THE EFFECTS OF WEALTH AND MA RKETS ON RUBBER TAPPER USE AND KNOWLEDGE OF FOREST RESOURCES IN ACRE, BRAZIL By RICHARD HOOD WALLACE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2004
Copyright 2004 by Richard Hood Wallace
For my family
iv ACKNOWLEDGMENTS I have completed this journey with the guidance, support, and encouragement of many people. At the University of Florid a, my dissertation committee members have been by my side every step along the way. I do not think it would be possible to find five faculty members more giving of their time and guidance. Dr. Karen Kainer in the School of Forest Resources and Conservation, a ment or and a friend, has always been a source for new ways of thinking about the role of forest products in rural livelihoods in Acre. Dr. Ronald Ward of the School of Food and Resource Economics worked with me patiently as I developed the economic models for the statistical analysis. The amount of learning that took place during those meetings was tremendous. Dr. Clyde Kiker, also of the School of Food and Resource Economics, he lped me to step back and think about what I really wanted to say. Over the past tw o years, I knew a trip to Dr. KikerÂ’s office would leave me with a better understanding of what I was trying to achieve. Dr. Russ Bernard of the Department of Anthropology ha s had a great effect on my development in graduate school and my dissert ation research. He has cha llenged me to think critically about my research questions and methods, a nd has shown me how exciting the field of anthropology can be. Finally, my chair, Dr. Marianne Schmink, has been a wonderful mentor and guide. She has taught me how mu ch one can achieve by pursuing small ideas and accepting challenges. As she told me once, Â“It doesnÂ’t hurt to dream.Â” Indeed, it does not. I thank her for her encouragement, patience, and support.
v I also would like to thank the many people and organizations in Acre, Brazil, that helped me in the implementation of my fiel dwork. I would like to thank the Research and Extension in Agroforestry Systems Gr oup (PESACRE) for providing institutional and logistical support during the study period. I also want to thank th e State Secretary for Forests and Extraction (SEFE), and in particul ar Arthur Leite, A ndrea Alexandre, MÃ¡rio Jorge da Silva Fadell and Alexandre Souza and Nuria Merched, who provided valuable insight into the changing extractive sector in Acre. At the Federal University of Acre, Francisco Kennedy de Souza was very helpful in developing my research instrument. The National Center for the Sustainable Development of Traditional Populations (CNPT)/Acre provided authorization to work in the Chico Mendes Reserve and I thank them for their continued support over the year s. In Xapuri, I would like to thank the Association of Inhabitants of the Chico Mendes Reserve in Xapuri (AMOREX), which supported my research in the three communitie s in the Chico Mendes Extractive Reserve, and the Rural WorkerÂ’s Union (STR), which assisted in arranging transport animals for my fieldwork. A number of individuals also assisted my research in diverse ways. Maria LÃºcia Rodrigues, a geography student at UFAC, work ed as my research assistant during the first round of fieldwork 2001. I would also lik e to thank Dione and Durival who assisted in fieldwork in 2002. Warm and special thanks go to the families in the communities of Rio Branco, Terra Alta, and SÃ£o JoÃ£o de Guarani in the Chico Mendes Extractive Reserve. I thank them for opening their homes and sharing their lives with me. Very special thanks to JoÃ£o Pereira da Silva, who befriended me on my first visit to Xapuri in 1995. Since then, he has contri buted to my research in so many ways, from assisting
vi with logistics of forest travel, serving as a guide, and helping with data collection. His friendship, and that of his family, I am privileged to have. Many colleagues also contributed ideas a nd support along the way. They include: Samantha Stone, with whom I passed many hour s discussing our research projects while sharing a house in Acre; Douglas Daly at the New York Botanical Garden and Evandro Ferreira of the National Institute for Amaz on Research (INPA) who provided assistance in the botanical iden tification of free-list items, and Ricardo Godoy, Beth Byron and Mike Gavin, all of whom helped me be tter understand data colle ction and valuation methods. I would also like to give speci al thanks to Ana Cristina Puentes for her friendship and encouragement as I neared my destination. Finally, a warm and heartfelt thank you to my family for encouraging me to pursue my dreams, and giving me the confidence to realize them. This degree is as much theirs as it is mine. Funding for this dissertation came from various sources and they deserve recognition. Pre-dissertati on research was funded by a Tropical Conservation and Development Program field grant and the Char les Wagley Fellowship, both of the Center for Latin American Studies at the University of Florida, and the Di ckinson Award of the School of Forest Resources and Conservation, also at the University of Florida. Fieldwork was funded by a National Scie nce Foundation Doctoral Dissertation Improvement Award and a grant from th e Conservation and Research Foundation.
vii TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...............................................................................................................x LIST OF FIGURES.........................................................................................................xiii ABSTRACT....................................................................................................................... xv CHAPTER 1 INTRODUCTION........................................................................................................1 The Emergence of Extractivism as a Sustainable Development Strategy....................5 Rubber Extraction in Acre: The History and Culture of the Seringueiro .....................8 Wealth, Markets and Extraction: Discussion and Hypotheses...................................15 Rubber Tapper Cultural Knowledge : Discussion and Hypotheses.............................21 Conclusion..................................................................................................................25 2 RESEARCH SITE AND METHODS........................................................................27 Acre: Geography, Climate and Ecology.....................................................................29 Acre: Political History, Economy and Conservation..................................................31 Large Scale Development Pl anning and Rural Conflicts...........................................36 The Acre State Â“Forest GovernmentÂ”.........................................................................39 Study Site: The Chico Mendes Extractive Reserve....................................................45 Research Design and Methods....................................................................................51 Use of Participatory Da ta Collection Tools.........................................................54 Operationalizing the Variables............................................................................56 The effects of wealth and markets on income..............................................56 The effects of wealth and ma rkets on cultural knowledge...........................62 Data Analysis.......................................................................................................65 Conclusion..................................................................................................................66 3 THE RUBBER TAPPER HOUSEHOLD ECONOMY.............................................68 Recent Household Economy Studies in the Chico Mendes Extractive Reserve........72 A Model of the Rubber Tapper Household Economy................................................77 The Rubber Tapper Household: A Demographic Portrait..........................................81
viii Age and Sex of Household Members..................................................................81 Education in the Forest........................................................................................84 Landholding Size and Land Use in the Forest.....................................................87 Social Organizations in the Forest.......................................................................89 The Diversity of Rubber Tapper Wealth....................................................................90 Household Wealth Holdings by Asset Category.................................................94 A Portrait of Changing Wealth Investments in the Reserve..............................100 Rubber Tapper Income in a Changing Forest Economy..........................................109 On-Farm Productive Income: Consump tion and Trade in the Forest...............113 Consumptive income in the forest: the important role of extraction..........114 Income from product trade.........................................................................125 Off-farm Labor Income.....................................................................................135 Conclusion................................................................................................................138 4 WEALTH, MARKETS, INCOME AND EXTRACTION......................................142 Statement of Hypotheses..........................................................................................144 Rubber Tapper Income and Wealth..........................................................................147 The Effects of Wealth and Market Integration on Household Income.....................153 Conclusion................................................................................................................173 5 RUBBER TAPPER CULTURAL KN OWLEDGE OF EXTRACTIVE RESOURCES...........................................................................................................177 Free-Listing of Non-Timber Forest Resources.........................................................182 The Pile Sort Test.....................................................................................................190 The Selection of Pile Sort Items........................................................................191 Results of the Pile Sort Test..............................................................................192 Initial Coding of Pile Sort Test Results.............................................................193 Pile Sort Test Results For All Respondents......................................................198 Consensus Analysis..................................................................................................209 Male and Female Cultural Knowledge.....................................................................210 Age and Cultural Knowledge...................................................................................217 Conclusion................................................................................................................224 6 WEALTH, MARKETS AND CULTURAL KNOWLEDGE..................................228 Statement of Hypotheses..........................................................................................229 Wealth, Market Integration and Cultural Consensus................................................233 Wealth, Market Integrati on and Cultural Knowledge..............................................236 Cultural Knowledge and Wealth.......................................................................237 Cultural Knowledge and Market Integration.....................................................242 Cultural knowledge and off-farm labor......................................................243 Cultural knowledge and product trade.......................................................246 Cultural knowledge and travel time to the city..........................................250 Conclusion................................................................................................................254
ix 7 CONCLUSION.........................................................................................................262 Summary of Key Study Findings.............................................................................264 Implications of Findings for Conservation...............................................................272 Implications for Policy Development.......................................................................273 Study Contribution to Research Methods.................................................................276 Study Contribution to Anthropology........................................................................278 APPENDIX A MULTIPLE REGRESSION MODELS...................................................................280 B RESULTS OF MULTIVARIATE REGR ESSION WITH ALL HOUSEHOLDS..282 C QUADRATIC ASSIGNMENT PROCEDURE RESULTS.....................................283 LIST OF REFERENCES.................................................................................................287 BIOGRAPHICAL SKETCH...........................................................................................303
x LIST OF TABLES Table page 2-1 The population of Acre, 1920 to 2000.....................................................................33 2-2 Federal and state conservation un its, agro-extractive settlements and indigenous lands in the state of Acre.......................................................................35 3-1 Age of household heads, number of household members, sex, and years of residence on landholding for 46 rubber tapper households......................................82 3-2 Previous residence of male and female heads of household....................................83 3-3 Literacy and formal schooling of male and female household heads......................86 3-4 Maximum level of education completed by study households................................87 3-5 Landholding size and land us e type in hectares in 2001..........................................88 3-6 Total and net wealth per capita for 46 rubber tapper households in the Chico Mendes Extractive Reserve in 2001.........................................................................93 3-7 Rubber tapper households ranked by net wealth pe r capita in 2001......................101 3-8 Percentage of mean wealth from diverse wealth holdings for households ranked by net wealth per capita..............................................................................102 3-9 Mean income of 46 rubber tapper ho useholds in the Chico Mendes Extractive Reserve for 12-month period in 2001....................................................................111 3-10 Income from extractive activities of 46 rubber tapper households in the Chico Mendes Extractive Reserve for 12-month period in 2001.....................................117 3-11 Number and value of game animal s and birds hunted by rubber tapper households over a 12-month period.......................................................................118 3-12 Extractive medicinal items used by households in the Chico Mendes Reserve.....120 3-13 Trading partners of r ubber tapper households in the Chico Mendes Reserve........126 3-14 Agricultural and extractive trade it ems and number of households involved in the trade of product............................................................................................128
xi 3-15 Principal trade income sources as a percent of trade income for 46 rubber tapper households...................................................................................................129 3-16 Trading patterns for rubber by 33 ho useholds in the Chico Mendes Extractive Reserve, 2001.........................................................................................................132 3-17 List of off-farm skilled and unskilled labor activities. (The number of households participating in act ivity is in parentheses)...........................................136 4-1 12-month income summary for rubber tapper households ranked by net wealth per capita.....................................................................................................148 4-2 12-month income summary for rubber tapper households ranked by net wealth per capita.....................................................................................................149 4-3 Multivariate regres sion model variables................................................................153 4-4 Rubber tapper households ranked by level of household integration into off-farm labor markets...........................................................................................154 4-5 Multivariate regression models relati ng wealth and market integration to measures of household income and income from extraction.................................156 4-6 Correlations between measures of ma rket integration and three proxies for deforestation...........................................................................................................172 5-1 Extractive resource free-list items a nd frequency of occurrence for 45 rubber tapper households...................................................................................................185 5-2 Free-list items categorized by plant family and genus..........................................187 5-3 Free list results with common name ite ms combined and listed by frequency......190 5-4 Frequency of higher-level codes fo r 24 pile sort item s. (Number of respondents of 118 total in parentheses)................................................................197 5-5 Aggregate proximity matrix generated from pile sort data of 118 respondents.....199 5-6 Consensus analysis eigenva lues for all 118 respondents.......................................210 5-7 Categorization of 118 respondents by sex and age................................................211 6-1 Households and indivi dual respondents grouped by sex and age, and categorized by net wealth per capita......................................................................232 6-2 Households and individual res pondents, grouped by sex and age, and categorized by percent of household produc tive income from off-farm labor......232
xii 6-3 Households and individual res pondents, grouped by sex and age, and categorized by percent of productive in come from barter and trade of agriculture and extractive products........................................................................232 6-4 Households and individual res pondents, grouped by sex and age, and categorized by travel time from household to city of Xapuri................................233 6-5 Gamma statistic measuring the associat ion between level of wealth and market integration and level of cultural knowledge for all respondents and sub-groups of rubber tappers.....................................................................................................235 B-1 Multivariate regression model findings for all 46 study households in Chico Mendes Extractive Reserve....................................................................................282 C-1 Quadratic Assignment Procedure (QAP) correlations for wealth for all respondents and respondents s ub-divided by sex and age.....................................283 C-2 Quadratic Assignment Procedure (QAP) correlations for integration into offfarm labor for all respondents and re spondents sub-divided by sex and age.........284 C-3 Quadratic Assignment Procedure (QAP ) correlations for integration into product markets all respondents and res pondents sub-divided by sex and age......285 C-4 Quadratic Assignment Procedure (QAP) correlations for travel time to the city of Xapuri for all respondents and res pondents sub-divided by sex and age..........286
xiii LIST OF FIGURES Figure page 2-1 Map of Acre, Brazil including defi nition of the Chico Mendes Extractive Reserve.....................................................................................................................30 2-2 Maps of study households in three research communites........................................49 3-1 Model of Rubber Tapper Household Economy.......................................................79 3-2 Wealth holdings for house holds by wealth rank group..........................................103 3-3 Productive wealth holdings of rubbe r tapper households by wealth rank category..................................................................................................................105 3-4 Animal wealth holdings of r ubber tapper households by wealth rank category..................................................................................................................105 3-5 Percent of rubber tapper on-farm in come earned from different production activities.................................................................................................................114 3-6 Percent of on-farm household consumption income contributed by different on-farm production activities..................................................................115 3-7 Percent of household on-farm production income from trade...............................125 3-8 Location and percent of rubber production traded by rubber tapper households..............................................................................................................132 3-9 Comparison of trade patterns for rubber (kilos) for 28 households in the Chico Mendes Extractive Reserve in 1996 and 2001............................................134 4-1 Percent of rubber tapper househol d productive income from on and off-farm activities by wealth rank category...........................................................150 4-2 Value of extraction from consump tion and trade by wealth rank group for rubber tapper households.......................................................................................151 4-3 Percentage contributi on of extractive activities to rubber tapper household extractive income by wealth group rank................................................................152
xiv 4-4 Simulation of the effects of a change in net wealth per capita on net productive income..................................................................................................159 4-5 Simulation of the effects of integration into off-farm labor markets on income from extractive activities...........................................................................161 4-6 Simulation of the effects of integr ation into product markets on income from extractive activities........................................................................................162 4-7 Simulation of the effects of wealth and integration into off-farm labor markets on percent of household productiv e income earned from extractive activities.................................................................................................................164 5-1 MDS scatterplot for pile-s ort test of 118 respondents............................................201 5-2 MDS scatterplot for pile-sort test of 118 respondents with items grouped............204 5-3 PROFIT Analysis for all 118 res pondents testing medicinal and eaten attributes.................................................................................................................206 5-4 Hierarchical cluster anal ysis for all 118 respondents.............................................207 5-5 MDS and PROFIT analysis for 73 male respondents with medicinal and eat attributes...........................................................................................................213 5-6 MDS and PROFIT analysis 45 female respondents with me dicinal and eat attributes.................................................................................................................213 5-7 Hierarchical cluster analysis for 45 females respondents......................................214 5-8 Hierarchical cluster analyses for 73 male respondents..........................................215 5-9 MDS and PROFIT analysis for 36 respondents in the youngest age group (age 1) with medicina l and eat attributes................................................................218 5-10 MDS and PROFIT analysis for 56 respondents in the young adult age group (age 2) with medici nal and eat attributes.....................................................218 5-11 MDS and PROFIT analysis for 26 respondents in the adult age group (age 3) with medicina l and eat attributes................................................................219 5-12 Hierarchical cluster analysis for 36 respondents in youngest age group(age 1)....220 5-13 Hierarchical cluster analysis for 56 respondents in adult group (age 2)................221 5-14 Hierarchical cluster analysis for 26 respondents in adult group (age 3)................222
xv Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE EFFECTS OF WEALTH AND MA RKETS ON RUBBER TAPPER USE AND KNOWLEDGE OF FOREST RESOURCES IN ACRE, BRAZIL By Richard Hood Wallace December 2004 Chair: Marianne Schmink Major Department: Anthropology This dissertation examines the changing liv elihoods of rubber ta ppers in the Chico Mendes Extractive Reserve in the Western Brazilian Amazon. Employing both theory and methods from the fields of microeconom ics and cognitive anthropology, this study explores how growing household wealth and integration into markets are transforming extractive activities in the reserve, and how these factors may be reshaping cultural knowledge of forest resources. It then cons iders the implications of these findings for conservation and development in the reserve, and for the extractiv e reserve concept. Multivariate regression analysis found that wealth had a strong positive effect on household income and a weak negative effect on percent of income from extraction. Households of varying wealth earn similar in comes from forest extraction, arguing that families still extract forest resources even as they accumulate and diversify wealth holdings. Yet, wealth was pos itively correlated to cattle wealth and pasture area, indicating that wealth ier households invest greater wea lth in production activities that
xvi involve cutting the forest. Integration into off-farm la bor markets and product markets had moderately negative effects on both inco me from extraction and percent of income from extraction. Households earning the gr eatest share of inco me off-farm, those carrying out skilled wage and salaried labor, showed the mo st dramatic decline in the percent of income from extraction. Ther e was no evidence that earnings were being invested in environmentally destructive land use, suggesting that the highest levels of integration into labor markets may have environmental benefits. Cognitive tests revealed that rubber tappe rs maintain a high degree of shared knowledge on the domain of non-timber forest resources. This was true of study participants sub-divided by se x and age, and dividing sub-groups of individuals by level of wealth and market integration. However, moderately strong positive associations were found between integration into product ma rkets and level of cultural knowledge for youth, and integration into off-farm labor markets and knowledge of young adults. Subtle trends in knowledge variation for rubber tapper youth suggest that youth from households at lower levels of market integra tion may be more likely to enter development projects promoting extraction.
