Modeling deforestation risk in the Maya Biosphere Reserve, Guatemala
Item Type text; Thesis-Reproduction (electronic)
Authors Grunberg, Wolfgang
Publisher The University of Arizona.
Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.
Download date 07/10/2021 09:45:22
Link to Item http://hdl.handle.net/10150/278736 INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter ^ce, while others may be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistirKt print, colored or poor quality illustrations and photographs, print bieedthrough, substandard margins, and improper alignment can adversely affect reproduction.
In the unlikely event that the author dki not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g.. maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overiaps.
Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6' x 9' black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additranal charge. Contact UMI directly to order.
Bell & Howell lnformatk>n and Learning 300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA 800-521-0600
MODELE^G DEFORESTATION RISK IN THE MAYA BIOSPHERE RESERVE, GUATEMALA
by
Wolfgang Griinberg
A Thesis Submitted to the Faculty of the
SCHOOL OF RENEWABLE NATURAL RESOURCES
In Partial Fulfillment of the Requirements For the Degree of
MASTER OF SCIENCE WITH A MAJOR IN RENEWABLE NATURAL RESOURCES STUDIES
In the Graduate College
THE UNIVERSITY OF ARIZONA
2000 UMI Number 1401059
® UMI
UMI Microfomi 1401059 Copyright 2000 by Bell & Howell Infomiation and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.
Bell & Howell Information and Leaming Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 -7
STATEMENT BY AUTHOR
This thesis has been subnfutted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.
Brief quotations from this thesis are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgement the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.
SIGNED:
APPROVAL BY THESIS COMMITTEE
This thesis has been approved on the date shown below:
William W. Shaw Date Professor of Renewable Natural Resources
D. Phillip Guertin Associate Professor of Watershed Management
Wissler Date Assistant Professor of Renewable Natural Resources 3
ACKNOWLEDGEMENT
The author would like to thank the following organizations and individuals for their indispensable help: Advanced Resource Technology Group - The University of
Arizona, CARE Guatemala, Centro de Monitoreo y Evaluacion (CEMEC) - Consejo
Nacional de Areas Protegidas (CONAP), ProPeten - Conservation International (CI),
Wildlife Conservation Society (WCS) - Gainesville, Roan Balas, Perfecto Carillo,
Teresita Chinchilla, Gary Christopherson, Reno Fiedler, Georg Griinberg, D. Phillip
Guertin, Vinicio Montero, Randy H. Gimblett, Michael J. Meitner, Gustavo Rodriguez
Ortiz, Marco Antonio Palacios, Victor Hugo Ramos, Steven A. Sader, Claudio Saito,
Norman B. Schwartz, William W. Shaw, Carlos Soza, Laura Stewart, and Craig
Wissler. 4
DEDICATION
This thesis is dedicated to my father Georg Griinberg for his enthusiastic and essential support of my studies and research.
Che ru Apyka Rendypeguara. 5
TABLE OF CONTENTS
1 LIST OF FIGURES : 7 2 LIST OF TABLES 8 3 LIST OF ACRONYMS 9 4 ABSTRACT 10 5 INTRODUCTION II 5.1 Problem Statement 11 5.2 Statement of Objective 13 5.3 Study Area 14 5.3.1 The Landscape 14 5.3.2 The People 14 5.3.2.1 Itza'Maya 15 5.3.2.2 Ladinos Peteneros 16 5.3.2.3 Highland Mayas 16 5.3.2.4 Ladinos Sureno 17 5.3.3 The Maya Biosphere Reserve 17 6 LITERATURE REVIEW 19 6.1 Man-Made Deforestation - Causes, Consequences, and Trends 19 6.2 Methodology for Modeling Deforestation 21 7 METHODS 24 7.1 Data 24 7.1.1 Forest Cover Change-Detection Images 24 7.1.2 Settlement Data 25 7.1.3 Road Data 27 7.1.4 Soil Data 27 7.1.5 Reference Data 28 7.2 Analysis 29 7.2.1 Settlement Analysis 29 7.2.2 Road Analysis 31 7.2.3 Soil Analysis 32 7.2.4 Deforestation Probability Model 32 7.2.5 Forecasting Deforestation Risk 34 8 RESULTS 37 8.1 Settlements and Deforestation 37 8.2 Roads and Deforestation 41 8.3 Soil Quality and Deforestation 42 8.4 Deforestation Probability 43 8.4.1 Logistic Regression Results 43 8.4.2 Deforestation Probability Model 44 8.5 Deforestation Risk Model 52 8.5.1 Testing the 1999 Forecast 52 8.5.2 The 2001 Deforestation Scenario 55 6
9 DISCUSSION 59 9.1 Modeling Deforestation 59 9.2 Strengths and Limitations of the Deforestation Risk Model 59 9.3 Management Implications and Recommendations 60 9.4 Future Analysis 61 10 APPENDIX A - SETTLEMENT DATABASE 63 11 APPENDIX B - SETTLEMENTS AND ROADS MAP 68 12 APPENDIX C - ARC/INFO COMMANDS 69 13 WORK CITED 70 7
1 LIST OF FIGURES Figure 1. The Maya Biosphere Reserve (MBR) in relation to Guatemala and the neighboring countries (Image source: ESRI ARC/INFO data set) 12 Figure 2. The Maya Biosphere Reserve and its core, buffer, and transition zones 13 Figure 3. Peten's population estimates according to Grandia (2000) and Schwartz (1990). 15 Figure 4. Annual deforestation rates of the Maya Biosphere Reserve from 1986 to 1999 according to Sader et al. (1997, 2000) 21 Figure 5. Buffering the El Naranjo settlement by the San Pedro River and the surrounding area in 1997 31 Figure 6. Steps taken to create the 1997 descriptive deforestation probability surface and values used to weight the grids 34 Figure 7. Steps taken to forecast the 1999 deforestation probability surface and values used to weight the grids 35 Figure 8. Forecasting the deforested area based on forecasted probability zones and observed deforestation 36 Figure 9. Deforestation distance decay curves of settlements according to primary economic occupation 40 Figure 10. Deforestation distance decay curves of settlements according to their ethnic majority 40 Figure 11. Deforestation trends on various soil classes 42 Figure 12. Percent deforestation observed in low (0) to high (1) deforestation probability zones over time 45 Figure 13. MBR's settlements, roads, and cumulative deforestation in 1986 46 Figure 14. Deforestation probability model for 1986 46 Figure 15. MBR's settlements, roads, and cumulative deforestation in 1990 47 Figure 16. Deforestation probability model for 1990 47 Figure 17. MBR's settlements, roads, and cumulative deforestation in 1993 48 Figure 18. Deforestation probability model for 1993 48 Figure 19. MBR's settlements, roads, and cumulative deforestation in 1995 49 Figure 20. Deforestation probability model for 1995 49 Figure 21. MBR's settlements, roads, and cumulative deforestation in 1997 50 Figure 22. Deforestation probability model for 1997 50 Figure 23. MBR's settlements, roads, and cumulative deforestation in 1999 including man-made wildfires from summer 1998 51 Figure 24. Deforestation probability model for 1999 51 Figure 25. Forecasted deforestation probability model for 1999 53 Figure 26. Observed vs. predicted deforestation area of the 1999 deforestation risk model 54 Figure 27. Forecasted deforestation probability model for 2001 56 Figure 28. 2001 deforestation scenario forecast compared to observed deforestation. ...58 8
2 LIST OF TABLES Table 1. The Maya Biosphere Reserve management units and their areas according to Griinberg and Ramos (1998) 18 Table 2. Soil reclassification according to drainage and soil depth 28 Table 3. Logistic regression results for the deforestation probability model 44 Table 4. Differences between predicted and observed deforestation for 1999 55 Table 5. Differences between the deforestation predicted by the 2001 scenario and observed deforestation for 1999 57 9
3 LIST OF ACRONYMS Acronym Name Translation GATE Centre Agronomico Tropical de Tropical Agronomy Center for Investigacidn y Ensenanza Research and Education CEMEC Centro de Monitoreo y Evaluacion del Center for Monitoring and CONAP Evaluation of CONAP CI Conservation International CONAP Consejo Nacional de Areas Protegidas National Council of Protected Areas ESRI Environmental Systems Research Institute, Inc. FAO Food and Agricultural Organization of the United Nations FDN Fundacion Defensores de la Naturaieza Defenders of Nature Foundation GIS Geographic Information System GPS Global Positioning System MAB UNESCO Man and the Biosphere Program MBR Maya Biosphere Reserve MIAL Maine Image Analysis Laboratory NDVI Normalized Difference Vegetation Index RBM Reserva de la Biosfera Maya Maya Biosphere Reserve SEGEPLAN Secretaria General del Consejo General Secretariat for National Nacional de Planificacion Economica Economic Planning UNESCO United Nations Educational, Scientific, and Cultural Organization USAID U.S. Agency for International Development WCS Wildlife Conservation Society ZAM Zona de Amortiguamiento Buffer Zone ZUM Zona de Usos Multiples Multiple Use Zone 10
4 ABSTRACT The tropical forest of Guatemala's 21,130 square kilometer Maya Biosphere
Reserve and buffer zone is being impacted by deforestation due to an increase of the local population and establishment of over 200 new settlements over the last 20 years.
Existing geographic information system databases and remote sensing data were used to determine how much of the observed deforestation could be explained by three factors: roads, human settlements, and soil quality. Each factor was analyzed separately using spatial and statistical analysis methods. These factors were then combined to create a final deforestation risk model. The deterministic model enables policy makers, as well as managers, to create scenarios that assess the impact of their actions on the forest on a regional scale. 11
5 INTRODUCTION
5.1 PROBLEM STATEMENT
Deforestation of tropical forests is a widespread management and conservation issue throughout the tropics. In the case of the tropical lowland forest in the northern
Peten, Guatemala, national and imemational conservation and development agencies have collected biophysical and socio-economic data that documents the area's deforestation trends and factors over the past years. The province of Peten, located in northern Guatemala, contains one of the largest continuous tropical forests of Central
America. In 1990, 40% of the forested area (21,130 square kilometers; CONAP et al.
1996) was set aside as the Maya Biosphere Reserve (MBR) with a surrounding Multiple- use Zone (Zona de Usos Multiples - ZUM - the equivalent of a Man and the Biosphere
Program - MAB - biosphere reserve buffer zone) and Buffer Zone (Zona de
Amortiguamiento - ZAM - the equivalent of a MAB biosphere reserve transition zone:
Figure 1, Figure 2, Table 1). However, the population of the Peten has increased dramatically through immigration of various Mayan peoples and Ladinos (Guatemalan people of mixed Hispanic and Mayan descent) into the region. This has resulted in the establishment of more than 200 new settlements over the last 20 years and has led to extended and rapidly increasing deforestation of the Peten including the reserve
(Griinberg and Ramos 1998). The settlements" subsistence economies - based on small- scale ranching and traditional agriculture - led not only to well-documented loss of biodiversity but also to economic loss due to soil degradation (CONAP et al. 1996). 12
Together, the reserves, national parks, ZAM, and ZUM compnse the MBR.
However, little is known regarding specific environmental, as well as socio-economic factors, that are contributing to the deforestation of the MBR. Quantifying the importance of economic and land use factors in terms of their relationship to deforestation would enable policy makers, as well as managers, to create scenarios and model the impact of their actions on the forest on a regional scale. This thesis will explore the influence of three factors: roads, human settlements, and soil quality on deforestation and how those factors can be used to model deforestation.
Figure 1. The Maya Biosphere Reserve (MBR) in relation to Guatemala and the neighboring countries (Image source: ESRI ARC/INFO data set). 13
Mexico
0 c - 0 ~ n o Maya Biosphere Reserve 0 Flores o o o o Core Zone oo c ~ (} 0 CJ Buffer Zone 0 Santa .Ana, 7. · . D Transition Zone LaLibertad 10 0 10 20 30 40 50 Kilometers - Guatemala.. - / .J_.-y.l ...... -1<.
Figure 2. The Maya Biosphere Reserve and its core, buffer, and transition zones.
