PUBLIC POLICY AND SPATIAL VARIATION IN LAND USE AND LAND COVER IN THE SOUTHEASTERN PERUVIAN AMAZON By ANDREA B. CHAVEZ A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT O F THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2009 Andrea B. Chavez
To the memory of my father, Jorge Francisco Chavez Pinedo Wherever he is now, I am honored to ma ke him proud.
ACKNOWLEDGEMENTS This dissertation would not have been possible without the constant motivation and belief in my accomplishments of my advisor, Mike Binford, and the rest of my committee: Stephen Perz, Marianne Schmink, and Jane Southworth. I could not have been more grateful in working with each of them and learn ing from them. I am very indebted due to how each member of my committee supported me, corrected me, encouraged me, and finally motivated me to finish the manuscript. In Peru, I am grateful to the Proyecto Especial Madre de Dios, in particular Froilan Barrientos, Santos Ikeda, and Omar Rengifo, for logistical support, an institutional base, and a place to work while in Iberia. Omar Rengifo provided his time and teammates from the Ag roforestry and Environment Program, to accompany me while in the field. Without his institutional support, I would not have had access to certain farms and interview accessibility, and the approach to local farmers would have been so much harder. In Iberia I further had the generous assistance of Honorato Pita Barra and Jorge Moreno, who provided precious historical information and Gilmer Gibaja provided historical information on the forestry sector around Iberia. I would like to thank the Agencia Agraria and INRENA in Iberia for valuable data at the district level. In Iapari, I am indebted to the support of the Cardozo family who made Iapari my second home while abroad and made me feel safe. In special, I would like to thank Deto Cardozo Mouzully, Abraham Cardozo Mouzully, Sandro Cardozo Mouzully, Alfonso Cardozo Mouzully, Veronica Cardozo Mouzully, and don Alberto Cardozo Segura. In Puerto Maldonado, I had the support from the Peruvian Titling Agency, PETT, Programa Especial de Titu lacin de Tierras, wh o assisted me with valuable cadastral maps and geographic information data. I am grateful to the Region Madre de Dios, who made secondary data available. In Lima, the
National Institute of Natural Resources, INRENA, provided me with helpful geographic inf ormation data and metadata. Throughout my field work years, I had the constant support and assistance of Leonor Mercedes Perales Yabar and Raul Flores. I have no words in thanking both of them in making each of my field work trips enjoyable, safe, fun, ve ry productive, and for the most part eye opening. I learned so much from them. I am also thankful to Angelica Almeyda and Eben Broadbent for exchanging information and experiences about our study area. Although our field visits did not always coincide duri ng the same time period I knew that I had their support I am gratefully thankful to Foster Brown, who grounded the base for my interest in my study area. My fieldwork and the purchase of sate llite images could not have been possible without the internati onal dissertation fellowship from the Compton Foundation, Environment and Sustainable Development, the Tropical and Development Conservation Research Fellows hip, and the Tropical and Development Conservation Field Research Grant. I received a grant from th e University of Florida Map & Imagery Library to cover the purchase of satellite imagery. I am further thankful to NSF HSD Project Infrastructure Change, Human Agency, and Resilience in Social -Ecological Systems project ( # 0527511) for providing me with additional satellite imageries. At the University of Florida, I have to thank the Department of Geography, who despite my long absences as a student due to professional work or field work, welcomed me every semester and made me feel part of the exceptional team of academic professionals and college students. I am indebted to Stephen Perz, who involved me with field study MAP projects and made several post -fieldwork trips available. Overall, these experiences added up to my growing
responsibility in bringin g back results to my study area and contributing with local counterparts in exchanging information. The International Center provided me with outstanding support. In special, I am tremendously grateful to the outstanding support of Debra Anderson. The Land Use and Environmental Change Institute enabled me permanent access to state -of the art computer facilities, a stable desk, technical support, and valuable advice from peer students. T he possibility to work in there increased my confidence, complemented my knowledge, endured my strength in keeping up with the writing, and overall strengthened friendship and soothed academic pain. I would like to thank Christian Russell, Amy Duchelle, Karla Rocha, Natalia Hoyos, Forrest Stevens, Matt Marsik, Lin Cassidy, Jac queline Hall, Rosa Cossio, Tracy van Holt Valerio Gomes, and Kwadwo Owusu I am further indebted to the friendship and continuous support of Patricia Ortiz Gudemann, Constanze Devine, Beatriz Schippner, Mariana Valqui, Claudia Consiglieri, and Sean Chenow eth. Although my family has been far away and was not able to experience closely the process of going through a foreign doctoral education, they unconditionally supported me and waited as long as I had to take. I am fo r ever grateful to my mother, Karin Chavez, and my sisters, Claudia Chavez de Lederbogen, Silvia Chavez de Fehlinger, and Isabel Chavez de Groos. At home, I owe my every day strength and motivation to Tony Dunn. I am thankful for his support, friendship, and love.
TABLE OF CONTENTS page ACKNOWLEDGEMENTS ................................................................................................................. 4 LIST OF TABLES ................................................................................................................................ 9 LIST OF FIGURES ............................................................................................................................ 11 ABSTRACT ........................................................................................................................................ 12 CHAPTER 1 INTRODUCTION ....................................................................................................................... 14 1.1 Statement of the Problem .................................................................................................. 15 1.2 Conversion of Small Agricultural Areas in a Newly Emerged Frontier Region in Southeastern Peru: A Trajectory of Policy Intervention Scenarios and Land Use Dynamics ........................................................................................................................... 17 1.3 Future Land Use Plans along the Inter Oceanic Highway in Southeastern Peru .......... 17 1.4 Landscape Dynamics of Amazonian Deforestation between 1986 and 2007 in Southeastern Peru: Policy Dri vers and Road Implications ............................................. 18 1.5 Relevance of Study ........................................................................................................... 19 2 CONVERSION OF SMALL AGRICULTURAL AREAS IN A NEWLY EMERGED FRONTIER REG ION IN SOUTHEASTERN PERU: A TRAJECTORY OF POLICY INTERVENTION SCENARIOS AND LAND USE DYNAMICS ........................................ 20 2.1 Introduction ....................................................................................................................... 20 2.2 Backgro und ........................................................................................................................ 22 2.2.1 Linkages of Government Policies to Landuse/land -cover change ................... 22 2.2.2 State Policies in the Peruvian Amazon ................................................................ 24 2.3 Study Case, Methods, and Data ........................................................................................ 26 2.3.1 Regional Setting .................................................................................................... 26 2.3. 2 Field Data Design and Sampling ......................................................................... 29 2.3.3 Land -Use Change Indicators ................................................................................ 30 2.3.4 Dependent Variables ............................................................................................. 30 2.3.5 Independent Variables .......................................................................................... 31 2.4 Findings ............................................................................................................................. 35 2.4.1 Forest Cover .......................................................................................................... 36 2.4.2 Crop Cultivation ................................................................................................... 36 2.4.3 Pasture Cultivation ................................................................................................ 36 2.4.4 Fallowing and Regrowth ...................................................................................... 37 2.5 Discussion and Conclusion ............................................................................................... 38 2.6 Notes .................................................................................................................................. 40
3 FUTURE LAND USE PLANS ALONG THE INTER OCEANIC HIGHWAY IN SOUTHEASTERN PERU .......................................................................................................... 48 3.1 Introduction ....................................................................................................................... 48 3.2 Complex Causality of Roads and Deforestat ion ............................................................. 49 3.3 Historical Background ...................................................................................................... 53 3.3.1 Infrastructure Improvements in the Study Region.............................................. 53 3.3.2 Connectivity and Cooperation ............................................................................. 55 3.4 Study Case, Methods and Data ......................................................................................... 57 3.4.1 Study Area ............................................................................................................. 57 3.4.2 Dependent Variables ............................................................................................. 58 3.4.3 Independent Variables .......................................................................................... 59 3.5 Findi ngs and Discussion ................................................................................................... 64 3.6 Conclusion ......................................................................................................................... 69 3.7 Notes .................................................................................................................................. 72 4 LANDSCAPE DYNAMICS OF AMAZONIAN DEFORESTATION BETWEEN 1986 AND 2007 IN SOUTHEASTERN PERU: POLICY DRIVERS AND ROAD IMPLICATIONS ......................................................................................................................... 81 4.1 Introduction ....................................................................................................................... 81 4.2 Study Area ......................................................................................................................... 86 4.3 Methods ............................................................................................................................. 88 4.3.1 Remote Sensing Data ........................................................................................... 89 4.3.2 Pattern Metrics ...................................................................................................... 91 4.4 Results and Discussion ..................................................................................................... 95 4.4.1 LCC from 1986 to 2007 ....................................................................................... 95 4.4.2 Landscape Metrics ................................................................................................ 97 4.4.3 Class Metrics ......................................................................................................... 99 4.4.4 Implications for LULCC .................................................................................... 102 4.5 Conclusion ....................................................................................................................... 104 5 CONCLUSION ......................................................................................................................... 114 5 .1 Significance of Findings ................................................................................................. 115 5 .2 Research Considerations and Future Work ................................................................... 116 LIST OF REFERENCES ................................................................................................................. 118 BIOGRAPHICAL SKETCH ........................................................................................................... 131
LIST OF TABLES Table page 2 1 Twenty -year frontier policy timeline .................................................................................... 42 2 2 Descriptive statistics for land use change indicators as total of net area and aggregated by location, farm households, Iapari -Iberia, 2004* ........................................ 44 2 3 Descriptive statistics for ex planatory variables, household background, biophysical endowment, and policies, aggregated statistics for Iapari (INA) and Iberia (IBE), expected landuse outcomes for forest (F), regrowth (R), crop (C), and pasture (P), + (expected increase), (expected increase or decrease), and (expected decrease), farm households, Iapari Iberia, 2004 .................................................................................. 45 2 4 Land use regression models regressed on background, location, institutional diversity, policy pe riod indicators and specific policy indicators, farm households, Iapari -Iberia, 2004. ............................................................................................................... 46 3 1 Population statistics for the districts Iapari Iberia ........................................................... 74 3 2 Mean descriptive statistic for future land use plans and their statistical significance of a relationship, farm households, Iapari -Iberia, 2004 ..................................................... 76 3 3 Mean descr iptive statistics for explanatory variables, household background, biophysical endowment, and policies, expected outcomes for land use plans: increase cattle (IC), reforestation (R), agriculture mechanization (AM), fish farming (F), and sell farm (SF), far m households, + (positive effects), (either positive or negative effects), and (negative effects), farm households, Iapari Iberia, 2004. ......................... 77 3 4 Future landuse plan binomial logistic regr ession models regressed on background, location, institutional diversity, and specific policy period indicators, farm households, Iapari Iberia, 2004. .......................................................................................... 78 3 5 Percentage of positive farmers p erception toward road paving followed by major critical concerns mentioned by surveyed farmers in regard to the road paving, farm households, Iapari Iberia, 2004. .......................................................................................... 80 4 1 Landsat TM, ETM+, and ASTER scenes of the study area linked to specific policies and expected land cover fragmentation outcomes. ............................................................ 108 4 2 Changes in landscape pattern indices within the boundaries of agricultural a rea and timber concession areas between 1986 and 2007 in the road axis Iapari Iberia, Southeastern Peru ................................................................................................................. 111
4 3 Class pattern indices for agricultural areas (AG): PLAND (percentage of land a rea), NP (number of patches), MPS (mean patch size), LPI (largest patch index), ENN_MN (mean Euclidean nearest neighbor distance distribution), and IJI (interspersion juxtaposition index. ...................................................................................... 112 4 4 Class pattern indices for timber concession areas (CONC): PLAND (percentage of land area), NP (number of patches), MPS (mean patch size), LPI (largest patch index), ENN_MN (mean Euclidean nearest neighbor distance distribution), and IJI (interspersion juxtaposition. ................................................................................................ 113
LIST OF FIGURES Figure page 2 1 Study site districts: Iapari Iberia, Madre de Dios, Peru. ................................................. 43 3 1 Theoretical framework based on policies adopted by farmers ............................................ 74 3 2 Composite RGB 231 ASTER image from 30 June 2006 showing the study site road axis: Iapari Iberia, Madre de Dios, Per, the two towns, the Inter Oceanic Highway, the cadastre of titled farms, and the farms that were sampled for the interviews. ............................................................................................................................... 75 4 1 Study area in the province of Ta huamanu, Madre de Dios Region, Southeastern Peru. 107 4 2 Land -cover change within the boundary of agricultural areas from 1986 to 2007. ......... 109 4 3 Land -cover change within the boundary of timber concession areas from 1986 to 2007. ...................................................................................................................................... 110
Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulf illment of the Requirements for the Degree of Doctor of Philosophy PUBLIC POLICY AND SPATIAL VARIATION IN LAND USE AND LAND COVER IN THE SOUTHEASTERN PERUVIAN AMAZON By Andrea B. Chavez August 2009 C hai r: Michael W. Binford Major: Geography There has been longstanding debate about the importance of government policies for land use decision-making in tropical frontier areas. Research has shown that past and current polic ies tied to particular historical and e conomic circumstances have altered decision -making processes and caused land cover conversion This dissertation therefore examines the question of the effects of policy regimes and specific policies on land use and deforestation in the tropical forest fro ntier of Southeastern Peru. This area exhibits high biodiversity, and land settlement there has seen changes over time in the context of policy shifts. However, few studies have focused systematically on the relationship between agricultural policies and p olicy regimes and land use/land cover change through time. This research is a land use / land -cover change study of the past 20 years (19862006) that explores t he following research hypotheses : (a) land cover changes in the face of shifting conditions or st ructures such as road accessibility and policies; (b) these shifts drive farmers to modify land as revealed through deforestation sequences as represented through the conversion and reversal of primary forest, crops and pasture regrowth and built/non-for est through time. The study included a comparative analysis of farmers wh o did and did not adopt agricultural policy incentives such as fish farming, agricultural mechanization, and government -sponsored cattle insemination I use multivariate statistical m odels to evaluate
the importance of policy variables for land use outcomes adopted by local landholders and combined methods used in landuse / land -cover classification and derived from remote sensing processing techniques T argeted household surveys provided an analysis of how these occurrences influenced their livelihood decision -making processes The evaluation of the changing driving forces focused on how policies influenced t he outcome of economic processes The results show that distinct policies are associated with different patterns of land use/land cover change. For example, policies that favored cattle expansion influenced an increase in pasture areas. P olicies associated with credit availability facilitated the expansion of agriculture areas, incr easing deforestation. The results of the dissertation have implications not only for understanding tropical deforestation and land use/land cover change, but also for policymaking in Peru and other countries that share the Amazon and other tropical rainfor ests.
CHAPTER 1 INTRODUCTION The dissertation focuses on the importance of government policies for land use decisionmaking in tropical frontier areas. I present an empirical analysis of the role of government policies in land use changes over two decad es of frontier expansion in Peru and evaluate future land use plans among farmers as rela ted to road paving. The analyses of government policies and road trajectories are then linked to a longitudinal remote sensing analysis, utilizing satellite images acq uired at dates to detect changes due to each policy and based on land use/land -cover change (LUL CC) methods. Th e research is presented as three separate papers, presented in publication style for submission to academic journals. Each paper is therefore a s tand alone document, addressing different aspects of the research problem described below. The first, Conversion of Small Agricultural Areas in a Newly Emerged Frontier Region in Southeastern Peru: A Trajectory of Policy Intervention Scenarios and Land Us e Dynamics deals with the question of the effects of policy regimes and specific policies on deforestation and ultimately land use in the tropical forest frontier of Southeastern Peru. I intend to submit this paper to World Development Journal. The second paper, Future Land Use Plans along the Inter Oceanic Highway in Southeastern Peru examines how transportation networks and policies have influenced land use among farmers along the Inter Oceanic Highway in Southeastern Peru. I intend to submit this pape r to the journal Global Environmental Change. The third paper, Landscape Dynamics of Amazonian Deforestation between 1986 and 2007: Policy Drivers and Road Implications focuses on landscape changes by calculating changes of landscape and class metrics and linked to socio-economic conditions. I intend to submit this paper to the journal Forest Ecology and Management.
1 1 Statement of the Problem The approach adopted here examines the degree to which policies constitute a significant part of the context for land use decisions in highly dynamic tropical frontier regions. I focus on the case of the Peruvian Amazon, where policies have shifted over time among policy regimes and where specific policies have frequently been adopted by some but not all landholders These considerations motivate two key questions. First, do distinct policy regimes associated with particular presidential administrations result in distinct LULCC patterns among landholders? And second, do adopters of specific policies exhibit different LULCC change patterns than nonadopters? The question of policies and LULCC is of particular importance for frontier regions, which in Latin America often retain tropical forest cover and have also been the target of state policies for frontier developme nt and regional integration (Andersen & Reis, 1997; Kaimowitz & Smith, 2001; Schmink & Wood, 1992). Wide -ranging discussions of drivers of tropical deforestation have pointed to a number of causes, including government policies ( Binswanger, 1991; Kaimowitz & Angelsen 1998; Geist & Lambin 2003; Liverman et al. 1998; Liverman & Vilas, 2006; Fox et al., 2003; Wood & Porro, 2002; McCracken et al. 2002; Rudel 2005; Walsh & Crews -Meyer, 2002; Turner et al 2004; Andersen et al. 2002; Lambin & Geist, 2006). Despite valuable contribution s by various discipline s much work remains to be done to understand the role and importance of socioeconomic processes through government policies and their linkage s with deforestation as they vary in different geographic se ttings. Of particular importance are variations in government policies and their impact on land use (Mather, 2006:376). For example, comparisons of land use in time periods associated with distinct policy
regimes1 would help to account for temporal changes in deforestation and land use. However, d iscussions of policies and deforestation which feature the analysis of time -series data have cer tain methodological limitations, because changes in deforestation over time can result from many factors changing at the same time. There is therefore a need for a complementary approach to analysis of policy effects on land use. One alternative is to compare land use among people who took advantage of a policy to those who did not. Such a comparison would provide a focus ed test of policy impacts on land use. The study of LULCC has revealed changes in deforestation along roads, with dramatic changes in forested areas occurring in combination with road building (Laurance et al. 2001; Nepstad et al. 2001; Carvalho et al. 2001). According to LULCC reports, t ropical forests constitute a region experiencing rapid change, especially in contexts where roads are present (Nepstad et. al ., 1999, 2001; Wood & Skole 1998; Mertens et al. 2002; Cropper et al. 2001; Wood & Porro, 2002) making it crucial to further analyze how people link landuse/land -cover (LULC) decision -making processes with road paving. T his dissertation examines a segment of the paving of the Inter O ceanic Highway, a last link in a major transportation developm ent project initiative that should open the product exchange between Brazil, Peru and other export markets, and promote economic development into Peruvian regions that have lacked investment and economic opportunities in the past. This initiative has raise d widespread questions about how the l ocal population will re act to changes in land processes By linking transportation networks and policies and quantifying deforestation trends in the Peruvian road axis Iapari Iberia, I address a long -debated controve rsy among development supporters and environmentalists regarding if and how roads have caused changes in tropical 1 Here policy regimes are defined by government policies characterized within a specific time period.
forests and their market access (Garcia, Luna & Boggio, 2008) Southeastern Peru represents thus an important region for studying global integration and processes related to global climate change 1.2 Conversion of Small Agricultural Areas in a Newly Emerged Frontier Region in Southeastern Peru: A Trajectory of Policy Intervention Scenarios and Land Use Dynamics The first paper, presented in Chap ter 2, features agrarian policies that changed over a 20 year period, represented by three distinct policy regimes, as well as specific policy incentives. The analysis considers policies associated with specific time periods of particular presidential admi nistrations and their distinct policies. The analysis focuses on comparisons of deforestation over time in light of different policy regimes, and on land use outcomes among adopters and non adopters of several specific policies in terms of their landuse portfolios. I employ statistical models to control for other socio -economic variables not related to the specific policies at the level of land users and consider the effects of specific policies on forest, pasture, crop, and regrowth hectares The findings confirm important effects of specific policies, some associated with distinct policy regimes. The effects of specific policies have implications not only for our understanding of tropical deforestation, but also for policymaking in Peru and other countri es that share the Amazon and other tropical rainforests. 1.3 Future Land Use Plans along the Inter -Oceanic Highway in Southeastern Peru In Chapter 3, I evaluate the importance of various policies and how farmers respond to them regarding future landuse pl ans in the context of recent paving of the Inter Oceanic Highway. I focus on how socioeconomic conditions have complemented and influenced land use decision -making processes among farmers at specific timeframes and between the two towns of Iapari and Ibe ria along the Inter Oceanic Highway in Southeastern Peru. Based upon semi structured surveys the article evaluates farmers future land use plans in light of policies, which
have impacted various parts of the study site in different ways. I use logistic re gression to estimate the probability of farmers engaging in specific activities, such as fish farming, government -sponsored cattle insemination or agricultural mechanization, based on socio economic indicators and under different road -policy planning outc omes. The results indicate that farmers respond differently to anticipate benefits of road paving based on pre -existing policies which vary from place to place and time to time. The paper contributes to a n increasing literature questioning issues of road c ausality and incorporating information relevant to global change impacts. 1.4 Landscape Dynamics of Amazonian Deforestation between 1986 and 2007 in Southeastern Peru: Policy Drivers and Road Implications This paper describes and explains the landscape cha nges between 1986 and 2007 in a frontier region located in Madre de Dios region in Southeastern Peru, which is currently undergoing increasing landscape transformation due to major infrastructure projects and has experienced varied periods of deforestation conditioned by distinct periods of socio-economic policies. I describe landscape dynamics, especially deforestation attributed to logging and agricultural activities that were responses to policy incentives in the districts of Iapari and Iberia, Southea stern Peru. Five -year interval Landsat TM/ETM+ /ASTER images from 1986 to 2001 and bi annual images from 2001 to 2007 were classified into five land-cover types: forest, crops and pasture, regrowth, built/non-forest, and cloud/haze producing a time series o f land cover maps. Landscape changes were examined by calculating landscape and class metrics related to size, density, connectivity, configuration, and the classified land -cover types, which were then linked to socio-economic processes. The results show t hat LULC dynamics are dependent on logging and agricultural activities emanating from national and regional policies. Landscape dynamics indicated a fluctuating conversion from forest to crops and pasture,
regrowth, and built/non-forest, with regrowth, crops and pasture and forest reverting back at times when policies were absent. T here was a continuous fluctuation in metric values, corresponding to dramatic changes in the landscape at the class and landscape level in short intervals (2 -year). I nterpreting landscape patterns in terms of soc io -economic processes remains a difficult task. However, complementing the assessment of landscape dynamics by incorporating a socio -economic component considerably improves our understanding of the impacts of complex poli cy frameworks on the environment. 1. 5 Relevance of Study Overall, these three papers contribute to the understanding of policy incentives within LULC over time and establish a relevant link from policies to land use and land -cover change The approach outl ined in the dissertation reveals how trajectories of policies have been controlled by different factors at different times and in different geographic spaces. The dissertation provides insights as to how policy implications played out at a specific scale a nd location and contributed to our understanding of LULCC and policymaking in Peru. Although this case remained at a local scale, the results confirm that the effects of policies on the landscape do have global implications such as forest losses and pastu re expansion Similar approaches can be used for countries that share the Amazon and other tropical rainforest and expand on the approaches and ideas taken on in this dissertation.
