<%BANNER%>

Defending public interests in private forests

Permanent Link: http://ufdc.ufl.edu/UFE0041028/00001

Material Information

Title: Defending public interests in private forests Land-use policy alternatives for the Xingu River headwaters region of southeastern Amazonia
Physical Description: 1 online resource (200 p.)
Language: english
Creator: Stickler, Claudia
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: agriculture, agroindustry, amazon, brazil, carbon, deforestation, frontier, landuse, market, policy
Interdisciplinary Ecology -- Dissertations, Academic -- UF
Genre: Interdisciplinary Ecology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: When native vegetation is cleared to establish agricultural lands, damage to ecosystem services such as air, water, and climate can outweigh the substantial benefits of agricultural production. Brazil has created ambitious laws and regulations for the purpose of regulating land use on private lands in Amazon forests. This dissertation analyzes the performance of the central piece of Brazilian environmental legislation in the Amazon region: the Forest Code. In the wake of escalating deforestation and international pressure in the mid-1990?s, the Brazilian Government modified the Forest Code, increasing from 50 to 80% the required area of each private landholding in the region that had to be maintained in native forest. I analyzed (1) the level of compliance with both the old and new Forest Code, the change in compliance over time, the costs of compliance, and the ecological services provided under the old versus the new regulations; (2) the potential for hybrid regulatory-economic policies (tradable forestland development rights and land-use zoning) to reduce the opportunity costs of the modified Forest Code while protecting ecosystem services and ecological integrity; and (3) the potential of the emerging forest carbon market to complement Forest Code and land-use zoning protection of public interests in Amazon forests. As a case study, I used the 178,000 km2 Xingu River headwaters region in the southeastern Amazon Basin. I developed a spatially-explicit land-cover simulation model in conjunction with a river discharge model and maps of potential economic rents under soy, cattle ranching, and logging, to conduct these analyses. When the Forest Code's legal reserve increased from 50 to 80% in 1996, compliance dropped immediately from 92 to 72%, then declined further to 46% by 2005. The regulatory change imposed approximately nine billion dollars in forgone profits from forest conversion to soy and cattle ranching. The Mato Grosso state zoning plan, if implemented, would potentially provide 4000 km2 more agricultural and pasture land, reducing the opportunity costs of strict compliance with the 80% legal reserve by one third, while protecting ecosystem services similarly well. Emerging carbon markets, if expanded to fully and fungibly include forest carbon could offset much of the opportunity cost of forest conservation in the Xingu region, increasing the viability of forest conservation.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Claudia Stickler.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Southworth, Jane.
Local: Co-adviser: Zarin, Daniel J.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0041028:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041028/00001

Material Information

Title: Defending public interests in private forests Land-use policy alternatives for the Xingu River headwaters region of southeastern Amazonia
Physical Description: 1 online resource (200 p.)
Language: english
Creator: Stickler, Claudia
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: agriculture, agroindustry, amazon, brazil, carbon, deforestation, frontier, landuse, market, policy
Interdisciplinary Ecology -- Dissertations, Academic -- UF
Genre: Interdisciplinary Ecology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: When native vegetation is cleared to establish agricultural lands, damage to ecosystem services such as air, water, and climate can outweigh the substantial benefits of agricultural production. Brazil has created ambitious laws and regulations for the purpose of regulating land use on private lands in Amazon forests. This dissertation analyzes the performance of the central piece of Brazilian environmental legislation in the Amazon region: the Forest Code. In the wake of escalating deforestation and international pressure in the mid-1990?s, the Brazilian Government modified the Forest Code, increasing from 50 to 80% the required area of each private landholding in the region that had to be maintained in native forest. I analyzed (1) the level of compliance with both the old and new Forest Code, the change in compliance over time, the costs of compliance, and the ecological services provided under the old versus the new regulations; (2) the potential for hybrid regulatory-economic policies (tradable forestland development rights and land-use zoning) to reduce the opportunity costs of the modified Forest Code while protecting ecosystem services and ecological integrity; and (3) the potential of the emerging forest carbon market to complement Forest Code and land-use zoning protection of public interests in Amazon forests. As a case study, I used the 178,000 km2 Xingu River headwaters region in the southeastern Amazon Basin. I developed a spatially-explicit land-cover simulation model in conjunction with a river discharge model and maps of potential economic rents under soy, cattle ranching, and logging, to conduct these analyses. When the Forest Code's legal reserve increased from 50 to 80% in 1996, compliance dropped immediately from 92 to 72%, then declined further to 46% by 2005. The regulatory change imposed approximately nine billion dollars in forgone profits from forest conversion to soy and cattle ranching. The Mato Grosso state zoning plan, if implemented, would potentially provide 4000 km2 more agricultural and pasture land, reducing the opportunity costs of strict compliance with the 80% legal reserve by one third, while protecting ecosystem services similarly well. Emerging carbon markets, if expanded to fully and fungibly include forest carbon could offset much of the opportunity cost of forest conservation in the Xingu region, increasing the viability of forest conservation.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Claudia Stickler.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Southworth, Jane.
Local: Co-adviser: Zarin, Daniel J.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0041028:00001


This item has the following downloads:


Full Text

PAGE 1

DEFENDING PUBLIC INTEREST S IN PRIVATE FORESTS: LAND-USE POLICY ALTERNATIVES FO R THE XINGU RIVER HEADWATERS REGION OF SOUTHEASTERN AMAZNIA By CLAUDIA MARGRET STICKLER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORID A IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009 1

PAGE 2

2009 Claudia Margret Stickler 2

PAGE 3

To my mother and my father, who gave me the courage, the in spiration, and the freedom to pursue my dreams even when they took me far away 3

PAGE 4

ACKNOWLEDGMENTS I extend my deep thanks to my advisors, Jane Southworth and Dan Zarin, for seeing me through this long process. W hen I was still a Masters student, Jane was instrumental in initiating me to the world of applying a spatia l perspective to social and ecological questions. Since then, she has hel ped orient in my re search, support my research in countless ways, and collaborated with me on various projects. She has also been unwavering in her support of my prof essional development. Dan opened a new world for me, introducing me to the Brazilian Amazon and, in the process, a wealth of new colleagues and mentors. His critical mi nd was invaluable in improving proposal after proposal, paper after paper, and finally, chapt er after chapter. I thank them both for their unflinching support I needed to accomplish this project. The other members of my committee were no less important. Grenville Barnes took me under his wing and guided me to wards critical resources and questions, provided important perspectives, expr essed confidence in me, and provided encouragement at crucial mome nts. Brian Child engaged me in important discussions about the realities of land management under imperfect governments. Chuck Wood helped to ensure that I had clarity and fo cus in developing and carrying out my research. Other faculty members at the University of Florida offered critical support throughout this stage of my education. Bob Buschbacher changed the course of my life when he suggested that I apply for a doctoral fellowship through the Working Forests in Tropics program. Under the leadership of Bob and Dan, the programs executive committee (especially Dan, Bob, Marianne Sc hmink, Karen Kainer, and Jack Putz) took a chance on me and provided me with an exc eptional professional and life experience. 4

PAGE 5

The Department of Geography provided a departmental base for me and assisted me with every aspect of education. In addition to Jane Southworths support as a mentor, special thanks go to Julia Williamswho had ably took the adm inistration of two of my fellowship grants, Desiree Pr icewho unfailingly made sure all administrative details related to enrollment and r equirements were looked after, and Pete Waylenwho made sure that SNRE students were incorpor ated on equal footing into the cadre of Geography graduate students. T he Tropical Conservation and Development Program helped me to maintain my roots in the prac tical and social aspects of conservation and development. The School of Natural Resources provided a flexible and rich environment for taking on the difficult task of gaining an interdisciplinary education in environmental science. Special thanks are due to Steven Humphrey, Cathy Ritchie, and Meisha Wade. Natalia Hoyos assisted me in many stages of satellite image processing and GIS database development and management, providi ng a level of organization and careful attention to detail that was an inspiration and a great sour ce of confidence for me. I had the good fortune to be hosted in my dissertation work in Brazil by the Instituto de Pesquisa Ambi ental da Amaznia (IPAM), the Woods Hole Research Center (WHRC), the Centro de Sensoriamento Remoto of the Federal University of Minas Gerais (UFMG), and the Aliana da Terra (AT). I was fortunate to be invited to take part in the Amazon Scenarios projec t by Dan Nepstad (WHRC/IPAM) and Britaldo Soares (UFMG). IPAM provided me an institut ional home in Brazil; IPAM staff oriented me in the development of my research and took me on as a team member. At IPAM, Oriana Almeida, Oswaldo de Carvalho, Jr ., Ane Alencar, Toby McGrath, Paulo Moutinho, Claudia Azevedo-Ramos, Gina Cardinot, Laura Dietzsch, Paulo Brando, 5

PAGE 6

Osvaldo Portela, Adriano Portela, Roberto Santos, Ricardo Melo, Elsa Mendoza, Lucimar Souza, Lou Spinelli, Osvaldo Stel la, Raquel Diegues, Andre Lima, and Isabel Castro all provided support, advice, and frie ndship. At CSR/UFMG, Britaldo SoaresFilho and Hermann Rodrigues took me under their wings and spent innumerable hours helping me to develop the simulation model at t he core of this dissertation. I also thank the CSR/UFMG support team, including William Lel es Costa, Rafaella Silvestrini, Leticia de Barros Viana Hissa, and Alerson Falieri, for their assistance. WHRC provided institutional and material support from the begi nning, as well as providing the support I needed to complete my dissertation. My spec ial thanks go to Wayne Walker, who spent countless hours helping and teaching me to cr eate the land-cover m aps that formed the basis of much of my work. Toby McGr ath read and re-read versions of outlines, proposals, and chapters at ever y stage of my dissertation pr ocess. Mike Coe helped me to work with hydrological models and looked out for my interests and made sure I stayed on track to finish my dissertation. At AT, John Carter introduced me to Mato Grosso and the obstacles that individual land -owners face as they attempt to maintain forest and a viable business. My friends and colleagues at the Universi ty of Florida, WHRC, IPAM, and AT and beyond have been invaluable. Over the years, Amy Duchelle, Miriam Wyman, Meredith Evans, Gaby Stocks, Anna Pri zzia, Lin Cassidy, Wendy-Lin Bartels, Franklin Paniagua, Alfredo Rios, David Buck, Ellie Harrison-Bu ck, Shoana Humphries, Maria DiGiano, Cara Rockwell, Joanna Tucker, Amy Daniels, Connie Clark, John Poulsen, Christine Engels, John Engels, Tracy Van Holt, Tita Alvira, Oswaldo de Carval ho, Jr., Paulo Brando, Ane Alencar, Oriana Almeida, Sergio Rivero, Paulo Moutinho, Claud ia Azevedo-Ramos, 6

PAGE 7

Lucimar Souza, Ricardo Mello, Raquel Diegues Osvaldo Stella, Lou Spinelli, Anaiza Portilho, Karine Carvalho, Vivian Zeidema nn, John Carter, Britaldo Soares, Hermann Rodrigues, Gina Cardinot, Laura Dietzsch, Wayne Walker, Jared Stabach, Pieter Beck, Karen Schwalbe, Josef Kellndorfer, Emily Ke llndorfer, Michelle West, Toby McGrath, Almira McGrath, Mike Coe, Wendy Kingerl ee, Scott Messner, Liz Braun, Frank Merry, Nora Greenglass, Ann Tarran t, Hope Neighbor, Heather Wr ight, and Maureen Geesey formed an extraordinary and patient support team, without whom I would not have arrived at this point with any humor or sanity. I thank my family for setting high standards for what individuals can achieve, no matter what their vocation. They provide me with inspiration and a sense of responsibility to the broader world. My paren ts afforded me great opportunities and freedom to follow my dreams, and continue to inspire me with their eagerness to always try new things. I thank my siblings for their encouragement. And I thank the newer members of my family, Isabel and Max, Laurie and Don, Steve and Kerri, for their patience and support. Last, but not least, I thank Daniel Nepst ad for his unwavering support and love at every step I have taken on the road toward completing this dissertation. He discussed every intricate detail of my research design, helped me to make contacts, and revised my proposal, manuscript, and dissertation drafts. My research and education were supported through a National Science Foundation Graduate Research Fellowship, a National Science Foundation Integrated Graduated Education and Research Training Fe llowship through the Working Forests in the Tropics Program at the University of Florida, a National Space and Aeronautics 7

PAGE 8

Administration Earth System Sciences Fellowship, a National Security Education Program David L. Boren Fello wship, and an Environmental Protection Agency Science to Achieve Results Fellowship. The Woods Hole Research Center and the Instituto de Pesquisa Ambiental da Amaznia also pr ovided support through grants from NASA LBA-ECO, the National Science Foundat ion, the Conselho Nacional de Desenvolvimento Cientifico e Tecnologico, the Gordon and Betty Moore Foundation, the David and Lucille Packard Foundation, and the Linden F oundation. I also thank the Woods Hole Research Center, the Institut o de Pesquisa Ambiental da Amaznia, and the Aliana da Terra for providing institutional support during the course of dissertation work. 8

PAGE 9

TABLE OF CONTENTS page ACKNOWLEDG MENTS..................................................................................................4 ABSTRACT ...................................................................................................................15 CHAPTER 1 INTRODUC TION....................................................................................................17 2 DEFENDING PUBLIC INTERESTS IN PRIVATE FORESTS: COMPLIANCE, OPPORTUNITY COSTS, AND POTENTIAL ECOLOGICAL PERFORMANCE OF THE BRAZILIAN FOREST CODE IN THE SOUTHEASTERN AMAZON.........23 Introducti on.............................................................................................................23 Governance of Common Pool Res ources...............................................................27 Environmental Gover nance in Br azil .......................................................................29 The Brazilian Fo rest C ode......................................................................................33 Current Requi rements......................................................................................33 History ..............................................................................................................35 Relevance of Ot her Policies.............................................................................38 Materials and Methods............................................................................................41 Study Ar ea........................................................................................................41 Land Cover Maps.............................................................................................42 Observed l andscapes ................................................................................42 Theoretical l andscape s..............................................................................43 Complianc e......................................................................................................46 Economic Aspects of Compli ance....................................................................47 Ecological C onsequences................................................................................48 Result s....................................................................................................................49 Complianc e......................................................................................................49 Legal rese rve.............................................................................................49 Riparian buffe r zone...................................................................................52 Economic Aspects of Compli ance....................................................................52 Opportunity cost.........................................................................................52 Cost of rest oration......................................................................................54 Ecological C onsequences................................................................................56 Carbon st ocks............................................................................................56 Hydrology and region al clim ate..................................................................56 Water qua lity..............................................................................................57 Habita t.......................................................................................................57 Discussio n..............................................................................................................58 Conclusi on..............................................................................................................66 9

PAGE 10

3 HYBRID REGULATORY-ECONOMIC POLICY INSTRUMENTS IN PRIVATE FOREST GOVER NANCE.......................................................................................81 Introducti on.............................................................................................................81 Forest Governance on Privat e Lands .....................................................................83 Policy Instruments for Natu ral Resource Managem ent....................................84 The Legal Reserve of the Federal Fore st Code...............................................86 Tradable Deforestat ion Right s..........................................................................88 State Socioeconomic-Ecological Zoning Plans................................................91 Materials and Methods............................................................................................94 Study Ar ea........................................................................................................94 Land Cover Maps.............................................................................................95 Observed l andscapes ................................................................................95 Modeled land scapes ..................................................................................96 Analyses .........................................................................................................103 Vegetation cover......................................................................................103 Economic as pects....................................................................................103 Ecological c onsequences ........................................................................104 Results ..................................................................................................................104 Remnant Forest and Cerrado .........................................................................104 Restored Forest and Cerrado .........................................................................105 Native Vegetati on Cover .................................................................................106 Potential Agricult ural Ar ea..............................................................................106 Potential Net Pres ent Valu e...........................................................................107 Costs of Restoring Fo rest and Ce rrado..........................................................108 Ecological Cons equences..............................................................................108 Carbon sto cks..........................................................................................108 Hydrology and region al climat e................................................................109 Water qualit y............................................................................................109 Habitat .....................................................................................................110 Discussio n............................................................................................................110 Conclusi on............................................................................................................113 4 THE COST OF CARBON TO OFFSET OPPORTUNITY COSTS ON PRIVATE LANDS UNDER ALTERNATIVE POLICI ES IN THE UPPER XINGU RIVER BASIN................................................................................................................... 123 Introducti on...........................................................................................................123 Study Ar ea............................................................................................................126 Materials and Methods..........................................................................................126 Model Devel opment....................................................................................... 127 Scenarios .......................................................................................................128 Business-as-usual scenario s...................................................................128 Policy scena rios.......................................................................................129 Analyses .........................................................................................................130 Emissions reductions and carbon enhancement......................................130 Price of ca rbon.........................................................................................131 10

PAGE 11

Eligible cr edits..........................................................................................132 Assessment of ecologica l co-benef its......................................................133 Results ..................................................................................................................133 Emissions Reductions and Carbon Enhancement.........................................133 Price of Ca rbon..............................................................................................135 All car bon.................................................................................................135 Eligible ca rbon.........................................................................................136 Ecological co -benefits ..............................................................................136 Discussio n............................................................................................................138 Conclusi on............................................................................................................143 5 CONCLUS ION......................................................................................................150 APPENDIX A LAND-COVER MAPS...........................................................................................155 Sensor Da ta..........................................................................................................155 Reference Data.....................................................................................................156 Classification and M apping Approac h...................................................................157 Segmentation and Attri bute Extrac tion...........................................................157 Spatial Database Joining................................................................................158 Classification and Mapping .............................................................................158 B DYNAMIC SPATIAL SIMULATION MODEL DEVELO PMENT.............................161 Overview ...............................................................................................................161 Model Calibra tion..................................................................................................163 Landcover Simu lation...........................................................................................164 Model Validat ion...................................................................................................165 C ECONOMIC AND ECOLOG ICAL ASSESSM ENT................................................167 Economic Assessment..........................................................................................167 Net Present Value..........................................................................................167 Restoration Costs...........................................................................................168 Ecological A ssessment .........................................................................................168 Carbon Sto cks................................................................................................169 Surface Hydrology and Local and Regional Climate......................................169 Indicators of Wa ter Qualit y.............................................................................171 Terrestrial H abitat...........................................................................................171 LIST OF REFE RENCES.............................................................................................174 BIOGRAPHICAL SKETCH ..........................................................................................200 11

PAGE 12

LIST OF TABLES Table page 2-1 The area and percent coverage of nat ive forest and cerrado in the Xingu River headwaters region as required by old and new Brazilian Forest Code regulations, and as observ ed for 1996, 1999 and 2005. ....................................68 2-2 The number and percentage of microbasins in the Xingu River headwaters region that were in compliance with Brazilian Forest Code forest cover requirements in 1996, 1999, and 2005. ..............................................................69 2-3 Evolution of compliance with old and new Brazilian Forest Code in the microbasins of the Xi ngu River headwaters region, by category of change for 2 time per iods.....................................................................................................70 2-4 Total area and mean area per microbasin of forest and cerrado within the Xingu River headwaters region that is available for conversion and/or needing to be restored under the old and new Brazilian Forest Code, measured at th ree dates .....................................................................................71 2-5 Potential net present value (NPV) associated with soybean cultivation or cattle production on forested and deforested land for the Xingu River headwaters region under alte rnative Brazilian Forest Code requirements at three dat es.........................................................................................................72 2-6 Mean potential net present value of soy cultivation or cattle ranching for cleared and forested land of compliant and non-compliant microbasins of the Xingu River headwaters regi on at three dates....................................................73 2-7 Mean potential net present value asso ciated with soy cultivation or cattle ranching on cleared and forested lands in microbasins of the Xingu River headwaters region at three dat es.......................................................................74 2-8 The average and range of estimated cost s of restoration (million USD) of the legal reserve and riparian zone under 2 iterations of the Forest Code in relation to three points in time, for the entire basin in the aggregate and for the average microbasin out of comp liance. ........................................................75 2-9 Ecological attributes of the Xi ngu River headwaters region (including protected areas) at three 3 dat es (1996, 1999, 2005) and of modeled landscapes representing t he region with application of 80% legal reserve or 50% legal reserve on private la nds.....................................................................76 3-1 Total area and mean area per microbas in of remnant forest and cerrado woodland within the Xi ngu River headwaters region in 2005 and under 3 land-use policy sc enarios.................................................................................114 12

PAGE 13

3-2 Total area and mean area per microbasin of forest and cerrado within the Xingu River headwaters region that will consist of restored vegetation in 2005 and under 3 land-use poli cy scenarios .............................................................115 3-3 The area and percent coverage of forest and cerrado vegetation (remnant and restored combined) in the Xingu Riv er headwaters region as observed for 2005 and modeled under three landuse policy scena rios..........................116 3-4 Total area and mean area per microbasin of forest and cerrado within the Xingu River headwaters region that is ava ilable for agricultural production in 2005 and in 3 land-use po licy scenario s...........................................................117 3-5 Potential net present value (NPV) associated with soybean cultivation or cattle production on cleared land for the Xingu River headwaters region under 3 alternative policy scenarios and in 2005..............................................118 3-6 The estimated costs of restoratio n of the legal rese rve and riparian zone under 3 alternative policies, for the ent ire basin in the aggregate and for the average microbasin out of compli ance.............................................................119 3-7 Ecological attributes of the Xi ngu River headwaters region (including protected areas) in 2005 and of modeled landscapes representing the region under 3 policy alternatives on private lands. .....................................................120 4-1 Forest carbon stocks on private lands of the Xingu Riv er headwaters region in 2005 and simulated for 2035 under six sc enarios. ........................................145 4-2 The effect of policy scenarios on the costs, carbon stocks, and costs per ton of carbon associated with additional fore st cover in the privately-owned land of the Xingu River hea dwater r egion................................................................ 146 4-3 The effect of policy scenarios on t he basin-wide costs, carbon stocks, and costs per ton of carbon associated with additional forest cover in excess of 50% cover of privately held lands in the forest biome of the Xingu River headwater region, represented by individual mi crobasins. ...............................147 4-4 Comparison of ecological features of the Xingu River headwaters regions with 2005 forest cover, under three simula tions of future fo rest cover under business-as-usual assumptions, and under three simulated policy scenarios..148 A-1 Image dates for Landsat-based mosa ics for 4 years (1996, 1999, 2005, 2007)................................................................................................................ 160 C-1 Estimated costs for restoration of native vegetation in legal reserves and riparian zones in the Xingu River headwaters region.......................................173 13

PAGE 14

LIST OF FIGURES Figure page 2-1 Maps showing the Xingu River headwaters region unde r different scenarios....77 2-2 Box diagram showing the fraction of microbasins (which serve as proxies for private properties) in the forest biome in each of three categories of legality and the change in that fraction from the previous dat e.......................................78 2-3 Map showing microbasins (which serve as proxies for private properties)out of compliance with two iterations (50% and 80% legal reserve requirement) of the Forest Code at three time poin ts: (a) 1996, (b) 1999, and (c) 2005..........79 2-4 Microbasins within two extreme categories of legality in relation to the modified Forest Code: (a) microbas ins with less than 50% forest cover remaining 1996 through 2005; (b) microbasins with more than 80% forest cover remainin g since 1996................................................................................ 80 3-1 Location and distribution of the major biophysical restrictions (shown in red) and the 4 major zones of the Mato Gro sso State Zoning Plan (ZSEE) within the Xingu River headwaters: (a) fl oodplain areas; (b) soil and slope restrictions; (c) wetland areas; (d) the 4 major zones (noted by number) combined with the 3 biophysi cal restrict ions.....................................................121 3-2 Comparison of land-cover in the basin for (a) 2005 (observed), (b) full compliance with 80% legal reserve in forest biome, no compensation (CFCNC) (modeled), (c) full compliance with 80% legal reserve in the forest biome with compensation (CFC-C) (modeled), and (d) Mato Grosso state socioeconomic ecological zoning plan (ZSEE) (modeled), with compensation.........122 4-1 Maps showing simulated future (2035) land cover of the Xingu River headwaters region under six different sc enarios. .............................................149 14

PAGE 15

Abstract of Dissertation Pr esented to the Graduate School of the University of Florida in Partial Fulf illment of the Requirements for t he Degree of Doctor of Philosophy DEFENDING PUBLIC INTEREST S IN PRIVATE FORESTS: LAND-USE POLICY ALTERNATIVES FO R THE XINGU RIVER HEADWATERS REGION OF SOUTHEASTERN AMAZNIA By Claudia Margret Stickler December 2009 Chair: Jane Southworth Cochair: Daniel J. Zarin Major: Interdisciplinary Ecology When native vegetation is cleared to es tablish agricultural lands, damage to ecosystem services such as air, water, and climate can outweigh the substantial benefits of agricultural producti on. Brazil has created ambitious laws and regulations for the purpose of regulating land use on private lands in Amazon forests. This dissertation analyzes the performance of the central piece of Brazilian environment al legislation in the Amazon region: the Forest Code. In the wake of escalating deforestation and international pressure in the mid-1990s, the Brazilian Government modified the Forest Code, increasing from 50 to 80% the require d area of each private landholding in the region that had to be maintained in native fo rest. I analyzed (1) the level of compliance with both the old and new Forest Code, the change in compliance over time, the costs of compliance, and the ecological services provided under the old versus the new regulations; (2) the potential for hybrid regul atory-economic policies (tradable forestland development rights and land-use zoning) to reduce the opportunity costs of the modified Forest Code while protecting ecosystem services and ecological integrity; and (3) the 15

PAGE 16

potential of the emerging forest carbon market to complement Forest Code and landuse zoning protection of public interests in Amazon forests. As a case study, I used the 178,000 km2 Xingu River headwaters region in t he southeastern Amazon Basin. I developed a spatially-exp licit land-cover simulation model in conjunction with a river discharge model and maps of po tential economic rents under soy, cattle ranching, and logging, to conduct these analyses. When the Forest Codes legal reserve increased from 50 to 80% in 1996, compliance dropped immediately from 92 to 72%, then declined further to 46% by 2005. The regulatory change imposed appr oximately nine billion dollars in forgone profits from forest conversion to soy and cattle ranching. The Mato Grosso state zoning plan, if implemented, would potentially provide 4000 km2 more agricultural and pasture land, reducing the opportunity costs of strict compliance with the 80% legal reserve by one third, while protecting ecosystem services si milarly well. Emerging carbon markets, if expanded to fully and fungibly in clude forest carbon could offset much of the opportunity cost of forest conservation in the Xingu region, increasing the viability of forest conservation. 16

PAGE 17

CHAPTER 1 INTRODUCTION In addition to producing food, fiber, and f uel for the worlds growing population, the agricultural sector is an important source of income to support infrastructure development and the provision of public goo ds throughout the world. However, when native vegetation is cleared to establish agricultural lands, damage to critical ecosystem services such as air, water, and climat e can outweigh the s ubstantial benefits of agricultural production. Nearly one fifth of world-wide, human-induced greenhouse gas emissions to the atmosphere originate from the cutting and burning of tropical forests (IPCC, 2007). On regional scales, silted rivers, deadly avalanches, mudslides and floods, displaced and decimated tribes of indigenous forest peoples, fragmented landscapes, and the loss of plant and animal species are among the negative consequences of large-scale forest conversion. In the face of projected land shortages, increased food prices (FAO, 2009) and the threat of climate destabilization (IPCC, 2007), the question of how to balance private and public gains from agriculture with the damages that often result from conversi on of native ecosystems to support these activities. A major policy response to rampant deforestation has been to set aside tracts of intact natural landscapes as parks, reserves public forests, terri torial reserves of indigenous and traditional peop les and other types of protected areas (Terborgh, 1999; Redford, 1994). In some places in the world, these protected areas have succeeded in defending public interests in forest s (Bruner et al., 2001; Nepstad et al., 2006) while in others they have failed (Curran et al., 2004). A growing body of evidence demonstrates, however, that many of the ecosystem services provided by forests that 17

PAGE 18

sustain human societies will require envir onmental conservati on on both private and public lands, and not just l ands controlled by government agencies (Soares Filho et al., 2006). An environmental conservation paradig m is needed that reconciles the defense of public interests in forest resources with the important social and economic benefits of forest conversion to agriculture and livestock expansion. In this dissertation, I evaluate the e ffectiveness of legally mandated land-use restrictions on private rural pr operty in protecting public inte rests related to forests and woodlands on the southeastern Amazons agro-i ndustrial frontier, using the Xingu River headwaters as a case study. I evaluated the economic and ecological trade-offs of the Brazilian Forest Codealong with alternatives adapted from this policyin protecting native vegetation cover, while permitting agricultural expansion, in the study region. I then assess the potential of a range of forest carbon price signals to compensate landholders for opportunity costs and forest restoration costs that they incur under alternative policy scenarios In response to international concern over the very high rate of Amazon deforestation in 1995, the Forest Code was modified to require that 80% of each individual property in rural Amaznia be maintained as native forest, increased from the 50% requirement that had been in place sinc e 1965. This sudden decision threw a large number of property holders who had already cleared more t han 20% of their properties into noncompliance, and was not acco mpanied by effective mechanisms and procedures by which these property holder s could bring their properties into compliance. Policy instruments that were cr eated to facilitate compliance with the new regulation were not implement ed. These included a provision allowing landholders to 18

PAGE 19

compensate the acreage of forest that they needed to comply with the 80% rule through the trade of land development rights, and the creation of socio-economic, ecological land-use zoning that relaxed the 80% requirement in zones of agricultural consolidation. Today, thirteen years after th e historic Forest Code modification was made, there is little quantitative information regarding the level of compliance with this law, the potential economic and ecological impacts of full compliance, and the degree to which these provisions, if implemented, could help to advance cost-effective reconciliation of forest conserva tion with agricultural expansion. The central goal of the research present ed here was to evaluate the effectiveness of legally mandated land-use restrictions on private rural property in protecting public interests related to forests and woodlands on the southeastern Amazons agro-industrial frontier, using the 178,000 km2 upper Xingu River basin regi on as a case study. I evaluated the economic and ecological tradeoffs of the Forest Codealong with alternatives adapted from this policyin pr otecting native vegetation cover, while permitting development of important econom ic activities, in the study region. Specifically, I asked the following questions: Does the current legislation to regulate land-use (especially forest clearing) reach its stated objective in practice? Does current legislation balance public and private interests in conservation and agricultural production on private lands in theory? If not, could it be adapted to be more effective? How are future land-use/land-cover trajectori es likely to vary in response to a suite of plausible future policy scenario s derived from the Forest Code? What are the environmental and economic trade-o ffs associated with those scenarios? 19

PAGE 20

The Xingu River is one of the Amazon Basins major southeastern tributaries. This landscape is three times the size of Costa Ric a, larger than 90% of tropical nations, and provides a mesocosm of the range of actors and land ca tegories encountered in the Brazilian Amazon, including modern agro-indus trial farms, indigenous reserves, nature reserves, cattle ranches, and smallholder settlements (Jepson, 2006). In selecting a region defined by watershed boundaries, I wa s able to examine the linkage between land-cover and river discharge, which is strongly regulated by forests. The Xingu headwaters region also spans across two majo r biomes, with the closed-canopy forest in the upper (northern) two-thirds, and t he cerrado woodland/savanna biome across the lower third. Chapter 2 of this dissertation provides an assessment of the effectiveness of the Brazilian Forest Code in protecting fore sts on private lands in the Xingu River headwaters. In particular, I focus on the legal and ecological effectiveness of the change in the policy that took place in 1996. This study consists of two parts, a qualitative analysis and a quantitat ive analysis. First, I describe the characteristics of the resources that are the obj ect of protection under the Forest Code and discuss the implications for effective policy structures for resources that represent both a private and a public good. Next, I describe the histor y and requirements of the Forest Code to gain an understanding of the policys charac teristics with respect to governing a common pool resource. Finally, I carry out a quantitative analysis focusing on three aspects of compliance with the Forest Code: (1) the level of compliance with the Forest Code for a landscape on the southeastern Amaz on frontier; (2) the cost of compliance to landowners; and (3) the protection of ec osystem services and the conservation of 20

PAGE 21

forested landscape integrity at three moment s in history and under full compliance with the 80% and 50% legal reserve requirement in modeled landscapes. I interpret these quantitative results in light of the political proce ss surrounding the creation and implementation of the decision to increas e the legal reserve requirement to 80%. Chapter 3 evaluates the potential for fa cilitating landholder compliance with the modified Forest Code without forfeiting the env ironmental protection potentially provided through two policy provisions that could s ubstantially lower the economic opportunity cost of compliance: tradable forest devel opment rights and state-level socio-economic, ecological zoning (ZSEE). First, I descri be these two policy in struments and compare them with the Forest Code, analyzing the extent to whic h they represent economic instruments or exclusively regulatory instru ments. Next, I carry out a quantitative analysis of each instrument, focusing on the balance between agricultural expansion and forest conservation. I compare modeled landscapes representi ng the land-cover consequences of full implementation of each of the three policy alternatives in terms of area available for agricultural activities, the opportunity costs and forest restoration costs of compliance with each theoretical la ndscape, and the extent to which ecosystem services are maintained. This chapter hi ghlights the potential of hybrid regulatoryeconomic policies to reconcile public and privat e interests in private-land forests, but identifies the need for additional economic incentives to fully achieve this reconciliation potential. Chapter 4 examines the potent ial of payments for forest carbon to complement the Forest Code and its hybrid provisions in prom oting land use decisions that protect public interests in private forest while allowin g for agriculture and ranching on a designated 21

PAGE 22

area of land that is suitable for these activiti es. More specifically, I estimate the per unit price of forest carbon that would be necessary to compensate landholders for the costs (both opportunity and forest restoration) incu rred in complying with land-use legislation under three policy scenarios: the reduced legal reserve (i.e., a 50% legal reserve in the forest biome compared to the current 80% requirement, the current Forest Code (80% legal reserve requirement, without tradabl e development rights), and the ZSEE scenario. I estimate the difference in fo rest carbon stocks (both remnant native ecosystems and enhanced carbon in restored forests) between each simulated policy scenario and three simulations of business-as-usual deforestation and calculate the carbon price using the co sts that are associated with each policy scenario. 22

PAGE 23

CHAPTER 2 DEFENDING PUBLIC INTERESTS IN PRIVATE FORESTS: COMPLIANCE, OPPORTUNITY COSTS, AND POTENTIA L ECOLOGICAL PERFORMANCE OF THE BRAZILIAN FOREST CODE IN THE SOUTHEASTERN AMAZON Introduction In 1995, deforestation in the Brazilian Amazon hit a record high of 29,000 km2 cleared in that year (INPE, 2009)shor tly after Brazil hosted the heralded United Nations Conference on Environment and Deve lopment (UNCED) in 1992. News of the Amazon forests accelerating destruction prec ipitated an international crisis, and the Brazilian government responded with widely publicized policy interventions 1 the most prominent of which was the in crease from 50 to 80% forest cover of rural properties in the legal reserve mandated by the Brazilia n Forest Code in the Amazon region. The effectiveness of this dramatic change in env ironmental policy for maintaining forests and the services they provide is still poorly understood even as the policy itself is hotly contested (Alencar et al., 2004). Three critical assumptions are inherent in the change in the Forest Code as the pillar of the policy response. First, it assu mes that the Forest C ode is fundamentally the correct approach to protecting forests and the services that they provide. Secondly, and in conjunction, it assumes t hat the effectiveness of the po licy is primarily a matter of adjusting the proportion of each landholding to be maintained in a private reserve. Thirdly, it assumes that there will be comp liance with such a dramatic change in policy with no additional provisions or incentives to assist landow ners in achieving compliance. The latter assumption may be one of the most important reasons for the heated debate 1 The other major interventions were (1) the declar ation of a 2-year moratorium on the harvest of mahogany, and (2) the prohibition of new clearing on properties already possessing abandoned or under-utilized areas or areas used inappropriatel y with respect to the capacity of the soil. 23

PAGE 24

over the policy change that has raged unabated since its inception in 1996 (Fearnside, 2008). The change in the Forest Code brings into stark relief the challenge of balancing the protection of native vegetation and ecosystem services, while permitting development of important economic activities From the perspective of (primarily) the agricultural lobby and its supporters, the new legal reserve requirement makes it economically unviable to maintain a working ranch or farm in the Amazon biome and jeopardizes the regional economy and the provis ion of public services (Camarga, 2008). For environmentalists and social justice adv ocates, the new policy is necessary to counter the alarming increase in deforesta tion in recent years and its possible consequences for biodiversity, regional clim ate stability, greenhouse gas emissions, air pollution, soil degradation, local and regiona l water quality, and in turn, human health. The critical question is what can be done to balance the numerous private and public gains from agro-industrial production with the damages borne by society as a result of these activities. Typically, the approach to protecting public interests in forests has been to put them under state ownership for either strict protection or limited use for timber harvest. State control over timber harvests on pub lic lands has frequently led to severe degradation of forests, as st ate oversight has often been inadequate to prevent open access regimes from developing, and as cronyism and corruption have influenced the allocation of forest concessions (Barbier an d Burgess, 1997). The effectiveness of parks and other protected areas in controlling def orestation has been mixed. Examples from Brazil (Nepstad et al., 2006) and Costa Rica (Sanchez-Azofeifa et al., 2003) indicate 24

PAGE 25

that protected area status is associated with decreased deforestation rates, whereas parks in Indonesia have been subject to massive deforestation (Curran et al., 2004). Increasingly, forest ownership or usufruct rights by indigenous and other traditional communities have been officially granted in common property arrangements that recognize not only the customary rights of these groups but also assume that these groups will more effectively protect forests if they implement customary institutions for regulating resource use. The resulting in digenous lands, extractive reserves, and community reserves encompass almost one thir d of the forests of the Brazilian Amazon and large percentages of the remaining forests of most tropical nations, although the enforcement of these reserves varies greatly. Although strongly advocated by economic theory (and more famously, Garrett Hardin (1968) in The Tragedy of the Commons ), private property arrangements for forests in the tropics are more rare: only 10% of tropical forests are held by firms or individuals (White and Martin, 2002; Agrawal et al., 2008). Proponents of individual property rights in forested frontier areas argue that secure tenure arrangements should induce more rational forms of exploitation of natural re sources and a reduction in deforestation rates (e.g., Southgate and Wh itaker, 1992). Establishing individual property rights does not automatically lead to reduced deforestation, however, especially when the prof itability of agricultural and ranching activities outweighs that of forest management (Jaramillo and Kelly, 1997) Restricted individual property rights have been suggested as one way to protect forest resources on private property (Jaramillo and Kelly, 1997). Most Latin Americ an countries place restrictions on tree cutting and require permits for any type of fore st use. Many also require that riparian 25

