Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
7 Validating Prehistoric and Current Social Phenomena upon the Landscape of the Peten, Guatemala Thomas L. Sever Remotely sensed data from airborne and satellite sensors have been used to identify archeological features, such as roadways, temples, cisterns, and agricul- tural areas. This information is critical in research on the Maya to help answer questions regarding subsistence, settlement patterns, population densities, soci- etal structure, communication, and transportation. These issues in turn may well relate to perhaps the most intriguing question of all: the reason for the Mayan collapse. From the perspective of space we can also view the effects of human activity upon the landscape. Satellite images of the Peten region of Guatemala allow us to see where, when, and how rapidly that landscape is changing. Satellite data can be used to provide quantifiable evidence for depletion rates and trends of deforestation, identify potential points of conflict, and create predictive models for the future. The data can also be combined with ground-truth information to help us better understand the social issues involved and develop an effective strategy for balancing the challenges of population increase, sustainability, and conservation. Because of unanticipated preliminary results, this project evolved from a small-scale study to a regional analysis. Originally, the project was to conduct an archeological and environmental analysis of a 1 x 1 degree study area along the Usumacinta river valley between Guatemala and Mexico. This work was to be performed in response to potential hydroelectrical projects and large-scale oil explorations that threatened the area. When the first Landsat Thematic Mapper (TM) imagery of the area was processed, it revealed the contrast between the tilled landscape of Mexico and the standing rainforest of Guatemala (Plate 7-1, 145
146 Mexico 100 VALIDATING SOCIAL PHENOMENA, PEIEN, GUATEMALA Guatemala am_ ,,~ ~_/ = Scale Belize FIGURE 7-1 Location of the Maya Biosphere Reserve and buffer zone in the Peten district of northern Guatemala. after page 150~. This image received national and international attention with its appearance in the October 1989 issue of National Geographic magazine. It was also taken before the Guatemala Congress by President Vinicio Cerezo, which resulted in the establishment and protection of the Maya Biosphere Reserve (MBR) in 1990. This image also led to an agreement between the National Aeronautics and Space Administration (NASA) and the Central American Commission on Environ- ment and Development (CCAD) to conduct joint research as part of the Mission to Planet Earth Program. As a result of these developments, the original study area was expanded to include both the Peten region of Northern Guatemala and the MBR, the protected area within the Peten's boundaries (see Figure 7-1~. Initially, the primary focus of this investigation was on the use of remote sensing technology for the identification of unrecorded archeological features. However, the utility of the remote sensing data for monitoring of current defores- tation activities was also apparent to conservationists, scientists, managers, and
THOMAS L. SEVER 147 politicians. Consequently, two avenues of research archeological inventory and deforestation have proceeded in parallel throughout the course of the project. These two areas of research are discussed separately here, although both are related to the social sciences. Using airborne and satellite remote sensing data, the project has identified prehistoric archeological features, such as roadways, temples, and agricultural areas, as well as documented a time-lapse sequence of deforestation from 1986 to the present (Seder et al., 19961. The clearing of tropical forests is one of the primary global environmental issues. Deforestation leads to three unintended impacts: loss of cultural diversity, loss of biodiversity, and loss of carbon storage capacity. Satellite imagery has proven to be one of the best techniques for monitoring forest clearing, shifting cultivation, and land-use conversion patterns (Seder, 1995; Sader et al., 1994~. While remotely sensed data provide quantifi- able information about forest and land-cover changes, they do not explain the anthropogenic causes for those changes. Hence ground-truth information is re- quired both to validate the satellite imagery and to supplement it with information from local harvesters, farmers, and ranchers regarding the criteria they employ for land-use conversion. Our research team consists of James Nations, an anthropologist and ecologist at Conservation International; Santiago Billy, a Guatemalan national and conser- vationist for Conservation International; Frank Miller, a forester at Mississippi State University; Daniel Lee, a geographic information system (GIS) expert at GeoTek; and the author, a remote sensing specialist and archeologist at NASA. This interdisciplinary team has been conducting field research in the Peten for 10 years to verify the results of our computer analysis. The primary focus of our research is environmental inventory and the detection of archeological features within the Peten. During the course of our work we have collaborated with many other researchers, particularly Steven Sader, University of Maine, in deforesta- tion studies and Patrick Culbert, University of Arizona, in archeological research. BACKGROUND The Peten, in northern Guatemala, covers 36,000 km2 a third of Guatemala's land mass. The wetlands and intact tropical forest of the Peten represent one of the world's richest areas of biological diversity. The ecosystem contains over 800 species of trees; 500 species of birds; and large populations of mammals, including monkeys, jaguars, and tapirs. The area also contains some of the most prehistorically significant Mayan archeological sites in Latin America. Since deforestation activities also include the destruction of archeological sites, conservation of these ruins is synonymous with forest protection. Only a few indigenous Mayan descendants still live in the Peten, although the population of inhabitants is increasing rapidly as a result of migration and settlement. Until 1970, nearly 90 percent of the Peten remained forested. Today over half of the