5.2 STATEMENT OF OBJECTIVE
The objective of this study can be summarized into five consecutive steps. The
first objectives was to collect, compile, normalize, and enhance available spatial data and their attributes of the MBR to create a geographic information system (GIS) for display and analysis purposes. The second step involved the separate analysis of the relationship of deforestation and man-made factors (such as settlements and roads) and environmental factors (such as bodies of water, soil quality, elevation, aspect, slope, and vegetation cover) over space and time. The third step concerned the combination and testing of relevant deforestation factors into a deterministic deforestation risk model that quantified deforestation probability and quantity. The fourth step was to prototype a deforestation forecast based on a future scenario. The fmal objective of this study was to discuss 14 weaknesses and strengths of the deforestation model and to identify future research needs. The study's results and source-data will be disseminated to the institutions and individuals who have contributed to this study or who may benefit from it.
5.3 STUDY AREA
5.3.1 The Landscape
The Peten is comprised of a hilly, limestone karst landscape with an elevation from ca. 100 to 300 m above mean sea level (Islebe et al. 1996). Its mean annual temperature is 25° C and the region's precipitation averages 1600 mm a year (Islebe et al.
1996). The region's vegetation zones consist of the predominating high canopy tropical lowland forest, occasionally inundated lowland areas or bajos of chaparral-like dense shrubs and small trees, wetlands along rivers and around lakes, and flat savanna-like grasslands.
5.3.2 The People
The area remained sparsely populated for most of the last 500 years by the native lowland
Mayan peoples and Ladinos. They traditionally practiced swidden agriculture - farming of fallow and forestland cleared by slash and bum - and agroforestry (Atran 1993).
Modem settlements in the MBR base their subsistence economy on ranching and swidden agriculture with increasingly shorter fallow periods (Santiso 1993, Schwartz 1990). Not until the 1950's was there a rapid population expansion in the Peten, from ca. 16,000 people in 1950 to 500,000 in 1999, observed (Grandia 2000; Schwartz 1990; Figure 3). 15
In 1998, an estimated 87,106 people lived in the greater MBR forming an estimated 28% of the Peten province's population (Griinberg and Ramos 1998). Most new settlers have uncertain status regarding land tenure, with no clear land rights.
Peten's Population from 1714 to 1999 500.000 - 450.000 - 400.000 - 350,000 - § 300,000 |- -i 250,000 - S. 200,000 - 150,000 100,000 50,000
1700 1750 1800 1850 1900 1950 2000 Year
Figure 3. Peten's population estimates according to Schwartz (1990) and Grandia (2000).
5.3.2.1 Itza' Maya
Originally, the Itza' Maya population moved from the Yucatan peninsula in the north into the area that is now part of the MBR's buffer zone before the Spanish conquest
(Atran 1993). The Itza', who rely heavily on forest products, are not considered to be related to the ancient Mayan civilizations that built the City States which, now in ruins, are famous archeological and tourist attractions for the region. Today, the small Itza' communities rely on growing com, agroforestry, wage labor, hunting, and collecting xate
(Chamaedorea elegans) - a palm whose ornamental leaves are exported to the U.S. and
Europe. San Jose is the only predominantly Itza' settlement within the greater MBR. 16
5.3.2.2 Ladinos Peteneros
The Peteneros, a small local non-immigrant population of mixed Mayan and
Hispanic descent, have continuously lived in the northern Peten and the greater MBR since the Spanish colonization period beginning in 1697 (Schwartz 1990, p. 48).
Traditionally, the Peteneros have been close to the Itza' and adopted much of their knowledge and agroforestry techniques such as home-gardens, rotational swidden farming with multi-cropping, and forest gardens (Atran 1993). Although agriculture, agroforestry. hunting, and xate collecting are also prevailing income and food sources, the Peteneros are currently shifting their economic mode to wage labor, commerce, and ranching. The Peteneros form the majority in 5 settlements within the greater MBR.
5.3.2.3 Highland Mayas
The immigrant Mayan peoples such as the Q'eqchi' and Kaqchikel entered the
Peten in the last 30 years from the south, due to socio-economic pressures such as a population boom in their original communities, limited access to land, and persecution
(Georg Griinberg 1998, personal communication). Similar to the Itza', their main staple is com. However, their agricultural techniques are mostly adapted to the often volcanic soils and more moderate climate of the highlands of central Guatemala, and not to the fragile soils of the Peten. Although the older immigrant communities are notably adapting their agriculture techniques to the new environmental conditions, soil degradation, pests, and low yields are common. Immigrant Mayans form the majority in
24 settlements within the greater MBR. 17
5.3.2.4 Ladinos Surefio
The vast majority of immigrants into the Peten are of mixed Hispanic and Mayan
descent - called Surenos or Ladinos - and have been moving into the MBR and its buffer
zone in the last 20 years for mostly economic reasons (Griinberg and Ramos 1998).
Similar to the highland Mayans, recent Ladino immigrants rely on subsistence farming of
com and beans using swidden-fallow agriculture and suffer from the same mal-adapted
agricultural techniques such as shortened fallow periods, easily eroded or inundated field
locations, and soil nutrients degradation through limited crop choices (Palma, 2000).
However, Ladinos tend to introduce small-scale ranching as soon as they can afford it.
The immigrant Ladinos form the majority in 145 settlements within the greater MBR.
5.3.3 The Maya Biosphere Reserve
The Reserva de la Biosfera Maya or Maya Biosphere Reserve was established in
1990 as part of UNESCO's Biosphere Reserve Network through the Man and the
Biosphere Program (MAB, Table 1). The MBR also includes the Tikal National Park, a
World Heritage Site since 1979. The objective of a Biosphere Reserve is to balance three functions: the conservation of biodiversity and ecosystems; human and economic development in an ecologically, economically, and culturally sustainable way; and
logistical support for an international network of research, monitoring, education, and
information exchange. A Biosphere Reserve is divided into three zones. The core zones are considered strictly protected areas with minimal human impact. Buffer zones, called
Multiple-Use Zone (ZUM) in the MBR, surround the core zones and theoretically allow IB only low-impact human activities such as research, recreation, and limited use of forest- products. Transition zones, called Buffer Zone (ZAM) in the MBR, form the outermost layer where sustainable use of natural resources by the local communities is encouraged.
The MBR and its buffer and multiple use zones are administrated by the
Guatemalan Consejo Nacional de Areas Protegidas (CONAP) with the aid of the Centro
Agronomico Tropical de Investigacion y Ensenanza (CATIE) and the U.S. Agency for
International Development (USAID) - Guatemala. However, several conservation agencies - such as Conservation International (CI), Fundacion Defensores de la
Naturaleza (FDN), The Nature Conservancy (TNC), the Wildlife Conservation Society
(WCS), and World Wildlife Fund (WWF) - and development aid agencies - such as
CARE International Guatemala - are involved in the monitoring and management of the greater MBR.
Table 1. The Maya Biosphere Reserve management units and their areas according to Griinberg and Ramos (1998). MBR Management Unit Name MAB Biosphere Reserve Function Area (km^) National Parks and Reserves Core Zone 7670 Multiple Use Zones (ZUM) Buffer Zone 8484 Buffer Zone (ZAM) Transition Zone 4975 21129 19
6 LITERATURE REVIEW
6.1 MAN-MADE DEFORESTATION - CAUSES, CONSEQUENCES, ANT) TRENT>S
Deforestation propagated by man is driven by various social and economic
causes. Many causes of land degradation, such as population changes, marginalization of
land, poverty, land ownership problems, and political or social instability can be
attributed to the world-wide loss of forest, woodland, and shrub cover (Barrow 1991).
Specifically, C. J. Barrow (1991) lists population increase, improved access ways, large
hydroelectric dams, expansion of shifting subsistence agriculture, large-scale agriculture, failure to assist the poor, increasing demand of forest products, administrative errors, and
land settlement schemes as the primary reasons for tropical forest degradation.
The main causes of deforestation observed in the MBR are due to population growth and land ownership conflicts in southern and central Guatemala and the resulting internal migration of landless farmers to the Peten province. To a lesser degree, large- scale cattle ranching, selective logging of hardwoods, oil exploration, and intrusion by
Mexican loggers along the northwestern border to the State of Tabasco, Mexico, contributed to the observed deforestation (Schwartz 1990, CONAP et al. 1996, Sader et al. 1997, Griinberg and Ramos 1998). The Guatemalan government opened the Peten for colonization and agricultural development in the 1960's (Schwartz 1990). Since then mostly new swidden agriculture settlements pushed an "agricultural frontier" from the southern Peten to the MBR in the north (Griinberg and Ramos 1998). 20
The effect of deforesting tropical forests can have global to local consequences.
On a global scale, deforestation has been linked to climate change, degradation of renewable natural resources - such as soils, water, and forest products - and loss of biodiversity (lUCN et al. 1991, Wilson 1992). Deforestation in Guatemala leads in local areas to direct and indirect rep>ercussions which have been observed also in the Peten.
Some of the direct consequences were: loss of tourist income; loss of precious timber, such as caoba (mahogany - Swietenia macrophvlla) and cedro (cedar - Cedrela sp.): loss of game for food consumption; loss of non-timber forest products, such as xate^ chicle
(rubber - Achras zapota) and pimienta gorda (allspice - Pimenta officinales); and the regional extinction of flora and fauna (Schwartz 1990, CONAP et al. 1996, Palma 2000).
Indirect effects of deforestation were: reduced precipitation; drying of wells, water holes, and springs; degradation of agricultural soils through nutrient depletion and compaction; erosion; and an increase in temperature (CONAP et al. 1996).
Close to 60% of the tropical lowland forest in Central America had been deforested by 1980 (Kricher 1997). By 1985, approximately 38% of the Peten had been deforested (Schwartz 1990). In the MBR, the yearly deforestation rate increased steadily from 1986 to 1995 within the MBR's core - an increase from 0.04% to 0.33% - and the multiple use zone (ZUM) - an increase from 0.05% to 0.25% (Figure 4, Sader et al 1997).
The yearly deforestation rate for the transition zone (ZAM) was not only higher but also increased dramatically over time - an increase from 0.74% to 3.76% - as can be expected from the settlement boom observed in the same area (Figure 4, Sader et al. 1997,
Griinberg and Ramos 1998). 21
The MBR's Deforestation Rate from 1986 to 1999 4.0
2.5
2.0 Reserves & National Parks m — Multiple Use Zone (ZUM) A Buffer Zone (ZAM)
0.5
0.0 1986-90 1990-93 1993-95 1995-97 1997-99 * Analysis Period * Up to 10% of trie areas were not included in the 1999 analysis due to cloud-cover or missing images.
Figure 4. Annual deforestation rates of the Maya Biosphere Reserve from 1986 to 1999 according to Saderet ai. (1997, 2000).
6.2 METHODOLOGY FOR MODELING DEFORESTATION
Deforestation is ultimately a spatial and temporal phenomenon. However, man- caused deforestation is only the symptom of socio-economic causes rooted in the complex relationship between physical and social needs of humans and their environment. Therefore, it comes as no surprise that deforestation studies and models run the gamut from a more statistical look at the relationship of economic and census data to deforestation (Pfaff 1996) to a more spatial approach, such as attempted in this study.
Deforestation, a form of land-use change, is most often detected, monitored, and analyzed by using remotely sensed aerial and satellite imagery (Sader 1995). The 22 deforested areas are most often identified by their spectral characteristics through supervised or unsupervised classification of the images' Normalized Difference
Vegetation Index (NDVI), and by visual identification of deforested areas (Sader 1995,
Chomitz and Gray 1996, Wood and Skole 1998).
The question of quantifying and locating deforestation leads often to the question of identifying deforestation causes. Due to the spatial nature of deforestation, deforestation studies, such as in Cameroon (Mertens and Lambin 1997), Costa Rica
(Juarez 1994), Belize (Chomitz and Gray 1996), Mexico (Mas et al. 1997), Philippines
(Apan and Peterson 1998), and Brazil (Dale and Pearson 1999), have employed a geographic information system (GIS) to compare the remotely sensed deforestation images with deforestation related spatial features such as roads, settlements, soil quality, and slope.
