|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 LAND USE ALLOCATION UNDER MULTIPLE OBJECTIVES IN THE BRAZILIAN AMAZON: CREATING A TOOL TO GUIDE ZONING OF PUBLIC PRODUCTION FORESTS By MARCO AURELIO WATANABE LENTINI A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007
2 2007 Marco Aurelio Watanabe Lentini
3 To my parents.
4 ACKNOWLEDGMENTS I express my sincere gratitude to Douglas R. Carter, my advisor, for his support and academic guidance. I am also grateful to my committee members, Daniel Zarin, Donna Lee and Janaki Alavalapati, for their a cademic advice and support. Jack Putz, Christopher Baraloto and Stephen Perz also provided important feedback on my thesis. I acknowledge a few colleagues and friends at University of Florida, Alexander Macpherson, Onil Banerjee, and Jamie Cotta, who reviewed earlier draf ts of this thesis, and provided se veral interesting comments on these drafts. I thank the research staff at the Amazon In stitute for the People and the Environment (IMAZON), for the institutional support they provided during my Masters studies at the University of Florida. Some data, as well as support in several methodological steps of this study, were provided by Rodney Sa lomo, Denys Pereira, Mark Schulze, Carlos Souza Jr, Eugenio Arima, Brenda Brito and Mrcio Sales. I am deeply grateful to Edson Vidal, currently a professor at the School of Agriculture Luiz de Queiroz, at the University of So Paulo (ESALQ/USP), for being an inte llectual mentor and for supporting the beginning of my career in the Brazilian Amazon. I thank Adalberto Verss imo, my supervisor at IMAZON for the last 7 years, for the opportunity to work on several in teresting research proj ects during this period. My colleagues at the forest resource ec onomics lab, Matthew Langholtz and Maitreyi Mandal, provided advice on computational issues that were useful in this study. Other UF colleagues offered a number of comments and su ggestions during the exec ution of this work, including Ane Alencar, Ana Eleuterio, and Clau dia Romero. The discussion about temporal forest zoning included in this thesis was bor n in a discussion with A. Alencar. Forestry professionals Ilana Gorayeb (Association of Certified Forest Producers in the Brazilian Amazon), Mauricio Voivodic (Inst itute for Management and Certification in Agriculture and
5 Forestry) and Leonardo Sobral (Cik el Brasil Verde) also provided valuable information for this work. Finally, I thank the Amazon Conservation L eadership Initiative (ACLI), the Tropical Conservation Development Program (TCD) and the School of Forest Resources and Conservation (SFRC) fo r providing financial and administra tive support during my M.Sc. studies at UF. I am immensely gratef ul to Robert Buschbacher, Hann ah Covert, Patrcia Sampaio, Victoria Gomez de la Torre, Wanda Carter and Cherie Arias for th eir generous and indispensable academic and administrative guidance throughout my Masters program.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES................................................................................................................ .........9 LIST OF ABBREVIATIONS........................................................................................................10 ABSTRACT....................................................................................................................... ............12 CHAPTER 1 INTRODUCTION..................................................................................................................14 Statement of the Problem....................................................................................................... .14 Study Objectives............................................................................................................... ......15 2 STUDY RATIONALE AND LITERATURE REVIEW.......................................................17 History and Socio-Economic Importance of the Logging Sector in the Amazon..................18 Structural Problems Related to Logging in the Brazilian Amazon........................................21 The Boom-Bust Ec onomic Pattern..................................................................................21 Land Tenure.................................................................................................................... .22 Illegal Logging................................................................................................................22 What is Needed to Address these Problems?..................................................................23 The Management of Public Forests Law................................................................................24 Literature Review on Spatial Modeling..................................................................................26 3 METHODS........................................................................................................................ .....34 The Calha Norte Region and Faro State Forest......................................................................35 The Spatially-Explicit Net Returns to Logging Model...........................................................36 The Land Use Allocation Model............................................................................................39 Building the Optimization Model...........................................................................................43 Simulations Executed........................................................................................................... ..47 Benefit-Cost Analyses (BCA) and Sensitivity Analyses........................................................49 The PPF Between Logging and Other Land Uses..................................................................51 4 RESULTS........................................................................................................................ .......58 Estimated Net Returns from Logging in Calha Norte............................................................58 Estimated Timber Supply within Faro State Forest................................................................59 Milling Capacity for Roundwood from Public Forests in Calha Norte..................................60 The Timber Supply Problem..................................................................................................60
7 Landscape Patterns Formed by the Optimization Model.......................................................62 Annual Profits from Logging and Government Revenues.....................................................64 Logging NPVs and IRRs under Increasing C onversion to Alternative Land Uses................65 The Production Possibility Frontier........................................................................................66 Opportunity Costs for Alternative Land Uses........................................................................67 5 DISCUSSION AND FINAL REMARKS..............................................................................82 Importance of Land Use Modeling in the Larg e-Scale Planning of Public Forest Use.........82 Model Assumptions.............................................................................................................. ..84 Model Limitations.............................................................................................................. ....86 Public Forests Sustainability and the Spatial and Temporal Zoning......................................89 Further Research............................................................................................................... ......90 APPENDIX: HOW CHANGES IN AN INPUT PRICE CAN AFFECT SUPPLY IN A PARTIAL EQUILIBRIUM MODEL: A MATHEMATICAL DEMONSTRATION...........91 LIST OF REFERENCES............................................................................................................. ..94 BIOGRAPHICAL SKETCH.......................................................................................................102
8 LIST OF TABLES Table page 3-1. Sources and year of the analyses of the datasets used in this study...................................54 3-2. Costs for roundwood transportation and fricti on coefficients used in the model of net economic returns for logging.............................................................................................55 3-3. Sawn-wood prices and typical costs of the logging activity in the logging centers of the Calha Norte region, 2004.............................................................................................56 4-1. Net economic returns from l ogging in the Calha Norte, 2004...........................................70 4-2. Net returns from logging in Faro State Forest, 2004.........................................................70 4-3. Total timber volume and estimated roundwood supply (in m3 ha-1) of merchantable species by timber value class in the sta nds within Faro Stat e Forest in 1976...................70 4-4. Milling capacity by logging cen ter within Calha Norte, 2004...........................................71 4-5. NPVs and IRRs for logging within Faro State Forest from the logging centers under decreasing number of stands used for concessions............................................................79 4-6. NPVs and IRRs for logging within Faro St ate Forest from closer urban centers under decreasing number of stands used for concessions............................................................79
9 LIST OF FIGURES Figure page 2-1. Logging centers and logging fron tiers in the Brazilian Amazon, 2004.............................31 2-2. Markets for the wood production from the Brazilian Amazon in 1998 and 2004.............32 2-3. Value of wood products exported from the Brazili an Amazon between 1998 and 2006........................................................................................................................... .........32 2-4. Destination of government revenu es generated by logging concessions...........................33 3-1. View of the Calha No rte region, northern Amazon...........................................................52 3-2. Spatially-explicit and economic data used to build the net economic returns and the land use optimization model..............................................................................................53 3-3. Potential for land use altern atives in Faro State Forest......................................................57 4-1. Net economic returns for logging in Calha Norte and within FSF, 2004..........................69 4-2. Results from the unconstrained logging scenario..............................................................72 4-3. Variation in the number of stands and volume logged within Faro State Forest under increasing number of stands converted to other land uses.................................................73 4-4. Available merchantable timber harves ted, and proportion of the regional milling capacity met by FSF, under increasing convers ion of stands to other land uses...............74 4-5. Landscape patterns formed by the optim ization model for increasing number of stands converted to alternative land uses...........................................................................75 4-6. Landscape patterns formed by the final opt imization model for increasing cumulative scores for livelihood systems and biodiversity conservation.............................................76 4-7. Annual profits from logging in Faro for government and loggers under increasing stand conversion to alternative land uses...........................................................................77 4-8. Spatial distribution of annual profit s from logging in the unconstrained logging scenario, with harvests performed from l ogging centers, and clos er urban centers..........78 4-9. Production possibility frontiers (PPF) for competing land uses within FSF.....................80 4-10. Marginal and average opportunity co sts for livelihood systems and biodiversity conservation in FSF...........................................................................................................81
10 LIST OF ABBREVIATIONS ACLI: Amazon Conservation Leadership Initiative BCA: Benefit-cost analyses BNDES: National Bank for Econo mic and Social Development CGFP: Commission for Manage ment of Public Forests DBH: Diameter at the Breast Height (1.3 m) DETER: Real-time detection system (det ection of deforestation in Amazon) DICOPT: Discrete and Continuous Optimizer EMBRAPA: Brazilian Agricultural Research Corporation ESALQ/USP: School of Agricultural Studies L uiz de Queiroz at University of So Paulo FAO: Food and Agriculture Orga nization of the United Nations FFT: Tropical Forest Foundation FLONA: National Forest FMP: Forest Management Plan FNDF: National Forestry Development Fund FSF: Faro State Forest GAMS: General Algebraic Modeling System GDP: Gross Domestic Product GIS: Geographic Information System IBAMA: Brazilian Institute for the Environment and the Renewable Natural Resources IBGE: Brazilian Institute of Geography and Statistics INPE: Brazilian Institute of Spatial Research IP: Integer programming IPEA: Institute for Applied Economic Research
11 IMAFLORA: Institute for Management and Ce rtification in Agriculture and Forestry IMAZON: Amazon Institute for the People and the Environment IRR: Internal rate of return ISA: Socio-environmental Institute LP: Linear programming MDIC: Brazilian Ministry of Developmen t, Industry and International Commerce MINLP: Mixed integer non-linear programming MMA: Brazilian Ministry of Environment mOC: Marginal opportunity cost MODIS: Moderate Resolution Imaging Spectroradiometer NGO: Non-governmental organization NPV: Net Present Value NTFP: Non-timber forest products PFCA: Association of Certified Forest Producers in the Brazilian Amazon PPF: Production possibility frontier RIL: Reduced impact logging techniques SBS: Brazilian Silviculture Society SFB: Brazilian Forest Service SFRC: School of Forest Resources and Conservation SIVAM: Amazonian Vigilance System TCD: Tropical Conservation and Development Program UF: University of Florida UL: Unconstrained logging scenario WFI: Wide Field Imager
12 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science LAND USE ALLOCATION UNDER MULTIPLE OBJECTIVES IN THE BRAZILIAN AMAZON: CREATING A TOOL TO GUIDE ZONING OF PUBLIC PRODUCTION FORESTS By Marco Aurelio Watanabe Lentini August 2007 Chair: Douglas R. Carter Major: Forest Resources and Conservation Logging in natural forests is a vital income generating activity for the Amazons fragile economy. However, illegal and unplanned logg ing, largely executed in public lands, is exhausting forests rapidly. In March 2006, a new forestry law (Lei 11,284 /2006) established the first directives to manage public lands, includ ing the multiple use of these forests through timber concessions, biodiversity c onservation, tourism, mining and other land uses demanded by society. From a social planners perspective, a new challenge lies in how to maximize the welfare that can be generated from these lands, using them in an efficient way while satisfying dynamic societal preferences. This study seeks to address the question, from a social planners perspective in the Brazilian Amazon, of how to optimize the alloca tion of forest units among multiple land use alternatives, satisfying criteri a that are often conflicting such as forest production and conservation objectives and assessing the tradeoffs associated with these different choices. To address this question, I developed an optimizatio n model of land use allocation in an Amazonian state forest, taking into account economic information about th e expected profits for logging within the state forest. The model developed is able to solve the timber supply problem for the
13 logging industry, considering the supply of r oundwood in public lands and capacity constraints of surrounding mills, and is also able to determ ine maximum profits from logging subject to a minimum area or score that must be achieved for alternative non-commodity uses. This research represents more than a single case study, as the datasets used in this study cover Amazons geographic extent and, therefore, results could be applied to a ny other Amazonian public forest. The model was then used to estimate Ne t Present Values for logging under different zoning configurations. In Faro, the State Forest us ed as a case study in this thesis, the NPV from logging concessions in a baseline scenario co uld achieve US$ 16.8 million. This value would naturally decline if a larger pr oportion of the public forest were allocated to other land uses. Next, simulation results and sensitivity anal yses were used to determine the production possibility frontier between the production of market and non-market goods. Using this method, opportunity costs were estimated for land uses related to the produc tion of non-market goods, such as biodiversity conser vation or livelihood systems prom oted by forest dwellers. The main result of this study is the development of a useful tool to aid social planners in the zoning of public forests in the Brazilian Amaz on. Simulations can be used to determine, for each area, efficient zoning alternatives for pr oducing market and non-market goods. In addition, this study revealed that marginal opportunity co sts increase when a larg er share of the public forest is assigned for alternative land uses, a f act that should be taken into account by planners during public forest zoning. In the case of Faro State Forest, for example, I found that marginal opportunity costs for biodiversity conserva tion begin around US$ 10,000 and can increase to US$ 170,000 when 43% of the public forest is assigne d for this land use. Such results are then discussed in relation to their policy implicati ons and, finally, future areas for research are presented.
14 CHAPTER 1 INTRODUCTION Statement of the Problem Selective logging in natural forests is curren tly the third most important economic activity in the Brazilian Amazon, after industrial mi ning and cattle ranching (IPEA, 2002; IBGE, 2004). However, since the beginning of its history, la rge scale timber loggi ng in upland Amazonian forests has been extensive, predatory and migrator y; and has been depleti ng forests rapidly, often in public lands (Uhl, 1997; Nepstad et al. 1999 ; Schneider et al., 2000; Asner et al., 2005). While a large proportion of the Amazonian timber production is still ille gally generated today (Lentini et al., 2005), a new forestry law, enacted in 2006, established the first directives to take control of and manage public lands, including forest concessions. Thus, this new law represents an opportunity to limit illegal activities and land speculation in public forests. Unsettled public lands cove r one third of the Brazilian Amazon, corresponding to 160 million hectares (IBGE, 1996). After the enactment of the new law, such lands can finally be used to generate economic deve lopment through forest concessions tourism, or mining; or to accommodate other land uses demanded by societ y, such as livelihood strategies promoted by traditional communities or biological conservation. From a social planners perspective, a new challenge lies in how to maximize the welfare that can be generated from these lands, using them in an efficient way while meeting dynamic societal preferences. This implie s that social planners must take into account the interest s of stakeholders involved in th e zoning of these forests, as well as the economic forces acting at multiple scales. The foundation of this problem is related to the question of how to optimize the allocation of forests among multiple land use alternatives, satisfying criteria that are often conflicting, such as timber production and conservation. This issue is important within th e current context of
15 Amazonian public lands, due to the scarcity of economic information to guide decisions regarding the zoning of public forests, includ ing estimates of the monetary value that government could raise through forest concessions. Study Objectives The main justification of this study is the need for the development of an analytical tool to guide the decision-making process in the zoning of Amazonian public forests. This tool could be used by social planners1 to assess efficient alternatives in the management of public lands and to estimate the tradeoffs among zoning alternativ es. This study has the following specific objectives: Develop a spatially-explicit optimization mode l to estimate the net returns from logging by taking into account multiple obj ectives in public forests; Apply this model to a specific public produc tion forest in the Brazilian Amazon; Determine the production possibility frontie r between the production of market and nonmarket goods, identifying efficient alternatives for land use allocation in the public forest; Assess the tradeoffs among land use altern atives by estimating the opportunity costs involved in the production of non-market goods within the public forest. The importance of this study is demonstrated by two facts: first, for the first time, a model is designed to solve the typical timber-s upply problem and the land allocation problem2 1 Social planners can be defined as bureaucrats and decision makers working for state and federal agencies that will be responsible for coordinating th e planning and the execution of Forest Management Plans (FMP) for public forests. Arguably, in some ca ses, such social planners can also be represented by members of supervisory committees, which are consulti ng bodies to be formed for each public forest to provide advice about management decisions. 2 Timber supply is commonly investigated in harvest scheduling and forest tactical planning problems in which the management of a given forest unit is op timized considering its supply of timber, capacity constraints of the surrounding mills, and other ph ysical constraints in the timber production. Furthermore, land allocation problems are concerned with finding the minimum number of units or the minimum cost to achieve a given score for an attribut e, such as biological conservation, or to achieve a minimum number of conserved species, among others. These problems are now capable of taking into account the integration of productio n and non-commodity uses. Both types of problems in natural resource management modeling will be reviewed in the next chapter.
16 simultaneously. Second, despite th e fact that the optimization m odel represents a case study in the sense that it was applied for a specific Amazonian public land, Faro State Forest, the databases used in this study cover the exte nt of the Brazilian Amazon, and could be used immediately for the zoning of any other Amazoni an public forest. Both issues will become evident in the next chapters.
17 CHAPTER 2 STUDY RATIONALE AND LI TERATURE REVIEW According to the Global Forest Resources Assessment 2005 (FAO, 2005), Brazil has, today, the second largest forested area in the world (478 million hectares), but it also has the worlds highest deforestation rate, equal to 3.1 million hectares per year. Most of this deforestation is concentrated in the Brazilian Amazon, where forests are still being converted to agricultural use. A large part of this deforestation is view ed, by Brazilian society, from a negative viewpoint, considering th e importance of such forests in conserving biological diversity and forest environmental services, watershed co nservation, the provision of livelihoods to local peoples, and global climate regulation. Further, it has been shown in the past couple of decades that some specific techniques can be used to conserve the forest st ructure during logging, mitigating the environmental impacts and improving the economic efficiency of harvesting operations (Barreto et al., 1998; Holmes et al., 2000). The Brazilian Amazon encompasses 9 states (~ 5 million km2) and covers roughly 60% of Brazils total area (IBGE, 1997). The region is st ill sparsely populated by 21 million inhabitants, concentrated mainly (70%) in urban areas. The Amazonian economy is strongly based on rural activities and contributed only 7% of Brazilian gross domestic product (GDP) in 2002 (IPEA, 2002). This is surprising considering the large av ailability of natural re sources, with 300 million hectares of forests (IBGE, 1997) having potenti al for timber and non-timber forest product (NTFP) production. On the other hand, historica lly Amazonian natural capital has not been efficiently used to generate social capital and economic development. While deforestation continues in the Brazilian Amazon, new mechanisms to promote the rational use of its forests are needed, including a comprehensive strategy to manage public lands.