48 VALIDATING SOCIAL PHENOMENA, PRIER, GUATEMALA forest has been cut, and if deforestation continues at its present rate, only 2 percent of the Peten's forest will remain by the year 2010 (Cameo, 1996~. The MBR lies within the Peten and covers 1.6 million hectares (ha), nearly 40 percent of the Peten and 14 percent of Guatemala. It is the largest protected area in of the Maya Tropical Forest, which stretches from Eastern Mexico to Belize, and as such is critical to regional conservation efforts. Deforestation rates in the area are accelerating as a result of increasing human migration. Over the past 20 years, the human population in the Peten has increased from 20,000 to more than 300,000 (Stuart, 1991~. Today, the Peten has a human population density of 26 inhabitants per square mile (J. Nations, personal communication); this figure contrasts significantly with the ancient Mayan population densities of 2,600 persons per square mile in the center and 500-1,300 per square mile in the more rural areas (Rice, 1991). New roads and pipelines being constructed in the Peten by logging and petroleum companies serve as conduits for human migration (Plate 7-2, after page 150~. Peasant farmers follow these corridors to clear the forest and establish maize-based agricultural plots (milpas). Slash-and-burn cultivation remains the predominant agricultural system in the Peten today. Soil fertility declines after 2 or 3 years of cultivation, and the milpas are abandoned (Lundell, 1937) as the farmers move to new areas. In addition, a major expansion of the Q'eqchi' Maya population has occurred in the southern Peten. The Q'eqchi', numbering over 400,000, show remarkable abilities in adapting to new environments and learning new technologies (Cahuec and Richards, 1994~. Cattle pastures are becoming increasingly present in the Peten, although they are not yet dominant as in other regions of Central America. The ancient Maya far outgrew the carrying capacity of slash-and-burn agri- culture, which is only 200-250 persons per square mile in this region (Rice, 1991~. The environmental effects of the current deforestation and settlement in the area are not known at this time. What is known is that monocultivation and cattle ranching are replacing the traditional adaptive system of the past. The forests of the Peten were nearly destroyed 1000 years ago by the ancient Maya, who, after centuries of successful adaptation, had finally overused their resources. Current inhabitants are threatening to do the same thing today in a shorter time period with a lesser population. ARCHEOLOGICAL RESEARCH The Classic Mayan period of the southern lowlands lasted from 250 to 900 A.D., when it collapsed in unparalleled fashion. That some disaster had befallen the Classic Maya became clear in the early days of Mayan studies when it was learned that the dating and inscription of monuments had ceased during the ninth century. The seventh and eighth centuries were a golden age for the Maya. A population of millions achieved new peaks in construction. Stone monuments
THOMAS L. SEVER 149 proclaimed the glory of the rulers in decorative masterpieces. Between 830 and 930 A.D., however, populations of both elites and commoners in the Mayan lowlands declined by two-thirds. By the Postclassic Period (after 930 A.D.), only a few scattered houses remained (Culbert, 1993~. The Mayan collapse may well be the greatest demographic disaster in human history. It occurred in the region of the Mayan lowlands. There was no exodus outward because the peripheral populations remained stable. What caused this disaster is not known. Potential explanations include internal conflict and war- fare, disease, climatic change, overpopulation, outside invasion, peasant rebel- lion, soil nutrient depletion, and failure of the agricultural system (Culbert, 1974, 1993; Hammond, 1990~. Rather than having a single explanation, however, the cataclysm may have had multiple causes. What is known is that most of the trees in the region had been cut down by the ninth century. Although there is recent evidence for a drought beginning around 800 A.D. (Sabloff, 1995; Curtis et al., 1996), the question emerges of why the Maya had survived previous droughts, not to mention other setbacks. Understanding how the Maya managed the land- scape so successfully for centuries and what the eventual effect was, if any, of their deforestation activities may provide an important lesson for the future of today's inhabitants. TM imagery reveals Mayan causeways that often cannot be seen from ground level. As part of the theoretical framework of landscape archeology, roadways reflect the interplay among technology, environment, social structure, and the values of a culture (Trombold, l991~. In this case, the causeways are the tangible evidence of the Maya's structural organization across geographic space. The Mayan causeways are formal routes. Formal routes reflect elements of planning and purposeful construction, including labor in construction, engineering, main- tenance, and an organizational structure that oversees the system. They tend to be straight in nature and are engineered to overcome natural obstacles in order to improve transportation and communication. They contrast with informal routes, such as paths, which show little if any evidence of construction, are irregular in design, and tend to avoid natural obstacles. From a remote sensing perspective, the Mayan roadways tell us how the cities were connected and to some degree suggest the level of sophistication of the city responsible for their construction and maintenance. Several analytical techniques have been employed to identify Mayan cause- ways in the TM imagery. Laplacian and Gaussian filtering techniques were used to isolate linear and curvilinear features in the imagery. These techniques have been employed successfully to detect Anasazi roads in New Mexico (Sever, 1990) and prehistoric footpaths in Costa Rica (Sheets and Sever, 1988~. In addition, a number of different ratios and other transformations, such as the Normalized Difference Vegetation Index (NDVI), principal component analysis, and a modified version of the Kauth-Thomas (KT) transformation (MKT) were used to isolate the causeway features. MKT was derived as follows:
150 VALIDATING SOCIAL PHENOMENA, PETEN, GUATEMALA . The KT tasseled cap transformation was performed using standard coeffi cients. · Bands 1 and 2 of the KT transformation were ratioed using Band 1 - (Band 2/Band 1 + Band 2~. In this format, the central processing unit (CPU) order of operations is to perform the division first and then subtract, yielding a different value from that of the NDVI calculation. These values are assigned to the red video color gun. · Bands 4 and 5 were ratioed as 5/4 and assigned to the green video gun. · Bands 5 and 7 were ratioed as 5/7 and assigned to the blue video gun. The combination of these techniques made it possible to identify various roadway segments. For instance, as the roads cross the seasonally flooded swamps (bajos), they support a more lush vegetation than the surrounding plant community. Since the TM data are collected near the end of the dry season (April-May), the dryness of the roads in the elevated areas can be detected, as can the wetness in some of the lower areas. The most distinguishing characteristic, however, is the straightness or linearity of these features. The karst topography of the landscape also reveals linear geologic faults, fractures, and drainages that can be confused with causeways. Ground-truth reconnaissance can often resolve this issue visually. In some cases, a ground survey is not sufficient, and excava- tion must be used to distinguish the features conclusively. It should be mentioned that some of the techniques used to identify roadways also reveal temple struc- tures and man-made water storage areas (see Figure 7-2~. The 30-m resolution of the satellite imagery precludes detection of canals, but airborne 5-m Thermal Infrared Multispectral Scanner (TIMS) and Calibrated Airborne Multispectral Scanner (CAMS) data, as well as simultaneously acquired color infrared (CIR) photographs, have been used to identify potential canals and field systems. A1- though these features have not been verified to date, their existence is supported by the discovery of canal features in nearby study areas. The problem is that the 5-m data were not acquired in the specific bajo areas where the canals were discovered through excavation. As discussed below, the detection of these canals will provide insight into the use of the bajos by the Maya. The CIR photography was also used to confirm Mayan causeways (Figure 7-3~. One of the leading debates in Mayan archeology concerns whether the low- land wetlands (bajos) of the Peten were or could be used for agriculture (Pope and Dahlin, 1989, 1993; Adams et al., 1990; Turner, 1993~. The bajos represent 40 percent of the land surface of the Peten, and it seems reasonable to assume that these areas were farmed by the ancient Maya. As noted earlier, slash-and-burn technology, still used by the inhabitants today, can support only 200-250 persons per square mile in the region. This level had been reached by the Maya by 300 A.D., far before their population peak in the Late Classic period (600-830 A.D.), when population densities ranged from 500 to 2,600 persons per square mile (Rice, 1991~.
THOMAS L. SEVER 151 FIGURE 7-2 Thematic Mapper satellite image created by ratioing Band 4 with Band 3. The light spots represent ancient Mayan temple structures beneath the forest canopy at the archeological site of Mirador, while the dark lines represent Mayan causeways and natu- ral geological features. Ground reconnaissance and sometimes excavation are required to separate the cultural from the natural features. Currently, our research team has joined forces with Patrick Culbert of the University of Arizona to address this issue. Both supervised and unsupervised classification techniques (Kelley, 1983) have been used to classify satellite im- ages and identify the bajo areas (Plate 7-3, on facing page). Three different types of bajos have been identified, in addition to seven additional land-cover classes, such as high forest, transition forest, and water. Local informants in the Peten, however, have identified to Culbert seven different types of bajos, each with
152 ;^ Ct Ct = . Ct o o ~ .= .= Cal s°- ;^ Cal so o Cal ·_. ~ .~ . _ Cal o ~ ~ _. ~ o Or, _. .= o A, . ~ Cal Ct Cal ·0 Hi_ Ct sly sol .0 o o ~ o o Cal Cd sly Or, L Cal =. ~ ~ Ct sly 5 o o ~ ·_. o Cal ;^ Ct · _. o ;^ Cal ~ Ct o
THOMAS L. SEVER 153 distinctive characteristics. Of these seven types, two or three remain sufficiently moist in the dry season to sustain agriculture. This information is pertinent today, and may provide insight into early Mayan farming practices. Currently, we are applying supervised classification techniques using information from two transects as the training samples. The transect data come from field data col- lected by Culbert and his research team during their May 1995 field season. They conducted two transects in the Bajo la Justa, Peten one a trail of approximately 7 km, the other along a road of approximately 17 km. They were assisted in the vegetation survey by a native (Don Felipe Lanza) employed by the Forestry Division of the Tikal National Park. During this field reconnaissance they learned that there are two primary classifications of bajo associations: palm and scrub. Within each of these two major types are "sections characterized by a predominance of particular species which provide subtypes such as 'escobal', 'coronal', and 