1 CHAPTER 1 INTRODUCTION This dissertation is about the changing lives of people living in the Amazon rainforest in Acre, Brazil. At its hear t, it is about the changing economy of the seringueiros , rubber tapper families, living in the Chico Mendes Extractive Reserve. It explores the changing economy in the reserv e, and examines economic factors that may be reshaping how rubber tappers use, and how th ey think about, the forest. It focuses in particular on the changing role of extr activism in the rubber tapper household economy. This study really began during my previous research with rubber tapper families in the Chico Mendes Extractive Reserve in 1996 and 1997. Extractive reserves, originally conceived by rubber tappers, were designed to meet both the social and economic needs of rubber tapper households, providing families long-term use rights to the forest resources on which their livelihoods depe nded (Allegretti 1990). The Chico Mendes Reserve had only recently been established in 1990, the result of a violent struggle for land rights and social justice pitting r ubber tappers against ranchers that culminated in the death of Francisco Â“ChicoÂ” Mendes, a rubber tapper and union lead er for which the reserve is named (Schmink 1992). At that time I was conducting a study on th e market system for traditional extractive products, rubber and Brazil nuts, examining how this system functioned, and the implications of trade patt erns for diversifying production, particularly for extractive products. My study was a response to the growing concern over deforestation in the Amazon, and a search fo r sustainable development strategies. One alternative attracting increased attention was the potential for non-timber forest products,
2 such as fruits and nuts, and artisan crafts , to raise forest household incomes in ways fitting with traditional livelihoods without cutting the forest. One of the things I learned during that period was how important extractivism was, both economically, and culturally, to forest hou seholds. At one of the first community meetings I attended to present my research project and ask households in the community if I might implement my study there, befo re it began, a women stood up in the stifling heat of the day and rallied everyone to sing the Hino de Seringueiro , the Rubber TappersÂ’ Anthem. In a short time, everyone was cl apping and singing, and smiles were breaking out as they sang . . . vamos dar valor ao seringueiro, letÂ’s give value to the rubber tapper. At that point, I began to lear n that extractivism seemed to be as much about who you are, as what is being extracted. Yet, as I hiked through the seringal , or rubber forests, visiting different colocaÃ§Ãµes , or rubber tapper landholdings, I also saw how different house holds were financially and economically, even as they held common cu ltural ties that 100 y ears of history had woven. Yes, households came together for community workdays, Sunday church services, fÃ»tbol , or soccer matches, and during co mmunity meetings they debated the benefits of ongoing projects and priorities for th e future, but when they went back to their landholdings, they resumed, to varying degree s, different economic lives. Some families no longer tapped rubber, some households had fa mily members with sa laried positions or who worked as carpenters, and a few households had a number of head of cattle. Some households maintained traditional home struct ures of palm slats and thatch, while other homes were made of sawn timber planks a nd wooden shingles. A few had small motors for making farinha , or manioc flour, while others us ed woven sifting baskets. One
3 household had a solar panel and battery to gene rate electricity to il luminate the house at night. Households were diversifying on-farm produc tion, working off-farm, and investing in their landholdi ngs in different ways. Looking back, I had a very simplistic unde rstanding of life in the Chico Mendes Reserve when I first ventured there. But then, at the time, the focus of academics, researchers, and policy makers writing about the reserve was on the success that rubber tappers had achieved through the establishmen t of the Chico Mendes Extractive Reserve, and subsequently debating its merits as a conservation and development strategy, rather than understanding the changes that were emer ging in the forest and the implications of these changes for conservation and development. This dissertation adds a small piece to our understanding of the changing lives and livelihoods of rubber tappers in the Chico Mendes Extractive Reserve. Drawing from the fields of microeconomics and cognitive anthr opology, I respond to tw o related sets of questions to examine how a changing econom y is transforming rubber tapper households. First, what are the effects of increasing household wealth and integration into markets on rubber tapper household income, and, more specifically, income from extractive resources? And second, what are the effects of increasing wealth a nd integration into markets on rubber tapper cultural knowl edge of extractiv e resources? In the following chapters, I argue that ec onomic forces, in the form of growing wealth, and integration into markets, are bringing changes to rubber tapper production activities as well as cu ltural knowledge of forest resource s. Not surprisingly, increasing household wealth positively affects rubbe r tapper household productive income, but wealthier households also hold more cattle and have converted more la nd to pasture. The
4 effects of wealth on extractiv e activities are less clear. Across wealth categories, households still rely heavily on extraction; however, rising wealth leads to a lower percent of household productive income ear ned from extraction. Concomitantly, increasing integration into both labor and product markets negativ ely affects household productive income from extraction, and increasi ng integration into labor markets leads to a fall in the percent of productiv e income earned from extraction. The effects of increasing wealth and in tegration into markets on rubber tapper cultural knowledge of extractive resources are less clear but th ey suggest that youth hold different ideas compared to older generations. Employing cognitive anthropological methods, rubber tappers across generations and sex showed a very high degree of similarity in their cultural unde rstanding of extractiv e resources. This similarity carried through when categorizing individuals by di fferent levels of household wealth and market integration. However, a closer examin ation of the data reveals that as households integrate into markets, and have better market access, they are thinking differently about the two extractive items traditionally associ ated with rubber tapper incomeÂ—rubber and Brazil nuts. It is a subtle change, but one with implicati ons. For rubber tapper youth, in particular, this suggests that future househol d heads hold different id eas about the role of these items in household production. Together, these findings argue that market forces are bringing change to rubbe r tapper livelihoods in the Chico Mendes Reserve, with implications for both conservation and development in the reserve. In this chapter, I introduce the reader to the theoretical view s that inform the research questions and the historical-cultural context in which this research was undertaken. This demands an understanding of the emergence of the extraction of non-
5 timber forest products as a conservation a nd development strategy for tropical forest lands, and the theoretical debate and empiri cal research which guide the study research questions. Into this discussion, I insert the rubber tappers, provi ding the historicalcultural context in which the research took pl ace to give the reader an understanding of who rubber tappers are and why extractivism ha s played such a critical role in their livelihoods. In the following chapter, which presents the research site, design and methods, I discuss the current policy environment of the Ac re State Â“forest governmentÂ” which has committed itself to a guiding vision of sustainable development for the state, a vision that actively promotes the sustainable exploitation of non-timbe r forest resources. The Emergence of Extractivism as a Sustainable Development Strategy The sheer scale as well as accelerating p ace of deforestation in the Brazilian Amazon during the 1970s and 1980s brought internat ional attention to the loss of tropical forestlands taking place in developing c ountries. In 1975, only 0.6 % of the Brazilian Amazon had been deforested. By 1988 approxi mately 12.0 % of the rainforest was gone, a twenty-fold increase in ju st 13 years (Mahar 1989). A lthough an alarming claim that the burning of the Amazon rainforest was dest roying the Â“lungs of the worldÂ” (Moran 1996: 157) was Â“scientifically invalidÂ” (Wood 2002: 2), the quickening rate of deforestation led policy makers, scientists and development pr actitioners to rethink largescale development projects in the tropics, in particular col onization schemes and livestock development, and search for so-cal led sustainable development strategies for the Amazon region. These strategies woul d be ecologically benign, economically sound (i.e., profitable to inhabitant s) and meet the social repr oductive needs of traditional peoples.
6 The extraction of non-timber forest resour ces was one development alternative that appeared to meet the stri ct criteria of sustainabl e development (Allegretti 1990, Schwartzman 1989). The harvest and sale of forest fruits, fibers, nuts and oils, among other forest resources seemed well suited to the objectives of the sustainable development paradigm. Traditional peoples would collect forest resources, with which they were familiar and probably already used, and sell them, in their natural form or as processed derivative products. The forest would rema in standing, diversity intact, and forest populations would earn higher incomes through a livelihood conforming to their social needs. Buoying early enthusiasm for this strategy was a study c onducted in rainforest near Iquitos, Peru, which estimated the annual net market va lue of non-timber tree resources (fruits and latex only) of one hect are of land at $422, and the net present value of these resources at $6,330, over twice the present value of cattle production (Peters Gentry and Mendelsohn 1989). The author s argued that, Â“Without question, the sustainable exploitation of nonwood forest resources repres ents the most immediate and profitable method for integrating the use and c onservation of Amazonian forestsÂ” (Peters, Gentry and Mendelsohn 1989: 656). Although this study has been widely cri ticized for unrealistic assumptions, it stimulated considerable debate regarding how forests might be bot h used and conserved.1 The numerous, diverse and increasingly spec ialized studies that have emerged on this topic over the past 15 years indicate that Pete rs and his colleagues had raised an important 1 A number of scholars have criticized the Peters, Gentry and Mendelsohn (1989) article for various reasons. Nugent (1991) argues that they make a number of unrealistic and simplifying assumptions about the stability of various economic factors, such as cl aiming rates of inflation are evenly distributed over markets and all products will be fully exploited. Godoy, Lubowski and Markandya (1993) argue that if value was based on product flow, (i.e., actual quan tities taken from the forest) rather than the entire inventory, the value would be 4.0% of their calculated value and equal the timber estimate for the same area. See also Crook and Clapp (1998).
