes mentioned in Webb et al. (2004). However, the Webb et al. (2004) profiles represent only one set of soil dynamic conditions whereas our profiles represent many different microsites within the study area. These environments include well drained upland soils, lacustrine and river floodplains, open canopy marshes and closed canopy bajos. These environments represent soil dynamic conditions which influence the movement of SOM in the soil and therefore the pattern of δ13C values with depth that include bioturbation and water percolation, water saturation due to poor drainage and soil deposition.

Of all the samples that were collected near Macabilero, 65% of them had a difference in SOM δ13C values from the surface to subsurface of 4.0‰ or greater and 30% of them had difference of 6.0‰ or greater (Table I; Fig.4). For discussion purposes, the profiles have been grouped according to their geographic location and landforms (Fig. 2,3). This seemed the most convenient way to discuss the profiles due to the somewhat unsystematic sampling strategy of the study. Their taxonomic description and other detailed characteristics are found in Johnson et al. (this volume).

“Near Usumacinta” (Fig. 2a). These profiles were located directly adjacent to the Usumacinta River, except for ARO2, which was located away from the river about 500m. One characteristic common to these profiles was shallow A horizons. PAPAYO3 and ARO3 had sandy C horizons and were classified as Oxyaquic Udorthents. The PAPAYO2 and ARO2 profiles were no more than 45 cm deep over bedrock classified as Lithic Haprendolls. AGUACATE1 had experienced more pedogenesis as it was located slightly farther away from the river (about 30 m) and was classified as Aquertic Argiudolls. Only the AGUACATE1 and ARO3 profiles showed moderate evidence of past C4 growth with SOM δ13C values above our cutoff value of 4.0‰ (4.2 and 4.7‰ respectively). In all the profiles, 13C enrichments were notable with depth and converged to about -24.2‰ at 30 to 45 cm depths compared to the average surface value of -27.6‰. This enrichment could have resulted from maize cultivation; however, the modest values cannot lead us to attribute them to only maize. These results support the interpretation that the Maya chose not to farm as heavily near the Usumacinta River and instead preferred more stable, inland soils (Johnson n.d. this volume).

“Near Water” (Fig. 2b,c). These profiles were located close to a natural source of water, either a seasonal wetland or the Laguneta Texcoco. The location of the CIVAL profile deserves some detailed discussion. Our purpose for sampling at this location was to test an area that was indicated by our agricultural potential map as “very poor”. It was located about 2 km away from the headwaters of the Rio Macabilero on the northern edge of the surrounding wetland. The sharply defined wetland-forest boundary was located only about 20 meters away. The wetland surrounded a branch of the Rio Macabilero where water movement seemed to be dominated by low energy vertical drainage (or saturation) instead of high energy lateral flow. The thin surface horizon (10 cm) had a black color (10YR 2/1) and was high in organic carbon (8.6%). Below this was a thicker (60 cm) lighter colored horizon (10YR 4/1) with less organic carbon (2.86%) which indicated illuviation from the surface horizon. From about 60–150 cm, the effects of seasonal inundation and break down of limestone parent material caused a soft, marl-like

37

material to develop. The leaves of two plant varieties that dominated the wetland were sampled close to the profile. The δ13C values were -28.1 and -27.5‰. SOM δ13C values of the profile itself strongly reflected the C3 vegetation and likelihood that these soils were avoided for maize agriculture. The values never exceeded -25.1‰ until 120 cm depth where they increase to -21.3‰. We speculate that in this soil accretionary environment that this enrichment is due to the establishment of some type of C4 plant, including maize, at an earlier time.

The TEXCOCO profile was similar to the CIVAL profile because it was also located in a lacustrine environment – about 50 meters from the shore of Laguneta Texcoco. Minimal ancient settlement was observed in the upland areas surrounding the lake. Only abandoned contemporary settlement was found along the shores of the lake which was the result of guerrilla refugee camps during Guatemala’s civil war. Again, this area was suspected to not have much potential for ancient agriculture. Indeed, as in the CIVAL profile, SOM δ13C values suggested that little or no maize was grown there in ancient times and only modest 13C enrichment with depth can be observed.

The CAMP profile was located at the 2003 field season base camp about 50 m from the Rio Macabilero and 2 km east of the Rio Usumacinta. This was the location of the abandoned refugee camp known as “La Esmeralda”. The average SOM δ13C values for the thirteen soil samples collected from this profile was only -24.1‰ for nearly two meters of depth which reflected little or no maize agriculture in the area (i.e. weak C4 strength). In contrast to the CIVAL and TEXCOCO profiles, this was an unexpected result as the landscape seemed very favorable to agriculture and soil properties were also desirable. No ancient settlement was observed near the profile. We can only surmise that the proximity to the river, which floods annually, would limit this area agriculturally.