PAGE 26

forests and forests on steep slopes be strict ly protected. Braz ilian and Paraguayan legislation goes one step further in requiring private landowners to maintain a certain percentage of the native forest or woodland vegetation in a type of private reserve. In Brazils Legal Amazon, 29% of the regi on is designated as indigenous land and protected areas, about 45% is c onsidered public untitled land ( terra devoluta), and private lands make up approximately one quarte r of the total area (Lentini et al., 2003). In contrast, in Mato Grosso, private lands comprise more than half of the total area, whereas indigenous reserves and state or feder al protected areas constitute a total of 18% (Congresso Nacional, 2000). Thus, fore st and woodland protection on private lands is of paramount importance in this r egion. However, it is unclear whether the Brazilian Forest Codethe primary instrument for protecting forests on private lands is adequate to protect common pool resource s, given the huge incentives to remove native vegetation in the face of agro-industria l growth, particularly following the increase in the proportion of private reserve to be maintained by each landholder. In this chapter, I present the results of an assessment of the legal and ecological effectiveness of the revised Fo rest Code. This assessment consists of two parts, a qualitative analysis and a quantit ative analysis. First, I describe the characteristics of ecosystem services and discuss the implicati ons for effective policy structures for ecosystem services. Next, I describe the hist ory and requirements of the Forest Code to gain an understanding of the policys char acteristics with respect to governing ecosystem services. Finally, I carry out a quantitative analysis focusing on three aspects of compliance with the Forest Code: (1) the level of compliance with the Forest Code for a landscape on the southeastern Amazon frontier; (2) the cost of compliance to 26

PAGE 27

landowners; and (3) the protection of ecosystem services and the conservation of forested landscape integrity at three moments in history and under full compliance with the pre-1996 and post-1996 legal reserve re quirement in modeled landscapes. Governance of Common Pool Resources The fundamental premise of the Forest Code is that fo rests and forest resources constitute a common good and t herefore, the right s of land owners to dispense with the land as each sees fit must be balanced with a set of responsibilities to protect this common good. At the time the first version of the Forest Code was drafted and signed into law in 1934 (described more extensiv ely below), there was widespread concern with erosion, loss of soil fertility, and the sedimentation a nd degradation of water bodies (Dean, 1995). In addition, when the legal re serve was first introduced, its original objective was to ensure that a minimum reserv e of forest resource s would be available on each property to supply local firewood, charcoal, and timber markets (Azevedo, 2009). Early on, the Forest Code also cited concern for the conservation of native species of flora and fauna. The major innovation (particularly for its temporal context) of the Forest Code is that it defined forests as a common good in relation to the services that flow from them because these services including the preventi on of sedimentation, the maintenance of water quality, and the cons ervation of native ecosystems, benefit people in society other than t he private landholder. As a c onsequence, the expansion of agro-industry in the Amazon constitutes a dilemma of the commons (Hardin, 1968; Ostrom et al., 1999), in which the costs and benefits of the use of natural resources common to many are not equi tably distributed among the users and inhabitants of the region. The costs include the loss of eco system services due to the large-scale conversion of native vegetation to pasture and cultivated lands that were not yet 27

PAGE 28

understood at the time the Forest Code was drafted, including increased carbon emissions and potential destabi lization of regional hydrol ogical and climate systems (Werth and Avissar, 2002; Alencar et al., 2004), and the displacem ent of traditional peoples and small producers (Hecht, 2005; ISA, 2005). The defining features of common-pool resources, such as the resources and services provided by forests, are that they are non-excludable and rival resources. An excludable resource is one whose ownership allows the owner to use it while simultaneously denying others the privilege (Daly and Farley, 2006). In contrast, when no institution or technology exists to make a good or service excludable, it is known as a non-excludable resource. Whereas the individual trees constituting the forest may be considered as excludable, the services that flow from them (e .g., maintenance of hydrological and climate regimes, clean air and water, h abitat) are non-excludable. Rivalness on the other hand, is an inherent characteristic of certain resources whereby consumption or use by one person reduces t he amount available fo r everyone else, or reduces the quality of the resource. Thus, a rival resource is one whose use by one person precludes its use by another, and a nonrival resource is one whose use by one person does not affect its us e by another. Unlike excludabilit y, rivalness is a physical characteristic of a good or service and is not affected by human institutions. Obviously, if trees are harvested for ti mber, they cannot be used by another person in the same way. However, ecosystem services that are maintained by forests are not traditionally considered to be rival, as the use of air by one person, for example, does not diminish the amount of air available for use by anot her. However, the quality or quantity of a service may be diminished by the use of the ri val resource (trees, in this case) from 28

PAGE 29

which the services flow. As such, it is usef ul to consider the services maintained by forests as rival, as well. The combination of these characteristics me ans that substantial free-riding is likely if people follow their own short-term interests exclusively, as assumed by Hardin (1968). Free-riding refers to natural resource use that maximizes private gain to the detriment of the common good and takes the form of (1 ) overuse without conc ern for the negative effects on others, and (2) a lack of contribut ed resources for maintaining and improving the common pool resource itself (Ostrom, 1990). To avoid these outcomes, effective rules to limit access and to define users rights and responsibilities regarding the resource must be established, and incentives for users to invest in the resource instead of overexploiting it must be created (Ostrom et al., 1999). In this paper, I consider the extent to wh ich these two conditions of limiting freeriding (effective rules and incentives) are met with respect to protecting forests. I specifically discuss the rules and issues relat ed to the Forest Code and its application in Mato Grosso in the subsequent section. Howeve r, it is useful to point out some general issues regarding environmental governance in Brazil and the Amaz on that affect the implementation and effectiveness of the Forest Code as we ll as other environmental legislation. Environmental Governance in Brazil In Brazil, the use of regulatory inst ruments has been the dominant policy approach to environmental problems. Regulatory (oft en referred to as command-and-control) instruments typically define performancebased or technology-based standards with which producers or landowners are required to comply by law. Regulation can take a variety of forms, including the outright prohibi tion of an activity or substance (e.g., the 29

PAGE 30

use of the pesticide DDT), or setting limits for the amount of a pollutant that may be produced (e.g., arsenic in drinking water), or defining the terms by which natural resources can be extracted (such as the l ength of the fishing season or the type of equipment that may be used in the fishery (Sterner, 2003). Failure to comply with regulations generally involves fines or other penalties. Thus, monitoring and enforcement are critical compon ents of direct regulation. This may be relatively easy in some cases, for example, where regulators mu st simply check that a firm or individual has installed a required piece of equipm ent. However, monitoring becomes more complicated in cases where compliance with regulations is more difficult to observe (e.g., non-point source pollution). Mor eover, if adequate numbers of trained personnel for monitoring and enforcement are not ava ilable, and if the force of law cannot be brought to bear on violators, adequate or sufficient moni toring and enforcement may become unviable. Critics argue that environmental policy bas ed on regulation tends to be inefficient because it generally does not capitalize on t he rent-seeking behavior of resource users (Sterner, 2003). Direct regulation is accused of being too prescriptive, and not providing enough flexibility to firms and individuals to innovate and meet standards in the most cost-effective way possible. Regulations are often ineffective, as well, because of budget short-falls and the complicated archit ecture of carrying out monitoring and enforcement, as well as sometimes sizeable economic incentives to circumvent legislation. Moreover, regulat ions may also create perverse incentives. For example, when ambitious legislation is not acco mpanied by adequate enforcement, legal resource users are, in effect, penalized re lative to the fraudulent resource users who 30

PAGE 31

incur lower costs of production (Carter, 2001; Tietenberg, 1996). Nevertheless, direct regulation may be the most appropriate me thod for integrating the biological requirements of some natural resources into environmental policy. For example, fish should not be caught during spawning s eason and timber harvesting should not eliminate the seed sto ck of tree populations. Although Brazilian legislation is consi dered to be among the most sophisticated and advanced in the world (Hochstettler and Keck, 2007), natural resource and environmental policies have been difficult to im plement and enforce in practice. This can be explained in large part by the mutually reinforcing charac teristics of (1) the Brazilian legal systemwhich follows the civil law traditionand (2) the command-and-control nature of most environmental l egislation in the country (C ampari, 2005; Ames and Keck, 1998). In civil law systems, the legal code cons ists of ideals and principles written as rules of law which are typically changed only by legislative action, in contrast with the common law system in which laws and legal procedure are progressively constructed in decisions made by judges and juries in succe ssive cases. For this reason, Brazilian environmental law tends to be idealistic and am bitious, but may be difficult to implement and enforce because the theoret ical truths represented in the law may not be compatible with social, ec onomic, and political realities among resource users. Furthermore, because laws are often perceiv ed by their creators and supporters to represent moral ideals, the conc ept of providing incentives to groups or individuals to facilitate or encourage compliance may be seen as antithetical. Finally, the design of laws as principles in the abstract stands in direct opposition to t he concept of negotiated 31

PAGE 32

solutions which are often necessary in co mplex natural resource problems and are more characteristic of the common law system. The command-and-control nat ure of legislation magni fies the importance of monitoring and enforcement, especially c oordination among agencies responsible for carrying out monitoring and enforcement, cl ear and consistent communication with resource-users regarding the rules, and efficient and appropriate monitoring systems. Unfortunately, environmental regulation in Brazil has been marked by poor coordination among responsible authorities (Mello, 2003), lack of human and financial resources (Pasquis, 2004; Lustosa et al., 2003), and a pervasive perception of environmental policy as generally obtrusive to economic development and, thus, of significantly lower priority than many other policies (Ascher, 1999; Mello, 2003). Many key policy decisions that affect natural resources are still primaril y the responsibility of other ministries, such as the Ministry of Agriculture, the Ministry of Transportation, the Ministry of Agrarian Development, and the Ministry of National Integration and Planning, which often leads to direct inconsistencies with environmental l egislation. Moreover, the responsibility for developing and implementing specific envir onmental regulations is divided among federal, state, and municipal authorities, o ften without sufficient clarity about which agency is responsible for which issues and with inadequate communication and coordination between these authorities (Ascher, 1999; May, 1999). Even where responsibility is clearer, lack of personnel or funding for enforcement can hamper policy implementation. Ascher (1999) also obser ves that government policies frequently are not designed to encourage landholders or land-users to invest in good land 32

PAGE 33

management, or instead provide incentives to invest in inappropriate technologies or methods. Effective governance also requires that t he rules of resource use are generally followed, with reasonable standards for tolera ting modest violations. As already pointed out, Brazilian environmental legislation itself typically provides few incentives for resource users to comply and resource s for enforcement are often lacking. Compounding these shortcomings, violations have tended to be overlooked, in large part because the judicial system is ill-equipp ed to prosecute violators (Ames and Keck, 1998). In 1998, the federal government signed into law a new Environmental Crimes Law which greatly broadens liability for envir onmental violators. The new law improves the ability of administrative agencies to apply administrative sanctions, establishes the responsibilities of corporations for envir onmental violations and damage, turns more environmental violations, such as illegal logging, into crimes with higher penalties (up to USD 16 million), and provides quicker judi cial procedures for many environmental crimes (Brito and Barreto, 2006). However, Br ito and Barreto (2006) conclude that the main obstacles to the enforcem ent of this law are difficultie s in locating violators and the lack of effective communication among the agencies responsible for applying the law, which results in delayed prosecutions. The Brazilian Forest Code Current Requirements The Brazilian Forest Code r egulates the use and conservation of forests and other native vegetation types on rural properties through three principal mechanisms: (1) Permanent Preservation Areas (APPreas de Preservao Permanente ), (2) the Legal Reserve ( RLReserva Legal ), and (3) hillslope forests APPs are designed to 33

PAGE 34

protect the most ecologically sensitive areas, and thus protect (1) the riparian vegetation around streams, rivers, lakes and artificial re servoirs, (2) slopes with greater than 45 inclination, and (3) hilltops Vegetation in the APP zone may not be harvested or cleared, except in exceptional situations defined by law (for pub lic utility or social interest). Where riparian v egetation has been cleared in excess of the limits defined within the Forest Code, the APP vegetation mu st be restored. The size of the APP zone within individual properties depends on the frequency and width of streams, springs, and on the topography of the property. The legal reserve constitutes a percentage of the area of the property in which native vegetation must be maintained; t he percentage is defi ned by the biome and region in which the property is located. In contrast with APPs, the RL may be sustainably harvested subject to an appr oved management plan that maintains the ecological function and composition of the native vegetation. Where native vegetation has already been cleared in exce ss of the limits stipulated by the Forest Code, property owners are required to restor e the RL. The current Fore st Code (modified in 1996 and officially adopted in 2001) determines that the legal reserve should constitute 80% of the area of the pr operty for properties located in t he forest biome of the Legal Amazon 2 35% in the cerrado biome of the Legal Amazon, and 20% in other regions of the country. Finally, the Forest Code also limits the use of forest on slopes between 25 and 45 inclination. It stipulates that only sust ainable forestry is allowed on these slopes, 2 The Legal Amazon extends beyond the natural boundaries of the Amazon River basin to include the entire administrative jurisdiction of the states of Mato Grosso, Tocantins, Maranho, Par, and Amap as well as those states that fall completely within the basin (Amazonas, Roraima, Acre, and Rondnia). 34

PAGE 35

and that these areas may not be counted as part of the Legal Reserve, nor do they constitute part of the APPs. In theory, the legal reserve, t he APP, and the hillslope forests are to be maintained separatel y and independently. For landowners having riparian and/or hillslope forests, the total area of land that must be set aside in restricted or sustainably managed reserves may sum to more than 80% of the property in the Amazon forest biome. History As noted previously, the fundamental premise of the Forest Code is that it defines forests and the services that flow from them as a common good and recognizes that private property ownership constitutes a bundle of both rights and responsibilities. The balance between rights and responsibilities has been the focus of intense and frequent debate between the farming and ranching l obby and environmental and social organizations in Brazil for at least 75 years. Although Braz il already had some regulations requiring the protection of forest resources on private lands prior to the 1930s, these were not formally codified until 1934. The Forest Code of 1934 ( Decretolei no 23.793 January 23, 1934) for the first time made protection of forest resources a responsibility of land-holders; previously fo rests were seen as the object of utilitarian rights accruing to property-holders. The firs t Forest Code required that property-owners maintain protection forest s serving a similar functi on as the permanent protection areas (APPs) of the modern Forest Code: to conserve hydrological functions, prevent soil erosion, support frontier defense, guarantee public health, protect sites of natural beauty, and provide protection for rare native s pecies of flora and fauna. Cutting of trees was strictly prohibited in these forests (Zakia, 2005). 35

PAGE 36

In 1946, the Brazilian Constitution was am ended to allow expropriation of private lands for agrarian reform if properties were deem ed not to fulfill their full social function. Although not formally enshrined in Brazilian law until 1946, the social function doctrine has been an important factor in property cl aims from the beginn ing of Portugals colonization of its South American territory (Alston et al., 1999; Colby, 2003; Ankersen and Ruppert, 2006). Until the Lei da Terra (Land Law) was enacted in 1850, individuals could legally declare posse (de jure usufruct rights; de facto, possession) simply by clearing land and establishing pasture or agriculture (Benatti, 2003; Ankersen and Ruppert, 2006). The justification for this clai m was that the actions of the claimant helped the land to fulfill its social function. However, the fear of expropriation on this basis also put the social function doctrine in direct conflict with the 1934 Forest Code, creating a disincentive for compliance with the Forest Code. To address this conflict, the Land Statute of 1964 ( Lei 4.504/64 ) accorded an environmental function to private properties for the first time, expanding the notion of productive lands to include the conservation of natural resources. One year later, the modern Forest Code ( Cdigo Florestal de 1965 Lei no 4.771 September 15, 1965) was written into law. It introduced the conc epts of the legal reserve (where sustainable timber harvests were permitted) in contra st with the permanent protection area (where strict protection is required, adapted from the protection forests of the 1934 Forest Code) and codified the notion that forest s are a common good and as such impose limits on the rights of the property owner. Furthermore, the new Forest Code created restrictions on the use of pr ivate property that would be enforced through fines and other penalties. The 1965 Forest Code determined that the Legal Reserve (RL) should 36

PAGE 37

constitute 50% of the area of the property for properties located in the Legal Amazon and 20% in other regions of the country. In 1989, the Forest Code was amended to reduce the RL to 20% in the cerrado and non -primary forest biomes in the Legal Amazon, to accommodate the agricultural ex pansion into the Center-West region of Brazil, particularly Mato Grosso (Azevedo, 2009). In 1995, deforestation in the Amazon reached a record high of over 29,000 km2 (INPE, 2009). Under international and domes tic pressure, the government undertook several policy changes. First, it declared a two-year moratorium on the harvest of mahogany. Secondly, and more importantly at the end of 1996, it adopted the provisionary measure ( medida provisoria) MP 1.511, increasing the RL on rural properties in the Amazon forest biome from 50% to 80% and prohibiting new clearing on properties already possessing abandoned or under-utilized areas or areas used inappropriately with respect to the capacit y of the soil. The MP 1.511 provoked an intense reaction from the agro-industrial lobb y; over the next 5 years, the measure was modified repeatedly as the agricultural and environmental lobbies battled back and forth, leaving landholders in doubt regarding their lega l obligations. After repeated substantial changes, in 2000, the federal governm ent re-edited the provisionary measure in favor of the Na tional Environmental Councils (CONAMA) recommendation, setting out the following requirements: (1) an RL of 80% for properties in the Amazon forest biome, 35% for t hose in the cerrado bi ome within the Legal Amazon, and 20% in all other areas of the country; (2) the area of APP could only be counted toward the RL in the 80% zones; (3) subject to specific conditions set out by state zoning plans ( Zoneamento Socio-Economico Ecologico ), the RL could be reduced 37

PAGE 38

to 50% from 80% in the Amazon forest bi ome; (4) the tradable legal reserve rights scheme required that an RL be compensat ed in another area of equal extent and ecological function and character within the same micro-basin (Chomitz, 2004); (5) special conditions for small properties (le ss than 100 ha) allowed only smallholders to restore their RL using exotic species, allowe d smallholders to subtract the area of APP from the RL when the APP exceeded 5% of the area of the parcel, and allowed smallholders to carry out sustainable agrof orestry management activities in the APP. CONAMAs text was maintained through MP 2.166 in 2001, at which point the National Congress passed a constitutional amendment limit ing the power of the executive branch of the federal government to edit provisionary measures. Provisionary measures that were active at that time became law. Relevance of Other Policies At the same time that the Forest Code was undergoing almost constant modification, environmental regulation was mo ving towards greater decentralization as part of an overall shift in governance structur e in Brazil in response to a mandate set out by the 1988 Constitution (Azevedo, 2009; Weiss, 2000). In 1997, CONAMA 3 approved a resolution requiring rural environmental licensing to be developed and implemented by states as a means of ens uring that the Forest Code would be carried out; this was part of an overall legal decision that states should elaborat e their own forest policies and regulate rural environmental licensing (related to ranching and agricultural production, especially deforestation, agrochemicals, and soil management). Although 3 Under current Brazilian law, CONAMA must approve any environmental legislation or policies (in addition to approval by the executive and/or legislative branches) before these policies can be officially enacted and implemented. 38

PAGE 39

the Federal Pact ( Pacto Federativo ), which turned greater decision-making and regulatory power over to states, was not signed until the beginn ing of 2000, Mato Grossos state envir onmental agencythe Fundao Estadual do Meio-Ambiente (FEMA)had already gained cont rol over implementation of the Forest Code (among other environmental legislati on) from IBAMA (the arm of the federal Environmental Ministry responsible for implementing and enforc ing all environmental legislation prior to decentralization) by 1999 (Azevedo, 2009). Simultaneously, FEMA began developing an environmental licensing system for private rural properties ( Sistema de Licenciamento Am biental para Propriedades Rurais SLAPR) as a means of monitoring co mpliance with the Forest Code, and particularly, of differentiating between legal and illegal deforestati on. Although INPE had been monitoring deforestation in the Amazon (PRODES) since 1988, the system was unable to discriminate legal and illegal clea ring. The system was implemented in 2000 and 2001. Now in control of implementation and enforcement of the Forest Code, Mato Grosso determined that it would require la ndholders within the Amazon forest biome in the state to observe a legal reserve of only 50% rather than the 80% required by the federal Forest Code. FEMA cited Mato Grossos state zoning plan as justification for this interpretation, despite the fact that the plan had not been legally approved by the state assembly yet (Azevedo, 2009). This provis ion was maintained until 2005, when the federal governments Operao Curupira extinguished FEMA and the federal government again exerted more control over environmental regulation in the state, including requiring establishment of a new state environmental agency, the State Environmental Secretariat (SEMA Secretaria Estadual de Meio-Ambiente ). 39

PAGE 40

Additionally, landholders faced uncertaint y regarding the definition of biomes and their boundaries, and thus, the proportion of a given property to be set aside, since the legal reserve requirement is biome-spec ific. The cerrado and forest biomes are frequently redefined in the Legal Amazon. That is, since the Forest Code specifies the percent of a rural property that must be ma intained in a legal reserve of the native vegetation, and this percentage is not the sa me for all biomes (i.e., 35% for cerrado, 80% for forest), landowners, lobb yists, and politicians often pursue legal redefinition of a vegetation type in regions where the di stinction between cerrado and so-called transition forests is not easily recognized and well-established. In general, the percent of vegetation required to be maintained is est ablished in the political arena, with little (if any) input from scientific asse ssment. Many properties located in what is considered to be transition forest were registered with a legal rese rve amounting to 20% of the property before 1996 since this vegetation type was not considered to be forest and was subjected the Forest Code am ended in 1989. In some cases, properties may be located in more than one biome, creating additio nal confusion regarding the percent and location of the legal reserve on the pr operty that must be maintained. Application of the Forest C ode on private rural properties is subject to conditions set out in the Agricultural Law of 1991 ( Lei no 8.171 1991). It established a term of 30 years for the property owner to restore the RLwhere it does not meet the minimum percentageon his land. Althoug h its goal is to protect t he landholder by assuring a generous period of time to comply with his le gal obligations under the Forest Code, it also states, unequivocally, that the rural pr ivate properties must comply with the APP and RL requirements. The Agricultural Law also exempts landholders from paying the 40

PAGE 41

rural property tax (ITR Imposto sobre a Propriedade Territorial Rural ) on the APP and RL areas of their land. Essentially, the Law thus assigns an economic value to the environmental function provided by these prot ected areas and is designed to serve as an incentive to reinforce the r equirements of the Forest Code. Materials and Methods Study Area The 177,780 km2 Xingu River headwaters region is located in the northeastern corner of Mato Grosso state, in cent ral Brazil (Figure 2-1). The regions soils, topography (100-300 m altitude, with flat interfluvial expanses) and climate are wellsuited for soybean production and cattle ranching. Native vegetation types in the region are comprised of forests (tall evergreen, transitional semi-deciduo us, and riparian) and savannas (cerrado woodland, mosaics of grassland, thickets, gallery forests). Ten indigenous territories are completely c ontained within the boundaries of the Xingu watershed in Mato Grosso (Figure 2-1). Indigenous territories cover approximately 42,200 km2 within the basin, representing 24% of the total area of the headwaters region. Private landholdings comprise nearly 70% of the total area, and smallholder settlements 4 comprise less than 5% of the region. The streams and rivers of the major protected forest area that lies at the center of t he regionthe PIX co mplex, which alone comprises nearly 20% of the headwaters areaare under gro wing threats from sedimentation, agrochemical run-off, and associated fish die-off from the unprotected headwaters regions outside of the park boundaries (Sanches, 2002). The Xingu region 4 Here, smallholder settlements are defined as government-sponsored settlement projects ( assentamentos) consisting of contiguous, individual 100-ha lots which are distributed to individuals or families who meet a number of criteria, including not owning any other real estate nor having the means or prospects to own any real estate. 41

PAGE 42

is representative of many areas along the Am azons agricultural frontier, but faces a more acute and immediate threat because it lies between two major federal highways (BR-158, BR-163) that are partially paved, and lies in the pathway of the northward expansion of Brazils grain belt. Land Cover Maps I developed maps of observed landscapes for 3 dates (1996, 1999, 2005) and 2 theoretical landscapes refl ecting land-cover under pre1996 and post-1996 Forest Code requirements. I used these maps to assess co mpliance with the requirements of each of these 2 versions of the Forest Code: (1) pre-1996, requiring 50% legal reserve in the forest biome (hereafter, re ferred to as pre-1996), and (2) post-1996, requiring 80% legal reserve in the forest bi ome (hereafter, referred to as post-1996). I also used these maps as a basis for assessing rate and proportion of legal and illegal deforestation following the change in the Forest Code. Furthermore, I calculated both the area available for conversion and in need of restoration re lative to the requirements of each version of the Forest Code. Finally I also combined them with maps of net present value for the region to es timate the costs of compliance. Observed landscapes I developed maps of land-cover for 3 dat es of significance for the change in legislation to assess the extent to whic h the change in the legal reserve slowed deforestation in the region: 1996, 1999, and 2005. 1996 is the year in which the legal reserve requirement in the Amazon Forest biome was raised from 50% to 80% of private property through a provisional meas ure. Mato Grossos state environmental agency took over responsibility for monitoring and enforcement at the beginning of 2000. Since only dry season imagery from th is region is cloud-free, it was more 42

PAGE 43

conservative to use images from July/Augus t 1999 than from the same period in 2000. 2005 is the year in which the federal gov ernment dissolved the Mato Grosso state environmental foundation (subsequently reor ganized as the state environmental secretariat) and mandated that landholders in Mato Grosso follow the requirements of the Forest Code to the letter of the law, rather than as interpreted by the state government since 2000. Maps with 4 classes (for est, cerrado, agricultural lands, other) were classified as described in Appendix A. Theoretical landscapes Two theoretical landscapes corresponding to the requirements set forth by the Forest Code before 1996 and after 1996 were developed using a spatial landscape simulation model. The basic architecture and function of the simulation model is described in Appendix B; details regardi ng the modeled landscapes follow here. The assumptions underlying each of the two scenario s differed only in the percent of native vegetation that was to be maintained or re stored on each private land-holding in the headwaters region. Since a complete map of property boundaries for the region is not available, I used micro-basins representi ng individual stream reaches (1:1,000,000 scale) as proxies for individual properties. T he mean, range, and distribution of sizes of the 2881 microbasin ( x = 5981 ha, 4-70,766 ha) is compar able to that of private properties in the region (Jepson, 2006; Fearnside, 2005; Appendix B). Furthermore, the mean percent clearing in the current (2005) landscape is comparable among microbasins and properties for which property lim it data are availabl e, indicating that microbasins are suitable substitutes for indi vidual properties in te rms of sampling the population of properties in t he region. It was necessary to use a spatial unit with 43

PAGE 44

complete coverage of the study region to si mulate the distributi on of vegetation across the landscape corresponding with the requirem ents of the preand post-1996 versions of the Forest Code. For both scenarios, full compliance wit h the law was assumed. All indigenous reserves and state and federal protected areas were strictly protected. Furthermore, a 50-m riparian buffer zone surrounding each str eam and river visible in a map derived from a thematic stream layer obtained from the Mato Grosso Stat e Regional Planning Secretariat (SEPLAN-MT) was strictly protec ted. The Forest Code stipulates that a riparian buffer zone of at least 50 m be prot ected around every natural body of water, and that the size of the bu ffer zone (up to 500 m) is de pendent on the width of the stream. However, as stream width is difficu lt to assess and no official map of riparian buffers exist, the estimation of variable-wid th riparian buffers was not possible and I assumed, conservatively, that all water bodies were surrounded by a 50-m wide buffer. Where necessary, vegetation was restored to this riparian buffer zone so that each theoretical landscape had 100% native vegetation cover within the boundaries of the riparian zone. The original 1965 Forest Code (pre-1996) required that the legal reserve be maintained in addition to the riparian zone (i.e., the riparian zone could not be counted toward the percent of vegetation to be maintained in the legal reserve). Various versions of the post-1996 provisionary meas ure alternately had the same requirement or allowed the riparian zone to count towards the legal reserve. The final version of the provisionary measure that was adopted in t he federal legal code required that each be maintained independently and completely, although subsequent interpretation at federal and state levels varied. To facilitate anal ysis of compliance with the legal reserve 44

PAGE 45

requirement separately from the riparian zone requirement under the two iterations of the Forest Code, I did not allo w the riparian zone area to count as part of the amount of legal reserve required in each microbasin in either the preor post-1996 scenario. To calculate the amount of deforestation or restoration that could or should take place, respectively, outside the riparian zone to meet the legal reserve requirements under each of the two scenarios, I classifi ed each micro-basin according to biome (cerrado or forest). The cerrado-forest bi ome map was obtained by merging a map of forest-non-forest derived from INPE Prodes maps with a map of biomes from the IBGE RADAM vegetation thematic map. Of 2881 micro-watersheds in the Xingu River headwaters region, 34 straddl ed both biomes. To facilitate model design and processing, each of these microbasins was assigned to the biom e representing more than 50% of that micro-watersheds area. At time step 0, the m odel calculates how much of each watersheds area consists of cl eared area. If this area is greater than the allowed amount (as determined by each version of the Forest Code) after the area of the riparian zone is subtracted from the to tal watershed area, the model reforests the area up to the allowable amount of cleared ar ea. If the cleared area is less than the allowable area, the model beg ins to deforest the watershed based on where the highest favorability for deforestation is indica ted (Appendix A). When a microbasin included protected areas, the microbasin was subdivided so that the part outside of the protected area was treated as an independent microbasin subject to the rules described above. The part in the protected area was prohibited from being cleared and did not count among those microbasins serv ing as proxies for private properties. 45

PAGE 46

Compliance To assess degree of compliance with the Forest Code for each of the three observed dates, I calculat ed the area and quantified the perc ent of remaining native vegetation by biome for the entire headw aters region outside protected areas and compared this to the amount required under each the preand post-1996 versions of the Forest Code. I also calculated the area and percent of remaining native vegetation for each microbasin located outside protected areas. I present the mean and standard error for the population of mi crobasins in each biome at each date. For each year, I assessed the absolute amount and the percent of vegetation remaining in the 50-m riparian buffer zone at both the landscape a nd microbasin levels. I also present the number of microbasins comply ing with both the legal re serve and the riparian zone requirement under each version of the Forest Code. Finally, I assessed the transitions of forest microbasins among three categories of compliance: (a) compliance with neither legal reserve regulation (below 50% forest cover), (b) compliance with the 50% legal reserve requirement (between 50 and 80% forest cover), and (c) compliance with both regulations (greater than or equal to 80% forest cover). Since the microbasin boundaries do not coincide with those of actual properties in the region, the results must be interpreted with care. Because of the si milarity in the range, mean size, size distribution, and percent clearing of t he microbasins compared with the group of properties for which data are available, it is possible to assume that the results presented here represent the overall trends likely to be seen in a survey using property boundaries. Nevertheless, some differences are likely to be present and the results presented here should be seen as describing gener al trends rather than specific results about the location and amount of actual compliance. 46

PAGE 47

To assess the extent to which the change in the Forest Code slowed illegal deforestation in the forest biome, I esti mated the amount and rate of deforestation outside protected areas in the Xingu River headwaters region for two periods: 19961999 and 1999-2005. I identified ea ch pixel that underwent deforestation during each of the two time periods. Pixels were then fu rther classified by whether they were considered to be in or out of compliance with each of the legal reserve requirements of each of the two iterations of the Forest Code. The final classification allowed pixels to fall into one of 3 categories in eac h of the two periods, as follows: Deforestation, legal : forest clearing in microbasins that were in compliance with the regulations at the beginn ing of the period and remained in compliance at the end of the period; Deforestation, illegal, new non-compliance : forest clearing in microbasins that were in compliance with the regulations at the beginning of the period and became non-compliant by the end of the period as a result of clearing; Deforestation, illegal, continued non-compliance : forest clearing in microbasins that were out of compliance with the regulations at the beginning of the period and remained out of comp liance at the end of the period. Economic Aspects of Compliance To evaluate the economic aspects related to compliance with the new Forest Code requirements in the forest biome for each of the three dates, I calculated the difference in the area of the watershed outside of federal and state protected areas (1) that would no longer be available for conversion to agric ultural land, and (2) that would need to be restored to come into compliance under the new regulations (80% legal reserve in forest biome, 35% in cerrado) in comparison with t he old regulations (50% in forest, 35% in cerrado). For each, I present the total area for the entire headwater s region as well as the mean per microbasin. Based on these figur es, I estimated the economic costs of 47

PAGE 48

complying with the policy change by estimati ng (1) the potential forgone profits to producers over the whole landscape and by mi crobasin using net present value as a proxy for the potential value of agricultural lands in the region, and (2) the cost of riparian forest restoration over the w hole landscape and by microbasin, where necessary (methods are described in detail in Appendix C). I estimated total maximum potential NPV for both forested and cleared lands if they were to be cleared or remain cleared within the forest biome as well as per microbasin. Furthermore, I estimated potential NPV for microbasins in and out of compliance with the preand post-1996 Forest Code requirements to estimate the potential for legal agricultural expansion under the 50 and 80% legal reserve requirement. In each case, I also ca lculated and compared the mean NPV ha-1 to test the hypothesis that mean potential earnings were associated with non-compliance. Restoration costs were esti mated for riparian zone and l egal reserve areas, at the aggregate and microbasin levels. Ecological Consequences I compared the final landscapes for each of the 5 alternative landscapes in terms of carbon stocks, river disc harge, annual evapotrans piration, terrestrial habitat quality, and water quality (methods are described in Appendix C). Unlik e all the preceding analyses, ecological consequences were assessed for the entire headwaters region, including all protected areas. 48

PAGE 49

Results Compliance Legal reserve At the scale of the entire Xingu River headw aters region, forest cover was greater than 80% (the new legal reserve requirement ) in 1996, but fell below 80% in 1999 and further still in 2005. The region was in compli ance with the 35% cerrado limit in all three periods (Table 2-1). In 1996, when the pr ovisional measure was first adopted, 83% (79,910 km2) of original forest cover and 64% (18,617 km2) of original cerrado cover still remained outside protected areas across th e Xingu headwaters region (Table 2-1, Figure 2-1), thus meeting both the old (50% legal forest re serve) and new (80% legal forest reserve) regulations. By the middl e of 1999, 6 months before the policy was permanently adopted, original forest cover had fallen to 78% (75,139 km2), and original cerrado cover had fallen to 54% (15,967 km2). During the subsequent 6 years, to 2005, further deforestation reduced fo rest cover to 64% (61,934 km2) and cerrado cover declined to 45% (13,065 km2) (Table 2-1, Figure 2-1). Tota l forest loss on lands outside of protected areas from 1996 to 2005 equaled 17,976 km2. The annual average area of forest loss increased from 1590 km2 between 1996 and 1999 to 2201 km2 between 1999 and 2005. In the cerrado biome, tota l native vegetation loss reached 5552 km2 over the 10-year time peri od; mean annual clearing decrea sed by nearly half in the second period, from 883 km2 to 484 km2. These aggregate statistics mask the degr ee of compliance by individual landholdings (represented here by microbasins, which I employ as proxies) with the regulations. Landholdings are the unit for which enforcement of t he legislation takes place. In 1996, when the legi slation changed, 8% of fore st microbasins had cleared 49

PAGE 50

more than 50% of forest vegetation and were therefore not in compliance with the old regulations (Table 2-2). However, with the increase in the legal reserve requirement from 50 to 80% of the property an additional 20% of fore st microbasins fell out of compliance, representing a sudden increase in illegality of 336%. The combined area of cleared land in excess of the new limit ( 20% clearing permitted) in these microbasins was 2404 km2, 15% of all existing cleared area that year. Eight percent of microbasins located in the cerrado biome were also out of compliance with the 35% native vegetation cover legal reserve requirement which did not change (Table 2-2). In 1999, 36% of forest microbasins failed to meet the new requirement s, representing an increase in illegality of 297% relative to compliance with the old requirements (Table 22, Figure 2-2). Eighteen percent of cerr ado watersheds did not meet the legal requirements. By 2005, following the post-1995 spik e in deforestation in the region, from 2002 to 2004, 54% of forest microbasins did not comply with the new requirements, whereas 21% would have been out of compliance under the old regulations. In contrast, more than three-quarters of cerrado microbasins met the requirements. Changes in microbasin-level compliance ta ke place through different pathways. I examined the transitions of forest microbas ins among three categories of compliance: (a) compliance with neither legal reserve r egulation (below 50% forest cover), (b) compliance with the 50% legal reserve require ment (between 50 and 80% forest cover), and (c) compliance with both regulations (gr eater than or equal to 80% forest cover) (Figures 2-2, 2-3). In 1996, the percentage of forest microbasins in each of these categories was 8, 20, and 72%, respectively (Fi gures 2-2, 2-3, Tabl e 2-2). In 1999, the new allocation of microbasins among the thr ee categories of compliance was 12, 24, 50

PAGE 51

and 64%, respectively (Figures 2-2, 2-3, Table 2-2). The largest shift between categories (8%) occurred from compliance with both regulations (>=80% forest cover) to compliance with the old regulations (bet ween 50 and 80% forest cover). Of the microbasins with less than 50% forest cove r in 1996, 98% remained non-compliant in 1999, with 86% continuing to deforest (F igure24). From 1999 to 2005, the new allocation of microbasins among the three categories of compliance was 21, 33, and 46%, respectively (Figure 2-2). Sixteen perc ent of forest microbasins shifted from compliance with both regulations to compliance only with the old regulation. Eight percent of forested watersheds were out of compliance with the old regulations in 1996 and continued to deforest through 2005 (Figur e 2-4). These ill egal microbasins are primarily concentrated on the eastern side of the Xingu Basin, where the profitability for soy cultivation was highest. Forty-six percent of the microbasins still retained 80% or more of their forest cover in 2005 (Figure 2-2), and are concentrated on the western side of the basin (Figure 2-3, 2-4), where selective loggi ng constitutes an important part of the regional economy; this indicates, however that remaining forests are likely to be degraded (Asner et al., 2005). The microbasin-level analysis also permits quantification of t he area of forest clearing that was in compliance with the old and new legal reserve requirements. Between 1996 and 1999, a total of 4867 km2 (5% of the original forest area on private lands) were deforested by 68% of forest mi crobasins (Table 2-3). Fifty-seven percent of total microbasins (representing 39% of tota l deforestation in the period) deforested without falling below the 50% minimum forest cover requirement. As a result of the increase in the legal reserve requirem ent, a greater proportion of 1996-1999 51

PAGE 52

deforestation (31%) was carried out on 430 (24%) microbasins already out of compliance with the laws requirements (Tab le 2-3). In the second period, from 19992005 (after the policy change was officially adopted as law), a total of 13,184 km2 of forest were deforested by 86% of microbasins. Fifty-eight per cent of this deforestation (8386 km2) was in compliance with the old regulation and only 15% (2006 km2) with the new regulation (Table 2-3). Riparian buffer zone A separate provision of the Forest Code man dates that the ripari an zone be strictly protected and that vegetation be restored if any clearing has taken place in the buffer zone. In 1996, a total of 12% of the ripar ian zone in the headwater s region, outside of protected areas, had been cleared, with a greater proportion of cerrado riparian area cleared (18%, Table 2-1). Only 12% of ce rrado microbasins were in compliance, one third the proportion of that encountered in t he forest biome (Table 2-2). In 1999, the area of riparian zone cleared climbed to 16% (Table 2-1). Only 32% of microbasins were in compliance with the law at this ti me (Table 2-2). By 2005, 23% of the riparian vegetation had been cleared, and compliance fell to 17% of microbasins (Tables 2-1, 22). This decrease in riparian zone vegetati on occurred mostly in forest microbasins, where compliance fell to 20% (Table 2-2). Economic Aspects of Compliance Opportunity cost The increase in the legal forest reserve requirement from 50 to 80% imposed costs on landholders. The first cost that I consider is that associated with foregone profits from soy or catt le ranching incurred through compliance with the mandatory forest legal reserve requirements, which dep ends on the area of forest land landholders 52