7 issue. Authors have argued that the extracti on and sale of non-timber forest resources can increase family income (Anderson and Ioris 1992, Clay 19 92, 1993, Gram et al., 2001, Gunatilake, Senaratne, and Abeygunawarde na 1993, Moreno-Black and Price 1993, Ziffer 1992) and help conserve tropical forests (Daly 1990, Freese 1998, Peters, Balick et al. 1989). Conversely, others have questioned th e long-term economic viability (Browder 1992b, Crook and Clap 1998, Dove 1993, Homm a 1993, 1992) and ecological soundness (Browder 1992a, Crook and Clap 1998, Pa ndit and Thapa 2003, Phillips 1993, Vasquez and Gentry 1989) of extractivism as a m eans to conserve tropical forests. In addition, a growing number of case st udies and overviews on non-timber forest resource extraction (Freese 1998, Neuma nn and Hirsch 2000, Shanley et al. 2003, Wollenberg and Ingles 1998) have helped synthe size this debate and led to more critical thinking about the role of research methods, resource valuation tec hniques, the role of markets, and Â“greenÂ” certification, among other is sues, that are at the core of linking nontimber forest product extraction to conser vation and development. Concurrently, warnings have been sounded regarding the damaging effects that entering markets may have on indigenous cultures (Gra y 1990), and the potential for this development alternative to undermine indigenous peoplesÂ’ quest for self-determination and land rights (Corry 1993). As studies have emerged, and this debate has matured, more recently, scholars have called for studies with a Â“situ ation-specific focusÂ” that recognize the different local values, uses and conservation of resources by diverse local populations (Arnold and Perez 2001: 445), and an examination of the multivariate factors that may influence extractive resource use (Barha m, Coomes and Takasaki 1999, Barham and Coomes 1996,
8 Shackleton 2001). It is within this site-specific context that I engage this debate, examining how economic forces are shaping use and knowledge of extractive resources among rubber tapper households in the Chic o Mendes Extractive Reserve. Thus, this dissertation begins with a jour ney back over 150 years into th e Amazon region, as it is in the forest, on the estradas de seringa , or rubber trails, that the history of the rubber tappers is told, where mÃ£e seringa, the mother of the rubbe r tree, watches over her children, the seringeira , or rubber trees, and where the rubber tapper has plied his trade for over a century. Rubber Extraction in Acre: Th e History and Culture of the Seringueiro With the perfection of the vulcanization process for rubbe r by Charles Goodyear in 1839, worldwide demand for products fashione d of this new weather resistant rubber grew immensely. Demand for natural rubber, dr iven first by the bicycl e craze of the late 1800s, then the invention of th e automobile in the early 1900 s, focused attention on the vast natural rubber stands of the Amazon region (Weinstein 1983). During the rubber boom, which lasted from the late 1870s to approximately 1912, the Brazilian Amazon region became the wo rldÂ’s supplier of rubber with production growing from approximately 7.9 metric tons in 1871 to 34.3 metric tons in 1910 (Barham and Coomes 1996). What is now the state of Acre was not the ini tial source for rubber, although it soon came to be. In the early year s of rubber extraction, rubber tree forests in the Eastern Amazon supplied the world market, but by the 1870s, capital and labor turned westward to the abundant, largely unexploite d, and river-accessible natural rubber stands of the upper Amazon (Rancy 1992). This regi on, including the stat e of Acre and the southern region of the state of Am azonas, was home to the rubber tree Hevea Brasiliensis, which produced the latex of choice of the interna tional market (Barham and
9 Coomes 1996). In addition, the Hevea tree, unlike c aucho , another type of rubber extracted from the costilloa tree ( C. elastica and C. ulei ) that required felling to extract latex, could be tapped on alte rnate days (allowing the tree to rest one day) through the tapping season, for years, with each carefully placed incision producing a small amount of latex. (Bakx 1996: 59-60, Barham and Coomes 1996: 37). Labor to work the rubber fields of Acre came largely from the northeast of Brazil, although caboclos , or mixed race forest dwellers a nd indigenous peoples living along the AmazonÂ’s tributaries participated in rubbe r extraction, both voluntary and coerced, to varying degrees (Barham and Coomers 1996, Weinstein 1983).2 Severe droughts in the Northeast of Brazil in 1877 and 1879 and fabled riches of Â“ ouro preto ,Â” or black gold, provided a supply of poor farmers to the re gion and helped resolve the severe labor shortage in the Upper Amazon at the onset of the rubber boom period. Estimates vary regarding the number of migrants to the Western Amazon, ranging from approximately 160,000 to 260,000 over the period 1872 to1900 (Santos 1980). Upon arrival in the seringal , the seringalista , or rubber baron, th at virtually owned the forests, generally assigned the rubber tapper to two or th ree rubber trails on which he would carry out his trade.3 Each rubber trail cove red considerable land area, approximately 100 hectares per trail, as rubbe r trees, although widely distributed in the upper Amazon region, have a very low density (Bakx 1986, Barham and Coomes 1996). 2 See Collier (1968), Furneaux (1969) and more recently Stanfield (1998) for detailed accounts of the brutal enslavement of Amerindians by rubber barons in the Putumayo region of Northern Peru. Their detailed portrayals of the atrocities leveled on indigenous peoples, including enslavement, brutal beatings and murder represent the darkest side of the rubber boom. 3 Weinstein (1980) states that in th e pre-boom period, some rubber tapper s were able to establish claims to land and a few hundred trees and thus avoid having to pay the rubber baron rent. In these cases, trade was often carried out with itinerant traders rather than a rubber baron.
10 New arrivals were provided toolsÂ—a small hand-axe to cut the rubber trees, small tin cups to collect latex from each tree, and a la rger bucket in which to pour latex from the cupsÂ—on credit. A number of scholars (Bakx 1986, Barham and Coomes 1996, Dean 1987 and Weinstein 1983) have similarly described th e rubber tapper workday during the rubber boom period. Tapping rubber consisted of risi ng early, often in the dark, and setting out on a rubber trail that wound past rubber trees in the area, making a small cut on each tree, and placing a small tin cup at the base of the slanted cut to catch the white milky latex as it dripped from the slash. Upon completing the looped trail, th e rubber tapper would return to his hut for lunch. Later that day, the rubber tapper would follow the same trail tapped in the morning to collect the still liquid latex from the tin cups. Trees were tapped only during the dry season, lasti ng eight months of the year from approximately March to October, as water would collect in the c up during the rainy season, contaminating the extracted latex (Bakx 1986, Barham and Coomes 1996).4 The late afternoon was spent smoking the co llected latex over a smoky fire, fueled by oily palm nuts, in a nearby hut called a Â“ defumador .Â” Latex was dripped onto a slowly turning long wooden pole where it woul d coagulate into a ball of black hard rubber. Rubber tappers woul d labor approximately 15 hours per day, six days per week in tapping, collecting and smoking activities (Bakx 1986: 60-64. See also Bakx 1988a: 146). At weekÂ’s end, the tapper would take his ball of rubber to the rubber baronÂ’s central trading post to trade for food and provi sions, where he might also socialize with other tappers (Burkhalter and Murphy 1989, Weinstein 1983: 16). 4 See especially Bakx (1986: 56-64) for a detailed description of the process of rubber extraction and processing, including different met hods employed for cutting the tree.
11 Exchange relations between rubber tappe r and rubber baron were governed by the aviamento or debt-merchandise system (B arham and Coomes 1996, Weinstein 1983).5 As rubber tappers had already run up signifi cant debt with the r ubber baron upon arrival in the forest, including the cost of passage, and tools and supplies, rubber tappers were obligated to sell their rubber to the rubbe r barons (Bakx 1986). Prices were low and rubber tappers were easily cheated regarding the weight of their pr oduct and the credit they received (Weinstein 1983, Wolf and Wolf 1936). Payment was almost never made in cash with rubber tappers ob liged to purchase high priced supplies with rubber earnings (Bakx 1986). Thus, Bakx (1986: 75-76) argue d that rubber baron control of rubber tappers went beyond the exchange relations hip, as armed guards were employed to ensure that rubber tappers di d not plant food crops and tappe d trees regularly. Maybe the most fitting account of this system comes fr om Euclides da Cunha (Cunha and Tocantins 1905: 59, my translation) who described the rubb er tapperÂ’s situation as one where Â“the man works to enslave himself.Â” The rubber Â“boomÂ” turned to Â“bustÂ” with the domestication of rubber in Malaysia and elsewhere in the early 1900s. Yet, de spite low world prices, many rubber tappers remained in the forest, now diversifying th eir production to incl ude food crops, hunting and fishing. Many turned to raising families in the forest, marrying locally or sending for their families in the east (Campbell 1996). The region underwent a second Â“boomÂ” when the Japanese took control of Asian rubber plan tations and the Allied powers turned to the natural rubber stands of the Amazon to s ecure a supply of rubber during World War II 5 Barham and Coomes (1996) provid e a compelling interpretation of the debt-merchandise system, arguing that the high risk and transactions costs inherent in rubber baron-rubber tapper exchange help explain this relationship.
12 (Dean 1987). A second wave of immigrants, again from the Northeast, who became known as the Â“ soldados da borracha ,Â” or rubber soldiers, arrive d in Acre during the war. However, with the end of the war, this br ief boom surrendered to a global market that once again demanded cheap latex. Despite the collapse in world rubbe r prices during the 1960s and 1970s, many rubber tappers and their families remained in the forest and continued extracting rubber, with the price of rubber often supported by government subsidies (Dean 1987). As the rubber barons abandoned their estate s, the Â“autonomousÂ” rubber tapper emerged, now with greater freedom to diversify household production, including the collecti on and sale of Brazil nuts (Alle gretti 1990). However, the aviamento system, in many respects, remained largel y in place, with itinerant traders and urban merchants assuming the role of the patrÃ£o . With the establishment of the Xapuri Ag ro-Extractive Cooperative (CAEX) in 1989 in the city of Xapuri, a small town in s outheastern Acre that sits approximately 1520 miles from the border of the Chico Mendes Reserve, rubber tapp ers essentially ended the unequal exchange that had dominated rubb er trade in the region for more than a century. And even as financial problems have plagued the cooperativeÂ’s existence, at times even unable to purchase rubber produc tion due to working capital problems, it has forced itinerant traders and me rchants that still buy rubber and Brazil nuts to raise the prices they pay for these goods, and lower th e price of basic supp lies they provide in exchange (Campbell 1996). But even as trade partners have change d and production has diversified, my own experience, as well as othersÂ’, suggests that rubber tappers still retain the accumulated knowledge of more than 100 years in the forest . These are found in the folk beliefs and
13 myths, hunting techniques, and preparation of medicinal remedies, many adapted from caboclo and indigenous populations (Esteves 1999, GalvÃ£o 1952, Smith 1996, Weigand 1996). Esteves (1999) argued that as new rubber tappers, or Â“ brabos, Â” arrived in the forest, the Â“ mansos ,Â” rubber tappers who had lived in the forest for long periods, would Â“socialize them in the rituals of hunting and fishing, in taki ng care in the wild, in the Â‘remediesÂ’ that cure sicknessÂ” (my translation, Esteves 1999: 49). The brabo was Â“ amansado, Â” his social identity reconstructed th rough new forms of social reproduction, the accumulation of cultural knowledge, and a new dependence on mae de seringueira , the rubber tree, Â“the woman w ho gives life,Â” as well as othe r rubber tappers for survival (Esteves 1999: 41). The Â“physical space of the forest, domesticated and incorporated as part of the identity of bei ng a rubber tapperÂ” (my translati on, Esteves n.d.: 5). In this sense, one might argue that the experience of the rubber tappers has led to the creation of Â“identity symbolsÂ” that have produced Â“i ntense collective cons ciousness and a high degree of internal solidarity,Â” contributing to Â“self-definition and an image of themselves as they have performed in the course of th eir historyÂ” and thus suggesting a Â“persistent cultural systemÂ” (Spi cer 1971: 798-799). Even today, rubber tappers in the reserve us e forest plants to make the medicinal remedies passed from generation to gene ration, and recount the myths that have humanized the forest and guide their use of natural resources. We igand (1996) detailed how rubber tappers have claime d to have seen or heard caboclinho , described as a young boy with his feet turned backward riding a de er (although also appearing in other forms), who will punish hunters if they take more from th e forest than they need. Indeed, I have overheard rubber tappers discussing panemas , a hex that prevents them from catching
14 game or fish (Wagley 1953). One rubber tapper was said to have not killed a deer after he referred to a deer as a Â“ burro, Â” or mule. Another rubber tapper asked my research assistant, as we were parting for another landholding, not to place the leg of deer meat that he was sending with us to a neighborÂ’s house on top of the mule, or it would give him a panema . Thus the rubber tappers have shared a hi story and culture intimately tied to the forest. The extraction of forest resources , both for consumption and the market, which has dominated their lives for decades, wa s governed by the accumulation of knowledge, and myths and beliefs, which have been passe d down over generations. But what are the driving forces that now guide socioeconomic practices in the rese rve? And how might these forces be reshaping rubber tapp er knowledge of extr active resources? Recent studies suggest that profound change s are taking place in the reserve, and that the intimate tie between rubber tappe rs and resources is being transformed. Campbell (1996: 115-116) found that many house holds in the reserve now view rubber tapping as Â“a necessary evil,Â” a Â“physically e xhaustingÂ” activity that has Â“lost its social and economic importance.Â” She argues that ther e is a Â“changing cultural identity within the seringal Â” and Â“changing livelihood st rategies [are] changing the reality and the image of who is a seringueiro .Â” Esteves (n.d.: 16-17), researching three gene rations of five rubber tapper families in the Chico Mendes Reserve, also finds that soci al changes are recasting the social identity of the rubber tapper. She argue d that the introduction of ag roforestry systems, producing fruits and nuts destined for the market rather than domestic use, is creating new types of workers in the forest. As a result, spaces w ithin the forest are now being recreated by the
15 younger generation. Despite new technologies fo r rubber tapping to increase quality and value, youth are not tapping rubber but sear ching for other econo mic alternatives. Esteves (n.d.) contends that these changes in land use are altering social relations of production as well as family social relations and bringing new genera tional conflicts to households. Fathers and grandfathers now feel threatened by their children and grandchildren, as they no longer dominate the reproductive activitie s on the landholdings. She argues that inheritance is no longer the transmission of Â“symbolic assets, social memory and identity, beliefs . . . knowledgeÂ” (Esteves n.d.: 6). Inheritance is now found in the landholding, reconstituted as an asset valued by new forms of market integration. This suggests that rubber tappers are Â“r econstituting their unique identities and territoriesÂ” in diverse ways (Schmink 2003: 236). Both Campbell (1996) and Esteves (1999, n.d.) argue that substantial changes are now taking place in the forest, both economically and culturall y. Their studies call for a deeper examination of factors that may be remaking social life in the forest. This dissertation responds to this demand by examining how economic factors may be transforming rubber tapper household inco me earning activities, and the cultural knowledge of the resources on which they have depended for generations. Wealth, Markets and Extraction : Discussion and Hypotheses Anthropologists have played an importan t role in understanding the rural household economy and, in particular, the role of w ealth and market activities in household livelihood activities. Earlier scholarly works (Foster 1967, Wo lf 1966) noted how social obligations could maintain equality among peasant households, while other studies (Bartlett 1982, Netting 1993) have found that di fferences in wealth endowments, such as land, can lead to diverse livelihood strategies . Enriching this di scussion were Marxist
16 studies that examined how access to produc tive wealth and power gives rise to inequalities (Plattner 1989). Complementing the work of anthropologist s, microeconomic studies have also played an important role in understandi ng peasant household economic behavior. Recently, studies have suggested that wealth is shaping peasant livelihood strategies and income generation. Demmer and Overman (2001: 106) argue that examining the effects of wealth on forest livelihoods is important, as wealth Â“makes it easier to achieve a better quality of life,Â” including meeting basic hum an needs, self-esteem, and freedom from servitude. Coomes (1996: 61) notes that we alth may be a better measure of peasant economic livelihood than income, as wealth refl ects both prior income earned as well as future earnings potential.6 Recent studies have shown that wealth can serve as a source for income and consumption smoothing (Deaton 1994, Morduc h 1995), while certain types of asset poverty (i.e., limited holdings in natural, human, or physical resources), can subvert diversification efforts by poor households (Reardon and Vosti 1995). In Tanzania, Dercon (1998) found that wealth ier households were able to invest in high return activities, such as ca ttle rearing, increasing the income disparity between rich and poor households. In their study in the Peruvian Amazon, Barham, Coomes and Takasaki. (1999; see also Takasaki, Barham and Coomes 2000) f ound that differences in household wealth holdings gave rise to considerable divers ity and specialization in economic livelihood. 6 Coomes (1996: 61) presents three main reasons why wealth may be a better measure of peasant economic livelihood than income: 1) peasants have few assets and their acquisition often is a memorable event; 2) assets are relatively apparent to the interviewer, and, as not ed above; 3) wealth reflects both prior income earned as well as future earnings potential.