To quantify, explain, and model the relationship between causes of deforestation and deforestation itself, various analysis methods and their combinations are used. The most common and basic analysis method apjjears to be the spatial overlay of deforestation related themes to see the spatial coincidence of deforestation factors and deforestation. Spatial overlay is often combined with nearest-neighbor analysis of deforestation and roads, bodies of water, or markets, etc. (Mas et al. 1997). Some deforestation analysis methods include mathematical framework models (Behrens 1996), statistical models (Mas et al. 1997, Apan and Peterson 1998), and stochastic dynamic simulations (Mertens and Lambin 1997, Dale and Pearson 1999). 23
To predict deforestation, some studies use empirical econometric models which determine the economically most valuable areas - based on travel time to markets, soil fertility, and access to water (Nelson and Hellerstein 1995, Chomitz and Gray 1996, Pfaff
1996). To predict patterns of land clearance, settlement patterns, and carbon use in the
Brazilian Amazon, Dale and Pearson (1999) created a dynamic stochastic simulation based on a settlement diffusion model, an ecological impact model, observed changes in road networks, and land-use changes. Because deforestation is not static but a process in a heterogeneous environment. Martens and Lambin (1997) approached deforestation probability prediction with a spatial model composed of a series of univariate models that each covered a homogenous area such as subsistence agriculture areas, peri-urban areas, roadside areas, and areas dominated by logging.
Although similar to other statistical probabilistic models such as those from
Manens and Lambin (1997) and Apan and Peterson (1998), this study's deforestation probability model based on logistic regressions (Chou 1997) was derived from Warren's
(1990) and Christopherson et al.'s (1996) predictive models of archeological site locations. Unlike stochastic simulations, this study's probabilistic model does not pinpoint deforested lots, but designates zones of high and low deforestation probabilities. 24
7 METHODS
7.1 DATA
The study's GIS database, compiled from various projections, sources, and
formats, was standardized to the UTM projection. Zone 16, units in meters, and NAD
Central America datum. The different vector and raster themes had a root mean square error (RMS) of approximately 400 meters to each other (Thapa and Bossier 1992). Most of the RMS error can be attributed to the small scale of the roads themes and to the selective availability or pseudo random error in the global positioning system (GPS) acquisition of settlement locations.
7.1.1 Forest Cover Change-Detection Images
The foundation of this study was a series of forest cover change-detection images for nearly the entire MBR developed by the Maine Image Analysis Laboratory (MIAL),
University of Maine, and provided by ProPeten - CI. MIAL derived the change-detection images from 3 Landsat Thematic Mapper (TM) satellite scenes for each analysis year.
The scenes' visible red band (TM 3) and near-infrared band (TM 4) were atmospherically corrected before a normalized difference vegetation index (NDVI = [near infrared - red]/[near infrared + red]) was calculated (Sader et al. 1997, 1998). NDVI is a measure of "greenness" that is correlated with green vegetation cover and leaf area index. The individual NDVI scenes were then reclassified by using an unsupervised clustering technique and further reclassified to change and no-change categories by using aerial photography and ground truthing (Sader et al. 1997). Finally, MIAL mosaiced the 3 separate scenes into one image. A NDVI can be used to detect forest cover changes especially if the forest is clear-cut which would show up as significant differences in greenness.
The forest cover change-detection images displayed forest cover changes at a 30 by 30 meter resolution for the first and second quarter of 1986, 1990, 1993, 1995, 1997, and 1999. Some TM scenes were from different months due to cloud-cover and image availability. The change-detection images represented a cumulative deforestation process and did not account for forest regrowth. The 1990 to 1997 forest cover change-detection images had an overall classification accuracy of 86.5% (Sader et al. 2000). The 1999 image, for the 1997 to 1999 period, did not cover the entire study area due to cloud- cover. In addition, the 1999 forest cover change-detection was hindered by widespread man-made wildfires in the summer of 1998 caused by drought and uncontrolled slash- and-bum forest clearing (Stanley 1998).
7.1.2 Settlement Data
Spatial and socio-economic data for 194 settlements up to March 2000, were provided by CARE Guatemala and the Centro de Monitoreo y Evaluacion del Consejo
Nacional de Areas Protegidas (CEMEC-CONAP; Griinberg and Ramos 1998, personal communication). Field agents from the Base de Datos sobre Poblacidn Tierras y Medio
Ambiente en la Reserva de la Bidsfera Maya project collected the main body of the socio-economic database from July 1997 to March 1998 (Griinberg and Ramos 1998).
Continuous updates of new settlements were added up to March 2000. Each settlement 26 was located by using GPS units, and socio-economic data was collected by field observations and questioning the representatives of the settlements. Settlement attributes collected included settlement type, ethnicity, language used, administrative and political affiliations, potable water sources, dominant economic activities, available health services, available educational services, foundation year of the settlement, census, and a short narrative description of the settlement s history and peculiarities.
The population count was often only based on the number of families within a settlement. An average of 5.5 inhabitants was estimated per family (Grunberg and
Ramos 1998). When available, previous census information from other studies was included. The census data was often incomplete, came from unreliable sources, and covered no more than three time periods. Because of its inadequacy, the census data was not used in the deforestation analysis.
The settlements were then reclassified and grouped into more generalized socio economic categories according to the settlements database and the help of anthropologists
Norman Schwartz and Georg Grunberg. The generalization was necessary to boost the settlement sample sizes for a deforestation analysis according to socio-economic grouping. The two new settlement attributes were ethnic majority and primary occupation. Categories for ethnic majority included Immigrant Ladinos, Immigrant
Maya, local Peteneros and Itza', and non-settlement sites such as ecological research stations, natural resource management stations, archeological sites, oil production camps, and other populated sites maintained by governmental and non-governmental (NGO) 27 institutions. Occupational categories included agriculture, transition from agriculture to small-scale ranching, small-scale ranching, forestry and forest products, and wage labor.
7.1.3 Road Data
The basic roads vector data, developed by the Secretaria General del Consejo
Nacional de Planificacion Economica (SEGEPLAN), were provided by CEMEC-
CONAP. Additional roads vector data covering the border areas of Belize (1:250,000 topographic map source from 1980) and Mexico (1:250,000 topographic map source from 1975) were provided by the WCS - Gainesville. Where identifiable, the deforestation images were used to correct spatial errors or add known roads to the roads vector data.
The roads vector data was then attributed according to quality and age. The road quality attribute differentiated between perennial versus intermittent and absence versus presence of public transportation. The age attribute included the year of perennial road and public transportation inception. The attributes were compiled with the help of Bus
Service entrepreneur Perfecto Carillo, anthropologist Georg Griinberg from CARE -
Guatemala, agronomist Vinicio Montero of CONAP, architect Marco Antonio Paiacios, and anthropologist Norman Schwartz from the University of Delaware.
7.1.4 Soil Data
An electronic 1:200,000 Food and Agricultural Organization of the United
Nations (FAO) soil series pK>lygon map was provided by CEMEC-CONAP. Erroneous 28 and missing attributes were corrected and completed with the help of Vinicio Montero and a hardcopy version of the soil series map from SEGEPLAN. The soil series map was
then reclassified into drainage and soil depth categories according to Vinicio Montero and local and FAO agricultural suitability guidelines (APESA/AHT 199L Carlos Collado
1998a and 1998b; Table 2). According to the agricultural soil suitability guidelines, deep soils are preferred over shallow soils and, to a lesser degree, well draining soils are preferred over poorly draining soils. The soil drainage and depth categories do not account for local variations and topography.
Table 2. Soil reclassification according to drainage and soil depth. Soil Series Soil Order Reclassification Chachadun Cambisoles Cromicos/Luvisoles Deep well draining soils Quinil-Cuxu Rendzinals/CambisolesA^ertisoles/Gleysoles Deep well draining soils Qiiinil-Yaxha-Chapayal- Rendzinals/Cambisoles/Vertisoles Deep well draining soils Uaxactun Sotz Cambisoles Cromicos-Gleicos/Luvisoles Cromicos Deep well draining soils Bolon Cambisoles Gleicos/Qleysoles Deep poorly draining soils Chocop-Saipuy Cambisoles Gleicos Deep poorly draining soils Exkixil Cambisoles Cromicos Deep poorly draining soils Mopan Cambisoles Verticos-Gleicos/Fluvisoles Deep poorly draining soils Petexbatun Fluvisoles Districos Deep poorly draining soils Sarstun Gleysoles Eutricos/Fluvisoles Deep poorly draining soils Usomacinta Fluvisoles Eutricos/Gteysoles Eutricos Deep poorly draining soils Chacalte-Cuxu Rendzinals/Utosoles/Cambisoles Shallow well draining soils Joija Cambisoles Calcicos/Vertisoles Shallow well draining soils Cromicos/Luvisoles Sacfuc Cambisoles Eutricos Shallow well draining soils (Gleysoles) Gleysoles Shallow poorly draining soils Macanche-Yaloch Vertisoles Pelicos Shallow poorly draining soils
7.1.5 Reference Data
Vector data on administrative boundaries, rivers, and lakes originally digitized from 1:50,000 and 1:250,000 topographic maps were provided by CEMEC-CONAP.
They also provided vector data on oil pipelines collected with GPS units within the greater MBR. WCS - Gainesville provided additional vector and point data on 29 administrative boundaries, rivers, lakes, towns, and archeological sites for northern Peten and the surrounding areas in Mexico and Belize (Selva Maya Database, WCS).
7.2 ANALYSIS
All spatial and statistical analyses were performed with the SPSS statistics package (http://www.spss.com) and the ARC/INFO and ArcView GIS software
(http;//www.esri.com) on Windows NT and Solaris workstations. The analysis of the relationship of landscape features and deforestation was limited to agricultural soil suitability. The initial intention of incorporating slope, aspect, elevation, and vegetation cover was hindered by the lack of appropriately scaled digital elevation and vegetation classification data. Those themes are, however, now available (Victor Hugo Ramos and
Steven Sader 2(XX), personal communication). The low resolution bodies of water data set was not included in the analysis because the overwhelming majority of water sources for the subsistence settlements, such as wells, springs, sink holes, and water holes, did not show up. Rivers and lakes, however, may be significant for medium to large-scale cattle ranching.
7.2.1 Settlement Analysis
Until 1998, there were no reliable land registries for most of the settlements. This made any direct deforestation impact analysis for specific settlements difficult. To estimate the deforestation impact of settlements, each settlement point was buffered by
20 concentric 1-km-diameter rings at increasing distances from the center (Figure 5).
Then, the forest cover change-detection images for 1986 to 1997 were used to determine 30 the total land surface area, forested area, and deforested area for each ring and settlement.
Parts of the outer rings of the settlements at the edge of the study fell outside the change- detection images. In those cases, the percentage of no data values of the total area was calculated for error estimation.
All area values for the individual distance rings of the settlements for 1986, 90,
93, 95, and 97 were related with the settlements' socio-economic attributes in a relational database using the Microsoft Access software package
(http://www.microsoft.com). The relational database enabled one to query the area deforested at various distance intervals according to the analysis year, percent of no data, and socio-economic attributes of the settlements. The query outputs were then used to calculate the average percent deforestation within the 20 one km rings of villages grouped according to their socio-economic attributes. The average percent-deforestation values were then graphed as distance decay curves for each analysis year to identify variations in deforestation trends caused by socio-economic settlement differences.
Using circular buffers to determine a settlement's deforestation impact is problematic because the method does not account for natural barriers such as rivers, the overlapping effects of neighboring villages, and non-settlement caused deforestation. It also assumes that settlements cut the forest in their immediate vicinity. The buffer method and average deforestation distance decay curves were used to isolate deforestation patterns linked to settlements based on distance alone regardless of the lack of land ownership delineation and regardless of the "background noise" of deforestation caused by overlapping areas of settlement influences and access-ways. 31
• Settlements N Perennial Roads D MBR Boundaries Change-Detection lm age EJ Wetland ,L;;.:~;< J~~]~~·::::.c~~~ D Forest - Water .,, "'-~ ...... ~ . ~ D Deforested 1995-97 D Deforested 1993-95 D Deforested 1990-93 D Deforested 1986-90 D Deforested < 1986 D NoData 5 0 5 10 Kilometers 1111 I
Figure 5. Buffering the El Naranjo settlement by the San Pedro River and the surrounding area in 1997.