CHAPTER 2 CONVERSION OF SMALL AGRICULTURAL AREAS I N A NEWLY EMERGED FRONTI ER REGION IN SOUTHEA STERN PERU: A TRAJEC TORY OF POLICY INTERVENTION SCENARI OS AND LAND USE DYNAMICS Summary: Do distinct policy regimes associated with particular presidential administrations result in distinct land use change patterns among landholders? Do adopters of specific policies exhibit different land use change patterns than non adopters? This paper presents an empirical analysis of the role of government policies in land use changes over two decades of frontier expansion in Peru. The analysis con siders policies associated with specific time periods of particular presidential administrations and their effects on forest, pasture, crop, and regrowth areas. The analysis focuses on policy adopters and nonadopters and controls for other socio -economic variables at the level of land users. The findings demonstrate that distinct policies cause land use changes at different time periods. For example, adopters of policies associated with cattle expansion cause an increase in pasture areas whereas an increas e in regrowth prevails once cattle expansion policies are eliminated. Incorporating a policy context strengthens the understanding of tropical deforestation in Peru and other countries that share the Amazon and other tropical rainforests. Key words land use, frontier development, policy trajectories, deforestation, OLS regression 2.1 Introduction Changes in the Amazon forest s and frontier regions continue to attract the scientific com munitys attention. Tropical frontier regions have enormous biological v alue, including their environmental services (biodiversity, carbon sequestration, nutrient cycling, among others). Land use changes represent a major environmental threat. This has prompted wideranging discussions of drivers of tropical deforestation whi ch have pointed to a number of human -
environmental causes, including government policies (Kaimowitz & Angelsen 1998; Geist & Lambin, 2003; Liverman et al. 1998; Liverman & Vilas, 2006; Fox et al., 2003; Wood & Porro, 2002; McCracken et al 2002; Rudel 2005; Walsh & Crews Meyer, 2002; Turner et al. 2004; Andersen et al. 2002; Lambin & Geist, 2006). Despite valuable contribution s by various discipline s much work remains to be done to understand the role and importance of government policies and their effects on land use/land cover changes as they vary in different geographic settings. Of particular importance are spatial and temporal variations in government policies and their impact on land use/land cover change (Mather, 2006:376; Pacheco, 2002). For example, comparisons of land use in time periods associated with distinct policy regimes2 would indicate temporal changes in deforestation and land use that have been caused by policies. Howe v er, d iscussions of policies and deforestation which feature the analysis of time series data have cer tain methodological limitations, because changes in deforestation over time can result from many factors changing at the same time. There is therefore a need for a complementary approach to analysis of policy effects on land use. One alternative is to compare land use and landuse change among people who took advantage of a policy to those who did not. This can reveal spatial variation in the extent of adoption, which in turn provides a basis for evaluating the effects o f policy incentives among adopters and non adopters. Such a comparison would provide a focused test of policy impacts on land use. This article examines the question of the effects of policy periods and specific policies on deforestation and ultimately lan d use in the tropical forest frontier of Southeastern Peru, an area of high biodiversity and land settlement that has seen changes over time in the context of policy shifts. I focus on the case of the Peruvian Amazon, where policies have shifted over time among 2 Here policy regimes are defined by government policies characterized within a speci fic time period.
policy regimes and where specific policy incentives have frequently been adopted by some but not all landholders. These considerations motivate two key questions for this inquiry. First, do distinct policy regimes associated with particular presidential administrations result in distinct land use patterns among landholders? And second, do adopters of specific policy incentives exhibit different land use patterns than nonadopters? The analysis focuses on comparisons of deforestation over a 20 -year pe riod in light of three distinct policy regimes, and on land use outcomes among adopters and nonadopters of several specific policies. I draw on survey data, including information on several types of land use and employ statistical models to control for ot her socio -economic factors not related to the specific policies at the level of land users. The findings confirm important effects of specific policies, some associated with distinct policy regimes. For example, policies that favored cattle expansion influenced an increase in pasture areas. P olicies associated with credit availability facilitated the expansion of agriculture areas, increasing deforestation. The results demonstrate the importance in incorporating a policy context in understanding tropical de forestation and land use/land -cover change, which has implications for policymaking in Peru and other countries that share the Amazon and other tropical rainforests. 2.2 Background 2.2.1 Linkages of Government Policies to Land -use/land-cover change Sortin g out the relative importance of multiple explanatory factors for tropical deforestation has been a core focus of the land use/land -cover change community (LULCC) (Turner et al ., 1994; IGBP IHDP, 1999; Lambin & Geist, 2006; Gutman et al. 2004; Geist & Lam bin, 2002; Angelsen & Kaimowitz, 1999; Mather et al., 1998). Many studies have related variations in deforestation patterns with numerous socio -economic explanatory factors (Geist & Lambin, 2001; 2003). Among such explanations are a variety of public polic ies, including
macroeconomic incentives (Lambin et al. 2000; Seto et al. 2002; Fearnside, 2000; Hecht, 2005; Pacheco, 2002), cattle ranching policies (Mertens et al., 2002), land tenure structure policies (Futemma & Brondizio, 2003; Walker et al., 2002), frontier development policies (Hecht & Cockburn, 1989; Schmink & Wood, 1992; Pichon, 1997), colonization policies (Smith, 1982; Moran, 1981; Walker et al., 2002), and road policies (Nepstad et al., 1999; 2001; Wood & Skole, 1998; Wood, 2002; Andersen et al., 2002). Considering the many governmental policies of different nature implemented on a routine base by specific administrations, it is important to analyze how these distinct policies relate to imprints on the landscape and how they vary on different ti mescales. As mentioned in Lambin & Geist (2006:174), the use of land is a highly political activity. The question of policies and land use/land cover change is of particular importance for frontier regions, which in Latin America often retain tropical fore st cover and have also been the target of state policy initiatives for frontier development and regional integration (Andersen & Reis, 1997; Kaimowitz & Smith, 2001; Schmink & Wood, 2002). While states often promulgate policies designed to integrate fronti er areas economically, c ertain conditions can undermine policy impacts in such regions, such as weak local economies (Simm ons, 2004), institutions absent or just emerging to implement policies (Perez 2007), or highly dynamic economic opportunities (Pichon & Bilsborrow 1999; Coomes, 1995, Alvarez & NaughtonTreves, 2003). Consequently, state policies may or may not bear their intended impacts, and their effects may themselves vary over time due to dynamics in frontier areas. By identifying time periods ass ociated with more or less distinct policy regimes promulgated by different presidential administrations, we gain one means of evaluating policy effects over time on land use in frontier areas.
Many studies on tropical deforestation emphasize the role of p olicies as distant macro level or intermediate meso -level factors that affect land use by influencing the decision making of landholders, who serve as proximate agents of land -cover change. Reviews of the causation behind land use and land-cover change hav e identified an array of global, national and regional driving forces that influence local factors which then more directly determine land use (Lambin et al. 2001; Geist & Lambin, 2002; Lambin & Geist, 2006). Wood (2002) proposes a hierarchical approach t hat explicitly links macro and meso -level processes to individual landholder decisions; in such a framework, landuse decisions are made in a complex context defined by numerous socioeconomic and biophysical drivers. From a hierarchical perspective, many meso and macro -scale socioeconomic factors can thus affect the decision -making process regarding whether to cut mature forest or a secondary forest; whether to grow annual or perennial crops, or use land for pasture; or alternatively, simply abandon land (Wood 2002; Perz 2002). In this perspective, the importance of policies for land use becomes evident via the adoption of policies by individual landholders (Rindfuss et al 2007; Gutman et al 2004). This then links policies to land use decisions by indi vidual farmers. Further, policies have distinctive effects and may be implemented differently at the farm level. For example, the implementation of certain agricultural policies may encourage different land uses, which will not necess arily lead to deforest ation: adopting cattle incentives will encourage more deforestation than engaging in agroforestry fruit species plantation ( e. g. copoazu). The question remains as to how policy implementations play out differently in different places that differ in the ex tent of policy adopters 2.2.2 State Policies in the Peruvian Amazon Worldwide, varied economic policy regimes, ranging from structuralism to neoliberalism and others, have influenced national land use dynamics (Liverman & Vilas, 2006; Klepeis &
Vance, 2003). In Latin America, economic vulnerability and uncertainty have prompted swings in policy regimes from military dictatorship to macroeconomic populism, structural adjustment and economic liberalization. However, the details of policy shifts have played out differently in different countries sharing the Amazon. While governments in different Amazon countries have offered credit programs and directed colonization efforts for the sake of agricultural frontier expansion, their timing and importance have vari ed, a reflection of distinct strategies and political views among countries. The case study presented here for Peru exhibits significant contrasts with other Amazon countries such as in neighboring Brazil, where directed colonization projects were accompan ied by state services (Mertens et al., 2002) which hardly occu rr ed in the Peruvian case. More generally speaking, t he history of policies and initiative s to develop the Amazon in Peru were different than in Brazil The Peruvian Amazon ha s seen a trajectory of changes mainly due to state centralism that envisioned the Amazon as a resource warehouse without the need for proper development. Since the 1960s in Peru, agricultural support and policy incentives were primarily directed toward coastal regions, discouraging infrastructure and market development in the Amazon. Instead, economic change in the Peruvian Amazon proceeded via short lived boom bust cycles, such as in rubber and gold impelled by short -term geopolitical interests along the Brazilian Bolivian border. Between the end of the 1960s and beginning of the 1980s, Perus policy regime shifted priorities from a long-standing export orientation supported by elites to a nationalist and militarist regime. The government of Alan Garcia came to power in 1985 with the support of a majority angry at elitist and top-down centralized policies. The Garcia administration (19851990) expressed greater interest in the agricultural sector. After 1990, a combination of fiscal austerity measures and market liberalizatio n tra nsformed the economy and
farmer s land practices. Since 2000, more liberalized and decentralized politics led to a major shift in economic opportunities. Following the understanding of the historical policy regime trajectory, it is highly probable that these described regimes catalyzed changes in land use. Therefore, the remainder of this article will exemplify specific policies pertaining to the past 20 -years and evaluate their implications as they influence land use (Table 2 1). By analyzing past im plications of policies, improved knowledge for the adoptions of future policy implications m ay be produced. 2 3 Study Case, Methods, and Data 2.3.1 Regional Setting T o evaluate the relationship between policy period and specific policies a nd land use, I emplo y data from a survey of small -household farmers in the Madre de Dios region of s outheastern Peru. This region is located along the newly paved Inter Oceanic Highway2, which links the Brazilian state of Acre with Peru. Madre de Dios comprises part of the tr i national MAP frontier constituted by M adre de Dios (Peru), A cre (Brazil) and P ando (Bolivia) The focus of my study area is the landscape around the towns of Iapari and Iberia, which covers an area of 2040 km in the province of Tahuamanu of Madre d e Dios Figure 2 -1 shows the location of both towns and the farms surveyed, along with property boundaries of all farms in the area. Madre de Dios is a frontier region with a specific historical trajectory and locally distinct landscapes. The last hundre d years in Madre de Dios can be summarized in terms of four boom bust extractive economic cycles: Rubber, Brazil nut ( castaa), gold mining, and timber. These cycles have not had equivalent impacts across the entire region. In particular, t he province of Tahuamanu experienced rubber boom cycles around 1900 and 1940 (Tahuamanu, 2001). Because there is little castaa in Tahuamanu, the effects of castaa cycles have been limited; the same is true for gold. Since the 1980s, Tahuamanu has been a key area for logging concessions
that support the ongoing timber boom (IIAP CTAR, 2001). Commercial agriculture has never played a strategic role in the regions economy (INADE OEA, 1998). Madre de Dios including Tahuamanu province underwent specific historical experi ences through national policies. Until the 1970s, Madre de Dios remained little noticed by a highly centralized government in Peru. This changed during the first administration of Alan Garcia (19851990), as the government adopted a heterodox supervision to macroeconomic management. The governments aim was to increase demand -led growth by expanding expenditure capacity among farmers, controlling production costs, and supplying demand s and market prices (Escobal, 1992). Two especially notable policies were implemented during this timeframe, namely credit and cattle expansion policies. Credit loans were administered by the Agrarian Bank, who gu aranteed the purchase of farmer s subsistence crops at fixed retail prices. Increased agricultural credit and suppo rt for cattle acquisition fostered expanded forest clearing, especially for cattle pasture during this time period ( Alvarez & Naughton Treves, 2003; Naughton Treves, 2004) However, because the government food policies subsidized both farmers and urban consumers, yearly losses increased and the credit program failed. Further, lack of competitiveness in markets among local producers, absence of technical information on appropriate agricultural practices, and limited access to markets all contributed to the c ollapse of these policies (Coomes, 1996) In 1990, Alberto Fujimori was elected and instituted a new policy regime. Agrarian policies were based on structural adjustment and fiscal austerity, which ended state marketing monopolies (Trivelli et al. 2003). U nder Fujimori, the Peruvian government closed the Agrarian Bank, halted agricultural credit and subsidies, and introduced new taxes. In this neo-liberal context from 1990 to 2000, the Fujimori administration formulated policies that fell within the
strateg ies contained in the Amazonian development plan through the Organization of American States (OAS) and Peruvian border communities (INADE OEA, 1998). Government programs assisted farmers in production, quality improvement, processing and marketing for crops such as coffee, cacao, pepper and agroforestry. In an attempt to promote agricultural diversification and increase crop yields and add the cultivation of trees to farmers current land-use practices, the specific policy introductions included reforestati on and seed improvement incentives, among others (CTAR, 2002). Studies showed that during this period, forest clearing declined in Madre de Dios (Alvarez & Naughton Treves, 2003; NaughtonTreves, 2004). Since the Fujimori administration, i nadequate financ ing opportunities remain a limiting facto r to agricultural expansion and intensification (Chapter 3). The agricultural sector has continued to operate since the 2000 under a neoliberal umbrella and has been exposed to various policies as attempts to incre ase agricultural output and prepare the region for trade integration (Tahuamanu, 2001). A crucial institution focusing on agriculture extension activities has been the Madre de Dios Special Project (Proyecto Especial Madre de Dios, or PEMD), a governmenta l institution operating through decentralized administrative units involved in agriculture extension work and other regional development initiatives3. A major focus of PEMDs activities has been the expansion of agriculture via development activities in pr e -existing colonization projects (INADE PEMD, 1998). To that end, PEMD instituted programs focused on three primary themes: infrastructure, agroforestry and environment4, and development of border communities. According to Dourojeanni (1990), deforestation increased by 2 million ha in areas where PEMD was active during 198087, providing a good reason to further evaluate more recent PEMD policies and their influence on land use.
This article therefore focuses on more recent PEMD policies and their area of i nfluence that belong to the agroforestry and environment program: government -sponsored cattle insemination, copoazu plantation, agricultural mechanization, and fish farming (INADE -PEMD, 1998). It is expected that these specific policies will have spatially explicit implications on land use change beginning with their adoption in 2000. For example, adopting agricultural mechanization is expected to expand crop areas and government -sponsored cattle insemination adoption anticipates an increase in pasture are as. 2.3.2 Field Data Design and Sampling From 2003 to 2005, I conducted household surveys of farms surrounding the towns of Iberia and Iapari in Tahuamanu province of Madre de Dios region. I chose to divide my field work between these two towns because they had experienced distinctive levels of exposure to state policy regimes during different policy periods and to specific policies. Importantly, some policies were more frequent ly adopted in one location than in the other. I sampled 125 households, divide d more or less evenly between Iberia and Iapari. I worked from a cadastral map from the Peruvian land titling agency (PETT). The map served as a sampling frame to help ensure a representative sample of farms distributed throughout the area titled for small -scale agriculture. The total area of sample farms equaled 11,279.8 ha out of 50,058.9 ha for all farms in the study area, or roughly a 20% sample by area. I sampled using other criteria relevant to this analysis, namely exposure to at least two policy pe riods (i.e., length of residence of at least 10 years), and a mixture of migration histories (e.g., inclusion of both migrants arriving in the area through a colonization program as well as natives born in Tahuamanu). Interviews employed structured questi onnaires that included sections on landuse practices, migration history, and landuse history, adoption of a variety of agricultural policy
incentives, soil fertility, land tenure, credit history, and governmental support. I emphasized farm households ex posure to agricultural, credit -based and other policy -related initiatives that affected a farmers decision to modify their land parcel uses (Laney, 2002). The resulting data set allows an examination of adoption of various Peruvian pol icy incentives for a griculture and their impacts on land use. 2. 3.3 Land -Use Change Indicators The analysis involves multivariate statistical models to evaluate the importance of policy variables for land use outcomes, controlling for the effec ts of other explanatory factors5. I evaluate the significance of adopting any of several policy incentives by comparing the land use by a dopters to that of nonadopters. I begin the analysis by outlining the dependent and explanatory variables included in the analysis. Table 2 2 shows d escriptive statistics for land use change indicators and data, and includes descriptive statistics aggregated by location. Table 2 3 presents descriptive statistic s for explanatory variables with separate data on location and including anticipated effects on landuse outcomes. 2.3.4 Dependent Variables The dependent variables refer to land use include measures of the number of hectares under crops fallow pasture and forest (Table 2 2). Due to non-normality, a log transform was performed to reduce skewn ess and obtain normal distributions. These land uses were targeted by many agricultural policies in Peru. In particular, larger pasture and crop areas are indicative of landholders adopting policies such as credit incentives and cattle expansion. Similarly changes in fallow and forest area may reflect policies promoting reforestation. The biggest land area is forest, with over 50% of total land area, followed by pasture and fallow. Crop area represents the smallest land use category, which suggests that cu ltivated land is still used for subsistence
agriculture (mainly rice, maize, beans, and some fruits). Fallow areas include regrowth between 1 15 years of age, according to the farmers definition of fallow. 2.3.5 Independent Variables I organize the indep endent variables into six groups: household life cycle, location and physical endowment, institutional context, three policy periods, namely 19851990, 19902000, and 20002005, and specific policies during 20002005. Household life cycle variables refer t o household migration history, population composition, and length of residence (Perz, 2001; Perz & Walker, 2002; McCracken, et al. 1999). I measure migration history via region of birth, whether in Madre de Dios or elsewhere, which would include other par ts of Peru as well as Brazil or Bolivia. Further, I differentiate between farmers who arrived as part of various colonization projects since the 1960s. Most colonists here migrated from the Peruvian region s of Arequipa and Puno during the late 1980s durin g the presidency of Alan Garcia6. According to field surveys, 44.8% respondents were born in Madre de Dios, and 41% of respondents arrived in colonization projects. I expect that landholders from Madre de Dios will preserve bigger areas of forest in compari son to migrants arriving in colonization programs, since the goal of colonization was agricultural expansion (Futemma & Brondizio, 2003). Table 2 3 shows that, on average, families consisted of four members. Small farmers in the Amazon who lack capital rel y on family labor, and larger families can thus clear larger land areas and bring them into production. Farm households are dependent on family members to help with labor intensive activities such as cutting down and removing trees, as well as burning, pla nting, and harvesting. I therefore expect that a higher number of family members will increase crop areas and pasture (Pichon, 1997; Coomes et al. 2000).
Length of residence has varied implications for land use because it affects farmer identification wit h the land, reflects experience with local agriculture, and may be related to economic opportunities. The sample shows that respondents had resided at their place of residence at the time of interview for 15 years. T he longer the years of residence, the mo re likely it is that a farm family will move from subsistence cultivation to commercial agriculture and thereby expand their land area in production, which implies more crops and pasture and less forest and fallow (Walker, 2003). However, if higher incomes are not guaranteed or if subsidies are necessary for the switch to commercial production are absent, a farmer will not risk market oriented production (Moran et al. 1994; Mertens et al. 2002; Walker & Homma, 1996). I nonetheless expect that length of re sidence will have a positive effect on crop and pasture areas. Farm size correlates with land use activities. The bigger the size of a farm, the higher the probability that higher forest areas will be intact, and more areas will be used for pasture and cr ops. Farms in the sample averaged 82 hectares in area7. Location/physical endowment refers to location, soil fertility, and distance to markets. Location distinguishes between the districts of Iapari and Iberia. These districts have experienced distinct h istorical trajectories (Moreno, 2000); one might characterize Iberia as a more transient and migrant -dominated town, whereas Iapari has remained a relatively stable, small, and locally unique community. These differences reflect their contrasting recent h istories. Iberia has seen a larger presence of public authorities headquartered in their territory, such as the Agrarian Bank until 1991 and the office of PEMD since 1983. I expect that location and area of influence of governmental projects will be strong er around Iberia than Iapari; consequently, I anticipate more cropland and pasture around Iberia.
Soil fertility refers to the respondents reported perception about the fertility of their land as in its productivity. Although according to INRENA OEA (1998), soils in the study area are extremely nutrient -poor and require shifting cultivation for most crops, more than half (56%) of the farmers in the survey indicated that they perceived their land as highly fertile. It is worth noting that the majority of fa rmers still engage in subsistence agriculture8. Farmers have hardly received training and extension services for improving soil fertility. I expect that perceptions of lower soil fertility will result in less land area under crops and pasture and more land area under fallows. Kilometers to the nearest local market describe the shortest distance to either Iapari or Iberia. Drawing on standard hypotheses from economic geography, I anticipate that farms located closer to towns will have less forest and fallo w, and more crops and pasture (Walker, 2004; Chomitz & Gray, 1996; Mertens & Lambin, 1997). This should especially be the case to the extent that farmers engage in commercial activities, whether crops for market or cattle raising. Institutional context9 re fer s to whether the respondent belonged to a local association and had a formal land title. Association membership is based on formal or informal social networks (Bebbington & Perrault, 1999). Successful associations may empower local producer groups, impr ove market access via cooperatives, and strengthen negotiating skills of members. Roughly 45% of interviewed landholders indicated some sort of association membership. A formal land title assures property rights, ensures credit eligibility and prepares far mers for participating in governmental projects. According to economic theory on tenure security, land titling increases resource conservation reducing forest exploitation and destruction (Walker et
al 2002; Mueller et al 1994). I therefore anticipate that possession of land titles increases forest areas. Policy indicators refer to variables that I expect to have implications for land use. I divide the policy indicators into three sub-groups based on specific policies tied to specific presidential admin istrations in Peru, namely the administrations of Alan Garcia (state intervention in the agricultural sector, 19851990), Alberto Fujimori (state withdrawal under neoliberalism, 19902000) and Alejandro Toledo (support for infrastructure for agriculture, 2000 2005). I expect that credit and colonization policies implemented during 1985 1990 decreased forest and increased pasture ( Alvarez & Naughton Treves, 2003; NaughtonTreves, 2004) By contrast, I anticipate that neoliberal policies that involved withdra wal of state support for agriculture during 19902000 increased forest and decreased crops. And finally, I expect that during 20002005, infrastructure policies again decreased forest and increased crops and pasture. Each policy regime is reflected in spe cific policies intended to affect land use, but which were only adopted by some farmers. Specific policy indicators include incentives for cattle a doption and agricultural credit implemented during the Garcia administration. Garcia introduced the cattle m odule policy incentive to increase cattle expansion. A module would comprise the acquisition of 11 heads of cattle, 10 cows and one bull per household This allowed farmers to acquire cattle at low prices and diversify land use practices beyond crops, the reby impelling forest clearing for pasture creation. Credit incentives were part of the Agrarian Bank, which handed out low -interest loans to farmers. I expect adoption of both of these incentives to decrease forest cover and increase pasture areas. I als o consider adoption of reforestation and improved (hybrid) seed varieties, connected with the presidency of Alberto Fujimori. Seed improvement and reforestation belong to incentive
projects implemented by various public authorities such as PEMD, Regional G overnment, and Agrarian Agency during the 1990s for diversifying agricultural production via agroforestry. The improved seed program included coffee, cacao, citrus, pepper, maize, and plantain (CTAR, 2002) Reforestation seedlings included peach palm, teak rubber and mahogany. I anticipate that adoption of improved seeds will reduce forest areas and increase cropland, whereas adoption of reforestation incentives will increase forest and fallow areas while reducing pasture and crops. Finally, I evaluate th e effects of incentives for agriculture mechanization, government sponsored cattle insemination, fish farming, and copoazu fruit plantations, promulgated by PEMD during the Toledo administration. In general all of these PEMD indicators refer to policies that promoted expanded agricultural production (INADE PEMD, 2004). The agricultural mechanization incentive offered the use of heavy equipment and access to fuel, as well as technical support and provision of seeds and fertilizer, in return for one ha of l abor input by farmers. Farms held by adopters of mechanization should therefore exhibit less forest and fallow and more cropland and pasture. Adoption of government -sponsored cattle insemination should have similar effects because it encourages investment in cattle ranching. Adoption of incentives for fish farming and copoazu plantations should also expand production, though they may not have exactly the same effects on the landscape as incentives for mechanization or government -sponsored cattle inseminatio n. Fish farming may reduce forest and fallows, but it may have the same impact on cropland and pasture. Copoazu plantations should also reduce forest and fallows and expand crops, but may also reduce pasture. 2.4 Findings Table 2 4 present s groups of O rdin ary L east S quares (OLS) models for forest, crop, pasture, and regrowth regressed on household life cycle, location/biophysical endowment, institutional context, and the policy variables. For each outcome variable, I present a group of
four models. The firs t model is a base model that includes the non -policy control variables. The second model has specific policy variables for the Garcia period, the third model has the Fujimori period, variables, and the fourth has the Toledo period PEMD variables. 2.4.1 Forest Cover Overall, the fo rest cover models are moderately strong (R from 0. 22 to 0. 28 ). The life cycle variables show significant effects on forest cover, in line with previous literature. Among the policy variables, the PEMD indicators show si gnificant effects and in the direction expected, in that adopters of government -sponsored cattle insemination and copoazu plantation had less forest F uture aims in acquiring cattle have recently increased in the province of Tahuamanu (Cha pter 3 ). This cou ld partly explain PEMD specific government -sponsored cattle insemination policy control for loss of forest after 2000. 2.4.2 Crop Cultivation The models for crop area indicate significant effects of location, family size, and colonization origin, all in t he direction expected. The inclusion of specific policy incentives (second model) reveals that the participation of farmers in credit incentives during the Garcia regime had a significant effect in raising crop areas. But contrary to expectations, taking part in reforestation also increased crop area, perhaps through diversification via agroforestry. Among the PEMD indicators, (model 4) adoption of copoazu plantations significantly increased crop areas, as expected. Lack of good transportation infrastruct ure systems prior to 2005 and uneven demographic spread, favored the tendency of farmers in Iberia toward engaging in agriculture and increasing crop areas. 2.4.3 Pasture Cultivation The pasture models are relatively strong (R from 0.33 to 0.34) and exhi bit expected findings. The life cycle variables are important: farm household s had significantly less pasture
when (1) farmers were born in Madre de Dios, (2) had moved to the study region as part of a colonization project, and (3) traveled longer distance to reach markets. These results correspond to expectations. The explanation for colonists exhibiting less pasture can be related to colonist cases that had to abandon their land. Although the timing of the colonization projects coincided with agricultural incentives such as loans and livestock units the lack of transportation or marketing mechanism s and inadequate prices all represented major drawbacks that prevented the success of the colonization sites (INRENA OEA, 1998) As a consequence, the majority of colonists abandoned or curtailed their agricultural activities10 (Eori, 1990). This explanation also relates to the credit incentive variable, which accounted for an increase in cropland but failed to do the same for pasture. However, the other policy i ndicators, including that for cattle insemination, did not exert significant effects on pasture area. These results suggest that ranching is complex and requires additional inquiry to better understand in light of public policy. 2.4.4 Fallowing and Regrowt h In the fallow/regrowth models, farmers born in Madre de Dios, those affiliated with a colonization project, and those with a longer duration of residence all exhibited larger areas under regrowth. Contrary to expectations, respondents reporting higher so il fertility also tended to have larger land areas under regrowth. Among the policy variables, findings in model 2 show that adoption of policies promulgated during 19851990 yielded greater regrowth for farmers exposed to the cattle module and a decrease for farmers adopting reforestation initiatives during 19902000. While adoption of cattle many years before might lead to regrowth via pasture degradation, the finding for reforestation decreasing regrowth was definitely not expected. Further unexpected fi ndings appear in model four, where adoption of copoazu plantations also reduced forest regrowth.