The CINE profile was located about 20 m from a small spring (not shown on topographical maps). The nearest ancient settlement was located about 0.5 km away. The water table was about 2 m below the surface and soil redoxomorphic feature indicate that the soils were frequently inundated in this area. Besides the surface value, the profile closely mimicked the SOM δ13C values of CAMP until a depth of about 1 m, where it departed and increased substantially to about 2 m. The moderate C4 strength was puzzling since there were no signs of a buried A horizon. Perhaps, soil from adjacent ancient maize fields eroded into this area.

“Bajo” (Fig. 2d). The term “bajo” is used loosely in the field to refer to any area where the vegetation changed and understory vegetation becomes thicker. In contrast to the upland areas, these areas are low-lying and accumulate soils. However, in several places we failed to find soils deeper than 30 cm in order to take samples. At the beginning of the rainy season, localized depressions were filled with water after downpours which indicated the poor infiltration compared to surrounding areas.

The ESMERALDA PLANADA and ARO1 profiles both showed somewhat similar SOM δ13C values with depth. They were located just outside of the center of Esmeralda within about 1 km of each other. Their strong C4 strengths were shown by SOM δ13C values that increased gradually with depth until -21‰ at 90 cm and had maximum differences of surface to subsurface SOM δ13C values of 7.4‰. Maximum 13C enrichment occurred at greater depths in these profiles compared to the other enriched profiles (e.g. GROUP44 and TEP2), perhaps because of greater rates of soil accumulation. This may have caused the data to appear more “spread out” than the shallower 13C enriched counterparts, even though they represent the same time period. The data suggests that heavy maize agriculture took place at these locations for long periods of time. Indeed, the calculated C4 composition of the SOM for the profile was 66%. contributed

38

Below 90 cm, the data show a divergence of SOM δ13C values for the two profiles. The maximum value of the ARO1 profile was -21.4‰ at 105 cm while the ESMERALDA PLANADA profile had a maximum value of -19.1‰ at a depth of 175 cm. ARO1 is interesting because the values actually decrease again until -22.7‰ at bedrock (150 cm). Possible explanations for this are that because ARO1 was not farmed intensively for as long due to its location at the periphery of the site and/or the processes responsible for the downward movement of the C4 signature was less pronounced at this location.

The local relief of the PAPAYO1 profile was different than the other bajo profiles because the adjacent slopes were much steeper. Thus, it was evident that much more soil had accumulated due to erosion. Although the area was capable of supporting agriculture on a small scale, it looked too confined to be considered for any significant large scale agriculture. Indeed, stable carbon isotope analyses of SOM revealed a small peak at a depth of -60 cm and a value of -25‰. The enrichment at this depth was greater than 4.0‰ which gave it a moderate C4 strength. The SOM δ13C gradually returned to more negative values below this depth to -28‰ at -180 cm.

“Upland” (Fig. 3a,b). These profiles were grouped together because they represent an area of shallow soils (<40 cm) and have much better drainage. In contrast to the soil accumulating areas of the “Bajo” and “Near Water” groups, the SOM δ13C values of the remaining “Upland” soils are more difficult to interpret because of their shallow depth. However, they still have moderate C4 strengths. GROUP44 and TEP2 were both located next to ancient housemounds and had very similar SOM δ13C values. Surface values were about -27.5‰ and at a 30 cm depth they were -22‰ for a difference of 5.5‰. This is a bit surprising considering their location which is not considered to be very conducive to large scale maize agriculture. A possible explanation for this is that the plaza floors and surrounding areas accumulated concentrated amounts of maize debris such as cobs and shucks. Further, their patterns closely resemble those observed by Webb et al. (2004), where bioturbation caused a downward movement of the signature and subsequent C3 additions diluted the signature.

The FAJARDO2 and FAJARDO CAMP profiles were located within 20 m of each other away from the core of the Fajardo site and the CUEVA profile was located closer to the site of Texcoco. All three had very similar δ13C signatures except that CUEVA’s values were generally a little more negative throughout the profile. The three profiles only show modest C4 strengths (average difference was 3.8‰). These three profiles therefore represent areas that reflect 13C enrichment, possibly from C4 plants, but for some reason did not have C4 strengths as high as the other upland soil profiles. Additionally, samples from the two upland profiles ESMERALDA MILPA and RANDOM2 (data not shown) all had SOM δ13C values greater than -25‰. The profile’s name, “ESMERALDA MILPA”, was given to note that the area was briefly cultivated by refugees of the Esmeralda camp and does not imply that the area was actually used anciently for maize cultivation.