18 This chapter is organized as follows. Firs t, I will provide some historical background about logging in the Brazilian Amazon and some data about its socio-economic importance. Second, the main structural problems related to the logging sector will be discussed. Third, justifications for the importanc e of managing public lands in th e region and the new Brazilian law focusing on this issue will be presented. The final section will discuss past literature on different types of spatial modeli ng in natural resource management. History and Socio-Economic Importance of the Logging Sector in the Amazon Selective logging has occurred in th e Amazonian estuary since the 17th century (Rankin, 1985). For the first two centuries, logging was re stricted to estuarine forests along the main Amazonian rivers, with high value species harv ested for European markets (Barros and Uhl, 1995; Zarin et al., 2001). It was only in the 1950s that the first industrial mills were established in the estuary to produce sawn-wood and veneer using foreign capital, w ith markets oriented toward exportation (Barros and Uhl, 1995; Pinedo-Vasquez et al., 2001). In the 1960s and 1970s, intensive government investments opened access to extensive portions of inland upland forests mainly through the construction of roads. Such investments enabled the development of extensive and highly predatory logging in upland forests, fueled by government subsidies for individu als to inhabit the region, the de pletion of hardwood species in southern Brazil, and the large av ailability of unclaimed lands in the Amazon (Uhl et al., 1997; Stone, 1998a; Verssimo et al., 1998; Verssimo et al., 2002b). The first areas explored during the road c onstruction period supplied some of the most important Amazonian logging centers scattered in these older logging frontiers, such as Paragominas and Sinop (Lentini et al., 2004) (Figure 21). Timber stocks in older frontiers have been largely depleted after 3 decades of unpl anned and predatory logging, in addition to conversion of forests to agricu ltural use. Timber companies in these frontiers have been
19 gradually deciding whether to cease operations, to invest in technological improvements, or to move to newer frontiers (Stone, 1997). Some official roads, such as the Cuiab -Santarm and the Transamazon, allowed the migration of timber firms from older to newer fron tiers. Such roads were also constructed during the late 1960s to early 1970s, as a purposeful strategy of the military Brazilian government to secure control over lands through human occupation. However, despite the fact that forests in the proximity of the Transamazon have been ha rvested for more than 30 years, the relative inaccessibility to this region during the rai ny season and the control exercised over local resources by smallholders who i nhabit the region partially saved th ese forests from the same fate of forests in older frontiers (Lima and Merry, 2003; Merry et al., in press). In 2004, 26 logging centers were located within intermediate frontie rs (Figure 2-1), representing of the timber production, revenues and jobs generated by the Amazonian logging industry. Then, as a consequence of timber firm migration, large av ailability of high value timber, and forest clearings for more intensive cattle ranching and agriculture, timber production within new frontiers, such as Novo Progresso and Castelo de Sonho (Par State), increased by more than 80% in the past 8 years (Lentin i et al., 2005) (Figure 2-1). Logging is, today, one of the three most impor tant economic activities in the Brazilian Amazon, after industrial mining and cattle ra nching (IPEA, 2002; IBGE, 2004). Logging also tends to rapidly develop inland ci ties since timber firms tend to be concentrated in urban logging centers, because of the availability of infra-st ructure (mainly roads), commerce and specialized services, electrical power, and la rge availability of labor (Sto ne, 1997; Uhl et al., 1997; Stone, 1998a; Verssimo et al., 2002b; Lentini et al., 2004) In 2004, the 82 logging centers established in the Brazilian Amazon encompassed 3,132 tim ber mills that consumed 24.5 million m3 of
20 roundwood, generating a gross revenue of US$ 2.3 b illion (Lentini et al., 2005). Logging is also an important activity for employment genera tion, representing 380,000 j obs (Lentini et al., 2005). Markets for Amazonian Wood. From the demand side, the exhaustion of the timber stocks from natural forests and the increasing restricti ons on harvesting forests in southern Brazil stimulated increased timber production in the Amazon. Most of this production is used to supply domestic markets with cheap civ il construction materials. In 2002, two-thirds of the Amazonian wood consumed in the state of So Paulo the primary consumer of Amazonian wood products in the world (Figure 2-2) was destined for low value added uses in civil construction, such as structures for roofing or forms fo r concrete structures (Sobral et al., 2002). The low value added to Amazonian wood products contributes to th e low interest from domestic markets in purchasing products generated through sound forest practices, such as forest certification (Verssimo et al., 2005). However, Amazonian participation in internat ional tropical timber markets is likely to increase in the future due to the exhaustion of natural timber stocks in Malaysia and Indonesia (Uhl et al., 1997; Vincent, 1997) In fact, the proportion of exported wood from the Brazilian Amazon increased from 14% of the total production in 1998, or 1.6 million m3 to 36% in 2004, or 3.7 million m3 (Verssimo and Smeraldi, 1999; Len tini et al., 2004, 2005) (Figure 2-2). Further, not only the quantity, bu t also the quality, of the Am azonian exported wood products has recently changed. According to the Brazi lian Ministry of Overseas Commerce (MDIC, 2007), the value of wood products exported from the Amazon increased from US$ 381 million in 1998 to US$ 1 billion in 2006, mainly due to the in crease in the participation of finished wood
21 products such as furniture, flooring or othe r value added wood parts in the Amazons wood exports (Figure 2-3). Structural Problems Related to Logging in the Brazilian Amazon Unfortunately, a significant share of Amazonian timber production comes from illegal sources. The poor adoption of sound forest mana gement practices by timber companies (Silva, 1997; Sabogal et al., 2006), and the easy access to public lands, stimulate a migratory and extensive pattern in th e logging industry and along with th e continuous conversion of exploited forests to low productivity agricult ural land. Timber companies continue to move to new forest frontiers while old ones are depl eted by logging and forest fires (Nepstad et al., 1999; Asner et al., 2005). Several factors are asso ciated with these problems. Human resources to control and monitor logging and to train forest workers to ap ply best practices are sc arce. Moreover, while public lands are abundant, forest regulations and inefficient enforcement systems encourage illegal logging. The Boom-Bust Economic Pattern There is a predominant pattern early on with respect to development of rural local economies dependent on selective logging. Often th ere is exponential grow th in the generation of revenues and jobs locally, as a result of harvesting of valuable timber and the establishment of sawmills (Schneider et al., 2000). Due to unsustainable logging practices, after a couple of decades forests become impoverished and are more susceptible to fires (Nepstad et al., 1999). Forest conversion to agricultural commodities su ch as cattle ranching and soybean production is often the next step for these lands. However, the climatic conditions in 45% of the Amazon (~ 225 million hectares) are not appropriate for establishing large scale agricult ure (Schneider et al., 2000). Without huge investments in technology, such as the development of new agricu ltural varieties more adapted to
22 the local climatic conditions a nd technical assistance in these regions, the social and economic negative impacts of the local forest resources exhaus tion are significant, provoking unemployment and an intensifica tion of poverty. The local logging sector eventually collapses or migrates to new regions due to the lack of raw material. Land Tenure At least 1/3 of the Brazilian Amazon (~ 160 million hectares) is represented by unsettled lands, designated terras devolutas that are neither completely i nventoried nor demarcated by the government (IBGSE, 1996; Lentini et al., 2004). These lands have a large potential for timber and NTFP production. A study conducted by Verssimo et al. (2000) estimated that there were 114 million hectares of forests (23% of the Brazilian Amazon) with potential to create production forests in that year. However, forest resources on these lands, frequently occupied by communities, ranchers or loggers, are jeopard ized by illegal logging and other unregulated activities (Barreto et al. 2006). At the same time, current Brazil ian regulations create perverse incentives for illegal activities on public lands because they fo rbid the licensing of forest management operations without definitive land us e rights. Without interventions, public lands will continue to be depleted by illegal activ ities and to be converted to extensive low productivity land uses, such as cattle ranching. Illegal Logging At least 40% of the Amazonian roundwood produc tion was harvested i llegally in recent years (Lentini et al., 2004, 2005; IBAMA, unpu blished). Less than 5% of the roundwood harvested in the Brazilian Amazon follows good di rectives for forest management and only 3% or less of this production comes from certi fied operations (Ilana Gorayeb, personal communication, September 1st, 2006). The reasons for illegal logging include factors on the demand side, such as the low interest in paym ent for sound environmental practices. On the
23 supply side, as discussed earlier the large availabili ty of non-monitored public lands is a disincentive for the adopti on of good forest practices. At the same time, there are several deficien cies in law enforcement and monitoring of illegal logging. First, enforcemen t agencies are fragile and ine fficient, allowing corruption of public agents. There is also a lack of resour ces to adopt modern technology systems in the monitoring of logging, such as remote sensing te chniques. Third, logger s typically argue that regulations to license forest management plans are excessive (Sabogal et al., 2006). Finally, there is a low probability that loggers fined for irregular pr actices will actually pay their environmental debts. A 2003 case study in State of Par showed that only 2% of the legal proceedings related to environmental crimes we re carried out until their conclusion (Brito and Barreto, 2005). What is Needed to A ddress these Problems? Brazilian society has recently taken several measures to reduce illegal logging. Hundreds of forest management plans were cancelled by the Federal Environmental Agency IBAMA in 2003-4 in an attempt to halt illeg al logging and decrease deforest ation rates in the Brazilian Amazon. In 2005, several governmental organizations conducted three large scale investigations into illegal behavior and corrupti on in the forest sector. Highly corruptible systems, such as the old system for authorizing roundwood transporta tion, which was based on paper documents, are gradually being replaced by electronic licensing sy stems. Specific regulations for forest management, often complicated enough to prev ent the adoption of re duced impact logging (RIL), were simplified, aiming to decrease tran saction costs for managed timber. Finally, electronic systems monitoring de forestation in the Brazilian Amazon are being improved to generate reports more frequently, such as the DETER (Real-Time Detection System), using
24 sensors such as the MODIS (Moderate Resolu tion Imaging Spectroradiometer) and the WFI (Wide Field Imager) (INPE, 2007). A second set of measures deals with problems related to land tenure. Federal and State governments have been pursuing regional pla nning and land rights regu larization. In 2006, a new law was enacted to provide legal instrument s to control and manage public lands in the Brazilian Amazon, including forest concessions. Concessions could help with land tenure problems by providing larger geographic and economic stability for timber companies, stabilizing the logging frontiers. They could also decrease conflicts between loggers and traditional communities inhabiting public lands, and decrease deforestation and the availability of illegal timber in public forests. Thus, conces sions might deliberately stimulate the adoption of sound forest management practices and forest certification, sin ce the main challenge to the adoption of these mechanisms, the chaotic land tenur e, will be mitigated (Schneider et al., 2000; Verssimo et al., 2000, Verssimo et al ., 2002a; Amaral and Amaral Neto, 2005). The Management of Public Forests Law The Brazilian federal law 11,284/2006 was enac ted in March 2006, generated from an intense debate between several se ctors within the society during the last decade. The new law creates rules to manage public forests, includi ng national and state fore sts already created and other public lands still not inventoried by the government. It also creates a Federal Forestry Fund (FNDF) to foment forest based activities in Brazil, and the Brazilian Forest Service (SFB), a federal agency focused on controlling the fund a nd supervising forest concession contracts. According to the Law, public forests can be de stined for three uses: creation of conservation units, as national and state forests; allocated as forest concessions; and for the direct use of traditional communities dwelling in such areas.
25 The law is gradually being implemente d. The federal government started a decentralization process in which licensing pr ocesses of forest management plans and deforestation will be transferred to state agenci es. Also, SFB staff is being formed. Finally, some state governments are also creating new state forests. For example, the State of Par, the most important timber producer in the Amazon, r ecently created three state forests, totaling an area of 7.8 million hectares. Forest Concessions in Brazil. The Brazilian Ministry of the Environment estimates that in the first 10 years after the implementation of the law, the forest area under concessions could reach 13 million hectares1, generating governmental revenue s of US$ 80 million and a total economic impact of US$ 820 million. The Minist ry estimates that 140,000 new jobs would be directly and indirectly creat ed through concessions (MMA, 2005). Forests designated for concessions will be divided into different pl ot sizes to guarantee access to small (< 10,000 hectares), medium (10,000 40,000 hectares), and large producers (40,000 200,000 hectares) (MMA, 2005). Criteria to be considered in the concession au ctions will include not only best prices, but also lower predicted environmental impacts, high er direct social bene fits, higher economic efficiency and higher aggregation of value in pr oduction. The minimum prices in the concession auctions will be established by the SFB and by the Commission for Management of Public Forests (CGFP), an independent consulting body formed to help the government in decisions related to the management of public forests. 1 On July 9, 2007, the SFB concluded the first pub lic forest inventory, including indigenous lands, conservation units, human settlements and other public fo rests. This inventory represents a starting point for land use planning in public lands, allocation of th ese areas for concessions, and elimination of illegal activities. The inventory totals 194 million hectares, of which 92% are located in the Brazilian Amazon (SBS, 2007).
26 The length of the contracts will vary between one forest cycle, typically 35 years according to forest management regulations, and 40 years. Concession contracts can be cancelled for several reasons, includi ng desistence and devolution of the contract by the company, bankruptcy, and non-adherence to the forest management plan. Beyond agencies monitoring measures, independent audits will be carri ed out in intervals no greater than three years. Revenues generated from concessions in public forests wi ll be divided among state government (30%), counties (40%) and FNDF (40%). In conservation units, such as national and state forests, IBAMA will also have a share of the revenues (Figure 2-4). Each public forest will be required to have an advisory committee and a specific management plan prior to the establishment of c oncessions and other land uses. The goal of this management plan is to establish directives for zoning potential land uses, meeting societal demands in relation to the public forest use. In the next sec tion, a brief literature review on spatial modeling in natural resources management will be presented, contextualizing the research question and the objectives of this stu dy, presented in the first chapter. Literature Review on Spatial Modeling Under the directives of the new forest law, so ciety has to make decisions in relation to the best use of the public forests to accommodate multiple objectives. These objectives may be conflicting, such that there are tradeoffs among different land-uses. From the public planner perspective, this also implies, in situations in which a win-win outcome cannot be reached, that the sought solution needs to expre ss societal values about effici ency and distributional equity (Loomis, 2002). In the literature, the foundati on of this problem is related to determining an appropriate allocation of land among competing uses. Economist s have explored this problem, using a social planners perspective, by mode ling land use alternatives that are able to present acceptable
27 tradeoffs among competing demands (Rothley, 19 99). In forestry and natural resource management literature, two main types of probl ems were investigated using land use modeling: (1) the reserve site selection problem, which aims to conserve the maximum number of species or biological features with a minimum cost or minimum number of reserves; and (2) harvest scheduling problems, which are c oncerned with the optimization of production forests to achieve higher efficiency in the harvest schedule, minimizing costs relate d to logging or dealing with ecological constraints. These two main streams of literature will be briefly presented bellow. Problems concerned with the reserve site selec tion issue started to be formulated in the earlier 1980s, originated in research fields such as conservation bi ology, operations research, and regional science (Church et al., 1996; Costello and Polask y, 2004). Two main types of subproblems are the set coverage problem, which deal s with how to have a given number of covered species with a minimum cost or number of re serves; and the maximal coverage problem, in which the objective is to maximize the number of conserved species for a fixed budget (Church et al., 1996; Ando et al., 1998; Onal and Briers, 2002). Latter work in this area used land prices and other economic information, such as timber stumpage values or net present value for productive activities, to make such models more realistic in choosing areas to create reserves (Ando et al., 1998; Polasky et al ., 2001; Polasky et al., 2005). Recent literature in reserve site selection problems mostly attempts to find solutions through mathematical programming, since these t echniques can guarantee an efficient solution2 (Rothley, 1999). Such problems are also rec ognized as intrinsically integer programming (IP) problems, since the several differe nt formulations reported in the literature typically creates a 2 In economics, the term efficiency refers to th e generation of an outcome with a minimum waste or minimum cost. In this way, an efficient solution is achieved when a given method generates the solution associated with the least cost or the highest possible total benefit.
28 binary variable expressing whether a given area will be assigned for a network of reserves. Other possible investigated approaches include goal programming (Onal and Briers, 2002) and heuristic techniques. The latter has been reported in this li terature as an alternative for computational difficulties generated by large IP problems normally solved using the branch-andbound algorithm, the standard algorithm to solve IP problems. The size of the branch-and-bound tree will increase and be exponentia lly more difficult to solve as the number of binary variables increases. The branch-and-bound algorithm is not of ten capable of solving these large problems. Therefore, other heuristic approaches have been developed (Onal and Briers, 2002). As mentioned earlier, a second common type of literature in natura l resource management and land use modeling is represented by harves t scheduling and forest tactical planning problems. Such problems are mainly concer ned with optimizing production forest use, maximizing their economic returns or minimizing costs related to logging. Road construction planning is an issue that can be addressed using these approach es (Murray, 1998). In this way, some of the past work in this area (e.g., Karlsson et al., 2004) has dealt with the classical timber supply problem. Such models are us ed in long-run planning of forest operations in which the best decisions regarding the use of equipment, harves ting of timber resources, processing capacity in the mill, allocation of harvest teams, transportati on, and storage, among several other factors, can be assessed jointly. Most of these probl ems can be solved th rough linear programming techniques using the simplex algorithm (Mu rray, 1999). Improvements in tools such as Geographical Information Systems and optimizati on software allowed the design of much more complex models in forestry dealing with other possible spatial constraints, such as avoiding adjacency among harvested stands to benefit native species in logged forests (Carter et al., 1997; Murray, 1999; McDill and Braze, 2000).