2.5 Discussion and Conclusion The findings reveal that government policy incentives, even those from many years before the time of interviews, influence farm l evel land use, at least in Tahuamanu province, Madre de Dios, Peru. Both variables related to earlier policy periods as well as recent policies such as PEMD activities exhibited significant effects on land use indicators. The findings also confirm that pol icies are not universally adopted, and that adoption of policy incentives can differentiate land use among farms. All that said, the relationships between public policies and land use are nonetheless complex, in part because there are other causal factors at work behind land-use decisions. As the impacts of policies become clearer, the adoption or non adoption of policies conditions the course of frontier development. While this analysis established a link from policies to land use, it is also necessary t o consider shifts in land use dynamics over time among policy periods, particularly reversals. This is beyond what one can conclude from the static models presented here, even with policy indicators referring to distinct periods, where different periods exhibit contrasting associations with a given land use outcome (e.g., Garcia and Fujimori periods and contrasting effects on pasture). The results show that the effects of policies are period -specific; however, not all outcomes can be explained by associatin g specific policies with specific time -periods, suggesting that more detail ed understanding is necessary for each policy period and for each specific policy, since more complex factors are at stake, such as lack of transportation, lack of markets and adequ ate prices, and lack of basic services. This is specifically important for the effects of cattle policies and policies related to cattle and crop expansion. P ast credit incentives and more recent cattle policies failed to explain a relationship with either less or more forest and pasture. The initial expectations made in this study could no t be met and may hint to farmer s lack of st rategic
competitive advantages and important questions. Why do some policies, when adopted, seem to make no difference? This r aises issues about 1.) the actual content of the policies, 2.) the logistical requirements to adopt a policy incentive, 3.) the characteristics of adopters and non adopters, and 4.) synergistic processes at hand while polices are being promulgated. For exa mple, a proper implementation of policies has to assure market profitability and proper technical support for a policy to succeed. Another factor can be the nature of the implemented policies. For example, in the Amazon region including the Peruvian Amazon agricultural opportunities have evolved unevenly, resulting in distinct landuse and land-cover trajectories as one looks across the basin. Within Peru, new agricultural policies have been targeted toward coastal areas, stimulating investments and incre asing productivity, whereas the Amazon region has continued to be excluded from an agricultural modernization process, and still exhibits low technological and educational levels, little organization, and high transaction costs (Perez, 2007). In Madre de Dios, the implementation of policies has occurred through a patchwork of political interests with strategic goals that have not always been congruent with each other or with local needs and local conditions. In this regard, further studies could expand the issue of adoption and nonadoption of policies and include a broader range of socio-economic variables. Further, focus should be directed to future events. Within the study region, a suite of new policies will influence farmers behavior with the current paving of the Inter Oceanic Highway (Dourojeanni, 2006). Here, a priority should be to monitor landscape changes underway before as compared to those now occurring during paving of the Inter Oceanic Highway (Chapter 3). In the larger context of policy regimes and land use change, results shown in this article offer reasonable justification, that future landscape
monitoring efforts should take a complementary approach, such as linking land use and policy (Reid et al ., 2006). The task of monitoring landscape changes should go hand in hand with monitoring policies. The article provided insights as to how policy implications played out at a specific scale and location and contributed to our understanding of tropical deforestation and policymaking in Peru. Alt hough this case remained at a local scale, it presented global implications such as forest losses and pasture expansions Similar approaches can be tested for countries that share the Amazon and other tropical rainforest as a means to keep oversight of the many policies that are being put into practice. 2.6 Notes 1 This does not exclude other timeframes that are relevant to the study areas agricultural frontier development. In this analysis, I chose to focus on the timeframe starting in 1986 until 2006. 2 The portion of the Inter O ceanic Highway in the Iapari Iberia study area was paved between 2005 and 2006 after the bridge over the Acre River in Iapari was opened in January 2005. 3 Proyecto Especial Madre de Dios (PEMD), (Special Project Madre de Dios) is a governmental institution that was created in 1981 with the objective to foster development throughout the region between Puerto Maldonado and the border town of Iapari. In 1983, PEMD fall under t he executive structure of INADE, the National Institute of D evelopment until recently, 2008 when it transferred to the Regional Government of Madre de Dios. PEMD operated as decentralized administrative mechanisms through agriculture extension work, transportation, credit and land tenure regu larization incorporat ing farmer s participation. 4 This program aims to assure sustainable development and the conservation of the environment through mitigation measures for productive activities. 5 In 1986, colonization efforts organized by Proyecto Especial Madre de Dios (PEMD) sen t 120 families from Arequipa and Puno to be settled in Primavera and Chilina along the Inter Oceanic Highway in the districts of Iapari and Iberia. Unfortunately, these colonization projects failed due to lack of support and possibilities for livelihood improvement.
6 Although I assumed that each location had experienced distinctive levels of exposure to state policy regimes during different policy periods and to specific policies I decided not to separate values for Iapari and Iberia because when run ning OLS models separately between the two towns the outcomes showed statistically weak models. A big concern was the nonrandom spatial aspect of who adopted policies. Some policies were solely adopted in Iberia, leading to a loss of statistical power w hen performing models separately. Further, a divided model between the two towns would pose a methodological problem. The focus of the paper is on policy issues characterized by policy adopters and non adopters and it was important to test the effects of g eographical location net of the other socio -economic factors. 7 Additional farm data: standard deviation 51.02, range 295 and median 70. 8 The most important annual and perennial crops ranked by crops the most farmed are: rice, maize, plantain, manioc, beans, fruits, citrus agroforestry, and coffee for household consumption and local market sales. 9 The survey had incorporated more variables related to institutional context such as farmers access to agricultural extension services and NGO access, participation in training and workshops by outside organizations. Since household s reported no participation or presence of any service in the region related to agricultural services, the se variables were not included. 10. An estimated 70% of migrants arriving with the colo nization programs relocated or returned to their regions ( INADE OEA, 1998)
Table 2 1. Tw enty -year frontier policy t imeline Policy Period I mplementation Specific Policies spatially explicit Mandate of A. Garcia (19851990) 1. Government -Sponso red Credit Incentive 2. Cattle Module Mandate of A. Fujimori (19902000) 1. Seed Improvement 2. Reforestation Mandate of A. Toledo (20002005) 1. Agriculture Mechanization 2. Fish Farming 3. Cattle Insemination 4. Copoazu Pla ntation Subsidized acquisition of 10 cows and one bull per household.
Figure 2 1. Study site districts: Iapari Iberia, Madre de Dios, Per
Table 2 2. Descriptive statistics for land use change i ndicators as total of net area and aggregated by lo cation, farm h ouseholds, Iapari -Iberia, 2004* Mean St. Dev. Skewness Forest area hectares (ha) 47.46 39.85 1.39 Cultivation area (ha) 3.39 3.43 1.80 Pasture area (ha) 18.17 22.87 1.82 Regrowth area (ha) 10.57 12.18 2.07 Total Area 79.59 48.60 1.55 Mean St. Dev Iberia1 Iapari2 Iberia Iapari Forest area hectares (ha) 42.96 51.75 36.53 42.63 Cultivation area (ha) 4.29 2.53 3.66 2.97 Pasture area (ha) 21.46 15.04 26.48 18.47 Regrowth area (ha) 9.75 11.34 12.65 11.76 *n=125 1n=6 1 2 n=64
Table 2 3. D escriptive statistic s for explanatory variables, household background, biophysical endowment, and policies, aggregated statistics for Iapari (INA) and Iberia (IBE), expected land use outcomes for forest (F), regrowth (R), crop (C), and pasture (P), + (expected incre ase), (expected increase or decrease), and (expected decrease), farm households, Iapari Iberia, 2004 Mean St. Dev. Skew ness Mean St. Dev. Expected F R C P ( Proportion Y es ) (INA/ I BE) (INA / IBE) Outcomes Background/Lifecycle Region of birth: Madre de Dios, (0=No, 1=Yes) .46 (.34 / .12) + Colonization Project Region* (0=No, 1=Yes) .41 (.10 / .30) + + Family size 3.92 1.92 .69 (4.06 / 3.77) (1.87 / 1.97) + + Years of residence 14.76 8.23 2.05 (14.95 /14.55) (7.75 / 8. 87) + + Natural log (ln) hectares (ha) farm plot 4.29 .62 .49 + + + Location / Physical Endowment Location (0=Iberia, 1= Iapari) .51 + Soil fertility (0=Low, 1=High) .56 (.37/.19) Kilometers to nearest market 10.22 5.54 .92 (9.19/11.29) (4.68 / 6.19) + Institutional Context Association membership (0=No, 1=Yes) .52 (.22 / .30) Formal land title (0=No, 1=Yes) 86 (.50 / .35) + Past Policy Period Implementation Indicators 1985 1990 Specific Policy Indicators Cattle module (0=No, 1=Yes) .16 (.08 / .07) + Credit incentive (0=No, 1=Yes) .07 (.22 / .25) + 1990 2000 Specific Policy Indicators Seed improvement (0=No, 1=Yes) .47 (.03 / .08) + Reforestation (0=No, 1=Yes) .42 (.06 / .12) + + 2000 2005 PEMD Specific Policy Indicators Agric. Mechanization (0=No, 1=Yes) .16 (.00/.16) + Cattle Insemination (0=No, 1=Yes) .04 (.00 / .04) + Fish Farming (0=No, 1=Yes) .07 (.00 / .07) + Copoazu Plantation (0=No, 1=Yes) .14 (.02 / .11) + Refers to household who migrated under a planned colonization project from Arequipa and Puno
Table 2 4. Land use regression m odels regressed on background, location, institutional diversity, policy period indicators and specific policy indicators, farm households, Iapari Iberia, 2004. DV 1 l n 2 (Forest Area) DV ln (Crop Area) Model 1 3 Model 2 4 Model 3 5 Model 4 6 Model 1 Model 2 Model 3 Model 4 Valid n =125 for all the models Model R (adjusted) .22 (.15) .22 (.13) .23 (.14) .28 (.19) .16 (.09) .20 (.12) .20 (.12) .19 (.09) Constant c .23 .13 .41 .06 .18 .45 .01 .00 Background/ life cycle Region of Birth: Madre de Dios (0=No, 1=Yes) .13 .11 .12 .21 .00 .01 .19 .04 Colonization Project Region (0=No, 1=Yes) .59* .59* .61* .62* .19 .16 .36* .23 Number of Family Members .09* .07 .08* .1 0* .05* .04 .04 .04 Years of Residence .02* .02* .02* .02* .00 .01 .01 .00 Ln Farm Plot Area .85*** .84*** .88*** .82*** .13 .07 .13 .15 Location, Physical Endowment Location (0=Iberia, 1=Iapari) .1 5 .13 .20 .59* .33** .31* .25 .16 Soil Fertility (0=Low, 1=High) .34 .35* .35* .33* .05 .06 .05 .04 Kilometers to Market .00 .00 .00 .00 .01 .02 .01 .02 Institutional Context Association Members hip (0=No, 1=Yes) .07 .05 .08 .09 .12 .15 .12 .12 Land Title (0=No, 1=Yes) .10 .06 .19 .42 .21 .13 .13 .07 Policy Period Implementation Indicators 1985 1990 Specific Policy Indicators Cattle Module (0=No, 1=Yes) .04 .18 Credit Incentive (0=No, 1=Yes) .17 .32** 1990 2000 Specific Policy Indicators Seed Improvement (0=No, 1=Yes) .49 .13 Reforestation (0=No, 1=Yes) .15 .40** 2000 2005 PEMD Specific Policy Indicators Agric. Mechanization (0=No, 1=Yes) .25 .04 Cattle Insemination (0=No, 1=Yes) 1.00* .31 Fish Farming (0=No, 1=Yes) .57 .19 Copoazu Plantation (0=No, 1=Yes) .67** .38* *p<0.15 **p< 0.05 ***p<0.01 1Dependent Variable 2natural log 3Base model that includes the nonpolicy control variables 4Adds the first earlier policy period specific policy variables 5Replaces those with the third specific policy variables associated with earlier policy periods 6Replaces those with the more recent (PEMD specific) policy variables.
Table 2 4. Continued DV 1 l n 2 (Pasture Area) DV ln (Regrowth Area) Model 1 3 Model 2 4 Model 3 5 Model 4 6 Model 1 Model 2 Model 3 Model 4 Valid n =125 for all the models Model R (adjusted) .33 (.27) .34 (.27) .33 (.26) .34 (.26) .14 (.07) .16 (.07) .17 (.09) .18 (.07) Constant c 1.16 .89 .1.41 1.00 1.19 1.20 1.41 .55 Background/life cycle Region of Birt h: Madre de Dios (0=No, 1=Yes) .79*** .84*** .64** .79** .72*** .65** .49* .82*** Colonization Project Region (0=No, 1=Yes) .44 .52* .31 .33 .59** .54* .39 .64** Number of Family Members .02 .03 .02 .03 .00 .02 .0 1 .01 Years of Residence .01 .00 .01 .01 .03** .02** .03** .03** Ln Farm Plot Area .93*** .86*** .95*** .88*** .06 .08 .06 .02 Location, Physical Endowment Location (0=Iberia, 1=Iapari) .29 .30 .26 .34 .05 .02 .06 .19 Soil Fertility (0=Low, 1=High) .03 .06 .02 .06 .39** .36* .39* .44** Kilometers to Market .05** .04** .05** .05** .02 .01 .02 .02 Institutional Context Association Membership (0= No, 1=Yes) .23 .28 .23 .26 .01 .00 .01 .03 Land Title (0=No, 1=Yes) .45 .45 .42 .39 .01 .07 .08 .06 Policy Period Implementation Indicators 1985 1990 Specific Policy Indicators Cattle Module (0=No, 1=Yes) .56* .46* Credit Incentive (0=No, 1=Yes) .09 .23 1990 2000 Specific Policy Indicators Seed Improvement (0=No, 1=Yes) .12 .17 Reforestation (0=No, 1=Yes) .41 .48* 2000 2005 PEMD Spec ific Policy Indicators Agric. Mechanization (0=No, 1=Yes) .46 .45 Cattle Insemination (0=No, 1=Yes) .59 .25 Fish Farming (0=No, 1=Yes) .14 .34 Copoazu Plantation (0=No, 1=Yes) .08 .46* *p<0.15 **p<0.05 ***p<0.01 1Dependent Variable 2natural log 3Base model that includes the nonpolicy control variables 4Adds the first earlier policy period specific policy variables 5Replaces those with the third specific policy variables associated with earlier policy periods 6Replaces those with the more recent (PEMD specific) policy variables.
48 CHAPTER 3 FUTURE LAND USE PLAN S ALONG THE INTER OCEANIC HIGHWAY IN SOUTHEASTERN PERU Abstract: This study examines how transportation networks and policies influence futur e land use plans among farmers along the Inter Oceanic Highway in Southeastern Peru. Logistic regression is used to estimate the probability of farmers engaging in specific activities, such as fish farming, government -sponsored cattle insemination or agri cultural mechanization, based on socio -economic indicators and under different road -policy planning outcomes. The results indicate that indicators of past policies influenced farmers perceptions and thus future landuse plans. However, further understandi ng is needed about context -specific local conditions such as settlement history, biophysical attributes, and institutional characteristics, and their effects on prospective landuse plans and resulting land transformation. Keywords : Road deforestation, Amazon, policy indicator, time -series analysis 3.1 Introduction Worldwide efforts to facilitate economic integration through infrastructure development projects have been at the core of the public development agenda. As transport ation infrastructures conti nue to improve, so does the desire to integrate some remaining regions that have lacked proper access. The Initiative for the Integration of Regional Infrastructure in South America (IIRSA ) 1 has opened the window for economic integration in South America, but at the same time it has raised increased concern about possible environmental impacts with global ramifications such as pasture expansion and forest loss (Dourojeanni, 2006). In this article, a major issue is to learn how the presence of a road relat es to land transformation in a region which has been waiting long for i mproved infrastructure. This study focuses on a segment of the paving of the Inter O ceanic Highway2, the last link in a key IIRSA Sur road project3 that should integrate product exchan ge between Brazil, Peru and their export
49 markets, and thereby promote economic development in affected Peruvian regions most of which have lacked investment and economic opportunities in the past. Past and current land transformation and development oppo rtunities related to infrastructure projects have been linked to governmental policies This has prompted a concern that specific policies at different time periods may influence changes in the landscape; moreover, public policies may further influence lan d changes via effects on future land use plans. In this regard, monitoring the case of IIRSA Sur paving process would benefit from the analysis of past policies and how farmer s reacted to them. As such, the article focuses on specific policies that existed at specific time -frames and affected land use and are subject to the population s perception of future land use plans. My hypothesis tests whether past and current agricultural policies have influenced f uture land use plans among farmers I assume that farmers land use plans are correlated with socio -economic indicators, the degree of development of the area in question, and specific policy regimes that are integrated with the road building process and ultimately affect land use change. Part of IIRSA S ur includes paving and improving approximately 2,586 km of roads in the Peruvian territory This will improve access to significa nt conservation areas and long term biodiversity hot spots (Terborg h 1999; Conservation International, 1999). As such, my stud y addresses a long debated controversy among development supporters and environmentalists regarding if and how roads have influenced changes in tropical forests and their market access (Garcia, Raez & Boggio, 2008) 3.2 Complex Causality of Roads and Def orestation T ransportation infrastruct ure projects positively affect a regions growth and development by fostering investment, market access, and technology transfer At the same time, infrastructure projects have led to massive deforestation rates and car bon emissions that have
50 influenced global climate change. Major transportation expansion projects in the Brazilian Amazon were implemented to incorporate the Amazon region with the national economy (Andersen & Reis, 1997; Chomitz & Gray, 1996). On the othe r hand, in major areas of the Peruvian Amazon, lack of transportation networks has largely hindered development and economic opportunities (GESUREMAD, 1998; INADE OEA, 1998). In most cases, tropical deforestation has been strongly linked to road constructi on (Eden, 1998; Kaimowitz & Angelsen 1998; Nepstad et al. 2001). For example, in the Brazilian Amazon, more than two thirds of Amazon deforestation has taken place within 50 km of major paved highways (Alves 2002) Brazil has experienced one of the larg est programs of public transportation expansion into tropical forests, resulting in major rates in deforestation (Carvalho et al ., 2001; Laurance et al ., 2001). Deforestation rates on the Peruvian side have been comparatively low, but the deforestation tha t has occurred is also located close to roads (Alvarez & Naughton Treves, 2003; Maki et al., 2001; Imbernon, 1999; Kaliola et al., 2001). The study of land use/land -cover change (LULCC) has revealed changes in deforestation along roads, with dramatic changes in forested areas occurring in combination with road building (Laurance et al. 2001; Nepstad et al. 2001; Carvalho et al. 2001). According to LULCC reports tropical forests constitute a region experiencing rapid change, especially in contexts where roads are present (Nepstad et. al ., 1999, 2001; Wood & Skole 1998; Mertens et al. 2002; Cropper et al. 2001; Wood & Porro, 2002; Nelson & Hellerstein, 1997; Via, Echavarria & Rundquist, 2004; Kaimowitz et al ., 2002), making it crucial to analyze furthe r how people link land use/land cover decision-making processes with road paving and deforestation. For example, it is important to understand the various local contexts and variables that affect deforestation, such as local perceptions about road corridor s, and to search for explanations among these varied
51 contexts. A variety of socio -economic conditions and past policies influence farmer perceptions which then determine land use plans. T he introduction of new roads is only one part of the equation of change and may lead to varied land transformation outcomes in combination with other factors such as: settlement stage, level of frontier development, policy intervention, technology, education, and variables unique to the region. As a consequence, in addit ion to consider ing the direct and simple causal effects of infrastructure activities on deforestation, road monitoring studies should include other indirect and complex effects modified by other factors as well (Weinhold & Reis, 2008). In this regard, the relationship between a road project and land change dynamics is based on the interaction of multiple causes with one another and can be understood under the concept of causal complexity. Causal complexity defines an outcome as the result from several diffe rent combinations of conditions (Ragin, 1987:20). This study will follow this approach A number of land transformation outcomes have linked government policies leading to logging, mining, agricultural development, cattle expansion, and increased rates of migration with deforestation (Kaimowitz & Angelsen 1998; Wood & Porro, 2002; Geist & Lambin, 2003; Alves, 2002; Wood & Skole, 1998; McCracken et al 2002). Roads lead to land -cover conversion by facilitating access to land and other natural resources, in the Amazon and elsewhere (Soares Filho, et al. 2004; Via, Echavarria & Rundquist, 2004). Road building fostered frontier expansion by reducing transport costs, potentially making local economic activities viable for competition in markets. A simulation s tudy by Cattaneo (2001) showed that with an improved infrastructure system, the transport costs for agricultural products produced in the Amazon would decrease. As agricultural production in the Amazon became more profitable, the price of arable land incre ased, thereby raising the incentive to deforest. To the extent that
52 roads facilitate market -oriented production, frontier expansion involves forest clearing beyond that needed for subsistence production. Thus, deforestation can be linked to farmers behavi or toward profit, as well as to the availability of roads. Changes in land use that trigger deforestation depend upon the interaction of a particular economic framework based on the presence of governmental policies, infrastructure levels, and farmers res ponses or expectations. This interaction can be further explained through a utility or profit -based approach. Farmers make land use decisions based on expected utility (whether production or profit), strategic competitive advantages and local economic con ditions (Walker & Homma 1996). It is expected that without opportunities to achieve profit s from sale of produce in local markets, farmers will not risk a shift from subsistence to commercial agriculture and they will there fore not cut much forest, only what is needed for subsistence. The strategic macro -economic advantages are related to the presence of biophysical factors, such as soil type and water, as well as socio -economic factors, such as market presence and road infrastructure. Context (location) specifics, such as easier product marketing over different locations than others play a crucial role as well. It is therefore relevant to further evaluate farmers perceptions or expectations about changes, if these are going to affect future planning for land use. T he theoretical framework argued upon here, is based on farmers expected utility landuse decision process and incorporates a temporal perspective. I link farmers land use decision s not only to policies in past periods but also to perceptions of future plans (Figure 31). For example, a key aspect is to acknowledge that farmers may modify plans given new infrastructure. The basic assumption is that farmers presumably plan for expansion of land use because of reduced transport costs and thus g reater profitability. But this plan in turn may be modified based on causal complexity via
53 the operation of other factors that affect planning. For example, land use planning may also be affected by past and current participation in public policies via ado ption of policy incentives. Thus both public policies as well as road paving can modify future land use planning among farmers. Farmers who participated in cattle incentives or credit incentives are better positioned to expand cattle or make higher investm ents in the context of new infrastructure. For example, in Southeastern Peru, farmers benefited from credit incentives through the Agrarian Bank during the late 1980s. According to previous studies, these socioeconomic incentives have been linked to accel erating deforestation rates along the transportation routes available at that time (Alvarez & NaughtonTreves, 2003, INRENA, 1996). However, once credit availability disappeared in the 1990s, returns declined, deforestation decreased, and regrowth increase d Currently, with the advancement of the paving of the Inter -Oceanic Highway since 2005, it is expected that policies favoring agricultural expansion in the presence of new infrastructure will again encourage forest clearing and crop cultivation and pastu re expansion (Chapter 2). 3.3 Historical Background 3.3.1 Infrastructure Improvement s in the Study Region Southeastern Peru offers a useful case study for analyzing the interaction between road development, policy shifts, and land use decisions having global environmental implications. Within Southeastern Peru, the region of Madre de Dios has been marked by boom and -bust economic dynamics driven by external market demand for specific regional commodities. Many forest resources in demand were depleted rapidl y; for example, rubber during the 1920s in Peru (INRENA OEA, 1998) As a consequence, economic opportunities and overall development in Peruvian tropical forests have been conditioned by unstable global market demand, and they lasted only as long as the re source itself, or until outside competitors found lower cost production options (INADE OEA, 1998). As a consequence, commodity booms have afforded limited
54 development in Madre de Dios before turning to busts. At the same time, forest resources in general have been plentiful, and deforestation rates have been historically low compared to neighboring areas such as in Brazil (Imbernon 1999; Chapter 4). Forests in Southeastern Peru were relatively intact, with moderately less dramatic increase in deforestation rates compared to other tropical regions of the Amazon, such as in neighboring Brazil (Imbernon, 1999). Besides some short -lived boom andbust economic cycles, such as rubber and gold, the exploitation of natural resources for commercial use was not signi ficant (IIAP CTAR, 2001). During the 1960s, the government initiated a development program to support the expansion of agriculture and cattle ranching, following the construction of the first unpaved trails in Madre de Dios (INRENA OEA, 1998). As a result pressure on the forest for settlement, agriculture, and grazing increased steadily. This occurred as a result of policy shifts at the national level in Peru. Between the end of the 1960s and beginning of the 1980s, Perus political government reflected priorities of a wide range of export -oriented elitists in a nationalist and militarist regime. During the administration of Alan Garcia (1985 1990), agricultural credit became more available and fostered landscape change. The government worked t oward expand ing expenditure capacity among farmers, controlling productio n costs, and supplying demands and market prices (Escobal, 1992). However, the credit program failed due to lack of economic competitiveness among local producers, absence of information on biophysical properties, and access to markets (Coomes, 1996) After 1990, reforms undertaken to liberalize input and product markets included the relaxation of price controls on inputs food and agricultural products as well as the end of state marketing monopolies (Trivelli et al. 2003) Since 2000, more liberalized and decentralized politics have provided a wider range of market opportunities.