17 Demmer and Overman (2001), working with Amerindians in Honduras found that wealthier households consumed more agri cultural products and i ndustrial products, and spent less time on forest activities and more time on agricultural activities. However, higher wealth also led to greater cash earnings from diverse sources, including agriculture, the forest, and non-forest/non-agricu ltural activities. Pattanayak and Sills (2001) suggest that extractive products can be used to smooth consumption across both poor and wealthy households, providing Â“natur al insurance,Â” and minimizing risks to forest families. In this study, I argue that rubber tapper hous eholds with greater wealth will be able to invest labor in riskier, higher return activ ities, leading to higher incomes. And despite prior findings (Pattanayak and Sills 2001) that non-timber forest products may smooth consumption for both poor and wealthy househol ds, I argue that households with greater wealth will rely less on the extraction of non-timber forest resources due to the opportunity cost of employing household labor in low return or subs istence activ ities. Therefore, I hypothesize that: H1: As household wealth increases, household income will rise. H2: As household wealth increases, the valu e of household income and consumption from non-timber forest res ource extraction will fall. H3: As household wealth increases, househol d consumption and income from nontimber forest resource extraction as a proportion of total income will fall. Just as studies have demonstrated that household wealth is shaping production decisions in the forest, a growing body of re search has argued that markets are also bringing changes to local communities. Indeed, an early anthropological study of extractive populations (Murphy and Steward 19 56) argued that it was work patterns driven by market linkages that brought culture change to communities, including
18 transforming social relations and bringing increased dependency on traders for goods. Burkhalter and Murphy (1989: 114) found that cas h transactions and paid labor activities among the MundurucÃº in Brazil have begun to Â“supersede traditional arrangementÂ” and Â“the transition from reciprocity to clientag e and then to active reliance on cash bespeaks a change in the quality of soci al relations, in which one pr ogressively eliminates . . . enduring relationships.Â” A number of recent studies have examined the marketÂ’s effects on tropical lands and people, responding to Godoy, Wilkie and FranksÂ’ (1997: 876) suggestion that the debate on the effects of markets on forest s has focused on Â“documenting deforestation rather than on presenting models or test able hypotheses about th e conditions under which markets may hurt or help conservation.Â” In a study carried out with four indigenous groups in Honduras and Bolivia, Godoy, Wilk ie and Franks (1997) found a positive association between sale of cash crops a nd deforestation and an opposite association between wage labor and deforestation, though ne ither of the variable s was significant. Pendleton and Howe (2002), employing agricultu ral prices as a measure of market integration, found that increasing prices led the Tsimane in Bolivia to greater old-growth forest clearance. Concomitantly, studies in the Peruvian Amazon have found that greater market-orientation for crops has led to a sl ightly higher deforest ation rate (Bedoya Garland 1995) and more intensive market-orien ted cultivation of crops (Henrich 1997). The effects of integration into market s on the extraction of non-timber forest products are also now undergoing empirical st udy. Two recent studies conducted with forest populations in India (Hedge et al. 1996, Hedge and Enters 2000) have argued that integration into labor markets negatively affects income from the collection of non-
19 timber forest products. In an innovative model, Robinson, Williams and Albers (2002) find that effects of markets on extraction are related to the heterogeneity of the population and spatial patterns of resour ce collection. They argue, fo r example, that increasing market access may lead to less pressure on periphery areas as some villagers leave extraction, but may intensify pre ssure on core, pristine areas. Maybe most notable among scholars examin ing the potential for extraction as a conservation and development strategy, is Al fred HommaÂ’s model of extractive resource use by forest households. Homma (1992) argues that the extraction of non-timber forest resources is economically unsustainable over the long-term due to eventual domestication and development of synthetic substitutes, and ev entual fall in prices for forest products. In HommaÂ’s model, forest dwel lers are rational actors. They operate under conditions of scarcity and will allocate scarce resources that maximize their financial return. Thus, socalled sustainable development projects th at are based on extractive activities are ultimately unsustainable. More specifically, in this model, production from th e domestication of forest products increases supply and leads to a fall in the equilibrium pr ice. This price reduction is particularly hard on the extracting population, as their cost per unit of production is much higher than that incurre d by cultivated production systems, such as plantation or agroforestry systems. Extrac tion involves collecting forest goods over large areas often requiring large amounts of labor. As a result, both land and labor employed in extractive production come under rising opport unity costs. Homma argues that as extractivists see their product from labor, their extractivist wage, drop with a falling price, they will leave low re turn extractive activities and undertake other higher return
20 activities, such as wage labor, crop producti on or agroforestry activities. With the expansion of the agricultural frontier into extractive pr oduction areas, labor opportunities become available and siphon off petty capitalist extractors into wage labor. Therefore, rubber tapping and other extractive activities pe rsist only when regional wages stay low. Eventually, however, the opportunity costs of extraction rise too high, and extractors will leave these activities. His model, however appealing and simply st ated, is challenged by past studies of economic behavior of peasant societies. Sah lins (1972) argued that th e domestic mode of production in Â“primitiveÂ” societies consistently under-produced while Chayanov (1967) posited that peasant households balanced the drudgery of mo re work with its expected benefits.7 Netting (1989: 229) rejected the idea that smallholders would under-produce, but also noted that they are not short-term ma ximizers, rather Â“serv[ ing] the survival and long-range interests of the domestic kin group.Â” Â“Agriculture as a household endeavor,Â” he maintained, Â“reflects a decision that the extra work and the quite modest returns to labor it brings are worth itÂ—you eat better, [and] live more s ecurely through the risks of unemployment, sickness and old age.Â” This dissertation empirically engages the above discussion by testing how integration into off-farm labor markets and product markets, and market access, measured by travel time to the nearest mark et, may affect forest household income and income from extractive activities. I argue that as households earn a greater share of income through off-farm labor, sell a greater percentage of production to the market, and gain greater access to markets through shorter travel times to the city, household income 7 Sahlins (1972: 87) defined the domestic mode of production as identified by a Â“small labor force differentiated by sex, simple technology, and finite production objectives.Â”
21 will rise. Further, as households integrat e into off-farm labor markets and product markets, and as they have better access to markets through shorte r travel times, the opportunity cost of allocating labor to extractive activities will fall. Thus I hypothesize that: H4: As the level of household market integrat ion increases, household income will rise; H5: As the level of household market integr ation increases, household income from non-timber forest resource ex traction will fall, and; H6: As the level of household market integr ation increases, household income from non-timber forest resource extraction as a proportion of total income will fall. Rubber Tapper Cultural Knowledge: Discussion and Hypotheses Cognitve anthropologists have argued th at culture can be viewed as shared cognitive representations, and that there can be considerable in tra-cultural variation among individuals within a culture (Bos ter 1986, Garro 1986, Romney et al. 1996, Weller 1987). An individualÂ’s understandi ng of a particular domain may vary Â“depending on the characteristics of the individuals, . . . the nature of the domain learned, . . . and the social situations in which l earning takes placeÂ” (Boster 1986: 429) and this can lead to differences in both the amount a nd types of information they know (Boster and Johnson 1989). Past studies have demonstrated intra-cu ltural variation in knowledge of folk biology (Ellen 1979, Gardner 1976, Hays 1976) . The field of folk biology, or ethnobotany, is a fertile area for explorati on by anthropologists (DÂ’Andrade 1995). By undertaking research on how peopl e think about, categorize a nd use plant resources, we can learn much about the relationship people ma intain with their environmentÂ—spiritual, economic and socio-cultural. The practical, applied use of this knowledge can be found in forest resource use and conservation pl anning that takes into account how families
22 think about and use the forest, the demarca tion of forest peoplesÂ’ land rights that recognize and protect local know ledge of plant resources (P lotkin and Fomalare 1992), and the possibility of identifying tropical fore st plant resources that can be tested for developing new drugs (Daly 1992, Plotkin 1993). The establishment of extractive reserves and the provision of land rights to rubber tappers based on the location of rubber and Brazil nut trees, is an example of how our understanding of lo cal knowledge and use of plant resources can inform conserva tion and development policy (Allegretti 1990). Ellen (1979: 346) suggested that intra-cult ural variation in folk biology among the Nuaulu in Indonesia is due, in part, to a Â“r elative absence of cogni tive sharing between different subgroups.Â” However, he also argue d that other factors must be considered to better understand knowledge variation, including: informant variabilit y, such as gender and age; contextual variability, such as social, economic, or ecological contexts; variability in the properties of the natural ent ity; and, methodological indeterminacy, including analytical errors by the researcher (Ellen 1979: 346-357). Ross (2002: 136) found that variation in knowledge of anim al-plant relationships among the Landacan Maya in Mexico was due to inter-generationa l differences in knowledge level, not due to Â“different stages in the development from novice to expert,Â” but because Â“members of the two generations subscribe to different underl ying models of the world.Â” Yet, ReyesGarcia et al. (2003) argued that the Tsim aneÂ’ Amerindians in Bolivia broadly share ethnobotanical knowledge. Might rubber tapp ers show intra-cultural variation, or consensus on their knowledge of forest resources? Kainer and Duryea (1992: 423), in thei r study in the Chico Mendes Extractive Reserve in Acre noted that men and women share much knowledge about plant use, but
23 argued that Â“women possess par ticular proficiency in plan t processing, especially of species used for food, spices, beverages and medicinesÂ” while Â“men are largely responsible for construction in the reserves, and are in ch arge of hewing fence postsÂ” (Kainer and Duryea 1992: 420). Further, they noted that Â“children are known for being the only ones interested in those fruits produced in small quantity that yield little pulpÂ” (Kainer and Duryea 1992: 422). Th eir findings hint at gender, as well as age-specific variation in cultural knowl edge of extractive resources among rubber tappers. As noted above, Esteves (n.d.), has argued th at the rubber tappersÂ’ struggle for land has brought new value to the forest, and land now serves a source of permanent tension among second generation Acreans seeking ne w forms of market integration. Her findings, combined with the work of Campbell (1996), as well as my own observations in the reserve, point toward a youth increasin gly interested in off-farm employment opportunities and a concomitant fall in rubbe r tapping, and suggest that there are generational shifts in how rubbe r tappers think about the fore st. Together, these studies suggest that there may be in tra-cultural variati on in knowledge of forest resources among sub-groups within the rubber tapper population. In this study, I employed cultural consensu s modeling to examine the intra-cultural variation in rubber tapper know ledge of the domain of non-t imber forest resources. Consensus modeling analyzes the Â“pattern of agreement or consensus among informants to make inferences about their differentia l competence in knowledge of the shared information pool constituting cultureÂ” (Rom ney, Weller and Batchelder 1986: 316). The domain of non-timber forest res ources is a particular ly relevant one to explore as the use and sale of extractive resources have played a critical role in su staining rubber tappers
24 over the past century. Indeed, who they are, i.e., Â“rubber tappers,Â” is defined by a native plant resource. Building on my hypotheses presented above, I employ consensus analysis to test for the effects of wealth and market integra tion on cultural knowledge of all respondents, household heads and the knowledge of subgr oups of household members (male-female, youth-young adults-adults). Usi ng the statistic gamma, I argue that greater levels of household wealth and increased integration into off-farm labor markets and product markets, as well as increased market access thro ugh shorter travel time to the city, will be associated with lower levels of consensu s analysis cultural knowledge scores. I hypothesize that: H7: As the level of household wealth incr eases, the cultural knowledge of non-timber forest resources of all respondents, th e household head, and knowledge sub-groups of individuals within the household (m ale-female, youth-young adults-adults) will fall, and; H8: As the level of household market integr ation increases, the cultural knowledge of non-timber forest resources of all responde nts, the household head, and knowledge of sub-groups of individuals within the household (male-female, youth-young adults-adults), will fall. Further, I employ quadratic assignment pr ocedure (Clark et al. 1998) to examine the similarity, or degree of fit, of cu ltural knowledge of different sub-groups of individuals (male-female, youthyoung adults-adults), categorized by levels of wealth and market integration. My hypotheses are: H9: The knowledge of sub-groups of indivi duals (male-female, youth-young adultsadults) in wealthier households will be si gnificantly different from the knowledge of sub-groups of individuals in poorer households, and; H10: The knowledge of sub-groups of individua ls in households (male-female, youthyoung adults-adults) with a higher level of market integration will be significantly different from the knowledge of sub-groups of individuals in households with a lower level of market integration.
25 Conclusion In this chapter I have laid out the principa l research questions that this dissertation will address. This study responds to the call for a greater understanding of factors that are shaping forest peoplesÂ’ us e and cultural understanding of natural resources. In the state of Acre, the creation of the Chico Me ndes Extractive Reserve has provided rubber tappers with long-term use rights to extrac tive resources. Development specialists and academics have hailed the creation of the reserve as sound policy for conservation and development in the tropics, ye t little is known about factors th at may be transforming the rubber tapper economy and cultura l understanding of resource s. Working with rubber tapper households in the reserv e, this dissertation contribut es to our understanding of how economic forces may be shaping livelihoods in the reserve by examin ing the effects of increasing household wealth and integration into market s on rubber tapper use and knowledge of extractive resources. I argue that these factors are transforming both resource use and knowledge, and these change s have implications for conservation and development in the reserve. In the following chapter, I present the rese arch siteÂ—introducing the reader to the state of Acre and the Chico Mendes Extrac tive ReserveÂ—and the research design and methods employed to carry out the research. This will include a discussion of the Acre state policy environment that is guided by a vision of sustainable development and placing considerable resources to ward developing the extracti ve sector. Chapters 3 and 4 will respond to hypotheses H1 through H6 above, describing and analyzing the diverse ways that rubber tappers hold wealth both on and off-farm, and the diverse income producing activities they undertake. Include d is a particular fo cus on the role of extraction in consumption and trade activities. Chapters 5 and 6 respond to hypotheses
26 H7 through H10, examining the effects of w ealth and markets on rubber tapper cultural knowledge of extractive resources. Chapter 5 also includes a deta iled discussion of the results of the free-listing and pile-sorting ac tivities, analyzing the different species of plants elicited by households as well as how they Â“think aboutÂ” these items. Finally, Chapter 7 will synthesize the discussion, su mmarizing the key findings and considering the implications of the study results for c onservation and development in the reserve.