29 Also, as discussed for the reserve site selec tion problems, those more complex problems in forestry required more sophisticated formulati ons such as IP programming. Integer variables allow models to assign for a given stand a specific management regime or to choose among several options to locate roads (Murray, 1998). Ho wever, as discussed before, IP formulations can be computationally challenging, requiring perhap s several heuristic algorithms to be used to address such problems, as can be illustrated by Clark et al. (2000), Richards and Gunn (2000), Bettinger et al. (2002), B oyland et al. (2004), and Batten and Zhu (2005). Spatial modeling optimization in forestry becam e so interesting for the field of operation research that in 2000 an entire is sue of Forest Science [46 (2)] wa s dedicated to such problems. In this issues introductory ar ticle, Murray and Snyder (2000), hi ghlighted five main areas for continuing research in spatial modeling optimization in forestry: (i) reserve site selection problems; (ii) adjacency concerns in harvest sc heduling; (iii) road acc ess network in harvest scheduling; (iv) hierarchical forest management planning proc esses; and (v) integration of production and conservation considera tions. The latter point is the main issue of this work and will be better discussed bellow. Some recent literature in spatial modeling in natural resource management has focused on the implicit tradeoffs between conservation a nd commodity production. The rationale is to determine through modeling and simulations the production possibility frontier (PPF) between these goals. Such PPF curves s how, at one extreme, the maximum economic value that a given area or landscape can generate without any produ ction of the non-market goal and, in the other extreme, the maximum generation of the ecol ogical attribute in th e absence of economic activities. Each point between these two extremes in the PPF is an efficient point, in the sense that it is impossible to increas e the production of one good without decreasing the production of
30 the other. There is a growing quantity of studies in the literat ure estimating the PPF or assessing marginal costs for the relationship between mark et and non-market goals. In forestry, economic goals often have being measured through stumpage values or the societal welfare generated through logging. A non-exhaustive list of studies in the literature on this issue includes Calkin et al. (2002), in which non-market goals were represented by wild life persistence in natural areas; Hurme et al. (2007), evaluating habitats for th e Siberian flying squirrel ( Pteromys volans ) in Finland; Montgomery et al. (1994), estimating a marginal cost curve associated with the survival likelihood of the northern spotted owl ( Strix occidentalis caurina ); Nalle et al. (2004), estimating the PPF between economic surpluses from timbe r management and porcupine and wild owl populations; and Rohweder et al. (2000), investigating the PPF betw een silvicultural regimes and habitats for several w ildlife species. Using the land value under economic uses, some studies also generated opportunity cost maps for cons ervation, such as Naidoo and Adamowicz (2006), for the Mbaracayu reserve in Pa raguay; and Chomitz et al. (2005), for a region in southern Bahia, Brazil. Rarely has this previous work in spatial modeling in natural resource management taken into account multiple objectives in relation to forest use, often assessing a dichotomist choice for land use productive and conserved forest units or commodity and non-commodity use. Also, this literature rarely investigated situations related to zoning of conservation units, with the exception of Sabatini et al. (2007), in which the authors evaluated effective zoning designs in conservation units to achieve biologi cal protection goals. As far as I have been able to ascertain, no published study assessed the issues described in this section for Amazonian forests. This study seeks to fill part of this gap.
31 Figure 2-1. Logging centers and logging frontie rs in the Brazilian Amazon, 2004. Reproduced with permission. Extracted from Lentini et al. (2005), figure 7, p. 38.
32 02,000,0004,000,0006,000,0008,000,00010,000,00012,000,0001998 2004Y ear Volume of wood products (m3) E xport S o P aulo S tate Other Brazilian regions R egional market 15 % 19 % 59 % 36 % 15 % 38 % 11 % 7 % Figure 2-2. Markets for the wood production fr om the Brazilian Amazon in 1998 and 2004. (Source: Adapted from Lentini et al. 2004, 2005). 0.00 100.00 200.00 300.00 400.00 500.00 600.00 700.00 800.00 900.00 1,000.00 1,100.00 199819992000200120022003200420052006 Y earE xports Value (US $ million) S awnwood Finished wood products P lywood and veneer wood Other products Total Figure 2-3. Evolution of the value of wood products exported from the Brazilian Amazon between 1998 and 2006. (Source: Brazilian Mi nistry of Development, Industry and International Commerce, http://aliceweb.mdic.gov.br Last accessed March 10, 2007).
33 0%10%20%30%40%50%60%70%80%90%100% S tate and National Forests P ublic Forests % of the revenues from concession s IBAMA S tate Government Municipality FNDF Figure 2-4. Destination of gove rnment revenues generated by logging concessions in conservation units, as national and state fore sts, and in public forests. (Source: Brazilian Law 11,284/2006).
34 CHAPTER 3 METHODS My study developed an optimization model used to allocate public forest across multiple land uses. To parameterize this model, and to provide economic information to evaluate the tradeoffs between the production of market and non-market goods, some measure of the economic value of the land under commodity use was necessary. According to economic theory, under perfect market conditions, th e value of the land should equa l its NPV in its highest and best use (Polasky et al., 2001). In the case of state forests in the Brazilian Amazon, the NPV under timber logging is a reasonable estimate, since it is a societal preference to maintain forest cover in such areas, even though economic uses su ch as cattle ranching and soybean typically could achieve higher economic re turns than logging (Schneider et al., 2000; Margulis, 2003; Arima et al., 2005). This also implies that if areas within state forests are not being used for logging concessions, but are being reserved for comm unity use or biological conservation, that the expected NPV under logging in such areas is a pr oxy for the opportunity cost of these uses. By definition, opportunity costs ar e the costs of foregone opportun ities (Naidoo et al., 2006), and should be taken into account by so cial planners when they decide to reserve public lands for the production of non-market goods. Social planne rs may choose non-commodity uses if such choices reflect societal preferences, or if they believe that the market land value does not reflect the full social value of the land due to externali ties or environmental amen ities produced by these lands (Polasky et al., 2001). Economic returns from logging were calculated using a model that predicts the spatial variation in net returns due to in fra-structure conditions in each fo rest and the expected costs to harvest timber. A Geographic Information System (GIS) was built to combine spatially-explicit
35 information to provide data for the optimizati on model. The third step was to generate a theoretical model to address th e research problem and program this model in optimization software. This model was used in simulations to estimate the producti on possibility frontier (PPF) between logging and alternative land uses and investigate the opportunity costs for these alternative uses in a sp ecific public forest. The chosen public forest is embedded in the Calha Norte region, which will be presented in the next section. Calha Norte was used to estimate the net economic returns from logging and the roundwood demand from the logging centers in th e surroundings of the public forest. Then, a case study represented by this pub lic forest was used in the op timization model. These steps will be detailed in the following sections. The Calha Norte Region and Faro State Forest Calha Norte (expression that literally means nor thern trench) is located in the north of the Amazon, including portions of the states of Par, Amazonas, Roraima and Amap; delimited by a window with coordinates 62 13W 4 16N (northwest corner) and 47 51W 5 19S (southeast corner) (Figure 3-1). The re gion totals 132 million hectares, equivalent to approximately 1/4 of the Brazilian Amazon. Calha Norte was chosen because it still has little evidence of human occupation and economic activitie s in large part due to its topography. At least 50% of its area has high slope s and 80% are lands that are uns uitable for agriculture (IBGE, 2002). Moreover, the region is surrounded by important logging cen ters, including older centers with increasingly exhausted local stocks of raw material (Verssimo et al., 2006). Recently, this region has been the focus of several governmental efforts aimed at guaranteeing the conservation of its natural resources. In fact, the region was the focus of different initiatives of land use planning, star ting in the 1980s, since the Brazilian military
36 government believed that Calha Norte was a strategic region for the development of infrastructure projects and measures to protect frontier zones1. In March 2007, Calha Norte had 64 million hectares of protected areas indigenous lands, conservation units, and military lands equivalent to 40% of its total ar ea. At least 30% of these prot ected areas are conservation units in which the direct use of the forests is legally allowed, including extractive reserves, or state and national forests. State and national forests tota l 20 million hectares in Calha Norte, including 7.4 million hectares of state forests created in D ecember 2006 by the government of Par State, such as Faro State Forest (FSF), the public forest us ed as a case study in th e optimization model. FSF (635,935.72 hectares) is surrounded by othe r protected areas on its northern (Trombetas-Mapuera indigenous land), eastern (FLONA Sarac-Taquera and Trombetas river biological reserve) and western (Nhamund-Mapuera indigenous land) borders (Figure 3-1). FSF is covered by dense forests and presents litt le evidence of logging and human occupation in the extreme eastern and southern portions (IMA ZON, 2006). Faro was chosen because it is the smallest public forest in Calha Norte region (0.4 % of Calha Norte and 4.4% of the public forest area in the region), facilitating comput ational procedures during the modeling. The Spatially-Explicit Net Returns to Logging Model This model was constructed over past spatial-economic models that estimated the viability for logging across Amazons forests (Stone, 1998b; Verssimo et al., 1998; Verssimo et al., 1 Calha Norte was the name of the first development plan for the region created after the re-establishment of the democratic civil government in Brazil, in 198 5. The plan proposed the implementation of infrastructure and colonization projects, taking advantage of the large extents of land without economic activities, the incipient population dwelling in th e region, and several frontier zones. Among these measures, the plan foresaw the construction of the Perimetral Norte highway, the development of the So Parado hydroelectric plant, the increase of military installations in the area, and the establishment of colonization projects settling indigenous peoples in agri cultural plots. The planning was carried out in secret by the military, pretentiously due to national security reasons (Hecht and Cockburn, 1990). Most of the measures foreseen by the plan were forgotten, and concerns about territorial security in frontier zones were re-assumed by the SIVAM project (Amazo nian Vigilance System), created in 1990 and implemented in 1997.
37 2000; Lentini et al., unpublished). Such models created maps identifying, in a binary classification (e.g., viable and non-viable), which forested area were economically accessible for logging, based on wood prices and the variable cost s associated with logging, such as harvesting, transportation and processing costs. This new model, however, estimates the value of the net returns from logging (per m3 of roundwood basis), defined here as the profits for loggers plus stumpage prices. The model uses an algorithm modified fr om a GIS routine in ArcView 3.2a called CostDistance. The algorithm calculates, for each forest cell, the highest economic return from the logging of 1 m3 of harvested roundwood by finding a l east cost path (e.g., the cheapest transportation option) between the mills and the harvested forests. As a result, the model calculates the maximum profit by assigning the logging of eac h cell on the map to the logging center able to harvest this cell with the largest difference between prices for processed wood and the variable costs. After the modeling, cells a ssociated with negative pr ofits are assigned as unviable for logging. The model evaluates, for each logging center, th ree different prices representing three timber value classes (high, medium and low value species). The model then generates a map in raster format (1 km2 cell size) for each class. Variable costs include harvesting, transportation2, transaction and processing costs. The latter two are assumed to be fixed across the Br azilian Amazon (Table 3-3). Transaction costs are related to the access of le gal logs (e.g., documents, elaborat ion of the forest management 2 All variable costs are expressed on a cubic meter ba sis. Since harvesting and transportation costs are partially a function of fixed costs associated with logging operations, such as the construction of roads and infra-structure, they should vary depending on th e logging intensity in a given harvested area. Logged forests with lower logging intensities should present relatively higher harvesting and transportation costs. Due to the l ack of data to calculate these varia tions in harvesting and transportation costs, however, in this study, I assu med such costs to be uniform in stands with different logging intensities.
38 plans, and licensing permits). Processing costs considered are average costs for medium and large size sawmills equipped with one band-saw, the most common type of timber mill in the Amazon (Stone, 1997; Lentini et al., 2005). The aver age efficiency in conve rting raw material to sawn-wood used in the calculations is the typica l efficiency of medium size sawmills in the Brazilian Amazon, equal to 39% (Lentini et al., 2005). Modeling is executed by combining a map expr essing transportation co sts in each cell of the map (a friction coefficients map) with a map expressing the logging center location. The centers are used by the algorithm as the origin cel ls of the modeling, and are associated in this second map with a maximum bearable transportati on cost for each center, equal to the difference between prices and ha rvesting, transaction, a nd processing costs. In the first map, friction coefficients were ge nerated using the average costs to transport roundwood in every access surface type (e.g., paved and dirty roads, rivers) (Table 3-2). Natural obstacles to roundwood transportation, such as high slopes, or hydroelectric dams, were arbitrarily assigned to have higher friction coefficients. Modeling also took into account the cost to transport timber through non-existent roads th at would have to be built to make harvesting possible, using data estimated by Lentini et al (unpublished). Beyond tran sportation costs, other spatial data used in the friction coefficients map were: (i) Endogenous roads Roads built by private agents were mapped by IMAZON in 2003 (Brando and Souza, 2006; IMAZON, unpublished) In 2003, the study region contained 67,200 km of roads, from which only 28,000 km were offi cial roads. The fric tion coefficient map was built over these pre-existing roads and the navigable rivers described below. (ii) Navigable rivers for roundwood transportation mapped by IMAZON using data provided by surveys in the timber in dustry until 1999 (IMAZON, unpublished).
39 (iii) Forest cover in the Brazilian Amazon (IBGE, 1997) provided information about potential areas for logging. This map was updated excluding the deforested areas until 2004 (INPE, 2006; compiled by IMAZON, unpublished). Prices and costs (Table 3-2 a nd 3-3) were collected in 2004 through field surveys executed by IMAZON interviewing 680 timber mills owners and managers, representing 27% of the Amazons firms (Lentini et al., 2005). From this database, 519 interviews, embedded in Calha Norte, were used in this work. I assumed in this work average prices for sawn-wood at the mill gate for the domestic Brazilian market. As disc ussed in Chapter 2, it is the main consumer of Amazonian wood. The Land Use Allocation Model I will use the following notation. First, (1,...,) in represents the forest stands within a given public forest. Then, (1,...,) jm denotes the logging centers surro unding the public forest. Following, (1,...,) kz represents different timber value cla sses in the forest stands. Finally, (1,...,) ur represents exclusive land use altern atives in each public forest stand. One of the possible formulations for the objec tive function of this problem is formally shown in Equation 3-1. The objective is to ma ximize economic returns from logging within the public forest (denoted by ) by choosing the volumes that will be harvested per hectare (ijk X ) in each stand i from each timber value class k to each logging center j and choosing which stands to assign to each land use u (iuY). Total returns from logging () will be equal to the sum of the product across the stands i, logging centers j and timber classes k of the harvested volumes per hectare (ijk X ), the profits per cubic meter (ijkP), and the area of the stand (iA). The equation is then multiplied by iu
40 which expresses the potential of each stand for each alternative land use. In this first case of logging only, values are equal to (no potential for logging) or (with potential). Then, all terms are multiplied by the binary decision variableiuY, which is equal to if a given stand is assigned for logging. A given sta nd would not be harvested, genera ting zero profits, if at least one of the two following conditions holds : (i) if the potential of the stand (iu ) for logging is equal to zero; (ii) if the decision variable iuYis equal to zero. 111maxijknmz ijkijkiiuiu XYiu ijk X PAY and uLOGGING (3-1) The best stands for logging have a combina tion of large harvestable timber stocks (ijk X ) and high profits on a m3 basis (ijkP). This optimization problem c ontains important constraints. In the first set, the model deals with the ty pical supply problem for the logging industry. The first constraint (equation 3-2) c onstraints harvest to the availabl e timber supply in each stand i for each have timber class k (denoted byikS). 1 m ijkik j X Si (3-2) A second constraint relates to the milling capac ity in the logging centers. The total volume harvested through concessions must ultimately be equal or lower than the total milling capacity of surrounding logging centers for each center j (denoted as j D). 11()nz ijkij ik X ADj (3-3) If the social planner decides to assign the entire public forest for logging and will only reserve unprofitable areas for logging to alternative us es, the decision variable iuY can be dropped from the objective function. This first scenario, named unconstrained logging, was simulated
41 using the two constraints expressed above. Ultim ately, compared to subsequent scenarios, it revealed the tradeoffs involved in considering different land uses with in the public forest. In the second scenario, logging a nd other alternative la nd uses are considered in the model. To incorporate the binary decision variable iuY in the model, an additional constraint was constructed to force, at most only one land use alternative u for each stand i There are two ways to construct this constraint. In equation 3-4A, it was imposed that the sum of iuY for each i has an upper bound equal to 1. In equation 3-4B, the constraint establis hes that the sum of iuY across all stands ( i ) and land uses ( u ) has an upper bound equal to n (i.e., the total number of stands). 11r iu uYi (3-4 A) 11 nr iu iuYn (3-4 B) In the second scenario, the social planner imposes a minimum cumulative number of stands or a minimum score for land use alte rnatives other than logging, expressed as*uL For example, if the social planner decide that at le ast 30% of the state forest should be reserved for livelihood systems, *uL should be at least 0.3 n considering that all potential areas for this land use have the same weight (i.e., uniform values foriu ), a condition that does not necessarily need to hold. This constraint, expre ssed in equation 3-5, es tablishes that the sum of the product of the decision variable iuY and the potential iu has a lower bound equal to this minimum cumulative score. 1n iuiuu iYLuLOGGING (3-5)
42 Since the optimization model may assign a given land use for stands that are not connected to each other, creating a fragmented landscape, a final concern was to create a constraint able to impose a minimum connectivity level among stands for certain uses. An extensive literature exists which presents the potential benefits for biological conservation when reserves are created in larger landscape patches (e .g., Harrison and Bruna, 1999). So me ecologists, however, argue that harvesting executed with reduced-impact logging (RIL) techniques would not create fragmentation of surrounding reserves (F.E Putz, personal communication, April 12th, 2007). In the same way, some connectivity for stands as signed for livelihood systems or logging would be arguably desirable to decrease overall costs for management of these units. It is not the scope of this study to give a verdict about fragmentation created by adjacent logged and unlogged stands, but rath er demonstrate that desired sp atial configurations can be programmed. In reality, these issues can quickly gain a lot of complexity if we consider that desired spatial configurations depend not only on the number of c onnected stands but also the shapes of the reserves and the habitat requirements of different species. Therefore, in a third simulation, the model in corporates a final cons traint (equation 3-6) which imposes a minimum connectivity score for stands assigned to biod iversity conservation and livelihood systems. This constraint can be built establishing that f can also represent the stands (1,...,) in Then, a matrix of distance among stands (ifdist ) of nxnelements was created, in which every ifaelement is equal to if two sta nds are adjacent and if they are not. The matrix diagonal, in which if is set equal to 0. This method was adapted from adjacency constraints found in the literature kno wn as NOAM (new ordinary adjacency matrix) constraints (McDill and Braze, 2000). If tw o stands assigned for the same land use u are adjacent, they add 1 to the total adjacenc y score for this land use type (noted as*Adj). An
43 important condition for this constraint work is that f i avoiding double-counting the adjacency among stands. 1n iufuif iYYdistAdju (3-6) Building the Optimization Model This section describes the data and operati onal details of the optimization model. Four potential land uses were considered (notation u in the land allocation model): logging, mining, biodiversity conservation and live lihood systems. Manipulation of the spatially-explicit data presented in this section was done using ESRI ArcView 3.3 and ESRI ArcMap 9.1. Timber supply in Faro. There is a lack of recent information about timber species distribution in the Brazilian Amazon. Detailed forest inventories we re conducted in a few specific sites (for example, IMAZON and FFT s ites in Paragominas, Par, and EMBRAPA sites in the FLONA Tapajs, Santarm, Par) or by tim ber companies for harvest planning purposes. For this reason, this study used data from Rada mBrasil, a project with a wide geographical range that compiled sample forest inventories in all Amazonian States in the 1970s. It is still the most comprehensive survey of tree species available for the region. Currently there is some literature focusing on estimating tree species distribution from RadamBrasil data using kriging techniques, as Steege et al. (2 003). However, such work does not estimate the distribution of merchantable timber across the Brazilian Amazon and this may be considered an important constraint to larg e-scale planning of logging activities in Amazonian public forests. In this study, a map of timber species volume above 35 cm of diameter at breast height (DBH), provided by Sale s & Souza Jr. (unpublished), was used, and was developed by applying a kriging technique to RadamBrasil data This map expresses timber volume on a cubic meter per hectare basis for cells of 1 km wide.