55 Migration, expansion and colonization into Southeastern Peru have been dependent on economic cycles and short term g eopolitical interests along the Brazilian Bolivian border (IIAP CTAR, 2001). As Table 3 1 shows, population influx began around the 1970s, when the first transportation dirt roads opened access to Puerto Maldonado, the capital of Madre de Dios. Population decreased between 1972 and 1981 in the district of Iberia, but it increased in the district of Iapari. By the end of the 1970s, the province of Tambopata attracted migrant workers to assist in the gold boom, and settlement expansion shifted from Tahuamanu toward Tambopata Province (INRENA OEA, 1998). During the 1990s (INEI, 1994) Iberia experienced a slight population increase, while Iapari remained stable. Interestingly, recent census data indicate population decrease in the Iapari area, whereas the po pulation of Iberia is rapidly increasing4. 3.3.2 Connectivity and Cooperation Although the Peruvian Amazon constitutes approximately 63% of the countrys territory, it has lacked integration due to a lack of sustained attention among policy makers, operati ng in the confines of a strictly urban, coastal and centralistic government. Communication and transportation deficits have proven a major hindrance to development throughout the region (Tahuamanu, 2001). Even though its infrastructure improved slightly, M adre de Dios hardly strengthened its relationship with the central government in the capital of Lima. In fact, the study region has had a closer relationship to western Brazil than with Lima, since Brazil has been favoring trade with Peru since the late ei ghteenth century5, when river traders from Brazil exchanged crude manufactured goods such as steel and clothing for Peruvian salted fish, wood resins, balsams, and wax (Herndon, 1952). Not surprisingly, Brazil has played a major role in integrating the reg ion through a series of cooperation treaties and by pushing for the pavement of the Inter Oceanic Highway (INRENA OEA, 1994).
56 Despite efforts at cooperation, road construction has been dependent on policy regimes as well. Prior to the paving of the Inter -O ceanic Highway, transport infrastructure consisted of dirt roads, mule trails, and highways under construction, all of which were passable only during the dry season. The main alternative, rivers, also afforded transport, but that option was limited by non -navigable stretches. It was not until 1981, when the Special Project Madre de Dios (PEMD) 6 was created and focused its work on the province of Tahuamanu, that the construction of a dirt road between Puerto Maldonado and the frontier town of Iapari was i nitiated. During the 1990s, segments of the prospective Inter -Oceanic Highway were traced, leveled, and maintained. However, they allowed safe passage only during the dry season. Beginning in the 2000s, the unpaved road was compacted and surfaced with gra vel and prepared for paving, which began at the end of 2005 (INADE -PEMD, 2004). In terms of connectivity and policy regimes, the timeframe of this study will be based on three important policy periods: a) from 19851990, which corresponds to the first pres idency of Alan Garcia in Peru and represents the time when dirt roads were improved in the region and attracted migration to the area; Despite efforts at cooperation, road construction has been dependent on policy regimes as well. Prior to the paving of th e Inter Oceanic Highway, transport infrastructure consisted of dirt roads, mule trails, and highways under construction, all of which were passable only during the dry season. The main alternative, rivers, also afforded transport, but that option was limit ed by non-navigable stretches. It was not until 1981, when the Special Project Madre de Dios (PEMD) 6 was created and focused its work on the province of Tahuamanu, that the construction of a dirt road between Puerto Maldonado and the frontier town of Iap ari was initiated. During the 1990s, segments of the prospective Inter Oceanic Highway were traced, leveled, and maintained. However, they allowed safe passage only during the dry
57 season. Beginning in the 2000s, the unpaved road was compacted and surfaced with gravel and prepared for paving, which began at the end of 2005 (INADE -PEMD, 2004). 3.4 Study Case, Methods and Data 3.4.1 Study Area T o document how socio -economic background and adoption of public policy incentives affect future land use planning by small -holders in the presence of prospective road improvements, I surveyed small household farmers in two districts in Tahuamanu province of Madre de Dios: Iapari and Iberia These districts have towns located on t he Inter Oceanic Highway, which links th e Brazilian state of Acre with Per u. The exact study area encompasses the road axis between the towns of Iapari and Iberia and covers an area of 2040 km. I conducted 125 farm household surveys outside the towns of Iapari and Iberia d uring 20032005. I worked from a 2000 cadastral map from the Peruvian land titling agency (PETT). The map served as a sampling frame to help ensure a geographically representative sample of farms distributed throughout the area titled for small -scale agriculture. The total a rea of sample farms equaled 11,279.8 ha out of 50,058.9 ha for all farms in the study area, or roughly a 20% sample by area. Figure 3 2 shows the location of farms sampled and the overall distribution of farms between Iberia and Iapari. In deriving this s ample, I also used other criteria relevant to this analysis, namely exposure to at least two policy periods (i.e., length of residence of at least 10 years), and a mixture of migration histories (e.g., inclusion of both migrants arriving in the area through a colonization program as well as natives born in Tahuamanu). Interviews employed structured questionnaires that included sections on migration history, land use history, and adoption of a variety of agricultural policies, soil fertility, land tenure, c redit history, governmental support, and landuse practices. I emphasized farm households adoption of incentives derived from various policies from the Garcia, Fujimori and
58 Toledo presidencies that were intended to affected land use decisions and therefor e land use planning (Laney, 2002). The resulting data set allows an examination of adoption of various Peruvian policy incentives for agriculture, tied to specific policy regimes from particular time periods. In addition to national policies promulgated by presidential administrations, I also evaluated regional development initiatives promulgated by the Madre de Dios Special Project (Proyecto Especial Madre de Dios, or PEMD). Many of the sampled farm households were supported by one or more of PEMDs major projects: agricultural mechanization, fish farming, copoazu plantation, or government -sponsored cattle insemination. This survey occurred before the actual paving of the Inter Oceanic highway, but during preparations for paving, which subsequently occurre d in 20062007. Hence, the timing of the survey is appropriate for a study of how prospective road improvements may affect future land use plans. Economic theory suggests that farmer expectations about future profitability influence present decision -making via planning for future land use. In the context of imminent road paving and thus reduced transportation costs, I expect farmers to report land use plans involving shifts toward expanded commercial production. I also anticipate that the specifics of such shifts in planning will be affected by whether a given farmer had taken advantage of a specific policy incentive (i.e., planning for cattle expansion among farmers that previously took advantage of incentives related to cattle). 3.4 .2 Dependent V ariables T he analysis focuses on a suite of 10 dependent variables indicating specific future plans tied to agricultural production. Descriptive statistics for these variables appear in Table 3 2. For purposes of the policy analysis, I focus on the first 5 out of the 10 total: fish farming, farm sale increase d cattle production reforestation, and agriculture mechanization These variables were more commonly reported plans, and/or showed greater differences between Iapari and Iberia (as
59 shown by the F test values) I focus on the prominent and spatially concentrated future plan variables because they are more likely to reflect previous adoption of policy initiatives that tended to be adopted in one district rather than the other. The analysis can thus pursue a key question, which is whether spatial concentration in policy adoption is in fact related to spatial differences in planning of future land use. 3.4 .3 Independent V ariables Table 3 3 presents descriptive statistics for the explanatory variables, along with t heir expected relationships with future landuse plans7. Separate descriptive statistics disaggregated for Iapari and Iberia have been reported in Chapter 2. The explanatory factors are organized into four groups: 1.) socioeconomic background and life cyc le, 2.) location and resource endowments, 3.) institutional context, and 4.) variables for policy periods and adoption of specific policies. Background/lifecycle variables refer to household migration history, population composition, and length of residence (Perz, 2001; Perz & Walker, 2002; McCracken, et al. 1999). I measured migration history as region of birth whether in Madre de Dios or outside the region, which would include other parts of Peru as well as Brazil or Bolivia. Further, I differentia te between farmers that arrived as part of various colonization projects since the 1960s. Most colonists here migrated from the Peruvian region s of Arequipa and Puno during the late 1980s during the presidency of Alan Garcia (19851990) through PEMD who se n t group s of farmers to work on rubber estates and agriculture8. According to field surveys, 44.8% respondents were born in Madre de Dios, and 41% of respondents arrived in colonization projects. In regard to anticipated land use plans, higher values for e ach of the background/life cycle variables will be associated with higher probability of planning cattle expansion. Landholders from Madre de Dios are expected to preserve their farm and not sell in the future
60 while landholders who came with a colonization project may be less probable to e ffect reforestation. Table 3 3 shows that, on average, families consisted of four members. Small farmers in the Amazon who lack capital rely on family labor, and larger families can thus clear larger land areas and bring them into production. Farm households are dependent on family members to help with labor intensive activities such as cutting down and removing trees, as well as burning, planting, and harvesting. Length of residence has varied implications for land use because it affects farmer identification with the land, it reflects experience with local agriculture, and it may be related to economic opportunities. The sample shows that on average respondents had resided at their place of residence at the time of int erview for 15 years. Although it is difficult to set a temporal line between the shift from subsistence to market -oriented systems (Perz & Walker, 2002), the longer the duration of residence, the more likely it is that a farm family will move from subsiste nce cultivation to commercial agriculture and thereby expand their land area in production and their land use plans. However, if higher incomes are not guaranteed or subsidies necessary for the switch to commercial production are absent, a farmer will not risk market -oriented production (Moran et al 1994; Mertens et al 2002; Walker & Homma, 1996). As a result, past experience should affect the farmers perception of their own land and decisions to transform land. It is expected that family size and leng th of residence will raise the probability of a farmer having plan s to acquire cattle and engage in agricultural mechanization. In terms of family size a larger family will be able to use a larger labor pool and will have more possibilities for improvin g farm conditions, and ultimately will show less interest in selling land. In regard to the variable farm area a larger area is more likely to affect plans for
61 prospective cattle acquisition, mechanization of fields, and retail sales. Cattle acquisition per se requires large areas for pasture management and agricultural mechanization usually involves a market oriented perspective in the near future, thus the need for larger farm areas. If a large farm area is in use and profitable, reforestation efforts are less likely. Location/physical endowment is measured as a binomial variable and distinguishes between the towns of Iapari and Iberia. Whereas Iberia has seen a major presence of public authorities headquartered in the territory, such as the Agraria n Bank until 1991 and the office of PEMD since 1983, Iapari has seen closer ties to Brazil and Bolivia due to its proximity. How these locational differences will influence prospective land use plans remains difficult to predict, since with the paving of the road, issues such as proximity and lack of connectivity will play a minor role. Nevertheless, some historical and demographic distinctions are still expected to influence future landuse plans. The variable s oil fertility describes farmers percepti on of the fertility of their own land in terms of productivity and yield. A critical point here is that farmers perceptions of soil fertility differ substantially from external evaluations. Despite studies confirming extremely nutrient poor soils in the s tudy area, 56% of farmers indicated that they perceived their land as highly fertile. This may be an indication that, for the most part, farmers still engages in subsistence agriculture. I expect that perceptions about soil fertility exert a major influenc e on future land use plans. These perceptions will be based on past experiences, local knowledge, and technical assistance availability, such as extension work. Regarding the variable distance to market it will very likely influence land use expansion because of reduced transport costs and thus greater profitability.
62 Institutional context refers to binomial variables such as association membership and formal land title. Among the farmers interviewed, 45% indicated some sort of membership in one or mor e associations. The functionality of the association membership variable falls within the concept of social capital as it reflect s a households ability to build trust and ties within and between members (Bebbington & Perrault, 1999). In this regard, membe rship in active farmer associations may strengthen negotiation skills and lower market risks for individual farmers. This suggests that association membership among farmers will increase the likelihood of future land use plans based on expanded market -orie nted production. A formal land title assures tenure security, facilitates access to credit, and is necessary for farmers to participate in government projects (Walker et al. 2002; Mueller et al. 1994). In the Amazon, this has in practice meant larger pro duction systems, since farmers with tenure security will risk investments in market oriented enterprises (Perz, 2002). I t is anticipate d that farmers with secured land titles will engage in future activities that expand production for markets. As described above, s pecific policy indicators focus on three specific policy periods: 1) 19851990, which corresponds to the first presidency of Alan Garcia; 2) 19911999, which encompasses the two terms of Alberto Fujimori, and 3) 20002005, which refers to the pr esidency of Alejandro Toledo. Each of these policy periods is represented by specific policy indicators defined by binomial variables that relate specific incentives to the periods when they were implemented. I asked the question if farmers had taken advantage of policies offered during the listed policy periods. In some cases, answers related to more than one specific policy indicator, and almost every interviewed household could relate to one of the described periods. I anticipated that past policy period s did influence the trajectory of frontier development, and thus
63 would also affect the trajectory of individual farm development, which in turn could affect future land use plans. Cattle module and credit incentive s were implemented during the presid ency of Alan Garcia; reforestation and seed implementation are connected with the presidency of Alberto Fujimori. Agriculture mechanization government -sponsored cattle insemination copoazu plantation and fish farming are detailed policies da ting to the presidency of Alejandro Toledo That period corresponds to the time of my field work and is therefore of particular interest for purposes of studying future plans since the Toledo presidency is most proximate to the period following the time of fieldwork9. Garcia introduced the module incentives to increase cattle expansion. A module c omprise s the subsidized acquisition of a herd of 11 cattle per household: 10 cows and one bull. This allowed farmer s to acquire cattle at low prices and to diver sify land use practices, increasing forest transformation and pasture creation and expansion. This policy period further introduced guaranteed agricultural markets and credit availability administered by the Agrarian Bank10. I expected that past exposure t o cattle policies would increase the probability of acquir ing cattle again while prospective plans for reforestation and agriculture mechanization would decrease. In regard to government -sponsored credit incentives landholders benefiting from past loans are more likely to engage in any of the landbased activities considering their level of preparedness, investment, and productivity output. Seed improvement projects refer to specific incentives implemented by various public authorities such as PEMD, the Regional Government, and the Agrarian Agency during the 1990s for coffee, cacao, citrus, pepper, maize, and plantain (CTAR, 2002) The r eforestation variable consists of planting trees for rubber, peach palm, teak, and mahogany timber11. Farmers re ported that they were unable to sell products, lacked technical assistance, and faced
64 high transportation cost s while markets were not guaranteed. Although most farmers indicated that these projects did not prevail and did not render economic stability for self -sustaining agriculture, a farmer s perception of similar land use plans is crucial and past experiences will likely impact prospective plans. It is probable that landholders involved in previous reforestation and seed improvement policies would cont inue to engage in these activities if agriculture markets are provided Farmers would be less likely to increase their cattle herds PEMD special project indicators focus on the promotion and diversification of the productive sector through the following activities: government -sponsored cattle insemination, agriculture mechanization, fish farming, and copoazu fruit plantation. PEMD policy adopters would benefit from agricultural extension support, seed and fertilizer availability, heavy machinery equipment usage, and participatory workshops, all of which are contributing factors to prepare landholders for market -oriented production (INADE -PEMD, 2004). Landholders linked to these specific activities will likely continue with the programs in an effort to i ncrease output efficiency. Farmers who adopted agricultural mechanization and government -sponsored cattle insemination activities are likely to prolong these activities and unlikely to engage in reforestation measures. Future land use plan predictions for fish farming and copoazu plantation are difficult to estimate; however, less interest in cattle expansion is possible. 3.5 Findings and Discussion Table 3 4 presents binomial logistic regression models for the five focal landuse planning outcome variables : increasing the number of cattle, engaging in reforestation, implementing agricultural mechanization, fish farming, and selling their farm property. These land use plans were regressed on the explanatory variables described in Table 3 3. The tables report odds ratios. Thus, r atios over 1 indicate a positive effect i.e. if the value of the independent variable is larger, the odds of the outcome occurring are larger. Conversely, a ratio
65 under 1 indicates a negative effect, which means that if the independen t variable has a higher value, the odds of the outcome occurring are lower. For each outcome variable, Table 3 4 presents three models. Of particular interest are the policy variables. For each land use variable, the first model includes only the non-poli cy variables and thus serves as a base model. In the second and third models for a given outcome variable, I added policy indicators associated with specific policy periods. This allows a comparative evaluation of the importance of different groups of poli cy variables for the land use planning outcomes, net of the effects of other explanatory factors. For the dependent variable increase cattle perceived soil fertility has a positive effect on the likelihood of a farmer having plans to expand cattle in th e future, while region of birth Madre de Dios and association membership have negative effects. Locally -born farmers highlighted the lack of expected cattle profits and proper technical support as well as high maintenance costs, which helps explain why t hey also lacked plans for expanded ranching. Very surprising is that none of the specific policies has any effect on plans to increase the number of cattle in the future. This means that adopters of any of several policy incentives are no more likely to pl an for cattle expansion than non adopters. These results were not anticipated, since I expected that farmers with previous cattle policy exposure, such as cattle acquisition through the cattle module policy, would report future plans to expand cattle hol dings. Further, farmers benefiting from current cattle incentives, such as PEMDs government -sponsored cattle insemination program, did not indicate future plans to increase the size of their herds more than those not benefiting from PEMD activities Alth ough all results related to cattle were unforeseen, field reports indicated that ranching required ample complementary knowledge regarding cattle
66 care and pasture management, as well as constant technological support, in order to be economically feasible. While the cattle module policy initially supported cattle acquisition, maintaining cattle was expensive, and after 1990, policy support was withdrawn and did not favor expansion of cattle. With the withdrawal of state support, ranching required continued investments by farmers, who came to perceive cattle maintenance as unprofitable. Consequently, farmers reported selling a large percentage of their cattle during the 1990s as well as herd losses due to cattle diseases (cf. INEI, 1995). All that said, futu re land use plans for cattle expansion require further analysis. The notion that cattle will improve a farmers livelihood has grown recently in the province of Tahuamanu and has led to an increase in future plans to acquire cattle. However, results shown here suggest that farmers are cautious when it comes to decide to acquire more cattle. Farmers suggested that they may engage in future cattle acquisition only if accessible complementary support (e.g., animal pest and sanitary consultancy, improved cattle breeds and better pasture quality) becomes available, as well as improved transportation. While the Inter Oceanic Highway has now been paved in Tahuamanu, the province still lacks sufficient government support, conditions that were important in cattle ori ented frontier areas of the Amazon (Mertens et al., 2002). Turning to models of plans for reforestation, region of birth and location have insignificant effects. Farmers born in Madre de Dios or living in Iapari were less likely to plan to reforest tha n farmers born outside of the region or those living in Iberia. These results were not foreseen, since it was expected that most migrants to Madre de Dios would not be attracted to the region for reforestation purposes but for agricultural activities. Furt her, Iapari is comprised
67 by family based timber extraction operations that focused heavily focused on timber extraction within agriculture. It was expected that r eforestation may have been a means to secure future timber supplies. These results thus indic ate that no distinction can be made between farmers origin, current location, and future reforestation plans. After adding policy indicators, results show that adopters of agricultural mechanization and copoazu programs are more likely to have future plan s to reforest than non adopters. Reforestation, agricultural mechanization and copoazu plantations all indicate intensification of land use occurring in similar places, and therefore could be correlated. The third set of models focuses on plans for agricul ture mechanization. Association membership and farm size area both increase the likelihood of having plans to pursue agricultural mechanization in the future; conversely, location in I apari, soil fertility, and larger family size all have negative impacts and thus reduce the likelihood of having plans to mechanize. Farmers living in Iapari are less likely to have plans to engage in agricultural mechanization than those in Iberia This result suggests that farm size and exposure to PEMD headquarters in Ibe ria plays a role in encouraging landholders to plan for agricultural mechanization in the future (see Chapter 2). The negative influence of soil fertility is an interesting result, because it indicates that farmers who perceive their soils as fertile are l ess likely to have future plans for agriculture mechanization. There is no obvious explanation for this finding other than that mechanization reflects intensification which includes greater use of chemical inputs, including fertilizers to sustain or raise production. The result for association membership is positive as expected and complemented by additional field data. According to field data, mechanized maize output increased from 60 tons to 750 tons between 2000 and 2003. This increase coincided with the creation of the Small Farmers Association of Tahuamanu
68 (ASPAT) in 2000. ASPAT has benefited members by increasing their productivity and guaranteeing markets. In terms of policy indicators, having acquired credit incentives and participating in copoazu pl antations increased the likelihood of planning to adopt agricultural mechanization. A group of credit incentive adopters seems to have gained knowledge and market exposure that enables them to plan for future expansion of mechanized agriculture. The relati onship between copoazu plantation and agriculture mechanization points to a broader engagement in diversifying agricultural activities. Models for fish farming exhibit few strong effects among the explanatory variables. Landholders with a higher number of family members and those living in Iapari are more likely to plan fish farming activities than families of smaller size or living in Iberia. Findings related to location correlate with field observations, since an increased number of fish ponds were visi ble in and around Iapari. None of the specific policies exert significant effects on plans to adopt fish farming. This is unanticipated, since it is expected that policies favoring adoption of fish farming would further motivate the expansion of fish farm s once profitability increases. Here again, technical support and fish maintenance costs may play a role for deciding to adopt fish farming or not. Farmers interviews indicated high initial investment costs with this activity that includes large quantitie s of fish food, labor, pond preservation, and proper transportation equipment. The fifth and final dependent variable is farmer s plans to sell their property Landholders who have lived longer in the study area are more likely to sell their farms than tho se who recently arrived. Also elderly households that lack labor for maintaining farm activities are more interested in selling land. Landholders living in Iapari are less willing to sell their farms than
69 farmers living in Iberia indicating that the population in Iapari has a stronger will to remain in the area than farmers from Iberia. Iapari has seen a more stable, smaller, and less migrant population than Iberia, which may explain the difference in retail plans. With regard to policy indicators, pri or participation in the cattle module positive ly influences farmers intentions to sell, whereas credit incentives have a negative effect. Besides previous cattle module policy adopters, other past policy engagement d oes not seem to influence real estate plans suggesting that farmers specifically related to past cattle activities had stronger motives to sell in comparison to farmers related to more recent activities. These results reinforce the observation that cattle business if not profitable unless sup ported by policy incentives, is hard to sustain by farmers in the study region. Overall, the results regarding future plans highlight the issue of farmer concerns about the lack of government support. Although this study concentrated on the adoption of po licies, it seems that lack thereof has been a factor affecting future plans even given the paving of the Inter Oceanic Highway. Table 3 5 show s that farmers view s about road paving are most ly positive. The population of the province of Tahuamanu has long required the construction of paved transportation routes indicating therefore a clear consensus toward road paving. A clear disagreement was demonstrated toward the government failure in facil itating better opportunities. S urveys indicated lack of government presence, lack of loans, lack of economic competitiveness, and lack of capacity training as major concerns vis a vis road presence. Even with past policy incentives and road paving, farmer intentions to shift toward expanded commercial production are limited due to lack of government support in key respects. 3.6 Conclusion This case study focused on how socioeconomic conditions influenced land-use planning among farmers along the Inter -Oceanic Highway in Southeastern Peru. Although the case study
70 is site specific and many explanatory variables are related to past events particular to Peru or Madre de Dios, the findings do suggest that past policy incentives in some instances do influence land use plans. Similar studies may be useful in other sites ex periencing infrastructure upgrades and can serve as a basis for attempting to understand the formation of plans for future actions involving forest and agricultural resource use. Overall, the findings could have indicated larger effects for understanding land use planning if farmers concerns about specific activities had been addressed. For example, one farm household reported the desire to increase cattle production, engage in fish farming, and/or undergo agricultural mechanization, once certain conditions were given. Conditionality is closely related to political, technological, and economic concerns, and it adds to the varied forces that modify the relationship between a road project and land change dynamics. Overall, the presence of the Inter Oceanic H ighway is only one part of the equation of change. Crucial to the equation are challenges such as the levels of public support, technological expertise, credit availability, education level, and increased fire hazards, to name a few. While farmers viewed r oad paving in positive terms, surveys also indicated that the lack of government presence, loans, economic competitiveness, and capacity training constituted major concerns as hindrances in pursuing future land use plans. In particular, the small effect of past policies related to cattle expansion on future plans acquiring cattle demonstrates that other conditions also influence farmer perceptions and thus their plans. The results add complexity to road causality studies. Future research on roads should fu rther analyze the interaction between road development, policy shifts, and land use decisions in specific local contexts and complement these studies with participatory approaches in order to depict local views. For example, Mendoza et al. (2007) implement ed multi -stakeholder
71 workshops along the Inter Oceanic Highway to facilitate knowledge exchange and public participation in highway planning. Stakeholders pointed to numerous problems that the authors organized under the headings of environmental, social, political, infrastructure, and economic issues. A key finding is that the principal problems mentioned varied substantially from place to place, and among areas with and without road paving. The results presented here are consonant insofar as I have highli ghted conditioning factors important to be recognized if future land use plans involving expansion of commercial activities are to predominate, even given past policy incentives and road paving. In regard to current attempts to understand local context, IIRSA Sur has provided environmental assessment studies that are publicly available, though descriptions of detailed impacts of road segments are not available, raising widespread questions about how the perspectives of local populations have been incorpor ated in the highway planning process (IIRSA SUR, 2007; Mendoza et al. 2007). Particularly in the IIRSA context, there is growing concern among civilian groups to interact with the public and private sector and participate in critical debates about changes which have occurred since the Inter O ceanic Highway was paved (Garcia, Raez Luna & Boggio, 2008). In the larger context of continent -wide and even global strategic plans for coordinated and integrated infrastructure projects, there remains a need for furt her research on local stakeholder responses, especially among related infrastructure projects if they involve increased connectivity among many localities across distinct landscapes. It is further crucial to understand the implications of past policies fo r trajectories of frontier development and change as a means to learn from past experiences and to promote future incentives leading to more sustainable actions.