27 CHAPTER 2 RESEARCH SITE AND METHODS I remember the first time I hiked to the Ch ico Mendes Extractive Reserve. It was during the rainy season in 1996, five years before the research for this dissertation would take place. I was in Xapuri, a small, qui et town in Southeast Acre. I had made arrangements to hike with a few rubber ta ppers from the community of Rio Branco, one of three communities where I would conduct my research, and also included in this study. We met in the morning, crossing the Xapur i River by rowboat, hiking up the tall riverbank into SibÃ©ria , a growing, poor neighborhood that sp rawls to the riverÂ’s edge, and started our hike up the Estrada Petropolis , or Petropolis Road. We would be hiking on this long and rolling, unkempt feeder-road, fi lled with cracks and crevicesÂ—only a large tractor could pass in the dry seasonÂ—for about two hours, then take a path through the forest for another three hours to reach the community center. A light rain began to fall not long after we left Xapuri. Soon after the rain began my heavy hiking boots were covered with mud, the hard rubber notch ed soles locking in the wet soil. With each step I left behi nd slide marks showing where I tried to push myself forward. Trying to scrape the mud fr om my boots became pointless. It felt like I had an extra 5 lbs of weight on each foot. I carefully followed the path of my rubber tapper colleagues, hoping to avoid the deep est mud holes. I ofte n fell behind, although my travel partners would wait patiently for me , stopping to allow me to catch my breath. I was exhausted when we arrived.
28 When I returned to Xapuri in 2000 to begi n pre-dissertation res earch for this study, I tracked down my good friend Paulo,8 the son of a rubber tapper now living in Xapuri, who would accompany me to the forest. I s uggested we get an early start the next morning. Even though it was the dry season, I was hoping to get an early start to reach the shade of the forest before the sun to too hot. To my surprise, he told me that if I wanted to pay, we could hire two motoqueiros , or motorcycle taxis, to take us to the community center in Rio Branco. The cost would be R$30.00 total, or about US$16.00 at the time. The Estrada Petropolis feeder road had been reformed, and another feeder road extended into the forest up to the community center in Rio Branco. Funds allocated by the Acre state government, known as the Â“forest government,Â” elected in 1998, had helped finance the road-building project, la rgely a dream when I had left in 1997. In 2001, the road was being extended further into the forest, approximately 40 kilometers, now reaching the community of SÃ£o JoÃ£o de Guar ani. Families still deeper in the forest talked about the road one day passing through their landholdings. I looked at Paulo with doubt, recalling the condition of the road when I left, but he assured me that we could get to the commun ity by motorcycle. That next day we hired two motorcycle taxis in SibÃ©ria . It took us a li ttle less than an hour to reach the community center. Things had changed in th e three years since I had been there, and these changes were bringing new opportuni ties and creating new hopes for forest communities. In this chapter, I introduce the state of Acre, and more specifically, the Chico Mendes Extractive Reserve and forest comm unities where this study was carried out. I 8 All of the names of individuals mentioned in this dissertation have been changed.
29 follow with a presentation of the research design and methods. I introduce the state of Acre by first situating the reader geographi cally, and also provide a brief background on the state, including the hi story, climate, ecology, economy, population, and natural resources of the region. I follow with a di scussion of how nati onal development policy for the Amazon region beginning in the 1960s th rough the 1980s led to land conflicts and deforestation in the state, and ultimately creat ed a political space for the emergence of the current Acre state Â“forest government.Â” The Acre state government, with strong ties with local grass-roots organizations , including the rubber tapper s, is pursuing a development agenda that aims to improve rural liveli hoods through both use and conservation of the stateÂ’s natural resources (State of Acre 1999). The rebuilding of the feeder road into the Chico Mendes Reserve is an example of a policy directed toward improving the livelihoods of forest populations by creating c onditions for better access to markets as well as public services, such as health and education. I follow this discussion by introducing the site of this studyÂ—three communities in the Chico Mendes Extractive ReserveÂ—and the research design and methods. Acre: Geography, Climate and Ecology The state of Acre (see Figure 2-1 on the following page) sits in the Southwest Brazilian Amazon and is one of six states th at make up the north region of Brazil. Other states include Amapa, Amazonas, ParÃ¡, RondÃ´ni a and RoraÃma. Acre forms part of the greater Amazon region, or AmazÃ´nia Legal , an area of over 5 million square kilometers including the entire north regi on of Brazil, and also extending to the northern region of the state of Mato Grosso and the eastern part of MaranhÃ£o. Th e state is cradled by Bolivia to the south and Peru to the southwest, while neighboring states of Amazonas and RondÃ´nia sit to the north and east. The state covers an area of 153,150 square kilometers,
30 approximately 3.9 % of the Brazilian Amazon region and 1.8 % of all Brazil (State of Acre 2000a). Acre Brazil Chico Mendes Extractive Reserve Acre Brazil Chico Mendes Extractive Reserve Brazil Chico Mendes Extractive Reserve Chico Mendes Extractive Reserve Source : Adapted from State of Acre (2000d). Figure 2-1. Map of Acre, Brazil including de finition of the Chico Mendes Extractive Reserve. The climate of Acre is tropical, with an average annual temperature of 24.5Â° C and a maximum temperature of appr oximately 32Â° C. However, friagens , or cold fronts, are not uncommon during the dry season and temperat ures can drop to as low as 10Â° C (State of Acre 2000a). The western region is much c ooler due to its extrat ropical location near the Andean foothills. Seasonal rainfall patte rns maintain considerable influence over transportation within the state. During th e rainy season from October or November to April or May, heavy rainfall facilitates rive r travel although it can temporarily cut off forest population access to urban areas due to the flooding of small forest streams, while dirt feeder roads are often impassable by moto r vehicles during this period. During the dry season from May through September, with wa ter levels at their lowest, river transport is more difficult, while land travel in the fore st permits the use of ve hicles on most feeder roads. Annual rainfall in Acre varies between 1.6 meters and 2.75 meters per year (State
31 of Acre 2000a). The far western region of th e state has a much shorter dry season than eastern Acre and thus receives greater rainfall. The rainforest in Acre is divided into two ecological types, dense tropical rainforest and open tropical rainforest, with open rainfo rest dominating 90.0 % of the state. Within these two ecological types, bamboo forests, palm forests and vine forests predominate (State of Acre 2000a). Climate differences between the western a nd eastern regions of the state have affected the distribution of tr ee species. For example, the Brazil nut tree ( Bertholletia excelsa H.B.K. ( Lecythidaceae ) does not occur in th e far western region, although the region has a much more abundant and diverse palm population than eastern Acre. This distribution has in turn influenced the economie s of these two regions. In eastern Acre, the collection of Brazil nuts often contribute s substantially to household incomes, while in western Acre, the palm fruit of buriti ( Maritius flexuosa ), among other palm products, plays an important role in trade. The biodiversity of AcreÂ’s flora and fauna has seen limited study. Silveira et al. (1997) referred to Acre as a Â“ buraco negro Â” or black hole, when it comes to the study of biological diversity. In 1999, approximately 316 species made up the Acre Flora Data Bank at the herbarium at the Zoobotanical Park at the Federal University of Acre. The State of Acre (2000a) notes that 1,319 species of vertebrate fauna ha ve been identified, including: birds (752 species ), fish (258 species), mammals (209 species), amphibians (119 species) and reptiles (94 species). Ac re is home to approximately 45.0 % of all birds, and 40.0 % of all mamma ls, that occur in Brazil. Acre: Political History, Economy and Conservation Although AcreÂ’s unique place in history extends back to the 1800s during the rubber boom, the region only became a Br azilian state in 1982. The land area now
32 encompassing the region officially became Br azilian territory at the turn of the 20th Century, with the signing of border treatie s with Bolivia and Peru. The Treaty of Petropolis of 1903 with Bolivia led to BrazilÂ’ s purchase of the re gion that now largely encompasses Acre. A great celebration was he ld in Acre in 2003 to commemorate the 100-year anniversary of the Acre revolution that led to the tr ansfer of the region to the Brazilian Republic. To the west, the BrazilPeru Treaty of 1909 officially resolved disputed land claims between Brazil and Per u, drawing the JuruÃ¡ region into the state. The capital of Acre is the city of Rio Br anco, located in the highly populated eastern region of the state. Politically, Acre is organi zed into five development regions: Lower Acre, Upper Acre, PurÃºs, TarÃ¡uca/Envira, and JuruÃ¡. Th ere are 22 municipalities, 10 of which were created in 1992, a sign of the increasing urbani zation of the state, as demonstrated in Table 2-1 on the following page. After a slight fall in population from 1920 to 1940, likely the result of out-migration at the end of the rubber boo m, the population has steadily increased, with the jump fr om 1940 to 1950 reflecting the influx of soldados da borracha during World War II. Table 2-1 also reveals the increasing urbanization and concomitant out-migration from rural areas th at began in the 1970s, a response to falling rubber prices as well as the state governmentÂ’ s promotion of large-scale cattle ranching (discussed in greater detail below), resulting in the flight of rural workers into cities. Although urban population growth slowed in the 1990s, a rising urban population combined with limited urban industrial growth has resulted in the ri se of urban periphery shantytowns and increasing unemployment (Bakx 1988b, Schwartzman 1992).
33 Table 2-1. The population of Acre, 1920 to 2000. 1920 1940 1950 1960 1970 1980 1991 2000 Population 92,370 79,768 114,755 158,852 215,299 301,303 417,718 557,226 Urban 59,307 132,169 258,520 370,018 Rural 155,992 169,134 159,198 187,208 Source: IBGE 2000, State of Acre 2000b. The growing urban population in Acre re flects the changing economy of the state. Although extractive products, notab ly rubber and Brazil nuts, s till maintain an important role in the regional economy, ot her sectors, such as public ad ministration, are emerging. State rubber production fell in the 1990s due to the fall of world rubber prices and decline of federal support. Between 1975 and 1996, rubber fell as a percentage of the value of production from agriculture, extractivism and cattle ranching, from 31.0 % to 6.0 % (State of Acre 2000b). This fall mirrors the de clining role of the agricultural sector in general. In 1985, agriculture (includi ng extractive activities) accounted for approximately 23.2 % of gross product in th e state, second to public administration, social security and defense, which accounted for 29.6 % of gross product (State of Acre 2000b). By 1997, agriculture fell to 3.9 % of state gross product while public administration rose to 47.9 %. The transf ormation of the Acre economy has been both dramatic and swift. As the state economy has changed, there ha s also been a transformation of the forest. Acre has undergone a relatively stea dy increase in gross deforestation over the past 25 years. In 1976, the earliest data availa ble on forest cover, only 1.6 % of the forest had been cut. By 1988, 5.8 % of the state was deforested, climbing to 8.7% in 1995. Gomes (2001) noted the invers e relationship between the fa ll of rubber production and increase in deforestation through the 1980s and 1990s. The regions that have undergone the greatest deforestation are Lower Acre and Upper Acre, found in the eastern half of the
34 state (Gomes 2001). The Upper Acre region straddles Highway BR-317 as it heads south and west toward Peru. This road open ed the region to exte nsive cattle ranching, particularly around the municipa lities of Xapuri, EpitaciolÃ¢ndi a and BrasilÃ©ia, leading to substantial deforestation where the Chico Me ndes Reserve was established. It was also the area that witnessed some of the wors t conflicts between fo rest populations and ranchers. In response to the steady increase in de forestation, the creat ion of conservation units and indigenous reserves, under both fede ral and state jurisdiction, has accelerated over the past decade. These areas now pr otect approximately 42.0 % of the state. Conservation units include extractive rese rves, state and national parks, and one ecological station. Together these area s claim approximately 4,282,842 hectares, or approximately 28.0 % of the state. (See Tabl e 2-2 on the following page). Federally administered indigenous lands, accounting fo r approximately 14.0 % of the state, also contribute to the growing conservatio n mosaic in the state of Acre. In addition to these protected areas, fede rally administered areas called AgroExtractive Settlements (PAE) have also been established for the direct sustainable use and management of natural resources. PAEs are often referred to as extractive reserves, and are very similar in use, although they differ in fundamental ways. PAEs are established and administered by the Nationa l Institute for Colonization and Agrarian Reform (INCRA) while extractive reserves are established and administered by the Brazilian Institute for the Environment and Renewable Natural Resources (IBAMA), the federal environmental agency. PAEs, befo re established, were already under the jurisdiction of INCRA, thus land use was alre ady under agrarian reform with titles held
35 by INCRA for distribution (Allegretti 1995). This facilitates th e provision of official title of use concession to resident families. Table 2-2. Federal and state conservation units, agro-extractive settlements and indigenous lands in the state of Acre. Area (ha) Creation Date Families Conservation Units Alto JuruÃ¡ Extractive Reserve 506,186 1/23/1990 695 Chico Mendes Extractive Reserve 976,570 3/12/1990 1,500 CazumbÃ¡-Iracema Extractive Reserve 750,794 9/19/2002 n.a. Alto TarauacÃ¡ Extractive Reserve 151,199 8/11/2000 n.a. Serra do Divisor National Park 843,012 6/16/1989 522 MacauÃ¡ National Forest 173,236 6/ 21/1988 18 SÃ£o Francisco National Forest 21,600 8/7/2001 n.a. Santa Rosa do Purus National Forest 230,258 8/7/2001 n.a. Rio Acre Ecological Station 77,500 6/2/1981 Antimari State Forest 66,168 2/7/1997 250 Mogno State Forest 143,897 5/10/2004 n.a. Rio Liberdade State Forest 126,360 5/10/2004 n.a Rio GregÃ³rio State Forest 216,062 5/10/2004 n.a Total Conservation Units 4,282,842 2,986 Agro-Extractive Settlements 8 Agro-extractive Units 193,449 Various since 1988 869 Indigenous Lands 28 Distinct Groups 2,167,146 Various n.a Total 6,643,437 3,855 Source : Adapted from Kainer et al. (2003), St ate of Acre (2000b and 2000c) and Amorim and Sousa (n.d.). N.A. indicates th e information is not available. Under the extractive reserve management regime, IBAMA faces a much more challenging task in providing titled use conces sions to resident populations. In fact, inhabitants of extractive rese rves do not have official ti tles for use concession. Unlike PAEs, areas appropriated for the creation of extractive re serves were not under the jurisdiction of either IBAMA or INCRA when appropriated, i.e., they were not under agrarian reform, and legislation for extractive reserves allows for regularization of land tenure after appropriation of areas (A llegretti 1995). As a result, land was appropriated for establishment of extractive reserves even as titles to these areas were often under dispute among private partie s (to be discussed below), including extractivists, who claimed areas but had no titles. Thus, despit e the appropriation of land and creation of
36 extractive reserves by IBAMA and the subseque nt establishment of utilization plans by reserve communities, resident families still do not have title for use concession. Without title, rubber tappers have difficulty proving where they live and the years they have worked in the forest. This complicates administrative procedures for securing retirement and other social benefits, as well as obtaining credit. The above discussion provides a general pi cture of Acre, a state increasingly urbanized, an economy with a growing admini strative sector and d eclining ag ricultural sector, a natural resource base undergoing defo restation, and in response, the more recent creation of protected areas. However, this brief sketch of AcreÂ’s changing socioeconomic and natural resource landscape doe s not expose the development planning and conflicts that have shaped the region, and ultimately given ri se to the current Acre state government, self-named the Â“forest government.Â” Thus, in the next section I want to go back a few decades and lay out the socio-politic al context that led to deforestation and the exodus of rural workers to the city. These policies fostered the gr owth of grassroots movements that led to the emer gence of a very different vision of forest use that now guides economic development in the state. Large Scale Development Planning and Rural Conflicts Just as the promotion of rubber extrac tion defined development in the Amazon beginning in the late 19th century through World War II, a different set of national development objectives for the Amazon region greatly transformed the state of Acre, beginning in the 1960s and through the 1980s. Large-scale remaking of the Amazon region began under Operation Amazonia in 1964 when the military di ctatorship assumed power in Brazil (Hecht and Cockburn 1989, Schmink and Wood 1992). Driven by geopolitical concerns and growi ng social problems in the south of Brazil, a product of the
37 industrialization of agricultu ral and land consolidation, Op eration AmazÃ´nia was a basinwide development program to colonize and in tegrate the Amazon regi on with the rest of Brazil. The programÂ’s key components incl uded extensive road building, agricultural colonization settlements, and generous fiscal incentives to attract agroindustry to the region (Mahar 1989).9 Up until the 1970s, Acre had remained relativel y isolated from the rest of Brazil. Lack of a paved road connecting Acre to the neighboring state of RondÃ´nia and onward to the south of Brazil made travel to the re gion difficult, and thus largely inaccessible. This began to change under Operation Amaz Ã´nia. In 1968, the construction of highway BR-364 from Cuiaba to Porto Velho, alt hough not yet paved, opened up the western Amazon frontier for the first time, and resulted in massive migration from the south of Brazil (Mahar 1989). By the 1970s , the highway finally reache d Acre, with the extension (again unpaved) from Porto Velho to Rio Br anco. In the 1980s, under the Polonoereste project, funded in part by the World Bank, highway BR-364 from Cuiaba-Porto Velho was paved. Further plans to pave the road into Acre, however, were only completed in the early 1990s. While new roads were providing easier acce ss to Acre, other national policies were shaping state development, resulting in a decline in the extractive sector and a concomitant leap in livestock development. The closing of the Amazon Credit Bank in the late 1960s ended credit to the rubber trade, leaving th e rubber barons who remained without credit little choice but to abandon or sell their forest lands (Costa Sobrinho 9 See Mahar (1989: 30-31) for a discussion of the mechanization of agriculture in the south of Brazil that led to land consolidation and the loss of small scal e farmers and release of agricultural laborers, greatly affecting migration patterns in Brazil.