'botonal' within the palm bajo, and 'jimbal', 'tintal', 'navajuelal' and 'huechal' within the scrub bajo" (Culbert et al., 1995:3~. Additionally, there are three distinctions that relate to ground-surface characteristics and inundation: "bajo piano," which has a flat surface and no noticeable slope; "bajo borbolal," which has an undulating sur- face; and "bajo inundable," which is at an elevation where seasonal flooding occurs and seems to be characterized by scrub vegetation. Lanza also informed Culbert and his team that palm bajo is excellent for milpa and is one of the environments of choice among native Peteneros. Along both transects the team recorded vegetation according to Lanza's classification and gathered Global Po- sitioning System (GPS) readings. This ground-truth information has been overlaid on the TM satellite data, and various classifications have been developed. In addition, we will acquire standard-beam (30-m) Radarsat C-band Synthetic Aperture Radar (SAR) data, which will then be merged with the TM data. This merging of TM/SAR data should provide the best opportunity for distinguishing class types. The various classification schemes will be tested during the 1998 field season to determine which technique is the most accurate. In the process of searching for the best satellite band combinations, ratios, and other transformations for discriminating sacbes and bajos, it was noted that these techniques also discriminated elevated "islands" of upland within the bajos. The best technique for discriminating such islands employed the MKT using TM bands 3, 4, and 5. Culbert's 1995 field season led to key findings with regard to these islands: A very important fact about the location of Maya sites in relation to bajos has become clear...our work suggests that there was occupation on almost every area of slightly higher land where patches of high forests occur as "islands" within bajo. We call these areas of occupation "bajo communities" and they will be one of the foci of our future project in the Bajo la Justa (Culbert et al., 1995:4~.
154 VALIDATING SOCIAL PHENOMENA, PETEN, GUATEMALA The discovery of these occupied islands within the bajo system raises many research questions related to chronology, settlement patterns, and artifactual as- semblages. Do these islands represent a long-term occupation, or are they the result of a sudden, brief occupation? Are the bajo communities similar to or different from other communities, and do they operate under a similar or different organizational structure? The answers to these questions are critical in under- standing the full adaptation of the Maya to their rainforest environment, and mapping these islands is the first step toward finding these answers. Recently, our research team participated in the identification of an unre- corded site that may prove to be Site Q (for ''que?''L''which?''l). The potential existence of this Mayan site has been based on relief sculptures found in various collections and museums, as first identified by Peter Mathews in the late 1970s. The location of Site Q. signified by a snake-head emblem glyph, has been a source of debate among scholars, with some postulating that the site has not yet been found and others claiming that it is the known site Calakmul in Mexico (Schuster, 1997~. The preeminence of Site Q in the Classic Maya world is evidenced by the fact that the snake-head emblem glyph has been found in inscriptions at sites in Belize, Guatemala, Honduras, and Mexico. While Calakmul is a strong candidate for Site Q. the fact remains that there are a number of large Mayan sites in the area where little or no research has been conducted, not to mention potential sites that have not yet been found. While studying scarlet macaw routes, team member Santiago Billy came across an unrecorded site in April 1996. His GPS measurements of this location were compared with TM satellite imagery in which potential causeway linea- ments could be seen by the author. Our team conducted an initial reconnaissance mission in July 1996. Pyramids, walls, plazas, structures, and hieroglyphic monu- ments could be seen. The research team reported this information to Ian Graham and David Stuart at the Harvard Peabody Museum. In May 1997, our research team met Graham and Stuart at the site, which has been named "La Carona." As they mapped the site and recorded the hieroglyphs, we surveyed the site perimeter, finding many additional but smaller structures. To date, sufficient analysis has been conducted to determine that if this location is not Site Q itself, it is nearby (Graham, 1997~. Our remote sensing imagery suggests that there are other archeological sites in the area that we intend to visit in the course of our research. (The difficulties of conducting ground-truth recon- naissance in these areas are discussed later in this chapter.) DEFORESTATION Central America has one of the highest rates of deforestation, on a percent- age basis, in the world (Food and Agriculture Organization, 1991~. Satellite imagery provides the primary source of quantifiable data about forest and land- cover changes. The remote sensing and GIS techniques used for deforestation