72 In this regard, the article linked past to future while recognizing differences among farmers, partly due to policy adoption as well as location. 3.7 N otes 1 IIRSA was launched in 2000 by 12 governments with support from the Inter -American Bank of Development (IBD), Andean Development Corporation (CAF), Fonplata, and the United Nations Development Pr ogram. The goal was to promote 10 integration networks composed by 335 projects valued at $ 37.4 billion. These projects focused primaril y on transportation and include infrastructure for roads, waterways, ports, and energy, as well as communication interc onnections. For more information see http://www.iirsa.org 2 Inter Ocenica or Transocenica denomination as referred to in South American official literature. 3 The portion of the Inter O ceanic Highway in the Iapari Iber ia study area was paved between 2005 and 2006 after the bridge over the Acre River in Iapari was opened in January 2005. 4 The current population census data does not include the rapid changes experienced in Iapari since 2005 due to the opening of the fron tier bridge to Brazil and the conclusion of paving of the Inter Oceanic in early 2007. P opulation numbers for Iberia may actually have been higher considering that construction camps were based in Iberia and attracted a high number of migrant workers and new services to the area. 5 Way back in 1853, the first steamship arrived in the region under a free trade and navigati on treaty signed with Brazil. The first economic cycle of wild rubber extraction and demand for rubber in Europe and the U.S. during the 1880s stimulated expansion from Brazil deep into the Madre de Dios region and continued until the severe recession of the 1910s when world rubber prices declined. 6 Proyecto Especial Madre de Dios (Special Project Madre de Dios) is a governmental institution that was created in 1981 with the objective to foster development throughout the region between Puerto Maldonado and the border town of Iapari. 7 Although I divided the sample area geographically between the two towns, the analysis of the case study did not allow for a separate statistical analysis of each town, due to difficulties in capturing a representative number of cases for each town meeting all criteria. Further, the analysis aimed to distinguish location specifics net of other socioeconomic variables and policy relevance. 8 In 1986, colonization efforts organized by Proyecto Especial Madre de Dios (PEMD) sen t 120 families from Arequipa and Puno to be settled in Primavera and Chilina along the Inter Oceanic Highway in the districts of Iapari and Iberia Unfortunately, these colonization projects failed due to lack of support and possibilities for livelihood improvement.
73 9 These policies belong to the program on agroforestry and the environment. This program aims to assure sustainable development and the conservation of the environment through mitigation measures for productive activities. Half of the sampled farm households had been supported by one of the major PEMD projects promoting the productive sector. 10. However, t hese incentives lacked the focus on long term competitive market opportunities, trading mechanism s and price stability. The Agrarian Bank disappeared, credit and subsidies were removed, and new taxes were introduced. As a result, agricultural production and forest clearing declined (Alvarez & Naughton Treves, 2003; NaughtonTreves, 2004). 11. These incentives represented a first step toward s agriculture diversification but they did not prevail. Some farmers reverted back to subsistence agriculture as a sign of lack in technological extension support and poor market availability (Alvarez, 2001).
74 Table 3 1 Population s tatistics for the d istricts Iapari Iberia Iapari Iberia Tahuamanu Madre de Dios Prov ince Region Population 1972 504 4015 5336 25174 Population 1981 812 3013 4928 35788 Population 1993 870 3772 6443 71636 Population 2005* 791 4868 7429 1 02174 Population 2007 N/A N/A 10742 109600 Growth Rate 19721981 4.1 2.4 0.2 4.9 Growth Rate 19811993 0.3 2.1 2.3 6.1 Growth Rate 19932005 0.5 1.9 1.2 3.5 Source: INEI, National Housing and Population Census 2005; 2007; INADE OEA 1998. *Th e 20 05 census was a de jure census. As a consequence, data shown for 2005 may be underestimated. Figure 3 1. Theoretical framework based on policies adopted by farmers Theoretical Framework Perceptions of future plans Mandate of A. Garcia (1985 1990) Credit Incentive Cattle Module* Past Policies Mandate of A. Fujimori (1990 2000) Seed Improvement Reforestation Mandate of A. Toledo (2000 2005) Future Plans Increase Cattle Sell Land Agricultural Mechanization Fish Farming Reforestation Linking past to future Expected utilities of past policies Influence Agricultural Mechanization Fish Farming Government Sponsored Cattle Insemination Copoazu Plantation *a module defines the acquisition of a herd of 11 cattle per household: 10 cows and one bull.
75 Figure 3 2. Composite RGB 231 ASTER image from 30 June 2006 showing the study site road axis: Iapari Iberia, Madre de Dios, Per, the two towns, the Inter Oceanic Highway, the cadastre of titled farms, and the farms that were sampled for the interviews.
76 Table 3 2. Mean d escriptive statis tic for fut ure land use plans and their statistical significance of a relationship, farm households, Iapari -Iberia, 2004 Mean F -score1 Future Land-Use Plans (analyzed) Increase Cattle (0=No, 1=Yes) .62 .04 Refo restation (0=No, 1=Y es) .15 8.58** Agric. Me chanization (0=No, 1=Yes) .14 13.36*** Fis h Farming (0=No, 1=Yes) .18 7.59* Sell Farm (0=No, 1=Yes) .32 2.98* Future Land-Use Plans (not included in models) Animal Div ersification .08 1.5 Fruit Plantation .06 .63 Citrus .11 1.07 Product Industrialization .14 .01 Services .05 2.61* *p<0.15 **p< 0.05 ***p<0.01 1indicating statistical significance of a relationship
77 Table 3 3. Mean d escriptive statistics for explanatory variables, household background, biophysical endowment, and policies, expected outcomes for land use plans: increase cattle (IC), refores tation (R), agriculture mechanization (AM), fish farming (F), and sell farm (SF), farm households, + (positive effects), (either positive or negative effects), and (negative effects), farm households, Iapari Iberia, 2004. Mean St. Dev. Expected Outcome IC R AM F SF Background/Lifecycle Region of Birth Madre de Dios, Peru (0=No, 1=Yes) .46 + + Colonization Project Region* (0=No, 1=Yes) .41 + Number of Family Members 3.92 1.92 + + Years of Residence 14.76 8.23 + + Natural log (ln) hectares (ha) farm plot 4.29 .62 + + + Locat ion / Physical Endowment Location (0=Iberia, 1= Iapari) .51 + + Soil Fertility (0=Low, 1=High) .56 Kilometers to nearest market 10.22 5.54 + + + + Institutional Context Association Membership (0=No, 1=Yes) .52 + + + Formal Land Title (0=No, 1=Yes) .86 + + + Past Policy Period Indicators 19851990 Specific Policy Indicators Cattle Module (0=No, 1=Yes) .16 + Credit Incentive (0=No, 1=Yes) .07 + + + 19902000 Specific Policy Indicators Seed Improvement (0=No, 1=Yes) .47 Reforestation (0=No, 1=Yes) .42 + 20002005 Specific Policy Indicators Agric. Mechanization (0=No, 1=Yes) .16 + Cattle Insemination (0=No, 1=Yes) .04 + Copoazu Plantation (0=No, 1=Yes) .14 *Refers to households who migrated under a planned colonization project from Arequipa and Puno Source: 2004 Dissertation Survey
78 Table 3 4. Future land use plan binomial logistic regression models regressed on background, location, institutional diversity, and specific policy period indicators, farm households, Iapari -Iberia, 2004. DV Increase Cattle DV Reforestation DV Agric. Mechanization Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Model 17.37* 17.56* 21.57 17.10* 20.77** 28.82** 27.29*** 34.56*** 41.71*** Valid n (n=125) Background/life cycle Region of Birth: Madre de Dios, (0=No, 1=Yes) .40* .33* .26* .25* .27* .33 .94 .94 3.03 Colonization Project Region (0=No, 1=Yes) .46 .47 .31* .77 .62 .61 1.71 2.16 4.89 Number of Family Members .96 .96 .97 1.10 1.04 1.05 .84 .67** .68* Years of Residence 1.00 1.01 .99 .98 .95 .95 .98 .97 1.01 Natural log (ln) hectares (ha) farm plot 1.55 1.60 1.44 1.65 1.29 1.88 2.59* 2.17 3.62* Location, Physical Endowment Location (0=Iberia, 1= I apari) .53 .52 .54 .25* .26* .53 .19* .21* .33 Soil Fertility (0=Low, 1=High) 3.22*** 3.27*** 3.25*** 1.66 1.52 2.16 .39 .29* .48 Kilometers to nearest market .97 .97 .96 1.05 1.06 1.09 1.07 1.06 1.08 Institutional Context Association Membership (0=No, 1=Yes) .47* .46* .41 .72 .59 .83 3.46* 2.50 5.58* Formal Land Title (0=No, 1=Yes) 1.82 1.84 1.67 1.36 1.03 .82 1.17 .67 .29 Past Policy Period Indicators 1985 1990 Specific Policy Indicators Cattle Module (0=No, 1=Yes) .78 2.47 .27 Credit Incentive (0=No, 1=Yes) .94 2.75 7.98** 1990 2000 Specific Policy Indicators Seed Improvement (0=No, 1=Yes) 2.46 1.27 1.27 Reforestation (0=No, 1=Yes) .55 1.42 1.13 2000 2005 Specific Policy Indicators Agric. Mechanization (0=No, 1=Yes) .91 5.56** 2.38 Cattle Insemination (0=No, 1=Yes) 3.62 .28 1.58 Copoazu Plantation (0=No, 1=Yes) .53 4.55* 12.59** Fish Farming (0=No, 1=Yes) 2.23 .70 .74 *p<0.15 **p<0.05 ***p<0.01 1 valid n=125 for all the models 2 values next to independent variables are odds ratios DV=Dependent Variable Source: Dissertation Survey 2004
79 Table 3 4. Continued DV Fish Farm ing DV Sell Farm Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 16.04* 16.29 20.83 13.20 17.75* 34.92*** Valid n (n=125) Background/life cycle Region of Birth: Madre de Dios, (0=No, 1=Yes) 2.64 3.52 4.03* .96 .60 .34 Colonization Project Region (0=No, 1=Yes) .94 1.31 1.28 .63 .43 .39 Number of Family Members 1.24* 1.25* 1.24* .89 .96 .94 Years of Residence .96 .97 .98 1.05* 1.05* 1.04 Natural l og (ln) hectares (ha) farm plot 1.22 1.25 1.39 1.53 1.48 1.25 Location, Physical Endowment Location (0=Iberia, 1=Iapari) 2.95* 3.52* 4.39* .38* .32** .26** Soil Fertility (0=Low, 1=High) 1.08 1.08 1.26 1.56 1.45 1.11 Kilometers to nearest market .95 .95 .93 1.00 1.02 1.04 Institutional Context Association Membershi p (0=No, 1=Yes) 2.36* 2.50* 3.51** 1.07 1.11 .78 Formal Land Title (0=No, 1=Yes) 1.10 .98 .99 1.09 1.45 1.52 Past Policy Period Indicators 1985 1990 Specific Policy Indicators Cattle Module (0=No, 1=Yes) .93 3.72** Credit Incentive (0=No, 1=Yes) .76 .45* 1990 2000 Specific Policy Indicato rs Seed Improvement (0=No, 1=Yes) 1.08 2.10 Reforestation (0=No, 1=Yes) 2.14 .44 2000 2005 Specific Policy Indicators Agric. Mechanization (0=No, 1=Yes) 1.58 .50 Cattle Insemination (0=No, 1=Yes) .00 123.33 Copoazu Plantation (0=No, 1=Yes) 2.68 .30 Fish Farming 3.06 .19 *p<0.15 **p<0.05 ***p<0.01 valid n=125 for all the models 2 values next to independent variables are odds ratios DV= Dependent Variable Source: Dissertation Survey 2004
80 Table 3 5. Percentage of positive farmers perception toward road paving followed by major critical concerns mentioned by surveyed farmers in regard to the road paving, far m households, Iapari Iberia, 2004. Positive attitude 85 % (out of 125 respondents) However Major concerns mentioned: Lack of government presence 27% Lack of Loans 17% Lack of competitiveness 30% Lack of capacity training 14% Socia l concern 3 % No position 9% 100%
81 CHAPTER 4 LANDSCAPE DYNAMICS OF AMAZONIAN DEFORESTATION BETWEEN 1986 AND 2007 IN SOUTHEASTERN PERU: POLICY DRIVERS AND ROAD IMPLICATIONS Abstract : The relationship between national development policies and land -use/land -cover change (LULCC) is of particular importance for frontier regions, which in Latin America are often forested areas which have been the target of state policies for regional integration. This paper evaluates landscape changes between 1986 and 2007 in Madre de Dios, a frontier region in Southeastern Peru, which is undergoing landscape transformation due to major infrastructure projects, and has experienced nonlinear trajectories of deforestation, conditioned by periods defined by distinct development policies. I describe landscape dynamics, especially land-cover changes (LCC) attributed to logging and agricultural activities that were responses to policy incentives in Southeastern Peru. Five -year interval Landsat TM/ETM+/Aster images from 1986 to 2001 and bi annual images from 2001 to 2007 were classified into five land-cover types (forest, crops and pasture, regrowth, built/non -forest, and cloud/haze), producing a time series of land -cover maps. Changes in landscape dynamics we re described through fragmentation processes and driven by the spatial arrangements of timber concessions and agricultural areas. I calculated landscape and class metrics related to size, density, connectivity, configuration, and the classified land -cover types. Results show that land related to agricultural policies exhibited a heterogeneous forest/non-forest mosaic with a larger number of patches, whereas timber concessions consisted of homogeneous forested patches at specific time frames and altered by n ew forestry management laws Results further show that not all LULCC and landscape dynamics can be attributed to policies since other causal factors transform the landscape as well. 4 1 Introduction Changes in land cover continue to represent an important occ urrence due to its impact on biodiversity and climate change. The highest total forest clearing takes place in tropical regions,
82 more specifically in Amazonia (Lambin & Geist 2006; FAO, 2006; Skole et al ., 1994; Kaimowitz, 2002; IGBP IHDP, 1999). Monitori ng these changes is imperative to understand how areas of rapid change leave imprints on the landscape as distinctive patterns of deforestation. Throughout the Amazon basin, shifting cultivation practices by small farmers, large cattle ranchers, and loggi ng activities are the main factors that lead to deforestation (Ferraz et al. 2005; Wood & Porro, 2002; Kaimowitz & Angelsen, 1998; Geist & Lambin, 2003; Cattaneo, 2002; Verissimo et al. 1995; Perz, 2002). These are drivers of deforestation that commonly take place within 50 km of main roads and have contributed to distinctive pattern changes on the landscape (Alves, 2002; Nepstad et al., 2001; Mertens et al. 2002; Arima et al. 2005). Changes in land cover are reflected on the landscape as a sequence of variously fragmented mosaics of land use and land cover (LULC) such as pastures, cultivated land, urban areas, and secondary growth. Identifying and describing these varied mosaics of LULC as quantitative indicators of landscape configuration facilitates the understanding of patterns affecting landscape processes (Forman, 1997; Shelhas & Greenberg, 1996; Lambin et al., 2001; Geist & Lambin, 2001; Southworth et al., 2004). Landscape ecology techniques have examined spatial and temporal differences and chang es in structure and function at the landscape level of organization, and these changes have been related to modifications in the pattern of land use (Forman & Godron, 1986; Turner & Gardner, 1990; Pearson, 2002; Lausch & Herzog, 2002). Landscape structure encompasses the spatial distribution of land cover patches and is based on the interaction between patches in terms of their size s shape s connectivity, and heterogeneity (Turner, 1989).
83 An understanding of the relationship between human behavior and la nd cover is required to understand the causes of changes in landscape structure. For example, analyzing the causes of changes in spatial pattern over time (i.e., under what political, environmental and economic contexts forest cover increases or decreases) provides knowledge about how socio-economic conditions affect forest dynamics and drive the changes (Brown et al. 2000; Forman, 1997; Walsh et al., 1999). Fragmentation processes have been of particular interest when increases in shifting cultivation an d small -scale logging operations create complex landscape patterns of forest patches in different stages of formation and at different time periods (Pichon & Bilsborrow 1999; Imbernon & Branthomme, 2001; Rudel & Horowitz, 1993; Ferraz et al., 2006). Forest -cover change patterns are shown through forest fragmentation which leads to i ncreased patchiness Relating the configuration of forest patches that appear spatially irregular with ecological processes has been more common than making an association with s ocio -economic processes (Turner et al ., 1989; Gardner et al ., 1991; Franklin & Forman, 1987; Peralta & Mather 2000; Browder, 1996). Nonetheless, deforestation patterns have been associated with a variety of socio -economic processes, (Geist & Lambin, 2001; 2003). These processes include macroeconomic incentives (Pacheco, 2002; Hecht, 2005; Seto et al., 2002), cattle ranching policies (Mertens et al. 2002), and colonization policies (Moran, 1981; Walker et al ., 2002), to name a few. Increasing attempts to li nk spatial and social processes have produced an array of approaches throughout multiple studies that link socio -economic processes and landscape change (Walsh & Crews -Meyer, 2002; Fox et al ., 2003; Turner et al., 2004; Wood & Porro, 2002; Moran & Ostrom, 2005).