PAGE 53

must leave standing. At the level of the ent ire Xingu River headwater s region, in 1996, nearly 33,000 km2 of forest could still be cleared by 1608 (92%) forest microbasins under the old regulations, whereas less than 10,000 km2 could still be cleared by 1259 (72%) microbasins under the new regulations in 1996 (Tabl es 2-2, 2-4). The average landholding still in compliance with the old regulations would have been permitted to clear a further 20 km2, whereas under the new regulations, only 7 km2 of further clearing would have been allowed (Table 2-4). By 2005, 20,000 km2 were potentially eligible for clearing under the old regulations in contrast with less than 4000 km2 under the new regulations. At the individua l microbasin level, these num bers had fallen to less than 15 km2 and 5 km2, respectively (Table 2-4). The total net present value (NPV, calculat ed over a 30-year time horizon) that could be achieved for lands outside of protected areas across the Xingu headwaters region, assuming that the maximum economic value were to be extracted from the landscape from soy or cattle ranching, r eaches 32.9 billion USD (Table 2-5). The potential NPV of the entire Xingu River watershed area out side of protected areas associated with soy and cattle production di ffers by approximately 5 billion USD between a landscape reflecting the old (50%) and new (80%) regulations (difference in NPV of forest area in compli ant microbasins, Table 2-5). This difference climbs to approximately USD 9 billion when the area of forest land available for conversion under the old vs. new regulation is compar ed for microbasins (T able 2-5). In 1996, producers in the aggregate still had the potential to realiz e an additional NPV of 11.6 billion USD under the old regulat ions, but only 2.9 billion US D under the new regulations (Table 2-5). By 1999, these values decreased to USD 9.6 vs. 0.9 billion, respectively, 53

PAGE 54

then to USD 4.5 and -4.2 billion in 2005 (Tabl e 2-5). Under the new regulations, land with a value of 4 billion had already been cleared illegally by 2005 (Table 2-5); this is income that producers out of compliance in the region would not have earned if they had complied with the new regulations. At the level of the individua l microbasin, the potential addi tional NPV that could be realized through legal clearing of land fell by an average of USD 4 million (Table 2-7). In other words, the change from a legal rese rve of 50 to 80% represented a USD 4 million dollar decline in the potential NPV of the regions landholdings. I tested the hypothesis that compliance with the legal reserve requirement is inversely proportional to the opportunity co sts of compliance by comparing the mean NPV ha-1 of cleared land for compliant and non-co mpliant microbasins (Table 2-6). The hypothesis was supported for all comparisons of NPV ha-1. The difference in NPV ha-1 between compliant and non-compliant microbasins became larger each successive year, and was greater for the 80% regulation than for the 50% regulation, where the per-hectare value in noncompliant basins remained 1.5 times greater than that of cleared lands in compliant microbasins (Table 25). Over time, this ratio increased, with the potential NPV ha-1 of cleared lands in non-co mpliant microbasins (3087 USD ha-1) nearly double that of cl eared lands in compliant microbasins (1772 USD ha-1) in 2005 (Table 2-6). Cost of restoration As the forest area available for clearing declined through the change in the Forest Code, the aggregate area of legal reserve to be restored in 1996 rose from 1200 km2 under the old regulations to 6500 km2 under the new regulations (Table 2-4). For the average individual microbasin already in viol ation of the old requirements in 1996, the 54

PAGE 55

legal reserve restoration require ment increased from 8 to 13 km2, going from the old to the new regulations (Table 2-4). By 2005, t he aggregate area to be restored increased to 3500 km2 and 16,000 km2, under the old and new regulations, respectively. By 2005, the average restoration requirement relative to the old and new requi rements rose from 9 to 17 km2. The total riparian area needing to be restored increased from approximately 1300 km2 in 1996 to 2500 km2 in 2005. For the average landholder, this meant an increase from 0.6 km2 to over 1 km2 of riparian forest re storation (Table 2-4). In 1996, restoration costs to bring all microbasins into compliance with the old regulations (including riparian zone restor ation) was 375 () million USD (Table 2-8). To come into compliance with the new regul ations, the cost more than doubled to 868 () million USD. These costs increased in 1999 to 534 and 1226 million USD, respectively; by 2005, the costs clim bed to 841 and 2018 million USD (Table 2-8). Whereas riparian restoration costs repr esented an average of 69% (under the old regulations) and 31% (under the new regulations) of the tota l restoration costs in 1996, by 1999 they had risen to 61% (old) and 27% (new) of the cost (Table 2-8). By 2005, these proportions had fallen to 51% (old) and 20% (new) of the cost. For the average individual landholder needing to come into comp liance, the cost rose from 1.2 to 1.6 million USD in moving from the old to new regulations in 1996 (Table 2-8). This cost rose in 1999to 1.4 and 2 million USDlargely because of a doubling in the area of riparian zone restoration. The overall average cost to l andholders increased further in 2005 (1.6 million USD to comply with the old regulations, 2.3 million USD to comply with the new regulations) due to an increase in t he amount of legal reserve restoration, despite the cost of riparian zone restorati on remaining stable. At the level of the 55

PAGE 56

individual landholder, riparian zone restoration costs repres ented a small proportion of the total restoration costs than for the landsc ape as a whole, never reaching more than 15% of the total costs (assuming a minimu m of active restor ation was carried out) (Table 2-8). Ecological Consequences Carbon stocks Carbon stocks of the Xingu River headwaters region forests varied by 212 million tons from the highest level, in 1996 (661 Mt C), to the lowest level, under the modeled 50% legal reserve regulation (449 MtC, Ta ble 2-9). Carbon stocks between the modeled 50% and 80% landscapes differ by 27% (120 MtC, Table 2-9). Wh ereas implementing the new regulations would have meant a loss of only 50 Mt C since 1996, retaining the old regulations would potentially have re sulted in emissions of 150 MtC (640 MtCO2e). Currently (2005), somewhat less than half t hat amount has been lost since 1996 (Table 2-9). Hydrology and regional climate Deforestation is known to reduce evapot ranspiration and incr ease stream flow because of the reduced leaf area index, decreased root density and depth, and increased soil compaction (Bruijnzeel, 1990; Costa, 2005; Sahin and Hall, 1996; Scanlon et al., 2007; Nepstad et al., 1994). In the Xingu River headwaters, modeled streamflow was higher than the original landscape in all five scenarios. Among the 1996, 1999 and % landscapes, mean annual stream discharge was similar ranging from 7 to 10% greater than that of t he control landscape. In the 2005 and % landscapes, stream discharge was higher, with 13 and 17% increases, respectively. 56

PAGE 57

This implies lower potential for flooding, and lo wer risk of overland flow and associated soil erosion under the 1996, 1999, and % landscapes. Mean annual evapotranspiration decreases as forest cover decreases on the landscape, ranging from a 3 to 7% reduction from the control scenario. Evapotranspiration decreases more than tw ice as much in the theoretical 50% landscape over the 1996 and 1999 landscape s and nearly twice as much as the theoretical 80% landscape. More water vapor, equal to 60 mm y-1, is therefore provided to the atmosphere under the high forest landscapes (1996, 1999, 80%) than under the low forest landscape. Water quality Whereas both the theoretical 50% and 80% landscapes had all riparian forests protected or restored ripari an forest cover was substantially lower (12 to 19%) in the historical landscapes. This suggests that str eams in the historical landscapes are more likely to be affected by sedimentation from point-source than those in the theoretical landscapes. Furthermore, by 2005, water temperatures are likely to be higher and dissolved oxygen levels lower in one-fifth of the land scapes stream network (Neill et al., 2006), affecting species populations and assemblages. Some of these streams may also be subject to grass invasion (Neill et al., 2006). Riparian zone forests ar e also important in the food chains of aquatic communities (N eill et al., 2006; Burcham, 1988; Bojsen and Barriga, 2002; Lorion and Kennedy 2009). Habitat Overall, habitat quality and quantity are lowe st in the theoretical 50% landscape, and highest in the 1996 landscape. The 1996 and 1999 landscapes are similar to one 57

PAGE 58

another, having the highest am ount of total forest and cerrado cover, the greatest mean fragment size (more than four times as la rge as the 50% landscape), as well as the highest amounts of interior (core) habitat ar ea (more than twice as much as the 50% landscape) and lowest amounts of edge habita t (nearly 3 times less than the 50% landscape). Notably, the theoretical 80% landscape has substantially more edge habitat (1.5 times more) than the most fragmented historical landscape (2005), despite having fewer fragments and a greater amount of interior habitat than that landscape (Table 29). This is likely to be explained by the spatia l stratification of legal reserves throughout the landscape in the theoretic al landscape, in comparison to the current landscape where some areas of the landscape are not yet fragmented and ot her areas no longer have any native (or regenerated) vegetation remaining. The Forest Code, if imposed at the level of the microbasin, increases the amount of edge relative to the natural pattern of forest cover that arises through frontier expansion. Discussion The Brazilian Forest Code was designed to pr otect public interests in private-land forests, an intent that was fully canoni zed in the new Constitution of 1988. To successfully protect the social function of private land forests, the Code must change the behavior of landholders. They must comply with the restrictions on forest clearing that are defined by the code, and adjust th is behavior when the Code changes. At the beginning of the study period, in 1996, most of the microbasin s (92%) of the Xingu River headwaters region, outside of protected areas, were in compliance with the requirement that at least 50% of each propertys forests be maintained (T able 2-2). With the announcement during this same year of a tempor ary regulation to raise the legal forest reserve requirement to 80%, the number of microbasins in compliance with the new 58

PAGE 59

requirement fell to 72%. This can be viewed as the beginning of the test of the effectiveness of the revised Forest Code. If the revised Code were viewed by landholders as carrying the full force of the law, then defores tation would have halted in microbasins that had already reached 20% fore st clearing, continuing only on those microbasins with more than 80% forest cover re maining. In practice however, the level of compliance with the 80% legal forest re serve requirement declined from 72% of microbasins in 1996 to 64% in 1999, then dropped precipitously to 46% in 2005, following the surge in clearing from 2002 to 2004 (Table 2-2). This illegal deforestation was manifested at the scale of the entire Xingu River headw aters region. Forest cover outside of protected areas declined from 83 to 64 % from 1996 to 2005 (Table 2-1). With the more restrictive Forest Code, annual deforestation in the region climbed from 1590 km2 during the 1996-1999 period to 2201 km2 during the 1999-2005 period, a further indication that the Code revision had little if any inhi bitory influence on forest clearing by landholders. It is not surprising that the level of compliance with the new legal reserve requirement was low. Compliance with environm ental regulation is highest when (a) the process for achieving compliance is clear and practical, (b) the probability of noncompliant landholders being identified is high, (c) the probability of apprehended landholders paying fines or facing imprisonment is high, and (d) the costs of compliance are low. In sum, compliance is highest when non-compliance is very expensive. The change in the Code was not accompanied by an effective program for helping landholders bring their properties into comp liance if they had al ready exceeded 20% clearing (28% of microbasins in 1996). It was only in 2000 that the requirement to bring 59

PAGE 60

properties into compliance began to be enforc ed through creation of the Mato Grosso rural environmental licensing system, coincidentally with the sudden change in legislation. Properties could be brought into compliance through restoration of forests to bring the propertys forest cover up to 80% compensation of the legal reserve through the purchase of excess forests on other properties (Chomitz, 2004; Azevedo, 2009), and payment into a fund that would maintain or expand state protected areas were created (Azevedo, 2009). Even in 2005, only 3992 km2 of forest could have been cleared legally and may have been set aside to compensate non-compliant properties (Table 2-3). However, the ar ea of illegal clearing by th is year had reached 16,000 km2, more than four times the area available for compensation (3788 km2). Hence, legal reserve compensation could have been used fo r, at most, one four th of the illegally cleared land. Very few landholders brought their properties into compliance through compensation on other proper ties; between 1999 and 2007, only 5 such applications were processed by Mato Grossos envir onmental agency (Azevedo, 2009). My data do not allow me to infer the degree to which landholders may have achieved compliance through payments into a state fund, although Azevedo (2009) i ndicates that it was also minimal. Compliance with the new forest Code may also have been low because of landholder uncertainty about its longevity. I identify two major sources of this uncertainty. First, the state of Mato Gro sso declared in 2000 that the legal reserve requirement for the transition forests, wh ich include approximately 76% of the forests in the Xingu River headwater region, was only 50%, despite the change to 80% for all forests in the Legal Amazon in the federal Forest Code. This state-level declaration 60

PAGE 61

went unchallenged by the federal government for 5 years, and was accompanied by a vigorous debate about the definition of transition forest As of 2005, the federal government over-ruled the state interpretation, and the entire forest biome returned to an 80% legal reserve requirement. Becaus e these percentages are decided in the political arena, rat her than through scientific assessmen t, some biomes (e.g., cerrado) and their ecological functions are undervalued re lative to others. Conversely, the legal reserve for other biomes (e.g ., forest) may be set higher than what is necessary to maintain ecological function and also result in low compliance as landholders attempt to gain parity with their peers in other biomes, ultimately to the detriment of the ecosystem. A second source of uncertainty regarding the longevity of the 80% requirement was the frequent attacks on the forest Code wit hin the Brazilian National Congress. The bancada ruralista (ruralist constituency, primarily the agricultural lobby) advanced proposals to reduce the legal reserve require ment of the Forest Code almost annually (Lima et al., 2005). The reducti on of the legal reserve re quirement was an important plank in the political platform of politici ans, agro-industrys representative organizations, such as FAMATO (Mato Grosso Agricultur e Foundation) and CNA (National Agriculture Confederation), and may have given landhol ders a sense of impunity regarding the Forest Code (Azevedo and Pasquis, 2006; Azevedo, 2009). Since 2005, Mato Grossos agro-industrial sector continues to seek ways to reduce the legal reserve and/or to legalize properties that have cleared in excess of the permi tted amount without requiring restoration (Camarga, 2008). These weaknesses in implementation of the Forest Code may have been reinforced by the costs incurred by landholders through compliance with the Code. The 61

PAGE 62

costs of registering with the state environmental licensing system (S LAPR) alone have been demonstrated to be prohibitive to many landholders, particularly if the landholder attempts to maintain a l egal reserve of 80% (Guimaraes and Almeida, 2007; Azevedo, 2009). Opportunity cost may present an even greater obstacle to compliance. The forgone profits from deforestation-dependent economic activities, such as cattle ranching and soy cultivation, were particularl y strong incentives for landholders to take the risk of getting caught and paying fines bec ause of the high potential profitability of the Xingu regions soils for agriculture and ranc hing. Using spatially-explicit rent models for soy cultivation and cattle ranching, I esti mate that the aggregate opportunity cost incurred by landholders through the increase in the legal reserve requirement was approximately USD 9 billion (T able 2-3). The mean opportunity cost for individual microbasins associated with the legal rese rve requirement shift was USD 4 million (Table 2-7). Further evidence of the contribution of opportunity costs to non-compliance is provided by the potential net present value of cleared land on a per-hectare basis for compliant and non-compliant properties. Me an NPV per hectare of cleared land was higher for non-compliant clearing than compli ant clearing, and this difference was greater for the 80% restricti on than for the 50% restriction, and it increased over time (Table 2-6). This suggests that landholders are more willing to break the law and run the risk of getting caught and punished if there is more profit to be made through illegal forest clearing. However, although it is uncl ear what proportion of violators were served with fines, of those who were, less than 2% of fines are estimated to have been 62

PAGE 63

collected, even after Mato Grossos environm ental authority was reorganized (Azevedo, 2009). In comparison to the forgone income from cleared lands, the costs to restore those lands are negligible, repres enting approximately 20% of t he opportunity cost of the lands to be restored in 1996 and 1999 under both the new and the old regulations (Tables 2-5, 2-8). In 2005, restorati on costs represented 13% and 18% of the opportunity cost of complying with the old and new regulations, respectively. This illustrates the point that t he opportunity cost is likely to be far more prohibitive for landholders in bringing illegally cleared lands into compliance, but that adding restoration costs to opportuni ty cost increases the total cost by one-fifth, on average. Although compliance with the change in the legal reserve requirement was low, simulation modeling allows us to understand what this policy decision would achieve if it were fully implemented. Under full compliance with the 80% forest cover, the level of biotic regulation (Bormann and Likens, 1979) of water flow through the Xingu River headwaters region increases. The higher level of evapotranspiration associated with the 80% legal reserve scenario relative to t he 50% scenario (3%, Table 2-9), although a small difference, signifies a lower potential for surface runoff and associated soil erosion (Stickler et al in press) and lower potential for stream and river flooding. Higher evapotranspiration also reduces the likelihood of deforestation-driven changes in the regional rainfall system, which some model s suggest can take place when clearing exceeds 60 to 70% original forest cove r (Sampaio et al., 2008; Werth and Avissar, 2002; Nobre et al., 1991). The Xingu headwaters region is likely to be severely impacted by rainfall reduction through climate change (Malhi et al., 2008) and the 63

PAGE 64

maintenance of high levels of evapotranspira tion could diminish the likelihood of these changes in rainfall (Nepstad et al., 2008). Bo th the 80% and 50% legal reserve scenario assume full restoration of riparian zone fore sts, potentially re-establishing the role of these forests in providing shade, fruits, and other forms of organic matter to the regions aquatic ecosystems. Riparian zone forests may prevent grass invasi on in streams that can be associated with low levels of oxygen (Neill et al., 2006). Improvements in water quality under a fully implemented Forest Code could have dire ct positive impacts on the livelihoods of the indigenous peoples w ho reside in the Parque Indigena do Xingu (Xingu Indigenous Park) located at the core of the headwaters r egion (Figure 2-1). Tribes such as the Kisedje have observed dec lines in the quality of the fish and turtles that they catch for subsistence consumption and have also noted changes in the timing and strength of rains at the beginning of the rainy season (pers. comm., Chief Cuiuci Kisedje, 16 July, 2005). O ne of the biggest impacts of the 80% scenario fully implemented is in the increase in interior hab itat of the regions forests (Table 2-9). Those species that depend upon forest interior habitats, such as ant birds (Bierregaard, 2001), would have more than twice the area of forest interior under the 80% scenario vs. the 50% scenario (Table 2-9). The 80% legal reserve scenario also retain s 120 million tons of carbon (450 million tons of CO2 equivalent (tCO2e), Table 2-9) more than the 50% legal reserve landscape. Carbon storage is the only ecosystem service in the Amazon region that is close to having a robust market mechanism to provi de incentives for its maintenance. If the REDD (Reduced Emissions from Deforestat ion and Forest Degradation) regime under negotiation within the United Nations Fram ework Convention on Climate Change 64

PAGE 65

(Gullison et al., 2007; Meridian In stitute, 2009) results in the development of a market for carbon through reductions in deforest ation, carbon could become a promising commodity for compensating l and holders for the opportunity costs that they incurred when the legal reserve was c hanged from 50 to 80% of each property. This opportunity cost, that we estimate at approximately USD 9 billion (Table 2-5), could be fully compensated at a price of USD 20 tCO2e-1 emission that is avoided; this price is within the range of current carbon pric es (approximately USD 21 tCO2e-1; Point Carbon, 2009). In sum, the Xingu headwaters region provid es insights into the limits of command-andcontrol regulations designed to defend public in terests in private-land resources. The change of the legal reserve r equirement from 50 to 80% of private properties imposed USD 9 billion in forgone potentia l present and future profits on the regions farmers and ranchersaveraging USD 3 to 4 million per microbasinand wa s ineffective in creating processes and procedures through which lan dholders who wished to comply with the law could do so. The validity of the 80% l egal reserve requirement was undermined by the state of Mato Grossos decision that the transition forests of the region had a 50% legal reserve requirement, by the frequent attacks on the Fore st Code within Brazilian legislature by the agricultural lobby, and by the low levels of enf orcement of the 80% rule. Non-compliance was reinforced by t he perceived risks associ ated with compliance, in which law enforcement officers punish t hose who are trying to comply with the law because of the threat that compliance poses to the culture of graft and corruption (Rosenthal, 2009). The origins of these defici encies in the Forest Code ma y be traced to the very system by which laws are created in Brazil and the extent to which this approach is 65

PAGE 66

suited for handling complex natural resource governance issues. If the governance of common pool resources depends on designing a system of clear rules and incentives, the Forest Code fails on both accounts. Whereas the Forest Code certainly lays out a body of rules, the modifications made a fter 1996 combined with the process of decentralization to lead to a set of rules that are far from cl ear and that inspire uncertainty. Concurrently, powerful economic drivers created an even greater need for incentives to comply with the regulations. However, the moral i dealism of the legal tradition in Brazil may have made the notion of incentives difficult to consider. With a focus on the ideal versus the practical solution to iss ues of the public good, and its defense, Brazils civil law system typically produces new regulations and laws without the dialogue and debate among interested stak eholders that could build into the design of the new law mechanisms for increasing its practicality and chances of successful implementation. However, multi-stakeholde r, participatory planning processes are already part of recent watershed management legislation (cite). More and more, civil society is initiating such processes for a variety of environmental governance issues, including infrastructure dev elopment (Campos and Nepstad, 2006; Soares-Filho et al., 2004), watershed management (ISA, 2005), and ot her regional planning processes (Perz et al., 2008), signaling an important trend in Brazilian environmental rule-making in the future. Conclusion This case study identifies a crucial c hallenge for lawmaking designed to defend public interests in privately controlled natural resources. At its core, it illustrates one legislative attempt to reconcile a trade-o ff that has repeated itself throughout human history. Are higher evapotrans piration, carbon stocks, greater rainfall security, reduced 66

PAGE 67

67 soils erosion, and the maintenance of native habitats over a 178,000 km2 watershed worth nine billion dollars? If the answer is affirmative, then creating incentives that facilitate private landholders compliance with the law would seem to be strategic. The Brazilian Forest Code is an innovative legisl ation, and one of the first to recognize and attempt to protect the broader public interests in private land forests in the tropics. Its potential for fostering the reconciliation of conservation with agricultural development is high, but in its current stat e is not being realized. The Br azilian government might have achieved the objectives of defending public intere sts in private forests if the shift to 80% had been implemented in a different way. First, the change should have been accompanied by an effective set of options by which landholders could bring their properties into compliance with the new law. Second, the government should have developed a system of positiv e incentives for complyi ng with the new regulation, potentially including compensat ion of at least part of t he opportunity cost associated with forgone profits from so y cultivation or cattle ranching. The carbon market represents an important oppor tunity to achieve these economic incentives.

PAGE 68

Table 2-1. The area and percent coverage of native forest and cerrado in the Xingu River headwaters region as required by old and new Brazilian Forest Code regulations, and as observed for 1996, 1999, and 2005. Data are presented both for the entire headwaters region and for the m ean values of the regions microbasins. Results are for lands outside of protected areas. Required Observed Old Regulations New Regulations 1996 1999 2005 Area (km2) % Area (km2) % Area (km2) % Area (km2) % Area (km2) % Native Vegetation Whole watershed (Total) Legal Reserve Forest 48,255 50 77,208 80 79,910 83 75,139 78 61,934 64 Cerrado 10,245 35 10,245 35 18,617 64 15,967 54 13,065 45 Riparian Zone Forest 8487 10 0 8487 100 7584 89 7354 87 6814 80 Cerrado 2571 10 0 2571 100 2117 82 1964 76 1733 67 Total 11,058 10 0 11,058 100 9700 88 9318 84 8547 77 Microbasins (Mean (s.e.)) Legal Reserve Forest 28 (0.6) 50 44 (0.9) 80 46 (1.0) 85 (0.5) 43 (1.0) 80 (0.6) 35 (0.8) 68 (0.6) Cerrado 22 (1.1) 35 22 (1.1) 35 41 (2.1) 66 (1.1) 35 (1.8) 58 (1.2) 28 (1.5) 49 (1.2) Riparian Zone Forest 4.8 (0.1) 10 0 4.8 (0.1) 100 4.3 (0.1) 90 (0.3) 4.0 (0.1) 88 (0.4) 3.9 (0.1) 80 (0.5) Cerrado 5.7 (0.3) 10 0 5.7 (0.3) 100 4.7 (0.2) 84 (0.7) 4.3 (0.2) 79 (0.8) 3.8 (0.2) 70 (0.9) Total 5.0 (0.1) 10 0 5.0 (0.1) 100 4.4 (0.1) 89 (0.3) 4.0 (0.1) 86 (0.4) 3.9 (0.1) 78 (0.5) 68

PAGE 69

Table 2-2. The number and percentage of microbasins in the Xingu River headwaters region that were in compliance with Brazilian Forest Code forest cover requirements in 1996, 1999, and 2005. Data are fo r microbasins outside of protected areas. Required Observed 1996 1999 2005 Number in Compliance % Number in Compliance % Number in Compliance % Number in Compliance % Microbasins Legal Reserve Forest 50% 1756 100 1608 92 1543 88 1383 79 80% 1756 100 1259 72 1123 64 810 46 Cerrado 35% 460 100 423 92 377 82 359 78 Riparian Zone Forest 100% 1749 100 748 43 640 37 344 20 Cerrado 100% 454 100 65 14 59 13 36 8 Total 100% 2203 100 813 37 699 32 380 17 69

PAGE 70

Table 2-3. Evolution of compliance with old and new Brazilian Forest Code in the microbasins of the Xingu River headwaters region, by category of change for 2 time periods. Illegal and l egal deforestation refers to those microbasins that were above or below, re spectively, the maximum percent of legal reserve clearing allowed by the Forest Code. Continued non-comp liance refers to mi crobasins that began the time period out of compliance. New non-compliance re fers to microbasins that moved from compliance to non-compliance during the time period. Period 1 (1996-1999) Period 2 (1999-2005) 50% Legal Reserve 80% Legal Reserve 50% Legal Reserve 80% Legal Reserve Area (km2) No. microbasins Area (km2) No. microbasins Area (km2) No. microbasins Area (km2) No. microbasins Deforestation Illegal Continued noncompliance 444 (9%) 125 (10%) 2129 (44%) 430 (36%) 854 (6%) 179 (12%) 5817 (44%) 570 (38%) New noncompliance 756 (16%) 68 (6%) 1474 (30%) 145 (12%) 3953 (27%) 171 (11%) 5361 (41%) 334 (22%) Legal In compliance 3667 (75%) 1008 (84%) 1263 (26%) 624 (52%) 8386 (58%) 1162 (77%) 2006 (15%) 603 (40%) 70

PAGE 71

71 Table 2-4. Total area and mean area per microbasin of forest and cerrado within the Xingu River headwaters region that is available for conversion and/or needing to be rest ored under the old and new Brazilian Forest Code, measured at three dates. Data are for microbasins outside of protected areas. Available for conversion refers to forest and cerrado that could be cleared without exceeding the maximum percent clearing allowable under the Forest Code. 1996 1999 2005 Total (km2) Mean ( s.e. ) per microbasin (km2) Total (km2) Mean ( s.e. ) per microbasin (km2) Total (km2) Mean ( s.e. ) per microbasin (km2) Area available for conversion Forest 50% 32,879 20.4 (0.5) 28,920 18.7 (0.5) 20,284 14.7 (0.4) 80% 9221 7.3 (0.2) 7403 6.6 (0.2) 3992 4.9 (0.2) Cerrado 35% 8499 20.1 (1.1) 6662 17.7 (1.0) 5362 14.9 (0.9) TOTAL 50% 41,378 20.4 (0.4) 35,582 18.5 (0.4) 25,646 14.7 (0.4) 80% 17,720 10.6 (0.3) 14,065 9.4 (0.3) 9354 8.0 (0.3) Area to be restored Legal Reserve Forest 50% 1224 8.3 (0.7) 1929 9.1 (0.7) 3490 9.4 (0.5) 80% 6520 13.1 (0.7) 9366 14.8 (0.7) 16,125 17.1 (0.6) Cerrado 35% 104 3.0 (0.5) 318 3.9 (0.5) 487 4.9 (0.8) TOTAL 50% 1328 7.3 (0.6) 2247 7.6 (0.5) 3977 8.4 (0.5) 80% 7744 12.5 (0.7) 3684 13.6 (0.6) 16,612 16.0 (0.6) Riparian Zone Forest 100% 894 0.5 (0.03) 1126 0.6 (0.03) 1673 1.0 (0.04) Cerrado 100% 449 1.0 (0.1) 603 1.3 (0.1) 839 1.9 (0.1) TOTAL 1344 0.6 (0.02) 1728 0.8 (0.03) 2512 1.1 (0.04)

PAGE 72

Table 2-5. Potential net present value (N PV) associated with soybean cultivation or cattle production on forested and deforested land for the Xingu River headwaters region under alte rnative Brazilian Forest Code requirements at three dates. Data are for areas outside of protected areas in the forest biome region, for microbasins in compli ance with 50 and 80% legal reserve requirements, and for microbasins out of compliance with 50 and 80% legal reserve requirements. Also, potential NPV that could be realized through additional legal clearing, summed ac ross microbasins, for 50 and 80% legal reserve requirements. 1996 1999 2005 NPV (million USD) NPV (million USD) NPV (million USD) Potential NPV All remaining forest (potential NPV) 26,215 24,266 19,138 All cleared land 6668 8617 13,746 Total 32,883 32,883 32,883 Compliant microbasins 50% Cleared 5070 5799 6905 Forest 25,388 22,963 16,330 80% Cleared 2673 2581 2541 Forest 20,536 16,793 9388 Non-compliant microbasins 50% Cleared 1598 2818 6841 Forest 827 1303 2809 80% Cleared 3996 6036 11,204 Forest 5680 7473 9750 Potential additional NPV through legal forest clearing Compliant microbasins 50% 11,595 9646 4518 80% 2853 904 -4224 72

PAGE 73

Table 2-6. Mean potential net present value of soy cultivation or cattle ranching for cleared and forested land of compliant and non-compliant microbasins of the Xingu River headwaters region at thr ee dates. Results are for microbasins outside of protected areas. 1996 1999 2005 Mean potential NPV per hectare (USD ha-1) Compliant microbasins 50% Cleared 2444 2467 2416 Forest 2668 2673 2478 80% Cleared 2009 1892 1772 Forest 2588 2552 2211 Non-compliant microbasins 50% Cleared 2480 2513 3102 Forest 2465 2510 3087 80% Cleared 2877 2865 3087 Forest 2961 2958 2997 73

PAGE 74

Table 2-7. Mean potential net present value associated with soy cultivation or cattle ranching on cleared and forested lands in microbasins of the Xingu River headwaters region at three dates. Data are presented for all microbasins, for compliant and on-compliant microbasins, and for forest land that could be legally cleared. 1996 1999 2005 Mean potential NPV per microbasin (s.e.) (million USD) Mean potential NPV per microbasin (s.e.) (million USD) Mean potential NPV per microbasin (s.e.) (million USD) Potential NPV All microbasins Forest (potential) 11.8 (0.4) 10.9 (0.4) 8.6 (0.3) Cleared 3.0 (0.2) 3.9 (0.2) 6.2 (0.3) Total 14.8 (0.5) 14.8 (0.5) 14.8 (0.5) Compliant microbasins 50% Cleared 2.5 (0.2) 3.0 (0.2) 4.0 (0.2) Forest 12.5 (0.4) 12.0 (0.4) 9.4 (0.3) 80% Cleared 1.6 (0.1) 1.7 (0.2) 2.2 (0.2) Forest 12.2 (0.4) 11.2 (0.4) 8.0 (0.4) Non-compliant microbasins 50% Cleared 8.8 (0.8) 9.6 (0.7) 14.5 (1.0) Forest 4.5 (0.5) 4.4 (0.4) 6.0 (0.4) 80% Cleared 7.5 (0.4) 8.4 (0.4) 10.7 (0.5) Forest 10.7 (0.7) 10.5 (0.6) 9.3 (0.4) Potential additional NPV on legally cleared lands 50% All 5.2 (0.2) 4.4 (0.2) 2.0 (0.1) 80% All 1.3 (0.1) 0.4 (0.1) -1.9 (0.2) 74

PAGE 75

Table 2-8. The average and range of estimated costs of restoration (million USD) of the legal reserve and riparian zone under 2 iterations of the Forest Code in relation to three points in time, for t he entire basin in the aggregate and for the average microbasin out of complianc e. Ranges refer to upper and lower estimates of restoration costs per hectare. 1996 1999 2005 Whole Basin (million USD) Whole Basin (million USD) Whole Basin (million USD) Legal Reserve Forest 50% 114 () 180 () 325 () 80% 607 () 872 (+/371) 1502 () Cerrado 35% 9 () 30 () 45 () Riparian Zone 100% 252 () 324 (+/232) 471 () Total 50% 375 () 534 () 841 () 80% 868 () 1226 () 2018 () Mean per microbasin (million USD) Mean per microbasin (million USD) Mean per microbasin (million USD) Legal Reserve Forest 50% 0.8 (.3) 0.8 (.3) 0.9 (.3) 80% 1.2 (+/0.5) 1.4 (.6) 1.6 (.7) Cerrado 35% 0.3 (.1) 0.4 (.2) 0.5 (.2) Riparian Zone 100% 0.1 (.1) 0.2 (.1) 0.2 (.2) Total 50% 1.2 (.6) 1.4 (.5) 1.6 (.8) 80% 1.6 (.7) 2.0 (.8) 2.3 (.1) 75

PAGE 76

Table 2-9. Ecological attributes of the Xingu River headwaters region (including protected areas) at three 3 dates (1996, 1999, 2005) and of modeled landscapes representing t he region with application of 80% legal reserve or 50% legal reserve on private lands. These attributes include: carbon stocks, surface hydrology and regional climate, indicators related to water quality, and terrestrial habitat quantity and quality. Scenarios Indicator 1996 1999 2005 50% 80% Carbon Stocks (MtC) 623 606 546 448 623 (MtCO2e) 2280 2217 1998 1638 2082 Surface Hydrology and Regional Climate Mean Annual Discharge (m3 s-1) (% change from potential) 2958 (7%) 3007 (9%) 3128 (13%) 3235 (17%) 3032 (10%) Mean Annual Evapotranspiration (m3 s-1) (% change from potential) 6754 (-3%) 6705 (-3%) 6583 (-5%) 6477 (-7%) 6679 (-4%) Water Quality Riparian forest cover, 50-m buffer (km2) 14,862 14,369 12,884 15,500 15,500 Mean % vegetation cover per microbasin (s.e.) 86 (0.4) 80 (0.5) 73 (0.5) 61 (0.5) 78 (0.4) % of microbasins with greater than 60% vegetation cover 88 79 69 31 84 % of microbasins with less than 40% vegetation cover 4 9 14 15 14 Terrestrial Habitat Vegetation cover (km2) Forest 86,944 82,130 71,336 51,561 80,005 Cerrado 16,496 14,326 13,238 9083 9083 Number of fragments Forest 9810 9609 12,295 23,673 11,439 Cerrado 7223 10,291 12,632 18,422 18,422 Mean distance to nearest neighbor fragment (m) Forest 383 372 367 376 390 Cerrado 404 368 385 354 354 Mean fragment size (ha) Forest 886 855 580 218 699 Cerrado 228 139 105 50 50 Total interior habitat area (km2) Forest 82,248 76,752 63,847 36,975 68,598 Cerrado 13,373 11,082 9338 4001 4001 Total edge habitat area (km2) Forest 4700 5378 7489 14,586 11,407 Cerrado 3123 3245 3900 5082 5082 76

PAGE 77

77 Figure 2-1. Maps showing the Xingu River headwaters region under different scenarios. (a) The Xingu River headwaters region ( outlined in blue), showing federal and state protected areas (ye llow), indigenous territories (white), paved roads (red), and other major unpaved roads (black). Land cover is shown for a Landsat 5 TM mosaic from 2005; greener areas indicate presence of more native vegetation or higher biomass r egeneration, pinker areas indicate cleared areas or areas of low native biom ass. (b-f) Comparison of land-cover in the basin representing 3 time points relevant to changes in the Brazilian Forest Code(b) 1996, (c) 1999, and (d) 2005and 2 alternative versions of the Forest Code, with a r equirement of (e) 50% legal reserve, and (f) 80% legal reserve on private properties in the Amazon forest biome.