7.2.2 Road Analysis
Only paved and dirt roads that were passable all year long with regular pickup trucks or public buses were included in the analysis. Intermittent or temporary dirt roads were not included because of their unknown status or extent. Also, a well maintained stretch of the "ruta petrolera" from the town El Naranjo by the San Pedro river to the north was not included until 1991 because access was limited through a ferry controlled by oil companies (Basic Resources International Ltd). It was assumed that the entire study area was easily penetrated on foot, with mules, or four-wheel drive pickup trucks during most of the year, except during the rainy season in summer. Perennial roads, however, are significant to settlements because they allow for cheap transportation, such 32 as public buses, and easy access for mobile merchants with trucks who buy most of the crop.
The change-detection images showed deforestation occurring along the immediate vicinity of perennial roads. However, the buffering and deforestation distance decay curve methods used with the settlement analysis could not be applied to roads due to their linear nature. Distance decay curves could describe differently aged road segments but the segments' comparison to each other was thwarted by difficulties in creating equal area sample units.
7.2.3 Soil Analysis
The reclassified soil quality categories or polygons and the change-detection images were used to determine the relationship between deforestation trends and soil quality. This was accomplished by calculating the percent deforestation within the area of each soil quality polygon for 1986, 90, 93, 95, and 97 resulting in percent deforestation time-series curves for the four agricultural soil suitability categories.
7.2.4 Deforestation Probability Model
To create a descriptive deforestation probability surface or model, a raster-based or cell-by-cell logistic regression was calculated for each of the analysis years - 1986, 90
93, 95, 97, and 99. The dependent variable for each year was a binary forested/deforested ARC/INFO raster surface (grid) derived from the forest change- detection images. The independent variables were grids representing well/poorly 33
draining soils, natural log transformed Euclidean distances to the closest road, and natural
log transformed Euclidean distances to the closest settlement. Although the logistic
regression assumes the independence of the independent variables used, roads,
settlements, and soil quality form a complex interactive relationship. The logistic
regression analysis is used to describe and quantify that relationship to deforestation on a
landscape level.
The study area's large size and the change-detection images' high resolution of 30
by 30 meters resulted in grids with over 22.8 million cells or records. Due to hardware
and statistical software limitations, 5% stratified random samples (> 1,100,000 cells)
were used for the statistical analysis - where 5% of the forested cells and 5% of the deforested cells were sampled independently to be combined later (Appendix C). The stratification of the samples was necessary in order to ensure the approximately 1:25 ratio of deforested to forested cells. In addition, the y-intercepts or constants of the logistic regressions were corrected for unequal sample sizes (Warren 1990, Appendix C).
The independent variable grids were then multiplied by their respective regression coefficients and the corrected y-intercept was added as a constant. Finally, the sums of the weighted grids were logistically transformed to create probability surfaces with values between 0 (lowest deforestation probability) and 1 (highest deforestation probability) for each analysis year (Figure 6, Appendix C). The resulting continuous probability surface was reclassified into deforestation probability zones of 20 intervals for analysis and 10 intervals for display purposes. 34
Variables: 1997 Logistic regression Sum of Dependent coefficients; weighted Deforestation -> 10.006 grids 1997 intercept
Independent Weighted Grids; LN site logisticaily transform^ distance 1997 X -1.087 =
LN road X -0.430 = 1 distance 1997 Deforestation > Soil drains probability X 0.955 = well/poorly surface 1997/
Figure 6. Steps taken to create the 1997 descriptive deforestation probability surface and values used to weight the grids.
7.2.5 Forecasting Deforestation Risk
The deforestation risk models were based on the combination of a deforestation probability forecast - "Where?" - and a percent deforestation forecast - "How much?" - for the deforestation probability zones. The forecasts are essentially based on past observed deforestation patterns applied to future road and settlement conditions. Trial and error showed that, compared to trying to extrapolate future deforestation patterns from past observations, this simple forecasting approach provided the best results.
Weighting the binary soil grid and forecasted roads and settlements distance grids with the logistic regression coefficients, determined in the previous observation period, created the forecasted deforestation probability surfaces (Figure 7). The percent deforestation observed in each probability zone of the observation period's deforestation 35 probability surface was used to determine the forecasted deforestation area in the probability zones of the forecasted surface (Figure 8).
To test the deforestation risk model, a 1999 deforestation risk forecast based on
1997's regression coefficients and roads and settlements observed in 1999 was compared with the 1999 deforestation change-detection image. The 2001 forecast was based on regression coefficients from 1999, observed road and settlement changes up to May 2000, intermittent roads expected to be upgraded, and proposed perennial roads. No potential new settlement sites were included. The 2001 deforestation scenario was developed with the help of Roan Balas (WCS - Peten), Georg Griinberg (CARE Guatemala), and Victor
Hugo Ramos (CEMEC - CONAP).
Variables: 1997 Logistic regression Sum of Dependent coefficients: weigfited Deforestation Corrected y- > 10.006 ->i grids 1997 intercept
Figure 7. Steps taken to forecast the 1999 deforestation probability surface and values used to weight the grids. 36
Deforestation 1999 Forecasted 1997 Observed 1999 Forecasted Probability Probability Zone Percent Deforestation Zone: Area (km^): Deforestation: (km^); 1 - 0.95 o X 66% 221
0.95 - 0.90 X 50% = 249
/—X
0.90 - 0.85 (o) X 41 % 227
Figure 8. Forecasting the deforested area based on forecasted probability zones and observed deforestation. See Table 4 and Table 5 for the complete forecasted deforestation tables. 37
8 RESULTS
8.1 SETTLEMENTS AND DEFORESTATION
Most new settlements are established along perennial roads or other access ways.
Navigable rivers, such as the San Pedro River and parts of the Usumacinta River, and oil pipeline maintenance roads often act as alternative access ways. Some, mostly immigrant
Mayan settlements (Buen Samaritano 2, Chinatal, Corozal, El Mango, Formacion, La
Bacadilla, Armenia, Los Cerritos, Nueva, Paso Caballos, Sagrado Corazon, San Juan
Villanueva, Tienra Linda Zapotal), establish themselves relatively far away from existing roads. The settlements, however, seek to upgrade the paths or intermittent roads connecting them to the perennial road network within a few years after their establishment. In general, immigrant Ladino settlements are often strung along roads, while settlements of predominantly immigrant Mayan ethnicity settle away from the road-side communities.
The average deforestation distance decay curves for settlements in 1997 show that different socio-economic backgrounds manifest different deforestation behaviors (Figure
9 and Figure 10). The settlement buffering method to determine the distance decay curves does not account for neighboring settlements. Therefore, the likelihood of the deforestation to be caused by neighboring settlements increases with increasing distance from the center.
Up to the distance decay curve's inflection point one can be relatively sure that the curve's average percent deforestation can be mostly attributed to the settlements in 38 question. The percent deforestation values beyond the inflection point or at distances greater than a two hour walk - less than 10 km Euclidean distance - only represent the local areas' deforestation "background noise." The deforestation background noise level equals the baseline or average deforestation within the distance's area. Distance decay curves with higher deforestation baselines occur in areas with higher deforestation impact than those with low baselines.
The two hour walk distance limit is based on the willingness of Maya farmers to walk on average up to two hours to their fields (Norman Schwartz, personal communication). The distance threshold for Ladino farmers is on average a one hour walk. In the few cases where fields are at greater distances, temporary accommodations on site are established which may turn into a new settlement.
Within a minimum distance of 4 km, settlements of all economic occupations, except wage labor had on average a noticeable deforestation impact on their surroundings
(Figure 9). Wage labor settlements had little impact on the immediate forest and were subsequently removed from the deforestation probability models. Agriculture settlements and settlements in transition from agriculture to small-scale ranching appear to have the greatest impact on their surrounding environment. Small-scale ranching settlements appear to have a lesser impact. This may be due to the relative remoteness of ranching settlements, such as those along the San Pedro River, and therefore less overlapping deforestation with neighboring settlements compared to agricultural settlements.
Settlements relying on forestry, agroforestry, and forest products, such as xate and game, appear to have a relatively small impact up to a 3 km distance. 39
The deforestation distance decay curves of settlements grouped according to their major ethnicity showed slight differences in behavior (Figure 10). The immigrant Ladino settlements show their inflection point around 3 km while the immigrant Mayan settlement curve tapers off only gradually without any clear inflection point. This could be explained with the Mayan's willingness to walk on average up to two hours to their fields while Ladinos are often only willing to walk half that distance. The shorter deforestation distance of Ladino versus Mayan immigrant settlements, however, was not believed to be significant enough on a regional scale to be included in the deforestation model. All non-settlement sites were supported by wage labor and were consequently excluded from the models. In addition, Comunidades de Poblacion en Resistencia (CPR) guerrilla settlements in the Sierra del Lacandon National Park were not included in the model because of their minimal impact on the forest in order to maintain cover from the
Guatemalan armed forces. Since 1998, the CPR settlements moved to non-protected areas of Peten as part of the Guatemalan peace accord. 40
Average Deforestation of Settlements in 1997 - Primary Occupation
7 8 9 10 11 12 13 U 15 16 17 18 19 20 Distance to Settlement (km) -Agriculture (85 Samples) -Transition from Agriculture to Ranching (52 Samples) -Ranching (16 Samples) - Forestry etc. (13 Samples) -Wage Labor (9 Samples) ^ „
Figure 9. Deforestation distance decay curves of settlements according to primary economic occupation.
Average Deforestation of Settlements in 1997 - Ethnic Majority
8 9 10 11 12 13 14 15 16 17 18 19 20 Distance to Settlement (km) Peteneros - Ladinos and Itza" (7 Samples) B Non-Settlement Sites (7 Samples) Immigrant Ladinos (134 Samples) K Immigrant Mayas (25 Samples)
Figure 10. Deforestation distance decay curves of settlements according to their ethnic majority. 41
8.2 ROADS AND DEFORESTATION
Roads had a clear but relatively short ranging impact on deforestation compared to settlements. On a regional scale, perennial roads, compared to intermittent roads, appeared to have the greatest and most consistent impact on the forest. This could be explained by the greater density and age of settlements along perennial roads than on intermittent roads. The roads mostly represented access ways for new settlements and in fewer cases for logging and oil exploration. Preliminary logistic regressions showed that bus routes with regular public transportation were poor deforestation probability indicators compared to perennial roads. The bus routes were therefore disregarded in favor of perennial roads as an independent variable for the deforestation probability models.
In two cases, travel restrictions on perennial roads led to fewer settlements and less deforestation than expected. In one case settlement access to the Ruta Petrolera route north of El Naranjo was limited by a ferry across the San Pedro River controlled by oil exploration companies. The area north of El Naranjo remained relatively sparsely populated until 1991 when the ferry was opened to the public which led to nine new
Ladino settlements within three years (Bella Vista, Cnice a Santa Amelia, El Petenero, La
Ceiba, Laguna Vista Hermosa, Los Reyes, Los Tubes, Rancho Sucely, Valle Nuevo). In the second case, access to the route to Uaxactun - through Tikal and its archeological attractions - is controlled by guards. The guards by Tikal are known to check the heavy tourist traffic for possible immigrants. The travel restriction led most likely to the lack of 42 any new settlements north of Tikal. However, the areas scarcity of perennial potable
water sources may also be a factor.
8.3 SOIL QUALITY AND DEFORESTATION
The agricultural suitability soil map showed that well draining soil types were
more likely to be deforested than poorly draining soils regardless of their soil depth
(Figure 11). In addition, over time, the deforestation rate increased on well draining soils while it remained close to 0 on poorly draining soils (Figure 11). The soil variable for the probability model was simplified to the binary categories of well draining (1) and poorly draining (0) soils because of the observed deforestation trends and the soil map's poor resolution (Table 2).