84 Linking spatial and social processes is important for understanding the implication of landscape dynamics. For example, it is known that forest patches are not only remnants of larger natural ecosystems, but represent social spaces that change over time (Browder, 1996). The size, shape and configuration of landscape patches are influenced by social forces emanating from local social actors (individuals, households and communities) as well as regional, national and global actors (governments and tran snational corporations). Individual decisions made by farmers influence the use of forest patches, and public policies create incentives to use forest lands in certain ways (Browder, 1988; Walker, 1987; 2003). Relating process to pattern to the context in which household land use decisions are made, such as economic standard of living, local and national government policies, commodity prices, and connectivity (e.g., roads), complements landscape dynamics. A process to pattern refers to the understandin g of how the mechanisms of a land-change process or more specifically in this study a socio -economic process, manifest as a pattern on the landscape (Geist, 2006). These socio -economic conditions are not static and do evolve, producing dynamic changing la ndscape structures seen as spatial patterns of LCC. This study incorporates the policy context in which household l and use decisions are created It expands the classical linkage between spatial patterns and ecological processes in landscape ecology to inc lude policy outcomes as drivers of LCC. In the Amazon, the proximate causes of deforestation are reasonably well understood, but the more indirect influence of policy factors on temporal changes has either been taken for granted or not systematically purs ued. This is important because policy regimes affecting the Amazon have changed substantially over time in most of the countries that share the basin. Further, most work on the Amazon has focused on Brazil, and this leaves aside many critical
85 Amazonian locations where deforestation is taking place. In the Peruvian Amazon, major infrastructure development projects related to the IIRSA Sur initiative1 (Initiative for the Integration of Regional Infrastructure in South America) have started to alter areas with significant ecological gradients and relatively high biodiversity (even by Amazonian standards) and diverse cultural and social landscapes (Dourojeanni, 2006). IIRSA Sur includes the paving and improvement of approximately 2,586 km of roads in Southeaster n Peru, a significant conservation and biodiversity hot spot area (Conservation International, 1999; Terborgh, 1999). This study focuses on Southeastern Peru, a region that has experienced varied periods of deforestation driven by changing socio -economic p olicies among different periods associated with different presidential administrations (Alvarez & Naughton Treves, 2003). This research tests the hypothesis that the mechanisms by which policy incentives motivate people to deforest (i.e., processes) are as sociated with increasing LCC due to logging and agriculture (i.e., patterns). To test this hypothesis I studied the province of Tahuamanu, in Southeastern Peru, particularly on the Inter Oceanic Highway corridor from Iapari to Iberia, for the period from the mid 1980s to 2007. I conducted this study in this region because of the importance in evaluating areas in close proximity to roads that pose a high risk of deforestation. Areas near roads and cities are more likely to be deforested than areas away from infrastructure routes (McCracken et al., 2002a; Carvalho et al. 2001; Nepstad et. al ., 1999, 2001; Wood & Skole 1998; Mertens et al. 2002). I examine if landscape structure changes along a systematic LCC gradient from forest to crops and pasture, to r egrowth, and built/non-forest, based on policy adoption. Forest reversal processes (regrowth and crops and pasture reverting back to forest) can be caused by policy abandonment or the policies that encouraged crop/pasture based activities ended. For example, 1 (see http://www.iirsa.org)
86 the concentration size and shape of land -cover patches will be dependent on that land covers location and its relationship to policy regimes within the respective land use history and distance to the road. I show how size, shape, and patches of land -cover are in terrelated, examine the process of converting forest and regrowth to crops and pasture or built/non-forest, and discern pattern s to associate policy modifications on the landscape (Ga rdner et al., 1987, Forman 1997). These spatial manifestation s on the ground are then linked to the land use history of the districts of Iapari and Iberia in the study area to facilitate the interpretation of socio economic processes based on policy regimes 4.2 Study Area The study area encompasses two districts in Southeastern Peru Iapari and Iberia within the province of Tahuamanu, Madre de Dios region (Figure 4 1) The Tahuamanu Province constitutes 74% of the Madre de Dios region, with an estimated human population of 10, 742 in 2007 (INEI, 2007). The specif ic study area is the path of t he newly paved Inter Oceanic Highway, which links the Brazilian southern state of Acre with Peru and the complete road path is designed to cover a road extension of 1580 km from the Brazilian border to the Peruvian. In particu lar, my study is narrowed to the road axis between the frontier town Iapari and Iberia and their secondary dirt roads which covers an area of 2040 km and an extension of approx. 70 k m. Historically, this area underwent specific historical experiences through national policies and offers an interesting case study for analyzing the interaction between road development, policy shifts, and LULCC. Currently, with the advancement of the paving of the Inter Oceanic Highway since 2005, policies favoring economi c returns may transform land change dynamics along the highway at a faster pace (Chapter 2 and 3). Iberia has seen a larger presence of public authorities headquartered in their territory, such as the Agrarian Bank until 1991 and the office of the
87 governme ntal institution Special Project Madre de Dios (PEMD). These proximity effects are expected to be play a major role in the way polices influence LULC. Since the major economic activities are defined through agriculture and logging activities, I focus on t he titled farms that encompass agriculture activities (crops and pasture) and a small area of timber concessions. The climate is hot and tropical, seasonally humid, with abundant rains from October to March and a short dry season from June to September. T he forests are evergreen and characterized by high diversity and high crowns. The vegetation consists of lowland rain forests, mainly alluvial and terra firme forests, as well as low hillside bamboo forest. Alluvial forests, which are low terraces found al ong the borders of the major rivers, are affected by river movement and probably frequent flooding. Terra firme is situated above the alluvial plains away from river courses and lakes and characterized by flat topography and variability of forest types. Th e province of Tahuamanu is also known for preserving big -leaf mahogany, one of the most critically endangered mature forests, and therefore, has raised attention among loggers and conservationists (ITTO, 2005). Land tenure comprises a mixture of indigenous reserves, small and middle -sized private holdings (mainly agriculture), state land, timber concessions, and unclaimed land (IIAP CTAR, 2001). Landuse patterns are associated with major productive activities and consist of subsistence agriculture, (mainl y rice, beans, and maize), forest extraction and, to a lesser extent, cattle ranching (INADE OEA, 1998). According to INRENA (1999), Madre de Dios region can be characterized by the following land uses: permanent crops (0.2% of the area), extensive agricul ture (0.5%), cultivated pasture for cattle ranching (3.4%), research units (0.1%), urban sectors (1%), and forest (94.8%). Although forest is still the most dominant land cover, forest
88 cover is expected to decrease with the paving of the Inter Oceanic Road as it occurred in most frontier developments elsewhere (Schmink & Wood, 1992; Pichon, 1997; Alves, 2002). In terms of agricultural profitability, areas with agricultural potential represent 28% of the total area, but only 0.6% is devoted to agricultural activities, an indication of the poor returns to farmers from cultivation, transportation problems, and lack of markets (INRENA, 1999; Mora, 1993). Despite the low population density, the region has experienced the highest migration and population growth r ates in Peru (annual population growth rate of 6%), but has lacked adequate communication and transportation infrastructure (Tahuamanu, 2001). Most government incentives have lacked the focus on competitive market opportunities, trading mechanisms, price s tability, and overall economic guidance. Within Southeastern Peru, the region of Madre de Dios has been marked by a series of boom andbust economic cycles. Most of these cycles did not last for a long time. Economic opportunities and overall development i n Peruvian tropical forests have been conditioned by unstable global markets, and have lasted only as long as the resources themselves or the absence of competition elsewhere (INADE OEA, 1998; Coomes, 1995; Giugale et al., 2007). 4.3 Methods This study constructs a LULC history of the Inter Oceanic Highway corridor from Iapari to Iberia from the mid 1980s to 2007, focusing on policies related to agricultural expansion and timber management The research builds on the following observations : a) LCC occur i n the face of shifting socio -economic conditions such as policies and road accessibility; (b) these shi fts create forest fragmentation through a systematic LCC gradient from forest to crops and pasture, to regrowth, and built/non -forest (includes conversi ons back from regrowth to forest and non-forest/built to regrowth and forest or to crops and pasture) through time along the road
89 Table 4 1 shows the three specific policy periods (I III) chosen for this study and the corresponding remote sensing data as well as the expected land -cover fragmentation outcomes measured through a variety of landscape and class metrics. Each policy period shows an increase or decrease in degree of fragmentation followed by expected metric indices measuring the changing fragme ntation. The policy periods are: a) 19851990, which corresponds to the first presidency of Alan Garcia in Peru and represents the time when credit and cattle incentives were available; b) 1991 1999, which includes the presidency of Alberto Fujimori and wi thdrawal of agricultural supports alongside the introduction of reforestation and seed improvement policies; and c) 20002005, which includes the presidential term of Alejandro Toledo and the implementation of more decentralized policy structures through c attle insemination, fish farming, copoazu plantation and agricultural mechanization. Further, since 2002, new forestry and wildlife reforms have been established to develop new concepts of sustainable forest management through the elaboration of forest management plans and new procedures to establish timber concessions In addition, the paving of the Inter Oceanic Road began in the study road axis at the end of 2005. 4.3.1 Remote Sensing Data Each policy period and their respective policies is linked to spe cific satellite images to show landscape dynamics through fragmentation based on the measurements of metric indices. Landsat satellite images for the study region were acquired for the years 1986, 1991, 1996, 2001, 2003, 2005 and 2007, for WRS Path/Row 003/068, which corresponds to the study region. These Landsat TM (1986, 1991, 1996, 2003, and 2005), ETM+ image (2001), and 2007 composite ASTER image were selected to evaluate the specific policies in five-year time increments since 1986. The frequency was decreased to a two -year increment since 2001 to capture the policy
90 events represented by shorter time intervals between the new forestry reforms in 2002 and the paving of the Inter Oceanic Highway beginning in 2005. The 2001 scene was geometrically rectifi ed and registered to a UTM WGS 1984 Zone 19S coordinate system based on a 1:100,000scale topographic map (Instituto Nacional Geogrfico, 2001 version). The 1986, 1991, 1996, 2003, 2005 and 2007 scenes were registered image -to -image with the rectified 2001 base image. The nearest -neighbor resampling technique was used, and a RMS error between 0.2 and 0.5 pixels was obtained for all images. All Landsat data were radiometrically calibrated using the CIPEC calibration method (Green et al., 2002) Ground sample data collected during fieldwork in 2003 were used to determine land -cover classes, and were divided into training and testing data sets. Field site visits collected data representative of each land cover and preliminary image clas sification identified areas to be surveyed along the road transect. Consecutive field visits during 20042006, served to revisit areas of classification uncertainty. Field observation provided insight into the structure of regrowth stages, mainly identifyi ng total height and ground cover of dominant species. Every field location was registered with a hand -held Global Positioning System (GPS) device to allow further integration with spatial data in the Geographical Information System (GIS) and image processi ng systems. In total, 295 plots covering five different land cover types were identified, and the socio economic land use history and policy implementation was recorded in 2004 via a field survey of 125 farmers and small -scale timber concessions. A detaile d description of household data collection can be found in Chapters 2 and 3. I used TC (Tasseled Cap) composite images to conduct an unsupervised classification through isodata procedure in which an original 255 -class image was aggregated to 5 -class
91 class ification. Using ground -data training signatures and through intensive familiarity with the site from field visits, spectrally similar classes were grouped and the final signatures were used to produce the five -class categorical land -cover map through a ma ximum likelihood supervised classification technique. Accuracy assessment for the 2003 image classification indicates an overall accuracy of 85.5%, which corresponds to the year most of the land-cover training samples were taken. Final land -cover classes i dentified and aggregated for the study area are: forest, built/non -forest, crops and pasture, regrowth, and haze/cloud. Classes described as forest encompassed alluvial and terra firme forest, as well as low hillside bamboo forest. Classes for built/ non -fo rest included urban areas (village settings and roads) and water (mainly rivers and creeks). Crops and pasture comprised agricultural land (mainly rice, maize, beans, and some fruit trees) and pasture Secondary forest or regrowth included secondary forest determined through farmers communication (1 2 year succession 3 5 year succession and 7 15 year succession) Ideally haze/clouds should have been omitted; however it was difficult to distinguish haze/cloud during very dry years. In particular, the 1996 and 2005 Landsat images presented extremely dry and fire prone haze conditions. Five -class LCC maps were produced for each image acquired during the study time interval. 4.3.2 Pattern M etrics Land -cover fragmentation analysis generates for each land -cover class a variety of metrics that describe fragmentation and spatial distribution from satellite -based land -cover classification (Southworth et al., 2004; Walsh et al., 2003). These algorithms quantify specific spatial characteristics that are being linked t o the processes causing LCC within a landscape. Pattern metrics were calculated for the five -class classification of the images of 1986, 1991, 1996, 2001, 2003, 2005 and 2007 for two separate areas: A digital 2006 agricultural plot
92 boundary file2 and an a rea within 2005 timber concession file3. Farm plots and timber concession areas were subset out of the classified TM, ETM+, and ASTER images (See Figure 4 1), used as agricultural (AG) and timber concession (CONC) areas and transformed to grids. AG and CON C grids were then input into FRAGSTATS to compute selected landscape metrics for each grid separately. Although there is no best choice for which metric best describes a particular pattern, specific pattern metrics were selected based on a literature rev iew of LULC and landscape fragmentation analysis (Geist, 2006; Turner, Gardner & ONeill, 2001; Nagendra et al., 2004; Southworth et al. 2002;Walsh et al. 2003). I analyzed metrics at the class level to compare forest, crops and pasture, regrowth, and b uilt/non -forest land -cover patterns between these classes and across AG and CONC. Class metrics correspond to patterns (composition and configuration) that are integrated over all the patches of a given LULC within a defined boundary (Pan et al., 2004; Yu & Ng, 2006). Further, I included landscape metrics to describe the composition and configuration of the overall landscape of the study area (Lausch & Herzog, 2002). Landscape level metrics represent patterns occurring within a preset boundary, which in thi s case represents agricultural plots and timber concessions. The landscape level and class level metrics utilized in this study were defined by McGarigal et al., (2002) as follows (Refer to Table 4 1 for expected outcomes): Landscape -level metrics chosen for this study: 1 Number of patches: total number of patches in the landscape. 2 Largest patch index in percentage: the area of the largest patch in the landscape divided by total landscape area, multiplied by 100. 2 Acquired through the Peruvian Titling Agency PETT. 3 Acquired through the Peruvian National Institute of Natural Resources INRENA.
93 3 Mean patch size: Average size of patches in h ectares. 4 Contagion in percentage: Patch type aggregation. When moving toward 0, patch types are disaggregated and interspersed. Range: 0
94 MPS. Subsistence agriculture may transition into commercial agriculture and/or larger cattle ranches, aggregating smaller patches into larger patches and increasing again MPS. This transition has rarely occurred in the study area but may be expect ed once the paving of the Inter Oceanic Road finalizes. Edge density distinguishes the amount of edge in the landscape. A less fragmented forest will have smaller ED whereas landscapes with an increased use of their land e.g. increased agricultural activ ities or timber extraction will show more ED. The largest patch index is an indication of dominance. For the forest class, a larger LPI indicates less human alteration, whereas a larger LPI for AG could be indicative of increased forms of agricultural acti vities. ENN_MN and IJI describe metrics of isolation and are measures of connectivity, adjacency, insularity, and spatial heterogeneity in the landscape. Interspersion refers to the intermixing of patches of different types. Land-cover classes describing an increase in IJI and a decrease in ENN_MN are expected less fragmented than a land -cover class with a low IJI and a higher ENN_MN. Contagion is a texture measure of the aggregation or clumping of patches. A contagion value approaching 100 indicates larg er contiguous patches in the landscape. Contagion is high when there are large aggregated patches on the landscape and low when there are many small clusters. Contagion decreases as farmers clear more small areas until small clearings coalesce into larger clearings, at which point contagion increases again (Frohn et al. 1996). Policy impacts and specific fragmentation outcomes will be associated with changing land -cover classes based on changes in spatial homogeneity and heterogeneity. For example a homog enous forested landscape will show a smaller degree of fragmentation and the following expected metric indices: smaller number of patches (NP), edge density (ED), interspersion -
95 juxtaposition index (IJI) and a bigger mean patch size (MPS), contagion (CONT), mean Euclidean nearest neighbor distance distribution (ENN_MN), and forest largest patch index (LPI). A heterogeneous forested landscape will show a larger degree of fragmentation and the following expected metric indices: larger NP, ED, IJI, and a smalle r MPS, CON, ENN_MN, and crops and pasture LPI (See detailed description in Table 4 1. However, not all policies will have the same impact. To account for time lag effects, I assume that policies will affect LULC within a two satellite image interval (five years or two years). Overall, I expect that the outcome will be dependent on specific policies at specific time periods and enhance or reduce spatial homogeneity and heterogeneity in the study region. 4.4 Results and Discussion 4.4.1 LCC from 1986 to 2007 Figure 4 2 displays the land -cover time -series patterns for AG area from 1986 to 2007. There is clear indication of a decrease in forest cover during the late 1980s and an increase in crops and pasture during the same time period, followed by an increase in regrowth at the end of 1990s. These changes are related to the policy cattle module and credit incentive policies implemented during the mid 1980s (see Table 4 1). Increased agricultural credit and support for cattle acquisition fostered expanded forest clearing, especially for cattle pasture, during this time period (Alvarez & Naughton Treves, 2003; NaughtonTreves, 2004; Chapter 2). However, these policies lasted only until the beginning of the 1990s when new policy regimes stopped agricultural credit and subsidies, and introduced new taxes. Garcias agrarian incentives of the late 1980s and their withdrawal (and the shift to other policies after 1990) accounts for a short term increase in regrowth and forest class as well as a decrease in crops and pas ture in the period of the early 1990s. A loss in forest is an indication of increased forest slash-and -burning events, whereas regrowth is a sign of regenerating forest/shifting cultivation or abandoned forest due to
96 an existing policy being discontinued. Changing areas in a state of clearing through crops and pasture is an indication of areas maintained as pasture supported by policy incentives for sustaining pasture. Maintaining pasture and large crop areas required investment and was coupled with high ma intenance costs for farmers at that time (Field surveys, 2004). Without proper policy incentives, market availability, and transportation services, growing areas of crops and pastures other than for subsistence means were difficult to sustain (Coomes, 1996; Tahuamanu, 2001). According to field data surveys, sites that remained as crops and pasture from period to period were usually pasture areas, commensurate with promoting cattle development in the late 1980s and into the 1990s, when the notion of cattle a cquisition still represented economic stability. The images since 2001 show a renewed increase in crops and pasture and built/non -forest especially around 2005, when road paving activities brought migratory settlements to the area of Iberia. LCC maps with in the forest concession boundary displayed in figure 4 3 show land cover fluctuating less than in agricultural areas. Some activity related to an increase in crops and pasture around timber concession roads is observed after the 2001 image, coinciding wit h the new forestry and wildlife law approved in 2000 and a new national forestry strategy initiated in 20024. In 2003 and 2005, significant forest loss happened close to the area of influence of the town of Iberia. This can be linked to demarcation and superimposition problems between timber concessions and agricultural areas, meaning an authorized advancement of crops and pasture activities into the boundaries of timber concessions (field notes, 2003, 2005). Contrary to the agricultural land -cover boundary the 2007 image suggests an increase of forest loss compared to the 2005 image. Although timber concessions follow strict forest management plans enforced by 4 INRENA website, http://www.inrena.gob.pe
97 the National Institute of Natural Resources (INRENA), increased deforestation close to improved c oncession roads is slightly visible. It is worth noting that the analysis included only a portion of a timber concession, mainly showing timber road accessibility. Overall, the amount of deforestation within the boundaries of agricultural and timber conces sion areas are significantly different. In the case of the agricultural area boundary, farming is mainly located along the major Inter -Oceanic Highway and secondary dirt roads, and tends to expand and contract in part according to the availability of polic y incentives and improved road, frontier conditions, and other complex conditions (Mertens et al., 2002; Geist & Lambin, 2003; Wood & Porro, 2002, NaughtonTreves, 2004, Chapter 2 and 3). In the case of the timber concessions, deforestation is related to e xpanded and improved timber roads (Kaimowitz & Angelsen, 1998; Nepstad et al ., 2001). 4 4 2 Landscape Metrics Table 4.2 displays changes in landscape patterns for agricultural and timber concession areas. The number of patches in the case of agriculture areas increases drastically until 1996. The next five year interval (19962001) shows only a slight increase, and then falls in 2003, and after that the number of patches sharply increases and almost doubles between 2005 and 2007. This progression indicates a se vere but discontinuous fragmentation process happening in the past 20years, which is being reinforced with the increase in edge density number from 41.63 in 1986 to 154.03 in 2007, as well as the decrease of mean patch size from 9 ha to only 1 ha by 2007. The contagion metric reveals a sequence of landscape disaggregation, since smaller clusters are shown after 1986 possibly reflecting socioeconomic circumstances resulting from policies adopted at the end of 1980s and later abandoned. These results agree with my predictions. After 1986, most farmers received loans, which they used to purchase cows, develop pasture, and increase crop profitability (field notes, 2004; Alvarez & NaughtonTreves, 2003). Thus,
98 contagion decreased as farmers cleared many small, disconnected areas. Until 1996, small agricultural areas were converted into pasture, but since pasture and cattle were not profitable and policies promoting cattle disappeared with 19902000 regime change, some of these areas reverted back to regrowth (Na ughton Treves, 2004). Contagion would have leveled off if enough farmers had entered the region, occupied land, adopted policies promoting agriculture and cattle, and cleared small areas for agriculture. However, this did not steadily happen in Southeaster n Peru (IIAP -CTAR, 2001; Tahuamanu, 2001). At some point a threshold was possibly reached whereby humaninduced changes cause more aggregation of patches into contiguous areas. This reversal is possible in the near future, and it remains crucial to monitor if the paving of a road will encourage expansion of pre existing agricultural activities (Maki et al ., 2001; Kaliola et al. 2001; Garcia, Raez & Boggio, 2008). The timber -concession landscape is dominated by large forested patches as shown by the stable largest patch and contagion indexes, both of which indicate relatively homogenous forested landscapes and the dominance of forest (Table 4 2). The number of patches and edge density display several high peaks (1996, 2001 and 2007), while mean patch size s hows an inverse pattern during the same years, assuming a slight increase in fragmentation. These results point toward a relationship between the development of the forestry policy framework and logging activity history in the region. Prior to 2002, forest s were harvested under the old regime which involved a large number of harvesting contracts, most of them with an area of less than 1,000 hectares (ITTO, 2005a). Specifically, the interval between 1996 and 2001 relates to limited implementation of forestry policies and weak institutions, which favored unsupervised and illegal timber activities. This situation in turn led to the creation of the new forestry law in 2001, followed by the laws implementation in 2002 (ITTC, 2003; Caillaux & Chirinos 2003). As
99 such, the patterns displayed for NP, LPI, and MPS can be linked to forestry harvesting and forestry policy processes before and after 2001. With the new forestry law, longterm concessions over larger areas were established, and compliance to forest management plans was required. Under this new strategy, public forests are classified as permanent production forests with harvesting units of 5,000 to 10,000 hectares (SPDA -INRENA, 2003). Although the new law sought to re -establish and monitor forest harvesting practices, its implementation has been characterized by inadequate state planning, as well as lack of financial and human capital and insufficient technical, business, and forest management experience among concessionaires (Elgegren & Lee, 2006). This sit uation may have pushed timber concessionaires to spend time solving their technical problems and adjusting to formal processes rather than in the field. At the same time, the new forestry regime may have motivated other more experienced illegal logging com panies to enter concession areas and disregard the strict regulations, which could explain a higher NP and ED, and a smaller MPS, CONT, and LPI after 2007 (Galarza & La Serna, 2005; Cossio, 2009). Currently, forestry still represents only a minor part of t he changing economic systems in the Peruvian Amazon, however future processes emanating from timber activities are expected to produce spatial patterns and transform the landscape, since livelihood systems revolve around the forestry sector in this area of Southeastern Peru. 4.4.3 Class Metrics Tables 4.3 and 4.4 show class metrics from 19862007 within agricultural (AG) and timber concession (CONC) boundaries, respectively. Land -cover change rates and the analyzed class metric indices relate to the expect ed fragmentation outcomes listed in Table 4 1. In the case of AG, dense forest decreased from 97.2% in 1986 to 85% in 1991, agreeing with my prediction and is therefore a strong indicator of policy outcomes associated with the mandate of the Garcia adminis tration and its policies offering credit incentives and support for cattle. Since farmers
100 benefited from the above mentioned policies, forests were lost to crops and pasture, shown by a noticeable increase in crops and pasture from 1.2% in 1986 to a high of 7.1% in 1991 after which it fell back to 5.3% in 1996. Once the Garcia policies ended at the beginning of the 1990s, forest had a chance to revert back up to 86.6% in 1996, and regrowth increased at the beginning of the 1990s as well. This can be shown a s well by a peak in NP in regrowth in 1996 suggesting a link to the absence of policies prior to 1990, resulting in replacement of crops and pasture by regrowth. Areas with lower crop productivity tend to have more regrowth due to longer fallowing periods and or higher land degradation ( Moran et al., 2000; Walker, 1999). Further, regrowth can be related to unproductive land or land abandonment during short term forest transition episodes (Perz & Skole, 2003a). In this case, the gradient of forest to crops a nd pasture, to regrowth and back to forest relates to cycles of frontier expansion and retraction matching policy cycles that foster or impede a frontier development. The 2000s have shown a renewed decrease in forest cover from 81.3% in 2001 to 80.1% in 2 003 and down to 73.5% in 2005 and 72% in 2007, probably related to a significant increase in crops and pasture after 2001 to 9.8% in 2003 and 13.9% in 2005, however falling slightly again to 12.3% in 2007. These results do agree partially with my predictions and have several possible explanations. First, policies such as agricultural mechanization, cattle insemination, and copoazu plantation during 20002005 did influence an increase in crops and pasture and a loss in forest, and consequently a higher degre e of fragmentation, an effect that decreased after 2005 once policies were implemented under a different administration (20052010). Further, the effects of road building definitely relates to an increase in non-forest/built land area. However, the effects of crops and pasture post 2005 could point toward a short term post -paving effect, meaning that the initial expected benefits from the road paving or the absence of policies
101 promoting agricultural activities did not further attract land based activities o n forest use. These results were unpredicted. Another explanation could be related to climatic conditions in 2005, when fire -prone conditions led to crop losses (Marengo et al. 2008). Very interesting is the trajectory of regrowth. Distances (ENN_MN -AG) a mong patches of regrowth have been reduced by more than a half from 1986 to 2007 from 169 m to 69 m. On the other hand, distances among forest and crops and pasture patches display a more uniform trend throughout the period. These trends can be related to the role of socio-economic determinants of regrowth (Perz & Skole, 2003; Coomes et al., 2000). As land becomes less profitable, due to shortages in agricultural credit, incentives and market, regrowth acts as a transition between forest, crops and pasture. The effect of road presence can be linked to IJI which shows defined arrangement of patches occurring close to each other or specifically adjoining to the main road. Overall in AG, forest is still the dominant class, however the largest forest patch decli ned from almost 18% in 1986 to 4% in 2003, and only crops and pasture showed a minor dominance in 1991 and 2005. Humaninduced processes act as underlying causes of spatial patterns on the landscape (Wood & Skole, 1998). The results for AG in this study confirm that landscape patterns measured through size, shape, connectivity, interspersion, and patches of land-cover are interrelated and can be linked to policy processes. No doubt, forest dynamics in AG have been conditioned by distinct periods of socio -ec onomic policies as seen throughout the class pattern indices. In the case of CONC, PLAND reveals a more or less stable forest cover class ranging from 99 to 97% throughout the 21 year period. LPI shows a steady dominant trend for forest as well and ENN MN reveals almost no change in distance between forest patches. Crops and pasture, regrowth, and built/non-forest classes show a consistent pattern of transformation, although the years after 2001 show the highest activities. In particular, forest loss occurr ed
102 primarily around the major road and as a consequence of farming activities superimposing timber concessions. Regrowth and crops and pasture have the largest total numbers of patches, indicating that these two classes are more fragmented than stable forest. In the case of CONC, fluctuating rates of change between 2001 and 2007 indicate that change dynamics are related to factors such as the presence and adoption of particular forestry policies. For example, NP for regrowth stands out with a peak in 2001 a nd a sharp decrease in MPS in correlation with timber reforms, agreeing with my predictions. Overall, deforestation through logging has been minimal, although the changing environment around timber reforms may affect the future of timber forest loss. These transitions capture the general trends of landuse dynamics in relation to policy implications; however it is imperative to add that other more complex processes could be interacting within main transition patterns, since not all LULCC and landscape dynam ics can be solely attributed to policy effects. 4 .4.4 Implications for LULCC Combining remote sensing, landscape dynamics and policy implications opens a new perspective in understanding the mechanism through which people alter land use. Characterizing th e nature of this change has been a major target in studies dealing with human environmental interactions and LULC (Lambin et al., 2001; Geist & Lambin, 2001; Southworth et al ., 2004). However, choosing certain patterns of change and associating governmenta l policies as the human agent shaping the structure and function of landscape change is less studied and will need further attention, considering the many policies that influence the environment and human behavior (Geist, 2006; Lambin & Geist, 2006). Pache co (2002) linked forest loss with the presence of public policies in Bolivia without incorporating a spatial and temporal dimension. Other studies such as in Fox et al (2003), Liverman et al. (1998), Turner et al (2004) and Walsh
103 & Crews -Meyer (2002) have all tried to link LULCC and remote sensing to local decision makers incorporating multiple levels of analysis and based on numerous theoretical frameworks. This study complements these approaches by not only focusing on a specific policy process and rela te process to pattern, but incorporating LULC, remote sensing, and socio -economic boundaries, such as AG and CONC. It is at this spatial level where LULCC influences the composition and structure of the landscape and may offer opportunities to not only so cialize the pixel but socialize the patch. Policy adoptions in AG and CONC were confirmed during field work. However, a set of problem facing the analysis are the following. AG and CONC units may not have been the proper units to establish a spatial and temporal relationship between people reacting toward policies and the landscape. The scale applied to the analysis may not entirely correspond to the scale at which farmers and logging operations decided to adopt a policy. However, relating farm plot and timber concession areas to assess the temporal imprints of socio -economic processes was the best approximation for linking process and pattern in this study. With regard to scale and grain, the agricultural and timber concession demarcation boundaries were defined as valid for 2005, however the legal boundaries may have been different in previous years of the analysis. Considering, that most agricultural areas were located close to the main road in the past, I am confident that the boundary used in this stu dy covers a great portion of past AG areas. Further, the measurement of timing and sequencing of changes in the character of policy effects is high priority since a time lag effect poses difficulties in regard to when policies are shown on the landscape. A methodological alternative is to relate satellite images as close to policy introduction/ending through time series analysis (before and after a change in policy) and confirm these policy periods through field work, as done in this study. I assumed that p olicy effects would be linked to a satellite image within a two -year interval.