“Planada” (Fig. 3c). The “Planada” group represented areas where soil characteristics were the most favorable to agriculture, i.e. gently sloped, well drained and deep. Therefore, we expected to see the strongest signs of C4 plants in these profiles. The pattern of PAPAYO1 SOM δ13C values with depth was similar to other soil profiles which indicated that C3 vegetation preceded a long stand of C4 plants and then C3 vegetation was reestablished. This pattern was observed more prominently in the RANDOM and TEP3 profiles from the “Planada” group. SOM δ13C values peak at about -21‰ and around 45 cm. The other two profiles, TEP1 and

39

LAPISTA also showed enrichments with depth, but no apparent peak. TEP1 maintained an average SOM δ13C value of -19‰ between 75 and 180 cm.

LAPISTA’s values were the highest of all the profiles reported and indicated no previous C3 vegetation for the area. At 105 cm the SOM δ13C value was -15.6‰ which was a difference from the surface value of -12.4‰. A suspected buried A horizon observed at 75 cm was also intriguing because buried A horizons in the Guatemalan Lowlands are reported to be the ancient surface (Fernandez n.d.; Beach et al. 2003). This extremely high SOM δ13C value is surprisingly close to that of the pure maize plant (about -12‰). One source of error could have been that the mineral carbon from the calcium carbonate parent material was not all removed by acidification. To check this, the samples were acidified a second time, but no differences in the SOM δ13C were observed. The profile was located in the outlying areas of Texcoco (about 1.5 km from the center) and housemounds were located about 50 m away.

The data for LAPISTA peaked at 40 cm but for RANDOM there is no apparent peak, even though they were located within 1 km of each other. Therefore, two patterns have thus been identified among the “Planada” group, which surround the ancient site of Texcoco. One shows long term ancient agriculture closer to the Texcoco site and the other only one agricultural event further from the site. We can only speculate the reasons for this which may involve factors dealing with population growth or the political situation. For example, we could be seeing the extent of ancient agriculture as it radiated from Texcoco. The differences of values at the 1 m depth represent earlier times where significant maize agriculture only took place near Texcoco. The higher values that are closer to the surface of all four “Planada” profiles reflect a later event, when population increased and therefore the extent of maize agriculture also increased. Ceramics analysis from the nearby site of Piedras Negras suggests that there was a peak in population around 200 B.C. followed by a hiatus and then another peak around 800 A.D. followed by a collapse (Houston et. al. 2000).

“Porvenir” (Fig. 3d). Although not part of the immediate study area, two profiles were collected near the site of Porvenir. This ancient site is located about 12 km to the northwest of Esmeralda and 4 km northwest of Piedras Negras. The area was sampled because it scored fairly high for ancient agriculture potential in our breadbasket prospection map. The soils were somewhat poorly drained yet were favorable due to relatively little variation in surface topography. The PORVENIR profile was located at a park station near the banks of the Usumacinta River. Modern buildings at the station led us to believe that the surface was probably artificially leveled in contemporary times. This would have removed the surface horizon and may be the reason why the surface SOM δ13C value was about -24‰ instead of the more typical value of -27‰. The PORVENIR2 profile was located near the Porvenir site in the clayey bajo soils and had a strong C4 strength (6.3‰ difference). The highest SOM δ13C values for both profiles were about -21‰ at 90 cm. Therefore, the cultivation of maize may have been fairly large scale and continuous in ancient times in these locations. These results also support the utility of our prospection map.

After the 2003 season, it was decided that a raster approach would allow for a more refined model to predict areas of ancient agriculture (Fig.4). This approach also produced results that were more easily compared to the isotope study (Fig.5). The raster slope values were reversed so that the higher slopes had the lowest values and the lower slopes had the highest values. The digitized soil series map was converted to raster. Different soil series reflected different levels of drainage and therefore levels of desirability. Each soil series was assigned a score where La Pasadita = 1, Yaxchilan = 3, Menche = 4, Lacandon = 4, and Cheqique = 5. The

40

two layers were then combined with the raster calculator tool where each pixel that resulted was the final score for agricultural potential, which ranged from about 4 to 22. Higher scores reflected areas of lower slopes and better soil conditions and therefore better agricultural potential. The scores where then classified into five categories by quantile distribution method. Next, results from isotope analyses could be overlaid to graphically show whether or not ancient maize agriculture had been present in our supposed breadbasket (Fig.5). The combination of the isotope analyses and GIS results allowed for a better understanding of the location, relative likelihood and extent of ancient agriculture. Although the map produced in Figure 5 is most useful on a regional scale, we see that on this local scale the strong C4 strength profiles (bigger circles) generally lie in areas that are more brightly colored. In contrast, the weak C4 strength profiles lie in areas that are less conducive to agriculture according to our map.