38 1992). In the 1970s, the creation of the Ba nk of Amazonia (BASA) made available highly subsidized loans for livestock produc tion (Schwartzman 1992; see also Hecht and Cockburn 1989: 166-168). Concom itantly, the Acre state gove rnment openly encouraged private investment from outsi de interests, namely ranche rs from the south of Brazil (Costa Sobrinho 1992). Costa Sobrinho (1992) notes that while small and medium investors were interested in ranching, the largest investors purchased land for speculative purposes. Some estimates suggested that up to 80.0 % of the territory of Acre may have been sold between 1970 and 1975. (Schwart zman 1992: 56). During this period, falsification of land titles was common, and ti tles issued by multiple entities, such as Brazil, Bolivia, and the state of Acre, furt her confused land ownership (Schwartzman 1992). As outside investors pur chased land in Acre, land consolidation also ensued. By 1980, 22.0 % of AcreÂ’s land was in landholdings of 10,000 hectares or more. The stateÂ’s cattle herd grew from 72,166 head in 1970 to 356,446 head in 1987 (Schwartzman 1992). The result of these policies was conflicts between new investorsÂ—ranchers from the southÂ—and rubber tappers. Ranchers clai med title to the lands where rubber tappers had been living for generations, and employe d threats, destroyed crops, and murdered residents to Â“clearÂ” rubber tappers from their lands (Cos ta Sobrinho 1992: 148). Rubber tappers, assisted by rural union organizers and the Catholic Ch urch, organized to resist them (Costa Sobrinho 1992, Keck 1995). To confront ranchers , the rubber tappers successfully employed empates , or non-violent stand-offs , during which men, women and children would surround trees to keep laborers hired by ranchers from cutting forests on their landholdings (Mendes 1989). The build ing of alliances with international environmental organizations sympathetic to their cause, placed pressure on the
39 international lending communities to withhol d funding for paving of the highway BR-364 to Acre, and also forced the Brazilian government to resolve the growing crisis, becoming increasingly violent (Schmink 1992, Schwartzman 1992). This violence culminated in 1988, when Chico Mendes, a rubb er tapper and rural union leader, who had traveled to the United States to take the rubber tapper cause to the World Bank and U.S. Congress, was murdered by hired gunmen out side of his home in Xapuri. The Chico Mendes Extractive Reserve, a symbol of the ru bber tappersÂ’ fight for their livelihoods, now bears his name. The victory of the rubbe r tappers not only secured their place in the forest, but also created a political space for new voices in state deve lopment. It is from this space that the current Â“forest governmentÂ” emerged. The Acre State Â“Forest GovernmentÂ” The guiding vision of the Acre state Â“for est government,Â” elected in 1998, is one that embraces the idea of Â“sustainable devel opment.Â” The government views Â“the forest as the basis for a new economic modelÂ” for the region, arguing that the largely intact forests can be both used and conserved (State of Acre 1999: 39). With strong ties to the rubber tapper movement and other grassroot s organizations representing and working with forest populations, the government ha s implemented an ambitious social and economic development plan that includes the sustainable management of both timber and non-timber forest products. The government wo rks directly with forest communities to involve them in community management of productive activ ities, diversifying production, improving production and processi ng techniques, and adding value to products locally (Kainer et al . 2003, State of Acre 1999). Th ese policies aim not only to raise rural incomes, but also to fix rural extr actors and workers in th e forest, in order to reduce rural to urban migrati on induced by development policies of the past 30 years.
40 A central element of the Acre government Â’s sustainable development vision is Â“neoextractivism.Â” Neoextractivism, a concept development by Fernando do RÃªgo, an economist and former Secretary of Production fo r the Â“forest government,Â” contends that we cannot view the extraction carried out by forest families as purely an economic activity, strictly defined as the exploitation of naturally occurring resources (RÃªgo 1999). He asks that we re-conceptualize extractivism as neo extractivism, a livelihood determined principally by the forest dwe llerÂ’s Â“cultural universe,Â” structured by his relationship with nature, rather than ra tional economic behavior (Rego 1999: 65). Neoextractivism involves carryi ng out not only traditional extr active activities, such as rubber tapping and Brazil nut collection, but also small farming activit ies, agroforestry, enrichment plantings, and processing activit ies that bring greater product value to traditional populations. RÃªgoÂ’s concept came as a reply to HommaÂ’s (1992) microeconomic model of extraction discusse d in the previous ch apter, which views extractivism as a purely economic activity gove rned by supply and demand and rational economic behavior. The application of RÃªgoÂ’s vision of Â“neoex tractivismÂ” might be best seen in the state governmentÂ’s creation of an Executive Secretary of Forestry and Extractivism (SEFE) charged to productively and su stainably develop the forest sector.10 SEFE had departments tasked to help develop both the extractivist and timber sectors, and also to work directly with communities and help build capacity in production and marketing activities of a diverse set of forest products. For the extractive sector, SEFE initially 10 In 2003, SEFE was divided into two Executive Secr etaries: Executive Secret ary for Family Production (SEPROF), which directs public sector activities in the extractive sector, and the Executive Secretary for Forestry (SEF), which conducts the same in the timber sector.
41 focused on a set of specific extractive products, but subsequently turned to working on a regional basis to attend to the different need s of forest communities across the state. Courses and workshops have been implemented in the forest to improve extraction and processing technologies for both traditiona l and Â“newÂ” products. To diversify production, twenty non-timber forest product species have been prioritized for development of sustainable management plans, including palm fruits, vines, medicinal plants, and seeds. Collaboration with the Fede ral University of Acre (UFAC) as well as private sector partners is aiding in the collection of ecological information for the development of management plans (Kainer et al. 2003). In addition, SEFE has provided technical assistance in marketing to COOP ERACRE, a cooperative consisting of local producer groups and cooperative, and works w ith the private sector to improve plant infrastructure and management to facilita te the in-state processing of products. While diversification of the forest ec onomy is a fundamental part of state government development policy, rehabilitation of the rubber sector is also a core initiative. The Lei Chico Mendes , or Chico Mendes Law, also referred to locally as the Â“rubber subsidy,Â” implemented in January 1999, was one of the governmentÂ’s first major policy initiatives. The subsidy, or environm ental service payment, initially provided an additional payment to rubber tappers of R $0.40 per kilogram of rubber, subsequently increased to R$0.60 in January 2002. The law wa s enacted to increase the income of the extractive populations by supporting traditional extractive activities, but also aimed to achieve a number of other objectives. These included: securing extractive populations in the forest, thus helping to reduce rural to ur ban migration; promoting the organization of rubber tappers into associationsÂ—membershi p of an association or cooperative is
42 required to receive the payment (this also w ould block itinerant traders still operating in the forest from receiving the payment destin ed for rubber tappers); improving the quality of rubber; and, assisting in the legal doc umentation of the extracting population as a means to provide proof of rural service, a requirement for receiving retirement benefits (Kainer et al. 2003). A first evaluation of the effects of th e rubber subsidy indicates some positive results. In 1998, rubber production register ed by the state was 962,025 kilos, growing to 1,252,000 kilos in 1999, to 2,830,000 in 2000, and 2,980,000 in 2001. Concurrently, the number of families that have benefited fr om the subsidy has also grown, from 1,480 in 1999 to 4,354 in 2000 and 6,154 in 2001 (Kainer et al. 2003, SEFE n.d.). In addition to the state subsidy payment, producers have also been able to negotia te slightly higher prices with outside buyers as production is centralized a nd sold through the cooperatives (Kainer et al. 2003). Further, the government has been able to capture 70.0% of the cost of the rubber subsidy through st ate taxes (Kainer et al. 2003). In addition, the policy has resulted in the organization of numerous c ooperatives and associations as well as the legal documentation of rural workers (SEFE, n.d.). However, policy effects on stemming rural-urban migration are less clear. Alt hough the local press has reported that many families are returning to the forest, studies to document these assertions have not been undertaken (Kainer et al. 2003). Finally, just as the SEFE worked to improve the prices paid to extractivists, they also helped facilitate paym ent of the federal subsidy fo r processed rubber to private sector partners. In 2001, the federal subsidy for a particular quality of processed rubber, known as GranulÃ£o Escuro Brasileiro (GEB), was R$.90 per kilogram . To receive this
43 subsidy, processors had to provide extensiv e documentation regarding buying, selling and transport, and once documentation was submitted, receipt of the federal subsidy could be delayed as long as nine months. To facilita te this payment, SEFE negotiated with the National Council for Agriculture and Suppl y (CONAB), the federal agency that administers the subsidy, to have the paymen t made directly to SEFE based on production estimates. This reduced the time period that private processors wait for the federal subsidy, to approximately one month. Despite strong support from forest commun ities, and the building of partnerships with private enterprise to help stimul ate the extractive sector, in some cases, implementation of the state policy has been gua rded and slow (Kainer et al. 2003). In the case of timber management, the strong associat ion that forest communities maintain with extractivism and the conflicting voices that have emerged among them regarding this alternative, have led to a cautious implemen tation process (Kainer et al. 2003). In the case of the extraction of non-timber products, di fferent challenges have emerged. First, for many potential extractive products, there is little information av ailable on the ecology of forest species; thus management plans, required for commerci alization, have been slow to develop. Second, i ndustrial buyers often require a very large supply of a particular product, which is of a special qual ity and purity, but production in the forest is still very low. And finally, although government reports have demonstrated the initial success of the Chico Mendes Law, the effects of the subsidy may be le ss clear than initial reports reveal. Results from my researc h, presented in Chapter 5, do find that rubber tappers have received a higher price for rubbe r due to the additional payment. However, comparing production data for 28 households from 1996 with 2001, my research shows
44 that five households left r ubber production, and among those that still produce rubber, production has fallen dramatically. This suggest s that simply increas ing the price of an extractive product may not be enough to re -stimulate or maintain production among households that are already diversifying into alternative production activities. Although substantial challenges remain, th e forest government remains committed to engage both forest communities and the private sector in the sustainable use of the stateÂ’s forests. In addition to forest-based initiatives, major infrastructure projects, including the paving of the two major state highways (one destined to provide road access to the Pacific via Peru in the coming years) (Brown et al. 2002), as well as the improvement of feeder roads in rural areas , such as in the Chico Mendes Extractive Reserve, aim to help rural producers and businesses reach new markets for locally collected and processed products , in hopes of capitalizing on th e stateÂ’s natural resource base (Kainer et al. 2003, State of Acre 1999). The re-election of the governmentÂ’s char ismatic leader, Jorge Viana, a forest engineer, in 2002, has provided the governme nt a mandate to pursue its vision of sustainable development. Further, the state should be able to count on substantial support from the federal government. The Minister of the Environment, Marina da Silva, a former Senator from Acre has strong ties to the state government , and her background, herself the daughter of rubber tapper, has provi ded rural extractors with an ally at the highest level of environmental policymaki ng in Brazil. In addition, with the 1992 election of the WorkerÂ’s Party (PT) candidate Lula Ignacio da Silva to President of the Republic, the Acre state government, also representing the PT, has gained national respect for its commitment to sustainable de velopment of the regionÂ’s forests (Kainer et
45 al. 2003). Recent loans from the Inter-Ameri can Development, a substantial portion of which will be utilized to support sustainable forest management and extractive activities in the state, should also help improve th e economic livelihoods of AcreÂ’s forest population (Kainer et al. 2003). The above discussion has described the soci al and political cont ext in which this study was undertaken, with th e current state Â“forest govern mentÂ” implementing a vision of development built on local alliances with forest populat ions and focused on the use and conservation of local resources, and emer ging from a previous development agenda shaped largely by outside interests. I now present the site of this study, the Chico Mendes Extractive Reserve, and the communities th at participated in the research. This is followed by a presentation of the research design and methods. Study Site: The Chico Mendes Extractive Reserve The Chico Mendes Extractive Reserve encompasses approximately 970,560 hectares of rainforest , stretching across six municÃpios , or counties, including: Assis Brasil, BrasilÃ©ia, Sena Madeireira, Xapuri, Rio Branco and Capixaba. The reserve was officially established by Presidential Decree 98.987 on January 30, 1990 under President Jose Sarney. It is one of four extractive reserves in Acre. The Alto JuruÃ¡ Extractive Reserve, located in the southwest corner of Acre, was the first extractive reserve established. The Chico Mendes Rese rve, located in the southeas tern corner of the state, followed shortly thereafter. Rubber tappers initially propos ed the concept of extractiv e reserves in 1985, during the first meeting of the National Rubber Tappe rsÂ’ Council held in Br asÃlia, the capital of Brazil. Extractive reserves are defined by Allegretti (1990: 252, 258) as Â“public lands designated for the specific purpose of sustaina ble use of forest productsÂ” with Â“property
46 rights . . . designated according to traditional patterns of land use rather than imported models of occupation.Â” Upon their creation, ex tractive reserves were immediately hailed as a model for sustainable development in the region, with forests managed and conserved by rural inhabitant s, raising local incomes a nd protecting rural livelihoods (Daly 1990, Fearnside 1989).11 As noted in the previous chapter, extractive reserves are unique areas of conserva tion, as landholdings, or colocaÃ§Ãµes , are determined by resources rather than by conventional geographic shap es. Landholdings in th e reserve are quite large, in general, 300 hectares or larger. Thus they provide extractive populations sufficient area to undertake market-oriented ex traction for items such as Brazil nuts and rubber, which are found in low densities, and subsistence activiti es, affording large hunting grounds and area for fallow. The Chico Mendes Reserve is administered by IBAMA, with the Centro Nacional de Desevolvimento Sustentado dos PopulacÃµes Tradicionais (CNPT), an agency within IBAMA, maintaining direct co ntact with the associations that represent rubber tappers living in the reserve. This study was conduc ted in the municipality of Xapuri, where residents are represented by the Associat ion of Inhabitants of the Chico Mendes Extractive Reserve in Xapuri (AMOREX).12 With the official establishment of the Chico Mendes Reserve, resident households were gi ven 30-year use rights to forest resources, excluding minerals, on their landholdings. The utilization plan for the reserves was developed through three rubber ta pper associations located in Assis Brasil (AMOREAB), 11 For views that question the economic and ecological sustainability of the extractive reserve concept, see Homma (1992) and Browder (1992a). 12 AMOREX is now known as AMORPREX, the Association of Inhabitants and Producers of the Chico Mendes Extractive Reserve in Xapuri.