44 To be useful in this work, it was necessary to estimate the proportion of this total volume that corresponds to mercha ntable timber species and, moreover, the proportion of the commercial volume that is formed by trees above 50 cm DBH the minimum cutting diameter required by Brazilian regulations. Then, this information was adapted to estimate the proportion of high, medium, and low value species in the merchant able timber volume > 50 cm DBH. M. Schulze and A. Macpherson (personal communication, March 12th, 2007) provided some data about the proportion of merchantable volume (> 50 cm DBH) from total volume (> 35 cm DBH) for three research sites in the Amazon Fazenda Agrosete (Paragominas, Par), Fazenda Cauaxi (Paragominas, Par) and FLONA Tapajs (Santarm, Par). These data also took into account the proportion of defective trees during the harvesting, such as hollowed trees or trees with defective stems. Data from these sites show that, for each cubic meter of roundwood > 35 cm DBH in the forest, on average, 0.41 m3 will be non-defective and merchantable > 50 cm DBH. Leonardo Sobral (personal communication, March 14th, 2007), provided some data on the distribution of volume by timber species in the F azenda Rio Capim, Parago minas, Par State. Timber species were aggregated into economic classes using the classification proposed by Lentini et al. (2005). The author s created timber value classes based on the confidence intervals ( = 5%) of the average sawn-wood prices of indicat or species in each econo mic value class. It was estimated that on average 6% of the explo itable volume is composed of high value species, 63% medium value, and 31% low value timber. From all estimated available merchantable volume in Faro, the optimization model took into account a maximum available supply of 30 m3 ha-1, the maximum harvestable volume allowed by Brazilian regulations for management cycles of 35 years. It was assumed for the
45 optimization model that timber stocks in each timber class are non-decl ining over subsequent harvest cycles. Implications of this assumpti on will be better discussed in the last chapter. Milling capacity in the logging centers. Lentini et al. (2005) provided estimates of the total capacity to proces s roundwood in the logging centers located in proximity to Faro. A portion of this amount of roundwood is currently being supplied by private lands. Using data from IBAMA (unpublished) of roundwood volume lega lly harvested in private lands in Calha Norte in 2004, I estimated the milling capacity of roundwood from public lands in the region as the difference between the roundwood consumpti on in the logging centers and the amount authorized through deforestation and forest mana gement plans on private lands. This amount is equal to 71% of the total regional milling capacity in 2004. Potential areas for biodiversity conservation. In 1999, ISA, a Brazilian NGO, led a wide consultation with several stakeholder gr oups (government, NGOs and research institutions) to identify hotspots for biodiversity, which would be recommended to be protected as conservation units. The map generated in th is consultation presents the hotspots in a classification of five differen t levels of importance for cons ervation based on a score given by specialists related to their biological diversity. This map (ISA et al., 1999) was used to identify areas within Faro with high potenti al for biodiversity conservation. Potential areas for livelihood systems. Barreto et al. (2006) and IMAZON (unpublished) created a map of forests with evidence of human occupation in the Amazon. In this map, areas with evidence of human occupation encompasse d mining permits, deforested areas until 2004, urban zones, official human settlements in rural ar eas, and areas located within a radius of 10 km from forest fires identified by satellites between 1996 and 2006. Th e latter areas are potentially being used by forest dwellers in their livelihood strategi es. As discussed before, according to the
46 Management of Public Forests Law, occupied fore sts will be reserved for exclusive use of their dwellers. However, to inves tigate the compromises among comp eting land uses, I considered that social planners can choose the total proportion of the public forest that will be reserved for the use of traditional forest communities. Therefore, this map was used to identify, in Faro, the areas with potential for livelihood systems, defined in categories as older occupied areas forest fires detected between 1996-2002 and recently o ccupied areas forest fires between 2003-6; and the areas with potential for mining. Potential areas for logging. The map of potential for agri culture in Brazil (IBGE, 2002) was used to identify areas within Faro with st eep or very steep slopes. High slopes would in theory increase the logging costs and, therefore, decrease the returns from logging. I assumed a decrease of 30% in the net retu rns in high slope forests. Then areas with very high slopes (> 45 ) were considered forbidden for logging, follow ing directives from the Brazilian Forest Code (Lei 4,771/1965). Forest stands inside Faro w ith such topographic condit ions were considered without potential for logging. Operational modeling aspects. Faro State Forest (FSF) ha s a total area, calculated in GIS, of 637,500 hectares. Faro was converted in to raster format formed by 255 stands (i.e., notation i in the land allocation model) of 5 km cell si ze (2,500 hectares each). Considering the requirements for forest management plans in the Brazilian Amazon3, each stand was divided into 35 annual harvesting units of 71.4 hectares each (notation iA in the land allocation model). As discussed in the Chapter 2, a cel l of 2,500 hectares represents a small concession. Considering 3 According to the Normative Instruction 05/2006 fr om IBAMA, a minimum cutting cycle of 35 years is required for harvested forests with a logging intensity of 30 m3 ha-1.
47 the same requirements for forest management plan s, this area would be sufficient to sustain a small mill that consumes 2,100 m3 year-1 (in average terms less than 3 logs per day). Optimization modeling and solvers. The software used in the analyses was IDE GAMS 22.4 (General Algebraic Modeling System). The unconstrained logging scenario, in which logging was the only land use alternative consid ered, is a linear programming (LP) problem, and was solved with CPLEX 10 solver. Simulations incorporating other land uses could not be solved through CPLEX, even though it is an efficient algorithm for mixed integer programming problems, which is the case when it wa s incorporated the binary variable iuY. Also, due to the non-linearities in the mode ls objective function, the problem was solved using an algorithm for mixed integer non-linear programming (MINLP ) problems, named DICOPT (Discrete and Continuous Optimizer), developed by Viswan athan and Grossmann (1999). DICOPT was chosen because it was proven to be an improve ment in robustness and time efficiency over several other solution methods for MINL P problems (Varvarezos et al., 1992). Simulations Executed Simulations were conducted with the optim ization model to estimate the production possibility frontier (PPF) and inve stigate the tradeoffs among different land uses in FSF. The first set of simulations consider ed a fixed weight of each potentia l stand for alternative land uses (i.e., the variable iu in the land allocation model) (Figure 33). Then, simulations were executed with an increasing number of stands assigne d to livelihood systems and/or biodiversity conservation. Three general scenarios were considered: (i) the unc onstrained logging (UL) scenario, as explained before, c onsidering logging as the only land use within Faro; (ii) the multiple use scenario without connectivity concer ns; and (iii) the multiple use scenario with connectivity concerns.
48 In the second set of simulations, the same set of scenarios were investigated considering differentiated weights for poten tial stands assigned to live lihood systems and biodiversity conservation. In the case of livelihood systems, I arbitrarily assigned fo r stands in which older forest fires were identified (1996-2002) a weight equal to 2/3 of the weight assi gned to stands of recent forest fire (2003-6). I also arbitrarily assi gned an increasing gradient of weights (1-5) for potential stands for biodiversity conservation, from the western to eastern portions of FSF, since its eastern portion contains hi gher biodiversity according to th e map from ISA et al. (1999) (Figure 3-3). It is not the scope of this st udy to discuss how accurate these scores are in representing the relative importa nce of stands but, rather, to demonstrate the models capabilities in taking this information into account in the optimization. In the third set of simulations, I assume that sawmills move to the urban centers located closest to Faro that currently have no sawmills under the assumption that new mills would be constructed to take advantage of the legal timber supply from FSF4. The main impact that such a decision would provoke, considering th at the cost of moving a firm can be considered as a fixed cost and therefore should not in fluence in the firms harvesting decision is a decrease in the transportation cost, allowing logging to be more profitable inside Faro and a larger number of stands to be harvestable5. I included 14 cities in this simulation: Barreirinha, Boa Vista do 4 As discussed in the second chapter, mills located in older frontiers are continuously migrating to newer frontiers, stimulated by the exhaustion of raw material sources in old logging centers. On the one hand, due to the abundance of roundwood in Calha Norte, it is expected that mills will start to migrate to the region in the coming years, even without the establis hment of concessions. On the other hand, factors such as the little infra-structure ava ilable in the cities closer to FSF (i.e., electric energy, paved roads and specialized services), the difficult acce ss to markets, and the low local availability of specialized labor might discourage the migration of firms for the next few years. 5 It is also true that the costs to transport the final production to markets would become relatively more expensive. To mitigate this effect I used, in the net returns model for the cities closer to FSF, average prices equal to the prices in mills located in the closest logging centers to these localities.
49 Ramos, Itapiranga, Maus, Nhamund, Parintins, Silves, So Sebastio do Uatum, Urucurituba and Urucar, in the State of Amazonas; and Faro Juruti, bidos, and Terra Santa, Par State. Benefit-Cost Analyses (BCA) and Sensitivity Analyses Recalling, the optimization model calculates in the objective function the annual profits for concessions. Harvested volum e is calculated on a per-hectare basis, and the total annual harvestable area in each stand was set to 71.4 hectar es. In the final set of analyses, annual profits from logging were used in a BCA to calculate the NPV from logging conc essions (equation 3-7). 1(1)1(1) (1)1nn audestab trr NPVCC rr (3-7) Where is the annual profits from logging calcu lated in the optimization model, and n represents the length of the concessions contract s set to 40 years, which is the maximum length allowed by law. The interest rate, r, is equal to 10% year-1, based on the long run interest rates of the Central Brazilian Ba nk in 2004 (BNDES, 2007). The second term of the equation calculates the present value of the ec onomic benefits from concessions, and the third term the presen t value of the audit costs, expressed byaudC. Every three years (time interval represented by the notation t) the law requires that concessions must be audited by independent bodies. Due to the lack of suitable data, au dit costs were assumed to be equivalent to current costs for certification a udits, which are typically estimated between US$ 0.15 ha-1 year-1 and US$ 1.00 ha-1 year-1, depending on the size of the management units (Mauricio Voivodic, personal communication, May 4th, 2007). In a baseline scenario, I used an estimate of US$ 3,000 stand-1 every 3 years (US$ 0.40 ha-1 year-1). However, larger concession units may have lower audit costs, with positive impacts in the NPV.
50 The notation estabC is the establishment costs for implementing forest concessions. The Federal Decree 6,063 (March 20th, 2007) establishes that such cost s include forest inventories to set minimum prices, preliminary studies, granti ng licenses, conducting the competitive proposal and auction processes. Such establishment cost s are case-specific and may be very different across countries and regions. However, suffi cient information exists to estimate the establishment costs. According to Sabogal et al. (2006), inventory costs for timber production executed by private companies in the Brazilian Amazon typically cost US$ 13.0 US$ 17.0 ha-1. While larger state forests may have lower costs due to economies of scale, establishment costs will be within an interval of US$ 15.0 US$ 40.0 ha-1. In a baseline scenario, I applied US$ 20.0 ha-1. Sensitivity analyses were conducted to investig ate the effect of the establishment and the audit costs in the NPV generated by forest conc essions within FSF. As mentioned before, a baseline scenario considered establishment co sts of US$ 20 ha-1 and audit costs of US$ 0.40 ha-1 year-1. Three additional simulations were performed. In the first, the low cost scenario, such values were assumed to be e qual, respectively, to US$ 10 ha-1 and US$ 0.20 ha-1 year-1. In an intermediate scenario, establishment costs were set at US$ 30 ha-1 and audit costs at US$ 0.70 ha1 year-1. Finally, in the high cost scenario, such valu es were assumed to be, respectively, equal to US$ 40 ha-1 and US$ 1.00 ha-1 year-1. Considering the baseline scenario costs, two ca lculations of NPV were made considering a societal and a loggers perspective6. In the latter, it was assume d that loggers can capture in concessions equivalent profits to logging in private lands, calcu lated by Verssimo et al. (2002b) 6 The private BCA appraises a given project under a fi rms perspective, evaluating its effects on revenues and costs and, therefore, in the firms profits. Wider implications of the project are investigated under the referent groups perspective, in the so cial BCA (Campbell and Brown, 2003).
51 as varying between 10% 26% over the prices fo r sawn-wood. In this study it was considered that loggers would capture 10% ove r the prices in the logging cente rs. The remaining profits are assumed to be captured by government through minimum prices in the auctions, volume harvested fees or profit fees7. Loggers also are assumed to reimburse the government for the establishment costs and pay the triannual audits. In both cases, the internal rates of return (IRR) were calculated. The PPF Between Logging and Other Land Uses The simulations and sensitivity analyses generated several NPVs according to an increasing number of stands or the cumulative scor e for land use alternatives other than logging. These values were used to estimate the PPF among land uses, such as logging and livelihood systems, or logging and biodiversit y conservation. Then, using th e establishment and audit costs of the baseline scenario, I tested the hypothesis that marginal opportunity costs (mOC) involved in these zoning decisions should be increasing when an increasing proportion of FSF is being assigned for land uses other than logging. Such increases in mOC are implicitly shown in the PPF generated to investigate tradeoffs between market and non-market goals in land use planning (e.g., Polasky et al., 2005). 7 From a theoretical point of view, while privat e agents are exclusively concerned with costs and revenues that affect the firms profits, from a social perspective the external costs provoked by a given project must be taken into account. The rationale for the fees and taxes imposed by the government is based on the idea that loggers, the group which w ill collect direct benefits from concessions, must compensate the rest of society by eventual external costs provoked by logging. This rationale is consistent with the concept of economic efficiency created by Kaldor and Hicks, where those who become better off under a given policy should compensat e those who are worse off, leading to a Pareto optimal outcome.
52 Figure 3-1. View of the Calha Norte region, northern Amazon. A) Map of Calha Norte highlighting the main roads, conservati on units, and logging cen ters. B) Surroundings of the Faro State Forest, highlighting the clos er urban centers to the state forest.
53 Figure 3-2. Spatially-explicit and economic data used to build th e net economic returns and the land use optimization model. Economic Viability for Selective Logging Net economic returns for logging Land use optimization model Timber volume > 35 cm DBH Human Occupation Priority areas for biodiversity conservation Topographic conditions Data used and previous A. Geographic Endogenous roads B. Economic data Navigable rivers for roundwood transportation Logging centers Urban centers Deforestation Typical costs for logging (transaction, harvesting, transportation, processing) Roundwood milling capacity in logging centers Roundwood volume consumed through deforestation permits and forest management plans Wood Prices Models generated and executed analyses
54 Table 3-1. Sources and year of the analyses of th e datasets used in the two models developed in this study. Data description Year of the Analyses Source Economic viability for selective logging 2004 Lentini et al. (unpublished) Endogenous roads 2003 (Brando and Souza, 2006), IMAZON (unpublished) Navigable rivers for roundwood transportation 1999 IMAZON (unpublished) Logging centers 1998, 2004 Lentini et al. (2004, 2005) Urban centers 1991 IBGE (1991) Deforestation 2004 INPE (2006), compiled by IMAZON (unpublished) Timber volume > 35 cm DBH 1976 Sales & Souza Jr. (unpublished) Priority areas for biological conservation 1999 ISA et al. (1999) Human occupation 2003 Barreto et al. (2006), IMAZON (unpublished) Topographic conditions 2002 IBGE (2002) Variable costs in logging activity (harvesting, transaction, transportation and processing) 2004 Stone (1997), Verssimo et al. (2000), Lentini et al. (2005), IMAZON (unpublished), and Lentini et al. (unpublished). Sawn-wood prices 2004 Verssimo et al. (2002), Lentini et al. (2005), and IMAZON (unpublished) Roundwood production characteristics milling capacity in logging centers 2004 Lentini et al. (2005) Roundwood volume consumed in the logging centers through deforestation permits and forest management plans 2004 IBAMA (unpublished)
55 Table 3-2. Costs for roundwood tran sportation and friction coefficients used in the model of net economic returns for logging. Access type to forestlands Transportation Cost1 (US$ m-3 km-1) Friction Coefficient River (barges or rafts) 0.03 0.05 4 Paved roads 0.07 0.14 11 Graveled dirt roads (high quality) 0.15 0.18 17 Non-graveled dirt roads (poor quality) 0.21 0.24 23 Non-trafficable roads 0.60 60 Construction of new roads through non-forest areas 0.92 92 Construction of new roads through forestlands 1.49 149 Natural obstacles 3,000 1 Source: Verssimo et al. (2000), Lentini et al. (2005), IMAZON (unpublished), and Lentini et al. (unpublished); and estimates made using data from Verssimo et al. (1995) and Stone (1997, 1998b).