PAGE 78

Figure 2-2. Box diagram showing the fraction of microbasi ns (which serve as proxies for private properties) in the forest biome in each of three categories of legality and the change in that fraction fr om the previous dat e. Categories are: (a) below 50% legal reserve, (b) between 50 and 80% native vegetation, and (c) greater than or equal to 80% native v egetation cover. The size of each box is proportional to the fraction of total forest microbasins in that category. 78

PAGE 79

Figure 2-3. Map showing microbasins (which serve as proxies for private properties) out of compliance with two iterat ions (50% and 80% legal reserve requirement) of the Forest Code at thr ee time points: (a) 1996, (b) 1999, and (c) 2005. Microbasins with <50% forest co ver (forest biome), 50 to 80% forest cover (forest biome), and <35% native vegetation ( cerrado biome) are highlighted. Map of net present value (N PV) (d) is provided for interpretation. 79

PAGE 80

Figure 2-4. Microbasins withi n two extreme categories of legality in relation to the modified Forest Code: (a) microbas ins with less than 50% forest cover remaining 1996 through 2005; (b) microbasins with more than 80% forest cover remaining since 1996. 80

PAGE 81

CHAPTER 3 HYBRID REGULATORY-ECONOMIC POLICY INSTRUMENTS IN PRIVATE FOREST GOVERNANCE Introduction The main instrument for protecting forests on private lands in Brazil is the federal Forest Code, which mandates that landholders maintain a certain percentage of native vegetation in a private reserve. In 1996, the Brazilian f ederal government responded to record high deforestation in the previous year by raising the proportion of private properties in the Amazon to be maintained in a forest reserve from 50% to 80%. This measure provoked an intense reac tion from the agro-industrial lobby, particularly in the state of Mato Grosso, Brazils only majo r agricultural producing and exporting state located in the Legal Amazon. Although it is unclear what the extent of non-compliance with the new Forest Code regulations is for the whole state of Mato Grosso, in the Xingu headwaters (a region representing 20% of the states area and more than one quarter of the states forest area), an estimated 28% of landholdings were non-compliant in 1996, increasing to 55% in 2005 (Chapter 2). A partial ex planation of this imperfect level of compliance is offered by the large potential earnings that can be made from soy cult ivation or cattle ranching in comparison with the earnings potential of selective loggi ng. In the Xingu headwaters, the change in the legal reserve requirement imposed USD 9 billion in potentially foregone profits on the regions farmers and r anchers (an average of USD 3 to 4 million per microbasin) (Chapter 2). Powerful economic drivers to cl ear forest were not met with any counter-incentive from the government to facilitate landholder compliance. Moreover, the change in the Forest Code wa s not accompanied by effective processes and procedures through which landholders w ho wished to comply with the law could do 81

PAGE 82

so. Finally, a number of other legal and adm inistrative factorsincluding lack of financial and human resources for monitori ng and conflicting messages from the state and federal governments about the exact regulationsconspired to leave landholders in doubt as to what rules they ought to adhere (Chapter 2). In recognition of the inefficiency and ineffe ctiveness of the Fore st Code, lobbyists and legislators have made several proposals for modifications or alternate policy instruments designed to reduce the cost of compliance both to individual landholders and to the States economy as a whole. Among these are two hybrid instruments that combine a regulatory framew ork with an economic inst rument with the goal of increasing the efficiency and effectiveness of t he Forest Code. This first is a provision (already incorporated into the current Fo rest Code) to permit landholders having less than the required area of forest reserve to bu y another landholders rights to clear forest on the second property (MP 2611). This mechanism is a type of economic policy instrument, akin to tradable permits that have been successfully used for reducing some pollutants and gas emissions. The second policy proposal focuses on designing a landuse zoning plan that incorporat es measures of agricultural suitability and ecological value and vulnerability to determine where and in what measure the requirements of the current Forest Code will be applied (SEPLAN-MT, 2008). This policy incorporates both economic and regulatory measures. In this chapter, I present the results of a study focusing on the provision to allow tradable development rights in the Forest C ode and the Mato Grosso state zoning plan these other existing legal inst ruments that could potentially lo wer the opportunity cost of complying with the new Forest Code regulati ons without forfeiting the environmental 82

PAGE 83

gains. First, I describe each of these inst ruments and compare th em with the Forest Code, analyzing the extent to which they repr esent economic instrum ents or exclusively regulatory instruments. Next, I carry out a quantitative analysis of the 3 instruments, focusing on the ability of each to balance agricultural production with ecological conservation. I compare modeled land scapes representing the land-cover consequences of implementing each of the th ree policy alternatives in a landscape on the southeastern Amazon frontier in terms of ar ea available for agricultural activities, the costs of compliance with each theoretical la ndscape, and the extent to which ecosystem services are maintained. Forest Governance on Private Lands Brazilian environmental legislation is considered to be among the most sophisticated in the world. The early part of the 20th century saw enactment of laws that recognized and protected forest resources as a common good (Chapter 2). Successive legislation, including th e 1988 Brazilian Constitution, created requirements for environmental impact assessments, environm ental quality standards, and licensing and monitoring of a range of activities with envir onmental impacts, as well as environmental crimes legislation. However, natural re source and environmental policies have been notoriously difficult to implement in practi ce. Primarily, this is due to poor coordination among responsible authorities, lack of human and financial resources, and a pervasive perception of environmental policy as generally obtrusive to economic development and, thus, of significantly lower priority than many other policies (Ascher, 1999). Many key policy decisions that greatly affect natural resources are still primarily the responsibility of ministries other than the Mini stry of Environment, such as the Ministry of Agriculture, the Ministry of Transportation, and the Ministry of National Integration 83

PAGE 84

and Planning, which often leads to direct in consistencies with envir onmental legislation. Moreover, the responsibility for developing and implementing specific environmental regulations is divided among federal, state, and municipal authorities, often without great clarity about which agency is responsib le for which issues and with inadequate communication and coordination between these authorities (Asc her, 1999; May, 1999). Even where responsibility is clearer, lack of personnel or funding for enforcement can hamper policy implementation. Ascher (1999) also observe s that government policies frequently are not designed to encourage landholders or landusers to invest in good land management and often do just the opposite, providing incentives to invest in inappropriate technologies or practices. Furt hermore, the judicial system is ill-equipped to help to enforce the laws, largely becaus e of its slowness and weakness in carrying out the rule of law (Ames and Keck, 1998; Brito and Barreto, 2006), and because of rampant corruption (Brito and Barreto, 2006). In this chapter, I examine examples of Brazilian policy instruments t hat attempt to overcome so me of these obstacles to protecting native ecosystems, and the services that they provide, on privately owned properties. Policy Instruments for Natural Resource Management Environmental policy instruments are typica lly divided into three broad categories: regulation, economic instru ments, and voluntary action. Whereas regulatory and occasionally economic instruments are c onsidered to be standard tools for government policy-making world-wide, market-based mechanisms and voluntary approaches are increasingly employed to complement or su bstitute traditional regulatory environmental policy approaches. Regulatory (often referred to as command-and-control) instruments typically define performance-based or technology-based standards with which 84

PAGE 85

producers or landowners are required to comply by law. Regulation can take a variety of forms, including the outright prohibition of an ac tivity or substance, setting limits for the amount of a pollutant that may be produced, or defining the terms by which natural resources can be extracted (Sterner, 2003). Fa ilure to comply with regulations generally involves fines or other penalties. Thus, monitoring and enforcement is a critical component of direct regulation. In contrast with regulatory instruments, economic instruments are generally cons idered to be more economically efficient (Carter, 2001). This is true because these instruments infl uence behavior, in effect, by increasing the profitability of a more desir able form of resource user behavior, hence harnessing the users rent-seeking behavior. They include ecological taxes, user charges, deposit refunds, tradable permits, and subsidies. Finally, voluntary approaches involve substitution of state-defined regulations with self-defined implementation standards for regulations, self-financed certification syst ems for enforcement, and elective public reporting for public disclosure requirement s (Ten Brink, 2001; Ca rter, 2001). In the private sector, voluntary approaches typically involve adoption of social responsibility standards and environmental management syst ems (EMS), which include standards such as ISO 14000 5 eco-labeling, and certificat ion. Recently, payment-forenvironmental-services (PES) schemes hav e become a prevalent type of voluntary approach based on the logic of an economic instrument; prominent among these are schemes focusing on carbon sequestration and avoided carbon emissions. In this chapter, I focus primarily on extant regulatory and economic instruments, returning to 5 The ISO (International Standards Organization) 14000 environmental managem ent standards provide a set of guidelines to help organizations develop envir onmental management system whose primary focus is on increasing management efficiency and reduci ng negative environmental impacts as a by-product, and on compliance with applicable environmental laws and regulations. 85

PAGE 86

voluntary approaches in the final discussion of the results and their implications. In the following sections, I discuss 3 policy instruments to regulate land-use on private lands. The three instruments include a standard regulat ory instrument and 2 hybrid economicregulatory instruments. The Legal Reserve of the Federal Forest Code In Brazil, use of regulatory instrument s has been the dominant policy approach to environmental problems. The Forest Code is an ex ample of such an instrument in that it requires private landholders to maintain forest reserves on their lands under penalty of fines. Since 1965, the Brazilian Forest Code requires landowners to maintain 2 types of native vegetation reserves on their properties : (1) a legal reserve, and (2) a riparian buffer zone. The legal reserve requires landholders to maintain a minimum percentagedetermined by the phytogeographic location of the propertyof native vegetation on their properties, which may be managed for sustainable timber harvests. In 1996, the legal reserve in the Amazon fore st biome was increased from 50% to 80%; the legal reserve in the cerrado continued to be set at 35% (Chapter 2). In practice, this requirement has been difficult to enforce, fo r a variety of reasons. On the one hand, monitoring and enforcement of the policy is human and financ ial resource intensive. Several studies carried out in Mato Gro ssothe state with the hi ghest deforestation rates for much of the past decadedemonstr ate that staff to conduct monitoring expeditions is in short suppl y, corruption in the legal an d regulatory system is rampant (Lima et al., 2005), and the costs of registering properti es and maintaining the legal reserve are prohibitive (Guimaraes and Alme ida, 2007). The effectiveness of the Forest Code in the Brazilian Amazon has been restricted by frequent changes in regulations, the lack of an effective mechanism for enabling landholders to comply with these 86

PAGE 87

changing regulations, and the absence of econom ic incentives for compliance, as described in Chapter 2. On the other hand, the potential income from lands cleared for ranching or soy farming (the principal agricultural activities in the region) is 2 to 20 times greater than the potential income from reduced-impact selective timber harvest (Nepstad et al. 2009). Thus, many landowners cons ider the risk of fines merely a cost of doing business. Both the absence of enforce ment and the profitability of agricultural production in the region (and thus, willingness to face sizeable fines when they are applied) help to explain this result (Chapter 2). In general, the legal reserve requirement has been found to be both an economically and ecologically inefficient way to meet a quantitative tar get for forest area to be preserved (Souli de Amaral, 1997; N ogueiro-Neto, 1998). According to Chomitz (2004), the reserve requirement is economically inefficient because it requires the same amount of preservation (or restoration) of forest cove r regardless of the propertys agricultural potential or market access. Mo reover, it is environmentally inefficient because (1) it does not prioritize forest co ver according to ecological or biodiversity value, and (2) it tends to result in fragmented forests (Chomitz, 2004). In addition, compliance with the Forest Code does not reach 100%. In the Xingu River headwaters region (in northeastern Mato Grosso), compliance with the increased legal reserve requirement fell from an estimated 72% of properties in 1996 to 46% in 2005 (Chapter 2). Simultaneously, the area requiring restorat ion of native vegetation under the law increased more than 2-fold and the opportunity cost (resulting from lost potential net present value) increased by nine billion dollar s. In an effort to increase compliance with the regulations by reducing the cost to pr oducers, lawmakers included 2 provisions in 87

PAGE 88

the modified version of the Forest Code that was finally adopted into law in 2001: (1) the tradable deforestation rights scheme required that an RL be compensated in another area of equal extent and ecol ogical function and character within the same micro-basin (Chomitz, 2004); (2) the legal reserve c ould be reduced to 50% from 80% in the Amazon forest biome within the context of state zoning plans. I consider both of these mechanisms to be a hybrid regulatory-economic instrument in that each essentially maintains the requirements of the Forest Code but permits individual landholders to determine the way in which he/she will m eet the requirements. I describe these instruments in furt her detail below. Tradable Deforestation Rights Tradable permits represent a rights-based mechanism for internalizing environmental externalities. The tradable permits (or capand-trade) system represents a relatively new institutional approach to manage the problem of rationing access to the commons (Tietenberg, 2002). It has been applied to a number of different resources or resource systems, including air pollution cont rol, fisheries, water resource management, water pollution control, and land-use c ontrol (Tietenberg, 2002; Chomitz 2004). Tradable permits address the commons problem by rationing access to the resource and privatizing the resulting access rights. The first step involves setting a limit (or cap) on user access to the re source. In the case of fis heries this would involve the total allowable catch, for example. For wate r supply, it would in volve the amount of water that could be extracted. For pollution control it typically specifies the aggregate amount of access to the resource that is au thorized or the total am ount of pollution that is permitted. These access rights are then distri buted (typically by direct allocation or by a public auction) to potential individual us ers. Depending on the s pecific system, these 88

PAGE 89

rights may be transferable to other users (o r tradable) and/or bankable for future use. Users who exceed limits imposed by the ri ghts they hold face penalties up to and including the loss of the right to participate (Tietenberg, 2002). The tradable permits system essentially creates a market for a common pool resource, if there is force of law accompanying it. Gener ally it will not work on a voluntary basis or under conditions of low enforcement, since those individuals or firms that comply accrue higher costs of production than those who do not (Tietenberg, 2002). The key difference between cap-and-trade syst ems and command-and-control is that the latter seeks to control emissions at the individual level, whereas cap-and-trade seeks to control emissions at the aggregate level. Su ch a system encourages innovation and incentivizes the least-costly means to meet the targets. This flexibility is likely to produce a higher level of compliance and thereby lead to m eeting the overall environmental target more readily. Furthermo re, as Carter (2001) explains, In short, regulations provide no incentive for polluters to reduce their pollution any further than that required by law. [Marke t-based instruments] are intended to provide that incentive. Like command-and-control syst ems, tradable permits systems require adequate monitoring and enforcement. When the legal reserve requirement in t he Amazon forest biome was increased to 80% in 1996, approximately one-fifth of l andholders who had been in compliance with the law up to that date found t hemselves suddenly in violation of the new Forest Code regulations (Chapter 2). Under the new law, they were required to reforest their land to meet the new legal reserve requirement. Ho wever, this represent ed both the collective loss of nine billion USD in potential net present value and a collective cost of 1 to 3 89

PAGE 90

billion USD in restoration costs (Chapter 2) As a means of bringing these landholders into compliance, a 1998 re-edition ( MP 1736 / 98) of the provisionary measure MP 1.511 modifying the legal reserve included a provision to allow landown ers whose property does not meet the legal reserve requirements to compensate the missing reserve area elsewhere within the same watershed, provided it has equal or greater ecological value. More specifically, landowners who had defores ted more than 20% of their land, but not more than 50% before December 31, 1999, were eligible to participate in the scheme. Instead of handing over title to the land, the owner of land having an amount of forest over and above the minimum requirement for t he legal reserve would sell the rights to develop that land to another owner with inadequate forest to meet the requirements. To date, however, the option to trade dev elopment rights within the Forest Code has been little exercised within the Legal Amaz on. The major barrier is the lack of sufficient current information about indivi dual properties forest reserves (due to inadequate monitoring and enforcem ent). Even in the state of Mato Grosso, where the innovative digital licensing system should fa cilitate the legal reserve compensation system (but see Azevedo, 2009), only 5 transfers have been documented as having taken place since 2000. This is due in la rge part to disorganization in data base management within the licensing system (Lima et al., 2005; Azevedo, 2009). In addition, however, the legislation does not make clear at what scale development rights trading may be carried out. Although it specifies t hat it may be carried out within the same watershed, it is unclear about what size of watershed. As Chomitz (2004) notes, if trading is carried out on a smaller scale (at t he level of lower-order watersheds), the feasibility of monitoring and enforcement is likely to in crease, thus decreasing the 90

PAGE 91

transaction costs (from societys point of view, if not the landow ners). Furthermore, trading on a smaller scale increases the homogenei ty or substitutabili ty of the forest areas involved. However, reducing the scale ma y also severely restrict the potential of the system to lower the economic costs of environmental management by reducing available options (Chomitz, 2004). State Socioeconomic-E cological Zoning Plans Since the enactment of the 1988 Constitution, the primar y instrument for regional land-use planning in Brazil within the National Environment Policy ( Politica Nacional de Meio-Ambiente, PNMA ) is intended to be the Zoneamento Socio-Economico Ecologico (ZSEE; Socioeconomic Ecological Zoning Pl an). The state zoning plans specify the types or intensities of land-uses which may occur throughout a given state, based on analyses incorporating biophysical and economic factors. It was not until 2002 that a federal decree ( Decreto Federal 4.297/2002 ) established criteria that all state zoning plans would be required to meet to be recognized by the national government, including being approved by the respective states legislative assembly and ZSEE Commission. Several federal technical agenciesincluding the National Institute for Space Research (INPE), the Brazilian Enterprise for Agricu ltural Development (EMBRAPA), and the Brazilian Institute of Geography and Statistics (IBGE) are charged with assisting states in developing their zoni ng plans. One of the most impor tant features of the state zoning plans is that they may supersede the Forest Code, relaxing or raising the size of the legal reserve in some zones, for example. Although the federal Ministry of Environment set a deadline of July 2009 for states to approve their ZSEE plans, only 2 of Br azils 9 Legal Amazon statesRondnia and Acremet this deadline. In Mato Gro sso, the ZSEE has been under development and 91

PAGE 92

discussion for 18 years (Folha de So Paulo, 2009). The state Planning Secretariat (SEPLAN) presented what it considered to be a final plan together with the state environmental agency (FEMA, at the time) to Governor Blai ro Maggi in 2004. However, the plan was heavily criticized by environmentalists and fo llowing reorganization of the states environmental regulator y and management structure by the federal government in 2005, a new proposal began to be developed. A new draft of the plan was presented to the governor in April 2008, but was strongly criticized by the agro-industrial sector as it increased the land area in protected areas to 27%, up from 20%. In part, the state Planning Secretariat increased the pr oportion of protec ted areas based on recommendations from INPE, which had predi cted increased deforestation rates in the second half of 2007, while the ZSEE proposal was still being modified. In contrast, the proportion of the state dedicated to agr icultural production (with more limited environmental requirements) amounted to only 11% of the total area. Since the proposal was unveiled, the ZSEE has undergone a lengthy consultation process (originally slated to be completed before the end of 2008, it continues into the second half of 2009) throughout the state with loca l stakeholders and the involvement of national civil society and government agencies. There is considerable concern about the validity of the plan, since the majority of the data on which the plan is based were collected between 1995 and 1997, before the ma jor expansion of agro-industry into the forest zones of the state had begun. The Mato Grosso ZSEE plan is a hybrid r egulatory-economic instrument in that it requires landholders to comply with the Forest Code, but redefines the proportion of a property that must be protect ed using a combination of econom ic and ecological criteria 92

PAGE 93

to reduce some of the inefficiencies prev iously identified in the Forest Code. Furthermore, in the version of the zoning plan currently being promoted by environmentalists in Mato Gro sso (A. Lima, pers. comm., Ap ril 15, 2009), application of the tradable deforestation rights scheme descr ibed above will be permitted. The plan identifies 4 major zones (Figure 3-1), each having specific rules and restrictions. The zone classification determines the fiscal incentives and public expenditures that will be made available, as well as the environm ental licensing requirements for an area. Among the major issues under debateaside from the proportion of land devoted to protected areasis whether the legal reserve in Zone 3 (see below) should remain at 80% and whether landholders ha ving less than 80% forest on their land must restore the full 80%, as currently stipulated under t he ZSEE. The zones are defined as follows: Zone 1 Areas of historical long-term dedica ted agricultural use or areas to be dedicated to agricultural production in the future In this zone, the Forest Code will be alte red, if approved. Private properties may be deforested up to 50%; if more t han 50% of the area has already been deforested, this excess must only be re stored to 50% of the property area. Properties in the cerrado biome may not clear more t han 65% of the total area and must restore up to 35% of native cerrado vegetation, if more has been cleared. Zone 2 Areas requiring correction of managem ent systems: In this zone, the Forest Code may be altered Private properties may not deforest more th an 20% of the total area; if more than 20% of the area has already been deforested, this excess must only be restored to 50% of the property ar ea. Properties in the cerrado biome may not clear more than 65% of the total area and must restore up to 35% of native ce rrado vegetation, if more has been cleared. Zone 3 Areas requiring special management: In this zone, the Forest Code may not be altered Thus, in the forest biome, private proper ties may not deforest more than 20% of the total area; if more t han 20% of the area has been deforested previously, this excess must be restored. However, eligib le properties may participate in trading their deforestation rights, as ex plained above. Properties in the cerrado biome may 93

PAGE 94

not clear more than 65% of the total ar ea and must restore up to 35% of native cerrado vegetation, if more has been cleared. Zone 4 Protected areas: this category incl udes all current and proposed protected areas No deforestation is allowed in this zone. Theoretically, reforestation will take place in proposed protected areas, as they cu rrently encompass privately-owned areas. Materials and Methods Study Area The 177,780 km2 Xingu River headwaters region is located in the northeastern corner of Mato Grosso state, in cent ral Brazil (Figure 3-2). The regions soils, topography (100-300 m altitude, with flat interfluvial expanses) and climate are wellsuited for soybean production and cattle ranching. Native vegetation types in the region are comprised of forests (tall evergreen, transitional semi-deciduo us, and riparian) and savannas (cerrado woodland, mosaics of grassland, th ickets, gallery forests) which comprise approximately two-thirds and one-third of region, respectively. Ten indigenous territories are completely contained wit hin the boundaries of the Xingu watershed in Mato Grosso (Figure 3-2). Indigenous territories cover approximately 42,200 km2 within the basin, representing 24% of the total area of the headwaters region. Private landholdings comprise nearly 70% of t he total area, and smallholder settlements comprise less than 5% of t he region. The streams and rive rs of the major protected forest area that lies at t he center of the regionthe PIX complex, which alone comprises nearly 20% of the headwaters areaare under gro wing threats from sedimentation, agrochemical run-off, and associated fish die-off from the unprotected headwaters regions outside of the park boundaries (Sanches, 2002). The Xingu region is representative of many areas along the Am azons agricultural frontier, but faces a 94

PAGE 95

more acute and immediate threat because it lies between two major federal highways (BR-158, BR-163) that are partially paved, and lies in the pathway of the northward expansion of Brazils grain belt. Land Cover Maps I developed maps of the obs erved landscape in 2005 and 3 theoretical landscapes reflecting land-cover corresponding to alternatives for regulati ng land-use on private lands in Mato Grosso. I used these maps to estimate and compare the amount of vegetationremnant native and restored or regeneratingt hat would be maintained under each scenario. Furthermore, I used these m aps to calculate the area available for agricultural production under each policy alternat ive. I also combined them with maps of net present value for the region to estimate and compare the potential net present value of cleared lands in the region under each scenario in the aggregate and on average per microbasin. Finally, the maps were also used as inputs for calculating several ecological indicators. Observed landscapes I developed maps of land-cover for 2 dates of significance r egarding private landuse legislation in Mato Grosso: 1999 and 2005. 2005 is the year in which the federal government dissolved the Mato Grosso st ate environmental f oundation (subsequently reorganized as the state environ mental secretariat) and ma ndated that landholders in Mato Grosso follow the require ments of the Forest Code as modified at the federal level in 2001, rather than as interpreted by the state government since 2000. Here, 2005 also represents the current landsca pe in the Xingu River headwaters, as very little new deforestation took place since that year. The map for 1999 was used to identify those landholders eligible to participate in the tr adable deforestation rights (or compensation) 95

PAGE 96

scheme. Maps with 4 classes (forest, cerrado, agricultural lands, other) were classified as described in Appendix A. Modeled landscapes Three theoretical landscapes corresponding to the requirements set forth by (a) the current Forest Code (after 1996) without the compensati on option (that I refer to here as CFC-NC; Chapter 2), (b) the Forest Code with the compensation option (CFC-C), and (c) the Mato Grosso Stat e socio-economic, ecological zoning plan (referred to here as ZSEE, SEPLAN-MT, 2009) were developed using a spatially explicit dynamic landscape simulation model. The basic architecture and function of the simulation model is described in Appendix A; details regarding the modeled landscapes follow here. The assumptions underlying each of the three scenarios differed only in the percent and location of native vegetation that was to be maintained or restored on each private land-holding in the headwaters region in accordanc e with the rules of each alternative mechanism, as described below Since a complete map of property boundaries for the region is not available, I used micro-basins representing individual stream reaches (1:1,000,000 scale) as prox ies for individual properties. The mean, range, and distribution of sizes of the 2881 microbasins microbasin ( x = 5981 ha, 470,766 ha) are comparable to that of privat e properties in the region (Jepson, 2006; Fearnside, 2005; Appendix B). Furthermore, the mean percent clearing in the current (2005) landscape is comparable among microbas ins and properties for which property limit data are available, indica ting that microbasins are suit able substitutes for individual properties in terms of sampling the populat ion of properties in the region. It was necessary to use a spatial unit with complete coverage of the study region to simulate 96

PAGE 97

the distribution of vegetation across the landscape corresponding with the requirements of the preand post-1996 versio ns of the Forest Code. The tradable deforestation rights provision in the Forest Code stipulates that property owners who have exceeded the legally permitted clearing of their land can com pensate this forest def icit within the same microbasin, as described below. For all three scenarios, full compliance wit h the law was assumed. All indigenous reserves and state and federal protected areas were strictly protected. Furthermore, a 50-m riparian buffer zone surrounding each str eam and river visible in a map derived from a thematic stream layer obtained from the Mato Grosso Stat e Regional Planning Secretariat (SEPLAN-MT) was strictly protec ted. The Forest Code stipulates that a riparian buffer zone of at least 50 m be prot ected around every natural body of water, and that the size of the bu ffer zone (up to 500 m) is de pendent on the width of the stream. However, as stream width is difficu lt to assess and no official map of riparian buffers exist, the estimation of variable-wid th riparian buffers was not possible and I assumed, conservatively, that all water bodies were surrounded by a 50-m wide buffer. Where necessary, vegetation was restored to this riparian buffer zone so that each theoretical landscape had 100% native vegetation cover within the boundaries of the riparian zone. To calculate the amount of deforestation or restoration that could or should take place, respectively, outside the riparian zone to meet the legal reserve requirements under each of the three scenariossubject to further restrictions or allowances, as described for each scenario belowI classifi ed each micro-basin according to biome (cerrado or forest). The cerrado-forest bi ome map was obtained by merging a map of 97

PAGE 98

forest/non-forest derived from INPE Prodes maps with a map of biomes from the IBGE RADAM vegetation thematic map. Of 2881 micro-watersheds in the Xingu River headwaters region, 34 straddl ed both biomes. To facilitate model design and processing, each of these microbasins was assigned to the biom e representing more than 50% of that micro-watersheds area. At time step 0, the m odel calculates how much of each watersheds area consists of cl eared area. If this ar ea is greater than the allowed amount (as determined by each iteration scenarios conditions) after the area of the riparian zone is subtracted from the to tal watershed area, the model reforests the area up to the allowable amount of cleared ar ea. If the cleared area is less than the allowable area, the model beg ins to deforest the watershed based on where the highest favorability for deforestation is indicated (Appendix A). In this study, I modeled 3 alternative scenarios over a 60-year time period (to ensure that the maximum allowable clear ing and regenerating under each scenarios assumptions would be achieved) in annual steps using 2005 as a starting point for: (1) Current Forest Code, No Compensation (CFC-NC), (2) Current Forest Code, Compensation (CFC-C); and (3) State Zoni ng Plan (ZSEE). The assumptions and conditions for each scenario are described below. Current Forest Code, no compensation The CFC-NC scenario assumes full compliance with the current Forest Code, such that properties (microbasins) located in the forest biome maintain a lega l reserve of 80% and those in the cerrado maintain a reserve of 35%. In addition, all properties mu st maintain 100% forest cover in riparian zones. 98

PAGE 99

To calculate the amount of deforestation or restoration that could or should take place, respectively, I classified each microbasin according to biome (cerrado or forest), as micro-basins serve as proxies for indi vidual properties in the model. The cerradoforest biome map was obtained by merging a map of forest-non -forest derived from INPE Prodes maps with a map of biomes fr om the IBGE RADAM vegetation thematic map. Of 2881 micro-watersheds, 34 straddled both biomes. To fac ilitate model design, each of these micro-watersheds was assign ed to the biome representing more than 50% of that micro-watersheds area. At time step 0, the model calc ulates how much of each watersheds area consists of cleared area. If this area is greater than the allowed amount (see amounts as defined by zone below) after the ar ea of the riparian zone is subtracted from the total wa tershed area, the model re forests the area up to the allowable amount of cleared area. If the clear ed area is less than the allowable area, the model begins to deforest the watershed bas ed on where the highest favorability for deforestation is indicated (see Simulation Model description). I generated a riparian buffer zone map by appl ying a 50-m buffer to either side of streams, as this width represents the minimum required under the Brazilian Forest Code. According to the legislation, the bu ffer width should vary with stream width. However, as I did not have access to data on stream width for the entire region, I assume that the minimum riparian zone width will be maintained, at a minimum. At time step 0 (the initial 2007 l andscape, in this case), the m odel determines how much, if any, of the area within the limits of the riparian buffer within eac h micro-watershed is cleared, then proceeds to reforest the cleared area su ch that 100% of the riparian zone is 99

PAGE 100

forested or in regeneration. Both curr ently forested and regenerating areas were prohibited from being cleared in the future. Current Forest Code, compensation Under the CFC-C scenario, landholders are required to comply with the requirement s of the Forest Code as described above. However, properties with less than the mini mum required forest cover and that meet certain other conditions, expl ained below, are permitted to bu y the rights to the forested area of another property to meet the legal reserve r equirement. The concept is essentially a version of tradable developmen t rights schemes, such as cap-and-trade programs, for deforestation ri ghts. Only properties which had already deforested more than 20% of the propertys area by January 1, 2000, are eligible to buy the deforestation rights of a property that curr ently still maintains more than 80% forest cover, up to the amount that must be restored on the first property. In this case, the second property gives up the right to deforest that portion for which the rights have been sold. To include the option of tradable defores tation rights, the model employs a mask that identifies properties eligib le for participation in the schem e to calculate, within each of the 6 sub-basins, the amount of remaini ng forest (from propert ies having more than 80% forest cover) that may not be deforested because it will be set aside in compensation for one of the microbasins that is eligible to reconstitute its own legally inadequate reserve by this means. Furthe rmore, the model determines how much reforestation is required using these paramet ers. If more than 20% of a micro-basins area was already cleared prior to January 1, 2000, then the m odel only requires it to reforest up to 50% of the tota l area of the microbasin outside the riparian zone, or in the case where a trade is made, to set aside t he equivalent of this area within the sub100

PAGE 101

basin. However, if less than 20% was clear ed by 1999, but more than 20% was cleared by 2008, then the model reforests (or sets as ide) up to 80% of the total area of the microbasin outside the riparian z one. These conditions conform to those set forth in the MP 2611 (the 2001 version of the Forest Code). State zoning plan The ZSEE scenario represents land-use/land-cover in the watershed if the Mato Grosso state zoni ng plan were to be appr oved and implemented (SEPLAN-MT, 2009). Although the plan requ ires that landowners observed the regulations of the Forest Code as described above, it allows restrictions imposed by the current Forest Code to be relaxed in specific zones, as dictated by the zoning plan. In addition, for eligible properties, as descri bed above, compensation is applied under the zoning plan. Finally, the state zoning plan hi ghlights a number of critical biophysical restrictions that limit agricultural deve lopment and encourage forest conservation or restoration. To develop the ZSEE scenario, I produced se veral maps delineating critical restrictions and zones in which the scenario conditions would be applied (Figure 3-1). The map of zones was derived from the offici al state zoning plan map, corresponding to the 4 major categories identified by the pl an, as described above (in State EcologicalEconomic Zoning Plans). In addition, I identified three critical restri ctions among the list of directives of the ZSEE that are explicitly spat ial in nature and that I determi ned to be limiting such that areas described by this set of restrictions would not be eligible for relaxation of the current Forest Code, regardless of whic h zone they fall in. Moreover, the areas described by this set of restri ctions would be subject to full protection of the native 101

PAGE 102

vegetation, and most likely, require restor ation of the native vegetation should the vegetation no longer be intact. The restri ctions (defined in the ZSEE; SEPLAN-MT, 2009) and methods I used for delineating spatial layers are as follows: Directive 7. Strictly protect floodplains, prohibiting any type of vegetation clearing, with the exception of clearing carried out fo r purposes of subsistence agriculture or for clearing of native pasture without the use of fire. I derived a map of floodplains in the basin from SRTM data (90-m resolution oversampled to 30-m; USGS 2007; Farr et al., 2007) for the region by dividing the region into 3704 6 micro-watersheds and calcul ating the negative exponential decay (-0.01) of distance upstream from each watersheds pourpoint to calculate the distance between the el evation at each point al ong each stream and the elevation at the pourpoint of the stream reach. I used this scale to estimate the reasonable flood/non-flood elevation at eac h point upstream from the pourpoint. Directive 42.a. Permit agro-pastoral uses only under suitable soil morphological conditions, prohibiting agro-pastoral uses in sensitive environments without appropriate slope and soil condit ions, particularly on sandy and alluvial soils that are not well drained, because of the importance of these areas to the stability of the local and regional hy drological regime. I obtained a thematic map of soil types for the r egion (SEPLAN/SEMA-MT) and identified all soil types that can be categorized as sandy or alluvial. In addition, I used the SRTM data to derive a map of slope and then identified all areas with a slope greater than 10% as being restricted, as this is the maximum value defined by the Brazilian Forest Code, above which permanent vegetation cover is required to be maintained. Finally, I combined both restricted soils and slopes into one layer to represent the spatial restrictions represented by this directive. Directive 81. Strictly protect aquifer recharge zones under campos umidos and murunduns which are sensitive habitats essential for the maintenance of water resources, eliminating any interference or establishment of structures that alter the hydrological regime and accelerate erosion processes; these areas should be assigned to legal reserves. I identified wetland areas by merging all ar eas identified as wetlands in 3 Landsatbased classifications for the years 1996, 2005, and 2007. Classification accuracy for wetlands in these classifications separately ranged from 0.88 to 0.91 (mean 0.90). 6 This set of microbasins includes all microbasins in the region, not ju st those outside of protected areas which comprise the set used in the remainder of the analysis. 102

PAGE 103

I merged the layers generated for each of these restrictions to obtain a single restriction map that was us ed in the development of the ZSEE scenarios. I adapted the CFC-NC scenario described above by assigning each micro-watershed to a zone and altering the percentage of allowable defores tation and required reforestation in each watershed according to zone as well as to biome. Furthermore, the model reforested both the riparian zone and the ar ea indicated by the composit e map of restrictions; both areas were subtracted fr om the total area of each watershed before the model calculated the amount of defores tation or reforestation that should take place in each zone. Analyses Vegetation cover For each scenario, I calculated the amount of remnant native forest and cerrado vegetation remaining. I also calculated t he amount of restored v egetation of each type that would be necessary to meet the requirements of each scenario. Restored vegetation also includes areas that ma y be left to regenerate naturally; however, throughout the chapter I will refer to restored vegetation to identify this category of vegetation. Economic aspects To evaluate the economic aspects related to compliance with each of three policy alternatives in the forest biome in compar ison with the current date, I calculated the difference in the area of t he watershed outside of federal and state protected areas (1) available for agricultural produc tion, and (2) that would need to be restored to come into compliance under each scenario. For each, I present the total area for the entire headwaters region as well as the mean per microbasin. Based on these figures, I 103

PAGE 104

estimated (1) the net present value (NPV) of agricultural activities carried out on legally cleared lands, and (2) the cost of riparian fo rest restoration over the whole landscape and by microbasin, where necessary (both me thods are described in detail in Appendix C). Ecological consequences I compared the final landscapes for each of the 4 alternative landscapes in terms of carbon stocks, river disc harge, annual evapotrans piration, terrestrial habitat quality, and water quality (methods described in Appendi x C). Unlike all the preceding analyses, ecological consequences were aggregated fo r the entire headwaters region, including all protected areas. Results Remnant Forest and Cerrado The impacts of the policy scenarios analyz ed here are best underst ood in light of the influence of each scenario on mandatory forest cover at the level of the microbasin. The areal coverage of each ecosystem (forest versus cerrado, remnant vegetation versus restored/regenerated vegetation), the potential area and associated economic value of land cleared for agriculture, and the ecological integrity of the Xingu River headwaters region, all varied as a function of the microbasin-level legal reserve requirement. The most restrictive scenario is the current forest code without compensation of the legal forest reserv e (CFC-NC), which requires that every microbasin outside of the prot ected areas maintain 80% fo rest cover or 35% cerrado cover. This policy scenario allows the clearing of approximately 6,500 km2 of extant forest and a similar area of cerrado vegetat ion present in the 2005 landscape that exceeded the microbasin-level legal reserve requirement representing a 17% reduction 104

PAGE 105

below the 2005 landscape (Table 3-1, Figure 31). The CFC-C scenario, which imposes the Forest Code but allows microbasins to achieve the legal reserve requirement through compensation in other microbasins, is virtually the same as the CFC-NC scenario since there was very little fore st or cerrado vegetation available for compensation. This is somewhat contrary to the expectation t hat remaining native vegetation would be substantially greater in the CFC-C scenario than the CFC-NC scenario. The ZSEE (ecological-economic zoning) scenario lowered the mandatory forest legal reserve to 50% in some regions, but only decreased the area of remnant vegetation remaining by 1,400 km2 relative to the CFC-NC and CFC-C scenarios. In other words, the ZSEE lowered the microbasin -level legal reserve requirement to 50% in some regions, but only 1,400 km2 of forest were in exce ss of 50% forest cover in these zones (Table 3-1, Figure 3-1). Restored Forest and Cerrado All three policy scenarios allowed some of the remnant vegetat ion present in 2005 to be cleared, but they also required the re generation/restoration of a greater area. Seventeen thousand square kilometers of restored/regenerated forest are required under these scenarios co mpared with 11,000 km2 under the Ecological-Economic Zoning scenario (Table 3-2). It is largely because of this requirement for regenerated/restored forest and cerrado that the total cove r of natural forests and cerrado was greatest in the CFC-C and CF C-NC scenarios with ap proximately 95,000 km2 (70% of the total area outside of protected areas in the Xingu River headwaters). The ZSEE scenario, which permits some micr obasins to fall to 50% forest cover, had only 3500 km2 (3%) less vegetation cover (Figure 3-2, Table 3-1). Hence, mean native 105

PAGE 106

vegetation cover per microbas in was greatest in the CF C scenarios3 (6scenarios (8 km2) and only half that in the ZSEE scenario (4 km2; Table 3-2). Native Vegetation Cover Native vegetation cover outside of protec ted areas and outside of the riparian zone was highest in the current (2005) landscape, in both the forest and cerrado biomes, owing to the fact that some microbasins could still legally clear vegetation under all three of the alternative po licy scenarios. Among those, however, the ZSEE scenario had the highest amount of native forest remaining, 8498 km2 (11%) less than in 2005 (Table 3-2). In contrast, both the CFC-NC and CFC-C scenarios had 17% less native vegetation than the current landscape. At the individual microbasin level, native vegetation cover ranged between 28 km2 and 34 per microbasin across the 4 landscapes (Table 3-1). Of the alternative land-use policy scenarios, the ZSEE scenario again had the highest mean vegetation cover per microbasin, after 2005, with 30 km2. Potential Agricultural Area The potential agricultural area associat ed with each scenario depends upon the microbasin-level requirements for natural vegetation cove r and the modeled clearing of native vegetation across the region. Land that is legally cleared, or that could be legally cleared for potential agricultural production was highest in 2005. The ZSEE scenario had only 1000 km2 less potential agricultural area than the 2005 landscape despite restoring nearly 10,000 km2 of vegetation throughout the l andscape (Table 3-3, 3-5). The ZSEE scenario had 4000 km2 less area available within the forest biome than in 2005, but nearly 3000 km2 more in the cerrado biome. Bu t lowering the legal reserve requirement in many microbasin s of the forest biome, t he ZSEE permits approximately 7,000 km2 more forestland to go to agriculture than the CFC scenarios but allows 3,000 106

PAGE 107

km2 less cerrado vegetation to be converted to agriculture (Table 3-4). This biomespecific influence of the ZSEE can be expl ained on the basis of the reduced demand for forest restoration in the ZSEE scenario. Mi crobasins in Zone 2 that had cleared more than 20% of their forests are not required to reforest back up to 80% coverage in the ZSEE as they are in the CFC scenarios. There was less cerrado conversion because the 3 biophysical restrictions included in t he ZSEE protected areas in the cerrado biome that were not protected in the CFC scenarios. In addition, requiring less reforestation in Zone 2 prevented clearing from being displac ed to the cerrado biome as extensively as in the CFC scenarios. The CFC-NC and CFCC scenarios had the least amount of area legally available for agricultura l production, approximately 41,000 km2 (Table 3-3). These rankings were maintained at t he microbasin level, with the ZSEE scenario allowing an average of 20 km2 per microbasin of legally cleared lands, compared with 21 km2 in the current landscape and 18.5 km2 in the CFC-NC and CFC-C landscapes. Potential Net Present Value The difference in area that could be lega lly cleared for agricultural activities between the ZSEE and 2005 landscapes represented a decrease of 1.6 billion dollars over the entire region (Table 3-4). In contrast, the CFC-NC and CFC-C scenarios would reduce potential NPV in the region by 4 billi on USD from the current landscape. Hence, the ZSEE reduces the opportunity cost of forest conservation by 2.4 billion dollars. For the individual landholder, the average potential NPV per micr obasin under the ZSEE scenario would be 0.7 million USD lower than in the 2005 landscape and 1.2 million USD more than the CFC-NC and CFC-C landscapes (Table 3-4). 107