47 BrasilÃ©ia (AMOREB) and Xapuri (AMORE X) and approved by IBAMA on April 18, 1994 (CNPT 1995). It stipulated that the resident population w ould enforce land use activities in the reserve. Some of th e specific land-use re gulations include: Brazil nut trees and rubber trees cannot be cut down; timber extraction is permitted only for construction of infras tructure on the landholding; Mineral extraction is proh ibited; hunting and fishing is permitted only for household consumption; Up to 10% of the landholding may be used fo r agricultural activ ities; up to 50% of agricultural area (or 5 % of the landholding) can be used for pasture, and; Sale or trade of use rights to the la ndholding can only be carried out with community approval (CNPT 1995). In 1992, there were an estimated 1,834 familie s living in the reserve, spread over 46 seringais , or rubber tree forest areas, which fall in whole or part in the reserve (CNS 1992). This figure has declined in recent ye ars, and currently th ere are an estimated 1,500 families (State of Acre 2000c). Concurrently, the reserve population has fallen from an estimated 9,000 in 1994 to approximately 6,000 in 1998, indicating heavy outmigration over this period. The forest areas serve as the basis for the organization of a range of socio-economic activities in the reserve, including co mmunity organization, elementary education, religious worship a nd soccer tournaments. CAEX, the rubber tapper cooperative in Xapuri, maintains trading posts in a few forest areas, to facilitate the selling of rubber and Brazil nuts as well as provide basic supplies. This study was carried out in three forest areas in the re serve, all located in the municipality of Xapuri. They were selected as I had previously car ried out research in each of these areas in 1996 and 1997. Thus, I was familiar with households living in these areas and would also be able to compare rubber production in two periods to evaluate the effects of the state rubber subsidy. The thre e study sites were: Seringal
48 Floresta, the community of Rio Branco; Seri ngal Boa Vista, the community of SÃ£o JoÃ£o de Guarani; and, Seringal Filipina s, the community of Terra Alta. These study sites were initi ally identified for st udy in 1996 using a purposive sampling framework. They were identified and selected in coordination with AMOREX.13 These three areas differ in terms of distance to the nearest town, Xapuri, and principal means of transport between th e forest area and XapuriÂ—land or river. Thus, they represent a gradient of lesser to greater market involvement which will allow me to test the impact of wealth and markets on household income and extractive activities, key questions of this study. Figure 2-2 found on the following page, is adapted from drawings by community members a nd displays the location of community households, including those that participat ed in the study, within each community, The community of Rio Branco in Seringa l Floresta is approximately 20 miles by feeder road from the city of Xapuri. As noted in the introduction to this chapter, the rebuilding of the Petropolis Road and the construction of a feed er to the community center has recently improved community access to the city. However, many residents still use animal transportation or walk, to tr avel from their landholdings to the city. Few are able to pay the R$15.00 to hire a moto rcycle from the community center of Rio Branco to Xapuri. From the community center in Rio Branco, travel time is approximately 5 hours. The community of SÃ£o JoÃ£o de Guarani in Seringal Boa Vista sits to the north of the Community of Rio Branco, deeper into the reserve. The feeder road that now 13All research activities were conducted in coordi nation with AMOREX, including: the initial study proposal presentation to AMOREX officers, a follow-up presentation at a general meeting, the coordination of site visits, and the scheduling of presentation of the study objectives at community meetings in the forest.
49 SÃ£o JoÃ£o do Guarani Rio Branco Terra Alta Figure 2-2. Maps of study households in three research communites. Maps were adapted from drawings by community members. Triangles represent households in the community. Black triangles represent households that participated in the study.
50 connects the city of Xapuri with Rio Branco has been extended to the community center of SÃ£o JoÃ£o de Guarani and trucks and even cars are able to use the road during the dry season. However, as the feeder road has not been sufficiently compacted, an afternoon rain can make travel difficult even in the dr y season. Travel time by foot from the city of Xapuri to the community cente r of SÃ£o JoÃ£o de Guarani is approximately 8-10 hours, depending on the route taken. It is much longer during the rainy season. During the rainy season, using pack animals to carry produ cts to Xapuri, one-way travel time can last up to 16 hours, thus taking nearly two days of travel to arrive in the city. By vehicle, travel time is approximately two hours. Motorcycles charge approximately R$30.00 from Xapuri to SÃ£o JoÃ£o de Guarani. The co st of renting a small pick-up (including the driver) from Xapuri to SÃ£o JoÃ£o de Guar ani was R$100.00 in 2002. Again, few residents can afford to pay to hire motorcycle s or vehicles to make this trip. Travel time and method of transport to ar rive at Terra Alta are governed by the seasons. During the rainy season, all size canoes can navigate the enti re trip, with small canoes with motors completing the voyage in 4 hours. However, during the dry season Riozinho becomes nearly impossible to navigate due to very shallow waters and exposed fallen trees. The main effect of the seasons is in determining the type and size of canoe that can reach the community. Larger canoe s that can carry production to the city can only pass in the rainy season while smaller canoe s can pass nearly all year. The cost of hiring a canoe for a one-way trip from Xapur i to near Terra Alta is approximately R$80.00. If the hired canoe and pilot must stay for more than an evening, an additional fee is charged.
51 Although residents of the Community of Terra Alta prefer to travel by river, land travel is often used, as river travel is not often available a nd is more expensive, even if one owns a canoe, due to the need to purchas e gas and oil. Land travel from Xapuri to Terra Alta is approximately 7-8 hours walk ing, 4-6 hours by animal. Residents of the community are pressing the state and Xapuri municipal governments for the construction of a feeder road linking the city of Xapuri with Terra Alta, but currently there are no plans for its construction. Research Design and Methods A convenience sample was used to identify the 46 households that participated in the study. To identify families, a map of each community was drawn with the assistance from individuals from the communities. H ousehold were then identified employing the following criteria: 1) the landholding would fa ll in part or whole in the Chico Mendes Extractive Reserve; 2) the hous ehold would not operate as a marreteiro , or itinerant trader, trading products with other reserve fam ilies; 3) the household participated in some way in the community (i.e., they attend comm unity meetings, provide labor at community work days, their children study in the community school); and, 4) the household had to be present on their landholding for a minimum of one year.14 The term community is more a social designation than geographical one. A household that lived outside of a specific forest area (but still in part within the reserve) could be included in the study if household members participated in community activities. 14 One household whose landholding sits on the border of the reserve in the Â“area of protection,Â” what I would term a buffer zone for the re serve, was also included in the st udy. Households in the area of protection are required to follow the guidelines of the reserve utilization plan.
52 Households that met the above criteria were then ranked according to their comparative financial situation, based on th e knowledge of the individuals from the communities and my own experience working there. Their location within the community was also considered. Thus families were identified based both on their financial situation and their location, to achieve a variation in wealth and market access. In some cases, landholdings had multiple households with the r ubber trails divided among the separate family households and in a few cases multiple households from the landholding were included. If a household iden tified for inclusion in the study was not home when we arrived at the landholding (o r we knew from other community members that the household members were not there), when possible, an attempt was made to return to the household later during the same fi eld trip, or during anot her visit to the area at a later time. However, this was not al ways possible. In these cases, we made adjustments in the field, identifying other households for participation. In the community of Rio Branco, 45 househol ds participated in the community. Of these, five fell out of the reserve (although one in the ar ea of protection was includedÂ— see above footnote), two act as itinerant traders, and two live on a small ranch. Of the remaining 36 households, 17 were selected and interviewed, for a total of 18 households. In the community of SÃ£o JoÃ£o de Guarani, 26 households participated in the community. Of these, six had lived in the area for less than a year. Of the 20 remaining households, 13 were interviewed. In the community of Te rra Alta, 24 households participated in the community. Of these 24 households, 15 were interviewed. Thus, of a possible 81 households (including one household in the ar ea of protection), 46 were interviewed, or approximately 57.0 % of potential participating households. During pr evious research in
53 the reserve, conducted in 1996 and 1997, I interv iewed 39 households in these same three forest areas. Of these households, 29 were included in this sample. This study employed both microeconomic and anthropological methods to test the hypotheses stated in Chapter 1. It consisted of three principal stages. The first stage, carried out in May and June 2000 involved presentation of the study objectives to administrators of the CNPT, IBAMA, lead ers of AMOREX and rubber tappers in the Chico Mendes Reserve, and pre-testing of re search methods and tools. All methods and tools utilized were pre-tested with a few se lected reserve households in the participating communities. Methods tested included a st ructured questionnaire, cognitive research methods tools including pile-sorting and triad tests, and a semi-structured questionnaire, or open-ended interview. The second stage of the research, conducted September through December 2001, involved the first round of data collection with participating households. Prior to implementation an open presentation of resear ch objectives and tool s for data collection was made in each community during a regul arly scheduled community meeting. A research schedule was made to organize visi ts to participating households. The first round of interviews employed a structured questionnaire and cued recall to collect demographic data on individual household me mbers, including age, sex, education and health, and household wealth and income. In addition, free-lists (Fleischer and Harrington 1998) were elicited from each fam ily on the domain of non-timber forest resources. Forty-six households were inte rviewed during the fou r-month field visit. The third stage of the research was ca rried out in June through October 2002. During this final visit, all but one of the house holds were visited again. One family in the
54 community of Terra Alta had moved from the forest area to a small farm area closer to the city of Xapuri. During this household visit, three activities were carried out: first, we clarified responses to the structured intervie w conducted the previous field visit; second, we conducted pile sorts (Roos 1998) with hous ehold members, and; third, we conducted semi-structured interviews that permitted all household members to respond to the same set of open-ended questions. Use of Participatory Data Collection Tools A participatory approach was employed to collect information on individual household members, such as age, sex, e ducation and health, household wealth and income. As the interview woul d require the participation (and focused attention) of all household members (where possible) and woul d last anywhere from approximately 1-3 hours, rather than ask ques tions about household member s and household structures, assets and income directly from the questionnai re, I developed an extensive set of tools in the form of diagrams to engage participants. To collect information on household member s, I diagrammed a set of 2Â” by 5Â” cards portraying men, women, boys and girl s. A card representing each household member was placed on the floor in front of th e group. Each card was then selected (in some cases, children were offered to hold the card), and then I asked for information about that person. For the youngest individua ls, the female head of household often responded. To obtain data regarding household assets and income, I developed an extensive set of 6Â” by 9Â” cards with diagrams of wealth items, including equipment, structures and consumer items, and agricultural and extractive resources which families collect, harvest or breed. This list of items was develope d based on my prior experience in the reserve
55 and in collaboration with researchers at UF AC and rubber tappers in the reserve (cf. Gunatilleke, Senaratne and Abeygunawarden a 1993). For wealth items, household members were asked questions regarding qua ntity, when and how acquired, and for highticket equipment items, such as chainsaws, their costs. For trade and consumption activities, household members were asked to recall the quantity of production, extraction, trade or barter of products and with whom any transactions took place. The cards encouraged the participation of all household members, not just the household heads, with one member (including children) frequently correcting the response of another. This method also encouraged the participation of female household members. In conducting the interviews, I displayed th e cards and spoke directly w ith household members while my research assistant recorded responses a nd insured that all questions listed on the questionnaire were asked. However, not all data collected during the structured interview was obtained in this manner. For example, questions regarding in come from salary, wage labor, contracted work, and social benefits, such as retireme nt income, maternity salary, health-related benefits, among others, were asked directly rather than through the use of cards. Diagrams were not used in these cases for two ge neral reasons: first, so as not to restrict the potential response of the household, such as the case of wage labor, where a diagram of one activity representing the category in ge neral might restrict th e potential response of the household, or potentially leaving out a wage labor activity not previously encountered; and second, some items, such as social security benefits due to health problems or membership in an organizati on, seemed better posed through a direct question rather than a diagram.