THOMAS L. SEVER 155 and vegetation mapping are well established (Fearnside, 1986; Sader and Joyce, 1988; Sader et al., 1990; Skole and Tucker, 1993~. These analytical techniques have been applied to present evidence of the current deforestation threat to the Peten region. As noted earlier, the ancient Maya had substantially deforested the region by the time of their collapse in the ninth century, while the current tropical forest, no more than 1,000 years old, is being destroyed at an alarming rate in the wake of human migration and settlement. Schwartz (1990) documents the social and political driving forces that have affected the people and the forest of the region. Until recently, the Peten enjoyed a largely stable sociocultural system dating back to the 1720s. From colonial times it had been a sparsely populated region, with one writer commenting, "After all, the Peten was then (1895) and is now (1940) a region almost unoccupied, little developed, and of small promise" (Jones, 1940, quoted in Schwartz [1990:2411~. The region remained sparsely populated until the 1960s, when the national government opened the Peten to colonization. Unequal land distribution, decreasing access to land, extensive deforestation, in- creasing economic inequality, and political unrest subsequently led to revolution. By the late 1970s and early 1980s revolutionary guerrilla forces had estab- lished a wide base of insurgency support against the national government in both the Peten and the Guatemala highlands. The conflict had reached a peak by the mid-1980s, but the struggle continued until the signing of a peace treaty between the rebels and the government in January 1997. Between 1960 and 1986, the Peten experienced a population increase from 26,000 to over 300,000 an in- crease of 1,100 percent in 25 years (Schwartz, 1990:256-257~. As will be seen, our satellite imagery demonstrates that between 1986 and 1995, human migration and deforestation continued to accelerate. Tropical rainforest deforestation results from complex combinations of so- cial, economic, and biological causes. It affects the flora and fauna and the lifestyles of indigenous populations (Nations, 1988~. Currently, cattle ranching and slash-and-burn agriculture are the most serious threats to the Peten. Others include road building, oil exploration, selective logging of mahogany, the taking of endangered species of animals and birds, and the looting of archeological treasures. Since the late 1970s, marijuana has also become an important though illegal export crop (Schwartz, 1990:260~. These practices stand in contrast to economic activities that have a negligible environmental impact, such as tourism and the harvesting of chicle, xate, and pimienta. Four Landsat TM satellite scenes were purchased for April 1986, April 1990, May 1993, and March 1995. As with many tropical regions, it is difficult to acquire cloud-free data over the Peten, and these satellite scenes represent the best data available. Image processing was conducted using Earth Resources Laboratory Applications Software (ELAS) (Graham et al., 1985) and ERDAS (1990) software. Data processing and analysis were conducted jointly under the direction of Steven Sader, University of Maine, and the author. A channel 3, 4,
156 VALIDATING SOCIAL PHENOMENA, PETEN, GUATEMALA and 5 subset was extracted from the seven-channel TM data. Atmospheric haze was reduced by subtracting the minimum values located in deep water (Lago Peten Itza) from the visible red (channel 3) and reflective near infrared (channel 4~. The NDVI was computed for all four dates. The equation for NDVI is: NDVI = (near infrared - visible red)/(near infrared + visible red) NDVI information correlates with measures of vegetation "greenness" (Tucker, 1979; Sellers, 1985) and can easily distinguish roads, water, and sparse vegetation from high biomass forests (Seder et al., 1991~. Change detection was performed by subtracting the 1990, 1993, and 1995 NDVI from the 1986 NDVI. The data were compared with false color composite images. The composites were made by loading TM bands 4, 5, and 3 onto the red, green, and blue image planes of the computer screen (Jensen, 1986~. In this way a time-sequence of change from 1986 can be identified and quantified. The results are alarming for the MBR, an area generally thought of as being protected. Increased forest clearing can be identified throughout various parts of the MBR. Comparison of the satellite images for the above four dates indicates that the greatest deforestation of primary forest occurred during 1990-1993, although the Belize-Guatemala border area saw the greatest increase during 1993-1995. Interior roadless regions within the MBR show little or no regrowth or clearing activity. There is a strong relationship between roads and distance to forest, with over 90 percent of the clear cutting being conducted within 2 km of roads. What is certain is that the ratio of clear-cutting to mature forest has been increasing since 1986. Results indicate a general 12:1 ratio (12 ha cut to 1 ha regrown) during 1986-1990. During 1990-1993, some areas experienced even higher ra- tios. We are currently completing our analysis for 1993-1995, but the same pattern of increasing deforestation is apparent as the forested landscape with scattered agricultural fields yields to an agricultural landscape with a fragmented forest. Forest-clearing ratios appear to be highest in frontier areas where there are new roads and an influx of settlers. These areas include the Sierra del Lacandon and the Mexico-Guatemala border region in the west. Forest clearing is less severe near lower, more stable resident populations, as in Carmelita in the central Peten, where more traditional farming techniques are practiced. Forest clearing is also less severe in areas where there are guards on duty, such as Tikal. It is not simply the ratios of deforestation that are alarming, but also the pattern of deforestation. Forest-clearing activities have created "vegetation is- lands" of primary forest surrounded by cleared areas. These vegetation islands are becoming more pronounced in response to new settlers and the construction of new roads and an oil pipeline in the area (apparent in the 1995 TM image). The extent to which this increasing deforestation is affecting the ecology of the