Conclusions GIS and stable carbon isotope analysis are tools that can be used to confirm, investigate and interpret ancient human activities. In this study GIS was used to “prospect” for large areas of potential ancient agriculture on a regional scale and in an understudied area of the Maya Lowlands. A map was produced from an “Agricultural Potential Score” based on slope and soil type. The map was used as a guide to develop a sampling strategy for soils during the 2003 field season of the Sierra del Lacandon Regional Archaeological Project. The map also identified a possible breadbasket between the major warring polities of Piedras Negras and Yaxchilan. After the soils were collected and studied, three tools (pedology, stable carbon isotope analysis and GIS) were used to confirm that the soils were indeed conducive to ancient agriculture and could have supported large populations. Stable carbon isotope analyses of soil organic matter from the sampled soil profiles confirmed the presence or absence of ancient maize agriculture. These results are presented under the assumption that our results were not confounded by other C4 plants. Some profiles showed no signs of ancient maize while 65% of the profiles showed moderate or strong signs of ancient maize. Also, profile isotope data suggested that ancient maize agriculture occurred continuously or, in some cases, as distinguishable events.

Acknowledgements Funds for this research were provided in part by Brigham Young University in collaboration and the Sierra del Lacandón Archaeological Project, funded by the Foundation for the Advancement of Mesoamerican Studies, Inc. (FAMSI). Additional support was provided to the Parque Nacional Sierra del Lacandón by the World Monuments Fund. Project participants included: Americo Ixcayau, Chico León, Pánfilo Regino Hernandez, Eduardo Martínez, Ismael Mijangos, José Luis Aldana Alvarado, Elder Urbelino Alvarado Rodríguez, Leonardo de Jesús Damián Méndez, Huber Edilberto Heredia Rodríguez, Graciano Saquij Barahona, Eduardo Isidoro Saquij Choc. Special thanks to Dr. Henry Schwarcz and Dr. Elizabeth Webb for their advice and development of the application of stable carbon isotope analysis to ancient agriculture. We thank the Defensores de la Naturaleza, particularly Marie-Claire Paíz and Rosa Maria Chan, whose collaboration has made this work possible. The authors would also like to thank the many hours dedicated by BYU undergraduate mentorees who helped with the soil analyses including: David Wright, Sarah Kendall, Susanna Peña and Marcos Alvarez. Statistical analyses were aided by The Center for Collaborative Research and Statistical Consulting at BYU. The project took place within a regional concession graciously granted by the Instituto de Antropología e Historia (IDAEH). The Consejo Nacional de Areas Protegidas (CONAP) granted permission for this work.

41

References Cited Aliphat, M. M. 1996 Ecology and settlement in the Usumacinta River Basin. University of Calgary.

Balesdent, J., Girardin, C., and Mariotti, A. 1993 Site-related d13C of tree leaves and soil organic matter in a temperate forest. Ecology 74(6):1713-1721.

Beach, T. 1998 Soil catenas, tropical deforestation, and ancient and contemporary soil erosion in the Peten, Guatemala. Physical Geography 19:378-405.

Beach, T. and Dunning, N. P. 1995 Ancient Maya terracing and modern conservation in the Peten rainforest in Guatemala. Journal of Soil and Water Conservation 50:138-145.

Bird, M. I. and Pousai, P. 1997 Variations of d13C in the surface soil organic carbon pool. Global Biogeochemical Cycles 11(3):313-322.

Boutton, T. W. 1991 Stable Carbon Isotope Ratios of Natural Materials: I. Sample Preparation and Mass Spectrometric Analysis. edited by Coleman, D. C. and Fry, B., pp. 155-171. Harcourt Brace Jovanovich.

Boutton, T. W. 1996 Stable Carbon Isotope Ratios of Soil Organic Matter and Their Use as Indicators of Vegetation and Climate Change. In Mass Spectrometry of Soils, edited by Boutton, T. W. and Yamasaki, S.. Marcel Dekker, Inc., New York.

Boutton, T. W., Archer, S. R., Midwood, A. J., Zitzer, S. F., and Bol, R. 1998 d13C values of soil organic carbon and their use in documenting vegetation change in a subtropical savanna ecosystem. Geoderma 82:5-41.

Dunning, N. and Beach, T. 2000 Stability and Instability in Prehispanic Maya Landscapes. In Imperfect Balance: Landscape Transformations in the Precolumbian Americas, edited by Lentz, D. L., pp. 179-202. University Press, New York.

Dunning, N., Rue, D. J., Beach, T., Covich, A., and Traverse, A. 1998 Human-Environment Interactions in a tropical watershed: The paleoecology of laguna Tamarindito, el Petén, Guatemala. Journal of Field Archeology 25:139-151.

42

Dunning, N. P. and Beach, T. 2004 Noxious or nurturing nature? Maya civilization in environmental context. In Golden, Charles W.// Borgstede, Greg, pp. 125-141. Routledge, New York.

Dunning, N. P. and Beach, T. 1994 Soil Erosion, slope management, and ancient terracing in the Maya lowlands. Latin American Antiquity 5:51-69.