56 Table 3-3. Sawn-wood prices and typical costs of the logging ac tivity in the logging centers of the Calha Norte region, 20041. Sawn-wood prices (US$/m3) per timber value class (s.d.) Costs per m3 of sawn-wood (US$) (s.d.) Zone (# of interviews) High value species2 Medium value species3Low value species4 Harvesting5 Transaction6 Processing7 Amazonas (129) 225.70 (266.3) 155.60 (179.2) 120.90 (130.0) 28.80 (20.3) Central Para (94) 404.70 (202.2) 189.74 (190.5) 153.30 (184.5) 27.30 (17.52) Estuary (204) 342.50 (230.9) 122.98 (132.9) 98.90 (156.6) 23.20 (8.41) Western Para (92) 338.10 (285.2) 165.38 (167.3) 104.30 (116.7) 30.90 (18.9) Amazonia (680) 303.90 (273.1) 161.23 (175.4) 111.00 (114.1) 27.60 (18.7) 28.20 (22.2) 34.40 (16.5) 1 Source: data from IMAZON (unpublished) and Lentini et al. (2005). 2 High value timber is formed by species such as cedro ( Cedrela spp .), freij ( Cordia spp .), jatob ( Hymenaea courbaril ), itaba ( Mezilaurus itauba ), ip ( Tabebuia spp .) and cerejeira ( Torresia acreana ). High value species were classi fied by Lentini et al. (2005) as species which have an average sawn-wood price in the domestic market higher than US$ 210 m-3. 3 Medium value timbers encompass non-high value species used to produce sawn-wood for civil construction parts, flooring, decking, etc. These species were classified by Lentini et al. (2005) as species whic h have an average sawn-wood price in the domestic market higher than US$ 130 m-3 and lower than US$ 210 m-3. 4 Low value timbers encompass species used to produce plywood or low value sawn-wood in regional Amazonian markets. Such timber was classified by Lentini et al. (2005) as species which have an average processe d price in the domestic market lower than US$ 130.0 m-3. 5 Harvesting costs include all operations conducted to extract timber from the forest, as cutting and bucking of trees, and skidding operations until the log decks, as well as loading the logs on trucks. 6 Transaction costs include all related costs to the documentation of the roundwood, including Forest Management Plans, environmental licen ses and documents to transport roundwood, etc. 7 Processing costs include operations carried out in the firms to transform roundwood to sawn-wood to be sold in the domestic Brazilian market in medium to large size sawmills in the Brazilian Amazon, which present an average industrial efficiency in converting roundwood to sawn-wood equal to 39% (Lentini et al., 2005).
57 Figure 3-3. Potential for land use alternatives in Faro State Fo rest and weights used in the simulations. A1) Location of areas with pot ential for livelihood systems or mining. A2) Differentiated weights for livelihood sy stems in the second set of simulations (see text). B1) Location of areas with high importance for biodiversity conservation. B2) Gradient of different we ights for biodiversity conserva tion used in the second set of simulations (see text). C1 ) Topographic conditions within Faro State Forest. Data respectively from Barreto et al. (2006) ISA et al. (1999) and IBGE (2002).
58 CHAPTER 4 RESULTS This chapter is organized as follows. First, I wi ll present the results of the net returns from logging model in the Calha Norte region and in Fa ro State Forest (FSF). Then, using methods described in the previous chapter, I will present the estimated supply and the milling capacity for roundwood from Faro. Results from the optimization model will be presented next, focusing on the timber supply problem for Faro. Results fr om the simulations will then be presented, showing different landscape mosaic alternatives in Faro as a consequence of the intended number of stands or the cumulative score pur sued for livelihood systems or biodiversity conservation. The variation in the annual profit s from logging due to such choices will then be reported. The last section will present the pr oduction possibility frontier (PPF) for logging and other alternative land uses and investigate th e tradeoffs between the NPV from logging and the provision of alternative land uses within Faro. Estimated Net Returns from Logging in Calha Norte Figure 4-1 shows how net returns from logging va ry in the Calha Norte region, including FSF. Table 4-1 shows the expected net return s from logging in Calha Norte forests for each timber value class. Considering only the forest ed lands in Calha Norte (80% of the region), approximately 20% of these forestlands are unvi able for logging of high value species (i.e., negative returns from logging). One quarter of these forests have economic returns between US$ 70 and US$ 100 m-3 for high value species. Almost 10% have returns higher than US$ 100 m-3. The more profitable forestlands are actually lo cated in the center of Calha Norte, within proximity of the Transamazon highway and close to important logging cente rs such as Altamira, Uruar and Pacaj. Some logging centers can access roundwood of high value species at distances greater than 300 km.
59 On the other hand, the harvesting of medium va lue species is viable in only 42% of the region, while the maximum harvestable distance s from these logging centers decreases to a maximum of 100 km. Only 11% of the region in dicates net returns from logging of medium value species above US$ 40 m-3. Finally, only 16% of the forest lands in the region are viable for logging low value species, since the maximum ha rvestable distance from the logging centers decreases to less than 50 km. Net returns within FSF. The most profitable stands for logging within Fa ro are located adjacent to its southern and eastern borders, in the proximities of rivers and the few existing roads. All stands within Faro are viable for l ogging of high value species, and almost 30% of its stands have net return s greater than US$ 70 m-3 for high value species (Table 4-2). However, only 22% of Faros stands are viable for logging medium value species and no stands are viable for the harvesting of low value species. If the in dustry moves to cities closer to FSF, 32% of its stands become feasible for loggi ng of medium value species and 7% for low value species (Table 4-2). Estimated Timber Supply within Faro State Forest According to the map of timber specie s volume provided by Sales & Souza Jr. (unpublished), the stands within FSF present, on average, 119 m3 of roundwood ha-1 (total stock of ~ 3,000 m3 stand-1). As expected, the variation among sta nds is very small (standard deviation of 4.0), since the original map was generated in a coarse scale for the entire Amazon. Most stands with higher timber volumes (> 120 m3 ha-1) are located in the cent ral-southern portion of FSF. As discussed in the last chapter, I estimat ed from this map the total merchantable timber inventory volume (> 50 cm DBH) which is on average, 49.1 m3 ha-1. From this volume, 2.9 m3 is assumed to be available from high value species, 18.1 m3 from medium value species, and 8.9 m3 from low value species (Table 4-3).
60 Milling Capacity for Roundwood from Public Forests in Calha Norte The total roundwood consumption from the 17 l ogging centers within the region was equal to 1.6 million m3 in 2004 equivalent to 7% of the total demand in the Brazilian Amazon (Lentini et al., 2005). Of this amount, at least 700,000 m3 was supplied from areas having forest management plans (FMP) and deforestation perm its granted by IBAMA. Therefore, around 0.9 million m3 was supplied by other sources (Table 4-4) which could include legal sources (e.g., FMP and legal deforestation) located outside the region, or illegal harvesting and illegal deforestation within the region. As discussed in the last chapter, assuming that the total milling capacity will remain constant in the region afte r the establishment of concessions, the proportion of the consumption currently supplied by other s ources is assumed to be the milling capacity of roundwood from public lands. The Timber Supply Problem The unconstrained logging (UL) scenario. Without considering any other alternative land uses within FSF, 250 stands 98% of Faro would be logged by the logging centers (Figure 4-2). The few unlogged st ands are located at the northeastern porti on of Faro, in which the existence of high slopes forbids harvesting activ ities. Forty four stands (17% of Faro) would have more than 70% of the volume logged from both medium and high value species. In the remaining area, only high value species would be logged. Total harvested volume would be equal to approximately 109,000 m3 equivalent to 12% of the total milling capacity of timber from public forests in Calha Norte. Logging companies would generate annual profits from logging of US$ 3.2 million (see the first bar in the Figures 4-3A and C). Logging would be carried out only by companies located in 3 loggi ng centers: Oriximin, Santarm and Uruar (Figure 4-4C).
61 As expected, Faro would become relatively mo re important for logging in the region if the industry moves to closer cities. In this scenar io, the same 250 stands would be logged, but 63 stands (25% of Faro) would have all the high and medium value species logged, and 18 stands (7% of Faro) would have all of the available timber harvested (Figure 4-2). Total volume harvested within Faro would rise to 168,519 m3, supplying 18% of the total milling capacity (first bar of the Figures 4-3BD and 4-4D). A nnual profits from logging would increase to US$ 4 million. From all the closer cities considered in this scenario, harvesting would be performed mainly by three urban centers: Faro, Nhamund and bidos. Interaction of logging and other land uses. Obviously, as the number of stands assigned for alternative land uses such as mining, bi odiversity conservation and livelihood systems is increased, there is a decrease in the number of stands logge d, timber volume harvested, and profits from logging. Figure 4-3 shows what ha ppens to the number of stands assigned for logging and harvested volume when logging is pe rformed by firms located in the current logging centers (Figure 4-3A and C) and in closer cities (Figure 4-3B and D). It is interesting to note that, in the case of the logging centers, the propo rtion of stands in which high and medium value species are harvested is always lower than 17% of the stands; in the second case (closer cities), always above 20% of the stands. In the case wh ere 48% of stands is under other land uses, only 133 stands are logged, and only 44% of the origin al available volume is harvested by the logging centers and 48% by closer cities. As will be disc ussed later, in this situation, annual profits would decrease to 46% and 47% of the UL scenario, respectively. Roundwood supply within FSF. In the UL scenario, with logging centers performing the harvests, only 20% of Faros total merchantable timber inventory would be logged. This proportion decreases to 9% of the merchantable timber inventory when 122 stands are assigned
62 to other land uses (48% of stands in other land uses). However, it is important to note that the timber stocks within the state forest are not logged uniformly 98% of the high value timber supply is logged in the UL scenario while onl y 17% of the medium value timber available is logged and 0% of the low value timber. Possible im plications of such findi ngs will be discussed in greater details the final chapter. Consider ing logging by closer cities, 31% of Faros total merchantable timber inventory is harvested in th e UL scenario, depleting 98% of the available timber inventory of high value species. Howe ver, logging performed by closer cities also resulted in a higher usage of medium value spec ies (32% of the availa ble inventory) and low value species (7% of the available inventory) (Figure 4-4A and B). Regional roundwood milling capacity. As discussed before, in the UL scenario with logging centers performing the harvests, only 12% of the regional milling capacity can be satisfied by Faro. This proportion falls to 5% in the extreme opposite situation in which 122 stands are assigned to other uses (Figure 4-4C). Interestingly, the esta blishment of concessions within Faro would be very important for Orixim in, a small timber pr ocessing center (~ 9,500 m3 year-1), which would be fully supplied by concessions in FSF. Other logging centers such as Santarm and Uruar would also be able to sa tisfy a large share of th eir capacity with roundwood from Faro. If the industry moves to closer citi es, Faro could satisfy 18% of the total milling capacity if the companies concentrate in the ci ties of Faro, Nhamund and bidos. In this situation, the city of Faro would be able to process between 75% and 93% of the roundwood harvested within FSF, depending on the number of sta nds assigned to other uses (Figure 4-4D). Landscape Patterns Formed by the Optimization Model Simulations shown in Figure 4-5 assumes each sta nd is weighted equally with respect to its contribution to the provision of an alternative la nd use (for stands with potential for providing the alternative land use). As such, the simula tions performed graduall y assigned an increasing
63 number of stands for livelihood systems and biodiv ersity in the least co st manner. The model attempts to avoid stands for conversion that have the highest logging profits. For livelihood systems, this later task becomes quickly difficult, since a large part of the areas with potential for livelihood systems are located near rivers and Faro s borders, stands which also have naturally high logging profits. Stands for biodiversity conservation can only be assigned on the eastern side of FSF (i.e., stands with pot ential for biodiversity conserva tion according to the map from ISA et al., 1999), and in the fi nal simulations they also occu py highly profitable stands for logging. Figure 4-5 compares landscape patterns gene rated by the optimization model developed both with and without co nnectivity constraints, considering harvesting as being performed from the current logging centers. Without connectivity constraints, individual patches for livelihood systems and biodiversity conservation tend to be smaller, more so for livelihood systems, which also implies that a larger number of patches will be created for a given number of stands assigned for these alternative land uses. With respect to number of patches, for livelihood systems, there were on average 3.5 patches crea ted versus 4.5 without the constraint, and for biodiversity conservation, 1.5 vers us 2.3 patches. With respect to patch size, livelihood systems average size was 3.9 stands wit hout the constraint versus 5.3 st ands with it. For biodiversity conservation, average patch size was 29.4 and 32.5 stands, respectively. Surprisingly, a very small decrease in the objective function value (a nnual profits from loggi ng) resulted from the constraint, varying from a 0.6% to 3.5%. Due to the assumed importance of landscape connectivity for planning purposes and due to the low decrease in the objective function generated by the constraint, the remaining result s reported in this chapter include connectivity constraints.
64 Figure 4-6 presents landscape patterns within Faro under increasing cumulative scores for livelihood systems and biodiversity conservation. St ands are weighted differentially within each use. In the case of scores for livelihood system s, the model first avoids to use more profitable stands for logging, as well as th e stands with lower potential for this land use. As the score constraint becomes larger, almost all potential sta nds are converted. The same logic is valid in the case of biodiversity, but there is more flexib ility in the assignment of stands with priority given to converting stands on the eastern side of Faro to achieve a given score. Annual Profits from Logging and Government Revenues As discussed in the beginning of this chapter, without alternative land uses within Faro, the total annual profits generated by logging performed from the surrounding logging centers would be equal to US$ 3.2 million. Such profits would naturally decrease if the social planner decides to allow other land uses within FSF. Then, in the opposite extreme, profits would decrease to US$ 1.5 million if 48% of Faros stands are conv erted to alternative land uses (assuming equal stand weights). A second question that could be raised by the soci al planner refers to how much of the logging profits could be taxed by the govern ment through, for exam ple, royalties, logging fees or profit taxes and still leave the loggers in a profitable position. As discussed in the methods, one assumption is to tax loggers only e nough such that they still earn a normal profit (i.e., equivalent to profits ear ned by loggers on private lands). Figure 4-7 shows the share of the profits for logger s and government under this assumption. Government would be able to extr act around 31% to 38% of the annual profits from logging if it is being performe d by firms located in the surrou nding logging centers (Figure 47A). Given total annual profits of US$ 3.2 million in the UL s cenario, US$ 2.3 million would be considered as normal profits for loggers and US$ 0.9 million would be paid to government (Figure 4-7A). Figure 4-7B shows the same sh are if harvesting is performed by firms that
65 moved to cities closer to Faro. In this second scenar io, government would be able to extract 31% to 32% of the total annual profits, which would be equivalent to US$ 1.2 million of the US$ 4 million generated if logging is the only land use considered within Faro. Figures 4-7C and D show the annual profits from logging under increasing scores for livelihood systems and biodiversity conservation, respectively. Both consider harvesting has been performed from the surround ing logging centers. The governments share of the annual profits under increasing scores for livelihood varies between 28% and 29% and, under increasing scores for biodiversity conser vation, between 26% and 29%. The base case UL scenario was used to inve stigate the spatial be havior of the annual profits from logging within Faro (Figure 4-8). Considering lo gging performed by firms in the current logging centers, on average, each stand ha s an annual profit of US$ 12,619. This value increases to US$ 15,744 when the ha rvests are performed by firms lo cated in cities closer to the state forest. In both scenarios, the most profit able stands are located in Faros southern and eastern portions. The highest prof it achieved by an individual stand, in the first case, was close to US$ 40,358, and in the second case, US$ 54,502. Logging NPVs and IRRs under Increasing Conversion to Alternative Land Uses Tables 4-5 and 4-6 show NPVs from a soci ety and loggers perspe ctive under increasing areas assigned to alternative la nd uses, with harvesting performe d by logging centers and closer cities. Again, in the UL scenario, annual prof its are US$ 3.2 million. Assuming the baseline scenario for the establishment and audit costs (respectively US$ 20 ha-1 and tri-annual audits of US$ 3,000 stand-1), this would generate a NPV of US$ 16.8 million from a societal perspective (i.e., considering all annual prof its from logging). This base ca se NPV declines as stands are converted to alternative land uses. With 48% converted, the societal NPV declines to US$ 6.7 million. From the loggers perspective, in whic h these agents would receive a normal profit over
66 the sawn-wood price while paying the government fo r the establishment costs, the excess profits and the audit costs, the NPV in the UL scenario would be equal to US$ 7.7 million. This would decline to US$ 2.6 million when 48% of stands are converted to other land uses. Internal rates of return (IRR) from the investments in logging fr om the loggers perspect ive are between 14% and 17%. Table 4-6 shows the same information, a ssuming that logging woul d be done by industry after migrating to cities closer to Faro. In the UL scenario and in the baseline establishment and audit costs scenario, the societ al NPV would increase to US$ 24. 6, and the loggers NPV would increase to US$ 12.5 million (IRR of 20.6%). It is also importan t to note that the share of the total economic benefits that could be captured by the government would rise to between 30.7% and 32.3% of the societal NPV. This is an expected finding since l oggers will continue to capture the same normal profits as in the origin al situation, but total economic benefits would increase. However, loggers would also have bene fits from moving to clos er cities since a higher number of species would become pr ofitable within the state forest and, as discussed before, they would be able to supply a larger share of their dema nd for roundwood from harvesting operations inside FSF. The Production Possibility Frontier As discussed before, PPF curves are valuable for social planners because they express efficient uses in the state forest, in the se nse that any zoning alte rnative below the PPF is suboptimal (i.e., it would be possible to increas e the production of one of the outputs without decreasing the other). Figure 4-9 shows PPFs ex amining tradeoffs between economic (NPV) and non-market objectives in Faro (i.e., livelihood systems or biodi versity conservation), assuming that harvests are performed from current loggi ng centers and varying the establishment and audit costs for concessions. In Figures 4-9A and B, potential stands for livelihood systems and
67 biodiversity conservation are assi gned equal weights w ithin uses. In both cases, the maximum economic value that can be reached lies under the UL scenario, genera ting a NPV of US$ 16.8 million in the baseline scenario. Lower establishment and audit costs would evidently generate higher NPVs, and higher costs (such as establishment costs of US$ 40 ha-1 and audit costs of US$ 1.0 ha-1 year-1) would generate very low or even negative NPVs. In Figure 4-9A, NPV would decrease with an increase in number of stands in livelihood systems. When 30 stands (12% of Faro) are in livelihood systems, the NP V in the baseline scenario drops to US$ 12.6 million. In Figure 4-9B, a NPV of US$ 8.0 milli on in the baseline scenario would be generated when 110 stands (43% of FSF) are assigned to bi odiversity conservation. Figures 4-9C and D shows basically the same frontiers assu ming stands not weighted equally. Opportunity Costs for Alternative Land Uses Until now, I demonstrated how the net re turns from logging can generate useful information to investigate the tradeoffs betw een revenue generation through logging and other alternative land uses that also ar e part of the management objectives in state forests. Hence, each zoning decision in public forests, assuming that logging is the activity th at most efficiently generates economic development, will involve some opportunity cost measurable in the decrease in NPVs for logging. However, as discussed in the last chapter, economic theory would say that such opportunity costs may not be uniform acr oss every management decision that social planners could take. For example, it was discussed how the first st ands assigned to an alternative land use will likely be stands with low opportunity costs in terms of lost NPV. While the social planner is increasing the proportion of the public forest in alternativ e uses, more profitable stands are assigned. If this is true, two main hypothe ses can be made about th e opportunity costs: (i) they should vary across the landscape, depending on the location of each state forest, since they influence the potential of the public land for po ssible land uses within its boundaries; (ii) they
68 should be increasing at the margin as a larger shar e of the public forest area is being assigned to alternative land uses. Figure 4-10 presents the behavior of four marginal opportunity costs (mOC) curves within FSF considering the baseline scenar io of establishment and audit costs (respectively, US$ 20 ha-1 and US$ 0.4 ha-1 year-1). All these curves were estimated using the final optimization model with connectivity constraints and harvesti ng being performed from the surrounding logging centers. Figures 4-10A and C investigate mOC as more stands are assigned to livelihood systems, assuming uniform stand weights and differe ntiated stand weights, respectively. Figures 4-10B and D shows the same results for areas assigned to biodiversity conservation. Comparisons between A and B or C and D re veal that the mOC and average opportunity costs for stands assigned to livelihood systems is considerably higher th an the opportunity costs for biodiversity conservation. This is an expe cted finding because, as discussed earlier, most stands with potential for livelihood systems are located in areas with high logging profits. Therefore, Figure 4-10A shows that mOC are around US$ 10,000 wh en few stands are assigned for livelihood systems and increase to US$ 270,00 0 when 30 stands are being assigned for this land use. Then, mOC for biodiversity conserva tion start around US$ 10,000 and can increase to US$ 170,000 when 43% of Faro is being assigned fo r this land use. Figures 4-10C and D depicts the same costs for an added unit score of liv elihood systems and biodiversity conservation. These results confirm the proposed hypothesis about the opportunity costs for alternative land uses. Social planners have to take into account, in the public forests zoning, that opportunity costs involved in the production of n on-market goods will vary spatially and also as a function of the proportion of a give n area that is being assigned fo r such uses. Implications of such findings for the planning of public forests use will be discussed in the next chapter.