Accumulated Deforestation vs. Soil Quality 25
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 Year " ^ - Well Draining Deep Soils Poorly Draining Deep Soils - ^ - Well Draining Shallow Soils • Poorly Draining Shallow Soils
Figure 11. Deforestation trends on various soil classes. 43
8.4 DEFORESTATION PROBABILITY
8.4.1 Logistic Regression Results
The logistic regression analysis showed that soil drainage and distances to roads and settlements were good predictors for forested cells but poor predictors for deforested cells (Table 3). The poor prediction of the deforested cells was affected by its rareness and small sample size compared to forested cells. Although the regressions' statistical significance improved over time, it remained to be of little significance (Nagelkerke R
Square = 0.282 to 0.375). Yet, the resulting deforestation probability surfaces reflected regional deforestation patterns quite well. Except for soil drainage, the regression coefficients change relatively little over time. The soil coefficients tenfold increase over time can be attributed to the increasing rate of deforestation observed in well draining soils (Figure 11). 44
Table 3. Logistic regression results for the deforestation probability model.
Analysis Year 1986 1990 1993 1995 1997 1999 5 % Sample Forested Cells 1,068,861 1.061,853 1,039,979 1.021,830 1.005.123 962,040 Size Deforested 40,338 47,280 69,648 87,471 103,934 112,259 Sum 1,109.199 1,109,133 1,109,627 1,109.301 1.109,057 1,074,299 Nagelkerke R Square 0.282 0.312 0.354 0.371 0.371 0.375 Hosmer & Chi-square 3,863.8 3,993.6 4.179.8 2.766.3 5,278.6 6,528.5 Lemeshow df 8 8 8 8 8 8 Test Significance >0.000 >0.000 >0.000 >0.000 >0.000 >0.000 Ll^ Road Distance -0.391 -0.441 -0.498 -0.503 -0.43 -0.408 Logistic S.E. 0.005 0.004 0.004 0.004 0.003 0.003 Regression LN Settlement Distance -1.005 -1.007 -1.060 -1.071 -1.087 -1.111 Coefficients S.E. 0.007 0.007 0.005 0.005 0.005 0.005 and their Soil Drainage 0.062 0.117 0.503 1.062 0.955 0.954 Standard S.E. 0.014 0.014 0.012 0.011 0.100 0.009 Errors y-lntercept 9.116 9.519 10.349 10.217 10.006 10.116 S.E. 0.051 0.049 0.045 0.041 0.038 0.037 Corrected v-lntercept 12.4021 12.6265 13.0546 12.6697 12.2751 12.264 Observed vs Forested % Correct 99.7 99.5 99.1 98.7 98.4 98.0 Predicted Deforested % Correct 7.0 9.5 16.1 20.0 21.6 22.8 (0.5 cutoff point) Overall % Correct 96.3 95.7 93.9 92.5 91.2 90.2
8.4.2 Deforestation Probability Model
As expected, the deforestation probability models did not predict the amount nor the exact location of observed deforestation, but were good indicators of deforestation distribution (Figure 12). Consistently for the analysis years, the higher the deforestation probability score, the higher the probability zone's observed deforestation rate (Figure 13 to Figure 24). This indicated that the chosen soil and distance variables captured the likelihood of deforestation. 45
Percent Deforestation Observed in the Probability Zones - 1986-99 70
"'C ~ 60 --+-1986 en ~ - 1990 .E 50 Q) 1993 0 -*- 1995 ~ 40 0 N --*- 1997 >. ~ 30 ---- 1999 :0 m -g 20 a..'- -0 10 ~0 0 - 10 N 10 10 "
Figure 12. Percent deforestation observed in low (0) to high (1) deforestation probability zones over time. 46
Settlements • 1820- 1986 Roads Unknown or in term itte nt N Perennial -1986 Deforestation
10 0 10 20 30 40 50 60 Kilometers ------
Figure 13. MBR' s settlements, roads, and cumulative deforestation in 1986.
Settlements .. 1820- 1986 Roads Unknown or in term itte nt N Perennial
10 0 10 20 30 40 50 60 70 Kilometers ------
Figure 14. Deforestation probability model for 1986. 47
Settlements • 1820- 1990 Roads Unknown or intermittent N Perennial ...... ···· -1990 Deforestation
'· "'I•' ...
...... ~. -.... \.,.:.
' ;/ 10------0 10 20 30 40 50 60 Kil~meters ' Figure 15. MBR's settlements, roads, and cumulative deforestation in 1990.
Settlements • 1820- 1990 Roads ··... · Unknown or intermittent N Perennial
·,......
10 0 10 20 30 40 50 60 70 Kilometers ------Figure 16. Deforestation probability model for 1990. 48
Settlements • 1820- 1993 Roads .. --.. Unknown or in term itte nt ·---~ ·...... ' N Perennial '••.t 1993 Deforestation
·,., __
·····;. \ ...
~ ... - - .. ____ ;:
·----...}
-: ... -t-~ ...-
10------0 10 20 30 40 50 60 Kilometers Figure 17. MBR' s settlements, roads, and cumulative deforestation in 1993.
Settlements • 1820- 1993 Roads -. , Unknown or in term itte nt N Perennial
10 0 10 20 30 40 50 60 70 Kilometers ------Figure 18. Deforestation probability model for 1993. 49
Settlements • 1820- 1995 Roads Unknown or interm itte nt N Perennial 1995 Deforestation
- '- - ·~. ··---=:.._
10-- -0 10- 20- 30- 40- 50- 60 Kilometer~ Figure 19. MBR' s settlements, roads, and cumulative deforestation in 1995.
Settlements • 1820- 1995 Roads .. 0 Unknown or intermittent N Perennial
10 0 10 20 30 40 50 60 70 Kilometers ------
Figure 20. Deforestation probability model for 1995. 50
Settlements • 1820- 1997 Roads Unknown or intermittent N Perennial 1997 Deforestation
- ... .. · ~ ...
10 0 10 20 30 40 50 60 Kilomete-rs .• ------"·· .. Figure 21. MBR' s settlements, roads, and cumulative deforestation in 1997.
Settlements • 1820- 1997 Roads Unknown or intermittent N Perennial
10------0 10 20 30 40 50 60 70 Kilometers
Figure 22. Deforestation probability model for 1997. 51
Settlements • 1820- 1999 Roads Unknown or intermittent N Perennial , 1999 Deforestation Deforestation Fire
10 0 10 20 30 40 50 60 Kilometers ------.,..._ .... . Figure 23. MBR' s settlements, roads, and cumulative deforestation in 1999 including man-made wildfires from summer 1998.
Settlements • 1820- 1999 Roads Unknown or intermittent N Perennial
-~."'....
10-- -0 10- 20--- 30- 40 50- 60 70 Kilometers Figure 24. Deforestation probability model for 1999. 52
8.5 DEFORESTATION RISK MODEL
8.5.1 Testing the 1999 Forecast
The forecasted 1999 deforestation probability surface, based on 1997's regression coefficients and observed roads and settlements of 1999 (Figure 7), is very similar to the
1999 probability surface based on observed deforestation (Figure 24 and Figure 25).
Although human-made wildfires in the 1997 to 1999 period - especially during the dry season of March, April, and May of 1998 - complicated a comparison, the distribution of observed and predicted deforestation area in 1999's predicted deforestation probability zones were similar to each other (Figure 26).
The deforestation prediction, based on multiplying the fraction of observed deforestation in the 1997 deforestation probability zones by the area of the projected
1999 deforestation probability zones, resulted in a close match to the observed deforestation in 1999 with an overall error of 0.49% or an overestimate of 9.9 km"
(Figure 26 and Table 4). The predicted deforestation was adjusted for the decrease of the
1999 observation area due to cloud-cover in the Sierra del Lacandon National Park area.
Lucidly, the missing area had a very low deforestation probability and the predicted deforestation was adjusted by subtracting 9 km" of deforestation from the unadjusted prediction.
The deforestation forecast was weakest in the low probability zones with an error up to 187%. It performed better in the high probability zones with an error ranging from
-9% to +8%. When comparing the difference in square kilometers deforested however. 53
2 low probability zones had smaller differences, between 1 and 8 km , while high
2 probability zones had larger differences, between 2 and 15 km . The sudden drop in the
area deforested in the 2 highest probability zones are due to the small area of those zones
(Figure 26). The probability zones' areas decrease with increasing deforestation
probability, because of the use of natural log transformed distance variables.
··. ,. r------, Settlements • 1820- 1999 Roads .··· .... · Unknown or ·· intermittent N Perennial
1999 Deforestation probability forecast ...... 0- 10 50- 60 c::J 10 - 20 60 - 70 0 20 - 30 70 - 80 D 30 - 40 80 - 90 10-- --0 10 20- 30- 40-- 50- 60 70 Kilometers c::J 40 - 50 90 - 1DO Figure 25. Forecasted deforestation probability model for 1999. 54
1999 Deforestation Risk Model - Predicted vs. Observed Deforestation
1999 Predicted Deforestation 250 .....,._ 1999 Observed Deforestation - -+ --1998 Fire Impact Areas
' • •' ~ ' ~ . •• L() N L() (") L() ' Figure 26. Observed vs. predicted deforestation area of the 1999 deforestation risk model. The predicted deforestation was adjusted for the decrease of the 1999 observation area due to cloud-cover in the Sierra del Lacand6n National Park area. Most of the man caused fires occurred during the dry season of March, April, and May, 1998. 