Dunning, N. P., Beach, T., and Rue, D. J. 1997 The paleoecology and ancient settlement of the Petexbatún region, Guatemala. Ancient Mesoamerica 8:255-266.

Dunning, N. P., Luzzadder-Beach, S., Beach, T., Jones, J. G., Scarborough, V., and Culbert, T. P. 2002 Arising from the Bajos: The Evolution of a neotropical landscape and the rise of the Maya Civilization. Annals of the Association of American Geographers 92(2):267-283.

Fedick, S. 1994 Ancient Maya agricultural terracing in the Upper Belize River area. Ancient Mesoamerica 5:107-127.

Fedick, S. L. 1996 An Interpretive Kaleidoscope: Alternative Perspectives on Ancient Agricultural Landscapes of the Maya Lowlands. In The Manage Mosaic: Ancient Maya Agriculture and Resource Use, edited by Fedick, S. L., pp. 107-131. University of Utah Press, Salt Lake City.

Fedick, S. L. 1995 Land evaluation and ancient Maya land use in the upper Belize River area, Belize, Central America. Latin American Antiquity 6:16-34.

Fernandez, F. G. 2002 Chemical and Physical Properties of Anthrosols in Rural Areas Near the Ancient Maya City of Piedras Negras, Guatemala. Brigham Young University.

Hansen, R. D. 2002 The architectureal development of an early Maya structure Nakbe, Peten, Guatemala. ed. FAMSI, FAMSI Website.

Hernandez, A. A., Guenter, S. P., and Zender, M. U. 2003 Sak Tz'i, a Classic Maya Center: A location model based on GIS and epigraphy. Latin American Antiquity 14(2):179-191.

Houston, S., Escobedo, H., Child, M., Golden, C., Terry, R., and Webster, D.

43

2000 In the land of the Turtle Lords, archaeological investigations at Piedras Negras, Guatemala, 2000. Mexicon 22:97-110.

Instituto Geografico Militar (IGM) and Defense Mapping Agency 1987 Laguneta Lacandon Guatemala, Scale 150,000, Sheet 1967 II. edited by Instituto Geografico Militar (IGM) and Defense Mapping Agency, E754. Defense Mapping Agency, Fairfax, VA.

Jacob, J. S. 1995 Archaeological pedology in the Maya lowlands. Pedological Perspectives in Archaeological Research 44:51-80.

Johnson , K. D., Wright, D. R., and Terry, R. E. n.d. Application of Carbon Isotope Analysis to Ancient Maize Agriculture in the Peten Region of Guatemala.

Leyden, B. W. 2002 Pollen evidence for climatic variability and cultural disturbance in the Maya Lowlands. Ancient Mesoamerica 13:85-101.

Leyden, B. W., Brenner, M., Hodell, D. A., and Curtis, J. H. 1993 In Late Pleistocene Climate in the Central American Lowlands, edited by Leyden, B. W., Brenner, M., Hodell, D. A., and Curtis, J. H., American Geophysical Union.

Martinelli, L. A., Pessenda, L. C. R., Espinoza, E., Camargo, P. B., Telles, E. C., Cerri, C. C., Victoria, R. L., Aravena, R., Richey, J., and Trumbore, S. 1996 Carbon-13 variation with depth in soils of Brazil and climate change during the Quaternary. Oecologia 106:376-381.

Nations, J. D. and Nigh, R. B. 1980 The evolutionary potential of Lancandon Maya sustained-yield tropical forest agriculture. Journal of Anthropological Research 36(1):1-30.

Parnell, J. J., Terry, R. E., and Nelson, Z. 2002 Soil chemical analysis applied as an interpretive tool for ancient human activities at Piedras Negras, Guatemala. Journal of Archaeological Science 29:379-404.

Pope, K. O., Pohl, M. E. D., Jones, J. G., Lentz, D. L., von Nagy, C., Vega, F. J., and Quitmyer, I. R. 2001 Origin and environmental setting of ancient agriculture in the lowlands of Mesoamerica. Science 292:1370-1373.

Rosenmeier, M. F., Hodell, D. A., Brenner, M., Curtis, J. H., and Guilderson, T. P.

44

2002 A 4000-year lacustrine record of environemental change in the southern Maya Lowlands, Peten Guatemala. Quaternary Research 57:183-190.

Sherer, A. K., Cassady, Y.J., Parks, M.S., and Wright, L.E. Paper presented at Classic Period diet at Piedras Negras, Guatemala: An isotopic perpective.

Taube, K. 2001 The Classic Maya Gods. In Maya: Divine Kings of the Rain Forest, edited by Grube, N., Eggebrecht, E., and Seidel, M., pp. 263-277. Konemann, Verlagsgesellschaft, Germany.