104 Future work should attempt to establish a more in depth quantifiable relationship between changes in polices and land use decision, and ultimately LULCC and metric values. This study further expanded into a two -year interval of analysis to capture additional temporal fluctuation in metric values at both the class and landscape level when socio -economic conditions fluctuated more frequently. This information would have been lost if the analysis included larger time -intervals in the presence of a rapid sequence of policy changes due to the paving of the Inter Oceanic Road and forestry reforms. Adding a two year interval distinguished changes in fragmentation in shorter time periods Choosing an appropriate temporal interval of analysis in accordance with the pace of shifts in policies is crucial for capturing relevant processes in the landscape. All these issues add complexity to linking LULCC, landscape dynamics and policy. Undoub tedly, land use decision -making is a dynamic process always changing in response to a number of variables, including both social and political policies, as well as economic variables. Certainly, not all outcomes can be explained by associating specific pol icies and their time periods with landscape dynamics, since more complex factors occur, such as lack of transportation, lack of markets and adequate prices, and lack of basic services. Other causal factors behind land use decision that transform the landsc ape do play a role as well. Describing how human actions, mediated and dictated by wider political -economic drivers, affect human patterns and LULCC is a difficult task and this study offers only an alternative complement to the many LULCC/landscape dynam ic efforts. 4.5 Conclusion In this paper, I described varied socio -economic related contexts associated with landscape dynamics over the past 21 years in Southeastern Peru. The fragmentation process was related to the spatial location of timber concession and agricultural areas and the associated road
105 network. Landscape dynamics indicated a varying level of conversion from forest to crops and pasture, regrowth, secondary forest and built/non -forest, with regrowth, crops and pasture and forest reverting bac k at times when policies supporting agriculture were absent. The adoption of policy incentives and the overall policy framework influenced the landuse/cover conversion process by small -scale farmers and small timber concessions, and added a social context to landscape patches. Although changes in pixels and patches cannot only attributed to policy effects and other more complex factors are at stake, the results show that policies are periodspecific and policy indicators referring to distinct periods bac k in time exhibit contrasting associations with a given LULC. The study suggests as well that more detailed understanding is necessary for each policy period and for each specific policy associated with LULCC. Trajectories of policies are controlled by dif ferent factors at different times and in different geographic spaces indicating other causal factors behind land use decisions that transform the landscape, e.g. levels of public support, technological expertise, credit availability, education level, and c limate change events. Further studies could expand the issue of analyzing policy implications by including a more comprehensive measurement of the local socio-economic contextual factors linking people to land and land to people. A major contribution o f this study is a more comprehensive appreciation between policies and landscape pattern changes. The socio economic processes considered here are helpful and indicate a relationship between fragmentation of the landscape and policy implementation. Complem enting the assessment of landscape dynamics by incorporating a socio-economic component considerably improves our understanding of the impacts of complex policy frameworks on the environment and the implications for LULCC and pattern metrics. This is
106 espec ially relevant if humans continue to alter the amount, pattern, and composition of global vegetation.
107 Figure 4 1. Study area in the province of Tahuamanu, Madre de Dios Region, Southeastern Peru.
108 Table 4 1. Landsat TM, ETM+, and ASTER scenes of the study area linked to specific policies and expected land cover fragmentation outcomes. Policies and Satellite Image Expected Land Cover Fragmentation Outcome Time Period Acquired as described through a Variety of Metrics1 Evaluation of initial conditions Initial = pre 1985 Landsat TM Homogenous forested landscape July 12, 1986 < degree of fragmentation Expected metric indices: < NP, >MPS, CONT >ENN_MN, Forest: >LPI I Mandate of A. Garcia (1985 1990) 1. Credit Incentives Landsat TM Heterogeneous forest with larger and 2. Cattle Module* Oct. 14, 1991 increased number of patches > degree of fragmentation Expected metri c indices: >NP, ED, >IJI, MPS, CONT >ENN_MN, Forest: >LPI III Mandate of A. Toledo (2000 2005) 1. Agricultural Mechanization Landsat ETM+ Throughout III and all ETM+/TMs: 2. Fish Farming July 29, 2001 Heterogeneous forest with larger and 3. Cattle Insemination increased number of patches 4. Copoazu Plantation > degree of fragmentation Expected metric indices: 5. New Forestry Reform Landsat TM >NP, ED, >IJI,
109 Figure 4 2. Land -cover change within the boundary of agricultural areas from 1986 to 2007.
110 Figure 4 3. Land -cover change within the boundary of timber concession areas from 1986 to 2007.
111 Table 4 2. Changes in landscape pattern indices within the boundaries of agricultural area and timber concession areas between 1986 and 2007 in the road axis Iapari Iberia, Southeastern Peru 1986 1991 1996 2001 2003 2005 2007 AGRICULTURAL AREAS Number of Patches 5827 10232 25846 28787 24346 29057 50007 Largest Patch Index 17.46 14.95 15.36 12.73 10.51 10.69 6.98 Mean Patch Size 8.68 4.94 1.96 1.76 2.08 1.74 1.02 Contagion 90.38 73.41 71.98 66.32 66.54 60.12 55.19 Edge Density 41.63 68.25 101.45 116.08 108.77 127.19 154.03 1986 1991 1996 2001 2003 2005 2007 TIMBER CONCESSION BOUNDARY AREAS Number of Patches 7432 7614 26001 27536 17561 13347 20392 Largest Patch Index 99.18 96.35 97.79 97.04 97.63 97.80 96.64 Mean Patch Size 18.21 17.77 5.20 4.91 7.70 10.14 6.42 Contagion 97.06 92.45 92.67 91.38 93.34 93.99 91.47 Edge Density 9.52 14.22 26.73 29.24 21.35 17.64 22.86
112 Table 4 3. Class pattern indices for agricultural areas (AG): PLAND (percentage of land area), NP (number of patches), MPS (mean patch size), LPI (largest patch index), ENN_MN (mean Euclidean nearest neighbor distance distribution), and I JI (interspersion juxtaposition index. 1986 1991 1996 2001 2003 2005 2007 PLAND AG (%) Forest 99.23 96.42 97.85 97.15 97.79 97.94 96.98 Built/Non Forest 0.43 0.25 0.22 0.59 0.58 0.35 0.64 Crops/Pasture 0.03 0.18 0.34 1.25 0.36 0.59 0.83 SF/Reg rowth 0.30 0.30 0.92 1.01 1.27 1.11 0.63 NP AG Forest 190 281 195 537 570 514 1531 Built/Non Forest 2738 801 787 5140 3149 1789 1514 Crops/Pasture 196 1469 4425 11917 1937 1081 4936 SF/Regrowth 4205 3415 11162 9895 11842 9923 6987 MPS AG (ha) Forest 706.70 464.30 678.96 244.79 232.13 257.84 82.99 Built/Non Forest 0.21 0.42 0.38 0.16 0.25 0.27 0.55 Crops/Pasture 0.17 0.16 0.10 0.14 0.25 0.74 0.22 SF/Regrowth 0.10 0.12 0.11 0.14 0.15 0.15 0.12 LPI AG (%) Forest 99.18 96.35 97.80 97.04 97.63 97.80 96.64 Built/Non Forest 0.01 0.01 0.01 0.01 0.01 0.01 0.01 Crops/Pasture 0.01 0.00 0.00 0.01 0.01 0.06 0.01 SF/Regrowth 0.00 0.00 0.00 0.00 0.01 0.01 0.00 ENN_MN AG Forest 54.41 53.09 53.8 5 52.83 48.50 51.49 51.69 Built/Non Forest 287.20 246.26 479.55 221.84 204.03 134.67 191.31 Crops/Pasture 382.21 176.37 236.77 132.35 172.81 104.41 131.10 SF/Regrowth 260.47 188.97 151.52 138.09 129.41 124.63 139.76 IJI AG (%) Forest 7 1.94 85.44 82.71 76.15 62.75 59.94 91.27 Built/Non Forest 31.82 83.31 58.48 43.65 48.99 68.06 78.35 Crops/Pasture 67.70 64.53 32.63 36.69 67.33 60.81 75.04 SF/Regrowth 6.66 36.16 13.66 35.81 13.58 14.61 40.15
113 Table 4 4. Class pattern indices for tim ber concession areas (CONC): PLAND (percentage of land area), NP (number of patches), MPS (mean patch size), LPI (largest patch index), ENN_MN (mean Euclidean nearest neighbor distance distribution), and IJI (interspersion juxtaposition. 1986 1991 1996 20 01 2003 2005 2007 PLAND CONC (%) Forest 99.23 96.42 97.85 97.15 97.79 97.94 96.98 Built/Non Forest 0.43 0.25 0.22 0.59 0.58 0.35 0.64 Crops/Pasture 0.03 0.18 0.34 1.25 0.36 0.59 0.83 SF/Regrowth 0.30 0.30 0.92 1.01 1.27 1.11 0.63 NP C ONC Forest 190 281 195 537 570 514 1531 Built/Non Forest 2738 801 787 5140 3149 1789 1514 Crops/Pasture 196 1469 4425 11917 1937 1081 4936 SF/Regrowth 4205 3415 11162 9895 11842 9923 6987 MPS CONC (ha) Forest 706.70 464.30 678. 96 244.79 232.13 257.84 82.99 Built/Non Forest 0.21 0.42 0.38 0.16 0.25 0.27 0.55 Crops/Pasture 0.17 0.16 0.10 0.14 0.25 0.74 0.22 SF/Regrowth 0.10 0.12 0.11 0.14 0.15 0.15 0.12 LPI CONC (%) Forest 99.18 96.35 97.80 97.04 97.63 97.80 9 6.64 Built/Non Forest 0.01 0.01 0.01 0.01 0.01 0.01 0.01 Crops/Pasture 0.01 0.00 0.00 0.01 0.01 0.06 0.01 SF/Regrowth 0.00 0.00 0.00 0.00 0.01 0.01 0.00 ENN_MN CONC Forest 54.41 53.09 53.85 52.83 48.50 51.49 51.69 Built/Non Forest 287 .20 246.26 479.55 221.84 204.03 134.67 191.31 Crops/Pasture 382.21 176.37 236.77 132.35 172.81 104.41 131.10 SF/Regrowth 260.47 188.97 151.52 138.09 129.41 124.63 139.76 IJI CONC (%) Forest 71.94 85.44 82.71 76.15 62.75 59.94 91.27 Bui lt/Non Forest 31.82 83.31 58.48 43.65 48.99 68.06 78.35 Crops/Pasture 67.70 64.53 32.63 36.69 67.33 60.81 75.04 SF/Regrowth 6.66 36.16 13.66 35.81 13.58 14.61 40.15
114 CHAPTER 5 CONCLUSION The objective of this dissertation has been to associate policy regimes and specific policies with land use/land -cover change over a period of 20 years. I studied landuse change and consequent land -cover change as responses to national socioeconomic development policy changes. To get a better understanding on this t opic, in this dissertation I test whether agricultural policy variables affected land use outcomes by local landholders, whether past and current agricultural policies have influenced land use decision -making processes of future landuse plans, whether the presence of specific policies conditioned land transformation changes of forest, built -non-forest, crops and pasture, and regrowth -cover classes, and whether agricultural and logging policies have had an effect on landscape dynamics. These research c oncerns were answered in the four research papers comprising this dissertation. The first paper in this dissertation shows that government policy incentives, even those from many years before the time of fieldwork, influence farm level land use. Variables related to earlier policy periods as well as more specific current policies such as PEMD activities exhibit significant effects on land use indicators. For example, adopters of policies associated with cattle expansion cause an increase in pasture areas a t some point whereas an increase in regrowth prevails once cattle expansion policies are eliminated. The findings also confirm that policies are not universally adopted, and that adoption of policy incentives can differentiate land use among farms. The fin dings also show that the relationship between public policies and land use are very complex, and not all land use decisions can be attributed to policy events and not all farmers have adopted policies at specific time periods. The second paper suggests tha t past polices in some instances influence farmers land use plans. The results serve as a basis for understanding the formation of plans for future actions
115 involving forest and agricultural resource use. The findings suggest as well that context -specific local conditions, such as settlement history, geographic location, biophysical attributes, and institutional characteristics are decisive influences on prospective landuse plans and the resulting land transformation. The results link past to future while recognizing differences among farmers, partly due to policy adoption. The third paper confirms the effects of the adoption of policy incentives and the overall policy framework over the LULC conversion process by small -scale farmers and small timber conce ssions. Further, changes in landscape dynamics through fragmentation process es are conditioned by the spatial location of timber concession and agricultural areas. Landscape metrics allow for comparisons of spatial patterns of deforestation at time interva ls and correlations of temporal pattern changes and policy affiliations. The results provide information on forest change associated with the specific actions of policies adding explanatory power to the social component of a patch. 5 .1 Significance of Fi ndings Collectively, these three papers demonstrated that policies do affect land use/land -cover change over time. By using remote sensing analysis and household survey methods, this dissertation revealed how trajectories of policies have been controlled b y different factors at different times and in different geographic spaces. In particular, this study has not only allowed for a comparison of land use in time periods associated with distinct policy regimes, but has added a focused test of policy effects on land use, by comparing land use and land use change among people who took advantage of a policy to those who did not. The study contributed to a better understanding of tropical deforestation and the causes affecting changes in deforestation by incorpor ating a policy context. As the effects of policies become clearer, focusing on the adoption or non adoption of policies will help monitor the course of forest transformations.
116 Similar approaches can be used for countries that share the Amazon and as a means to keep oversight of the many policies that are being put into practice. 5 .2 Research Considerations and Future Work Although the results show that the effects of policies are period-specific and spatially explicit, not all outcomes can be explained by associating specific policies with specific time periods, indicating that for each policy period and for each specific policy, there are more complex factors are at stake, such as lack of transportation, lack of markets and adequate prices, and lack of ba sic services. Indeed, the relationships between public policies and LULC are complex, in part because there are other causal factors behind land use decisions that transform the landscape, e.g. levels of public support, technological expertise, credit ava ilability, education level, and climate change events. These and other factors comprise the different combinations of conditions that play a relevant role in determining farmers perception s and expectation s toward land uses. In this regard, further studie s could expand the issue of adoption and non adoption of policies and include a more comprehensive measurement of the local socio-economic contextual factors linking people to land. Future work studying social change as stemming from policy change in analysis of land use/land -cover should further expand in the measurement of social organizations and institutions that shape the nature of daily activities and the social organization of behavior. The ability to view physical attributes of land cover from above does not provide any immediate tools for direct measurements of changes over time in the character of organizations and institutions, and especially how quickly they affect the decision-making processes. In addition, the measurement of the timing and sequ encing of changes in the character of policy organizations and institutions is a high priority.
117 Household surveys should further study human interaction with land to measure the experiences and activities of individual people and groups, communities, gove rnment agencies, firms, NGOs, and their plans, expectations, and preferences. The focus into these measurements would be extremely valuable for understanding how people use land, how they plan to use land in the future, and factors that may promote changes in land use over time. In this dissertation, the adoption and non adoption of policies has been used to connect between people and land use; however, other tools can be used to measure the systems of policy organization that link local actors in the past to predict current and future policyland associations. Theoretical and empirical attention to a broad range of institutions and organizations is needed to construct a more comprehensive understanding of the links between people and LULCC that goes beyond what has been presented in this dissertation. The empirical focus and measurement strategies of LULC science studies will need to move beyond what is being studied, such as using combinations of ethnographic, household and individual surveys, and remote se nsing and links with a policy component. As new tools for measuring the social context of human and land-cover interactions incorporating a policy component become available, tools and approaches used in this study offer a potential starting point. Future analysis of deforestation could benefit from integrating a policy component into the analysis equation, and expand ing on the approaches and ideas taken on in this dissertation, and could serve as a guide for policy makers and resource managers to make app ropriate judgments towards deforestation.
118 LIST OF REFERENCES Alvarez L. N., (2001). Trayectorias de la deforestacin en la zona de Tambopata, Madre de Dios Reporte preliminar del anlisis de imgenes de satlite para los aos 1986, 1991 y 1997. (Unpubli shed). Alvarez, N. & Naughton Treves, L. (2003). Linking national agrarian policy to deforestation in the Peruvian Amazon: A case study of Tambopata, 19861997. Ambio, 23, 269274. Alves, D. (2002). An analysis of the geographical patterns of deforestati on in the Brazilian Amazon in the period 19911996. In C.H. Wood & R. Porro (Eds.), Deforestation and land use in the Amazon ( pp. 95 106). Gainesville: University of Florida Press. Alves, D. (2002a). Space Time Dynamics of Deforestation in Brazilian Amaz on. International Journal of Remote Sensing, 23, 29032908. Andersen, L.E. & Reis, E.J. (1997). Deforestation, development and government policy in the Brazilian Amazon: an econometric analysis. Working Paper 513, Rio de Janeiro: Instituto de Pesquisa Eco nmica Aplicada (IPEA). Andersen, L.E., Granger, C.W., Reis, E.J., Weinhold, D. & Wonder, S. (2002). The dynamics of deforestation and economic growth in the Brazilian Amazon. New York: Cambridge University Press. Angelsen, A. & Kaimowitz D. (1999). Ret hinking the causes of deforestation: lessons from economic models. The World Bank Research Observer 14(1), 7398. Arima, E.Y., Walker, R., Perz, S.G. & Caldas, M. (2005). Loggers and Forest Fragmentation: Behavioral Models of Road Building in the Amazon B asin. Annals of the Association of American Geographers, 95(3), 525 541. Bebbington, A. & Perrault, T. (1999). Social capital, development and access to resources in highland Ecuador. Economic Geography 75(4) 395418. Binswanger, H. (1991). Brazilian pol itics that encourage deforestation in the Amazon. World Development 19(7), 821829 Browder, J.O. (1996). Reading colonist landscapes: social interpretation of tropical forest patches in an Amazonian frontier. In J. Schelhas & R. Greenberg (Eds.), Forest Patches in Tropical Landscapes (pp.285 299). Washington, D.C.: Island Press. Browder, J.O. (1988). Public Policy and Deforestation in the Brazilian Amazon. In R. Repetto & M. Gillis (Eds.), Public Policies and the Misuse of Forest Resources (pp. 247297). Cambridge: Cambridge University Press.
119 Brown, D.G., Duh, J. D. & Drzyzga, S.A. (2000). Estimating error in an analysis of forest fragmentation change using North American Landscape Characterization (NALC) data. Remote Sensing of the Environment, 71, 1061 17. Carvalho, G., Barros, A.C., Moutinho, P. & Nepstad, D., (2001). Sensitive development could protect Amazonia instead of destroying it. Nature 409, 131. Cattaneo, A., (2001). Deforestation in the Brazilian Amazon: Comparing the impacts of macroeconomi c shocks, land tenure, and technological change. Land Economics 77(2), 219240. Cattaneo, A. 2002. Balancing Agricultural Development and Deforestation in the Brazilian Amazon Research Report 129, IFPRI. International Food Policy Research Institute, Cai llaux, J. & Chirinos, C. (2003). El caso Tahuamanu: cuando el bosque toca m adera. Lima: SPDA. Chomitz, K.M., & Gray, D.A., (1996). Roads, land use, and deforestation: A spatial model applied to Belize. World Bank Economic Review 10, 487512. Conservation International, (1999). Complexity and fragility: Biodiversity conservation and oil exploration in Tambopata-Candamo, Peru. CI -Peru Ediciones. Coomes, O. T. (1995). A century of rain forest use in Western Amazonia: lessons for extraction based conservatio n of tropical forest resources. Forest & Conservation History 39, 108120. Coomes, O.T. (1996). State credit programs and the peasantry under populist regimes: lessons from the APRA experience in the Peruvian Amazon. World Development 24(8), 13331346. Coomes, O. T., Grimard, F. & Burt, G.J. (2000). Tropical forests and shifting cultivation: secondary forest fallow dynamics among traditional farmers of the Peruvian Amazon. Ecological Economics 32, 109 124. Cossio, R. (2009). Capacity for Forest Manageme nt among Private Small Medium Forest Enterprises in Madre de Dios Dissertation Manuscript. Gainesville: University of Florida. Cropper, M., Puri, J., & Griffiths, C., (2001). Predicting the location of deforestation. Land Economics 77(2), 172186. CTAR (2002). Plan estratgico concertado de desarrollo Madre de Dios 20022004. Puerto Maldonado: Per Dourojeanni, M. (1990). Amazonia: que hacer? Iquitos, Per: Centro de Estudios Teolgicos de la Amazonia (CETA). Dourojeanni, M. J., (2006). Estudio de caso sobre la carretera Interocenica en la Amazonia sur del Per. SERVIGRAPH E.I.R.L., Lima, Per.
120 Eden, M.J. (1998). Forest and environmental degradation. In: F.B. Goldsmith (Ed.) Tropical Rain Forest: A Wider Perspective, London, UK: Chapman & Hall. Elgegr en, J. & Lee, D. (2006 ). Deforestation in Peru Working Paper. Washington, D.C.: World Bank. EORI (Centro de Investigacin y P romocin Regional (1990). Borrador de lineamientos de poltica para el desarrollo de la selva de Madre de Dios. Documento present ado a la Secretara de Planificacin del Gobierno Inka. (unpublished). Escobal DAngelo, J., (1992). I mpacto de la poltica de precios y de crdito agrcola sobre la distribucin del ingreso en el Per: 1985-1990. Documento de trabajo 18, Lima: GRADE. Fe arnside, P.M. (2000). Global warming and tropical landuse change: greenhouse gas emissions from biomass burning, decomposition and soils in forest conversion, shifting cultivation and secondary vegetation. Climatic Change 46(12), 115158. Ferraz S.F.B., Cap o, L.M.S.A.C. & Vettorazzi, C.A. (2006). Temporal scale and spatial resolution effects on Amazon forest fragmentation assessment in Rondnia. International Journal of Remote Sensing 27(3), 459 472. Ferraz, S.F.B., Vettorazzi, C.A., Theobald, D.M. & B allester, M.V.R. (2005). Landscape dynamics of Amazonian deforestation between 1984 and 2002 in central Rondnia, Brazil: assessment and future scenarios. Forest Ecology and Management 204, 6783. Food and Agricultural Organization of the U.N. (FAO) (2006). Global Forest Resources Assessment 2005: Progress towards Sustainable Forest Management Rome, Italy: FAO. Forman, R. (1997). Land Mosaics The ecology of landscapes and regions Cambridge: Cambridge University Press. Forman, R.T.T & Godron, M. (198 6). Landscape Ecology New York: John Wiley and Sons. Fox, J., Rindfuss, R., Walsh, S. & Mishra, V. (Eds.) (2003). People and the Environment. Approaches for linking household and community surveys to remote sensing and GIS. Boston: Kluwer Academic Publis hers. Franklin, J.F. & Forman, R.T.T. (1987). Creating landscape patterns by forest cutting: ecological consequences and principles. Landscape Ecology 1, 5 18. Frohn, R.C., McGwire, C., Dale, H. & Estes, J.E. (1996). Using satellite remote sensing analys is to evaluate a socio -economic and ecological model of deforestation in Rondnia, Brazil. International Journal of Remote Sensing, 17(16), 32333255. Futemma, C. & Brondizio, E. (2003). Land reform and landuse changes in the lower Amazon: implications f or agricultural intensification. Human Ecology 31 (3), 369402
121 Galarza, E. & La Serna, K. (2005). Las concesiones forestales en el Per : Cmo hacerlas sostenibles? In R. Barrantes (Ed.) La poltica forestal en la Amazonia andina. Estudio de casos: Boliv ia, Ecuador y Per (pp. 445600). Serie Diagnostico y Propuesta. Consorcio de Investigacin Econmica y Social (CIES) Garca, A., Raez Luna, E. & Escamilo Boggio, L. (2008). Grupo de trabajo de la sociedad civil para la Interocenica Sur: Una experiencia de accin colectiva ciudadan a ante un megaproyecto. Accin efectiva contra amenazas a la diversidad biolgica y cultural en el mbito de la Interocenica Sur (IOS) en el Corredor de conservacin Vilcabamba Ambor Fase II. Puerto Maldonado: Pronaturaleza. Gardner, R.H., Milne, B.Y., Turner, M.G. & ONeill, R.V. (1987). Neutral models for the analysis of broad-scale landscape patterns. Landscape Ecology 1, 1928. Gardner, R.H., Turner, M.G., ONeill, R.V. & Lavorel, S. (1991). Simulation of the scale depen dent effects of landscape boundaries on species persistence and dispersal. In M.M. Holland, P.G. Risser & R.J. Naiman (Eds.) Ecotones: the Role of Landscape Boundaries in the Management and Restoration of Changing Environments (pp.7689). New York: Chapman and Hall. Geist, H.J. (Ed.) (2006). Our Earths Changing Land. An Encyclopedia of Land-Use and LandCover Change. Vol. 2. Westport, CA: Greenwood Press. Geist, H.J. & Lambin, E.F. (2001). What drives tropical deforestation? A metaanalysis of proximate a nd underlying causes of deforestation based on subnational scale study evidence. LUCC Report Series N. 4 : University of Louvain. Geist, H.J. & Lambin, E.F. (2002). Proximate causes and underlying driving forces of tropical deforestation. BioScience 52 (2), 143150. Geist, H.J. & Lambin, E.F. (2003). Regional differences in tropical deforestation. Environment 10 (6), 22 36. GESUREMAD, (1998). Plan estratgico de desarrollo del departamento de Madre de Dios 19982008. Gerencia Sub Regional de Madre de D ios, Puerto Maldonado: GESUREMAD. Giugale, M., Fretes -Ciblis V. & Newman J. (2007) (Eds.). An opportunity for a different Per: prosperous, equitable, and governable. Washington D.C.: The International Bank for Reconstruction and Development, The World Bank. Green, G.M., Schweik, C.M. & Hanson, M. (2002). Radiometric Calibration of LANDSAT Multi -Spectral Scanner and Thematic Mapper Images: Guidelines for the Global Change Community. CIPEC Working Paper CWP 0203, Bloomington: Center for the Study of In stitutions, Population, and Environment Change (CIPEC), Indiana University.