55 Table 4. Differences between predicted and observed deforestation for 1999. The predicted deforestation was adjusted for the decrease of the 1999 observation area due to cloud-cover in the Sierra del Lacandon National Park area. Deforestation % Forecasted 1999 Deforestation (km^) Difference % Probability deforested Probability Surface (km^) Difference Zone in 1997 Area (km^) Predicted Observed 0-0.5 0.39 X 2562.9 — 9.9 7.2 2.7 38.01 0.05 - 0.1 0.14 X 3463.9 — 4.9 1.7 3.2 186.59 = 0.1 - 0.15 0.54 X 1839.8 10.0 11.2 -1.2 -10.40 = 0.15-0.2 1.33 X 1474.7 19.7 21.1 -1.4 -6.72 = 0.2 - 0.25 2.12 X 1240.7 26.3 28.7 -2.3 -8.16 0.25 - 0.3 2.15 X 1025.7 — 22.1 26.3 -4.2 -16.10 0.3 - 0.35 2.76 X 882.4 — 24.3 31.9 -7.6 -23.81 = 0.35 - 0.4 5.38 X 788.2 42.4 45.2 -2.7 -6.03 0.4 - 0.45 9.05 X 749.8 = 67.8 71.9 -4.1 -5.71 0.45 - 0.5 11.06 X 693.1 = 76.7 81.6 -4.9 -6.05 = 0.5 - 0.55 12.94 X 680.3 88.1 96.9 -8.9 -9.16 0.55 - 0.6 16.01 X 697.7 = 111.7 121.5 -9.8 -8.08 0.6 - 0.65 18.22 X 698.1 127.2 137.8 -10.6 -7.71 0.65 - 0.7 20.29 X 682.8 — 138.5 146.9 -8.4 -5.71 = 0.7 - 0.75 25.40 X 655.6 166.5 168.5 -2.0 -1.20 = 0.75 - 0.8 30.30 X 620.3 187.9 181.1 6.8 3.77 = 0.8 - 0.85 35.09 X 596.1 209.2 201.0 8.2 4.06 0.85 - 0.9 40.52 X 559.3 = 226.6 214.3 12.3 5.76 0.9 - 0.95 50.10 X 497.0 = 249.0 235.2 13.8 5.88 = 0.95 - 1 66.37 X 333.6 221.4 206.0 15.5 7.51 2030.4 2036.1 9.9 0.49 8.5.2 The 2001 Deforestation Scenario The 2001 forecasted deforestation probability surface is based on the regression coefficients of the 1999 observed deforestation probability surface, known settlements until 1999, and a 2001 [>erennial roads scenario. All but 2 road stretches in the scenario already existed as intermittent roads or pipeline maintenance paths which have been upgraded or will be by the end of the year 2000 (Victor Hugo Ramos and Georg Grunberg 2000, personal communication). The two exceptions are proposed roads. One may go from Carmelita in the Center of the MBR to the El Mirador archeological site in the El Mirador National Park to the north (Victor Hugo Ramos 2(XX), personal 56 communication). The Xpujil-Tikal highway, a paved road, may go from Tikal through the Rio Azul National Park to Mexico (Roan Balas and Victor Hugo Ramos 2000, personal communication). The 2001 scenario deforestation forecast is based on the percent deforestation observed in the 1999 deforestation probability surface. The 2001 scenario forecasts a deforestation increase of 295 km2 or a 14.5% increase over the deforestation observed in 1999 (Table 4). Most of the additional deforestation is predicted to be in the high deforestation probability zones. ~his follows the trend of previous years where most of the additional deforestation has been observed in the high probability zones (Figure 28). Settlements • 1820- 1999 Roads Unknown or intermittent N Perennial Estimate 2001 Deforestation probability forecast o- 10 D 50-60 c:J 10- 20 CJ 60- 70 D 20 - 30 70 - 80 D 30 - 40 80 - 90 10------0 10 20 30 40 50 60 70 Kilometers c:J40- 50 90- 100 Figure 27. Forecasted deforestation probability model for 2001. 57 Table 5. Differences between the deforestation predicted by the 2001 scenario and observed deforestation for 1999. Deforestation Deforestation (km^) Difference % Probability Observed Predicted (km^) Difference Zone 1999 2001 0-0.5 7.2 4.0 -3.2 -44.10 0.05-0.1 1.7 1.8 0.1 3.92 0.1 -0.15 11.2 13.2 2.1 18.43 0.15-0.2 21.1 23.4 2.4 11.16 0.2 - 0.25 28.7 29.9 1.3 4.40 0.25 - 0.3 26.3 28.3 1.9 7.34 0.3 - 0.35 31.9 34.8 2.8 8.89 0.35 - 0.4 45.2 49.5 4.3 9.52 0.4 - 0.45 71.9 77.3 5.4 7.45 0.45 - 0.5 81.6 87.1 5.5 6.77 0.5 - 0.55 96.9 101.3 4.4 4.52 0.55 - 0.6 121.5 126.1 4.6 3.81 0.6 - 0.65 137.8 145.4 7.6 5.53 0.65 - 0.7 146.9 159.8 12.9 8.78 0.7 - 0.75 168.5 191.1 22.5 13.37 0.75 - 0.8 181.1 218.9 37.8 20.90 0.8 - 0.85 201.0 244.0 43.0 21.39 0.85 - 0.9 214.3 259.0 44.7 20.87 0.9 - 0.95 235.2 285.1 49.9 21.23 0.95 -1 206.0 251.1 45.2 21.93 2036.1 2331.4 295.3 14.50 58 Observed Deforestation and the 2001 Scenario Prediction 300 -- .. -- 1986 250 ~------~~----~ N' -- .. -- 1990 < ~ 200 ·- 1993~------~~~~-- co - -~- 1995 ~ 0 ,...._ I{) I{) N I{) ('t) I{) ...;t I{) I{) I{) Figure 28. 2001 deforestation scenario forecast compared to observed deforestation. 59 9 DISCUSSION 9.1 MODELING DEFORESTATION Several studies (Chomitz and Gray 1996, Mas et al. 1997, Mertens and Lambin 1997) have shown a clear relationship between the presence of roads and human settlements, and deforestation. As expected, the results of this study support these findings. This study shows that soil quality and natural log transformed distances for roads and settlements are good indicators for deforestation trends in an agricultural frontier such as the Peten. 9.2 STRENGTHS AND LIMITATIONS OF THE DEFORESTATION RISK MODEL The deforestation risk model does not predict deforestation locations per se but can be used to estimate the possible impact of new roads and settlements. For example, it can be used as an environmental impact assessment tool for a proposed road through the MBR to Mexico as shown in the 2001 scenario. Its greatest strengths are its simplicity as a general model with good results and that it uses common spatial features such as roads, settlement points, and simple soil maps. The existence of adequate census data might have improved the deforestation risk model significantly but their lack did not hinder it - which is advantageous for modeling in frontier regions which often lack census data. Once the coefficients are calculated, a deforestation probability or risk model can be easily calculated using map algebra on common spatial features. As a general model, it may prove to be useful not only for northern Peten but also for other agricultural frontiers. 60 The deforestation model does not, however, account for spatial or temporal autocorrelation nor the age of roads and settlements. The analysis methods used in this study may have been successful on a regional scale, but better methods that combine spatial and temporal analysis are needed for modeling deforestation on a local level. For example, such an analysis method would need to account for roads which have segments of different ages that "grow" over time and it would need to quantify the overlapping effect of those road segments. Panel data or pooled time data analysis combined with a stochastic model may be more adequate methods. Also, any forecasts beyond two years may prove to be questionable because the model does not account for changes in the peoples' behavior. Eventually, the subsistence settlements' economy will grow strong enough to afford other means of income such as ranching, wage labor, and tourism. 9.3 MANAGEMENT IMPLICATIONS AND RECOMMENDATIONS This study led to several suggestions regarding reducing deforestation risks. The following recommendations are based solely on the goal of reducing deforestation. There are, of course, many social and political reasons why deforestation might be appropriate in spite of its impact on biodiversity and natural resources. As long as immigration is talcing place, access to the MBR needs to be controlled to avoid any new settlements in undesired areas. The easiest way of controlling access appears to be by controlling road passages, such as in the case of the ferry by the ruta petrolera and the Tikal check station to Uaxactun. Another significant form of access control is by avoiding building perennial roads or upgrading existing intermittent roads to 61 a perennial status. This, of course, may lead to a diminished standard of living for settlements relying on those roads. Besides roads, pipelines and rivers need to be considered as possible access routes and hence precursors of deforestation. To avoid habitat fragmentation, any new settlements and roads should be avoided in low deforestation risk areas. The settlement analysis also strongly suggests supporting forest resources or wage-labor based economy because of their lesser and relatively shorter ranged impact on the forest. Finally, in an agricultural frontier, regional deforestation trends are not only controlled by access, but also by agricultural considerations such as soil quality. With the increasing importance of cattle ranching, water sources may become more important than soil quality. 9.4 FUTURE ANALYSIS The study's analysis has room for improvement. The model needs to be tailored to and applied in scenarios of regional significance, such as that of the proposed road to Mexico. Spatial and temporal autocorrelation need to be accounted for. Settlements and road age could be included in the form of modified nearest neighbor analysis to create a years of roads or settlement influence surface. Settlement ethnicity and their land tenure status may be of significance in modeling deforestation. Water availability for ranching and agriculture could be incorporated into the regression analysis. River traffic and oil- pipelines need to be considered as access routes in the model and settlements may be differentiated according to their socio-economic qualities. A stochastic component could be incorporated into the model for better visualization of deforestation risk. Such a 62 stochastic model assigns a forested cell to be deforested according to its deforestation probability and according to the existence of neighboring deforested cells to account for spatial autocorrelation. The model may also have uses for estimating areas of high wildfire risks. Finally, slope, aspect, and elevation data of adequate resolution in combination with better soil maps may turn this regional model into a more localized version. 