PAGE 108

Costs of Restoring Forest and Cerrado Restoration costs were highest in the CFC-NC and CFC-C scenarios, totaling just over 2000 million USD for the entire region, 1.6 times more than for the ZSEE scenario (Table 3-6). For the average individual landh older, the restoration burden was lowest both in terms of area and cost in the ZSEE scenario, as well. The mean area of legal reserve to be restored in the ZSEE scenario (4.4 km2) is approximately half that of the other two policy scenari os (approximately 8 km2), with the bulk of restoration needing to occur in the forest biome. The mean cost of restoration (for legal reserve and riparian zone restoration combined) was approximatel y 0.8 million USD per microbasin in the ZSEE scenario, approximately one-third less than in the CFC-C and CFC-NC scenarios (Table 3-6). Overall, when combined with the loss in potential NPV, restoration costs doubled the cost of adopting the ZSEE scenario co mpared to the current (2005) scenario, reaching 1.5 million USD. Under the CF C-NC and CFC-C scenarios, restoration increased the average cost to the landholder to 3.2 and 3.1 million USD. The total cost of legal compliance under the CFC-C AND CFC-NC scenarios, combining the opportunity costs (Table 3-5) and the increase in restoration costs relative to the current (2005) landscape (Table 3-6) was approximately USD 6620 m illion for the entire region and averaged USD 3.1 million per microbasin. Ecological Consequences Carbon stocks Contrary to expectations, the ZSEE sc enario achieved higher overall emissions reductions than either the CFC-NC or CF C-C scenarios when each is compared to the 2005 (current) landscape (only 33 MtC emitted vs. 47 MtC) (Table 3-7). This can 108

PAGE 109

primarily be attributed to t he greater area of forest and cerrado woodland savanna that is maintained or restored under the former scenario in areas defined as restricted by the zoning plan, despite also relaxing the require ments of the Forest Code in 60% of the area outside protected areas. Although exis ting protected areas contain the highest biomass stands in the region, since both these and the riparian zones are strictly protected under all three alternative policy scenarios, the additional carbon stocks in the CFC-NC and CFC-C scenarios are entirely attributable to pr otection and restoration of forest and woodlands on private lands. Rest ored vegetation contributed an estimated 23 MtC to the total carbon stock in both the CFC-C and CFC-NC landscapes over a 30year time period, about 1.6 times more than in the ZSEE landscape (Table 3-7). Hydrology and regional climate All three scenarios show decreases in stream discharge in comparison with the 2005 landscape, ranging from 2 to 3% less t han that of the curr ent landscape (Table 37). The greatest decrease in discharge occu rs in the CFC-C and CFC-NC scenarios. Mean annual evapotranspiration decreases as forest cover decreases in the 3 alternative policy scenarios, although all ex hibit a 4% reduction from the control scenario (Table 3-7). However, all three hav e increased evapotranspiration by 1% over the current landscape. Water quality All three of the alternative policy scenarios had all of the riparian forests protected or restored, thus riparian forest cover increased by approximately 2600 km2 (20%) over the 2005 landscape (Table 3-7). Thus, in all ca ses, 20% of streams would be likely to have lower temperatures and higher dissolved oxygen levels (Neill et al. 2006) than the current landscape, affecting species popula tions and assemblages. The proportion of 109

PAGE 110

microbasins having less than 40% vegetation co ver was lowest in the ZSEE scenario (Table 3-7). This indicates that although this scenario did not have the highest proportion of microbasins with 60% or more vegetation cover (both CFC scenarios had more), more microbasins would be likely to maintain basic hydrological functions (Coe et al., 2009). Habitat Overall, habitat quality and quantity differed between scenarios by biome. For the forest biome, habitat quality and quantity ar e lowest in the 2005 landscape, and highest in the CFC-NC and CFC-C scenarios (Table 3-7). The CF C-NC and CFC-C landscapes have the highest amount of total forest cover, the smallest number of forest fragments, the greatest mean forest fragment size, as well as the greatest amount of core or interior area (representing 90% of the total fo rest area in the landscape) (Table 3-7). In contrast, the 2005 landscape compared most fa vorably for the cerrado biome, with the ZSEE scenario indicating the next best outcome for the biome. Tota l vegetation, mean fragment size and interior habitat area were greatest in the 2005 landscape. Total edge area was lowest by a factor of 1.5 (in the fo rest biome) to 1.3 (in the cerrado biome) in the 2005 in comparison with the three alternative policy sc enarios. Among the modeled scenarios, the CFC-C had the least amount of edge forest habitat. Discussion The hybrid regulatory-economic policy instru ments examined in this study differ in their degree of flexibility and responsiveness to frontier dynami cs, with important implications for ecological and economic processes in the Xingu River headwaters region. The Brazilian Forest Code provision that allows compensation of the legal reserve among property holders (CFC-C scenario) is intended to reduce the economic 110

PAGE 111

costs of the Forest Code by allowing more agricultural expansi on on lands with higher returns from deforestation-dependent agricultura l activities, such as soy production and cattle ranching, in exchange for reduced deforestation on lands with lower potential returns from soy and cattle. In the case of the Xingu Rive r headwaters region, however, this provision had virtually no effect on native ecosystem cover when compared with the Forest Code with no compensation (CFC-NC sc enario), in part because of the strict eligibility criteria, t he small amount of excess forest or cerrado vegetati on in microbasins having more than 80% or 35% cover, respecti vely, and possibly because of restrictions in the model that limited trading to a 4th-order watershed level (Table 3-4). This form of flexibility in land use regul ations must be functional ear ly in the evolution of an agricultural frontier to signific antly reduce the economic costs of compliance. In the case of the Xingu River headwaters region, there may have also been a strong reluctance on the part of landholders to us e the compensation pr ovision after it was enacted, in 2000 7 because of the great uncertai nty surrounding the legal reserv e requirement itself. To use the compensation provisi on, landholders must alter t heir land titles to forgo development rights on all remaining forests or cerrado on their property as they enter into a financial compensation contract with another property holder. These restrictions become unpalatable for many when there is a perceived likelihood that the legal forest reserve requirement could be reduced perm anently, especially when severe forms of punishment for non-compliance can be av oided through bribes (Rosenthal, 2009). The Brazilian socio-economic and ecologic al zoning policy (ZSEE) is a far more sophisticated instrument than the legal rese rve compensation provision in that it 7 The provision was included in the revised Forest Code in late 1998, but landholders were identified as eligible depending on their vegetati on cover status at the end of 1999. 111

PAGE 112

responds to biophysical and infra-structural (transportation, urban centers, electrical supply) features of landscapes to restrict clearing for agriculture and landscape in areas of marginal suitability and liberate clearing in areas of agricultura l consolidation. The modeled ZSEE scenario, using economic rent models that are responsive to the same factors that impinge upon the profitability of agriculture that the ZSEE is designed to represent, demonstrates the potential for this instrument to considerably reduce the economic costs of legal compliance while pr otecting public interests in private land native ecosystems and their services. T he ZSEE reduces the opportunity costs associated with full compliance with the Br azilian Forest Code (with no compensation) from 4.2 to 1.6 billion dollars (Table 3-5), primarily by allo wing a greater share of cleared lands that are non-compliant wit h the 80% or 35% legal reserve requirement to stay in production. More specifically, in Zone 2 of the ZSEE scenario, microbasins that have cleared more than 80% of forest cover or 35% of cerrado cover are not required to reforest (or compensate) the difference. For the average l andholder, this means that the net present value of a landholding is worth 1.2 million dollars more than under the CFC scenarios, and only 0.7 million dollars less than in the current (2005) landscape (Table 3-5). The ZSEE also has features that impr ove the ecological performance of the landscape in ways that are not captured by par ameters such as the total cover of native vegetation. For example, agriculturally marginal lands, with steep slopes and high erodibility or that are locat ed in wetland or floodplain area s, are protected under the ZSEE. In the Xingu River headwaters region, th is feature of ZSEE is manifested as a greater coverage of cerrado vegetat ion. The ZSEE protects 3,100 km2 more remnant 112

PAGE 113

113 cerrado vegetation than the CFC scenarios, largely because of their marginal suitability for agriculture (Table 3-1). By most ecological parameters, the Z SEE scenario is nearly indistinguishable from the CF C scenarios, despite its far lower opportunity costs. The ZSEE scenario has the same river discharge, evapotranspiration, and nearly the same number of microbasins with at least 60% native ecosystem coverage (75% vs. 80% and 81% for the CFC scenarios). The ZSEE scenario has smaller carbon stocks (1924 vs. 1909 and 1907) than the CFC scenario landsc apes, but less interior habitat (99,000 km2 vs. 105,000 km2) than the CFC scenario landscapes. Conclusion Hybrid regulatory-economic policy instru ments may be particu larly appropriate for achieving land-based economic a nd ecological outcomes. Un like pollution regulation, where environmental impacts are lowered by reducing em issions of a pollutant regardless of the exact location of this emi ssion, ecosystem services are best protected through instruments that guarantee that these services will be s patially distributed. This is a positive feature of the Forest Code, that requires t hat all riparian zone forests be protected and a portion of the native ecosyst ems of every microbasin. The economic component of one of the hybrid instruments that I evaluat ed, the ZSEE, lowered the cost to landholders and to secondary and te rtiary industries that depend upon agricultural production, with only a small sacrifice of ecological performance. Albeit not a perfect solution to balancing agricultural produc tivity with environmental sustainability, in this landscape it may be among t he optimal policy frameworks.

PAGE 114

Table 3-1. Total area and mean area per microbasin of re mnant forest and cerrado woodla nd within the Xingu River headwaters region in 2005 and under 3 land-use policy scenarios. Data are for microbasins outside of protected areas. CFC-NC is the abbreviation for Current Forest Code, No Compensation. CFC-C is the abbreviation for Current Forest Code with Compensation. ZSEE is the abbreviation for Zoneamento Socio-Economico Ecologico (Socioeconomic Ecological Zoning). 2005 CFC-NC CFC-C ZSEE Total (km2) Mean ( s.e. ) per microbasin (km2) Total (km2) Mean ( s.e. ) per microbasin (km2) Total (km2) Mean (s.e. ) per microbasin (km2) Total (km2) Mean ( s.e. ) per microbasin (km2) Remnant native vegetation Forest 61,934 35.3 (0.8) 55,416 31.6 (0.7) 55,410 31.6 (0.7) 56,882 32.4 (0.8) Cerrado 13,065 28.5 (1.5) 6547 14.3 (0.8) 6558 14.3 (0.8) 9619 21.0 (1.1) TOTAL 74,999 33.9 (0.7) 61,963 28.0 (0.6) 61,968 28.0 (0.6) 66,501 30.0 (0.6) 114

PAGE 115

Table 3-2. Total area and mean area per microbasin of forest and cerrado within the Xingu River headwaters region that will consist of restored vegetation in 2005 and under 3 land-use policy scenarios. Data are for microbasins outside of protected areas. Scenario abbr eviations are as in Table 3-1. CFC-NC CFC-C ZSEE Restored Vegetation Total (km2) Mean (s.e. ) per microbasin (km2) Total (km2) Mean ( s.e. ) per microbasin (km2) Total (km2) Mean ( s.e. ) per microbasin (km2) Legal Reserve Forest 17,101 9.7 (0.4) 17,235 9.8 (0.4) 9090 5.2 (0.3) Cerrado 722 1.6 (0.2) 687 1.5 (0.3) 565 1.2 (0.1) TOTAL 17,823 8.0 (0.3) 17,922 8.1 (0.3) 9654 4.4 (0.2) Riparian Zone Forest 1666 1.0 (0.04) 1666 1.0 (0.04) 1666 1.0 (0.04) Cerrado 835 1.9 (0.1) 835 1.8 (0.1) 835 1.8 (0.1) TOTAL 2501 0.9 (0.03) 2501 0.9 (0.03) 2501 0.9 (0.03) 115

PAGE 116

Table 3-3. The area and percent coverage of forest and cerrado vegetation (remnant and restored combined) in the Xingu River headwaters region as observed for 2005 and modeled under three land-use policy scenarios. Data are presented both for the ent ire headwaters region and for the mean va lues of the regions microbasins. Results are for lands outside of protected areas Scenario abbreviations are as in Table 3-1. 2005 CFC-NC CFC-C ZSEE Area (km2) % Area (km2) % Area (km2) % Area (km2) % Total Natural Vegetation (remnant and restored) Whole watershed (Total) Legal Reserve Forest 61,934 64 75,590 80 75,718 80 69,044 71 Cerrado 13,065 45 9265 35 9241 35 12,194 42 Total 74,999 60 84,855 68 84,959 67 81,138 64 Riparian Zone Forest 6052 78 7736 100 7736 100 7736 100 Cerrado 1695 67 2530 100 2530 100 2530 100 Total 7747 76 10,266 100 10,266 100 10,266 100 Legal Reserve and Riparian Zone 82,746 60 95,121 69 95,225 70 91,404 67 Microbasins (Mean (s.e.)) Legal Reserve Forest 35 (0.8) 68 (0.6) 43 (0.9) 80 (0.3) 43 (0.9) 78 (0.3) 39 (0.8) 74 (0.4) Cerrado 28 (1.5) 49 (1.2) 20 (1.1) 35 (0.8) 20 (1.1) 33 (0.8) 27 (1.3) 45 (1.0) Riparian Zone Forest 3.5 (0.1) 79 (0.5) 4.4 (0.1) 100 (0) 4.4 (0.1) 100 (0) 4.4 (0.1) 100 (0) Cerrado 3.7 (0.2) 69 (0.9) 5.6 (0.3) 100 (0) 5.6 (0.3) 100 (0) 5.6 (0.3) 100 (0) Total 3.5 (0.1) 77 (0.5) 4.7 (0.1) 100 (0) 4.7 (0.1) 100 (0) 4.6 (0.1) 100 (0) 116

PAGE 117

117 Table 3-4. Total area and mean area per microbasin of fore st and cerrado within the Xingu River headwaters region that is available for agricultural production in 2005 and in 3 land-use policy scenarios. Data are for microbasins outside of protected areas. Available for agricultural produc tion refers to areas that already are or could be legally cleared without ex ceeding the maximum percent clearing allowable under each scenario. 2005 CFC-NC CFC-C ZSEE Area available for agriculture Total (km2) Mean ( s.e. ) per microbasin (km2) Total (km2) Mean ( s.e. ) per microbasin (km2) Total (km2) Mean (s.e. ) per microbasin (km2) Total (km2) Mean ( s.e. ) per microbasin (km2) Forest 31,419 17.9 (0.6) 20,812 11.8 (0.3) 20,881 11.9 (0.3) 27,429 15.6 (0.5) Cerrado 14,110 30.8 (2.0) 19,931 43.5 (2.2) 20,082 43.8 (2.2) 17,003 37.1 (2.2) TOTAL 45,529 20.6 (0.6) 40,743 18.4 (0.6) 40,963 18.5 (0.6) 44,432 20.0 (0.6)

PAGE 118

Table 3-5. Potential net present value (NPV) associat ed with soybean cultivation or cattle production on cleared land for the Xingu River headwat ers region under 3 alternative policy scenarios and in 2005. The opportunity cost of adopting each scenario (relative to the 2005 land scape) is also presented. Data are presented for the entire region and per microbasin, for areas outside of protected areas in the forest biome region. 2005 CFC-NC CFC-C ZSEE NPV (million USD) NPV (million USD) NPV (million USD) NPV (million USD) Potential NPV Whole basin 13,755 9522 9529 12,104 Mean per microbasin (s.e.) 6.2 (0.3) 4.3 (0.2) 4.3 (0.2) 5.5 (0.2) Opportunity Cost Whole basin n/a 4233 4228 1651 Mean per microbasin (s.e.) n/a 1.9 (0.2) 1.9 (0.2) 0.7 (0.1) 118

PAGE 119

Table 3-6. The estimated costs of restor ation of the legal re serve and riparian zone under 3 alternative policies, for the ent ire basin in the aggregate and for the average microbasin out of compliance. Ranges refer to upper and lower estimates of restoration costs per hectare, as defined in Table C-1. CFC-NC CFC-C ZSEE Whole Basin (million USD) Whole Basin (million USD) Whole Basin (million USD) Legal Reserve Forest 1593 (+/676) 1605 () 847 () Cerrado 67 () 64 () 53 () Riparian Zone 366 () 366() 366() Total 2026 () 2035 () 1266 () Mean per microbasin (million USD) Mean per microbasin (million USD) Mean per microbasin (million USD) Legal Reserve Forest 0.9 (.4) 0.9 (.4) 0.5 (.2) Cerrado 0.2 (.1) 0.1 (.1) 0.1 (.1) Riparian Zone 0.2 (. 1) 0.2 (.1) 0.2 (.1) Total 1.3 (0.6) 1.2 (.6) 0.8 (.4) 119

PAGE 120

Table 3-7. Ecological attributes of the Xingu River headwaters region (including protected areas) in 2005 and of modeled la ndscapes representing the region under 3 policy alternatives on private lands. These attributes include: carbon stocks, surface hydrology and regional c limate, indicators related to water quality, and terrestrial habi tat quantity and quality. Scenarios Indicator 2005 CFC-NC CFC-C ZSEE Carbon Stocks Stored in native vegetation (MtC) 544 497 497 511 (MtCO2e) 1995 1824 1822 1873 Stored in restored vegetation (30 years) (MtC) 5 23 23 14 (MtCO2e) 17 85 85 51 Total (MtC) 549 520 520 525 (MtCO2e) 2012 1909 1907 1924 Emissions since initial year (MtC) n/a 47 47 33 (MtCO2e) n/a 171 173 122 Surface Hydrology and Regional Climate Mean Annual Discharge (m3 s-1) (% change from potential) 3128 (13%) 3032 (10%) 3033 (10%) 3070 (11%) Mean Annual Evapotranspiration (m3 s-1) (% change from potential) 6583 (-5%) 6679 (-4%) 6678 (-4%) 6641 (-4%) Water Quality Riparian forest cover, 50-m buffer (km2) 12,931 15,509 15,509 15,509 Mean % native vegetation cover per microbasin (s.e.) 69 (0.5) 73 (0.4) 73 (0.4) 73 (0.4) % of microbasins with greater than 60% vegetation cover 65 80 81 75 % of microbasins with less than 40% vegetation cover 18 16 16 11 Terrestrial Habitat Vegetation cover (km2) Forest 107,789 116,487 116,395 110,902 Cerrado 17,037 13,297 13,149 15,209 Number of fragments Forest 13,427 12,666 12,958 13,777 Cerrado 13,285 18,238 18,138 15,571 Mean distance to nearest neighbor fragment (m) Forest 361 381 379 363 Cerrado 406 371 371 378 Mean fragment size (ha) Forest 803 920 898 805 Cerrado 128 73 73 98 Total interior habitat area (km2) Forest 99,978 104,754 105,287 98,886 Cerrado 12,737 7652 7433 9762 Total edge habitat area (km2) Forest 7810 11,732 11,108 11,969 Cerrado 4300 5645 5716 5472 120

PAGE 121

Figure 3-1. Location and distri bution of the major biophysical restrictions (shown in red) and the 4 major zones of the Mato Gro sso State Zoning Plan (ZSEE) within the Xingu River headwaters: (a) fl oodplain areas; (b) soil and slope restrictions; (c) wetland areas; (d) the 4 major zones (noted by number) combined with the 3 biophysical restrictions. 121

PAGE 122

Figure 3-2. Comparison of land-cover in the basin for (a) 2005 (observed), (b) full compliance with 80% legal reserve in forest biome, no compensation (CFCNC) (modeled), (c) full compliance with 80% legal reserve in the forest biome with compensation (CFC-C) (modeled), and (d) Mato Grosso state socioeconomic ecological zoning plan (ZSEE) (modeled), with compensation. 122

PAGE 123

CHAPTER 4 THE COST OF CARBON TO OFFSET OP PORTUNITY COSTS ON PRIVATE LANDS UNDER ALTERNATIVE POLICIES IN THE UPPER XINGU RIVER BASIN Introduction Approximately 17% of globa l greenhouse gas emissions are estimated to come from the clearing and degradation of tropical forests (I PCC, 2007). In an attempt to reduce these emissions, negotiators within the UN Framework Convention on Climate Change (UNFCCC) are designing a mechanism for compensating tropical nations that succeed in reducing carbon emissions from deforestation and forest degradation, known by the acronym REDD (from R educing Emissions from Deforestation and Degradation; Gullison et al., 2007). REDD has t he potential to both increase the scale of tropical forest conservation and significantly reduce greenhouse gas emissions. REDD could also become the vehicle for protecting and restoring the role of tropical forests in the provision and regulation of pure freshwater, biodiversity conservation, soil conservation, and the protection of regional climate systems (Stickler et al., 2009), channeling financial resources from either a new forest carbon market or from expanded carbon-relat ed donations from developed countries. As negotiations of the REDD mechanism within the UNFCCC framework move toward fruition, federal and state governments of tropi cal nations are developing programs to achieve reductions in deforesta tion and forest degradation. Brazil has been developing a program to reduce emissions from deforestation in the Amazonthe countrys largest source of GHG emissions (76%, 1.2 billion tons CO2e; MCT, 2009). Under the Brazilian National Climate Policy, the government established a target of 80% reduction in deforestation in the Amazon region by 2020 in 2009 (GOB 2008; Nepstad et al., 2009). With a USD 1 billion commitm ent from the Norweg ian government, Brazil 123

PAGE 124

has already established a fund to finance pr ojects and programs for achieving these emissions reductions (BNDES, 2009). The Amazon Fund, as negotiated with Norway, would be replenished as long as Brazil continues to reduce deforestation in the Amazon region below its historical baseline. The initial payment of USD 110 million from Norway was in recognition of deforestation reduction s already achieved since 2005. Though not officially considered a REDD program by t he Brazilian government for all intents and purposes, it serves as the first regionalor national-scale REDD experiment in the world. However, the Amazon Fund does not have a well-developed plan for achieving reductions in deforestation and forest degradation in the Amazon region. REDD land use policy design is advancin g within the Amazon states of Brazil. Under the Brazilian National Climate Policy, Amazon state governm ents are expected to establish their own deforestation reduction tar gets that, combined, pr ovide the requisite reductions for the entire region (Tollefson, 20 09). Mato Grosso state has historically been responsible for the greatest amount of deforestation (39% of annual regional average for 1996-2005 time period; INPE, 2009), but from 2006-2009, it has been responsible for 59% of deforestation reducti on already achieved under the BNCP goals over the last 4 years (IN PE, 2009; GOMT, 2009). Furthermo re, in its own plan, Mato Grosso is committing to 89% reduction in deforestation by 2020, 9% more than the national plan. This target represents more than 60% of the national target for deforestation reduction in the Amazon, and approx imately 40% of Brazil s total goal of GHG emission reduction by 2020. According to the states draft REDD plan, the implementation of REDD is considered vital to achieve Mato Grossos deforestation reduction targets to finance 124

PAGE 125

credits for emissions reductions financed by the Amazon Fund and other voluntary and potential official market mechanisms. The em issions reductions effectively achieved during a given period of time are intended to provide financi ng for future deforestation reductions. REDD Certificates are proposed to be allocated between six programs for (1) indigenous peoples; (2) protected areas; (3) private forests; (4) smallholder settlements; (5) state governance ; and (6) an insurance fund. During the first phase of implementation, indigenous lands, protec ted areas, and smallholder settlements will participate in the mechanism through priority pilot projects that will help to develop adequate approaches and methodolog ies for state-wide programs. The private forests program, linked to the State environmental licensing syst em of rural properties (Azevedo, 2009), will provide payments for forest reserves located in rural properties that fully comply with the feder al and state environmental laws. In this paper, I attempt to pr ovide input to the policy fr amework design of the Mato Grosso REDD program, specifically to the design of the program to reduce deforestation on private lands. I examine the extent to whic h existing forest and landuse policies might be used as a basis for ef fectively applying REDD or other carbon credits in the upper Xingu River basin in nor theastern Mato Grosso one of 3 pilot sites for designing and testing the state level program (GOMT, 2009). I estimate the emissions reductions below (a) a historical baseline and (b) 3 BAU scenarios that could be achieved on private lands in the region under 3 alternative policy frameworks. I also estimate the price of carbon that woul d be needed to offset the opportunity costs incurred by private landholders of complyi ng with each alternative framework used to achieve target (portion coming from private lands). Finally, I compare the alternative 125

PAGE 126

scenarios with respect to a series of ecologi cal indicators to evaluate the performance of each in providing co-benefits beyond carbon stocks. Study Area The Xingu headwaters region is representative of many areas along the Amazons agricultural frontier, with expanding production of cattle and soy (70% of the area) surrounding smallholder settlements (3 %) and largely-forested indigenous lands (approximately one quarter) (Figure 2-1a). T he stream and river ecosystems are under growing threats from sediment ation, agrochemical run-off, and associated fish die-off from the unprotected headwaters regions outside of the indigenous reserve, which is located at the core of the region (Figure 2-1a). The Basin supports cerrado (savanna woodland) in the south and dense humid forest in the no rth (Figure 2-1a). The Xingu region is also an advanced laboratory for explor ing the potential ecological co-benefits of al ternative approaches to REDD plans. The region (177,780 km2) is larger than 90% of t he tropical nations that could seek participation in REDD within the UNFCCC. The Xingu is also the si te of a 5-year multi-stakeholder campaign to protect water resources, particularly thr ough efforts to protect and reforest riparian forests in the region (Y ikatu Xingu, 2009). The campaign has been moderately successful, and the prospect of carbon funds represents an opportunity to continue funding and expanding efforts related to st ream health and other ecosystem services that are important to the regions inhabitants. Materials and Methods I compared 3 alternative land-use policie s that represent po ssible approaches to the implementation of emissions reductions targets over 30 years in the Xingu headwaters region. I developed land-cover maps representi ng each of the alternative 126

PAGE 127

scenarios, as well as 3 business-as-usual (BAU) scenarios. I used these maps to estimate and compare the volume of carb on stocks that would be maintained under each scenario, in both remaining native and restored or regener ating forest or cerrado I then used the maps to calculate the cost of (1) reducing carbon emissions, and (2) restoring or enhancing carbon stocks under co mpliance with each of the 3 alternative policy scenarios relative to each of three business-as-usual (BAU) scenarios. Finally, the maps were also used as inputs for calcul ating several ecological indicators as a basis for assessing the potential ecological co-benefits of REDD. Model Development Six land cover maps corresponding to thr ee alternative policy scenarios and three business-as-usual (BAU) scenarios were deve loped using a spatiall y explicit dynamic landscape simulation model descr ibed in Appendix A. This spat ial-statistical model of land-use change was derived from a land-us e/land-cover change analysis and a GIS consisting of data related to the location an d neighborhood attributes (e.g., distance to roads, distance to streams, slope, agricultural suitability) of 4 focal land-use transitions: (1) forest agriculture (pasture or annual crops); (2) cerrado agriculture; (3) agriculture regenerating forest; and (4) agriculture regenerating cerrado For the BAU scenarios, the model simulates land-cover change over 30 time steps, beginning in 2005 and ending in 2035, using land-cover c onversion probabilities and rates calculated from the 1996-2005 reference period (Appendix A). High and low BAU scenarios were developed in an attempt to bracket th e range of likely future deforestation by applying the highest and lowest historical obse rved deforestation rates, respectively, as described below. For the policy scenarios, maximum required reforestation and maximum allowed clearing were imposed upon the Xingu River headwaters region 127

PAGE 128

landscape, independent of the amount of time required to reach these levels. The basic architecture and function of the simulation model is described in further detail in Appendix A. Scenarios Each of the scenarios is based on recen t, existing and/or proposed legislation (Chapter 2, Chapter 3) and was compared with a range of BAU simulations that assume no REDD interventions. The basic assumptions underlying each scenario, including the reference scenarios, are as follows: Business-as-usual scenarios Business as Usual (BAU) assumes that the histor ical rate and pattern of deforestation continues into the future, and thus serves as a baseline model against which to compare other options. Here, I develop 3 BAU scenarios that represent alternative development trajectories with no additional REDD or other governance interventions (e.g., creating or enforcing protected areas). The average BAU scenario uses the average deforestation rate calculated for this region over the 1996-2005 period and applies it over 30 years. The refer ence period corresponds to that set for determining crediting levels for the Amazon Fund (GOB 2009). Under the Amazon Fund, the reference scenario is estimated by extending the average rate for the 10-year period from 1996-2005 as an absolute (gross) rate into the future. In the analysis presented here, the BAU scenario is somewhat more conservative, applying the same annual rate of clearing as a percentage (net rate) of the remaining forest; thus the absolute amount cleared each year decreases proportionally with the decrease in total forest cover. This is a more realistic reference scenario fo r a region such as the Xingu River headwaters, as it has historically high levels of deforestation which are unlikely to 128

PAGE 129

be sustained at the same absolute level in the future. Under the Amazon Fund, the reference scenario is extended only thr ough 2020; here, however, extend the simulation through 2035 to provide an assessment of how pol icies aided by carbon offset to create incentives might protect ecologica l resources in the longer term. I also model two further BAU scenarios BAU High and BAU Low using a higher and lower annual deforestation rate, respectively, than the original BAU uses. The scenarios use deforestation rates observ ed in the region for the 2005-2007 and the 2001-2003 periods, respectively, to provide a range of reference values that reflect business-as-usual under alternative economic conditions. These periods represent the periods of highest and lowest observed def orestation rates in the region to date and thus represent realistic endmembers for a ttempting to bracket historic land-cover change patterns into the future. Policy scenarios I modeled three scenarios representing the Xingu River headwaters landscape under alternative extant or proposed federal and st ate policies (Chapter 2, Chapter 3). The Current Forest Code (CFC) scenario represents the landscape under the assumption that the current Federal Fo rest Code was perfectly implemented and enforced. Since 1996, the Fore st Code requires that properti es located in the forest biome in the Legal Amazon maintain 80% of the native vegetation in a permanent legal reserve; properties in the cerrado biome are required to maintain 35% of the native vegetation in a legal reserve (Chapter 2). Where less than this amount is present, the vegetation must be restored. In addition, vegetation within 50-m of each stream on private properties must be strictly pr otected or restored if it is absent. 129

PAGE 130

The Reduced Legal Reserve (RLR) scenario models the Forest Code under the law before 1996 and as was required by the Ma to Grosso state government until as recently as 2005, enforcing a legal reserve of only 50% on properties located in the Amazon forest biome (Chapter 2). The legal reserve on cerrado properties remains 35% and the riparian zone is also strictly protected. The Socio-Economic Ecological Zoning Plan (ZSEE) scenario assumes that the proposed Mato Grosso state zoning plan is implemented, as descr ibed in Chapter 3. The zoning plan has 4 major zones which determine the percent of legal reserve that is required for properties fallin g within each zone. The scenario modeled here also assumes that the tradable deforestation rights option included in the federal Forest Code in 1998 will be exercised (Chapter 3). Furthermore, t he scenario assumes strict protection of areas falling in any one of 3 areas described as requiring special attention and protection under the ZSEE (Chapter 3). Fina lly, as in the other two scenarios, all riparian areas within 50-m of streams ar e strictly protected and reforested. Analyses Emissions reductions and carbon enhancement For the initial landscape (2005) and each alternative scenario, I calculated the carbon (CO2e) stocks stored in remaining nativ e vegetation using a map of forest biomass developed for the entire Amazon basin (Saatchi et al., 2007, adapted in Nepstad et al., 2007a and Nepstad et al., 2009). To estimate the gain in carbon stocks re sulting from reforestation under each alternative policy scenario, I assigned a value of 1.5 tC ha-1y-1 and 0.5 tC ha-1y-1 (Houghton et al., 2000; Zarin et al., 2001) per pi xel, respectively, and multiplied by the 130

PAGE 131

number of years that regeneration in each pixel had tak en place (ranging from 1 to 29 years). I present the distributio n of carbon stocks among riparian and non-riparian zones separately. Furthermore, I present r egenerating and maintained carbon stocks aggregated over the entire Xingu River headwaters region outside of protected areas. Thus, I obtained estimates of total carbon emissions, of carbon emissions due to clearing of native vegetation, the potential emissions reductions that could be achieved, and the total carbon enhancement repres ented by each policy alternative. Price of carbon To calculate the minimum pr ice of a ton of carbon (CO2e, in this case) needed to offset the opportunity cost represented by each of the 3 policy scenarios for private lands in the basin in the aggregate, I divided t he difference in total opportunity cost of a given scenario and each of the BAU scenarios by the corresponding difference in carbon stocks in remnant vegetation. The opport unity cost was calculated by estimating the potential net present value (NPV) of forested (native or regenerating) lands maintained under each scenario associated wit h soy or cattle produc tion (the higher of the two) (Appendix C), and then calculating the difference in potential NPV between each policy scenario and each of the BAU scenar ios. I obtained a range of prices per ton of CO2e to offset the aggregate opportunity co st faced by private landowners under each policy scenario. In addition, I calculated the price per ton of carbon necessary to compensate private landholders for the co sts of restoring lands where restoration is necessary to comply with each alternative policy scenario. The costs of restoration are two-fold: (1) the foregone profits (opportunity cost) of th e land taken out of production, and (2) the 131

PAGE 132

direct expenditures for restoration. The di rect costs of restoration are likely to be negligible outside riparian areas, as landhold ers will simply allow the land to regenerate naturally (Chapter 3). However, here, I pr esent the range of pot ential direct costs associated with active restoration for illustrative purposes. The opportunity cost is calculated as above and the range of direct costs is calcul ated as described in Appendix C. The per unit price of car bon is calculated by dividing the sum of opportunity costs and direct restoration costs for each scenario by the corresponding volume of carbon accumulated on regenerating lands over a 30 year time horizon. We assume no significant amount of natural r egeneration or active restoration occurs in any of the BAU scenarios; thus, the amount of accumulate d carbon for each scenario represents the difference between the total observed regenerat ion for that scenario and the initial volume of regenerating vegetation in 2005. Eligible credits Finally, for microbasins locat ed in the forest biome only, where the legal reserve was 50% of private landholdings in the Amazon until 1996 (C hapter 2), I also present the total volume of credits and the necessary price for a restrict ed portion of carbon stocks which represent the carbon stocks held in forests in excess of 50% of total potential forest in each microbasin. This re striction represents a scenario under which governments limit compensation for the costs of forest maintenance and restoration to those landowners who contribute more to the maintenance of carbon stocks and provision of ecosystem services than t he minimum required under the law until 1996 (and until 2005, in Mato Gro sso) (Chapter 2). In this ca se, I assume that the law requires landholders to set aside at least 50% of their land in a forest reserve, for reasons explained in Chapter 2. This restriction addresses concerns both domestically 132

PAGE 133

and internationally that individuals should not be paid to comply with the law, and furthermore, that individuals who violate the law not be compensated for whatever amount of forest they do maintain. The cost of reducing emissions and the price of carbon for only this subset of the total car bon stock in the region were calculated as described above. The per unit price to compens ate restoration costs was not included. I present the results for this subset of microbasins in the region at the aggregate (landscape) level. Assessment of ecological co-benefits I compared the final landscapes for each of the 3 alternative scenarios and for each BAU scenario (for a total of nine comparisons) in terms of river discharge, annual evapotranspiration, habitat quality, and wa ter quality (methods described in Appendix C). Results Emissions Reductions and Carbon Enhancement The simulated policy scenarios affected land cover in the Xingu River headwater region by imposing restrictions on privateland forest and cerrado clearing and by defining forest/ cerrado restoration requirements for eac h privately-held microbasin. The reduced legal reserve (RLR) scenario was the most permissive, allowi ng forest clearing to proceed to 50% of each microbasin and requi ring forest restoration only up to 50% of each microbasin in cases where there was less than 50% forest cover during the base year, 2005. It is therefore not surprising t hat the RLR scenario resulted in the lowest amount of carbon in remnant native vegetation and restored vegetation of the three policy scenarios for private lands in the region (Table 4-1, Figure 4-1). 133

PAGE 134

The current Forest Code (CFC) scenario was the least permissive policy among those simulated. It restricts forest clearing to 20% of indi vidual microbasins and requires landholders to restore forest cover to 80% if clearing had already exceeded 20% during the base year, 2005. The socio-economic, ecol ogical zoning plan (ZSEE) scenario was the same as the CFC scenario except for one difference: landholders were required to restore forest cover only up to 50%, instead of 80%. As a resu lt of this difference, the CFC scenario had five times more restored fo rest carbon than the ZSEE and six times more than the RLR (Table 4-1). But the CFC had slightly less remnant native vegetation (1113 MtCO2e) than the ZSEE (1241 MtCO2e) (Table 4-1), because of the additional restricted areas created under the plan, in addition to the regulations defined for each zone (Chapter 3). The policy scenarios exerted a large impact on forest/ cerrado carbon stocks in the Xingu River headwaters region. The ZSEE a nd CFC scenarios maintain the greatest amount of carbon in remnant native vegetation relative to the other modeled scenarios (Table 4-1, Figure 4-1). They maintain approx imately 1.5 times as much carbon as the BAU Low and RLR scenario and 2 to 5 ti mes more than the BAU and BAU High scenarios. Notably, more native carbon sto cks are maintained (not emitted) under the BAU Low scenario than the RLR scenario. Under the worst case BAU scenario, BAU High, the least amount of ca rbon stocks were maintained, approximately 5 times less than either the ZSEE or CFC scenarios. In co ntrast, only a 2 to 3-fold difference in carbon stocks was observed between t he average BAU scenario and the CFC and ZSEE scenarios. 134

PAGE 135

Price of Carbon All carbon The price per ton of CO2e necessary to compensate all private property holders in the region for maintaining or restoring ca rbon stocks over and above the amount that would be maintained under the BAU scenario was greatest for the RLR scenario (Table 4-2). The price ranged from 25 to 54 USD tCO2e-1 (average 32 USD tCO2e-1), whereas under the CFC and ZSEE scenarios, the range of prices was both lower and narrower (22-26 and 22-28 USD tCO2e-1, respectively). However, the total cost of maintaining and restoring forests to reach compliance with each of the latter scenarios was 2 to nearly 3 times higher, respectively than for the RLR scenario (Table 4-2). Removing compensation for the cost (bot h opportunity and direct costs) of restoration reduces the per ton price substantially, in the case of the RLR scenario, with prices ranging from 18 to 21 USD tCO2e-1. Again, the range of prices for the other two scenarios is narrower and slightly lower, rangi ng around 18 and 17 USD tCO2e-1 for the CFC and ZSEE scenarios, respectively. Under the RL R and ZSEE scenarios, compensating for regeneration is expensive (the per unit price ranges around 102 and 103 USD tCO2e-1, respectively), as the amount of carbon accu mulated is relatively low and the costs are high (including both foregone potential income and the direct cost of restoration). In contrast, under the CFC scenario, where far mo re restoration is required, something of an economy of scale is achieved, and the pric e falls to a far more reasonable 36 USD tCO2e-1. This surprising result can be expl ained as the result of higher carbon accumulation in restored forests under this scenario and lower oppor tunity costs. In other words, this scenario forces the rest oration of a large area of land that once 135

PAGE 136

supported forests with high carbon density and t hat has low potential profitability for soy and cattle ranching. Eligible carbon When compensation for carbon maintenance (reduced emissions) is restricted only to that portion of forest that exceeds 50 % of the total possible native forest in each microbasin, a 3-fold reduction in per unit prices for CO2e is observed (Table 4-3). Prices for compensating landholders for their excess carbon maintenance under the RLR scenario range from 11 to 12 USD tCO2e-1, compared with 15-16 USD tCO2e-1 and 1415 USD tCO2e-1 under the CFC and ZSEE scenarios, respectively. Ecological co-benefits Hydrology and regional climate. Different changes in forest cover caused by each of the three policy scenarios resulted in corresponding changes in key ecological parameters. The CFC and ZSEE scenarios induced twice as much forest cover as the average BAU scenario and therefore had far mo re evapotranspiration (approximately 300 mm yr-1, Table 4-4). With more evapotranspi ration, the two high-forest-cover scenarios allowed less river discharge than any of the other scenarios that I modeled (Table 4-4). The RLR scenario, which induce d forest cover inte rmediate to the BAUhigh and BAU-average scenarios, had correspondi ngly little effect on evapotranspiration and discharge. Water quality. All three of the alter native policy scenarios forced full protection or restoration of riparian forest s, thus riparian forest cove r increased by approximately 2600 km2 (20%) over the 2005 landscape (Table 44). Thus, in all cases, 20% of streams would be likely to have lower temper atures and higher dissolved oxygen levels (Neill et al., 2006) than the current land scape, affecting species populations and 136