Turner, B. L. I. and Miksicek, C. H. 1984 Economic plant species associated with prehistoric agriculture in the Maya lowlands. Economic Botany 38:179-193.

Turner II, B. L. 1983 Once beneath the forest: Prehistoric terracing in the Rio Bec region of the Maya lowlands. ed. Westview, Boulder, CO.

Turner II, B. L., Klepeis, P., and Schneider, L. C. 2003 Chapter 20 Three millennia in the southern Yucatan Peninsula: Implications for occupancy use, and carrying capacity. In The Lowland Maya Area: Three Millenia at the Human-Wildland Interface, edited by Gomez-Pompa, A., Allen, M. F., Fedick, S. L., and Jimenez-Osornio, J. J., pp. 361-387. Food Products Press, New York.

Webb, E. A., Schwarcz, H. P., and Healy, P. F. 2004 Detection of ancient maize agriculture in the Maya lowlands using the stable carbon isotope compositions of soil organic matter: Evidence from Caracol, Belize. Submitted to Journal of Archaeological Science .

Webster, D. 2002 The Fall of the Ancient Maya. ed. Thames and Hudson Ltd., London.

Westcott, K. L. and Brandon, J. R. 2000 Practical Applications of GIS: A predictive modeling kit. Taylor and Francis Inc, Philadelphia.

Wheatley, D. and Gillings, M. 2002 Spatial Technology and Archaeology: The Archaeological Applications of GIS. Taylor and Francis Inc, New York.

White, C. D. and Schwarcz, H. P. 1989 Ancient Maya diet: As Inferred from isotopic and elemental analysis of human bone. Journal of Archaeological Science 16:451-474.

45

46

47

TABLES

Table 1. Profile C4 strengths and percent C4 carbon.

FIGURES

Figure 1. Ancient agriculture prospection map for the study area.

Figure 2. SOM δ13C values with depth for the Near Usumacinta, Near Water1,2 and Bajo soil profile groups.

Figure 3. SOM δ13C values with depth for Upland1,2, Planada, and Porvenir soil profile groups.

Figure 4. Refined raster map for ancient agricultural potential in the study area.

Figure 5. Agricultural potential map overlaid with C4 Strengths of soil profiles.

Table 1. Profile C4 strengths and percent C4 carbon.

Group

Profile Surf SOM δ13C Subsurf SOM δ13C Diff* C4 Strength** C4-Carbon*** ‰ ‰ highest value ‰ % Near Usumacinta

AGUACATE1 -28.4 -24.2 4.2 Moderate 28 ARO3 -28.6 -23.9 4.7 Moderate 32PAPAYO3 -26.4 -23.8 2.7 Weak 18PAPAYO2 -27.7 -24.2 3.5 Weak 23ARO2 -27.3 -25.2 2.1 Weak 14

Near Water1

SIVAL -26.8 -21.3 5.5 Moderate 37 TEXCOCO -28.1 -23.9 2.3 Weak 28

Near Water2

CINE -29.1 -20.0 5.5 Moderate 61 CAMP -25.3 -23.0 2.3 Weak 15

Porvenir

PORVENIR -24.4 -21.8 2.6 Weak 18PORVENIR2 -27.8 -21.5 6.3 Strong 42

Upland1

GROUP44 -27.5 -22.0 5.5 Moderate 37TEP2 -27.8 -22.2 5.6 Moderate

37

Upland2

CUEVA -27.0 -23.6 3.4 Weak 23FAJARDO2 -29.1 -25.6 3.5 Weak 23FAJARDO CAMP

-28.3 -21.5 4.5 Moderate 46

Bajo ARO1 -27.8 -20.4 7.4 Strong 49E. PLANADA -29.0 -19.1 7.4 Strong 66PAPAYO1 -29.6 -25.0 4.5 Moderate

30

Planada

LAPISTA -28.0 -16.6 6.4 Strong 76RANDOM -28.5 -20.9 7.5 Strong 50TEP3 -27.0 -19.9 7.1 Strong 47TEP1 -26.9 -18.1 8.7 Strong 58AVERAGES -27.7 -22.1 4.9 37

* Difference = Subsurface - Surface ** C4 Strength = Weak (<4.0‰ difference); Moderate (4.1-6.0‰ difference); Strong (>6.1‰) *** %C4-Carbon = (δSubsurface-δSurface)/(δC4-δC3) **** %C4-Carbon = (δSubsurface-δSurface Average)/(δC4-δC3)

Figure 1. Ancient agriculture propection map for the study area.