THOMAS L. SEVER region, in particular the m time. 157 . igration routes of birds and animals, is unknown at this What is known is that the increasing human migration into the area is adding to the social stress as the competition for land and resources increases. Satellite change-detection images can be used to identify potential areas of conflict as the forces of expansion meet protected land. The peace treaty of 1997 provides for the resettlement within the MBR of refugees who fled the area during the con- flict. The resulting impact on the forest will be seen in future satellite images. Between March and May 1997, there were two serious incidents inside the MBR, in which government and nongovernmental organization (NGO) hostages were captured at gunpoint, and a biological station was burned to the ground by a force of 60 armed men. Eventually, the release of the hostages was negotiated, and an agreement was made with the perpetrators to rebuild the biological station in exchange for logging, farming, and hunting concessions within the reserve. In May 1997, Carlos Catalan, a Guatemalan spokesman for conservation, was mur- dered at gunpoint in the village of Carmelita. In June 1997, archeologist Peter Mathews and his five-man research team were captured and beaten along the Usumacinta River as they attempted to remove a Mayan altar to a local museum. Although Matthews had the necessary government and local permissions, the captives told him that "they don't respect any authority" (Associated Press, 1997~. These recent incidents illustrate the reality of the situation and suggest the likeli- hood that similar incidents will occur in the future unless an effective land-tenure and migration policy can be established and enforced. RESEARCH CHALLENGES: COLLECTING GROUND-TRUTH INFORMATION The gathering of ground-truth information, often referred to as "reference data," involves collecting measurements or observations about objects, areas, or phenomena that are being remotely sensed (Lillesand and Kiefer, 1994~. Social scientists can use ground-truth information in two ways: first, it can aid in the analysis, interpretation, and validation of the remotely sensed data; second, such information helps in understanding the socioeconomic forces behind land-cover modifications due to human activities. Ground truthing is expensive and time-consuming. While the price of com- puter hardware and software for remote sensing analysis has dropped dramati- cally in recent years, the costs associated with ground-truth activities have gener- ally increased. Airfare, lodging, vehicle rental, food, labor, and the like remain expensive elements of a research design, although it may be noted that recent advances in affordable, portable GPS receivers and digital field recorders give the field investigator greater flexibility in the sampling design. Our ground-truth activities require that we visit as many sites as possible in the remote and difficult terrain of the Peten in order to document as many study
158 VALIDATING SOCIAL PHENOMENA, PETEN, GUATEMALA sites as possible. Consequently, we tend to break camp each day and move on to the next location. This feature of our research contrasts with the approach of those who remain at a village or site for an extended period of time. While we map and verify the existence of archeological features, we never excavate. Some of the challenges we have encountered while conducting fieldwork in the Peten include logistical problems, communication problems, equipment failure, inad- equate maps, physical stress, suspicion, danger, and unstable political environ- ments. In fact, our research team was captured and held at gunpoint by leftist guerrillas for several hours before being released at nightfall. Certain phenomena (such as prehistoric Mayan roadways) that appear in the nonvisible bandwidths of the remote sensing data are sometimes simply not visible from ground level and require clearing and/or excavation before being verified. This situation is related to the fact that certain features that cannot be seen in visible light can nevertheless be detected in the infrared or microwave portions of the spectrum by remote sensing instrumentation (Sever, 1990~. Field reconnaissance also provides us with other information not visible in the imag- ery, such as the selective cutting of mahogany trees. Generally, our interviews with local farmers and ranchers provide accurate information. Sometimes, how- ever, the respondents provide us with either information they think we want to hear or false information intended to mask illegal activity. Logistics are probably the major constraint on our field work. Often we are the first professionals to visit an unrecorded archeological site. We must sched- ule in advance the jeeps, boats, aircraft, mules, horses, and workers that will get us to our destination. Since many areas of the Peten do not have telephone service, a member of our team who lives in Guatemala must travel weeks and months in advance to arrange these rentals with the local villagers. The more inaccessible the location, the more difficult the arrangements are. Generally, our field missions last 2 to 3 weeks. As we travel and switch from jeeps to boats to horses and mules, it is critical that the dates, times, and locations for these ar- rangements be finalized in advance. Occasionally, we are met with suspicion regarding the true purpose of our research. We have successfully combatted this situation by taking the time to explain our research goals and objectives to the local residents. We always take a large number of satellite images and, after explaining how we are using the imagery, leave a copy of it with the personas) involved. Through years of expo- sure and word of mouth, we have gained acceptance and support for our data acquisition activities, as well as the confidence of the inhabitants. As the years have passed, many of the inhabitants have become better educated about satellite imagery and GPS units, so that when we stop in a village and present the images, they can often help us interpret some of the features and anomalies displayed. Having a Guatemalan national on the research team is also a positive benefit to our research activities. GPS measurements are a critical component of our field research. Initially,