PAGE 137

assemblages. In contrast, riparian forest cover decreased by an average of 3200 km2 (ranging from 2500 km2 to 6400 km2) under the BAU scenarios. Thus, more streams would be likely to have higher temperatures and lower dissolved oxygen levels than the current landscape. More importantly, an average across BAU scenarios of nearly 40% of all streams would be subject to forest removal and potential oxygen depletion relative to any of the three alte rnative policy scenarios. Among the policy scenarios, the proportion of microbasins having less than 40% vegetation cover was lowest in the ZSEE scenar io (Table 4-4) because of differences in model predictions for cerrado microbasins. T he RLR and CFC scenarios were similar to the current landscape, whereas under the BAU scenarios, t he proportion of microbasins with this level of native vegetation loss from 41 to 72%. Only the CFC and ZSEE scenarios had a greater perc entage of microbasins (10 to 16%) with more than 60% forest cover than the 2005 landscape. The RL R was equivalent to the BAU scenario, with 37% fewer microbasins having over 60% v egetation cover. These results indicate that any of the policy scenarios would be likely to maintain basic hydrological functions in more microbasins (Coe et al., 2009) in the region than the BAU scenarios, although the RLR scenario would be the least favorabl e option of the three policy scenarios. Habitat. Overall, habitat quality and quantity diffe red between scenarios by biome, but was highest in the CFC and ZSEE scenari os. Perhaps most surprising, the RLR scenario had lower habitat quality than the BAU Low scenario Although the BAU scenarios maintained nearly as much or more forest cover than the RLR scenario (Table 4-4), cerrado cover was generally much lower. This can be explained by the models representation of t he most likely pattern and trajectory of forest and cerrado 137

PAGE 138

deforestation, which is more likely close to roads and previously cleared areas, if no controls on land-use that regulate the spatial distribution of native vegetation are put in place. Forest fragmentation was consistently 2 to 3 times higher in the BAU scenarios than in the alternative policy scenarios; cerrado fragmentation was 1.4 to 1.7 times greater under BAU scenarios than in the al ternative policy scenarios. Mean fragment sizes were 4to 9-fold lower in the fo rest biome under BAU sc enarios than under CFC or ZSEE scenarios. Differences were fa r less pronounced between fragment sizes under the RLR and BAU scenarios. A strikingly different pattern emerges in comparing edge and interior habitat area among the scenarios. The RLR scenario has the highest amount of forest edge habitat and even less forest interior habitat than the BAU high scenario. All of the alternative policy scenarios have more total edge habit at than the BAU scena rios in both the cerrado and forest biomes. Additionally, the BAU Low scenario maintains more total interior habitat in both the forest and cerrado biomes than the RLR scenario. Discussion On balance, the results of this analysi s indicate that even in an active agroindustrial frontier, with very high oppor tunity costs associated with forest conservation (Chapter 2, Chapter 3), compensation of opportunity costs can be achieved through carbon prices that are at the low end of estimate s of future carbon markets. Larsen and Heilmayr (2009) esti mate a substantial amount (300 MtCO2e) of greenhouse gas emissions reductions in the Un ited States could be achieved for less than 50 USD tCO2e-1 by 2030. Currently, CO2e is trading for approximately 20 USD tCO2e-1 under the European Unions Emissions Trading Scheme (ETS; Point Carbon, 138

PAGE 139

2009). In the CFC and ZSEE scenarios, I estimate the price of CO2e at approximately 20 to 30 USD per ton. Overall, the ZSEE scenario performed best in balancing ecosystem service protection with the costs of providing incentiv es to private landholders in the region to comply with the zoning plan, although its perfo rmance in terms of price was surprisingly similar to that of the CFC scenario. Emi ssions reductions are maximized under the ZSEE scenario, although overall carbon stocks are greatest under the CFC scenario by approximately 100 MtCO2e due to forest restoration. The per unit price of CO2e to achieve the desired emissions reductions is lower under the CFC and ZSEE scenarios than the RLR scenario, although the total cost to achieve the reductions is more than 2fold greater under the former scenarios. No tably, the RLR scenario provides only modest gains against BAU scenarios in terms of forest carbon stocks compared to the CFC and ZSEE scenarios. Although this scenario permits the greatest amount of agricultural development and the least opportunity cost to landholders (Chapter 2), its relatively poor performance (in compar ison with the CFC and ZSEE scenarios) with respect to carbon stocks and other ecosystem services strengthens the argument for some type of compensation or incentive to encourage increased forest protection in the region. Enhancing carbon stocks through active rest oration doubles the total cost of complying with the RLR and CFC scenarios, but only increases the total cost of complying with the ZSEE scenario by 50%. In general, compensating restoration costs is expensive when the amount of accumulated carbon is relatively low given both the foregone potential income and the direct cost of restoration. In contrast, when a large 139

PAGE 140

volume of carbon is accumulated, as under compliance with the CFC scenario, the per unit price of CO2e falls markedly. This large volume of carbon in restored forests implies a much larger absolute cash outlay will be necessary to make the CFC scenario feasible. Nevertheless, even compensation for restoration at the same level as for the opportunity cost alone could potentially be important in enc ouraging riparian zone restoration, which is critic al for overall stream health. Critics of the Amazon Fundand REDD more broadlycite concerns that the proposal intends to reward landholders for comp lying with legislation to protect forests, arguing that individuals or companies shou ld not be paid to obey the law (FOEI 2008). However, land-use policies which set out ambitious goals for forest and ecosystem service protection in the Amazon impose high costs on landholders and therefore achieve only low compliance levels (Chapt er 2). Such policies could become more feasible under a compensation mechanism. Ho wever, it may be important to restrict eligibility for compensation so as not to inadvertently punish la ndholders that have maintained a minimum forest reserve of 50% by rewarding those who blatantly disregarded regulations (Chapt er 2). When the subset of carbon for which landholders may be compensated is restricted to that portion that exceeds the stocks contained in legally required forest reserves, the ZSEE sc enario is demonstrated to be the most efficient although the per unit price is not the lowest. This is because it retains 24 times the carbon stocks on the land than the lowe st priced policy alternative, the RLR scenario, at only 1.3 times the per unit pric e. This pattern holds true at all three examined reference levels. The CFC scenario also retains high carbon stocks (although not as high as the ZSEE scenario ), but the per unit price to achieve them is even higher 140

PAGE 141

than for the ZSEE. Overall, compensating landholders in good standing for only part of their forest reserve reduces the cost of reducing emissions by nearly 4 USD tCO2e-1 when compared to a plan that compensates all landholders in the regi on for protecting all remnant native vegetation. This price w ould put compensation well in the range of current carbon prices (Point Carbon, 2009). REDD is also criticized for its narrow focus on carbon and the concern that noncarbon ecosystem services (e.g., the prov ision and regulation of pure freshwater, biodiversity conservation, and the maintenan ce of soil resources) and social issues (e.g., poverty reduction and the protection of land and human rights) will be neglected or affected detrimentally (Daviet et al., 2007; Br own et al., 2008, Dooley et al., 2008). These are important concerns, especially given the poor perfo rmance of previous global initiatives to protect tropical forests (Winte rbottom, 1990). The potential ecological costs and co-benefits of REDD will depend upon the types of land management interventions that this regime will eventually allow, whic h is the focus of considerable debate within the United Nations Framework Conv ention on Climate C hange (UNFCCC). An integrated zoning plan that takes into a ccount agricultural suitability as well as environmental sensitivity may be one of the best ways to carry out REDD-related policies. As demonstrated in the Xingu River headwaters case study, the ZSEE scenario performs best with respect to a num ber of ecological i ndicators, including surface hydrology, several measures of water quality, and several measures of terrestrial habitat quality (e.g., increased fo rest connectivity and reduced edge effects). Furthermore, it optimizes agricultural production (Chapter 3) and carbon stocks relative to other existing policy alternatives. The case study demonstrates that the ecological co141

PAGE 142

benefits of REDD are sensitive not only to the quantity of fo rests and woodlands remaining on the landscape, but also to their spatial distribution. Even small flows of carbon revenue properly targetedfor ex ample, toward the conservation and restoration of riparian zone forestscould confer enormous ecological benefits for aquatic ecosystems. The results suggest that overall watershed function would be best protected under a more even distribution of forests and that REDD co-benefits could be maximized in the context of an integrated re gional plan. In addition, a zoning plan, if extended over an entire region or nation (in this case, first to Ma to Grosso, then the Brazilian Amazon), could reduce potential leakage effects, in which forest protection in one location leads to increased cleari ng or degradation in other location. In this analysis, I have only explored the possibility of providing incentives and compensation to private landho lders, largely because in t he study region altering the behavior of landholders with respect to forest protection will be critical to an overall deforestation reduction strategy and thus is one of the major components of the Mato Grosso draft REDD plan (GOMT, 2009). In Br azil, the agro-industrial lobby is very powerful due to its large and growing shar e in generating nationa l income. Currently, representatives of the sector are successfully lobbying to significantly weaken or overturn the bulk of Brazils environmental legislation, arguably in response to the environmental lobbys largely uncompromisi ng stance in recent years (Folha de So Paulo, 2009). Thus, including agricultural producers and ranchers in a scheme to provide incentives for good forest stewardshi p will be crucial for achieving deforestation reduction targets. However, for REDD or any other carbon-based compensation mechanism to be successful in Brazil and el sewhere, an integrated approach that 142

PAGE 143

addresses a broad range of stakeholders and re source use types will be necessary. In the case study presented here, the ZSEE scenario addresses indigenous territories, smallholder settlements, and state and federal protected areas, as well as private landholdings. Furthermore, t he extended plan includes a la rge number of non-spatial considerations related to optimizing ecological integrity and agricultural production for the state of Mato Grosso. Conclusion The regulatory framework created by the frameworks that I analyzed could greatly facilitate the implementati on of markets for ecosystem services. Carbon storage and sequestration by forests is the ecosystem service with the most advanced market, and may soon expand precipitously for tropical forests through the UNFCCC (Gullison et al., 2007). It is very likely that REDD funding will flow to nations and regions that have mechanisms in place for rewarding comm unities and private landholders who are keeping their forests standing, and for identifying those rural land users who are out of compliance with land use legislation. The ZSEE under negotiation in Mato Grosso, and the property-level land registry developed in this state (Fearnside, 2003; Azevedo, 2009), provide an excellent framework for de veloping and implementing deforestation reduction targets that are tied to the emerging REDD carbon market. Furthermore, if well executed, the potential social and ecological benefits of REDD are numerous. The protection of water res ources, local and regional climate, soil resources, and biodiversity could contribut e to the social benefits derived from REDD since they are ecosystem services on whic h local and regional populations depend. Because of its focus on carbon emissions reduction needed to stabilize the global climate system, REDD has access to a pool of non-local stakeholders who are 143

PAGE 144

144 interested in paying to maintain carbon in forests and thereby potentially provide a cascade of ecosystem services to local st akeholders who would ot herwise be unable to pay for the benefits those services provide.

PAGE 145

Table 4-1. Forest carbon stocks on private lands of t he Xingu River headwaters region in 2005 and simulated for 2035 under six scenarios: Business-as-usual (BAU) with low, high, and average rate s of deforestation; reduced legal reserve (RLR, legal reserve of 50% in the forest biome and 35% in the ce rrado biome), current Forest Code (legal reserve of 80% in the forest biome and 35% in the Cerrado biome), and the so cio-economic, ecological zoning plan of Mato Grosso state (ZSEE). Scenario (2035) 2005 BAU Low BAU High BAU RLR CFC ZSEE Non-riparian zone Stored in native vegetation (MtCO2e) 1161 770 215 501 684 993 1041 Stored in restored vegetation (30 years) (MtCO2e) 15 n/a n/a n/a 38 241 47 Total (MtCO2e) 1176 770 215 501 722 1234 1088 Riparian zone Stored in native vegetation (MtCO2e) 120 90 39 80 120 120 120 Stored in restored vegetation (30 years) (MtCO2e) 2 n/a n/a n/a 7 7 7 Total (MtCO2e) 122 90 39 80 127 127 127 Grand total Stored in native vegetation (MtCO2e) 1281 860 254 581 804 1113 1241 Stored in restored vegetation (30 years) (MtCO2e) 17 n/a n/a n/a 45 248 54 Total (MtCO2e) 1298 860 254 581 849 1361 1295 145

PAGE 146

146 Table 4-2. The effect of policy scenarios on the costs, carbon stocks, and costs per ton of carbon associated with additional forest cover in the privat ely-owned land of the Xingu River headwat er region. Calculations are made relative to each of three business-as-usual (BAU) scenar ios of future forest cover. Costs include the total economic costs (opportunity costs of for egone profits from agriculture or ranching plus forest restoration costs). Policy scenarios (RLR, CFC, ZSEE ) are described in Table 4-1. BAU (average) BAU (high rate) BAU (low rate) Reduced Emissions Enhancement Total Reduced Emissions Enhancement Total Reduced Emissions Enhancement Total Cost (million USD) RLR 4872 4316 (99) 9188 (99) 8898 4316 (99) 13,214 (99) 1477 4316 (99) 5793 (99) CFC 9562 8709 (05) 18,271 (05) 13,818 8709 (05) 22,527 (05) 4938 8709 (05) 13,647 (05) ZSEE 9940 4917 (83) 14,857 (83) 14,256 4917 (83) 19,173 (83) 5274 4917 (83) 10,191 (83) Carbon stocks (MtCO2e) (relative to BAU stocks) RLR 247 38 285 486 38 524 70 38 108 CFC 521 241 762 789 241 1030 276 241 517 ZSEE 559 54 613 834 54 888 304 54 358 Price of carbon (USD tCO2e-1) RLR 20.1 102.3 (1.9) 32.2 (.2) 18.3 102.3 (1.9) 25.2 (.7) 21.2 102.3 (1.9) 53.6 (.4) CFC 18.3 36.1 (.7) 24.0 (.2) 17.5 36.1 (.7) 21.9 (.9) 17.9 36.1 (.7) 26.4 (.7) ZSEE 17.8 103.9 (2.3) 24.2 (.0) 17.1 103.9 (2.3) 21.6 (.8) 17.4 103.9 (2.3) 28.5 (.6)

PAGE 147

147 Table 4-3. The effect of policy scenarios on the basin-wide costs, carbon stocks, and costs per ton of carbon associated with additional forest cover in excess of 50% cover of privately held lands in the forest biome of the Xingu River headwater region, represented by individual microbasins. Calculations are made relative to each of three business-as-usual (BAU) scenarios of future forest cover. Policy scenarios (RLR, CF C, ZSEE) are described in Table 4-1. Costs include the total economic costs (opportunity costs of foregone profits from agriculture or ranching plus forest restoration costs). BAU (average) BAU (hi gh rate) BAU (low rate) Reduced Emissions Reduced Emissions Reduced Emissions Cost (million USD) RLR 84 92 59 CFC 2456 2556 2103 ZSEE 2758 2873 2396 Carbon (MtCO2e) RLR 8 8 5 CFC 162 171 134 ZSEE 192 204 162 Price of carbon (USD tCO2e-1) RLR 11.2 11.0 12.2 CFC 15.2 14.9 15.7 ZSEE 14.4 14.1 14.8

PAGE 148

148 Table 4-4. Comparison of ecological f eatures of the Xingu River headwaters regions with 2005 forest cover, under three simulations of future forest cove r under business-as-usual assumptions, and under three simulated policy scenarios. The policy scenarios include RL R, CFC, and ZSEE, as described in Table 4-1. Ecological features include parameters for surface hydrology and regional climate, indicators related to water quality, and terrestrial habitat quantity and quality. Scenarios 2005 BAU Low BAU High BAU RLR CFC ZSEE Surface Hydrology and Regional Climate Mean Annual Discharge (m3 s-1) (% change from potential) 3113 (13%) 3314 (21%) 3579 (31%) 3410 (25%) 3235 (17%) 3055 (11%) 3018 (10%) Mean Annual Evapotranspiration (m3 s-1) (% change from potential) 6570 (-5%) 6368 (-8%) 6103 (-12%) 6272 (-10%) 6477 (-7%) 6665 (-4%) 6628 (-4%) Water Quality Riparian forest cover (km2) 12,931 10,431 6527 9697 15,509 15,509 15,509 Mean % vegetation cover per microbasin 69 (0.5) 53 30 43 57 (0.4) 73 (0.4) 73 (0.4) % of microbasins with greater than 60% vegetation cover 65 34 23 28 28 81 75 % of microbasins with less than 40% vegetation cover 18 41 72 61 18 16 11 Terrestrial Habitat Vegetation cover (km2) Forest 107,789 81,483 45,246 64,602 51,561 116,395 110,902 Cerrado 17,037 8567 3988 6305 9083 13,149 15,209 Number of fragments Forest 13,427 40,568 49,780 46,510 23,673 12,958 13,770 Cerrado 13,285 24,881 25,467 26,702 18,422 18,138 15,573 Mean distance to nearest neighbor fragment (m) Forest 361 403 564 465 376 379 363 Cerrado 406 427 490 442 354 371 378 Mean fragment size (ha) Forest 803 201 91 139 218 898 805 Cerrado 128 34 16 24 50 73 98 Total interior habitat area (km2) Forest 99,978 72,744 40,498 56,977 36,975 105,287 98,973 Cerrado 12,737 5758 2030 3569 4001 7433 9753 Total edge habitat area (km2) Forest 7810 8739 4748 7624 14,586 11,108 11,930 Cerrado 4300 2809 1958 2736 5082 5716 5456

PAGE 149

Figure 4-1. Maps showing simulated futu re (2035) land cove r of the Xingu River headwaters region under six different scenarios. These scenarios are described in Table 4-1 and include: (a) Business as Usual (BAU), average rate; (b) BAU, low rate; (c) BAU, high rate; (d) RLR (reduced legal reserve); (e) CFC (current Forest Code); and (f) ZSEE (socio-economic, ecological zoning plan). Indigenous territories and protected areas are indicated with white hatching. 149

PAGE 150

CHAPTER 5 CONCLUSION By suddenly increasing the legal forest rese rve of the Brazilian Forest Code from 50 to 80% of private landholdings in the Amazon region, in 1996, the Brazilian Government imposed substantial costs on fa rmers and cattle ranchers, reaching nine billion dollars in the Xingu River headwaters region alone (three to four million dollars, on average, for each of the regions landholders). By mainta ining the new measure in a state of perpetual modification and revision fo r the next five years, the government may have further undermined the perception, by landho lders, of the new rules as binding and permanent. Finally, by failing to provide progr ams that facilitate landholder compliance with the new regulation, the potential role of the law to protect private land forests may have been further diminished. Compliance with the new Forest Code fell to only 46% of landholders in 2005 (Chapter 2). The potential ecological benefits of a fully implemented 80% legal reserve requirement are substantia l. Compared to the fully implemented 50% requirement, it would provide an additional 120 million tons of carbon (450 tons of CO2 equivalent (tCO2e)) stored in forests, higher evapotr anspiration, lower stream and river discharge (with reduced risks of flooding), and near ly twice the amount of interior forest habitat (Table 2-9). Problems with the modified Forest Code we re foreseen. The 1988 Constitution established a new land use policy instrument known as the socio-economic, ecological zoning plan (ZSEE). These state-level plans, upon approval by the state legislative assembly, permit the definition of specific geographical zones in which the legal forest reserve requirement of the Forest Code c an be reduced to 50%. A second instrument for addressing the economic costs of implementing the 80% modification of the Forest 150

PAGE 151

Code is the provision for legal reserve com pensation. In the case of the Xingu River headwaters region, the Mato Grosso state ZSEE has not been finalized. If it were implemented, it could greatly in crease the feasibility of movi ng agricultural regions into compliance with the law while protecting eco system services. The economic opportunity cost of compliance with land-use regulations incurred by landho lders in the region as a whole would decline by USD 2. 5 billion; for the average landholder, the cost of compliance would decline from 4.2 m illion USD to 1.6 million USD through implementation of the ZSEE. In contrast, the tradable development rights provision of the modified Forest Code would have virtually no effect on the cost of legal compliance in the Xingu River headwater s region, since very few of the regions microbasins (the proxy for landholdings in this study) have mo re native vegetation than is mandated by the law. This sharp reduction in opportunity costs achieved through the ZSEE does not imply deep sacrifices in the provision of ecological services. Evapotranspiration and stream discharge remain virtually the same in ZSEE compared with full compliance with 80% legal reserve, although forest carbon stocks and forest interior habitat decline. Furthermore, the ZSEE protects more native cerrado vegetation than the post-1996 Forest Code because the reduction in restor ation requirements in the forest biome reduce leakage of clearing for agricultural lands into the cerrado biome. The major policy alternatives under discussion for the Xingu River headwaters region, including a reduced legal reserve (RLR returning to 50% from the current 80% requirement), the current legal code (C FC, 80%), and the ZSEE, would result in different amounts and patterns of coverage of remaining and restored forests and cerrado with important implications for the amount of carbon contained on the 151

PAGE 152

landscape, the costs to produc ers, and the associated ecosystem services. The gain in forest carbon would be much smaller under the RLR scenario (285 to 584 MtCO2e), than under the CFC (762 to 1030) and ZSEE (613 to 888 MtCO2e) relative to the range of BAU scenarios. The associ ated cost per ton of CO2e would be somewhat higher for the RLR scenario (18-21 USD tCO2e-1), than the CFC (18 USD tCO2e-1) and ZSEE (1718 USD tCO2e-1) scenarios. The current price of CO2 in the carbon market is approximately USD 21 per t on (Point Carbon, 2009), sugges ting that the potential for carbon offsets to provide one source of financia l incentives for landholders in the region to comply with regulations is high. There is concern, however, that emergi ng carbon markets, such as REDD, could generate negative ecological consequences as it fosters greater carbon storage on the land. REDD could displace agricultural exp ansion from a region of high carbon density to a region of low carbon density, potentially eradicating native ecosystems that are rare and contain species threatened with extirpation or extinction (Stickl er et al., 2009). In the case of the Xingu River headwaters region, under a zoning plan, incentives tied to carbon stock maintenance and enhancement c ould also foster improved watershed function, and increase habitat quality and quantity (e.g., through increased forest connectivity, reduced edge effects) while allo wing for reasonable areas of land to be made available for agricultural production. The Xingu headwaters case study demonstrates that the ecolog ical co-benefits of REDD are sensitive not only to the quantity of forests and woodlands remaining on the landscape, but also to their spatial distribution. 152

PAGE 153

This case study of the Xingu River headwaters region indicates that the prospects for protecting public interests in private forest s in regions such as the Brazilian Amazon are high. Although no policy framework is lik ely to present a perfect solution to balancing two public goodsagricultural pr oduction and the provision of ecosystem servicesthis assessment does indicate that a happy medium may be achieved by identifying areas likely to be more suitable for agriculture as well as those which are more environmentally sensitive. With 83% of t he forest remaining, t here is still time to correct the mistakes made in modifying and implementing the Forest Code. However, the moral idealism of the legal tradition in Brazil may make the notion of positive economic incentives for compliance with the law difficult to consider With a focus on the ideal versus the practical solution to issues of the public good, and its defense, Brazils civil law system typically produc es new regulations and laws without the dialogue and debate among interested stakeholders that could build into the design of the new law mechanisms for increasing it s practicality and chances of successful implementation. However, multi-stakeholde r, participatory planning processes are already part of recent watershed management l egislation. More and more, civil society is initiating such processes for a variety of environmental governance issues, including infrastructure development (Campos and Nepstad, 2006; Soares-Filho et al., 2004), watershed management (ISA, 2005), and other regi onal planning processes (Perz et al., 2008), signaling an important trend in Brazilian environment al rule-making in the future. This study presents the results of an in tegrated assessment that attempts to evaluate a more complete set of ecologic al and economic trade-offs associated with alternative policies for managing a large-scale agro-industrial landscape by projecting 153

PAGE 154

the likely land-cover outcomes of each of those policies into the future. This approach is important for policy makers and stakeholders in the region because it allows them to make informed decisions regarding landuse planning based on a quantitative analysis of the effects of each policy option on a series of indicators. For the first time, this research brings such a quantitative analysi s into a debate that has been historically ruled by ideological perspectives. Thus, the results presented her e are already being used to improve policy design in the region. 154

PAGE 155

APPENDIX A LAND-COVER MAPS I developed land-cover classification maps for the years 1996, 1999, 2005, and 2007 using image segmentation and object-oriented classification techniques of 30-m resolution Landsat TM image mosaics corres ponding to each year. I generated maps with 5 land-cover classes for each year: forest, cerrado, agriculture in the forest biome, agriculture in the cerrado biom e, and a miscellaneous category. Sensor Data Twelve Landsat-5 TM (L1G) scenes (bands 1-7) were acquired for each of 4 years (1996, 1999, 2005, 2007). All images were acqui red during the dry season to minimize cloud cover (Table A-1). Pre-processing of the L1G scenes included radiometric calibration, atmospheric correction, georeferencing, and mosaicing. The radiometric data were converted from radiance to ground reflectance (Jensen, 2000) to eliminate variability and noise due to differences in satellite instrumentation, Earth-Sun distance, solar elevation angle, solar curve, and atmospheric effects on different acquisition dates and times using the calibration technique of Green et al. (1999). G eometric rectification was performed using nearest neighbor resampli ng to co-register each scene to its orthorectified GeoCover analogu e obtained from the Global Landcover Facility (GLCF). Using a combination of m anual and automatic registration methods, all scenes were transformed to the UTM coordinate system and WGS 84 datum. A root mean square (RMS) error of <0.5 pixels was achieved for all scenes. Automatic mosaicing was conducted after manual exclusion of areas covered by clouds, shadows, and/or smoke, using histogram matching and seam-li ne generation. The 1996, 1999, and 2005 155

PAGE 156

mosaics were co-registered to the 2007 mosa ic to increase consistency between the images and subsequent maps. Reference Data A total of 989 reference points were used for classification mo del calibration and validation. Of these, 535 were ground re ference plots sampled across 15 landcover/land-use types during an eight-week, dr y-season field campaign between 26-June and 23-August 2006 (Walker et al., in prep.). At each location, a 900-m2 plot was established, and the land-cover/land-use class, land-use histor y, mean vegetation height, and geographic coordinates were recor ded. All plots were located along or within the Xingu basin watershed boundary, and for accessibility reasons, tended to be concentrated along primary and secondary tr ansportation routes outside of major protected areas in the region. Landcover classes for all years were assigned using landcover/land-use class in 2006, land-use hi story information, and cross-referencing between years. The remaining 454 points were collect ed via on-screen identif ication using the 2007 Landsat 5 mosaic as a base map and in conjunction with the ground reference dataset. Point locations were determined through the random supe rimposition of a uniform 35 km2 point grid on the Landsat image mosaic with the overall goal of obtaining supplemental reference data points in regi ons previously unsampled, most notably areas that were inaccessible for field sa mpling. Land-cover classes for the two other years were assigned by cross-referencing between years. 156

PAGE 157

Classification and Mapping Approach Segmentation and Attribute Extraction Segmentation refers to t he automated and optimal groupi ng of image pixels, i.e., the partitioning of a landscape, into spectrally homogenous and spatially distinct regions or image objects based on some predefined, knowledge-based criteria (Baatz et al., 2004; Navulur, 2007). Unlike individual image pixels, im age objects can be smartly characterized by hundreds of object-level a ttributes, including measures of shape, size, texture, and morphology, in addition to st andard spectral descriptions, and numerous investigators have demonstrated the superiority of object-based strategies to empirical modeling and mapping over traditional pixe l-based approaches (Baatz et al., 2004; Walker et al., 2007). The eCognition software package provi ded the computational framework for image segmentation and subsequent object-level attribute extraction. For the purposes of producing a second segmentation product suitable for classification and mapping, fourteen individual image data la yers were used as inputs to this segmentation including raw bands 3-7, principal components 1-3, Tasseled Cap brightness, greenness, and wetness, NDVI, the 3-arcsecond SRTM DEM, and the SRTM-derived slope. The segmentation process is guided primarily by three crit eria: 1) scale, 2) shape, and 3) smoothness/compactness. Whereas the scale parameter controls the average size of the image objects to be generated, the latter two criteria determine the homogeneity of image objects. Hierarchica l (top-down), bi-level segmentation procedures were parameterized using scale parameters of 40 and 20, shape factors of 0.1, and compactness/smoot hness factors of 0.5. Parameters were determined empirically using an iterative post-segm entation inspection/refinement approach. 157

PAGE 158

Further information on eCogniti on parameterization can be found in Baatz et al. (2004). The final segmentation gener ated an average of 474,265 image objects with an average object size of 0.7 km2 (range: 0.001-36.0 km2) for each image. Fifty putative predictor variables were computed for each of the segments. Of the 50 variables, 26 spectral, 13 textural, 7 ancillary-spatial, and 4 ancillar y-topographic variables were included. Spatial Database Joining To compile the tabular databases requir ed for use in the development of the classification models, spatial joins were established between the reference point dataset and each of the segmentation products (i.e., for each year). The joining procedure resulted in a database file of 989 rows by 53 columns (52 predictor and 1 response variables). Classification and Mapping Empirical-statistical machine learning techniques such as classification and regression trees (CART) have become increas ingly popular within the remote sensing research and applications community, particu larly when the objectives involve broadscale mapping (e.g., Baccini et al., 2009; Bla ckard et al., 2008; Goetz et al., 2005; Hansen et al., 2008; Walker et al., 2007). T hese approaches tend to be user-friendly, computationally less demanding, and can achieve high predictive accuracies when well calibrated for the region of interest. For the purposes of this classification effort, the randomForest (RF) algorithm, implemented in the open sour ce R statistical programming environment, was used. Firs t proposed by Breiman (2001), the RF algorithm falls into the category of ensembl e learning methods where the goal is to construct a forest (i.e., ensem ble) of individual classifier s that are later combined to improve predictive accuracy. In the case of RF, independent trees are constructed using 158

PAGE 159

a bootstrap sample of the dat a set in a process called bagging a term derived from b ootstrap agg regation (Bauer and Kohav i, 1999; Dietterich, 2000; Liaw and Wiener, 2002). In each bootstrap sample, approximately 1/3rd of the reference cases are left out. These out-of-bag samples are predicted during each bootstrap iteration (i.e., generation of each tree), and later aggregated to produce an out-of-bag (OOB) estimate of error. Direct comparisons between OOB e rror rates and error rates computed under independent validation scenar ios have found the OOB estimates to be quite robust (Liaw and Wiener, 2002; Walker et al., 2007). Three separate randomForest classification s (corresponding to each year in the study) with 6 classes (agriculture, forest, cerrado, open water, wetlands, sandbars in areas of open water) were generated and eval uated based on overall classification accuracy and Cohens Kappa statistic. Subsequently, some classes were merged (water, wetlands, and sandbars to a miscellaneous class) and post-classification sorting (Janssen et al., 1990; Vogelmann et al., 1998) split some of the automatically generated classes to achieve a final classificati on with 7 classes for each year (forest, cerrado, regenerating forest, regenerating cerrado agriculture in the forest biome, agriculture in the cerrado biome, other). Specifically, cleared areas were recoded by biome (forest and cerrado ) to differentiate between dynami cs of clearing in the forest and cerrado biomes. The cerrado -forest biome map was obtained by merging a map of forest/nonforest derived from INPE Prodes maps (INPE, 2009) with a map of biomes from the IBGE RADAM vegetation thematic map (IBGE, 1981). Regeneration and/or abandonment were also recoded as necessary by cross-referencing between years. 159

PAGE 160

Table A-1. Image dates for Landsat-bas ed mosaics for 4 years (1996, 1999, 2005, 2007). YEAR (MM/DD) PATH/ROW (WRS-2) 1996 1999 2005 2007 P224/R67 07/03 10/08 07/28 06/16 P224/R68 07/03 08/21 07/12 06/16 P224/R69 06/17 08/21 08/13 08/19 P224/R70 07/03 08/21 06/10 08/19 P225/R67 07/10 08/12 07/03 07/09 P225/R68 07/10 08/12 08/04 07/09 P225/R69 07/10 08/12 08/20 07/09 P225/R70 06/24 08/12 04/14 09/09 P226/R67 07/01 08/19 07/10 08/17 P226/R68 07/01 07/02 07/26 08/17 P226/R69 06/15 07/02 05/07 08/17 P226/R70 07/17 08/03 05/07 08/17 160

PAGE 161

APPENDIX B DYNAMIC SPATIAL SIMULATION MODEL DEVELOPMENT Overview I developed a dynamic landscape simulati on model to model future landscape trajectories corresponding to a set of alte rnative policy proposals. The model is based on a spatial-statistical model of land-use change, derived from a land-use/land-cover change analysis and a GIS consisting of data related to the location and neighborhood context of 4 focal land-cover transitions: (1) forest agriculture (pasture or annual crops); (2) cerrado agriculture; (3) agriculture regenerating forest; and (4) agriculture regenerating cerrado. The model simulates land-cover change over 14 time steps, beginning in 2007 and ending in 2020, using land-cover conversion rates calculated from the 1996-2005 reference period. The model integrates coupled components developed within two spatial st ructures: (1) subregions defined by hydrographic sub-basins, and (2) raster cells (4510x5963) at 1-ha resolution. The model uses a nested, sub-regiona l approach based on hydrographic basins, to better reflect the ecological and legal realit y of land-use policy in the region, as well as to facilitate model processing. The model is calibrated and run separately for 6 3rdorder sub-basins (explained in more detail be low), the results of which are merged at every time step. This step decreases processing memory requirements and better simulates actual land-cover change processes by regionalizing rates, relationships to proximal drivers, and patterns. Each of the 6 sub-basins is further subdivided into micro-watersheds representing individual stream reaches (1:1 ,000,000 scale), which interact such that the proximity of a deforestation front in one micro-basin infl uences deforestation in a neighboring micro161

PAGE 162

basin. Within a sub-basin, all microbasins are subject to the total annual land-cover change rate for the whole basin. For the baseline (Business as Usual) scenario simulation, the sub-division into microbasins does not affect the location of the focal land-cover change events. However, this be comes important in modeling alternative land-use policy scenarios in which regulations are established for the property level. As a complete map of property boundaries for the region is not yet available, I employ the micro-basins as proxies for individual proper ties to better simulate policy outcomes on private lands. The mean size and range of sizes of the 2881 micro-basins (x = 5981 ha, 4-70,766 ha) is comparable to that of priv ate properties in the region (Jepson, 2006; Fearnside, 2005). Furthermore, Brazilian wate r law requires management plans at the watershed level and the Brazilian Forest Code st ipulates that deforestation rights may be traded by property owners wit hin watersheds in certain cases (MP 2166-67, 2001; Chomitz, 2004; Stickler, 2009). The model has 4 basic steps: First, annual deforestation rates are calculated for each 3rd-order watershed based on conversion rate s calculated for the period 1996 to 2005. Next, annual deforestation probability in rela tion to a set of spatial variables is obtained using weights of evidence analysis (S oares-Filho et al., 2004) for each of the 6 sub-basins. Third, for each sub-basin, I developed a unique spatial simulation model. Finally, I validated the model by comparing the simulated and observed landscapes for the year 2007. I developed all modeling ph ases using the Dinamica EGO graphical interface platform ( http://www.csr.ufmg.br/dinamica/ ) that has the capacity to process multiple large map sets and has special features for advanced spatial modeling and simulation (Soares-Filho et al., 2009). 162

PAGE 163

Model Calibration First, annual deforestation rates are calculated for each 3rd-order watershed based on conversion rates calculated for the period 1996 to 2005. The amount of change in each transition of interest is comp uted from a Markov matrix obtained through the comparison of land use/co ver maps for the two dates. Second, annual deforestation probability in rela tion to a set of spatial variables is obtained using weights of evidence analysis fo r each of the 6 sub-basins. I used the Weights of Evidence method (Soares-Filho et al., 2004) to select the variables most related to observed landscape changes as well as to quantify their influences on each of the modeled transitions. Weights of Evidenc e is a Bayesian method traditionally used to derive favorability maps for spatial point phenomena (Agterberg and Bonham-Carter, 1990; Bonham-Carter, 1994). In this study, weights of evidence ( Wk +) are calculated for every k category of each spatial variable under analysis and can be interpreted as the influence of that category on t he chances of a deforestation event occurring. Since this method only applies to categorical data, it is necessary to categorize continuous graytone variables, such as distance-decay m aps; this is done using a method adapted from Agterberg and Bonham-Carter (1990) in Dinamica EGO. The variables I examined comprised a se t of biophysical and socioeconomic (or proximate) factors that spatially determine the location of the changes. This set includes slope, elevation, soil type, protected areas (including indigen ous territories), suitability for annual crops and cattle ranching, INCRA small-holder settlements, distances to rivers, major and secondary roads, and urban centers, and distances to forested, regenerating and deforest ed areas. A basic assumption for the Weights of Evidence method is that the variables must be s patially independent. I tested the spatial 163

PAGE 164

independence of the af orementioned variables using t he Crammer coefficient (V) and found that all variable s, except the pair soil type and suitability, have values lower than an empirical threshold (V < 0.45), and th us are spatially independent (Almeida et al., 2003). I retained suitability as it encompassed more information than the soil variable. Landcover Simulation For each of the 6 sub-basins, I developed interacting spatial simulation models with customized parameters co nsisting of (1) a cellular automata type model that simulates the spatial patterns of (a) deforestation or clearing of cerrado, based on a probability map depicting t he integrated influence of proxim ate drivers on the location of clearing at each time step (Soares-Filho et al ., 2002), and (b) regenerat ion, according to a set of exogenous rules and assumptions; and (2) a road constructor model that projects the expansio n of secondary road network, and thereby incorporates the effect of road expansion on the evolvi ng spatial patterns of deforestation (Soares-Filho et al., 2004). Deforestation or clearing of cerrado occurs in accordance with the rate determined for each sub-basin by the Mark ovian transition matrix and the probability map determined by the set of spatial variab les. As microbasins become saturated with deforestation, the model chooses the pixel in a neighboring microbasin with the next highest probability of being deforested within a given sub-basin. Depending on the assumptions of a given scenario, certain ar eas (e.g., protected areas) and individual micro-basins within a sub-basin can be forced to become saturated despite having a high probability of being deforested, thus causing the model to search for highprobability pixels in neighbor ing micro-basins that may still be cleared. The model includes regeneration only when a scenario requires that regeneration take place. 164