69 Figure 4-1. Net economic returns for logging in Calha Norte and w ithin Faro State Forest, 2004. A) Net returns from the logging of high va lue species. B) Net returns from the logging of medium value species. C) Net returns from the logging of low value species.
70 Table 4-1. Net economic returns from logging in the Calha Norte, 2004. Proportion (%) of the forests in Calha Norte Net returns for logging (US$/m3) High value species Medium value species Low value species Unviable for logging (negative returns) 18.9 58.1 83.7 0.00 20.00 5.9 31.2 15.7 20.00 40.00 10.3 10.7 0.6 40.00 70.00 30.1 70.00 100.00 25.3 > 100.00 9.5 1 The region (131,757,900 hectares) in 2004 was 78.3% forested, 17.1% deforested, and 4.6% occupied by rivers and other water bodies. Table 4-2. Net returns from loggi ng in Faro State Forest, 2004. Proportion (%) of Faro State Forest Harvesting from the logging centers Harvesting from closer cities Net returns for logging (US$/m3) High value species Medium value species Low value species High value species Medium value species Low value species Unviable for logging (negative returns) 0.0 78.1 100 0.1 67.6 93.3 0.00 20.00 6.8 21.9 7.1 30.0 6.7 20.00 40.00 22.3 22.8 2.3 40.00 70.00 41.8 41.3 70.00 100.00 29.1 28.7 Table 4-3. Total timber volume and estimated roundwood supply (in m3 ha-1) of merchantable species by timber value class in the stands within Faro State Forest in 1976. Total volume from trees > 50 cm DBH assigned for harvesting Total volume from trees > 35 cm DBH Total volume from trees > 50 cm DBH High value species Medium value species Low value Species Mean 119.3 49.1 2.9 18.1 8.9 Std deviation 4.06 1.670.100.07 0.03 Minimum 106.9 43.9 2.6 18.0 8.9 Maximum 125.9 51.7 3.1 18.3 9.0
71 Table 4-4. Milling capacity by loggi ng center within Calha Norte, 2004. Milling capacity of roundwood in 2004 (m3) Logging Center Logging Zone Total Supplied through FMP1 Supplied through Deforestation2 Demand from public forests3 Altamira central Par 172,31600 172,316 Anapu central Par 53,68146,1376,186 1,358 Itacoatiara Amazona s 200,00014,526301 185,173 Itaituba western Par 88,4620420 88,042 Manacapuru Amazonas 28,598030 28,568 Manaus Amazonas 126,4410978 125,463 Medicilndia central Par 27,68400 27,684 Novo Aripuan Amaz onas 6,327222,88812,787 0 Oriximin western Par 22,401012,854 9,547 Placas central Par 71,47711,9826,610 52,885 Porto de Moz estuary Par 110,00095,8440 14,156 Rurpolis western Par 43,98942,523351 1,114 Santarm western Pa r 167,599107,2904,829 55,480 Sen. Jos Porfrio central Par 130,00000 130,000 Trairo western Par 197,95296,9210 101,031 Uruar central Par 168,52314,124240 154,159 Vila Km 30 western Par 19,59700 19,597 Calha Norte region 1,635,047652,23645,586 1,166,574 1 Refers to roundwood authorized for logging through forest management plans. 2 Refers to roundwood legally generated though deforestation permits from IBAMA. 3 Defined as the difference between the total milling capac ity and the volume authorized by IBAMA through forest management plans and deforestation permits.
72 Figure 4-2. Results from the unc onstrained logging scenario. A) Location of the current logging centers potentially consuming roundwood fr om the Faro State Forest, and B) Location of the harvested stands and propor tion of the total available timber volume in each stand logged from current logging cen ters and from urban centers (cities).
73 A) B) 0 20 40 60 80 100 0%7%16%20%24%36%48% L and under other us e s % of F AR O Only high value spp. High + Medium Value spp. 0 20 40 60 80 100 0%7%16%20%24%36%48% L and under other us e s % of F AR O Only High value spp. High + Medium Value spp. All spp. C) D) 0 20 40 60 80 100 120 140 160 180 0%7%16%20%24%36%48% L and under other us e s Harves ted volume (1000 m3) High + Medium Value spp. Only high value spp. 0 20 40 60 80 100 120 140 160 180 0%7%16%20%24%36%48% L and under other us e s Harves ted volume (1,000 m3) Only High value spp. High + Medium Value spp. All spp. Figure 4-3. Variation in the num ber of stands and volume logged within Faro State Forest under increasing number of stands converted to other land uses (biodi versity and livelihood systems). A) Proportion of stands logged by species value class from current logging centers. B) Proportion of stands logged by species value class from closer urban centers. C) Volume harvested by species valu e class from current logging centers. D) Volume harvested by species value class fr om closer urban centers. All results include connectivity constraints.
74 A) B) 0 20 40 60 80 100 0%11%16%20%24%36%48% L and under other us e s % s upply/clas s (bars ) 0 5 10 15 20 25 30 35% of total s upply (line) Medium Value High Value Total 0 20 40 60 80 100 0%11%16%20%24%36%48% L and under other us e s % s upply/clas s (bars ) 0 5 10 15 20 25 30 35% of total s upply (line) High Value Medium Value Low Value Total C) D) 0 20 40 60 80 100 0%11%16%20%24%36%48% L and under other us e s % demand/center(bar s ) 0 2 4 6 8 10 12% total demand (line) Oriximin P lacas S antarm Uruar Total Demand 0 4 8 12 16 20 0%11%16%20%24%36%48% L and under other us e s % of total demand F aro Nhamund bidos Figure 4-4. Variation in the pr oportion of the available merchant able timber harvested (supply), and in the proportion of the regional milling capacity met by Faro harvests, under increasing conversion of stands to othe r land uses (biodiversity and livelihood systems). A) Proportion of timber volume ha rvested (total and by timber value class) from current logging centers. B) Proportion of timber volume harvested (total and by timber value class) from closer urban cen ters. C) Proportion of the regional milling capacity being supplied by Faro to curre nt logging centers. D) Proportion of the regional milling capacity being supplied by Faro if the industry moves to closer urban centers. All results include connectivity constraints.
75 Figure 4-5. Landscape patterns formed by the optimization model for increasing number of stands converted to alternative land uses (A) with and (B) wit hout the connectivity constraint. These simulations assume that e ach stand contributes e qually within either biodiversity conservation or livelihood systems (for those stands with potential for these uses). Logging is performe d by current logging centers.
76 Figure 4-6. Landscape patterns formed by the fina l optimization model for increasing cumulative scores for livelihood systems and biodiversity conservation. Stands are weighted differently within each use. All results include connectiv ity constraints. Logging is performed by current logging centers.
77 A) B) 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 0%7%16%20%24%36%48% L and under other us e s Annual Profits (thous and US $) Loggers G overnment 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 0%7%16%20%24%36%48% L and under other us e s Annual Profits (thous and US $) Loggers G overnment C) D) 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 020406080100 C ummulative s core for livelihood s ys tem s Annual Profits (thous and US $) Loggers G overnment 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 0100200300400500 C ummulative s core for biodivers ity cons ervationAnnual Profits (thous and US $) Loggers G overnment Figure 4-7. Annual profits from logging in Fa ro for government and loggers under increasing stand conversion to alternative land uses A) Increasing stand conversion assuming stands weighted equally and logged by cu rrent logging centers. B) Increasing stand conversion assuming stands weighted equa lly and logged from closer cities. C) Increasing stand conversion to livelih ood systems assuming stands weighted differentially and logged by current logging centers. D) Increasing stand conversion to biodiversity conservation assuming sta nds weighted differentially and logged by current logging centers. All results include connectivity constraints.
78 Figure 4-8. Spatial distributi on of annual profits from loggi ng in the unconstrained logging scenario, with harvests performed from th e (A) logging centers, and (B) closer urban centers. Zero values represent stands in which logging is forbidden due to the occurrence of high slopes.
79 Table 4-5. NPVs and IRRs for logging within Fa ro State Forest from the logging centers, from the society and loggers perspectives1, under decreasing number of stands used for concessions. Concession costs and fees Societal perspective Private perspective Proportion of Faro logged Annual Profits from logging (thousand US$) Establishment costs (thousand US$) Royalties and fees NPV (thousand US$) IRR NPV (thousand US$) IRR 98.0% 3,217.7 12,500.0 28.8% 16,755.224.2%7,699.9 16.6% 92.2% 3,096.4 11,750.0 29.7% 16,451.524.8%7,448.6 16.8% 78.4% 2,725.3 10,000.0 30.6% 14,882.425.7%6,726.7 17.2% 78.0% 2,618.0 9,950.0 29.4% 13,891.524.7%6,364.1 16.8% 74.1% 2,374.3 9,450.0 30.0% 12,097.223.5%5,137.0 15.8% 62.4% 1,994.6 7,950.0 30.2% 10,149.223.5%4,255.5 15.7% 52.2% 1,482.9 6,650.0 28.0% 6,674.820.6%2,614.5 14.2% 1 The societal perspective includes the to tal annual profits from logging in the calculations. The loggers perspective discounts the values paid to government in royalties and fees (4th column). Table 4-6. NPVs and IRRs for logging within Faro State Forest from closer urban centers, from the society and loggers perspectives1, under decreasing number of stands used for concessions. Concession costs and fees Societal perspective Private perspective Proportion of Faro logged Annual Profits from logging (thousand US$) Establishment costs (thousand US$) Royalties and fees NPV (thousand US$) IRR NPV (thousand US$) IRR 98.0% 4,024.2 12,500.0 30.9% 24,641.630.7%12,492.8 20.6% 85.5% 3,709.6 10,900.0 31.7% 23,448.932.6%11,953.8 21.6% 70.6% 3,165.5 9,000.0 32.3% 20,363.833.7%10,354.1 22.2% 78.0% 3,268.2 9,950.0 31.8% 20,249.831.4%10,076.2 20.7% 74.1% 3,022.7 9,450.0 32.3% 18,438.230.5%8,895.2 20.0% 62.4% 2,585.5 7,950.0 31.6% 15,927.631.0%7,947.0 20.6% 52.2% 1,909.2 6,650.0 30.7% 10,843.827.2%5,113.9 18.2% 1 The societal perspective includes the to tal annual profits from logging in the calculations. The loggers perspective discounts the values paid to government in royalties and fees (4th column).
80 A) B) -5.00 0.00 5.00 10.00 15.00 20.00 25.00 0.0%2.0%3.9%5.9%7.8%9.8%11.8%% of F AR O under livelihood s ys tem s NPV (million US $) L ow costs Baseline Intermediate High C osts -5.00 0.00 5.00 10.00 15.00 20.00 25.00 0%4%8%16%24%31%39%43%% of F AR O under biodivers iy cons ervationNPV (million US $) Low costs Baseline Intermediate High C osts C) D) -5 0 5 10 15 20 25 020406080100C ummulative s core for livelihood s ys tem s NPV (million US $) L ow costs Baseline Intermediate High C osts -5 0 5 10 15 20 25 0100200300400500C ummulative s core for biodivers ity cons ervationNPV (million US $) Low costs Baseline Intermediate High C osts Figure 4-9. Production possibility frontiers (PPF) for competing land uses within Faro State Forest with logging being performed by fi rms located in the logging centers and different costs for establishing and aud iting concessions. The low cost simulation considers establishment costs of US$ 10 ha-1 and audit costs of US$ 0.2 ha-1 year-1. The baseline simulation consid ers, respectively, US$ 20 ha-1 and US$ 0.4 ha-1 year-1. The intermediate simulation considers, respectively, US$ 30 ha-1 and US$ 0.7 ha-1 year-1. The high costs simulation cons iders, respectively, US$ 40 ha-1 and US$ 1.0 ha1 year-1. A) NPVs generated by logging and livelihood systems, assuming stands weighted equally for livelihood systems. B) NPVs genera ted by logging and biodiversity conservation, assuming sta nds weighted equally for biodiversity conservation. C) NPVs generated by logging and livelihood systems, assuming differential weights for livelihood system s. D) NPVs generated by logging and biodiversity conservation, assuming di fferential weights for biodiversity conservation. All results include connectivity constraints.
81 A) B) 0 50 100 150 200 250 300 0%2%4%6%8%10%12% % of F AR O under livelihood s ys tem s C os t (thous and US $) Marginal opportunity cost (per stand) Average opportunity cost (per stand) 0 40 80 120 160 200 0%4%8%16%24%31%39%43% % of F AR O under biodivers ity cons ervationC os t (thous and US $) Marginal opportunity cost (per stand) Average opportunity cost (per stand) C) D) 0 20 40 60 80 100 020406080100 C ummulative s core for livelihood s ys tem s C os t (thous and US $) Marginal opportunity cost (per score unit) Average opportunity cost (per score unit) 0 5 10 15 20 25 30 35 0100200300400500 C ummulative s core for biodivers ity cons ervationC os t (thous and US $) Marginal opportunity cost (per score unit) Average opportunity cost Figure 4-10. Marginal and average opportunity costs for livelihood systems and biodiversity conservation in Faro State Forest, assumi ng harvests are performed by firms located in the logging centers. A) Opportunity costs under increasing number of stands assigned to livelihood systems, assuming uni form stand weights. B) Opportunity costs under increasing number of stands assigned to biodiversity conservation, assuming stand uniform weights. C) Oppor tunity costs under increasing cumulative score for livelihood systems, assuming differe ntiated stand weights. D) Opportunity costs under increasing cumulative scores for biodiversity conservation, assuming differentiated stand weights. All results include connectivity constraints.
82 CHAPTER 5 DISCUSSION AND FINAL REMARKS Importance of Land Use Modeling in the La rge-Scale Planning of Public Forest Use The models developed in this study proved us eful for Amazonian public forest land use planning. First, based on wood prices and typi cal variable logging co sts, a spatially-explicit model was built to estimate net economic returns. Such estimates represent, in areas within public forests designated for logging, the resource rent plus the normal private profits that can potentially be generated. The net returns from logging are also a proxy for the opportunity costs associated with allocating to other land uses, including the production of non market goods and services, such as biodiversity c onservation or livelihood systems. Second, an optimization model was developed to guide decisions regarding the desirable level of profits that should be raised through c oncessions in a public fore st, subject to a minimum area or a minimum subjective score to which no n-commodity uses are assigned. Then, the compromises between competitive land uses can be determined through the production possibility frontier (PPF), sin ce it reveals efficient altern atives for land allocation. One of the key findings of this study is th at marginal opportunity costs for non-commodity uses in Faro State Forest (FSF) are increasing when a larger pr oportion of the area is assigned for such uses. Social planners need to take into account in public forest zoning that opportunity costs vary across the landscape, and also w ithin the same area, depending on the zoning alternative chosen. Regulations establish that each public forest will have a supervisory committee responsible for making land use planning decisions, guided by an overarching forest-level management plan (Federal Decree 4340/2002). Management plans w ill be generated based on accurate surveys in each public forest, including forest inventories, important sites for biol ogical conservation and
83 tourism, and location and needs of the traditional communities dwelling in these forests. The models developed in this study do not aim to re place these surveys to establish the management plans. Instead, the models can be improved by incorporating the data generated during these surveys, improving their utility to decision-makers. It is expected that so cial planners will ofte n not be able to decide about land use in a specific public forest, since this zoning configura tion will be generated in negotiation processes among several stakeholders. Models are still us eful in this situati on to evaluate zoning alternatives in relation to land allocation efficien cy within the public forest, as demonstrated in this study. Such models become even more important in the Amazoni an context considering that, frequently, planners are locat ed in capital cities thousands of kilometers from the forests they are planning and they have very scarce information about the economic potential of such lands. At the landscape level, these models can help to determine the optimum level of timber production from public lands in a given region, considering future production trends of the logging industry and, at the same time, maximize so cietal benefits from public forests. In the example used in this study, the Calha Nort e region (~ 130 million hectares) encompasses 20 million hectares of public forests that will be pa rtially used to supply approximately 1.2 million of m3 of roundwood to the regional timber industry. In one extreme scenario, Faro (only 4.4% of the public forests in the study area) would be ab le to supply 18% of the regional milling capacity if the regional industry moved to cities close to the state forest. However, this milling capacity may increase in the study area if, due to the esta blishment of concessions, more companies are attracted to migrate due to the local stocks of raw material.