THOMAS L. SEVER 159 in 1988, there were only a few satellites in orbit, and we often found ourselves climbing a temple at midnight to collect a position that would be available only between 1:00 and 4:00 a.m. Also, intense vegetation cover of the tropical forest limited the collection of GPS readings. Today there is a complete constellation of satellites, and readings can be gathered nearly around the clock. New GPS units with better software allow for the collection of more data with greater accuracy. A problem we often encounter is the inaccuracy of available maps. We consistently find that lakes, rivers, archeological sites, and cultural features are not located where the map indicates. Apart from the obvious in-field confusion, there is the major concern that if these inaccurate maps are digitized and incorpo- rated into a GIS, they will lead to false results for predictive models. We resolve this dilemma by constantly comparing our GPS measurements, imagery, and maps to eliminate as much confusion as possible. It should also be noted that the names on a map are not necessarily the names used by the local inhabitants. As we studied deforestation trends in the Peten over the last several years, we designed our ground-truth activities primarily to identify the difference between new forest clearings and regrowth. Now we are expanding our ground-truth activities to include information on the decision processes associated with land use and land conversion. Currently, the various socioeconomic factors associ- ated with deforestation are poorly understood. In addition, much of the current uncertainty involved in modeling the terrestrial carbon budget arises from inad- equate data on tropical deforestation rates and trends in land-use conversion. To address these scientific issues, we will acquire additional information as we interview local farmers and ranchers. Through these interviews, we will deter- mine such factors as the crop-to-fallow ratios, the decision process for converting land to pasture or to shifting agriculture, the forest fragmentation indices and spatial characteristics of cleared land, the associated socioeconomic factors, and how the driving forces differ by zones or management units. The results will be correlated over the time scale of our database, providing better analytical infor- mation for management decisions. SUMMARY AND CONCLUSIONS The delicate ecosystem of the MBR in the Peten region of northern Guate- mala was managed successfully for centuries by the ancient Maya until their collapse in the ninth century A.D. The archeological evidence indicates, how- ever, that by the time of their collapse, the Maya had deforested most of the region. For the next 1,000 years the forest regenerated. Today, human migration and so-called "modern" subsistence techniques once again threaten the sustain- ability of the area. Through the use of remote sensing and GIS research, we are attempting to answer questions about the past in order to protect and manage the resources of the future. The protection of the tropical rainforest is also synonymous with the
160 VALIDATING SOCIAL PHENOMENA, PETEN, GUATEMALA protection of archeological sites, for the cutting and burning of an area destroy not only the trees, but also archeological features and materials. In addition, the construction of modern roads and pipelines provides a conduit for the illegal looting of endangered animal species and archeological treasures. Remote sensing data have successfully located archeological features in the Peten region that are difficult to discern from traditional survey and analysis techniques. These features include ancient roadways, temples, cisterns, and agri- cultural areas. The detection and subsequent analysis of these features may help answer questions regarding ancient Mayan subsistence, transportation, and water management and the factors that led to the eventual abandonment of the area at the time of the Mayan collapse. At the same time, however, verification of archeological features detected in the remote sensing data is expensive, time- consuming, difficult, and dangerous because of the dense vegetation, remoteness, and hazards associated with the Peten landscape. After the ninth century, the Peten remained a sparsely populated area until the Guatemalan government opened the area to colonization in the 1960s. Since that time the forest has continued to be cut down at an accelerating rate as human migration and settlement have introduced nontraditional agricultural techniques. The driving forces behind deforestation activities in the region result from a complex combination of social, economic, and political issues. Change-detec- tion analysis using remote sensing imagery has been used to document and quan- tify the extent and rate of change during 1986-1995. Data analysis reveals that the ratio of clear-cutting to mature forest has been increasing over that time. The highest forest-cutting ratios are associated with the construction of new roads and increases in migrating settlers. Forest-clearing ratios are lowest near resident populations who practice traditional farming techniques. Cattle ranching and slash-and-burn agriculture are the greatest threats to the Peten, although road building, oil exploration, selective logging of mahogany, and illicit crops also adversely impact the ecology of the region. Increasing human migration and settlement in the next few years will only add to the deforestation and social stress as the competition for resources in- creases. According to current estimates, only 2 percent of the Peten's forest will survive by the year 2010. The use of aerial and satellite imagery can help identify areas of possible conflict, inventory natural and cultural resources, identify ille- gal practices, and monitor protected areas. It is hoped that the results of this interdisciplinary research will provide information that will help managers, sci- entists, and politicians responsible for the MBR and the Peten region in general make more informed decisions and thereby avoid the collapse that occurred in this area a little over 1,000 years ago.