PAGE 165

Regeneration occurs in one of two, not mu tually exclusive, ways: (1) within predetermined zones (e.g., ripar ian buffer zone, proposed prot ected areas), and/or (2) within individual micro-basins according to the requirements of th e assumptions of a given scenario (e.g., according to state or federal legislati on or some other proposal for private lands). In the latter case, regenerat ion preferentially occurs adjacent to the defined riparian zone, adjacent to remnant lar ge blocks of forest, and/or farther away from roads and urban c enters. At each time step, s ub-basin level maps depicting saturation, time of residence (for pixels that have been subjected to a transition), landcover, probability, and cumulative roads are merged and passed on to begin the next time step. Baseline Scenario Development. The baseline (or Business as Usual, BAU) scenario assumes that the current rate of deforestation, level of compliance with environmental legislation, and accompanying land-use/land-cover change will continue, and thus serves as a baseline model against which to compare other regulatory options. The baseline scenario, referred to as the business-as-usual (BAU) scenario, assumes that historical trends will continue into the future, projecting regional rates using 19962005 figures. This scenario applies a Markov ian approach, simply projecting the changes into the future using transiti on rates annualized from the 1996-2005 timeperiod transition matrix using the general spatial allocation approach described above (for details see Soares-Filho et al., 2002). Model Validation Finally, I validated the model by comparing its predictions to the observed data for 2007 using a fuzzy map comparison which co mpares simulated deforestation to observed deforestation (Soares-Filho et al ., 2009; Almeida et al., 2008; Hagen, 2003). 165

PAGE 166

Two-way fuzzy comparison using a constant decay function provides a detailed assessment both of categorical and spatial si milarity of two maps that more closely mimics human visual comparison by overcoming restrictions induced by hard pixel limits like pattern quantification and exclusive cell state (Hagen-Zanker, 2006; Visser and Nijs, 2006). This method compares the number of cells of a certai n class in a simulated map with the number of thes e cells in a reference map that fall within a central cell neighborhood, as defined by a window size. By using a constant decay function, if a matching cell is found within the window, fit is assigned to 1, otherwise 0. Windows with increasing sizes convolute over the map and a mean is computed for each window size. This method employs a reciprocal appr oach, comparing the match between the observed map and the simulated map, and vice versa, ultimately choosing the minimum mean in order to penalize random maps, which tend to overestimate the fit. In this manner, this method accounts for both omission and commission errors. Our comparison employed increasing windows size s from 1 to 11 cells, which in terms of map resolution represent a range of 100x100-m to 1100x1100 -m. The overall agreement between the two maps ranged from 60 to 77%. 166

PAGE 167

APPENDIX C ECONOMIC AND ECOLOGICAL ASSESSMENT Economic Assessment I estimated the economic costs of complying with the policy change by estimating (1) the potential forgone profits to producers over the whole landscape and by microbasin using net present value as a pr oxy for the potential value of agricultural lands in the region, and (2) the cost of ri parian forest restorat ion over the whole landscape and by microbasin, where necessary. Net Present Value Net present value (NPV) for the region was estimated using spatially-explicit rent models for soy production (Vera Diaz et al., 2007; Nepstad et al., 2007a), cattle ranching (Nepstad et al., 2009), and sustain able timber harvest (Merry et al., 2009) the three major economic activities in the region. These models estimate the potential rent of each economic activity on analyses of the costs of production (several of which are spatially-dependent, such as transportation costs), yields, and prices. For each of the three economic activities, the NPV was estimated for 30 years into the future assuming a 5% annual discount rate and a plausible schedule of highway paving (Soares et al., 2006). Agricultural land values are typically appraised by determining the production value of the land, as determined by the NPV of the specific use to which that land is or will be put (Van Kooten and Bulte, 2000). In this case, the range and distribution of NPV in the region is similar to that of actual land values for which prices are only available at the municipal le vel (FNP, 2005). Although the models do not account for short-term fluctuations, they use a set of assumptions that provide conservative projections of profit for each activity. 167

PAGE 168

The layers derived from the rent models were combined such that, for any given pixel, the NPV of timber har vests (where this value was gr eater than 0) was subtracted from the highest value from either cattle or so y activities. Negative values resulting from this calculation were set equivalent to a net present value of zero. I used the resulting map of combined NPV of the Xingu headwaters to estimate total maximum potential NPV for both forested and cleared landsif they were to be cleared or remain cleared within the forest biome as well as per microbasin. Restoration Costs I estimated the costs to restore both legal reserve and riparian areas where required under the modeled scenarios. To estimate the cost of riparian zone restoration, I used figures generated in field trials ca rried out by non-governmental organizations active in the region: Instituto Socioambi ental, Aliana da Terra, and Instituto de Pesquisa Ambiental da Amaznia. I adapted t he riparian restoration costs to estimate legal reserve restoration costs, eliminat ing methods that involve out-planting and maintenance of seedlings as this would be ex tremely costly and labor-intensive over the relatively larger areas needi ng restoration. Restoration costs range from USD 536 ha-1 to USD 3217 ha-1, depending on the type of adjacent land-use and the intensity of treatments required to restore forest (Table C-1). Although the cost could be as low as USD 0 ha-1, regeneration would likely be slow and success would be highly dependent on the type and intensity of previous land-us e, as well as distance to nearest seed source. Ecological Assessment I compared the final landscapes for the observed and modeled landscapes in terms of carbon stocks, river discharge, annual evapotranspiration, terrestrial habitat 168

PAGE 169

quality, and water quality. Unlike other analyses in this study, which were restricted to private lands in the Xingu Basin, ecological consequences were assessed for the entire headwaters region, including all protected areas. Here, I describe how each indicator was assessed. Carbon Stocks Carbon stocks under each scenario were calculated using a map of aboveand belowground forest biomass developed and adapted for the entir e Amazon basin (Saatchi et al., 2007, adapted in Nepstad et al., 2009). I overlaid the map of biomass with the final outcome map for each scenario and assigned biomass values to each land-use/land-cover class in the simulated m ap. For intact native vegetation classes (intact forest, intact cerrado, native wetlands), the values in the aboveground biomass map were assigned directly. I c onverted biomass values to CO2-equivalent (CO2e) values for each pixel. For areas of agricultu re or pasture, I assigned a value equivalent to 15% of the original carbon stock, which represents the reduction in biomass following clearing since sufficient information to a ccurately identify pasture and soy expansion was not available (Houghton et al., 2000). For areas of regenerating forest and cerrado, I assigned a value of 1.5 tC ha-1y-1 and 0.5 tC ha-1y-1 (Houghton et al., 2000; Zarin et al., 2001) per pixel, respectively, and multiplied by the number of years that regeneration in each pixel had taken place (rang ing from 1 to 29 years). Surface Hydrology and Local and Regional Climate To investigate the impact of each scenario on the surfac e hydrology of the Xingu River in the absence of atmospheric feedbacks to precipitation, simulations with a land surface model (IBIS; Kucharik et al., 2000) and a river transport model (THMB; Coe et al., 2009) were carried out. I carri ed out offline simulations (as described in Stickler et 169

PAGE 170

al., 2009; Coe et al., 2009) fo r the landscape maps for all scenarios (3 historical, 2 modeled), as well as for a scenario describing potential (historical) land-cover in the region prior to settlement (referred as the Control (CTL) scenario). As the THMB model has a resolution of 5-min, whereas the l and-cover maps have a resolution of 100 m, new 5-min cell size land-cover maps to se rve as input to THMB were derived by calculating the fraction of each 5-min gr id cell size represented by disturbed and undisturbed vegetation for eac h scenario. The land-cover maps corresponding to each scenario were first recoded such that t hey contained only 2 classes, disturbed and undisturbed vegetation (corresponding to the two extreme scenarios used by IBIS to calculate mean monthly surfac e and sub-surface runoff fo r THMB, IBIS-POT (which assumes all intact vegetation) and IBIS-GRASS (which assumes all disturbed or cleared vegetation). In the 2 theoret ical Forest Code scenarios, regenerating forest and cerrado were reclassified as intact vegetation to refl ect the intention that these lands will remain protected and allowed to conti nue to regenerate. The differences from CTL in simulated discharge of the three obser ved landscapes and the two theoretical landscapes quantify the sensitivity of the surf ace hydrology (discharge and flooding) to land cover changes. For each scenario, I extracted simulated monthly discharge values for 32 years (1968-2000) for the main trunk of the Xingu River at the border between Mato Grosso and Par states. I calculated annual mean discharge (N = 32) for each scenario. I calculated mean annual evapotra nspiration by subtracting mean annual discharge from mean annual precipitation. M ean annual precipitation was derived for the same location from an interpolation of obs erved climate data over the same time period (1968-2000) over which the discharge simulations were carried out. For each scenario, the total 170

PAGE 171

volume and the percent change from the potential in annual discharge and annual evapotranspiration for each scenario is presented. Indicators of Water Quality To compare the effects of native vegetat ion distribution for water quality, I assessed a series of landscape-level meas ures that associated with broad physical and chemical changes in water. The primar y landscape measure associated with water quality is the presence of ripar ian zone vegetation. For each landscape, I calculated the amount of each land-cover type within the riparian zone by overlaying each landscape with the riparian buffer map described above. Nex t, I calculated the percent of forest or cerrado remaining in each of 2881 micro-basins comprising the headwaters region. Together with the presence of ri parian vegetation cover, this percent forest cover serves as an indicator of the propor tion of small streams that are likely to have higher temperatures and lower dissolved oxygen due to the lack of forest cover (Neill et al., 2006; Nepstad et al., 2007b). Terrestrial Habitat To evaluate differences in habitat quantity and quality am ong the scenarios, habitat fragmentation and the potential extent of edge effects were assessed for forest and cerrado cover associated with areas of grai n or cattle production by calculating a series of simple landscape metrics for eac h landscape. I assessed quantity (total class area for both cerrado and forest classes), degr ee of fragmentation (number of patches, mean patch size), habitat quality (total co re area, total edge ar ea, edge-to-core-area ratio), and connectivity (patch nearest neighbor distance). All analyses were carried out using Fragstats 3.3 spatial pattern analysi s software (McGarigal et al., 2002). The proportion of edge vs. interior habitat fo r forest was calculated using edge influence 171

PAGE 172

values derived from empirical observations by researchers studying edge effects related to the effects of fire on forests in t he region (Balch, 2008). I applied an edge depth value of 150 m to forest patches adjacent to agric ultural areas. This was rounded to 100 m (1 pixel) due to the resolution of the land-co ver maps. For forest patches adjacent to cerrado or regenerating forest or cerrado patches, edge infl uence was considered to be negligible relative to map resolution. Si milarly, edge influence depth in cerrado was considered to be negligible as cerrado is a more open land-cover type that is welladapted to regular fire disturbance. 172

PAGE 173

Table C-1. Estimated costs for restoration of native veget ation in legal reserves and riparian zones in the Xing u River headwaters region. Description of Restoration Method Legal Reserve (USD ha-1) Riparian Zone (USD ha-1) Natural regeneration, adjacent to forest or cropland 0 0 Natural regeneration, adjacent to active pasture (fencing, fire breaks only) 536 536 Natural regeneration with enrichment, adjacent to active pasture (fencing, fire breaks, enrichment with native seeds) 644 644 Mechanized restoration, adjac ent to forest or cropland (grading, leveling, plan ting native and some exotic seeds) 791 791 Mechanized restoration, adjacent to active pasture (fencing, fire breaks, gr ading, leveling, planting native and some exotic seeds) 1327 1327 Restoration with seedlings (fencing, fire breaks, seedlings, planting labor, maintenance labor) 2nd year enrichment (seedlings, native seeds, planting labor) NA 3217 Exchange rate: 0.53 BRL/USD (August 28, 2009; Source: Instituto Socioambiental & Aliana da Terra) 173

PAGE 174

LIST OF REFERENCES Abell, R., Allan, J.D., 2002. Riparian shade and stream temperatur es in an agricultural catchment, Michigan, USA. Verhandl ungen Internationale Vereinigung fur Theoretische und Angewandt e Limnologie 28, 232. Achard, F., Defries, R., 2007. Pan-tropical monitoring of deforestation. Environmental Research Letters 2, 045022. doi: 10.1088/1748-9326/2/4/045022 Acevedo, M.A., Villanueva-Rivera, L.J., 2006. Using automated digital recording systems as effective tools for the moni toring of birds and amphibians. Wildlife Society Bulletin 34, 211-214. Adams, J.B., Sabol, D.E., K apos, V., (1995) Classification of multispectral images based on fractions of endmembers application to land-cover change in the Brazilian Amazon 52(2), 137-154. Ad-Hoc Working Group on Long Term Cooperative Action, 5th Session (AWGLTCA5), 2009. Reducing Emissions from Deforestat ion and Forest Degradation and the role of Conservation, Sustainable Management of Forests, and the Enhancement of Forest Carbon Stocks. Paper No. 1. B onn, April 2009, UNFCCC. Accessible at http://unfccc.int/resource/ docs/2009/awglca5/eng/misc01a04.pdf (accessed on November 2, 2009) Agencia Nacional de guas (ANA), 2009.. Ottobacias. Accessible at http://www.ana.gov.br/biblio tecavirtual/ottobacias.asp (Accesed on November 2, 2009) Agterberg, F.P., Bonham-Carter, G.F., 1990. Deriving weights of evidence from geoscience contour maps for the prediction of discrete events. Proceedings of the 22nd APCOM Symposium, Technical University of Berlin, Berlin, Germany, 381-395. Akcakaya, H.R., 2000. Viability analyses with habitat-based metapopulation models. Population Ecology 42, 45-53. Alencar, A., Nepstad D., McGrath, D., Mouti nho, D., Pacheco, P., Vera Diaz, M.D.C., and Soares Filho, B., 2004. Desmat amento na Amaznia: Indo Alm da Emergncia Crnica. Instituto de Pesquisa Ambiental da Amazn ia (IPAM), Belem, Par, Brasil. Alencar, A.A.C., Nepstad, D.C., Vera-Diaz, M.D.C., 2006. Forest U nderstory Fire in the Brazilian Amazon in ENSO and Non-ENSO Years: Area Burned and Committed Carbon Emissions. Earth In teractions 10(6), 1-17. Allan, J.D., 2004. Landscapes and riverscapes: the influence of land use on stream ecosystems. Annual Review of Ecology and Systematics 35, 257. 174

PAGE 175

Almeida, C.M., Batty, M., Monteiro, A.M.V., Camara, G., Soares-Fil ho, B.S., Cerqueira, G.C., Pennachin, C.L., 2003. Stochastic cellular automata modeling of urban land use dynamics: empirical dev elopment and estimation. Computers, Environment, and Urban Systems 27, 481-509. Almeida, C.M., Gleriani J.M., Castejon, E.F., Soares F ilho, B.S., 2008. Neural networks and cellular automata for modeling intraurban land use dynamics. International Journal of Geographical Info rmation Science 22, 943-963. Alston, L.J., Libecap, G.D., and Mueller, B., 1999. Titles, conflict, and land use: the development of property rights and land reform on the Brazilian Amazon frontier. University of Michigan Press, Ann Arbor. Ames, B., Keck, M., 1998. The politics of sustainable development: environmental policy making in four Brazilian states. J ournal of Interamerican Studies and World Affairs 39 (4), 5-40. Andreae, M.O., Rosenfeld, D., Artaxo, P., Co sta, A.A., Frank, J.P ., Longo, K.M., SilvaDias, M.A.F., 2004. Smoking rain clouds over the Amazon. Science 303, 1337 1342. (doi:10.1126/science.1092779 ) Angelsen, A., Streck, C ., Peskett, L., Brown, J., Luttrell, C. 2008. What is the right scale for REDD? The implications of national, s ubnational and nested approaches. Center for International Forestry Research (C IFOR), Bogor, Indonesia. Accessible at http://www.cifor.cgiar.org/carbofor /projects/globalredd/Publications.htm (accessed on November 2, 2009) Ankersen, T.T., Ruppert, T.R. 2006. Tierra y Libertad: the social function doctrine and land reform in Latin America. Tulane Environmental Law Journal 69, 70-120. Ascher, W., 1999. Why Governments Waste Na tural Resources: Policy Failures in Developing Countries. The Johns Hopkins University Press, Baltimore. Asner, G.P, Lobell, D.B., 2000. A biogeophy sical approach for automated SWIR unmixing of soils and vegetation. Remote Sensing of Environment 74 (1), 99-112. Asner, G.P., Vitousek, P.M., 2005. Remote analysis of biological invasion and biogeochemical change. Proceedings of t he National Academy of Sciences of the United States of Am erica 102 (12), 4383-4386. Asner, G.P., Martin, R.E., 2009. Airborne sp ectranomics: mapping canopy chemical and taxonomic diversity in tropical forests. Frontiers in Ecology and the Environment 7 (5), 269-276. Asner G.P., Knapp D.E., Broadbent E.N., 2005. Selective logging in the Brazilian Amazon. Science 310 (5747), 480-482. 175

PAGE 176

Automated Remote Biodiversity Monitori ng Network (ARBIMION), 2009. Overview. University of Puerto Rico, San Juan, Puerto Rico. Accessible at http://arbimon.uprrp. edu/arbimonweb/index.php (accessed on November 2, 2009) Azevedo, A.A., 2009. Legitimao da in sustentabilidade? Anl ise do Sistema de Licenciamento Ambiental de Propriedades Rurais SLAPR (Mato Grosso). Ph.D. Dissertation, Federal University of Braslia, Braslia, DF. Azevedo, A.A., Pasquis, R., 2006. PROJET O DILOGOS. Identificao de temas e espaos de formao e dilogo pelo se tor produtivo ligado cadeia da soja no estado do Mato Grosso. WWF, Braslia. Accessible at www.dialogos.org (accessed on November 2, 2009) Azevedo-Ramos, C., Kalif, K., de Carvalho, O., 2004. As madeiras e a conservao da fauna. Revista Cincia Hoje 34, 111-120. Baatz, M., Benz, U., Dehghani, S., 2004. eCognition: User Guide 4 Definiens Imaging GmbH, Munich, Germany. 72 pp. Accessible at http://www.gis.unbc.ca/help/softw are/ecognition4/ELuserguide.pdf (accessed on November 29, 2009) Baccini, A., Laporte, N., Goetz, S.J., Sun, M., Dong, H.,2008. A first map of tropical Africa's above-ground biomass derived from satellite imagery. Environmental Research Letters 3, 045011 Baguette, M., Schtickzelle, N., 2003. Local population dynamics are important to the conservation of metapopulations in highly fragmented landscapes. Journal of Applied Ecology 40 (2), 404-412. Balch, J.K. 2008. Effects of Recurrent Fi re on Transitional Forest Dynamics in the Amazons Wildfire Frontier in Mato Grosso, Brazil. PhD dissertation, Yale University, New Haven, CT. Ballester, M.V.R., Victoria, D.D.C., Krusche, A.V., Coburn, R., Victoria, R.L., Richey, J.E., Logsdon, M.G., Mayorga, E., Matricardi, E., 2003. A remote sensing/GIS-based physical template to understand the biogeoc hemistry of the Ji-Paran river basin (Western Amazon). Remote Sensing of the Environment 87, 429-445. Banco Nacional de Desenvolvimento (BNDES) 2009. Fundo Amaznia. Accessible at http://www.bndes.gov.br/SiteBND ES/bndes/bndes_pt/Fundo_Amazonia/Fundo/ (accessed November 2, 2009) Barbier, E.B., Burgess, J.C., 1997. Econom ic Analysis of Forest Land Use Options. Land Economics 73 (2), 174-195. 176

PAGE 177

Barlow J., 2007. Quantifying the biodiversity value of tropical primary, secondary, and plantation forests. Proceedings of the Nati onal Academy of Sciences of the United States of America 104 (47), 18555-18560. Barlow, J., Haugaasen, T., Pe res, C.A., 2002 Effects of ground fires on understory bird assemblages in Amazonian forests. Biological Conservation 105, 157-169. Barton, D.R., Taylor, W.D., Biette, R.M., 1985. Dimensions of riparian buffer strips required to maintain trout habitat in s outhern Ontario streams. North American Journal of Fisheries Management 5, 364. Bauer, E., Kohavi, R., 1999. An empirical comparison of vo ting classification algorithms: Bagging, boosting, and variants. Machine Learning 36, 105-139. Beaulac, M.N., Reckhow, K.H., 1982. An examination of land use-nutrient export relationships. Water Resources Bulletin 18, 1013-1024 Benatti, J.H., 2003. Direito de Propriedade e Proteo Ambiental no Brasil: apropriao e o uso dos recursos naturais no imvel rural. Ph.D. Dissertation, Center for Advanced Amazonian Studies, Federal Un iversity of Par, Belm, Brasil. Beissinger, S.R., Westphal, M.I., 1998. On the use of demographic models of population viability in endang ered species management. Journal of Wildlife Management 62 (3), 821-841. Beissinger, S.R., McCullough, D. R., 2002. Population viability analysis. University of Chicago Press, Chicago. Benstead, J.P., Pringle, C.M., 2004. Defo restation alters the resource base and biomass of endemic stream insects in eas tern Madagascar. Freshwater Biology 49, 490. Bierregaard, R.O., 2001. Lessons from Amaz onia: the ecology and conservation of a fragmented forest. Yale University Press, New Haven. Biggs, T.W., Dunne, T., Domingues, T. F., Marti nelli, L.A., 2002. The relative influence of natural watershed properties and hum an disturbance on stream solute concentrations in the southwestern Br azilian Amazon basin. Water Resources Research 38, 10.1029/2001WR000271. Biggs, T.W., Dunne, T., Mart inelli,L.A., 2004. Natural controls and human impacts on stream nutrient concentrations in a devel oping region of the Brazilian Amazon basin, Biogeochemistry 68, 227-257. 177

PAGE 178

Blackard, J.A., Finco, M. V., Helmer, E.H., 2008. Mapping U.S. forest biomass using nationwide forest inventory data and moderate resolution information. Remote Sensing of Environment 112, 1658-1677. Bojsen, B.H., Barriga R., 2002. E ffects of deforestation on fish community structure in Ecuadorian Amazon streams. Fres hwater Biology 47, 2246. Bonham-Carter, G., 1994. G eographic Information Systems fo r Geoscientists: Modelling with GIS. Pergamon, Kidlingt on, United Ki ngdom. 398 pp. Bonan, G.B., Pollard, D., Thompson, S.L., 1992. Effects of boreal fo rest vegetation on global climate. Nature 359 (6397), 716-718. Bormann, F.H., Likens, G.E., 1979. Pattern and Process in a Forested Ecosystem. Springer-Verlag, New York, 253 pp. Bosch, J.M., Hewlett, J.D., 1982. A review of catchment experiments to determine the effect of vegetation changes on water yi eld and evapotranspiration. Journal of Hydrology 103, 323-333. Bowman, M.S., Amacher, G.S., Merry F.D., 2008. Fire us e and prevention by traditional households in the Brazilian Amazon. Ecological Economics 67 (1), 117-130. Breiman, L., 2001. Random fo rests. Machine Learning 45 5. Brito, B., and Barreto, P. A Eficcia da aplicao da lei de crim es ambientais pelo IBAMA para proteo de florestas no Pa r. IMAZON, Belm, Brasil, 33 pp. Brown, D., Seymour, F., Pe skett, L., 2008. How do we ac hieve REDD co-benefits and avoid doing harm? In Moving Ahead with REDD (Ed. A. Angelsen). CIFOR, Bogor, Indonesia, 107-118. Accessible at http://www.cifor.cgiar.org/publications/pdf_files/Books/BAngelsen0801.pdf (accessed on November 2, 2009) Bruijnzeel, L.A., 1990. Hydrology of moist tr opical forests and effects of conversion: a state of knowledge review. UNESCO, 224. Burcham, J., 1988. Fish communities and envir onmental characterist ics of two lowland streams in Costa Rica. Revist a de Biologia Tropical 36, 273. Busch, J., Strassburg, B., Cattaneo, A., Comparing REDD mechanism design options with an open source economic model. Ec ological Economics, in review. Camarga, 2008. http://www.inesc.org.br/noticia s/noticias-gerais/2008/julho/bancadaruralista-a-um-passo-da-vitoria/ (accessed on November 2, 2009) 178

PAGE 179

Campari, J.S., 2005. The Economics of Deforestation in the Amazon: Dispelling The Myths. Edward Elgar Publishing. Campos, M.T., Nepstad, D.C., 2006. Smallholders, The Amazons New Conservationists. Conservation Biology 20 (5), 1553-1556. Carpenter, S.R., Caraco, N.F. Correll, D.L., Howarth, R. W., Sharpley, A.N., Smith, V.H., 1998. Nonpoint pollution of surfac e waters with phosphorus and nitrogen. Ecological Applications 8, 559-568. Carter, N., 2001. The Politics of the En vironment: Ideas, Activism, Policy. The Cambridge University Press, Cambridge. Chan, K.M.A., Pringle, R.M., Ranganathan, J., 2007. When agendas collide: Human welfare and biological conservation. Conservation Biology 21 (1), 59-68. Chaves, J., 2008. Land management impacts on runoff sources in small Amazon watersheds, Hydrological Processes 22, 1766-1775. Chomitz, K.M., 2004. Transfe rable development rights and forest protection: An exploratory analysis. International Regi onal Science Review 27 (3), 348-373. Climate, Community and Biodiversity Alliance (CCBA), 2008. Climate, Community and Biodiversity Project Design Standards Second Edition. CCBA, Arlington, VA. Accessible at http://www.climatestandards.org/standards/pdf/ccb_standards_second_edition_december_2008.pdf (accessed on November 2, 2009) Cochrane, M.A., 2003. Fire Science fo r Rainforests. Nature 421, 913-919. Cochrane, M.A., Souza, C.M., 1998. Linear mixture model classification of burned forests in the Eastern Amazon. Internati onal Journal of Remote Sensing 19 (17), 3433-3440. Coe, M.T., 2000. Modeling terrestrial hydr ological systems at t he continental scale: Testing the accuracy of an atmospheric GCM. Journal of Climate 13, 686-704. Coe, M.T., Costa,M.H., S oares-Filho,B.S., 2009. The In fluence of historical and potential future deforestation on the stream flow of the Amazon River Land surface processes and atmospheric feed backs. Journal of Hydrology doi:10.1016/j.jhydrol.2009.02.043 Colby, K.E., 2003. Brazil and the MST: land reform and human rights. New York International Law Review 1,10-11. 179

PAGE 180

Cook, W.M., Lane, K.T., Foster, B.L., Holt, R.D. 2002. Island theory, matrix effects and species richness patterns in habitat fragments. Ecology Letters 5, 619. Congresso Nacional (CN), 2000. Commis o Mista destinada a examinar e emitir parecer sobre a Medida Provisria N 1.956 -48. Congresso Nacional, Brasilia, April 17. Accessible at http://webthes.senado.gov.br/sil/Comissoes /Mistas/MP/Comissoes/20001956048/At as/20000417RO006.rtf (accessed on November 2, 2009) Costa, M.H., Foley,J.A., 1997. Water balanc e of the Amazon Basin: Dependence on vegetation cover and canopy conductance. Journal of Geophysical ResearchAtmospheres 102 (D20), 23,973-23,989. Costa, M.H., Foley, J.A., 2000. Comb ined effects of deforestation and doubled atmospheric CO2 concentrations on the climate of Amazonia, Journal of Climatology 13, 18 34. Costa, M.H., Botta, A., Cardille, J.A., 2003. Effects of large-scale changes in land cover on the discharge of the Tocantins River southeastern Amazonia. Journal of Hydrology 283, 206-217. di o-10.1016/S0022-1694(03)00267-1. Costa, M.H., Yanagi, S.N.M., Souza, P.J.O.P ., Ribeiro, A., Rocha, E.J.P., 2007. Climate change in Amazonia caused by soybean cr opland expansion, as compared to caused by pastureland expansion. Geoph ysical Research Letters 34, L07706. doi:10.1029/2007GL029271 Costanza, R., Darge, R., De Cross, R., Farber, S., Hannon, B., Raskin, R.C., Sutton, P., 1997. The value of the worlds ecosystem services and natural capital. Nature 338: 253-260. Cox, P.M., Betts, R.A., 2000. Accelera tion of global warming due to carbon cycle feedbacks in a coupled climate model. Nature 408, 184-187. Cumberlidge, N., Ng, P.K.L., Yeo, D.C.J., Magalhes, C., Campos, M.R., Alvarez, F., Naruse, T., Daniels, S.R., E sser, L.J., Attipoe, F.Y.K., Clotilde-Ba, F.L., Darwall, W., McIvor, A., Baillie, J.E.M. Collen, B., Ram, M., 2009. Freshwater crabs and the biodiversity crisis: Importance, threat s, status, and conservation challenges. Biological Conser vation 142:1665. Curran, L.M., Trigg, S.N., McDonald, A.K., As tiani, D., 2004. Lowland forest loss in protected areas of Indonesian Borneo. Science 303 (5660),1000-1003. Daly, H.E., Farley, J., 2003. Ec ological Economics: Principl es and Applications. Island Press, Washington, DC. 180

PAGE 181

Daily, G. C., 1997. Introduction: What are ecosystem services? In Nature's Services: Societal Dependence on Nature's Ecosystems (Ed. G. Dailey). Island Press, Washington, D.C. Davidson, E.A., Neill, C., Krusch, A.V., Ballester, V.V.R., Markewitz, D., Figueiredo,R.O., 2004. Loss of nutrients from terrestrial ecosystems to streams and the atmosphere following land use change in Amazonia. In Ecosystems and Land Use Change (Eds. DeFries, R., Asner, G., Houghton R.). G eophysical Monograph Series 153, 147-158. American Geophysi cal Union, Washington, D.C. Davidson, E.A., de Carvalho, C.J.R., Figuei ra, A.M., Ishida, F. Y., Ometto, J.P.B., Nardoto, G.B., Sab, R.T., Hayashi, S.N., Leal, E.C., Vieira, I.C.G., Martinelli, L.A., 2007. Recuperation of nitrogen cycling in Am azonian forests following agricultural abandonment. Nature 447, 995-998. Daviet, F., McMahon, H., Sto lle, F., 2007. REDD Flags: what we need to know about the options. WRI, Washingt on, D.C. Accessible at http://pdf.wri.org/redd-flags.pdf (accessed on November 2, 2009) Dean, W., 1995. With Broadax an d Firebrand: Destruction of the Brazilian Atlantic Forest. University of California Press, Berkeley, CA. Delire, C., Behling, P ., Coe, M.T., Foley, J.A., Jacob, R., Kutzbach, J., Liu, Z., Vavrus, S., 2001. Simulated response of the atmos phere-ocean system to deforestation in the Indonesian Archipelago. Geophysical Research Le tters 28 (10), 2081-2084. Dickinson, R.E., Henderson-Sellers, A., 1988. Modelling tropical def orestation: a study of GCM land-surface parameterizati ons. Quaternary Jour nal of the Royal Metereological Society 114, 439-62. Dietterich, T.G., 2000. An experimental comparison of three methods for constructing ensembles of decision trees: baggi ng, boosting, and randomization. Machine Learning 40, 139-157. Dietz, J.M., Dietz, L.A., 1994. The effective us e of flagship species for conservation of biodiversity: the example of lion tamarins in Brazil. In Creative Conservation (Eds. P. Olney, G. Mace), 145-161. Dietz, T., Ostrom, E., Stern, P.C., 2003. The struggle to govern the commons. Science 302, 1907-1912. Dooley, K., Griffiths, T., Leake, H., Ozi nga, S., 2008. Cutting Corners World Banks forest and carbon fund fails forests and peoples. FERN / Forest Peoples Programme, London. Accessible at http://www.fern.org/medi a/documents/document_4312_4313.pdf (accessed on November 2, 2009) 181

PAGE 182

Engel, S., Pagiola, S., Wunder, S., 2008. Designing payments for environmental services in theory and practice: an overview of the issues. Ecological Economics 65 (4), 663-674. Fearnside, P.M., 2003. Deforest ation control in Mato Grosso: a new model for slowing the loss of Brazils Amaz on forest. Ambio 32, 343-345. Fearnside, P.M., 2005. Deforestation in Brazilian Amazonia: History, Rates, and Consequences. Conservati on Biology 19 (3), 680. Fearnside, P.M. 2008. The roles and movement s of actors in the deforestation of Brazilian Amazonia. Ecology and Society 13: 23. Fimbel, R.A., Grajal, A., Robi nson, J.G., 2001. Loggingwildlife issues in the tropics: an overview. In The Cutting Edge: Conserving W ildlife in Logged Tropical Forests (Eds. R.A. Fimbel, A. Grajal, J.G. Robinson), 3-9, Columbia Un iversity Press, New York. Fischer, J., Lindenmayer, D.B., 2006. Beyond fragmentation: the continuum model for fauna research and conservation in human-modified landscapes. Oikos 112, 473 480. Flecker, A. S. 1992. Fish trophic guilds and the structure of a tr opical stream: weak direct vs. strong indirect effects. Ecology 73, 927. FNP Consultoria e Comrcio (FNP), 2005. Anlise de Mercado de Terras: Relatrio Bimestral. Accessible at www.fnp.com.br (accessed on December 1, 2009) Foley, J.A., Prentice, I.C., Ramankutty, N., Levi s, S., Pollard, D., Sitch, S., Haxeltine, A., 1996. An integrated biosphere model of land surface processes, terrestrial carbon balance, and vegetation dynamics. Global Biogeochemical Cycles 10 (4), 603-628. Folha de So Paulo, October 15, 2009. Ruralistas obtm comando de debate sobre Cdigo Florestal. Accessible at http://www1.folha.uol.com.b r/fsp/brasil/fc1510200915.htm (accessed on October 29, 2009) Food and Agriculture Organization (FAO), 2009. The State of Agricultural Commodity Markets 2009: High food prices and the food crisisexperiences and lessons learned. Accessible at http://www.fao.org/docrep/012/i0854e/i0854e00.htm (accessed on November 29, 2009) Forest Carbon Partnership Facility (FCP F), 2009. About the FCPF. Accessible at http://www.forestcarbonpar tnership.org/fcp/node/12 (accessed on November 2, 2009) 182

PAGE 183

Friends of the Earth Inte rnational (FOEI), 2008. REDD myths : a critical review of proposed mechanisms to reduce emissions from deforestation and degradation in developing countries. FOEI, London. www.foei.org.uk (accessed on November 2, 2009) Gibbs, H.K, Brown, S.L., 2007. Monitoring a nd estimating tropical forest carbon stocks: making REDD a reality. Environmental Research Letters 2, 045023. doi: 10.1088/1748-9326/2/4/045023 (accessed on November 2, 2009) Goetz, S.J., Bunn, A.G., Fi ske, G., Houghton, R.A., 2005. Satellite-observed photosynthetic trends across boreal No rth America associated with climate and fire disturbance. Proceedings of the National Academy of Sciences 102, 13521-13525. GOFC-GOLD, 2009. Reducing greenhouse gas emissions from deforestation and degradation in developing countries: a sour cebook of methods and procedures for monitoring, measuring and reporting, GOFC-GOLD Report version COP14-2, GOFC-GOLD Project Office, Natural Resources Canada, Alberta, Canada. Accessible at http://www.gofc-gold.unijena.de/redd/sourcebook/Sourceb ook_Version_July_2009_cop14-2.pdf (accessed on November 2, 2009) Goldammer, J.G., ed., 1990. Fire in the tr opical biota. Ecosystem processes and global challenges. Ecological Studies 84, Springe r-Verlag, Berlin-Heidelberg-New York, 497 p. Goodwin, R.A., Pandey, V., Kiker, G.A., 2007. Spatially-explicit population models with complex decisions.. In Environmental Se curity in Harbors and Coastal Areas, (Linkov, I., Kiker, G.A., et al. Ed s.), 293-306, Springer Netherlands. Goulding, M., 1980. The fishes and the fore st: explorations in Amazonian natural history. University of Californi a Press, Berkeley, California. Government of Norway (GON), 2009. International Climate and Forest Initiative Funding Scheme. Accessible at http://www.norway.org.vn/development/E nvironment/Norway+Intl+Climate+and+For est+Initiative.htm (accessed on November 2, 2009) Government of Brazil (GOB), 2008. Pol tica Nacional sobre Mudana do Clima. Accessible at http://www.mma.gov.br/estruturas /imprensa/_arquivos/96_11122008040728.pdf (accessed on November 2, 2009) Government of Mato Grosso (GOMT), 2009. Reference Document for the Development of Mato Grosso States REDD Program. November 26, 2009, Cuiab, Mato Grosso. 183

PAGE 184

Green, G.M., Schweik, C., Hanson, M., 2002. Radiometric calibration of LANDSAT multi-spectral scanner and thematic mapper images:Guidelines for the global change community. Center for the Study of Institutions, Population, and Environmental Ch ange (CIPEC), Indiana University, Bloomington, Indiana. Gregory, S.V., Swanson, F. J., McKee, W.A., Cummins, K.W., 1991. An ecosystem perspective of riparian zones. BioScience 41, 540. Gullison, R.E., Frumhoff, P.C., Canadell, J.G., 2007. Tropical forests and climate policy. Science 316 (5827), 985-986. Guimaraes, J. Almeida, O., 2007. Anlise de custos de produo da pecuria e da soja na regio nordeste do Mato Grosso. CNPq/ IPAM, Belm, PA. Gustafson, E.J., Crow, T.R. 1996. Simulating the effects of alternative forest management strategies on landscape stru cture. Journal of Environmental Management 46, 77-94. Hagen, A., 2003. Fuzzy set approach to asse ssing similarity of categorical maps. International Journal of Geographic Information Science 17, 235-249. Hagen-Zanker, A., 2006. Map comparison methods that simultaneously address overlap and structure. Journal of Geographic Systems 8, 165-185. Hansen, M.C., Stehman, S.V., Potapov, P.V., 2008. Humid tropical forest clearing from 2000 to 2005 quantified by using multitempora l and multiresolution remotely sensed data. Proceedings of the National Ac ademy of Sciences 105, 9439-9444. Hardin, G., 1968. The Tragedy of t he Commons. Science 162, 1243-1248. Hecht, S., 2005. Soybeans, Development an d Conservation on the Amazon Frontier. Development and Change 36(2), 375. Herold, M., Johns, T. 2007. Linking requirements with capabilitie s for deforestation monitoring in the context of the UNFCCCREDD process. Environmental Research Letters 2, 045025, doi: 10.1088/1748-9326/2/4/04502 Hochstettler, K., Keck, M.E., 2007. Greening Br azil: Environmental Activism in State and Society. Duke University Press, Durham, NC. Hoffmann, W.A., Orthen, B., 200 3. Comparative fire ecology of tropical savanna and forest trees. Functional Ecology 17, 720-726. Holdsworth, A.R., Uhl, C., 1997. Fire in Am azonian selectively logged rain forest and the potential for fire reduction. Ec ological Applications 7, 713. 184