Near Usumacinta

-180

-160

-140

-120

-100

-80

-60

-40

-20

0-35 -30 -25 -20 -15

δ13C, ‰

Soil

Dep

th, c

m

AGUACATE1

ARO3

PAPAYO3

PAPAYO2

ARO2

-24.2

Near Water1

-180

-160

-140

-120

-100

-80

-60

-40

-20

0-35 -30 -25 -20 -15

δ13C, ‰

Soil

Dep

th, c

mTEXCOCOSIVAL

Near Water2

-180

-160

-140

-120

-100

-80

-60

-40

-20

0-35 -30 -25 -20 -15

δ13C, ‰

Soil

Dep

th, c

m

CINECAMP

Bajo

-180

-160

-140

-120

-100

-80

-60

-40

-20

0-35 -30 -25 -20 -15

δ13C, ‰

Soil

Dep

th, c

m

ARO1E. PLANADAPAPAYO1

Figure 2. SOM δ13C values with depth for the Near Usumacinta, Near Water1,2 and Bajo soil profile groups.

Porvenir

-180

-160

-140

-120

-100

-80

-60

-40

-20

0-35 -30 -25 -20 -15

δ13C, ‰

Soil

Dep

th, c

m

PORVENIRPORVENIR2

Upland1

-120

-100

-80

-60

-40

-20

0-35 -30 -25 -20 -15

δ13C, ‰

Soil

Dep

th, c

m

-180

-160

-140

GROUP44

TEP2

Upland2

-120

-100

-80

-60

-40

-20

0-35 -30 -25 -20 -15

δ13C, ‰

Soil

Dep

th, c

m

-180

-160

-140 CUEVAFAJARDO2FAJARDO CAMP

Planada

-180

-160

-140

-120

-100

-80

-60

-40

-20

0-35 -30 -25 -20 -15

δ13C, ‰

Soil

Dep

th, c

m

LAPISTARANDOMTEP3TEP1

Figure 3. SOM δ13C values with depth for the Upland1,2, Planada and Porvenir soil profile groups.

51

Figure 4. Refined raster map for ancient agricultural potential in the study area.

Figure 5. Agricultural potential map overlaid with C4 Strengths of soil profiles.

APPENDIX

KEY: Group 1: Haprendolls Group 2: Argiudolls

The NPAR1WAY Procedure Wilcoxon Scores (Rank Sums) for Variable Drainage_Code Classified by Variable group Sum of Expected Std Dev Mean group N Scores Under H0 Under H0 Score ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ 1 10 140.0 105.0 12.711371 14.0 2 10 70.0 105.0 12.711371 7.0 Average scores were used for ties. Wilcoxon Two-Sample Test Statistic (S) 140.0000 Normal Approximation Z 2.7141 One-Sided Pr > Z 0.0033 Two-Sided Pr > |Z| 0.0066 t Approximation One-Sided Pr > Z 0.0069 Two-Sided Pr > |Z| 0.0138 Exact Test One-Sided Pr >= S 0.0020 Two-Sided Pr >= |S - Mean| 0.0040 Z includes a continuity correction of 0.5. Kruskal-Wallis Test Chi-Square 7.5814 DF 1 Pr > Chi-Square 0.0059 The NPAR1WAY Procedure Wilcoxon Scores (Rank Sums) for Variable P__ppm_ Classified by Variable group Sum of Expected Std Dev Mean group N Scores Under H0 Under H0 Score ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ 1 9 92.0 85.50 11.324752 10.222222 2 9 79.0 85.50 11.324752 8.777778 Wilcoxon Two-Sample Test Statistic (S) 92.0000 Normal Approximation Z 0.5298 One-Sided Pr > Z 0.2981 Two-Sided Pr > |Z| 0.5962 t Approximation One-Sided Pr > Z 0.3015 Two-Sided Pr > |Z| 0.6031 Exact Test One-Sided Pr >= S 0.3024 Two-Sided Pr >= |S - Mean| 0.6048 Z includes a continuity correction of 0.5. Kruskal-Wallis Test

Chi-Square 0.3294 DF 1 Pr > Chi-Square 0.5660 The NPAR1WAY Procedure Wilcoxon Scores (Rank Sums) for Variable Total_O_N_ Classified by Variable group Sum of Expected Std Dev Mean group N Scores Under H0 Under H0 Score ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ 1 10 120.0 105.0 13.228757 12.0 2 10 90.0 105.0 13.228757 9.0 Wilcoxon Two-Sample Test Statistic (S) 120.0000 Normal Approximation Z 1.0961 One-Sided Pr > Z 0.1365 Two-Sided Pr > |Z| 0.2730 t Approximation One-Sided Pr > Z 0.1434 Two-Sided Pr > |Z| 0.2867 Exact Test One-Sided Pr >= S 0.1399 Two-Sided Pr >= |S - Mean| 0.2799 Z includes a continuity correction of 0.5. Kruskal-Wallis Test Chi-Square 1.2857 DF 1 Pr > Chi-Square 0.2568 The NPAR1WAY Procedure Wilcoxon Scores (Rank Sums) for Variable Total_O_C_ Classified by Variable group Sum of Expected Std Dev Mean group N Scores Under H0 Under H0 Score ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ 1 10 120.0 105.0 13.228757 12.0 2 10 90.0 105.0 13.228757 9.0 Wilcoxon Two-Sample Test Statistic (S) 120.0000 Normal Approximation Z 1.0961 One-Sided Pr > Z 0.1365 Two-Sided Pr > |Z| 0.2730 t Approximation One-Sided Pr > Z 0.1434 Two-Sided Pr > |Z| 0.2867 Exact Test One-Sided Pr >= S 0.1399 Two-Sided Pr >= |S - Mean| 0.2799 Z includes a continuity correction of 0.5. Kruskal-Wallis Test Chi-Square 1.2857 DF 1 Pr > Chi-Square 0.2568 The NPAR1WAY Procedure