THOMAS L. SEVER REFERENCES Adams, R.E.W., T.P. Culbert, W.E. Brown, P.D. Harrison, and L.J Levi 1990 Rebuttal to Pope and Dahlin. Journal of Field Archaeology 17:241-243. Associated Press 1997 Archeologists barely survive ambush in Mexican jungle. CON Interactive World News, July 1. Cahuec, E., and J. Richards 1994 La variacion sociolinguistica del Maya Q'eqchi. Boletin Linguistico 40 Guatemala. Canteo, C. 1996 Destruccion de Biosfera Maya avanza ano con ano. Siglo Veintinno (Guatemala), Novem- ber21. Culbert, T.P. 1993 Maya Civilization. Washington, D.C.: St. Remy Press and Smithsonian Institution. 1974 The Lost Civilization: The Story of the Classic Maya. New York: Harper and Row. Culbert, T.P., L. Levi, B. McKee, and J. Kunen 1995 Investigaciones Arqueologicas en el Bajo La Justa, Entre Yaxha y Nakun. Unpublished field report, Department of Anthropology, University of Arizona. Curtis, J.H., D.A. Hodell, and M. Brenner 1996 Climate variability on the Yucatan Peninsula (Mexico) during the past 3500 years, and implications for Maya cultural evolution. Quaternary Research 46, Article No. 0042:37 47. ERDAS 1990 Field Guide, Version 7.4. ERDAS, Inc., Atlanta, Ga. Fearnside, P.M. 1986 Spatial concentration of deforestation in the Brazilian Amazon. Ambio 15:74-81. Food and Agriculture Organization 1991 Second Interim Report on the State of Tropical Forests. Unpublished paper, Forest Re- sources Assessment 1990 Project, United Nations FAO, Rome, Italy. Graham, I. 1997 Mission to La Carona. Archaeology, September/October:46. Graham, M.H., B.G. Junkin, M.T. Kalcic, R.W. Pearson, and B.R. Seyfarth 1985 Earth Resources Laboratory Applications Software (ELAS), Revision 3. NASA-NSTL- Earth Resources Laboratory, Report No. 183, Stennis Space Center, Miss. Hammond, N. 1990 Ancient Maya Civilization. New Brunswick, N.J.: Rutgers University Press. Jensen, J.R. 1986 Introductory Image Processing-A Remote Sensing Perspective. Englewood Cliffs, N.J.: Prentice-Hall. Jones, C.L. 1940 Guatemala: Past and Present. Minneapolis: University of Minnesota Press. Kelley, P.S. 1983 Digital analysis in remote sensing: from data to information. Pp. 218-240 In Remote Sensing of the Environment, B.F. Richason, Jr. ed. Lillesand, T.M., and R. Kiefer 1994 Remote Sensing and Image Interpretation. New York: John Wiley and Sons. Lundell, C.L. 1937 The Vegetation of the Peten. Carnegie Institute Publication No. 478, Washington, D.C. Nations, J. 1988 Tropical Rainforest: Endangered Environment. New York: Franklin Watts. 161
62 VALIDATING SOCIAL PHENOMENA, PETEN, GUATEMALA Pope, K.D., and B.H. Dahlin 1989 Ancient Maya wetland agriculture: New insights from ecological and remote sensing. Journal of Field Archaeology 16:87-106. Radar detection and ecology of ancient Maya canal systems. Journal of Field Archaeol- ogy 20:379-383. Rice, D.S. 1991 Roots. Natural History February:10-14. Sabloff, J.A. 1995 Drought and decline. Nature 375: 357. Sader, S.A. 1995 Spatial characteristics of forest clearing and vegetation regrowth as detected by Landsat Thematic Mapper imagery. Photogrammetric Engineering and Remote Sensing 61: 1145- 1151. Sader. S.A., and A.T. Joyce 1988 Deforestation rates and trends in Costa Rica, 1940-1983. Biotropica 20(1): 11-19. Sader, S.A., T.A. Stone, and A.T. Joyce 1990 Remote sensing of tropical forests: an overview of research and applications using non- photographic sensors. Photogrammetric Engineering and Remote Sensing 55(10):1343- 1351. Sader, S.A., G.V.N. Powell, and J.H. Rappole 1991 Migratory bird habitat monitoring through remote sensing. International Journal of Re mote Sensing 12(3):363-372. Sader, S.A., T. Sever, J.C. Smoot, and M. Richards 1994 Forest change estimates for the northern Peten region of Guatemala 1986 to 1990. Hu man Ecology 22(3):317-322. Sader, S.A., T. Sever, and J.C. Smoot 1996 Time series tropical forest change detection: A visual and quantitative approach. Interna- tional Symposium on Optical Science, Engineering and Instrumentation, Denver, Colo. SPIE, Volume 2818-2-12. Schuster, A.M.H. 1997 The search for Site Q. Archaeology September/October:42-45. Schwartz, N.B. 1990 Forest Society: A Social History of Peten, Guatemala. Philadelphia: University of Penn- sylvania Press. Sellers, P.J. 1985 Canopy reflectance, photosynthesis and transpiration. Sensing 6(8):1335-1372. International Journal of Remote Sever, T. 1990 Remote Sensing Applications in Archeological Research: Tracing Prehistoric Human Impact Upon the Environment. Ph.D. dissertation, University of Colorado, Boulder. Available: University Microfilms, Ann Arbor, MI. Sheets, P., and T. Sever 1988 High tech wizardry. Archaeology 41(6) Nov/Dec:28-35. Skole, D., and C. Tucker 1993 Tropical deforestation and habitat fragmentation in the Amazon: Satellite data from 1978 to 1988. Science 260: 1905-1910. Stuart, G.E. 1991 Maya heartland under siege. National Geographic 10:94-107. Trombold, C.D. 1991 Ancient Road Networks and Settlement Hierarchies in the New World. Cambridge, En- gland: Cambridge University Press.
THOMAS L. SEVER 163 Tucker, C.J. 1979 Red and photographic infrared linear combinations for monitoring vegetation. Remote Sensing Environment 8:127-150. Turner II, B.L. 1993 Rethinking the "new orthodoxy:" Interpreting ancient Maya agriculture. Pp. 57-88 In Culture, Form, and Place: Essays in Cultural and Historical Geography, K. Mathewson, ed. Geoscience and Man 32. Baton Rouge, La.: Louisiana State University.