122 Gutman, G., Janetos, A., Justice, C., Moran, E., Mustard, J., Rindfuss, R., Skole, D., & Turner II, B.L. (Eds.). (2004). Land change science: observing, monitoring, and understanding trajectories of change on the earths surface. New York: Kluwer. Hecht, S. & Cockburn, A. (1989). The fate of the forests Verso: Routledge. Hecht, S. (2005). Soybeans, development, and conservation on the Amazon Frontier. Development and Change 36 ( 2), 375404. Herndon, W.L. (1952). Exploration of the valley of the Amazon. New York: McGraw -Hill. IGBP IHDP, (1999). Land-use and land-cover change, implementation strategy. IGBP Report 48/IHDP Report 10. Prepared by Scientific Steering Committee and I nternational Project Office of LUCC. Stockholm and Bonn. IIAP CTAR, (2001). Madre de Dios, camino al desarrollo sostenible: Propuesta de zonificacin ecolgica econmica como base para el ordenamiento territorial. Iquitos, Per: Impresiones C.E.T.A. IIRSA SUR (2007). Estudio de Impacto Socio-Ambiental (EISA) del Corredor Vial Interocenico Sur, Per -Brasil, Etapas II y III Tramo 2: Urcos Inambari, Tramo 3: Inambari Iapari. Marzo, 2007. IIRSA -SUR, Walsh. Imbernon, J. (1999). A comparison of the driving forces behind deforestation in the Peruvian and the Brazilian Amazon. Ambio 28, 509513. INADE OEA, (1998). Plan de desarrollo de las comunidades fronterizas Peruanobrasileas, Iapari -Assis Brasil. CIDI/OEA. INADE PEMD, (1998). Proyecto Especial Ma dre de Dios. Memoria 1998. Puerto Maldonado, Lima, Per: CANO Asociados S.A.C. INADE PEMD, (2004). Plan operativo institucional para el ao 2004. Puerto Maldonado: PEMD. INEI, (1994). Censos nacionales de poblacin y vivienda de 1993: Directorio nacional de centros poblados Lima, Per: INEI. INEI, (1995). III Censo Nacional Agropecuario. Madre de Dios Perfil Agropecuario. Lima, Per : INEI. INEI, (2005). Instituto nacional de estadstica e informtica. Lima, Per: INEI INEI, (2007). Anlisis sociodemo grfico del departamento de Madre de Dios Lima, Per: INEI. INRENA (1996). Monitoreo de la Deforestacin en la Amazonia Peruana. Lima. Per.
123 INRENA (1999). Monitoreo de la Deforestacin y Zonificacin Ecolgica Econmica en la Amazonia de Per y Brasil Informe Ejecutivo Seminario y Taller Internacional. INRENA, Instituto Nacional de Recursos Naturales, 2022 Octubre. Lima, Per: ESERGRAF INRENA OEA. (1994). Zonificacin ecolgica-econmica Yaco-Iapari e IberiaIapari, Madre de Dios: Lima, Per. IN RENA OEA (1998). Zonificacin ecolgica-econmica Yaco-Iapari e IberiaIapari, Madre de Dios : INRENA, OEA. Imbernon, J. & Branthomme, A. (2001). Characterization of landscape patterns of deforestation in tropical rain forests. International Journal of Remote Sensing 22, 17531765. ITTC (2003) Desarrollo e Implementacin de Lineamientos de Control de la Extraccin Ilegal para un Manejo Forestal Sostenible en el Per. Informe Final de Estudio. SPDA CIMT. ITTC (XXXIV)/15. ITTO (2005). Status of Tropical Forest Management 2005. Yokohama, Japan: ITTO. ITTO (2005 a ). Evaluation of commercial stocks and strategy for the sustainable management of mahogany (Swietenia macrophylla) in Peru. Project Document. PD 251/03 Rev.3 ( F). Kaimowitz, D. (2002). Amazon Def orestation Revisited. Latin American Research Review 37, 221235. Kaimowitz D. & Angelsen, A. (1998). Economic models of tropical deforestation: A review Bogor, Indonesia: CIFOR. Kaimowitz, D. & Smith, J. (2001). Soybean technology and the loss of natural vegetation in Brazil and Bolivia. In A. Angelsen & D. Kaimowitz (Eds.), Agricultural technologies and tropical deforestation (pp. 195 211). New York: CABI Publishing Kaimowitz, D., Mendez, P., Puntodewo, A. & Vanclay, J. (2002). Spatial Regression Ana lysis of Deforestation in Santa Cruz, Bolivia. In C.H. Wood & R. Porro (Eds.), Deforestation and land use in the Amazon (pp. 4165). Gainesville: University of Florida Press. Kaliola, R., Maki, S. & Vuorinen, K. (2001). Road consequences in the Peruvian A mazon: Process, causes and consequences. Environmental Conservation 28(3), 199 214. Klepeis, P. & Vance, C. (2003). Neoliberal policy and deforestation in Southeastern Mexico: an assessment of the PROCAMPO program. Economic Geography 79, 221240. Lambin, E.F., Mertens, B., Sunderlin, W., & Ndoye, O. (2000). Impact of macroeconomic change in deforestation in South Cameroon: Integration of household survey and remotelysensed data. World Development 28(6), 983999.
124 Lambin, E.F., Turner II, B.L., Geist, H.J., Agbola, S.B., Angelsen, A., Bruce, J.W., Coomes, O.T., Dirzo, R., Fischer, G., Folke, C., George, P.S., Homewood, K., Imbernon, J., Leemans, R., Li, X., Moran, E.F., Mortimore, M., Ramakrishnan, P.S., Richards, J.F., Skanes, H., Steffen, W., Stone, G.D., Svedin, U., Veldkamp, T.A., Vogel, C., & Xu, J. (2001). The causes of landuse and land -cover change: moving beyond the myths. Global Environment Change 11, 261269. Lambin, E.F. & Geist, H.J. (Eds.) (2006). Land-use and land-cover change: local processe s to global impacts Heidelberg: Springer Verlag. Laney, R.M. (2002). Disaggregating induced intensification for land-change analysis: a case study from Madagascar. Annals of the Association of American Geographers 92(4), 702726. Laurance, W.F., Cochran e, M.A., Bergen, S., Fearnside, P.M., Delmnica, P., Barber, C., DAngelo, S. & Fernandes, T., (2001). The future of the Brazilian Amazon. Science 291, 438439. Lausch, A. & Herzog, F. (2002). Applicability of landscape metrics for the monitoring of lands cape change: issues of scale, resolution and interpretability. Ecological Indicators 2, 3 15. Liverman, D., Moran, E., Rindfuss, R. & Stern, P. (Eds.) (1998), People and pixels. linking remote sensing and social science Washington D.C.: National Academy Press. Liverman, D. & Vilas, S. (2006). Neoliberalism and the environment in Latin America. Annual Review of Environment & Resources 31(1), 327363. Mki, S., Kaliola, R. & Vuorinen, K. (2001). Road consequences in the Peruvian Amazon: Process, causes and consequences. Environmental Conservation 28 (3), 199214. Marengo, J.A., Nobre, C.A., Tomasella, J., Oyama, M.D., Oliveira, G.S., Oliveira, R., Camargo, H. Alves, L.M. & Brown, I.F. (2008). The Drought of Amazonia in 2005. Journal of Climate 21(3), 495516. Mather, A.S. (2006). Land use policies. In H.J. Geist (Eds.), Our Earths changing land: An encyclopedia of landuse and land-cover change. Vol. 2 (pp.375379).Westport, London: Greenwood Press. Mather, A. S. Needle C.L., & Fairbairn, J ( 1998) The human drivers of global land cover change: The case of forests. Hydrological Processes 12, 1983 1994. McCracken, S., Brondizio, E.S., Nelson, D., Moran, E.F., Siqueira, A. D., & Rodriguez -Pedraza, C. (1999). Remote sensing and GIS at farm property level: demography and deforestation in the Brazilian Amazon. Photogrammetric Engineering & Remote Sensing 65(11), 13111320. McCracken, S. Boucek, B. & Moran, E. (200 2). Deforestation trajectories in a frontier region of the Brazilian Amazon. In S. Walsh & K. Crews -Meyer (Eds.), Linking people, place and policy: A GIScience approach (pp. 215 234) London: Kluwer Academic Publishers.
125 McCracken, S.D., Siqueira, A.D., Mo ran, E. F. & Brondizio, E.S. (2002a). Land Use Patterns on an Agricultural Frontier in Brazil. In C.H. Wood & R. Porro (Eds.), Deforestation and Land Use in the Amazon (pp.162192). Gainesville: University of Florida Press. McGarigal, K., S. A. Cushman, M C. Neel, & Ene, E. (2002). FRAGSTATS: Spatial Pattern Analysis Program for Categorical Maps. Computer software program produced by the authors at the University of Massachusetts, Amherst. Available at the following web site: www.umass.edu/landeco/research/fragstats/fragstats.html Mendoza, E., Perz, S., Schmink, M. & Nepstad, D. (2007). Participatory stakeholder workshops to mitigate impacts of road paving in Southwestern Amaz on. Conservation and Society 5(3), 382407. Mertens, B. & Lambin, E.F. (1997). Spatial modeling of deforestation in southern Cameroon Applied Geography 17(2), 119. Mertens, B., Poccard -Chapuis, R., Piketty, M. G., Lacques, A. E. & Venturieri, A. (2002) Crossing spatial analyses and livestock economics to understand deforestation processes in the Brazilian Amazon: the case of Sao Felix do Xingu in South Para. Agricultural Economics 27, 269294. Mora, C. (1993). Expansin de la frontera agrcola en la a mazonia peruana. In J. Snchez & G. Pajares (Eds.), Hacia la recuperacin econmica y social Lima, Per: CEDEP, CIED, CIPA, DESCO, CIPCA, Arariwa. Moran, E.F. (1981). Developing the Amazon. Bloomington: Indiana University Press. Moran, E., Brondizio, E. Mausel, P & Wu, Y. (1994). Integrating Amazonian vegetation, land use and satellite data. BioScience 44(5), 329338. Moran, E. F., Brondizio, E.S., Tucker, J., Si lva Forsberg, M.C., Falesi, I. & McCracken, S.D. (2000). Strategies for Amazonian forest re storation: evidence of afforestation in five regions of the Brazilian Amazon. In A. Hall (Ed.), Amazonia at the Crossroads: The Challenge of Sustainable Development (pp. 129149). London: Institute of Latin American Studies. Moran, E.F. & Ostrom, E. (Eds .) (2005. Seeing the Forest and the Trees. Human-Environment Interactions in Forest Ecosystems. Cambridge, Massachusetts and London, England: The MIT Press Moreno, J. (2000). Aproximacin a una historia de Iapari (unpublished). Mueller, B., Alston, L., Libecap, G.D., & Schne ider, R. (1994). Land, property rights and privatization in Brazil. The Quarterly Review of Economics and Finance 34, 26180.
126 Nagendra, H., Munroe, D.K. & Southworth, J. (2004). From Pattern to Process: Landscape Fragmentation and t he Analysis of Land Use/Land Cover Change. Agriculture, Ecosystems and Environment, 101, 111115. Naughton Treves, L. (2004). Deforestation and carbon emissions at tropical frontiers: a case study from the Peruvian Amazon. World Development 32 (1), 173190. Nelson, G. & Hellerstein, D. (1997). Do Roads Cause Deforestation? Using Satellite Images in Econometric Analysis of Land Use. Staff paper 95 E488. American Journal of Agricultural Economics 79, 8088. Nelson, G.C. & Geoghegan, J. (2002). Deforestati on and land use change: Sparse Data Environments. Agricultural Economics, 27, 201216 Nepstad, D., Verissimo, A., Alencar, A., Nobre, C., Lima, E., Lefebre, P., Schlesinger, P., Potter, C., Moutinho, P., Mendoza, E., Cochrane, M. & Brooks, V. (1999). Larg e -scale impoverishment of Amazonian forests by logging and fire. Nature 398, 505508. Nepstad, D., Carvalho, A. C., Barros, A., Alencar, A., Capobianco, J.P., Bishop, J., Moutinho, P., Lefebre, P., Silva U.L. & Prins, E. (2001). Road paving, fire regime, feedbacks, and the future of Amazon forests. Forest Ecology and Management 154 395407. Pacheco, P. (2002). Deforestation and forest degradation in Lowland Bolivia. In C.H. Wood & R. Porro (Eds.), Land use and deforestation in the Amazon (pp.66 94). Gainesville: University of Florida Press. Pan, W.K.Y., Walsh, S.J., Bilsborrow, R.E., Frizzelle, B.G., Erlien, C.M. & Baquero, F. (2004). Farm -level models of spatial patterns of land use and land cover dynamics in the Ecuadorian Amazon. Agriculture, Ecosyste ms and Environment 101, 117134. Pearson, D. M. (2002). The application of local measures of spatial autocorrelation for describing pattern in north Australian landscapes. Journal of Environmental Management 64, 8595. Peralta, P. & Mather, P. (2000). A n analysis of deforestation patterns in the extractive reserves of Acre, Amazonia from satellite imagery: A landscape ecological approach. International Journal of Remote Sensing, 21, 25552570. Perez, A. (2007). Toward agricultural competitiveness and rural social equity. In M. Giugale, V. Fretes -Ciblis & J. Newman (Eds.), An opportunity for a different Per: prosperous, equitable, and governable. (pp.233249) Washington D.C.: The International Bank for Reconstruction and Development, The World Bank. Per z, S. (2001). From sustainable development to productive conservation: forest conservation options and agricultural incomes and assets in the Brazilian Amazon. Rural Sociology 66(1), 93 112.
127 Perz, S. (2002). The changing social contexts of deforestation in the Brazilian Amazon. Social Science Quarterly 83(1), 34 52. Perz, S. & Walker, S. (2002). Household life cycles and secondary forest cover among small farm colonists in the Amazon. World Development 30(6), 100972 Perz, S. & Skole, D.L. (2003). Social determinants of secondary forests in the Brazilian Amazon. Social Science Research 32 25 60. Perz, S. & Skole, D.L. (2003a). Secondary Forest Expansion in the Brazilian Amazon and the Refinement of Forest Transition Theory. Society and Natural Resources 16 277294. Pichon, F. (1997). Settler households and landuse strategies in the Amazon frontier: farm -level evidence from Ecuador. World Development 25(1), 67 91. Pichon, F. & Bilsborrow, R. (1999). Landuse systems, deforestation, and demographic fact ors in the humid tropics: farm level evidence from Ecuador. In R. Bilsborrow & D. Hogan (Eds.), Population and deforestation in the humid tropics (pp.175 92). Belgium: International Union for Scientific Study of Population (IUSSP). Ragin, C.C. (1987). The Comparative Method: Moving beyond Qualitative and Quantitative Strategies Berkeley: University of California Press. Reid, R.S., Tomich, T.P., Xu, J., Geist, H., DeFries, R., Liu, J., Alves, D., Agbola, B., Lambin, E.F., Chabbra, A., Veldkamp, T., Kok, K ., van Noordwijk, M., Thomas, D., Palm, C. & Verburg, P.H. (2006). Linking landuse/cover change science and policy: current lessons and future integration. In E.F. Lambin & H. Geist (Eds.), Land-use landcover change: local processes, global impacts (pp.159171). Berlin: Springer -Verlag. Rindfuss, R.R., Entwisle, B., Walsh, S.J., Mena, C.F., Erlien, C.M. & Gray, C.L. (2007). Frontier land use change: synthesis, challenges, and next steps. Annals of the Association of American Geographers 97(4), 739754. Rudel, T. & Horowitz, B. (1993). Tropical Deforestation: Small Farmers and Forest Clearing in the Ecuadorian Amazon. New York: Columbia University Press. Rudel, T. (2005). Tropical forests: regional paths of destruction and regeneration in the late twen tieth century. New York: Columbia University Press. Schelhas, J. & Greenberg, R. (1996) (Eds.), Forest Patches in Tropical Landscapes Washington, D.C.: Island Press. Schmink, M. & Wood, C.H. (1992). Contested frontiers in Amazonia. New York: Columbia Un iversity Press.
128 Seto, K., Kaufmann, R. & Woodcock, C. (2002). Monitoring land use change in the Pearl River Delta, China. In S. Walsh & K. Crews -Meyer (Eds.), Linking People, Place and Policy: A GIScience Approach (pp. 69 90). London: Kluwer Academic Publ ishers. Simmons, C. (2004). The political economy of land conflict in the Eastern Brazilian Amazon. Annals of the Association of American Geographers 94(1), 183 206. Skole, D.L.; Chomentowski, W.A.; Sallas, W.A. & Nobre, A.D. (1994). Physical and Human Dimensions of Deforestation in Amazonia. BioScience, 44, 314322. Smith, N.J.H. (1982). Rainforest corridors: the Transamazon colonization scheme Berkeley: University of California Press. Soares Filho, B., Alencar, A., Nepstad, D., Cerqueira, G., Vera D iaz, M., Rivero, S., Solorzano, L. & Voll, E. (2004). Simulating the response of land-cover changes to road paving and governance along a major Amazon Highway: The Santarem Cuiaba corridor. Global Change Biology 10, 745764. Southworth, J., Nagendra, H. T ucker C. (2002). Fragmentation of a Landscape: incorporating landscape metrics into satellite analyses of land -cover change. Landscape Research 27 (3) 253269. Southworth, J ., Munroe, D. & Nagendra, H. (2004) Land Cover Change and Landscape Fragmentatio n Comparing the Utility of Continuous and Discrete Analysis for Western Honduras Region Agriculture, Ecosystems and Environment 101, 185205. SPDA INRENA (2003). Compendio de Legislacin Forestal y de Fauna Silvestre. Lima: SPDA INRENA. Tahuamanu Munic ipalidad Provincial, (2001). Plan estratgico de desarrollo. Documento de trabajo, Tahuamanu, Madre de Dios, Per: Municipalidad Provincial Tahuamanu. Terborgh, J. (1999). Requiem for nature Washington, D.C.: Island Press. Trivelli, C., Shimzu, T., & Gl ave, M. (2003). Economic liberalization and evolution of rural agricultural sector in Per. L.A.S. Series No. 2. Institute of Developing Economies IDE JETRO. Turner II, B.L., Geoghagen, J.G. & Foster, D.R. (2004). Integrated land change science and tropic al deforestation in the Southern Yucatn: final frontiers Oxford: Clarendon Press/Oxford University Press. Turner II, B.L., Meyer, W.B. & Skole, D.L. (1994). Global land use/land cover change: toward an integrated study. Ambio 23 (1): 9195.
129 Turner, M. G. (1989). Landscape ecology: the effect of pattern on process. Ann. Rev. Ecol. Syst. 20, 171197. Turner, M. G., Gardner, R. H., Dale, V. H. & O'Neill, R. V. (1989). Predicting the spread of disturbance across heterogeneous landscapes. Oikos 55, 121129. Turner, M.G. & Gardner, R.H. (Eds.) (1990). Quantitative method in landscape ecology: the analysis and interpretation of landscape heterogeneity New York: Springer. Turner, M.G., Gardner, R.H. & ONeill, R.V. (2001) Landscape Ecology in Theory and Pra ctice: Pattern and Process. New York: Springer. Verissimo, A., Barreto, P., Tarifa, R. & Uhl, C. (1995). Extraction of a high -value natural resource in Amazonia: The case of mahogany. Forest Ecology and Management 72(1), 39 60. Via, A., Echavarria, F.R & Rundquist, D. C. (2004). Satellite Change Detection Analysis of Deforestation Rates and Patterns along the Colombia -Ecuador Border. Ambio (33)3, 118125. Walker, R. (1987). Land use transitions and deforestation in developing countries. Geographical Ana lysis 19(1), 1830. Walker, R. (1999). The structure of uncultivated wilderness: land use beyond the extensive margin. Journal of Regional Science 39 (2), 387410. Walker, R. (2003). Mapping process to pattern in the landscape change of the Amazonian fro ntier. Annals of the Association of American Geographers 93(2), 376398. Walker, R. (2004). Theorizing land use and land cover change: The case of tropical deforestation. International Regional Science Review 27, 247270. Walker, R. & Homma, A. (1996). L and use and land -cover dynamics in the Brazilian Amazon: an overview. Ecological Economics 18(1), 67 80. Walker, R., Wood, Ch., Skole, D. & Chomentowski, W. (2002). The impact of land titling on tropical forest resources. In S. Walsh & K. Crews Meyer (Eds .), Linking people, place and policy: A GIScience approach (pp. 131 153). London: Kluwer Academic Publishers. Walsh, S.; Evans, T.P.; Welsh, W.F. Entwisle, B. & Rindfuss, R.R. (1999). Scale dependent relationships between the population and environment in northeast Thailand. Photogrammetric Engineering and Remote Sensing, 65(1), 97 105. Walsh, S. & K. Crews-Meyer, K. (Eds.). (2002). Linking people, place and policy: A GIScience approach (pp.) London: Kluwer Academic Publishers. Walsh, S.J., Bilsborrow, R.E., McGregor, S.J., Frizzelle, B.G., Messina, J.P., Pan, W.K.Y., Crews -Meyer, K.A., Taff, G.N. & Baquero, F. (2003). Integration of longitudinal surveys,
130 remote sensing time -series, and spatial analyses: approaches for linking people and place. In J. Fo x, R.R. Rindfuss, S.J. Walsh & V. Mishra (Eds.), People and the Environment: Approaches for Linking Household and Community Surveys to Remote Sensing and GIS (pp. 91 130). Boston: Kluwer Academic Publishers. Weinhold, D. & Reis, E. (2008). Land use and tr ansportation costs in the Brazilian Amazon. Global Environmental Change 18(1), 5468. Wood, C. & Skole, D. (1998). Linking data to study deforestation in the Brazilian Amazon. In D. Liverman, E. Moran, R. Rindfuss & P. Stern (Eds.), People and pixels. li nking remote sensing and social science (pp. 70 93). Washington D.C.: National Academy Press. Wood, C. (2002). Land use and deforestation in the Amazon. In C.H. Wood & R. Porro (Eds.), Deforestation and land use in the Amazon (pp. 1 40). Gainesville: University of Florida Press. Wood, C. H. & Porro, R. (Eds.) (2002). Deforestation and land use in the Amazon. Gainesville: University of Florida Press. Yu, X. & Ng, C. (2006). An integrated evaluation of landscape change using remote sensing and landscape me trics: a case study of Panyu, Guangzhou. International Journal of Remote Sensing, 27(6) 10751092.
131 BIOGRAPHICAL SKETCH Andrea B Chavez was born in Lima, Peru. She obtained a Magister Artium in Political Science, Environmental Law, and Geography from the Karl Ruprechts -Universitt in Heidelberg, Germany. She then completed a Master of Arts at the University of Miami, Florida. She worked for several years in different non -profit (NGOs) and governmental institutions in Germany, Peru, and the United States i n the field of communitybased natural resource management, ecotourism, and climate change. Before starting her doctoral degree at the University of Florida, Andrea spent two years as a Visiting Scholar in Syracuse University, NY. While at Florida, Andrea has been concerned with the environmental impacts of natural and human induced disturbances in tropical regions. During her graduate education, Andrea accepted a public professional position within G eographic I nformation S ystem and remote sensing technolog ies. After her graduation as a PhD in Geography, s he plans to continue to work at the international level advocating the cooperation and collaboration among the natural and social sciences. Her future professional plans include the promotion of global change interpretation in the Latin American public and private sector.