84 Concessions may have a major role in stabi lizing logging frontiers in regions where most roundwood is potentially harvested ill egally, as in Calha Norte. Ho wever, if monitor and control systems are not well implemented, concessions coul d have the perverse effect of creating a trap for the current protected areas. In this study, it was discussed how c oncessions could increase economic returns from logging by promoting the migr ation of mills to urban centers closer to FSF. This also implies that large extents of protected forests such as indigenous lands and conservation units near Faro also would beco me more profitable for illegal logging if the industry relocated to closer cities. Model Assumptions The first important assumption of the optimization model is related to land use specialization. Each stand within a given public fo rest is assumed to have an exclusive land use, which is a strong assumption, given that many areas can feasibly have overlapping uses. Tourism, for example, is described in some re gulations for forest concessions (e.g., Federal Decree 6,063, from March 20th, 2007) as a land use that is permitted within harvesting areas. However, this study assumes that management effo rts in each public forest stand will be more efficiently allocated if these stands have a unique use, decreasing th e likelihood of conflicts among different management beneficiaries. Mul tiple use is then a concept applied at the landscape-level. According to Vincent and Binkl ey (1993), an exclusive use at the stand-level typically can prevail in terms of economic effici ency over multiple stand level uses, mainly due to diminishing returns from the management effort for different activities within the same stand. The second important assumption of the model is non-declining timber stocks in the stands over time. Conveniently, since th e model of this study is static it was considered a time length of 40 years (equal to the maximum length for c oncession contracts) and harvest cycles of 35 years. Even if this assumption is wrong, it would cause only small impacts in the NPV
85 calculated because only 1/7 of the harvested area would be used for a second cycle and the potential impacts would be di scounted far in the future. However, there is a large literature discussi ng how the forest structure and the composition of merchantable species can change over time un der logging, even using RIL practices (Putz et al., 2001; Phillips et al., 2004; Zarin et al., in press). This is an important concern considering that this study found that, in the more optimistic scenario, at least 70% of the stands in FSF would be harvested only for the extraction of high value timber a loggin g pattern described in economic literature as high grading (Repetto and Gillis, 1988; Hyde and Sedjo, 1992). High grading is generally undesirable because it leads to degradation of the forest conditions, since these species can require specific post-logging silv icultural practices to regenerate (Schulze et al., 2005; Zarin et al., in pr ess). It is also possibl e that some of these spec ies would require larger disturbances provoked by more intensive harv esting to regenerate (F.E. Putz, personal communication, April 12th, 2007). Some improvements in th e optimization model that could better address these issues will be further discussed. A third important assumption is that prices we re considered static during the time length of the analyses. As mentioned befo re, wood prices used are equal to the average prices for sawnwood at the mills gate for the Brazilian domestic ma rket in 2004. In this way, the map of net economic returns can change if more firms in Calha Norte start to e xport due to the larger proportion of roundwood legally ha rvested under concessions. Currently, these firms have difficulties in exporting because they cannot prov e the legal origin of the roundwood. Second, wood markets may be affected if large extents of public forests will be conceded to timber firms in the short run, since changes in roundwood pric es as an input to timber firms will have an
86 effect in the equilibrium price and quanti ty sold for wood products (a mathematical demonstration of such effects can be seeing in Appendix A). Model Limitations The main limitation of the analyses presented in this study is related to the coarse scale of the data used in the optimization model. As men tioned earlier, due to th e lack of site specific forest inventory data for FSF, I used a timber volum e map in a coarse scale (cells of 1 km wide) to estimate the exploitable roundwood volume within Faro. I also used coarse scale maps to identify areas within FSF with potential for bi odiversity conservation or with evidences of human occupation. It is important to highlight that accurate site specific information will be collected by the government to generate the FMP for the public forests before the establishment of concessions and other land uses Forest inventories, for example, are critical steps to be executed before the establishment of concession s to guide government in the best decisions regarding minimum prices and acceptable harves ting intensity. Therefore, the optimization model developed in this study can be used by social planners to identify critical gaps of information to be collected for the generation of the FMP. Clearly, the models capabilities will be further enhanced with more accurate and site specific information collected. Economic Efficiency, Government Revenues and Distributional Issues This study presented an estimate of the share of the profits from concessions that could be captured by the government as resource rent, consid ering that loggers coul d raise normal profits over the wood prices as they usua lly have in private managed fore sts. Concessions could indeed represent an important source of funds for public agencies consider ing their current budgets. For example, in an extreme scenario modeled, rents captured by government c ould be equivalent to US$ 1.24 million year-1 in FSF. If other public forests in Ca lha Norte were able to generate the same rents per unit of area, Brazilian gove rnment could generate US$ 28 million year-1 from
87 concessions in Calha Norte region. Intuitively, this amount is al so a rough estimate of the excess profits that loggers could potentially raise from illegal logging in a scenario in which these lands continue to be uncontrolled by government. For comparison r easons, the budget for the SFB in 2007 is estimated in approximately US$ 13 million (Schulze et al., in press). Such estimates consider that government would be able to entirely capture resource rents applicable to logging in public forests. Experien ces in other countries sh ow that this assumption can be too optimistic, since governments have captured a small proportion of the total rents (Repetto and Gillis, 1988; Vincent, 1990). Lowe r pricing policies in such countries created several negative impacts in th e past, such as reduced govern mental revenues, excessive expansion of the forest sector and forest da mages provoked by logging (Repetto and Gillis, 1988; Vincent, 1990). These governments failures, associated with several cases of economic subsidies given to concessions in public lands, generated doubt s about the Brazilian governments capacity to successfully implement a forest concession system ( Merry et al., 2003; Barreto, 2004a; Barreto, 2004b). The economic literature contains many studies discussing the best strategies to maximize rent capture according to political settings and institutional arrangements. The main goal, as explained by Vincent (1990), is to convert as much of the resource rent as possible into royalty revenue, while minimizing the excess profits for l oggers and excessive damages or timber left in the harvesting sites. Efficiency, then, can be meas ure in terms of the prop ortion of the rent that is converted to royalties (Vincent, 1990). In we ak institutional and orga nizational environments like Brazil, the association of a lump sum tax (c harged over the overall c oncession) and a harvest ad valorem tax (charged over every m3 or value extracted) can be the best setting to capture
88 rents, reducing high grading and improving fore st condition (Hyde and Sedjo, 1992; Amacher, 1999; Amacher et al., 2001). Distribution is an issue that also deserves some attention. It is assumed in this study that the government is the best destin ation for resources that otherw ise would be excess profits for loggers. This does not necessarily need to be true. If government permitted excess profits, some interesting questions that can be raised are rela ted to what amount of such resources would be reinvested locally, or would be transferred to regional hous eholds (Hyde and Sedjo, 1992)1. This is a relevant question consideri ng the historical fragility and corruption of government forest agencies in the Brazilian Amazon. Another point that deserves some attention rega rding distributional issues is related to the size of firms acquiring forest concessions in the Brazilian Amazon. Brazilian regulations require that different sizes of concession units will be offered, as mentioned earlier in this study (MMA, 2005). The government must decide the best way to distribute overall benefits from logging on public lands. In one model, forests can be con ceded to small loggers a nd traditional community associations, which can have direct benefits from logging. Such an idea is strongly suggested by community forest management advocates (Lima et al., 2003; Merry et al., 2003). In a second model, harvest rights can be granted to more cap italized and economically efficient firms, which would generate higher revenue s that could be raised for the Forestry Fund (FNDF), and indirectly could be used to foment small-scale forestry in the Brazilian Amazon. 1 In addition, in a competitive concessions market higher profits must drive concession prices higher, such that governments should be able to captu re opportunistic rents (J. Alavalapati, personal communication, July 3, 2007).
89 Public Forests Sustainability and the Spatial and Temporal Zoning As any forestry law in different parts of th e world, the Brazilian Management of Public Forests Law was created to utilize Amazonian public forests in a sustainable way. Then, a question that is posed to society is what sustainable means. Za rin et al. (in press) propose that harvesting carried out on any public land > 10,000 hectares should be able to maintain the volume production at the level of the individual species logged. As mentioned before, for FSF, as well as probably many other Amazonian public lands, this goal is certainly challenging considering that large extents of these lands w ould only be harvested fo r the extraction of high value species. Furthermore, recent computer si mulations, executed with data from harvested study sites in the Brazilian Amazon, show that this sustainability goal in harvesting areas may be impossible without specific silvicultural measures to improve natural regeneration (Favrichon, 1998; Sist et al., 2003; van Gardi ngen et al., 2006; Keller et al., 2007; Valle et al., 2007). The national forestry fund could be us ed to subsidize the adoption of silvicultural treatments in harvested stands in public forests, guar anteeing species level sustainability. At the landscape-level, sustainability goals are intuitively related to the idea of spatial zoning. During the last decade, state governme nts in the Brazilian Amazon have performed macro-scale economic and ecological zoning in cons ultation with several st akeholders. A main thrust of these efforts is to identify areas that would be excl usively reserved for production of market and non-market goods based on their potential and societal preferences about such uses. However, this is not the only possible zoning co nfiguration in public la nds. A novel idea would be to implement temporal zoning. For example, taking Faro as an example, a given number of stands can be assigned for logging in the first cycle of the contracts (40 years), and can be afterwards reserved for biodiversit y conservation. At the long r un, this zoning could achieve the sustainability goals as proposed by Zarin et al. (in press). Under this idea, government could
90 even consider investing intens ively in harvesting concession s during the first cycle of the contracts and generate an endowment for the nati onal forestry fund, reserv ing logged stands for biodiversity conservation in further cycles. Thus, the fund could be used to establish silvicultural treatments in logged forest units in which natu ral regeneration may not provide full recovery at the species-level (D. Zarin, personal communication, July 3rd, 2007). Further Research Further research could investigate what can happen to profits and harvested volumes in public forests over further cycles, using dynamic optimization models. Data predicting the variation in timber stocks afte r logging could be extracted from studies modeling the post harvest timber species volume through time (Alder and Silv a, 2000; Phillips et al., 2004; Macpherson et al. forthcoming). Second, furthe r research could deal with diffe rent spatial configurations and sizes for stands destined for c oncessions, satisfying, for a given public forest, legal requirements regarding different types of timber firms as holde rs of concessions contracts. The optimization model developed can be improved with more accu rate estimates of concession establishment costs, transaction costs generated by the licensing of forest management plans and audit costs.
91 APPENDIX A HOW CHANGES IN AN INPUT PRICE CAN AFFECT SUPPLY IN A PARTIAL EQUILIBRIUM MODEL: A MATHEMATICAL DEMONSTRATION Given the following general equations re lating the supply and the demand for wood products: (,,,)SS WWQfPwvr (,,)DD WWSQfPIP Where S WQrepresents the supply quant ity of wood been produced, S WPis the wood price in the supply side, w is the wage rate, v is the capital price and r is the roundwood price. First equation describes how the quantity in the supply side is function of the prices for such inputs. As the same way, the second equation expresses how the demand quantityS WQis a function of the prices in the demand side D WP, the income I and the prices of substitute goods PS. Then, in the equilibrium point, the following co nditions need to be true: SDe WWWQQQ and SDe WWWPPP In words, the demand quantity should be equa l to the supply quantit y and therefore they should equal the equilibrium. The same is valid fo r prices. Then, we can proceed with the total differentiation of our first equations. ....SSSS SS WWWW WW S WQQQQ dQdPdwdvdr Pwvr ...DDD DD WWW WWS D WSQQQ dQdPdIdP PIP As discussed before, since SDe WWWQQQ and SDe WWWPPP the differentiations above may assume the following notation in the equilibrium point: ....SSSS ee WWWW WW S WQQQQ dQdPdwdvdr Pwvr
92 ...DDD ee WWW WWS D WSQQQ dQdPdIdP PIP Now, it is possible to inve stigate the effect that ch anges in the input price r will provoke in the supply of wood products. Evidently, demand will not be affected. Then, if 0SdwdvdIdP using the ceteris paribus assumption, and therefore 0dr .eSeS WWWW S WdQQdPQ drPdrr .0eDe WWW D WdQQdP drPdr To solve such equations, they can be expresse d in matrix form, which can be later solved through the Cramers rule: 1 0 1S e W S W S W W De WW D WQ dQ Q P dr r QdP Pdr First, Cramers rule is used to investigate the changes in the equilibrium quantity when there is a shock represented by the change in the roundwood input price. As we can see, both will interact in an opposite way. In other words, this is a mathematical demonstration that, when the input price rise, the equilibrium quantity will drop, and vice-versa. 0 0 0SD WW D e W W DS WW DS WWQQ rP dQ dr QQ PP And, at the same way, it is possible to dem onstrate that, when the input price rises, the output equilibrium price will also rise:
93 0S W e W DS WW DS WWQ r dP dr QQ PP So, ceteris paribus, changes in the roundwood input prices will provoke a shift in the supply curve and changes in the equilibrium quan tity and output prices. The magnitude of the final effect on the supply will depend upon the elasticities. Shift in the supply curve assuming rising roundwood input prices.
94 LIST OF REFERENCES Alder, D., Silva, J.N.M., 2000. An empirical co hort model for management of Terra Firme forests in the Brazilian Amazon. Fore st Ecology and Management 130, 141-157. Amacher, G.S., 1999. Government preferences a nd public forest harvesting: A second-best approach. American Journal of Agricultural Economics 81, 14-28. Amacher, G.S., Brazee, R.J., Witvliet, M., 2001. Royalty systems, government revenues, and forest condition: An application from Malaysia. Land Economics 77, 300-313. Amaral, P., Amaral Neto, M., 2005. Manejo Flor estal Comunitrio. Proc essos e aprendizagens na Amaznia Brasileira e na Amrica La tina. IMAZON and IEB, Belm (Brazil). Ando, A., Camm, J., Polasky, S., Solow, A., 1998. Species distributions, land values, and efficient conservation. Science 279, 2126-2128. Arima, E., Barreto, P., Brito, M., 2005. Pecuria na Amaznia: Tendncias e Implicaes para a Conservao Ambiental. IMAZON, Belm (Brazil). Asner, G.P., Knapp, D.E., Broadbent, E.N., O liveira, P.J.C., Keller, M., Silva, J.N., 2005. Selective logging in the Brazilian Amazon. Science 310, 480-482. Barreto, P., 2004a. Concesses florestais no Peru : reflexes para o Brasil. Technical Report. IMAZON, Belm (Brazil). Barreto, P., 2004b. Impresso da estrutura, funcio namento e performance do modelo de gesto de florestas pblicas de New South Wales Austrlia Technical Report. IMAZON, Belm (Brazil). Barreto, P., Amaral, P., Vidal, E., Uhl, C., 1998. Costs and benefits of forest management for timber production in eastern Amazonia. Forest Ecology and Management 108, 9-26. Barreto, P., Souza Jr., C., Noguern, R., Ande rson, A., Salomo, R., 2006. Human pressure in the Brazilian Amazon forests. World Resour ces Institute, IMAZON and Global Forest Watch. WRI Report. Barros, A.C., Uhl, C., 1995. Logging along the Am azon River and estuary patterns, problems and potential. Forest Ecology and Management 77, 87-105. Batten, W.H., Zhu, J.P., 2005. The effects of spatia l harvest scheduling co nstraints on the value of a medium-sized forest holding in the sout heastern United States. Southern Journal of Applied Forestry 29, 185-193. Bettinger, P., Graetz, D., Boston, K., Sessions J., Chung, W.D., 2002. Eight heuristic planning techniques applied to three increasingly difficult wildlife pl anning problems. Silva Fennica 36, 561-584.
95 BNDES. 2007. Taxas de Juros de Long Prazo. Ba nco Nacional de Desenvolvimento Econmico e Social. http://www.bndes.gov.br/produtos/download/tjlp.pdf Accessed March 15, 2007. Boyland, M., Nelson, J., Bunnell, F.L., 2004. Cr eating land allocation zones for forest management: a simulated annealing approach. Ca nadian Journal of Forest Research-Revue Canadienne de Recher che Forestiere 34, 1669-1682. Brando, A.O., Souza, C.M., 2006. Mapping unofficial roads with Landsat images: a new tool to improve the monitoring of the Brazilian Am azon rainforest. International Journal of Remote Sensing 27, 177-189. Brito, B., Barreto, P., 2005. Challenges for the En vironmental Crimes Law in Para. State of the Amazon No 5. IMAZON, Belm (Brazil). Calkin, D.E., Montgomery, C.A., Schumaker, N. H., Polasky, S., Arthur, J.L., Nalle, D.J., 2002. Developing a production possibility set of w ildlife species persistence and timber harvest value. Canadian Journal of Forest Research -Revue Canadienne de Recherche Forestiere 32, 1329-1342. Campbell, H.F., Brown, R.P.C., 2003. Benefit-Cost Analysis: Financial and Economic Appraisal using Spreadsheets, Cambridge University Press, Cambridge. Carter, D.R., Vogiatzis, M., Moss, C.B., Ar vanitis, L.G., 1997. Ecosystem management or infeasible guidelines? Im plications of adjacency restrictions for wild life habitat and timber production. Canadian Journal of Forest Research-Revue Canadienne de Recherche Forestiere 27, 1302-1310. Chomitz, K.M., Alger, K., Thomas, T.S., Orla ndo, H., Nova, P.V., 2005. Opportunity costs of conservation in a biodiversity hotspot: the case of southern Bahia. Environment and Development Economics 10, 293-312. Church, R.L., Stoms, D.M., Davi s, F.W., 1996. Reserve selection as a maximal covering location problem. Biological Conservation 76, 105-112. Clark, M.M., Meller, R.D., McDonald, T.P., 2000. A three-stage heuristic for harvest scheduling with access road network developm ent. Forest Science 46, 204-218. Costello, C., Polasky, S., 2004. Dynamic reserve site selection. Resource and Energy Economics 26, 157-174. FAO. 2005. Global Forest Resources Assessmen t 2005. Food and Agriculture Organization of the United Nations, Rome, Italy. Favrichon, V., 1998. Modeling the dynamics and sp ecies composition of a tropical mixedspecies uneven-aged natural forest: Effects of alternative cutting regimes. Forest Science 44, 113-124.