PAGE 185

Holmes, T.P., Blate, G.M., Zweede, J.C., Pereir a, R., Barreto, P., Boltz, F., Bauch, R., 2002. Financial and ecological indicators of reduced impact logging performance in the eastern Amazon. Forest Ecol ogy and Management 163 (1-3), 93-110. Hlscher, S, D., T.D.d.A., Ba stos, T.X., Denich, M., Flste r, H., 1997. Evaporation from young secondary vegetation in eastern Amaz nia. Journal of Hydrology 193, 293305. Hooijer, A., Silvius, M., 2006. PEAT-CO2, Assessment of CO2 emissions from drained peatlands in SE Asia. Delft Hy draulics report Q3943, 7. Accessible at www.wetlands.org/ckpp/publication.aspx ?ID=f84f160f-d851-45c6-acc4-d67e78b39699 (accessed on November 2, 2009) Houghton, R., Skole, D., Nobre, C., Hackler J., Lawrence, K., Chementowski, W., 2000. Annual fluxes of carbon from deforesta tion and regrowth in the Brazilian Amazon. Nature 403: 301. Howarth, R.W., Billen, G., Swaney, D., Towns end, A., Jaworski, N., Lajtha, K., Downing, J.A., Elmgren, R., Caraco, N., Jordan, T., Berendse, F. Freney, J., Kudeyarov, V., Murdoch, P., Zhao-Liang, Z., 1996. R egional nitrogen budgets and riverine N & P fluxes for the drainages to the North Atlantic Ocean: Natural and human influences. Biogeochemistry 35, 75-139 Howe, H.F., Smallwood,J.F., 1982. Ecology of seed dispersal. Annual Review of Ecology and Systematics 13, 201-228. HR 2454, American Clean Energy and Securi ty Act of 2009, 111th Congress, 2009. Instituto Brasileiro de Geografia e Estatsti ca (IBGE), 1981. Pr ojeto RADAMBRASIL, Vol. 22, Folha SC.22 Tocantins. Brasilia, Brasil. Instituto Nacional de Pesquisa Espacia is (INPE), 2009. Pr ojeto Desmatamento (Prodes): Monitoramento Da Floresta Amazonica Por Satelite. Accessible at http://www.obt.inpe.br/prodes/ .(accessed on November 2, 2009) Instituto de Pesquisa Am biental da Amaznia (IPAM), 2009. Xingu Social Carbon Program: Potential for REDD. Presentat ion at the Forum for REDD Readiness South-South Collaboration Workshop, Marc h 15-20, Manaus, Brazil. Accessible at http://www.whrc.org/Policy/REDD/Reports/FlaviaGabriela-Introduction_Ipam.pdf (accessed on November 2, 2009) Instituto Sociambiental (ISA), 2005. Cam panha `Y Ikatu : Mobilizao para salvar nascentes do Rio Xingu. Accessible at http://www.socioambiental.org/in st/camp/xingu/enc/acarta.html (accessed on November 2, 2009) 185

PAGE 186

Intergovernmental Panel on Climate Change (IPCC), 2007. Climate change 2007: the physical science basis. Contribution of work ing group I to the fourth assessment report of the IPCC. Cambridge University Press, Cambridge. Jansen, P.A., Zuidema, P.A., 2001. Logging, seed dispersal by vertebrates, and natural regeneration of tropical tim ber trees. In The Cutting Ed ge: Conserving Wildlife in Logged Tropical Forests (Eds. R.A. Fimbel A. Grajal, J.G. Robinson), 35-59. Columbia University Press, New York. Janssen, L.L.F., Jaarsma, M.N., van der Lind en, E.T.M., 1990. In tegrating topographic data with remote sensing for land-cover classification. Photogrammetric Engineering and Remote Sensing 56, 1503-1506. Jaramillo, C., Kelly, T., 1997. Deforestation and Property Rights in Latin America. InterAmerican Development Bank, Washington, DC. Accessible at: http://www.iadb.org/sds/publication/publication_1030_e.htm (accessed on November 2, 2009) Jensen, J.R., 2004. Introductory Digital Image Processing, 3rd Ed. Prentice Hall Series in Geographic Information Science (Ed. K.C. Clark), Prentice Hall, Upper Saddle River, New Jersey. 526 pp. Jepson, W., 2006. Private agricultural coloni zation on a Brazilian frontier, 1970-1980. Journal of Historica l Geography 32, 839-863. Jipp, P., Nepstad, D., Casse l, K., de Carvalho, C.R., 1998. Deep soil moisture storage and transpiration in forests and pastures of seasonally-dry Amaznia. Climatic Change 39, 395-412. Johns, T., Merry, F., Stickler, C., Nepstad, D., Laporte, N., Goetz, S., 2008. A three-fund approach to incorporating government, public and private forest stewards into a REDD funding mechanism. Internationa l Forestry Review 10 (3), 458-464. Justice, C.O., Giglio, L., Korontzi, S., Owens, J., Morisette, J.T., Roy, D., Descloitres, J., Alleaume, S., Petitcolin, F., Kaufman, Y., 2003. The MODIS fire products. Remote Sensing of Environment 83, 244-262. Karr, J.R. 1991. Biological integrity: A long-neglected aspect of water resource management. Ecological Applications 1, 66. Karr, J. R., Schlosser, I.J., 1978. Wate r resources and the landwater interface. Science 201, 229. 186

PAGE 187

Kauffman, J.B., Cummings, D.L., Ward, D.E ., Babbitt, R., 1995. Fire in the Brazilian Amazon: biomass, nutrient pools, and losses in slashed primary forests. Oecologia 104, 397-409. Kellndorfer, J.M., Shimada, M., Rosenqvist, A., Walker, W.S., Kirsch, K, Nepstad, D.C., Laporte, N., Stickler, C.M., Lefebvre, P., 2007. New Eyes in the Sky: Cloud-Free Tropical Forest Monitoring for REDD with the Japanese Advanced Land Observation Satellite (ALOS). Report fo r the United Nations Framew ork Convention on Climate Change (UNFCCC) Conferenc e of the Parties (COP), Thirteenth session, 3-14 December, The Woods Hole Research Center, Falmouth, MA. Accessible at http://whrc.org/BaliRepor ts/assets/Bali_ALOS.pdf (accessed on November 2, 2009) Kindermann, G., Oberstei ner, M., Sohngen, B., Sathaye, J., Andrakso, K., Rametsteiner, E., Schlamadinger, B., Wunder S., Beach, R., 2008. Global cost estimates for reducing carbon emissi ons through avoided deforestation. Proceedings of the Nati onal Academy of Science of the Unites States of America. http://www.pnas.org/content/105/30/10302.full (accessed on November 2, 2009) Krishnaswamy, J., Bawa, K.S., Ganesahi ah, K.N., 2009. Quantifying and mapping biodiversity and ecosystem services: Utility of a multi-season NDVI based Mahalanobis distance surrogate. Remote Sens ing of Environment 113 (4), 857-867. Kucharik C.J., Foley J.A., Delire C., Fis her V.A., Coe M.T., Lenters J., Young-Molling C., Ramankutty N., Norman J.M., Gower S.T., 2000. Testing the performance of a dynamic global ecosystem model: Water balance, carbon balance and vegetation structure. Global Biogeochemic al Cycles 14 (3), 795-825. Lal, R., 1995. Global soil erosion by water and carbon dynamics. In Soils and global change (Eds. R. Lal, J. Kimble, E. Levi ne, B.A. Stewart) 131-142. CRC/Lewis Publishers, Boca Raton, FL. Landell-Mills, N., Porras, T.I., 2002. Silver bullet or fools gold? A global review of markets for forest environmental services and their impact on the poor. Instruments for sustainable private sector forestry seri es. International inst itute for Environment and Development, London. www.iied.org (accessed on November 2, 2009) Lang, C., 2009. Halt climate change. Halt fo rest destruction. Halt plantations. REDD Monitor, June 10. Accessible at http://www.redd-monitor.org/2009/06/10/haltclimate-change-halt-forestdestruction-halt-plantations/ (accessed on November 2, 2009) Larsen, J., Heilmayr, R., 2009. Emissions Reductions under the American Clean Energy and Security Act of 2009. World Resources In stitute, Washington, D.C. Accessible at http://pdf.wri.org/usclimatetargets_2009-05-19.pdf (accessed on November 26, 2009) 187

PAGE 188

Latrubesse, E.M., Chen, Z., St evaux, J.C. Short and long te rm processes, landforms and responses in large rivers Geomorphology, in press. doi:10.1016/j.geom orph.2009.03.018 Laurance, W.F., Laurance, S.G. W., 1996. Responses of five arboreal marsupials to recent selective logging in tropica l Australia. Biotropica 28, 310-322. Laurance, W.F., Bierregaard, R.O., Jr., eds. 1997. Tropical Forest Remnants: Ecology, Management and Conservation of Fragmented Communities. University of Chicago Press, Chicago. Laurance, W. F., Cochrane, M.A., Bergen, S., Fearnside, P.M., Delamonica, P., Barber, C., D'Angelo, S., Fernandes, T., 2001. The fu ture of the Brazilian Amazon. Science 291, 438-439. Lentini, M., Verssimo, A., Sobral,L., 2003. Fa tos Florestais da Amaznia 2003. Imazon, Belm, Brasil. Accesible at: http://www.imazon.org.br/upload/im_livros_002.pdf (accessed on November 2, 2009) Li,W., Fu, R., Dickinson, R.E., 2006. Rainfall and its seasonali ty over the Amazon in the 21st century as assessed by the coupled models for the IPCC AR4. Journal of Geophysical Research 111, D02111, doi:10.1029/2005JD006355. Liaw, A., Wiener, M., 2002. Classification and regression by randomForest. R News 2/3, 18-22. Lima, A., Irigaray, C.T., Silva, R.T., Guimaraes, S., Araujo, S., 2005. Sistema de Licenciamento Ambiental em Propriedades Ru rais do Estado de Mato Grosso: Anlise de Lies na Sua Implementao (Rel atrio Final). Ministrio do Meio Ambiente/Secretaria de Coor denao da Amaznia/Programa Piloto para a Proteo das Florestas Tropicais do Brasil/Projet o de Apoio ao Monitoramento e Anlise (AMA), Brasilia, July. (Projeto PNUD: BRA 98/0005). Lindenmayer, D.B., Margules, C.R., Botkin, D.R., 2000. Indicators of Biodiversity for Ecologically Sustainable Forest Managemen t. Conservation Biology 14 (4), 941. doi:10.1046/j.1523-1739.2000.98533.x Lorion, C.M., Kennedy, B.P. 2009. Relationships Between Deforestation, Riparian Forest Buffers And Benthic Macroinvertebrates In Neotropical Headwater Streams. Freshwater Biology 54, 165. Lorion, C. M., Kennedy, B.P. 2009. Riparian forest buffers mitigate the effects of deforestation on fish assemblages in tropical headwater streams. Ecological Applications 19 (2), 468. 188

PAGE 189

Lowrance, R., Altier, L.S., Newbold, J.D., Schnabel, R.R., Groffman, P.M., Denver, J.M., Correll, D.L., Gilliam, J.W., Robinson, J. L., BRinsfield, R.B., 1997. Water quality functions of riparian forest buffers in Chesapeake Bay watersheds. Environmental Management 21, 687. Malhi, Y., Timmons, R.J., 2008. Climate c hange, deforestation and the fate of the Amazon. Science 319, 169. doi:10.1126/science.1146961 Malhi, Y., Aragao, L.E.O.C., Galbraith, D., 2009. Exploring the likelihood and mechanism of a climate-change-induced dieback of the Amazon rainforest. Proceedings of the Nati onal Academy of Science of the United States of America Mansourian, S., Vallauri, D., Dudley, N., 2005. Forest restoration in landscapes: beyond planting trees. Springer, New York. Markewitz, D., Davidson, E. A., Figueiredo, R., Victoria, R. L., Krusche, A.V., 2001. Control of cation concentrations in stream waters by surface soil processes in an Amazonian watershed. Nature 410, 802-805. May, P.H., 1999. Natural Resource Valuation and Policy in Brazil. Columbia University Press, New York. McFarland, A.M.S., Hauck, L.M., 1999 Relating agricultura l land uses to in-stream stormwater quality. Journal of Environmental Quality 28, 836-844. McGarigal K., Cushman, S.A., Neel, M.C., Ene, E., 2002. FRAGSTATS: Spatial Pattern Analysis Program for Categorical Maps. University of Massachusetts, Amherst. Accesible at: www.umass.edu/landeco/research /fragstats/fragstats.html (accessed on November 2, 2009) McGrath, D.A., Smith, C.K., Gholz, H.L., Oliveira, F.D. A. 2001. Effects of land-use change on soil nutrient dynamics in Am aznia, Ecosystems 4, 625-645. Medida Provisoria (MP) 2.166-67, 2001. Presidncia da Repblic a, Casa Civil, Brasila, Brasil, August 24, 2001. Accessible at http://www.planalto.gov.br/ccivil_03/MPV/2166-67.htm#art1 (accessed on November 29, 2009) Melo, C.E., Machado, F.A., Pinto-Silva, V., 2003. Diversidade de peixes em um crrego de Cerrado no Brasil Centra l. Brazilian Journal of Ecology 1-2, 17-23. Mendona, F.P., Magnusson, W.E., Zuanon, J. 2005. Relationships Between Habitat Characteristics and Fish Assemblages in Small Streams of Central Amazonia. Copeia 2005 (4), 751 189

PAGE 190

Meridian Institute, 2009. Reducing Emissions form Deforestation and Degradation (REDD): an options assessment report. Pre pared for the Government of Norway. Meridian Institute, Washi ngton, DC. Accessible at http://www.REDD-OAR.org (accessed on November 2, 2009) Merry, F.D., Soares-Filho, B.S., Nepstad, D.C., Amacher, G., Rodrigues, H.O., 2009. Balancing conservation and ec onomic sustainability: the fu ture of the Amazon timber industry. Environmental Management 44 (3), 395-407. Ministrio da Cincia e Tecnologia (MCT), 2009. Inventrio Brasileiro das Emisses e Remoes Antrpicas de Gases de Efeito Estufa. Accessible at http://mct.gov.br (accessed November 25, 2009) Miles, L., Kapos, V., 2008. Reducing Greenh ouse Gas Emissions from Deforestation and Forest Degradation: Global Land-Use Im plications. Science 320 (5882), 14541455. Molnar, A., Scherr, S., Khar e, A., 2003. Who conserves the world's forests? A new assessment of conservation and investment trends. Ecoagriculture Partners, Washington, D.C. Montgomery, D.R., 2007. Soil er osion and agricultural sustainability. Proceedings of National Academy of Scienc es 104 (33), 13268. Moorcroft, P.R., Hurtt, G.C., Pacala, S.W., 2001. A method for scaling vegetation dynamics: the ecosystem demography model (ED). Ecological Monographs 71 557585. Moraes, J.M. de, Schuler, A. E., Dunne, T., Figueiredo, R. de O., Victoria, R.L., 2006. Water storage and runoff processes in plinth ic soils under forest and pasture in eastern Amazonia. Hydrologi cal Processes 20, 2509-2526. Moreira, A.G., 2000. Effects of fire protection on savanna st ructure in Central Brazil. Journal of Biogeography 27, 1021-1029. Myers, N., 1988. Threatened biotas : Hot spots in tropical fo rests. The Environmentalist 8, 1. Myers, N., 1990. The biodiversity cha llenge: Expanded hot spots analysis. The Environmentalist 10, 243. Nagendra, H., Rocchini, D., 2008. High re solution satellite imagery for tropical biodiversity studies: the devil is in the det ail. Biodiversity and Conservation 17 (14), 3431-3442. 190

PAGE 191

Navulur, K., 2007. Multispectral Image Analysis Using the Object Oriented Paradigm. CRC Press, Boca Raton, Florida. 184 pp. Naiman, R.J., Dcamps, H., 1997. The ecol ogy of interfaces: riparian zones. Annual Review of Ecology and Systematics 28, 621. Neeff, T., Ascui, F., 2009. Lessons from carbon markets for designing an effective REDD architecture. Climate Policy 9 (3), 306-315. Neill C., Davidson, E.A., 2000. Soil carbon a ccumulation or loss following deforestation for pasture in the Brazilian Amazon. In Global climate change and tropical ecosystems (Eds. Lal, R. J.M. Kimble, B.A. Stewart), 197-211. CRC Press, Boca Raton. Neill, C., Deegan, L.A., Thom as, S.M., Cerri, C.C., 2001. Deforestation for pasture alters nitrogen and phosphorus in soil solution and streamwater of small Amazonian watersheds. Ecological Applications 11, 1817-1828. Neill, C., Deegan, L.A., Thom as, S.M., Haupert, C.L., Krusc he, A.V., Ballester, V.M., Victoria, RL., 2006. Deforestation alters channel hydraulic and biogeochemical characteristics of small lowland Amazonian streams. Hydrological Processes 20, 2563-2580. Neill C., Piccolo, M.C., Steudler, P.A., Melill o, J.M., Feigl, B.J., Cerri, C.C., 1995. Nitrogen dynamics in soils of forests and ac tive pastures in the Western Brazilian Amazon basin, Soil Biology and Biochemistry 27, 1167-1175. Nepstad, D., de Carvalho, C.R., Davidson, E ., Jipp, P., Lefebvre, P., Hees Negreiros, G., Silva, E., Stone, T., Trumbore, S., Vieira S., 1994. The role of deep roots in the hydrologic and carbon cycles of Amazonian forests and pastures. Nature 372, 666669. Nepstad, D.C., Alencar, A.C., McGrath, D.G., Moutinho, P., Barros, A.C. 2001. Road paving, fire regime feedbacks, and the future of Amazon forests. Forest Ecology and Management 154, 395. doi:10.1016/S03 78-1127(01)00511-4 Nepstad, D.C., Schwartzmann, S., Bamberger, B., Santilli, M., Ray, D., Schlesinger, P., Lefebvre, P., Alencar, A., Prinz, E., Fiske, G., Rolla, A., 2006. Inhibition of Amazon deforestation and fire by parks and indigenous lands. Conservation Biology 20 (1), 6573. Nepstad, D.C., Soares-Filho, B., Merry, F., Moutinho, P., Rodrigues, H.O., Bowman, M., Schwartzman, S., Almeida, O., River o, S., 2007a. The Costs and Benefits of Reducing Carbon Emissions from Defore station and Forest Degradation in the Brazilian Amazon. The Woods Hole Rese arch Center/Instituto de Pesquisa Ambiental da Amazon ia, Falmouth, MA. 191

PAGE 192

Nepstad, D.C., Carvalho Jr., O., Carter, J. Moita, A., Neu, V ., Cardinot, G., 2007b. Manejo e Recuperao de Mata Ciliar em Regies Florestais da Amaznia (Management and restoration of riparian zone forests of Amazon forest regions).Instituto de Pesquisa Ambiental da Amaznia (IPAM), Be lm, Brazil. 68 pp. Nepstad, D.C., Stickler, C.M., 2008. Managing the Tropical Agriculture Revolution Journal of Sustainable Fo restry 27 (1-2), 43 56. Nepstad, D.C., Stickler, C.M., Soares Filho, B.S., Merry, F., 2008. Ecological, economic, and climatic tipping points of an Amazon fore st dieback. Philosophical Transactions of the Royal Society B 363, 1737. Nepstad, D.C., Soares, B.S., Merry, F.D., Lima, A., Moutinho, P., Carter, J., Bowman, M.S., Cattaneo, A., Rodrigues, H., Schwartz man, S., McGrath, D.G., Stickler, C., Lubowski, R., Piris-Cabezas, P., Rivero, S ., Alencar, A., Stella, O., Almeida, O. 2009. The end of deforestation in the Brazilian Amazon. Science 326, 1350-1351. Nobre, C.A., Sellers, P.J., Shulka, J. 1991. Amazonian deforestation and regional climate change. Journal of Climatology 4, 957. Noss, R., 1990. Indicators for monitoring biodiversity A hierarchical approach. Conservation Biology 4, 355-364. Oliveira, P.J.C., Asner, G.P., Knapp, D.E ., Almeyda, A., Galvan-Gildemeister, R., Keene, S., Raybin, R.F., Smith, R.C ., 2007. Land-use allocation protects the Peruvian Amazon. Science 317 (5842), 1233-1236. Oliveira, P.S., Marquis, R.J., 2002. The Cerra dos of Brazil: Ecology and Natural History of a Neotropical Savanna. Columbia University Press, New York. Olson, D.M., Dinerstein, E., 1998. The Global 200: A representation approach to conserving the Earths most biologically valuable ecoregions. Conservation Biology 12, 502. Olson, D.M., Dinerstein, E., Wikramanayake, E.D., 2001. Terrestrial ecoregions of the world: a new map of life on Ea rth. Bioscience 51 (11), 933-938. Omernik, J.M., 1995. Ecoregi ons a framework for envir onmental management, In Biological assessment and crit eria-tools for water resource planning and decision making, (Eds. W.S. Davis, T.P. Simon) 49-62, Lewis Publishers, Boca Raton, Florida. Ostrom, E., 1990. Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press, Cambridge. 192

PAGE 193

Ostrom, E., Burger, J., Field, C.B., Norgaar d, R.B., Policansky, D., 1999. Revisiting the Commons: Local Lessons, Global C hallenges. Science 284, 278-282. Oyama, M.A., Nobre, C.A., 2003. A new climate-vegetation equilibrium state for Tropical South America Geophysical Re search Letters 30 (23), 2199. doi:10.1029/2003GL018600 Page, S.E., Siegert, F., Rieley, J.O., Boehm H.D.V., Jayak, A., Limink, S., 2002 The amount of carbon released fr om peat and forest fires in Indonesia during 1997. Nature 420, 61-65. Parrotta, J.A., Turnbull, J.W. 1997. Catalyzing native forest regeneration on degraded tropical lands. Forest Ecology and Management 99 (1-2), 1-7. Peres, C.A., Barlow, J., Haugaasen, T., 2003. Ve rtebrate responses to surface fires in a central Amazonian forest. Oryx 37, 97-109. Perz, S., Brilhante, S., Brow n, F., Caldas, M., Ikeda, S. Mendoza, E., Overdevest, C., Reis, V., Reyes, J.F., Roja s, D., Schmink, M., Souza, C., Walker, R., 2008. Road building, land use and climat e change: prospects for environmental governance in the Amazon. Philosophical Transactions of the Royal Society of London B. 363 (1498),1889-1895. doi: 10.1098/rstb.2007.0017 Peskett, L., Luttrell, C., Iwat a, M., 2007. Can standards fo r voluntary carbon offsets ensure development benefits? ODI Forestry Briefing 13, Overseas Development Institute, London. Accessible at http://www.odi.org.uk/resources/download/11.pdf (accessed on November 2, 2009) Peskett, L., Huberman, D., Bowen-Jones, E., Edwards, G., Brown, J., 2008. Making REDD work for the poor. London, Overseas Development Institute. Accessible at http://www.odi.org.uk/ccef/resources/ reports/s0179_redd-_final_report.pdf (accessed on November 2, 2009) Pickett, S.T.A., Parker, V.T., Fiedler, P.L. 1992. The new paradigm in ecology: implications for conservation biology abov e the species level. In Conservation Biology: The Theory and Practice of Na ture Conservation, Preservation, and Management (Eds. P.L. Fiedler, S. K. Ja in), 37-56, Chapman and Hall, New York. Pielke, R.A., Eastman, J., C hase, T. N., Knaff, J., Kitte l, T.G. F., 1998. The 1973-1996 trends in depth-averaged tropospheric tem perature, Journal of Geophysical Research 103, 16,927-16,933. Plumptre, A.J., 2001. The e ffects of habitat change due to selective logging on the fauna of forests in Africa, in African Rain Forest Ecology and Conservation (Eds. W. Webber, L.J.T. White, A. Vedder, L. Naughton-Treves), 463-479, Yale University Press, New Haven, CT. 193

PAGE 194

Poiani, K.A., Richter, B.D ., Anderson, M.G., Richter, H.E., 2000. Biodiversity conservation at multiple scales: f unctional sites, landscapes, and networks. Bioscience 50 (2), 133-146. Point Carbon, 2009. Carbon Market Monitor January 2009: a review of 2008. Point Carbon, Oslo. Accessible at http://www.pointcarbon.com/research/carbonmarketresearch/monitor/1.1034635 (accessed on November 2, 2009) Pringle, C.M., Hamazaki, T., 1997. Effects of fishes on algal response to storms in a tropical stream. Ecology 78, 2432. Pusey, B.J., Arthington, A.H. 2003. Importance of the ripari an zone to the conservation and management of freshwater fish: a review. Marine and Freshwater Research 54, 1. Putz, F.E., Sirot, L.K., Pinar d, M.A., 2001. Tropical forest management and wildlife: silvicultural effects on forest structure, fruit production, and locomotion of arboreal animals. In The Cutting Edge: Conserving Wildlife in Logged Tropical Forests (Eds. R.A. Fimbel, A. Grajal, J.G. Robinson), 11-34, Columbia University Press, New York. Putz, F.E., Sist, P., Fredericksen, T.S., Dykstra, D. 2008. Reduced-impact logging: Challenges and opportunities. Forest Ecology and Management 256, 1427-1433. Ray, D.G., Nepstad, D.C., M outinho, P., 2005. Micrometeorol ogical and canopy controls of fire susceptibility in mature and di sturbed forests of an east-central Amazon landscape. Ecological Appl ications 15 (5), 1664-1678. Rosenthal, E., 2009. In Brazil, Paying Farmers to Let the Trees Stand. New York Times, August 22. Saatchi, S.S., Houghton, R.A., Dos Santos Alvala, R.C., Soares, J.V., Yu, Y., 2007. Distribution of aboveground live biomass in the Amazon basin. Global Change Biology 13 (4), 816-837. Sahin, V., Hall, M.J., 1996. The effects of afforestation and deforestation on water yields. Journal of Hydrology 178, 293-309. Sampaio, G., Nobre, C., Cost a, M.H., Satyamurty P., Soares-Filho, B.S., Cardoso, M., 2007. Regional climate change over easte rn Amazonia caused by pasture and soybean cropland expansion. Geophysi cal Research Letters 34, L17709. doi:10.1029/2007GL030612 Sampson, G.P., 2000. Trade, Environment and the WTO: the Post-Seattle Agenda. Overseas Development Counc il, Washington, DC. 151 pp. 194

PAGE 195

Sanches, R.A., 2002. Desmatamento na regi o dos formadores do Rio Xingu, Mato Grosso, Brasil. ISA, S o Paulo. Sanchez-Azofeifa, G.A., Daily, G.C., Pfaff, A.S.P., Busch, C. 2003. Integrity and isolation of Costa Ricas national parks and biological reserves: examining the dynamics of land-cover change. Biol ogical Conservation 109, 123-135. Sanchez-Azofeifa, G.A., Pfaff, A., Robalino, J.A., Boomhower, J.P., 2007. Costa Rica's Payment for Environmental Services Pr ogram: Intention, Implementation, and Impact. Conservation Biology 21 (5),1165-73. Sanderson, E.W., Redford, K.H., Vedder, A., Coppolillo, P., Ward, S.E., 2002. A conceptual model for conservation planning based on landscape species requirements. Landscape and Ur ban Planning 58, 51-56. Santilli, M., Moutinho, P., Schwartzman, S., Nepstad, D.C., Curran, L., 2005 Tropical deforestation and the Kyoto Protocol: an editorial essay. Climate Change, 71, 267 76. Sawaya, K.E., Olmanson, L.G., Heinert, N.J., Brezonik, P.L., Bauer, M.E., 2003. Extending satellite remote sensing to local scales: land and water resource monitoring using high-resolution imagery. Re mote Sensing of Environment 88 (1-2), 144-156. SBI Reports, 2009. Carbon Emissions Tradi ng Markets Worldwide. SBI Reports, Rockville, MD. 186 pp. Accessible at http://www.sbireports.com/Carbon-EmissionsTrading-1926753/ (accessed on November 2, 2009) Scanlon, T.M., Caylor, K.K., Levin, S.A., R odriguez-Iturbe, I., 2007. Positive feedbacks promote power-law clustering of Kalahar i vegetation. Nature 449, 209-213. Schafer, J.S., Eck, T.F., Holben, B.N., Arta xo, P., Yamasoe, M.A., Procopio, A.S., 2002. Observed reductions of total solar irradi ance by biomassburning aerosols in the Brazilian Amazon and Zambian Savanna. Geophysical Research Letters, 29. Schafer, J.S., Holben, B.N., Eck, T.F., Yamasoe, M.A., Artaxo, P., 2002. Atmospheric effects on insolation in the Brazilian Am azon: Observed modification of solar radiation by clouds and smok e and derived single scattering al bedo of fire aerosols. Journal of Geophysical Research-Atmospheres, 107. Schwarzenbach, R.P., Escher, B.I., Fenner, K., Hofstetter, T.B., Johnson, C.A., von Gunten, U., Wehrli, B., 2006. The challenge of micropollu tants in aquatic systems. Science, 313, 1072 1077. 195

PAGE 196

Secretaria de Estado de Pl anejamento e Coordenao Geral de Mato Grosso (SEPLAN-MT), 2008. Projeto Zoneamento Socioeconm ico Ecolgico do Estado de Mato Grosso Cuiba, Mato Grosso. Accessible at http://www.geo.seplan.mt.gov.br/zsee/ (accessed on November 29, 2009) Soares-Filho, B.S., Cerqueira, G.C., Pennachin, C.L., 2002. DINAMICAa stochastic cellular automata model designed to si mulate the landscape dynamics in an Amazonian colonization frontier. Ecological Modelling 154, 217-235. Soares-Filho, B., Alencar, A., Nepstad, D. Cerqueira, G., Diaz, M.C.V., Rivero, S., Solorzano, L., Voll, E., 2004. Simulation of deforestation and forest regrowth along a major Amazon highway: the ca se of the Santarm-Cuia b highway. Global Change Biology 10, 745-764. Soares-Filho, B.,Nepstad, D., Curran, L.. Cer queira, G., Garcia, R., Ramos, C., Voll, E., McDonald, A., Lefebvre, P., Schlesinger, P., 2006. Modeling Amazon conservation. Nature 440, 520. Soares-Filho, B.S., Rodrigues, H.O., Co sta, W.L., 2009. Mode ling Environmental Dynamics with Dinamica EGO. Accessible at: http://www.csr.ufmg.br/dinamica (accessed on November 28, 2009) Sonzogni, W.C., Chesters, G., Coote, D.R., Jeffs, J.C., Robinson, J. B., 1980, Pollution from land runoff. Environmental Science and Technology 14, 148-153. Southgate, D., Whitaker, M., 1992. Promoting Resource Degr adation in Latin America: Tropical Deforestation, Soil Erosion, and Coastal Ecosystem Disturbance in Ecuador. Economic Development and Cultural Change 49, 787-807. Sterner, T., 2003. Policy instruments fo r environmental and natural resource management. RFF Press, Washington, DC. Stickler, C.M., Nepstad, D. C., Coe, M.T., Rodrigues, H .O., McGrath, D.G., Walker, W.S., Soares-Filho, B.S., Davidson, E.A. 2009. The potential ecological costs and co-benefits of REDD: a critical review and case study from the Amazon region. Global Change Biology 15, 2803-2824. Stocking, M.A., 2003. Tropical soils and food security: the next 50 years. Science 302, 1356. DOI: 10.1126/science.1088579 Sweeney, B.W., Bott, T.L., Jacks on, J.K., Kaplan, L.A., Newbol d, J.D., Standley, L. J., Hession, W.C., Horwitz, R.J., 2004. Riparian deforestation, stream narrowing, and loss of stream ecosystem services. Proceedings of the National Academy of Sciences 101, 14132. 196

PAGE 197

Tarnocai, C., Canadell, J.G., Schuur, E.A.G., Kuhry, P., Mazhitova, G., Zimov, S., 2009. Soil organic carbon pools in the northern circumpolar pe rmafrost region. Global Biogeochemical Cycles 23, GB2023; doi:10.1029/2008GB003327. Terborgh, J., 1999. Requiem for Nature. Island Press, Washington, DC, 426 pp. Thanapakpawin, P., Richey, J., Thomas, D ., Rodda, S., Campbell, B., Logsdon, M., 2007. Effects of landuse change on the hydrologic regime of the Mae Chaem river basin, NW Thailand. Journal of Hydrology 334, 215-230. Thiollay, J.M., 1992. Influence of selective logging on bird species diversity in a Guianan rain forest. Conservati on Biology 6, 47-63. Tietenberg, T., 1996. Pr ivate Enforcement of Environm ental Regulations in Latin America and the Caribbean: An Effectiv e Instrument for Environmental Management? Inter-American Development Bank, Washington, D.C., June. No. ENV-101. Accessible at http://www.iadb.org/sds/doc/env-101e.pdf (accessed on November 2, 2009) Tollefson, J., 2009. Paying to save t he rainforests. Nature 460, 936-937. Torgersen, C.E., Faux, R.N ., McIntosh, B.A., Poage, N.J ., Norton, D.J., 2001. Airborne thermal remote sensing for water temperature assessment in rivers and streams. Remote Sensing of Environment 76, 386-398. Tortajada, C., 2001. Instituti ons for Integrated River Basin Management in Latin America. Water Resources Development 17 (3), 289-301. Turner, M.G., 1989. Landscape ecology: the e ffect of pattern and process. Annual Review of Ecology and Systematics 20, 171-197. Van Kooten, G.C., Bulte, E.H., 2000. The Economics of Nature: Managing Biological Assets. Oxford : Blackwell Publishers. Vannote, R.L., Minshall, J.V., Cummins, K.W., Seddell, J.R., Cushing, C.E., 1980. The river continuum concept. Canadian. Journal of Fisheries and A quatic Science 37, 130-137. Vayda, A.P., 2006 Causal explan ation of Indonesia's forest fi res: concepts, applications, and research priorities. Human Ecology 34 (5), 615-635. Vera-Diaz, M.D.C., Kaufmann, R.K., Neps tad, D.C., Schlesinger, P., 2007. An interdisciplinary model of soybean yield in the Amazon Basin: the climatic, edaphic, and economic determinants. Ecol ogical Economics 13, 134-141. 197

PAGE 198

Vira, B., Adams, W.M., 2009. Ecosystem se rvices and conservation strategy: beware the silver bullet. Conservation Letters 49, 6-7. Visser, H., de Nijs, T., 2006. The map comparison kit. Environmental Modelling and Software 21, 346-358. Vogelmann, J.E., Sohl, T.L., Campbell, P.V., Shaw, D.M ., 1998. Regional land cover characterization using Landsat Thematic Mapper data and ancillary data sources. Environmental Monitoring and Assessment 51, 415-428. Vollenweider, R.A., 1971. The scientific f undamentals of lake and stream eutrophication with particular reference to phosphorus and nitrogen as eutrophication factors. OECD, Paris. Walker, W.S., Stickler, C.M., Kellndorfer, J.M., Kirsch, K., Neps tad, D.C. Cloud-free mapping of tropical forests with satellite rada r. Nature Geoscience, in preparation. Walker, W.S., Kellndorfer, J.M. Lapoint, E., Hoppus, M., Westfall, J., 2007. An empirical InSAR-optical fusion approach to mapping vegetation canopy height. Remote Sensing of Environment 109, 482-499. Watchman, P., Delfino, A., Addison, J., 2007. EP 2: The revised Equator Principles: Why hard-nosed bankers are embracing so ft law principles. Law and Financial Markets Review 1 (2), 79-90. Weiss, T.G., 2000. Governance, Good Governance and Global Governance: Conceptual and Actual Challenges. Th ird World Quarterly 21 (5), 795-814. Werth, D., Avissar, R., 2002. The local and global effects of Amazon deforestation. The Journal of Geophysical Research, 107 (D20), 8087. Werth, D., Avissar, R., 2005. The loca l and global effects of Southeast Asian deforestation. Geophysical Research Letters, 2005GL022970 Werth, D., Avissar, R., 2005. The local an d global effects of African deforestation. Geophysical Research Letters, 12704, 2005GL022969 Wilcox, B.A., 1984. In situ conservation of genetic resources: Determinants of minimum area requirements. In Nati onal Parks, Conservation and Development, Proceedings of the World Congress on National Parks (E ds. JA McNeely, KR Miller), 18-30, Smithsonian Institution Pr ess, Washington, D.C. Williams, M.R., Melack, J.M., 1997. Solute ex port from forested and partially deforested catchments in the central Amazon. Biogeochemistry 38, 67-102. 198

PAGE 199

Williams, E., Rosenfeld, D., Madden, N., Gerlach, J., Gears, N., Atkinson, L., Dunnemann, N., Frostrom, G., Antonio, M., Biazon, B., Camargo, R., Franca, H., Gomes, A., Lima, M., Machado, R., Manhaes, S. Nachtigall, L., Piva, H., Quintiliano, W., Machado, L., Artaxo, P., Roberts, G., Renno, N., Blakeslee, R., Bailey, J., Boccippio, D., Betts, A., Wol ff, D., Roy, B., Halverson, J., Rickenbach, T., Fuentes, J., Avelino, E., 2002. Contrasting convective regimes over the Amazon: Implications for cloud electrification. Journal of Geophysical Research-Atmospheres, 107. Winterbottom, R. 1990. Taking stock: the Tropica l Forestry Action Plan after five years. World Resources Institute (W RI), Washington, DC, 59 pp. http://db.jhuccp.org/icswpd/exec/icswppro.dll?BU =http://db.jhuccp.org/icswpd/exec/icswppro.dll&QF0=DocNo& QI0=075491&TN=Popline&AC=QBE_QUERY &MR=30%25DL=1&&RL=1&&RF=LongRecordDisplay&DF=LongRecordDisplay (accessed on November 2, 2009) Woodroffe, R., Ginsberg, J.R. 1998. Edge effects and the extincition of populations inside protected areas. Science 280, 2126-2128. Wunder, S. 2006. Are direct pay ments for environmental services spelling doom for sustainable forest management in the tropics? Ecology and Society 11 (2), 23. URL: http://www.ecologyandsociet y.org/vol11/iss2/art23/ Wunder, S., 2007. The efficiency of payments fo r environmental services in tropical conservation. Conservation Biology 21 (1), 48-58. Y ikatu Xingu. 2009. Campaign to sa ve the Xingu River. Accessible at http://www.yikatuxingu.org.br/home (accessed on November 2, 2009) Zakia, M.J., 2005. A silvicultura em reas declivosas e o topo de morro. Ministrio do Meio Ambiente, Braslia, DF. Zarin, D.J., Ducey, M.J., Tucker, J.M. Salas, W.A., 2001. Potential biomass accumulation in Amazonian regrowth forests. Ecosystems 4(7), 658-668. Zarin, D. J., Davidson, E.A., Brondizio, E., Vieira, I.C.G., S, T., Feldpausch, T., Schuur, E.A.G., Mesquita, R., Moran, E., Delamonica, P., Ducey, M.J., Hurtt, G.C., Salimon, C., Denich, M., 2005. Legacy of fire slows carbon accumula tion in Amazonian forest regrowth. Frontiers in Ecology and the Environment. 3, 365-369. 199

PAGE 200

BIOGRAPHICAL SKETCH Claudia Stickler received her B.S. degree in biology and international political economy from the University of Puget Sound Tacoma, Washington in May 1996. She spent three years in Cameroon, Central Africa, first as an agroforestry extension agent with the Peace Corps and then as a resear ch assistant on a San Francisco State University project focusing on tropical fore st regeneration and plant-animal interactions. In 2000, she worked in Suriname, South Amer ica, as a research assistant on a longterm University of Florida study of primat e behavior and ecology. She received her M.S. degree from the College of Natural Resources and Environment, University of Florida, in 2004. Her M.S. research focused on the ef fect of logging on habitat selection by primates in Kibale National Park in Uganda. She is affiliated wit h the Department of Geography, the Tropical Conservation and Development Program, and the Land Use and Environmental Change Institut e at the University of Fl orida. Since 2004, she has been working as a Visiting Scholar with the Amazon Environmental Research Institute ( Instituto de Pesquisa Ambiental da Amaznia ) in Brazil and as a Graduate Fellow with the Woods Hole Research Center in Falmouth, Massachusetts. 200