Wilcoxon Scores (Rank Sums) for Variable Depth Classified by Variable group Sum of Expected Std Dev Mean group N Scores Under H0 Under H0 Score ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ 1 10 80.50 105.0 12.942179 8.050 2 10 129.50 105.0 12.942179 12.950 Average scores were used for ties. Wilcoxon Two-Sample Test Statistic (S) 80.5000 Normal Approximation Z -1.8544 One-Sided Pr < Z 0.0318 Two-Sided Pr > |Z| 0.0637 t Approximation One-Sided Pr < Z 0.0396 Two-Sided Pr > |Z| 0.0793 Exact Test One-Sided Pr <= S 0.0299 Two-Sided Pr >= |S - Mean| 0.0598 Z includes a continuity correction of 0.5. Kruskal-Wallis Test Chi-Square 3.5836 DF 1 Pr > Chi-Square 0.0584 The NPAR1WAY Procedure Wilcoxon Scores (Rank Sums) for Variable Clay Classified by Variable group Sum of Expected Std Dev Mean group N Scores Under H0 Under H0 Score ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ 1 10 127.50 105.0 13.223782 12.750 2 10 82.50 105.0 13.223782 8.250 Average scores were used for ties. Wilcoxon Two-Sample Test Statistic (S) 127.5000 Normal Approximation Z 1.6637 One-Sided Pr > Z 0.0481 Two-Sided Pr > |Z| 0.0962 t Approximation One-Sided Pr > Z 0.0563 Two-Sided Pr > |Z| 0.1126 Exact Test One-Sided Pr >= S 0.0466 Two-Sided Pr >= |S - Mean| 0.0931 Z includes a continuity correction of 0.5. Kruskal-Wallis Test Chi-Square 2.8950 DF 1 Pr > Chi-Square 0.0889

KEY: Group 1: Argiaquolls Group 2: Udorthents Group 3: Argiudolls Group 4: Haprendolls

1 2 3 4

2

3

4

5

6

Drainage

Code

group

1 2 3 4

0

5

10

15

20

25

LE

group

1 2 3 4

5. 0

5. 5

6. 0

6. 5

7. 0

7. 5

8. 0

pH

group

1 2 3 4

0

5

10

15

20

25

30

35

P

(ppm)

group

1 2 3 4

0

0. 25

0. 50

0. 75

1. 00

1. 25

Total

O#N#

group

1 2 3 4

0

2. 5

5. 0

7. 5

10. 0

12. 5

Total

O#C#

group

1 2 3 4

0

20

40

60

80

100

Depth

group

1 2 3 4

20

40

60

80

100

Clay

group

Table. UTM coordinates of profiles and their series according to Aliphat (1994).

East UTM North UTM Profile Series 694056 1893394 CUEVA Cheqique I 696758 1891342 FAJARDO2 Cheqique I 694678 1894795 TEXCOCO Cheqique I 695419 1892892 TEP2 Cheqique I 695419 1892892 TEP3 Cheqique I 696712 1891357 FAJARDO CAMP Cheqique I 691733 1889448 CAMP Cheqique II 690711 1889051 PAPAYO1 Cheqique II 694122 1893238 TEP1 Cheqique II 685475 1898031 PNFAB Lacandon 690228 1887312 Aguacate 1 Menche 690016 889989 ARO2 Menche 689516 1889569 ARO3 Menche 689907 1888505 PAPAYO2 Menche 689892 1888623 PAPAYO3 Menche 690562 1890490 ARO 1 Yaxchilan 694503 1892974 EL CINE Yaxchilan 691206 1890719 ESMERALDA PLANADA Yaxchilan 697135 1890689 GROUP44 Yaxchilan 693949 1893081 LAPISTA Yaxchilan 683093 1900464 PORVENIR Yaxchilan 682755 1900947 PORVENIR2 Yaxchilan 693530 1892556 RANDOM Yaxchilan 692378 1891212 SIVAL Yaxchilan