96 Harrison, S., Bruna, E., 1999. Habita t fragmentation and large-scal e conservation: what do we know for sure? Ecography 22, 225-232. Hecht, S., Cockburn, A., 1990. The Fate of the Fore st: Developers, Destroyers and Defenders of the Amazon, Harper Perennial, New York. Holmes, P., Blate, G., Zweede, J., Pereira, R ., Barreto, P., Boltz, F., Bauch, R., 2000. Financial Costs and Benefits of Reduced-Impact Loggi ng Relative to Conven tional Logging in the Eastern Amazon. USDA Forest Servi ce, TFF, IFT, IMAZON and SFRC. Hurme, E., Kurttila, M., Monkkonen, M., Hei nonen, T., Pukkala, T., 2007. Maintenance of flying squirrel habitat and timbe r harvest: a site-specific spatial model in forest planning calculations. Landscape Ecology 22, 243-256. Hyde, W.F., Sedjo, R.A., 1992. Managing tropical fore sts reflections on the rent distribution discussion. Land Economics 68, 343-350. IBGE. 1991. Geografia do Brasil: Regio Norte. Brazilian Institute of Geography and Statistics, Rio de Janeiro (Brazil). IBGE. 1996. Censo Agropecurio. Brazilian Institute of Geography and Statistics. http://www.sidra.ibge.gov.br Accessed February 2, 2005. IBGE. 1997. Diagnstico ambiental da Amaznia Legal. Brazilian Inst itute of Geography and Statistics, IBGE/DGC/Derna-Degeo-Decar, Rio de Janeiro (Brazil). IBGE. 2002. Mapa do Potencial Agrcola e Topogr afia do Brasil. Atlas Nacional do Brasil. 4th edition. Brazilian Institute of Geography a nd Statistics, Rio de Janeiro (Brazil). IBGE. 2004. Banco de Dados Agregados: SIDR A. Brazilian Institute of Geography and Statistics. http://www.sidra.ibge.gov.br Accessed February 20, 2006. IMAZON. 2006. Detalhamento do Macrozoneamen to Ecolgico Econmico do Estado do Par: reas para Produo Florestal Manejada. Re latrio Preliminar para Discusso. IMAZON and Par State Government. Secretaria Espe cial de Estado de Pr oduo, Belm (Brazil). INPE. 2006. Deforestation map of the Brazilian Am azon (2004). Brazilian Institute of Spatial Research, Projeto Prodes Digital. Monitorament o da Floresta Amaznica Brasileira por Satlite So Jos dos Campos (Brazil). INPE. 2007. Sistema DETER: Deteco de Desmatamen to em Tempo Real. Br azilian Institute of Spatial Research. http://www.obt.inpe.br/deter/ Accessed February 10, 2007. IPEA. 2002. Dados Macroeconmicos e Regionais. Institute for Applied Economic Research. http://www.ipeadata.gov.br Accessed February 10, 2007.
97 ISA, IMAZON, IPAM, ISPN, GTA, and CI. 1999. Se minrio Consulta de Macap 99: Avaliao e Identificao de Aes Prioritrias para a Conservao, Utilizao Sustentvel e Repartio dos Benefcios da Biodiversidad e na Amaznia. ISA, So Paulo, Brazil. Karlsson, J., Ronnqvist, M., Bergstrom, J., 2004. An optimization model for annual harvest planning. Canadian Journal of Forest Research -Revue Canadienne de Recherche Forestiere 34, 1747-1754. Keller, M., Asner, G.P., Blate, G., McGlocklin, J., Merry, F., Pea-Claros, M., Zweede, J., 2007. Timber production in selectively logged tropi cal forests in South America. Frontiers in Ecology and the Environment 5, 213-216. Lentini, M., Pereira, D., Celentano, D., Pere ira, R., 2005. Fatos Florestais da Amaznia 2005. IMAZON, Belm (Brazil). Lentini, M., Verssimo, A., Sobral, L., 2004. Forest Facts in the Brazilian Amazon 2003. IMAZON, Belm (Brazil). Lima, E., Merry, F., 2003. Views of Brazilian producers increasing and sustaining exports. Macqueen, D. (ed.). Growing Timber Exports: The Brazilian Tropical Timber Industry and International Markets. IIED, IIED Small a nd Medium Enterprise Series, No. 1. London, UK. Lima, E., Leite, A.A., Nepstad, D., Kalif, K., A zevedo-Ramos, C., Pereira, C., Alencar, A., Lopes Silva Jr., U., Merry, F., 2003. Florestas Fa miliares: um pacto scio-ambiental entre a indstria madeireira e a agricultura fa miliar na Amaznia. Ipam, Belm (Brazil). Loomis, J.B., 2002. Integrated Public Lands Ma nagement: Principles and Applications to National Forests, Parks, Wildlife Refuges, and BLM Lands, Columbia University Press, New York. Margulis, S., 2003. Causas do desmatamento na Amaznia Brasileira. World Bank, Braslia (Brazil). McDill, M.E., Braze, J., 2000. Comparing adjacenc y constraint formulations for randomly generated forest planning problems with four age-class distributions. Forest Science 46, 423-436. MDIC. 2007. Banco de dados de exportaes Brasile iras. Brazilian Ministry of Development, Industry and International Commerce. http://aliceweb.mdic.gov.br Accessed March 10, 2007. Merry, F., Amacher, G., Nepstad, D., Lima E., Lefebvre, P., Bauch, S., IN PRESS. Industrial development on logging frontiers in the Br azilian Amazon. Internat ional Journal of Sustainable Development. Merry, F.D., Amacher, G.S., Lima, E., Nepsta d, D.C., 2003. A risky forest policy in the Amazon? Science 299, 1843.
98 MMA. 2005. Projeto de Lei que regulamenta o uso su stentvel das florestas brasileiras e cria o Servio Florestal Brasileiro. Braz ilian Ministry of Environment. http://www.mma.gov.br/estruturas/ pnf/arquivos/resumoplgestao.pdf Accessed March 15, 2006. Montgomery, C.A., Brown, G.M., Adams, D.M., 1994. The marginal cost of species preservation the northern spotted owl. Journal of Environmental Economics and Management 26, 111-128. Murray, A.T., 1998. Route planning for harvest si te access. Canadian Journal of Forest Research-Revue Canadienne de Recherche Forestiere 28, 1084-1087. Murray, A.T., 1999. Spatial restrictions in ha rvest scheduling. Forest Science 45, 45-52. Murray, A.T., Snyder, S., 2000. Introduction: Spatia l modeling in forest management and natural resource planning. Forest Science 46, 153-156. Naidoo, R., Adamowicz, W.L., 2006. Modeling opport unity costs of conser vation in transitional landscapes. Conservation Biology 20, 490-500. Naidoo, R., Balmford, A., Ferraro, P.J., Po lasky, S., Ricketts, T.H., Rouget, M., 2006. Integrating economic costs into conservati on planning. Trends in Ecology & Evolution 21, 681-687. Nalle, D.J., Montgomery, C.A., Arthur, J.L., Po lasky, S., Schumaker, N.H., 2004. Modeling joint production of wildlife and timber. Journal of Environmental Economics and Management 48, 997-1017. Nepstad, D., Verssimo, A., Alencar, A., Nobre, C., Lima, E., Lefebvre, P., Schlesinger, P., Potter, C., Moutinho, P., Mendoza, E., Co chrane, M., Brooks, V., 1999. Large-scale impoverishment of Amazonian forests by logging and fire. Nature 398, 505-508. Onal, H., Briers, R.A., 2002. Incorporating spatial criteria in optimum rese rve network selection. Proceedings of the Royal Society of L ondon Series B-Biological Sciences 269, 2437-2441. Phillips, P.D., de Azevedo, C.P., Degen, B., Thom pson, I.S., Silva, J.N.M., van Gardingen, P.R., 2004. An individual-based spatially explicit simulation model for strategic forest management planning in the eastern Amazon. Ecological Modeling 173, 335-354. Pinedo-Vasquez, M., Zarin, D.J., Coffey, K., Pa doch, C., Rabelo, F., 2001. Post-boom logging in Amazonia. Human Ecology 29, 219-239. Polasky, S., Camm, J.D., Garber-Yonts, B., 2001. Se lecting biological reserv es cost-effectively: An application to terrestrial vertebrate conservation in Oregon. Land Economics 77, 68-78. Polasky, S., Nelson, E., Lonsdorf, E., Fackler, P., Starfield, A., 2005. Conserving species in a working landscape: Land use with biol ogical and economic obj ectives. Ecological Applications 15, 1387-1401.
99 Putz, F.E., Blate, G.M., Redford, K.H., Fi mbel, R., Robinson, J., 2001. Tropical forest management and conservation of biodivers ity: an overview. Conservation Biology 15, 720. Rankin, J.M., 1985. Forestry in the Brazilian Amaz on. In: Prance, G. and T. Lovejoy (Eds.), Amazonia, Key environmental series. Pergamon Press, Oxford, pp. 369-392. Repetto, R., Gillis, M., 1988. Pub lic policies and the misuse of forest resources. World Resources Institute, University of Cambridge. Richards, E.W., Gunn, E.A., 2000. A model and ta bu search method to optimize stand harvest and road construction schedul es. Forest Science 46, 188-203. Rohweder, M.R., McKetta, C.W., Riggs, R.A., 2 000. Economic and biological compatibility of timber and wildlife production: an illustrative use of production possibilities frontier. Wildlife Society Bulletin 28, 435-447. Rothley, K.D., 1999. Designing bioreserve networ ks to satisfy multiple, conflicting demands. Ecological Applications 9, 741-750. Sabatini, M.D., Verdiell, A., Ig lesias, R.M.R., Vidal, M., 2007. A quantitative method for zoning of protected areas and its spatial ecological implications. Journal of Environmental Management 83, 198-206. Sabogal, C., Lentini, M., Pokorny, B., Silva, J. N., Zweede, J., Verissimo, A., Boscolo, M., 2006. Manejo Florestal Empresarial na Amaznia Brasileira: Restries e Oportunidades. Relatrio Sntese. Cifor, IMAZON, Embrapa e IFT, Belm (Brazil). SBS. 2007. Cadastro Nacional de Florests P blicas. Brazilian S ilviculture Society. http://www.sbs.org.br Accessed July 10, 2007. Schneider, R., Arima, E., Verssimo, A., Barre to, P., Souza Jr., C., 2000. Sustainable Amazon: limitations and opportunities fo r rural development. World Bank and IMAZON, Braslia (Brazil). Schulze, M., Grogan, J., Vidal, E., IN PRESS. Technical Challenges to Sustainable Forest Management in Concessions on Public La nds in the Brazilian Amazon. Journal of Sustainable Forestry. Schulze, M., Vidal, E., Grogan, J., Zweede, J ., Zarin, D., 2005. Madeir as nobres em perigo: Prticas e leias atuais de manejo florestal no garantem explorao sustentvel. Cincia Hoje 36, 66-69. Silva, N., 1997. Avaliao de Planos de Manejo Fl orestal Sustentvel na Regio de Paragominas, Par. Embrapa, Belm (Brazil).
100 Sist, P., Sheil, D., Kartawinata, K., Priyad i, H., 2003. Reduced-impact logging in Indonesian Borneo: some results confirming the need fo r new silvicultural prescriptions. Forest Ecology and Management 179, 415-427. Sobral, L., Verssimo, A., Lima, E., Azeve do, T., Smeraldi, R., 2002. Acertando o Alvo 2: Consumo de Madeira Amaznica e Certifica o Florestal no Estado de So Paulo. IMAZON, Imaflora and Amigos da Terra, Belm (Brazil). Steege, H.T., Pitman, N., Sabatier, D., Castella nos, H., Van der Hout, P., Daly, D.C., Silveira, M., Phillips, O., Vasquez, R., Van Andel, T ., Duivenvoorden, J., Oliveira, A.A., EK, R., Lilwah, R., Thomas, R., Van Essen, J., Baider C., Maas, P., Mori, S., Terborgh, J., NezVargas, P., Mogolln, H., Morawetz, W., 2003. A spatial model of tree a-diversity and tree density for the Amazon. Biodivers ity and Conservation 12, 2255-2277. Stone, S.W., 1997. Economic trends in the timber industry of Amazonia: Survey results from Para State, 1990-1995. Journal of Developing Areas 32, 97-121. Stone, S.W., 1998a. Evolution of the Timber I ndustry Along an Aging Frontier: The Case of Paragominas (1990-95). World Development 26, 433-448. Stone, S.W., 1998b. Using a geographic information sy stem for applied policy analysis: the case of logging in the Eastern Am azon. Ecological Economics 27, 43-61. Uhl, C., Barreto, P., Verssimo, A., Vidal, E., Amaral, P., Barros, A.C., Souza, C., Johns, J., Gerwing, J., 1997. Natural resource management in the Brazilian Amazon. Bioscience 47, 160-168. Valle, D., Phillips, P.D., Vidal, E., Schultz, M ., Grogan, J., Sales, M., van Gardingen, P.R., 2007. Adaptation of a spatially e xplicit tree-based growth a nd yield model and long term comparison between reduced impact and c onventional logging in Eastern Amaznia, Brazil. Forest Ecology and Management 243, 187-198. van Gardingen, P.R., Valle, D., Thompson, I., 2006. Evaluation of yield regulation options for primary forest in Tapajos National Forest, Brazil. Forest Ecolo gy and Management 231, 184-195. Varvarezos, D.K., Grossmann, I.E., Biegler, L.T., 1992. An outer-approximation method for multiperiod design optimization. Industrial & Engineering Chemistry Research 31, 14661477. Verssimo, A., Barreto, P., Tarifa, R., Uhl, C., 1995. Extraction of a high-value natural-resource in Amaznia the case of mahogany. Fo rest Ecology and Management 72, 39-60. Verssimo, A., Celentano, D., Souza Jr., C., Sa lomo, R., 2006. Zoneamento de reas para Manejo Florestal no Par. Estado da Amaznia 8. IMAZON, Belm (Brazil). Verssimo, A., Cochrane, M.A., Souza, C., Salo mo, R., 2002a. Priority areas for establishing national forests in the Brazilian Amazon. Conservation Ecology 6, 4.
101 Verssimo, A., Lima, E., Lentini, M., 2002b. P los Madeireiros do Estado do Par. IMAZON, Belm (Brazil). Verssimo, A., Smeraldi, R., 1999. Acertando o alvo: consumo de madeira no mercado domstico brasileiro e promoo da certificao florestal. IMAZON, Imaflora and Amigos da Terra, Piracicaba (Brazil). Verssimo, A., Smeraldi, R., Azevedo, T.,. 2005. Fo rest Certification in Brazil: advances, innovations and challenges. In: Burger, D ., J. Hess, and B. Lang (Eds.), Forest Certification: An innovative instrument in th e service of sustainable development?. GTZ, Eschborn, pp. 207-217 Verssimo, A., Souza Jr., C., Amaral, P., 2000. Id entificao de reas com Potencial para a Criao de Florestas Nacionais na Amaznia Le gal. Brazilian Ministry of Environment, Braslia (Brazil). Verssimo, A., Souza Jr., C., Stone, S., Uhl, C., 1998. Zoning of timbe r extraction in the Brazilian Amazon. Conservation Biology 12, 128-136. Vincent, J.R., 1990. Rent capture and the feasibility of tropical forest management. Land Economics 66, 212-223. Vincent, J.R., 1997. Resource depletion and economi c sustainability in Malaysia. Environment and Development Economics 2, 19-37. Vincent, J.R., Binkley, C.S., 1993. Efficient multiple-use forestry may require land-use specialization. Land Economics 69, 370-376. Viswanathan, J., Grossmann, I.E., 1999. Discre te and Continuous Optimizer (DICOPT). Engineering Design Research Center (EDRC), Carnegie Mellon University. Zarin, D., Pereira, V.F.G., Raffles, H., Rabe lo, F.G., Pinedo-Vasquez, M., Congalton, R.G., 2001. Landscape change in tidal floodplain near the mouth of the Amazon River. Forest Ecology and Management 154, 383-393. Zarin, D.J., Schulze, M.D., Vidal, E., Lentini, M., IN PRESS. Beyond Reaping the First Harvest: Management Objectives for Timber Producti on in the Brazilian Amazon. Conservation Biology.
102 BIOGRAPHICAL SKETCH Marco A. W. Lentini was born in So Paulo, th e largest Brazilian city, in southeastern Brazil. From 1995-1999, he completed his undergrad uate studies in Forestry at ESALQ, the Agriculture School of the University of So Pa ulo, in Piracicaba. During his undergraduate experience, he participated in research projects relating to th e ecology and recovery of forest fragments in dry Atlantic forests in So Paulo State. In 2000, he moved to the Brazilian Amazon to work as an assistant researcher at the Am azon Institute for the People and the Environment (IMAZON), an independent resear ch institute located in Belm, in the State of Par (www.imazon.org.br ). The bulk of his work at IMAZ ON relates to economics and policies focused on timber logging and forest certifica tion. In 2004, Marco served as the field coordinator of a wide survey of the timber industry, in which 680 mill owners and managers, located in 82 logging centers of the Brazilian Amazon, were interviewed. Another of his primary assignments at IMAZON was carried out in 2004-5, when he worked as Executive Secretary of the Certified Forest Producers Association (PFCA, www.pfca.org.br ). Marco moved to the United States in August 2005 to pursue a Master of Science degree at the University of Florida, as a scholar in the Am azon Conservation Leadership Initiative (ACLI) program, supported by the Gordon and Betty Moore Foundation.