63 10 APPENDIX A - SETTLEMENT DATABASE Settlements in the Maya Biosphere Reserve up to 1999 and their Attributes According to Grunberg and Ramos (1998, 2000 personal communication) Key to Ethnic Majorities Key to Primary Occupations Key to MBR Management Units IL Immigrant Ladino Ag Agriculture Ldt NP Laguna del Tigre National Park Itza' Itza' Maya FR Forest Resources LdT Re Laguna del Tigre Reserve OM Other immigrant Mayas Ra Small-scale Ranching SdL NP Sierra del Lacandon National Park PL Peteneros Ladinos Tr Transition from Ag to Ra Tikal NP Tikal National park Qe Q'eqchi" WL Wage Labor ZUM Multiple-use Zone ZAM Buffer Zone O-MBR Outside of MBR Census Primary Manage Code Settlement Name Alternative Name Year of Ethnic Occupa ment Note Inception Count Year Majority tion Unit SA0051 Arroyo Chocop Los Cemtos 1992 138 1998 IL Ra. LdTNP SA0037 Bella Vista 1993 402 1997 IL Tr. LdTNP SA0039 Campamento Xan Campiamento Basic 1986 100 1997 - WL LdTNP Oil Camp SA0027 Cruce a Santa 1993 825 1997 IL Tr. LdTNP Amelia SA0042 El Mirador 1991 193 1998 IL Ra. LdTNP SA0033 El Petenero Rt'o Escondido 1993 161 1997 IL Tr. LdTNP SA0041 EI Zacatal 1998 4 1998 - WL LdTNP CONAP SA0044 Guayacan 1996 10 1997 - WL LdTNP CONAP SA0034 La Casuala 1994 99 1998 IL& Oe Ag. LdTNP SA0045 La Ceiba 1993 66 1997 IL Tr. LdTNP SA0049 La Profundidad 1983 39 1998 IL Tr. LdTNP SA0032 Laguna Vista 1994 253 1997 IL Ra. LdTNP Hermosa SA0036 Las Guacamayas Estacion Biologica 1995 4 1998 WL LdTNP ProPeten SA0028 Los Tres Reyes Los Reyes 1994 165 1997 IL Ag. LdTNP SA0029 Los T utx» 1994 224 1997 IL Tr. LdTNP SA0035 Paso Catiallos 1994 525 1996 Ag. LdTNP SA0038 Rancho Sucely 1994 28 1998 IL Ra. LdTNP SA0043 Valle Nuevo La Mancomadora 1993 33 1997 IL Tr. LdTNP SA0031 Buenos Aires 1994 150 1997 IL Ag. LdT Res. SA0048 Parcelamiento Rio 1989 61 1998 IL Ra. LdT Res. Escondido LL0097 Alt^eno 1992 110 1998 IL Ag. SdLNP CPR LL0076 Argueta 1992 11 1997 IL Ag. SdL NP CPR LL0016 Centro Campesino 1 1992 55 1996 IL Ag. SdL NP LL0077 El Oesempeno El Aguacate 1993 28 1997 IL Ag. SdLNP CPR LL0008 El Esfuerzo 1986 686 1997 IL Ag. SdL NP LL0080 El Pital 1995 17 1998 IL Ag. SdL NP LL0091 El Repasto 1989 10 1998 IL Ag. SdL NP LL0011 El Retalteco 1978 700 1996 IL Ag. SdLNP LL0078 Esmeralda 1992 440 1998 IL Ag. SdLNP CPR LL0075 Estacibn Aforo 1996 15 1997 WL SdL NP CONAP LL0082 Fajardo 1992 165 1997 IL Ag. SdLNP CPR LL0064 Guayacan 1986 44 1998 IL Tr. SdL NP LL0012 La Lucha 1990 218 1997 IL Ag. SdL NP LL0065 Las Estacas Guayacan Centro 1986 220 1998 IL Tr. SdL NP LL0094 Las Victorias 1991 176 1998 IL Tr. SdLNP LL0067 Manantialito El Manantial 1996 77 1996 IL Tr. SdLNP LL0066 Nueva Jerusalen II 1985 385 1997 IL Tr. SdL NP LL0041 Nuevo San Jose Km 96 1985 330 1997 IL Ag. SdL NP LL0069 Poza Azul 1992 314 1997 IL&Oe Ag. SdLNP 64 Settiements in the Maya Biosphere Reserve up to 1999 and their Attributes According to Grunberg and Ramos (1998, 2000 personal communication) Key to Ethnic Maiorities Key to Primary Occupations Key to MBR Management Units IL Immigrant Ladino Ag Agriculture Ldt NP Laguna del Tigre National Park Itza' Itza' Maya FR Forest Resources LdT Re Laguna del Tigre Reserve OM Other immigrant Mayas Ra Small-scale Ranching SdL NP Sierra del Lacandon National Park PL Peteneros Ladinos Tr Transition from Ag to Ra Tikal NP Tikal National park Qe Q'eqchi" WL Wage Lat)or ZUM Multiple-use Zone ZAM Buffer Zone O-MBR Outside of MBR Census Primary Manage Code Settlement Name Alternative Name Year of Ethnic Occupa ment Note Inception Count Year Majority tion Unit LL0070 San Juan Villanueva 1987 578 1997 Oe Ag. SdL NP LL0043 Santa Amelia Km 101 1985 638 1997 IL Tr. SdL NP LLOOOO Tierra Colorada Se kaq'i ch'och" 1999 - - Qe Ag. SdL NP LL0013 Union Maya Itza' La Quetzal 1995 1150 1997 OM Ag. SdLNP LL0068 Villa Hermosa ElJutal 1987 583 1998 IL Tr. SdL NP LL0081 Virgilio 1992 110 1998 IL Ag. SdLNP CPR FLOODS Tikal 1948 50 1997 - WL Tikal NP IDAEH SA0023 Carmelita 1940 355 1997 PL FR ZUM FL0012 Champoxte EI Ramonal 1992 138 1998 IL Ag. ZUM SJ0005 Corozal 1982 280 1996 IL&Qe Ag. ZUM SAG019 Corozal Pasadita 1980 146 1997 IL&Qe FR ZUM SA0022 Cruce a la Colorada 1985 468 1997 IL Ag. ZUM SA0020 Cruce a la Naranjita 1992 28 1997 IL Ag. ZUM SA0025 Cruce a Pescaditos 1992 83 1997 IL Ag. ZUM SAGO13 Cruce dos Aguadas 1976 1216 1997 IL&Qe Ag. ZUM LL0051 El Ceibo 1985 40 1997 IL Ra. ZUM FLOOOO El Ramonal 1999 - - IL Ag. ZUM FL0012 La Bendicidn Champoxte 1992 - - IL Ag. ZUM SA0047 La Caot^a 1992 83 1998 IL Ra. ZUM SA0026 La Colorada 1988 121 1998 IL Tr. ZUM MM0027 La Coloradita Paso de Julio 1994 17 1998 IL Ag. ZUM FL0015 La Maquina 1962 470 1997 IL Tr. ZUM SAGO16 La Milpa 1990 42 1997 PL FR ZUM SA0018 La Pasadita 1980 394 1997 IL FR ZUM FL0017 La Pochitoca 1995 39 1997 IL Ag. ZUM MM0G09 La Polvora 1966 700 1997 IL Ra. ZUM SA0046 La Tuberia 1997 22 1998 - WL ZUM Oil Camp MM0010 Melchor de Mencos Fallatxin 1962 11315 1997 IL Ag. ZUM FLGGGG Nakum 1996 - - IL WL ZUM FLGOOO Paso del Carmen 1995 - - - WL ZUM CPR SAG040 San Luis Frontera 1992 165 1997 IL Tr. ZUM SA0015 San Miguel la 1975 151 1997 PL FR ZUM Palotada SA0030 Santa Rosita 1976 330 1998 IL Ra. ZUM MM0020 Santa Teresita la 1982 100 1997 IL Tr. ZUM Zarca SA0021 Sibal 1992 15 1997 IL FR ZUM FLG007 Uaxactun 1931 693 1997 PL FR ZUM SAGO17 Yarche 1992 18 1997 PL FR ZUM FLOG16 Yaxha 1990 149 1997 IL FR ZUM LLOOOO 17 de Abril 1998 - - IL Ag. ZAM FLG013 Aguada Nueva 1981 198 1998 IL Tr. ZAM LL0004 Amienia 1992 55 1997 Qe Ag. ZAM MM0G01 Bajo del Venado 1993 105 1995 IL Ra. ZAM LLGGG7 Betliania 1981 533 1996 IL Tr. ZAM 65 Settlements in the Maya Biosphere Reserve up to 1999 and their Attributes According to Grunberg and Ramos (1998,2000 personal communication) Key to Ethnic Majorities Key to Primary Occupations Key to MBR Management Units IL Immigrant Ladino Ag Agriculture Ldt NP Laguna del Tigre National Park Itza' itza' Maya FR Forest Resources LdT Re Laguna del Tigre Reserve OM Other immigrant Mayas Ra Small-scale Ranching SdL NP Sierra del Lacandon National Park PL Peteneros Ladinos Tr Transition from Ag to Ra Tikal NP Tikal National park Qe Q'eqchi" WL Wage Latjor ZUM Multiple-use Zone ZAM Buffer Zone O-MBR Outside of MBR Census Primary Manage Code Settlement Name Alternative Name Year of Ethnic Occupa ment Note Inception Count Year Majority tion Unit LL0010 Bethel 1968 441 1997 IL Tr. ZAM LL0090 Campamento 1997 50 1998 IL WL ZAM Chocop MM0026 Casa de Piedra 1975 33 1998 IL Ra. ZAM SA0001 Centro Campesino II 1985 439 1997 IL&OM Ag. ZAM LL0034 Chinatal 1987 100 1997 Qe Ag. ZAM MM0012 Cidabenque 1968 150 1997 IL Tr. ZAM LL0035 Corozal 1989 440 1997 Qe Ag. ZAM LL0095 Cruce los Esclavos 1986 55 1998 IL Tr. ZAM SA0024 Cruce Perdido 1980 222 1997 IL Tr. ZAM MM0013 Cruzadero 1967 2000 1996 IL Ra. ZAM LL0059 El Aguacate 1990 440 1997 IL Tr. ZAM SA0009 El Aguacate 1993 163 1997 IL Ag. ZAM SA0007 El Almendro 1988 165 1997 OM Ag. ZAM LLOOOO El Buen Samaritano 1990 - - Qe Ag- ZAM MM0005 El Camalote 1986 110 1997 IL Ra. ZAM MM0022 El Camalote 1975 94 1998 IL Ra. ZAM FL0003 El Caot>a 1960 1160 1996 IL FR ZAM FL0002 El Capulinar 1970 165 1997 Qe&OM Ag. ZAM LL0032 El Esqueleto 1985 242 1997 IL Ag. ZAM SA0008 El Jobo 1984 102 1997 IL Ag. ZAM LL0042 El Manantial 1986 479 1997 IL Ag. ZAM LL0058 EI Mango 1993 501 1998 Qe Ag. ZAM LL0063 El Matrimonio Eden 1988 143 1997 IL Ag. ZAM FL0008 El Naranjo 1962 1300 1996 IL Tr. ZAM LL0049 El Naranjo 1981 3500 1996 IL Tr. ZAM MM0016 El Naranjo 1981 600 1996 IL Ra. ZAM LL0044 El Nuevo Parai'so Km 107 1977 1760 1997 IL Tr. ZAM FL0004 El Porvenir 1987 300 1997 - FR ZAM FLOOD1 El Remate 1937 1100 1997 IL Ag. ZAM LL0087 El Remo 1994 39 1997 IL Ra. ZAM MM0023 El Rondon 1972 275 1998 IL Ra. ZAM LL0061 El Sinai 1992 121 1997 IL Ag. ZAM LL0054 El Tamtxj 1992 99 1997 Qe Ag. ZAM LL0057 El Tuxpan 1996 33 1997 IL Tr. ZAM FL0009 El Zapote 1962 755 1997 IL Tr. ZAM LL0022 Guadalupe 1994 66 1997 IL Ag. ZAM SA0014 Ixhuacut 1960 385 1997 IL Tr. ZAM SJ00C4 Jobompiche 1958 880 1997 IL&Oe Ag. ZAM LL0045 Km 114 1989 60 1997 IL Ag. ZAM SA0002 La Bacadilla 1990 148 1997 IL Ag. ZAM MM0006 La Blanca 1963 1200 1997 IL Ra. ZAM LL0031 La Bomtia 1988 605 1996 IL Tr. ZAM LL0023 La Caoba 1990 440 1997 Qe Ag. ZAM LL0027 La Casaca Altamira 1985 1359 1997 IL Tr. ZAM 66 Settlements in the Maya Biosphere Reserve up to 1999 and their Attributes According to Grunberg and Ramos (1998,2000 personal communication) Key to Ethnic Majorities Key to Primary Occupations Key to MBR Management Units IL Immigrant Ladlno Ag Agriculture Ldt NP Laguna del Tigre National Park Itza' Itza' Maya FR Forest Resources LdT Re Laguna del Tigre Reserve OM Other immigrant Mayas Ra Small-scale Ranching SdL NP Sierra del Lacandon National Park PL Peteneros Ladinos Tr Transition from Ag to Ra Tikal NP Tikal National park Qe Q'eqchi" WL Wage Labor ZUM Multiple-use Zone ZAM Buffer Zone O-MBR Outside of MBR Census Primary Manage- Code Settlement Name Alternative Name Year of Ethnic Occupa ment Inception Count Year Majority tion Unit SA0006 La Ceibita 1985 186 1997 IL&Qe Ag. ZAM LL0060 La Esperancita 1996 121 1998 Qe Ag. ZAM LL0014 La Felicidad 1968 102 1994 IL Ag. ZAM LL0056 La Isia 1995 11 1997 IL Ag. ZAM LL0048 La Jicotea 1986 220 1997 IL Ra. ZAM SA0011 La Juventud 1971 165 1997 IL Ag. ZAM LL0085 La Leona 1994 15 1997 IL Ra. ZAM LL0030 La Limonada El Rlin 1994 132 1997 IL Ag. ZAM U.0093 La Uorona 1980 138 1998 IL Tr. ZAM LL0019 La Nueva Sur de San Diego 1997 - - Qe Ag. ZAM LL0033 La Nueva Km 75 1983 237 1997 IL Ag. ZAM Candelaria LL0053 La Pata de Faisan 1995 22 1997 IL Ag. ZAM LL0029 La Pista Nuevo Paraiso 1987 748 1997 IL Ag. ZAM LL0019 La Poza Maya Sur de San Diego 1997 297 1997 Qe Ag. ZAM MM0002 La Puerta del Cielo 1977 72 1997 IL&OM Tr. ZAM LL0015 La Tunica 1990 231 1997 IL Ag. ZAM Agropecuaria LL0052 La Tortuga La Isia de Gongora 1996 44 1997 IL Ag. ZAM FL0021 La Union 1988 110 1998 IL Ra. ZAM LL0047 Lagunitas Km 114 1981 1683 1996 IL Tr. ZAM LL0038 Las Flores Km 86 1987 605 1997 IL Tr. ZAM LL0025 Las Marias Santa Maria 1990 523 1997 IL Tr. ZAM LL0037 Las Ruinas 1982 1750 1996 IL&Qe Tr. ZAM FL0010 Las Vinas 1970 800 1996 IL Ra. ZAM LL0028 Los Angeles 1989 413 1997 IL Tr. ZAM LL0084 Los Cerritos 1990 330 1997 Qe Ag. ZAM LL0055 Los Chicos 1994 28 1997 IL Ag. ZAM MM0024 Los Encuentros 1973 275 1998 IL Tr. ZAM MM0007 Los Lagartos 1985 121 1997 IL Ag. ZAM LL0002 Los Manueles 1976 385 1997 IL Tr. ZAM FL0014 Los Tulipanes 1970 150 1994 IL Ag. ZAM LL0009 Monte Sinai 1987 111 1996 IL Ag. ZAM FL0022 Monterrico 1973 468 1998 OM Tr. ZAM LL0086 Nueva Florida 1994 66 1997 IL Ag. ZAM LL0062 Nueva Jerusalen 1 Naranjita 1987 165 1997 IL Ag. ZAM LL0036 Nuevo Canaan Km 77 1987 99 1997 Qe Ag. ZAM LL0039 Nuevo Eden Km 91 1984 413 1997 IL Tr. ZAM SJ0002 Nuevo San Jose 1978 433 1996 IL Tr. ZAM LL0026 Poza del Macho Km 63 1985 1100 1996 IL Tr. ZAM LL0083 Rio Taman's Las Vegas 1989 770 1997 Qe Ag. ZAM LL0040 Sagrado Corazbn 1985 644 1997 Qe Ag. ZAM MM0014 Salpet 1967 150 1997 IL Ra. ZAM MM0003 Salsipuedes 1992 242 1995 IL& OM Ra. ZAM LL0020 San Diego Guatelinda 1979 1073 1997 IL Tr. ZAM 67 Settlements in the Maya Biosphere Reserve up to 1999 and their Attributes According to Grunberg and Ramos (1998, 2000 personal communication) Key to Ethnic Majorities Key to Primary Occupations Key to MBR Manaejement Units IL Immigrant Ladino Ag Agriculture Ldt NP Laguna del Tigre National Park Itza' Itza' Maya FR Forest Resources LdT Re Laguna del Tigre Reserve OM Other immigrant Mayas Ra Small-scale Ranching SdL NP Sierra del Lacandon National Park PL Peteneros Ladinos Tr Transition from Ag to Ra Tikal NP Ttkal National park Oa Q'eqchi' WL Wage Labor ZUM Multiple-use Zone ZAM Buffer Zone O-MBR Outside of MBR Census Primary Manage Code Settlement Name Altematlve Name Year of Ethnic Occupa ment Inception Count Year Majority tion Unit SA0003 San Jorge 1985 39S 1997 IL Ag. ZAM LL0021 San Jose El Triunfo 1978 984 1996 IL Ag. ZAM LL0050 San Julian 1991 94 1997 Qe Ag. ZAM SJ0003 San Pedro 1960 450 1997 IL&Oe Ag. ZAM MM0025 Santa Rosa 1978 160 1998 IL Tr. ZAM Chiqulbul LL0046 Santiaguito 1989 523 1996 OM Ag. ZAM LL0024 Tierra Linda Zapotal 1991 204 1997 Qe Ag. ZAM MM0015 Valle Nuevo 1994 400 1997 IL Ra. ZAM LL0005 Vista Hermosa LosChorros 1977 1430 1997 IL Tr. ZAM MM0004 Yaltutu 1981 572 1996 IL Ra. ZAM LL0088 Yanaf 1968 109 1994 IL&Qe Ag. ZAM FL0005 Zocotzal 1987 285 1995 IL FR ZAM MM0018 Arroyo del Guarda 1982 7 1998 IL Ag. O-MBR LL0006 Bonanza 1991 160 1997 IL Ag. O-MBR MM0011 El Arenal 1932 1000 1997 IL Ag. O-MBR SA0010 El Bayalito 1986 148 1997 IL& Oe Ag. O-MBR SA0004 El Habanero 1989 197 1997 IL Ag. O-MBR FL0018 El Limbn 1996 88 1998 Qe Ag. O-MBR FL0019 Ixlii 1980 1127 1997 IL Tr. O-MBR LL0003 La Bacadllla 1991 94 1997 Qe Ag. O-MBR LL0017 La Gloria 1978 1700 1997 IL Tr. O-MBR SA0050 Laguna Perdida La Union 1998 220 1998 IL Ag. O-MBR FL0020 Macanche 1952 1250 1996 IL Ag. O-MBR LL0018 Nueva Formacion 1986 314 1997 Qe Ag. O-MBR LL0001 Palestina 1976 2200 1996 IL Tr. O-MBR MMOOOa Pichelito 1 1983 110 1997 IL Tr. O-MBR MM0021 Pichellto II 1983 220 1998 IL Tr. O-MBR SA0005 Rey Balantun 1979 118 1997 IL Ag. O-MBR SA0012 San Andres 1820 3517 1997 PL FR O-MBR SJ0001 San Jose 1851 1042 1996 Itza' FR O-MBR MM0019 Santa Rosa la Zarca 1976 90 1996 IL Tr. O-MBR MM0017 Tikallto 1984 467 1997 IL Ra. O-MBR ~ ~ Ye ar of In cep ti on 0 1985 and be fore > 0 1986- 1994 ""d 0 1995-1 999 Pe ren nial Roads ~ Yea r of Incepti on z N 198 5 and before ~ ~ /\; 1986-1994 >< _, /\:' 1995- 1999 eo 00 Biosphere Re serve trj Natio nal Parks & Reserv es ZUM ~ ZAM ~ ~ trj z~ ~ 00 > ~ ~ 0 > \ ~ .,"' 00 ~ 10 '··· ,:20 Kilometers ~:n 1 ~\ o . , ! ·~<~ ~ ··.,,' ·· ··\ . ~ (. \ \ , ""d Settlements and Roads of the Maya Biosphere Reserve, Guatemala 0\ 00 69 12 APPENDIX C - ARC/INFO COMMANDS 1. Statement used to create a stratified 5% random sample grid of the study area's deforestation grid: In ARC/INFO's GRID module: out_gfrid = con (in crrid eq 1, con(rand (.) < 0.05, 1, in crrid eq 0, con(rand (.) <0.05, 0) Where 0.05 is the proportion of the total number of cells in the grid that are desired as deforested (1) or forested (0) cells. 2. Formula for calculating the corrected y-intercept or the constant regression coefficient. Correcting the y-intercept was necessary to account for the unequal sample sizes between deforested (1) and forested (0) cells. a' = a + In (n2/nl) Where a is the y-intercept in the regression, a' is the corrected y-intercept. In is a natural logarithm, nl is the number of cases in the smaller sample (deforested cells), and n2 is the number of cases in the larger sample (forested cells). This equation follows Warren (1990). 3. Formula for the logistic transformation of the probability surfaces: In ARC/TNFO's GRID module: docell tenpl:= e:^ (-1 * in grid) out grid = 1/(1 + tenpl) end Where in grid is the sum of the corrected y-intercept and the dependent variable grids weighted by the regression coefficients and where out grid is the resulting probability surface with a range of values between I (high probability) and 0 (low probability). 70 13 WORK CITED Apan, Armando A. and James A. Peterson. 1998. Probing tropical deforestation - the use of GIS and statistical analysis of georeferenced data. Applied Geography. 18(2): 137-152. APESA/AHT. 1991. Plan de Desarrollo Integrado de Peten - Volumen 1: Diagnostico general de Peten. APESA, Guatemala. Atran, Scott. 1993. Itza Maya tropical agro-forestry. Current Anthropology. 34(5):633- 700. Barrow, C. J. 1991. Land Degradation - Development and Breakdown of Terrestrial Environments. Cambridge University Press, Cambridge. 295pp. Behrens, Clifford A. 1996. A Formal Justification for the Application of GIS to the Cultural Ecological Analysis of Land-Use Intensification and Deforestation in the Amazon p55-77 in Anthropology, Space, and Geographic Information Systems. Oxford University Press, Oxford. 294pp. Chomitz, Kenneth M. and David A. Gray. 1996. Roads, land, markets and deforestation - a spatial model of land use in Belize. Environmental, Infrastructure, and Agriculture Division Policy Research Department, PERDEI, The World Bank. 50pp. Chou, Yue-Hong. 1997. Exploring Spatial Analysis in Geographic Information Systems. OneWord Press, Santa Fe, New Mexico. 474pp. Christopherson, Gary L., D. Phillip Guertin, and Karen A. Borstad. 1996. GIS and rcheology: Using Arclnfo to increase our understanding of ancient Jordan. 16"' Annual ESRI International Users Conference Proceedings. ESRI, California. http://www.esri .com Collado, Carlos. 1998a. Los Suelos de Peten - Clasificacion Parte I. Somos Uno. l(2):7. Collado, Carlos. 1998b. Los Suelos de Peten - Clasificacion Parte II. Somos Uno. 1(3):7. CONAP/USAID/Peregrine Foundation. 1996. El estado de la Reserva de la Biosfera Maya en 1996. CONAP, Peten, Guatemala. 27pp. 71 Dale, Virginia H. and Scott M. Pearson. 1999. Modeling the driving factors and ecological consequences of deforestation in the Brazilian Amazon in Spatial modeling of forest landscape change: approaches and applications. Cambridge University Press, Cambridge. Grandia, Liza. 2000. Cuantas personas quiere usted que vivan en Peten? Encuentro Intemacional de Investigadores - Nuevas Perspectivas de Desarrollo Sostenible en Peten. FLACSO, Guatemala. In print. Griinberg, Georg and Victor Hugo Ramos. 1998. Base de Datos sobre poblacion, tierras y medio ambiente en la Reserva de la Biosfera Maya. CARE Guatemala and CONAP. Peten, Guatemala. 91pp. Islebe, Gerald A., Henry Hooghiemstra, Mark Brenner, Jason H. Curties and David A. Hodell. 1996. A Holocene vegetation history from lowland Guatemala. The Holocene. 6:265-271. lUCN/UNEPAVWF. 1991. Caring for the Earth - A Strategy for Sustainable Living. Giand, Switzerland. 228pp. Juarez, Jenny. 1994. A GIS Assessment of Deforestation in the Coto Brus Valley of Costa Rica. M.S. Thesis, Oklahoma State University. Stillwater, Oklahoma. USA. Kircher, John. 1997. A Neotropical Companion. Princeton University Press, Chichester, West Sussex. 451pp. Mas, Jean-Francoise, Azucena Perez Vega, Gerardo Palacio Aponte, and David Zarate Lomeli. 1997. Assessing deforestation in the coastal zone of Campetche State. Mexico. Proceeding of the IV International Conference on Remote Sensing for Marine Coastal Environments. Orlando, Florida. 1:593-602. Mertens, Benoit, and Eric F. Lambin. 1997. Spatial modeling of deforestation in southern Cameroon spatial disaggregation of diverse deforestation processes. Applied Geography. 17(2): 143-162. Nelson, Gerald C. and Daniel Hellerstein. 1995. Do roads cause deforestation? Using satellite images in econometric analysis of land use. University of Illinois, Department of Agricultural Economics, Urbana-Champaign. Staff Paper 95 E- 488. Palma, Edgar. 2000. Como vivir mejor de nuestras parcelas - Permacultura Petenera. CARE Guatemala. Peten, Guatemala. 147pp. 72 Pfaff. Alexander S. P. 1996. What Drives Deforestation in the Brazilian Amazon? - Evidence from Satellite and Socioeconomic Data. The World Ban, Policy Research Department, Environmental, Infrastructure, and Agricultural Division. Report. Sader, Steven A. 1995. Spatial characteristics of forest clearing and vegetation regrowth as detected by Landsat Thematic Mapper Imagery. Photogrammetric Engineering and Remote Sensing. 61(9); 1145-1151. Sader, Steven A., Conrad Reining, Thomas Sever, and Carlos Soza. 1997. Human migration and agricultural expansion - a threat to the Maya Tropical Forest. Journal of Forestry. 95(I2):27-32. Sader, Steven A., Edgar Bemales Martinez, Daniel E. Irwin, and Hector Tuy Yax. 2000. Estimacion de la Deforestacion En la Reserva de la Biosfera Maya 1997 - 1999. Encuentro Intemacional de Investigadores - Nuevas Perspectivas de Desarrollo Sostenible en Peten. FLACSO, Guatemala. In print. Sader, Steven A., Michael Coan, and Daniel Hayes. 1998. Time-series tropical forest change-detection for the Maya Biosphere Reserve: Updated Estimates for 1995 to 1997. Maine Image Analysis Laboratory, Department of Forest Management, University of Maine. Unpublished report. Cl-ProPeten. Peten, Guatemala. Santiso, Claudio. 1993. The Maya Biosphere Reserve: an alternative for the sustainable use of resources. Nature and Resources. 28(l-4):6-l 1. Schwartz, Norman B. 1990. Forest Society - a social history of Peten, Guatemala. University of Pennsylvania Press, Philadelphia. 367pp. Stanley, Scott A.. 1998. Evaluacion Rapida de los Incendios Forestales de 1998 en la Reserva de Biosfera Maya, Peten, Guatemala. CONA. Peten, Guatemala. Report. Thapa, Khagendra and John Bossier. 1992. Accuracy of spatial data used in geographic information systems. Photogrammetric Engineering & Remote Sensing. 58(6):835-841. Warren, R. E. 1990. Predictive modeling of archeological site location: a case study in the Midwest. p210-215 in Interpreting space: GIS and archeology. London, Taylor and Francis. Wilson, Edward O. 1992. The Diversity of Life. Beknap Press of Harvard University Press, Cambridge, Massachusetts. 424pp. 73 Wood, Charles H. and David Skole. 1998. Linking Satellite. Census, and Survey Data to Study Deforestation in the Brazilian Amazon p70-93 in People and Pixels - Linking Remote Sensing and Social Sciences. National Academy Press. Washington, D.C. 244pp.