56 Operationalizing the Variables To test the hypotheses, variables were cons tructed to measure ne t wealth per capita and market integration, as well as house hold income and income from extraction. Subsequently, the measures for net wealth per capita and market integration were categorized to rank individuals by level of net wealth per capit a and market integration to test for differences in cult ural knowledge of extractive re sources across these measures. Explanatory and dependent variables used in the multivariate regression analysis to test hypotheses H1-H6, along with the ordinal variab les used to calculate gamma to test hypotheses H7 and H8 and for employing quadra tic assignment procedure (QAP) to test hypotheses H9 and H10, are operationalized below. The effects of wealth and markets on income Hypotheses H1 to H6 test for the eff ects of wealth and markets on household productive income and income from extraction. I hypothesized that as wealth and market integration increase, household income would rise, but that income fr om extraction, and the percent of productive inco me from exraction would fall. Variables for the multivariate regression analyses to test the hypotheses were operationalized as follows. Explanatory variable: net wealth per capita. At the outset of the study, wealth was to be measured by land and non-land assets (Barham, Coomes and Takasaki 1999). Non-land assets would include wealth object s, such as household items, equipment, animals, and structures on the househol d landholding, such as the house, storage buildings, animal pens, fenced gardens, and fences. In addition, as some families own land and homes as well as other items in the city of Xapuri; these assets were to be included in the wealth measurement. For equipment and i nputs, through explicit questions regarding specifi c objects owned by the household (Demmer and Overman
57 2001), I would learn the type, quantity owne d and used, how acquired, and the year acquired for all items still in use.15 I would conduct focus grou ps with rubber tappers from the research communities (Bernard 1995) , to elicit the usef ul life of household assets and use straight-line de preciation to calculate their present net value (Demmer and Overman 2001). For animals, households were asked the number of animals they owned, both for transportation and production. For structures, households were asked what structures they maintained on their landholdi ng, when they were constructed, materials used, and costs of labor and materials. Despite pre-testing, collection of data to operationalize wealth in this manner ran into various problems. As an example, fo r equipment, although families were able to respond with the quantity of each object they owned, recall of when items were acquired, if more than five years in the past, was difficult. In addition, many items were acquired second hand, either purchased or donated. Other items were donated new through local development projects but rubber tappers were unsure of the year they were acquired. Thus use of straight -line depreciation encountered problems. In addition, straight-line de preciation (i.e., liquidation value) did not capture the productive value of these items to their ow ners. Therefore, rather than depreciate equipment, I used replacement value fo r valuing them (Byron 2003, R. Godoy, personal communication). This method over-estimates the liquidation value of rubber tapper wealth holdings, though I believe it is more fitting to rubber tapper productive activities in the forest. 15 Households were asked if the equipment they owned was still in use. For example, a household might have 10 machetes, but only three might be used. Theref ore, the number of machet es for which information was recorded was for thr ee machetes, not for 10.
58 Another difficulty emerged in obtaining a value for land.16 To obtain this information, I asked households how much they thought their landholding was worth should they decide to sell it. One household had purchased the landholding approximately one year prior, and gave th e actual price they paid. However, many immediately noted (not surp risingly) that they had no intention of selling their landholdings or moving. Thus, many specula ted on what they thought their landholding might be worth, or provided an es timate based on what price it would take them to move. This resulted in inflated valu es that in many cases were clearly beyond the market value of the landholding. Valuation of land became more complicated with the building of the feeder road into the forest, which has im proved access to urban areas. Due to these problems, I did not include a value for landholding.17 Wealth was divided into three categories: on-farm productive wealth, other on-farm wealth, and off-farm wealth. The total net wealth measurement was divide d by the number of household members to calculate the household net wealth per capit a. Wealth was measured as follows. On-farm productive wealth includes: Equipment. Equipment included the value of 43 items. Equipment was valued at replacement cost based on the average of pri ces collected at four sellers in the city of Xapuri. For a small number of high-ti cket items not found in Xapuri, such as chainsaws and rifles, equipment was valu ed at average purchase price by study households who purchased these goods in 2001. If stated purchase value was higher than this average purchase price, then the higher stated price was used; Animals. Animals (except for pigs) were va lued at the average farm gate price of each community. If this price was not available, then the animal was valued at 16 Reserve residents have use rights to resources, not land ownership, so when they purchase landholdings, they purchase the right to resources rather than the land itself. However, rubber tappers in the reserve consider themselves the owners of their landholdings. 17 Non-structural investments on land, such as the planting of perennials, also were not included in the wealth measure. This value would have been reflected in the householdÂ’s estimate of land value, which ultimately was not included.
59 overall average forest gate price. Pigs ar e sold by weight, not per unit. Thus, pigs were valued by first calculating the av erage weight of pigs consumed on all landholdings, multiplying this by the forest gate itinerant trader price to get an average pig weight, and multiplying this by number of pigs;18 Productive structures. Productive structures (not including reside nce) were valued at actual labor and material costs, or estimated based on like structure and materials. All fencing, regard less of material, was valued at the cost of building the stated meters of fencing in plac e with sawn wood posts and wire; Stocks of inputs. Seventeen input items we re used to calculate input value. Stock items were valued at replacement costs base d on average of prices collected at four sellers in the city of Xapuri, and; Stock of goods. Stocks of goods for eventual sale were valued at average farm-gate price for the community. Other on-farm wealth includes: Household structure. The family reside nce was valued the same as productive structures, and; Household consumer goods. Nine consum er goods were valued at replacement cost using the average of prices collected in the city of Xapuri. Off-farm wealth: Urban wealth. Urban wealth included land and structures, the value of seven consumer items, and animals held in the city of Xapuri. Land and structures (residences) were valued at stated value. If the owner did not know the value of land, or could not estimate the value, then the land or property was not included. The seven consumer items were valued at purchase price unless the price was not provided. If not provided, the item was valued at the average of prices collected at merchants in the city of Xapuri. Animals were valued at urban market price. Other rural land. Other rural land was valued at the stated value. Three households share other rural land with relatives but did not know and could not estimate the value of their share. They were not included. Bank Account. Five categorical replies were possible. The bank account was valued at the lowest valu e of the category indicated. 18 As pigs are generally slaughtered when they are full grown and fattened to produce a large amount of lard for cooking, this value is greater than the average selling price per pig across all landholdings.
60 Explanatory variable: market integration. Godoy, Brokaw and Wilkie (1998) operationalized market integration using th ree different measures: the share of rice harvest sold, share of cash income earned th rough wage labor, a nd share of income earned from the sale of forest goods. Godoy, Franks and Claudio (1998) measured market integration by rice sales, while Godoy, Wilkie and Franks (1997) used cash income from the sale of both crops and labor as proxy for mark et integration. Godoy (personal communication) also suggests other measures of market integration, including town-to-village distance, credit received, share of goods sold, and number of days worked for a wage. I used three different indicators of market integration: 1) pr oportion of household income from wage labor, measured by cash income received for wage labor over the past 12 months divided by productive income over the past 12 months; 2) proportion of productive income from cash and barter trade, measured by the tota l value of cash sales and barter transactions over the past 12 months divided by total household income over the past 12 months; and, 3) tr avel time from household to the city of Xapuri by principal means of transport during the season when travel time is shortest. Dependent variable: net household income. Net household income includes onfarm productive income, off-farm income, soci al income, and gift income minus hired labor costs earned over the previous 12 mont hs. Focus groups with rubber tappers were employed to elicit rubber tapper valuation of non-traded resources. Categories include the following items and measures. On-farm productive income includes:
61 Basic crops. Basic crops include rice, bean s, corn, and manioc flour. Consumption income is community average farm-gate price and trade income is actual value received or cash equivalent of barter good received; Animals. Animal income for production and work animals is valued as above; Extraction. Extraction items include rubber, Brazil nuts, fruits (7 items used for making fruit drinks), artisan items, medicinal items (23 items), honey, copaÃba oil , and game.19 Fruit, medicinal items, and game were valued by two focus groups with rubber tappers. Other items were valued as above; Plants. Plant items include six perennial crops and other planted items valued by the value received or cash equiva lent of barter good received; Processed goods. Processed goods include cheese, rapadura , a candy made from sugar cane juice, and goma , a thick liquid by-product from manioc processing. Cheese was valued by focus group. Consumption and trade income for rapadura and goma was valued as above. Off-farm productive income includes: Skilled and unskilled labor. Salary and wa ge income were valued as reported. Other income includes: Social income. Income from monthly re tirement, physical and mental disability benefits, and maternity benef its were valued as reported. Gift income. Gift income was valued as reported. Dependent variable: net productive income. Net productive income was valued as above minus social income. Dependent variable: income from extraction . Income and consumption from non-timber forest resource extraction was m easured by the value of barter and cash transactions and consumption of extraction, as defined above. Dependent variable: proportion of productive income from extraction . Proportion of income and consumption from extraction was measured by the value of 19 Fish were not included; pre-testing found that r ecall for this item was ill suited to the one-shot recall method. Medicinal items includes only forest-extracted medicinal items, not planted medicinal items. Fuelwood is also not incl uded in this measure.
62 barter and cash transactions and consumption of extractive activities, as defined above, as a proportion of net productive income. The effects of wealth and ma rkets on cultural knowledge Hypotheses H7-H10 test the e ffects of wealth and mark et integration on cultural knowledge of extractive resources. For H7 and H8, I hypothesized that as household wealth or market integration increases, cultu ral knowledge of indivi duals and sub-groups of individuals would fall, while for H9 and H10, I hypothesized that the cultural knowledge of individuals and sub-groups of i ndividuals from househol ds more integrated into markets would be different from the cu ltural knowledge scores of individuals and sub-groups of individuals from households with lower wealth and less integrated into markets. Data on the cultural knowledge of non-timber forest resources were collected through free-lists (Fleisher and Harrington 1998) and pile-sorts (Roos 1998). Free-lists on the domain of non-timber forest products were obtained from the household as a group during the second stage of the studyÂ—the first data colle ction visit. The household was asked, Â“Can you give a list of things from th e forest that you use, except for wood and animals?Â” The free-lists were elicited at the outset of the structured interview before any questions regarding production or extraction were asked, to avoid response bias by the previous mention of forest resources. From the initial free-lists, 24 items were selected for conducting pile-sorts. Pile-sorts were obtained from household members 13 years and older.20 The results of two pile sort res pondents were dropped before running the data: one was assisted by another family me mber, and a second continually observed the 20 In three cases, younger household members were pres ent and carried out the pile sort activity. In one case, the 12-year old son was assisted by his mother. In another case, a 12-year old daughter repeatedly watched her mother sort the cards and sorted her cards in nearly the exact same manner. In another case, a 10-year old grandson, who had completed the fourth grade, asked if he could carry out the activity. As other children his age were not given the opportunity to participate, his pile sorts were not included.
63 pile-sorting of another family member. Mo re details regarding the implementation of cognitive methods will be discussed when the results ar e presented in Chapters 5 and 6. Pile-sort cards included bot h a colored pencil drawing of the resource and the written name of the resource. For each partic ipant, cards were placed in front of the individual, one card at a time, with the name of each item read to the sorter as it was placed in on the floor or table. Cards were shuffled after each use. Individuals were asked to place the cards in piles however they wished. They were instructed that they could form as many groups as they preferre dÂ—they could have a lot of groups or very few groups. The only two limiting criteria stated were that all cards could not be placed in a single pile, and that all cards could not remain separate. They were also advised that if an item did not have a group, then it coul d be alone. Respondents were reassured that however they did the assignment was correctÂ—they could not do it wrong. On one occasion, an elderly adult male was unable to carry out the exercise. Four sets of pile-sort cards were used to work with household members simultaneously to avoid one member viewi ng the sorting of another member. For households with more than four individuals old enough to undertake the pile-sort activity, it was emphasized that the inclusion of thei r pile-sorting was important to the study, but that their work could only be included if they did not view the work of other household members. This meant that those waiting usuall y left the room or area where pile sorting was taking place. After each household member completed th e pile-sort exercise, the individual was asked why he or she had placed cards togeth er in the group, or the Â”nameÂ” of the group.
64 This information was noted as well as the items sorted together.21 Results of pile sorts were used to conduct consensus analysis on the domain of non-timber forest resources among the study individuals to identify the degr ee of agreement of each individual in the sample. Thus, for each individual, a score of estimated knowledge, or degree of cultural competence, was produced. Hypotheses H7-H10 test the e ffects of level of net weal th per capita and level of market integration on cultural knowledge. Thus , to test for the effects of wealth on cultural knowledge (H7 and H9) individuals were categorized into four wealth rank groups based on their household ne t wealth per capita. To te st the effects of level of market integration on cultural knowledge (H8 a nd H10) individuals we re categorized into four (or three, in the case of travel time) mark et integration rank groups. For hypotheses H7 and H8, level of cu ltural knowledge is measured by the individualÂ’s consensus analys is score. Individuals we re divided into sub-groups, including household heads, males, females, and three age categor ies (youth, young adults, and adults) to examine like individuals across levels of wealth and market integration. Individuals in each sub-group were then ranke d by their consensus analysis score and categorized into two groups, one with the highest knowledge scores (top 50 percentile), the other with the lowest knowledge scores (lowest 50 percentile). For hypotheses H9 and H10, individuals were categorized into sub-groups as noted above, and the results of the pile sort test we re used to conduct QAP analysis to test for 21 In recording the results of pile sorts, if the respondent could not indicate the purpose for grouping items together into piles, i.e., unable to explain the rationale for creating most or all of the sorted groups, this was noted in the field book. The results of these individual s were eventually included in the study after running consensus analysis and finding that a ll received a positive consensus score.
65 differences in cultural knowledge across rank le vels of net wealth per capita and market integration. Data Analysis Hypotheses H1 through H6 were tested em ploying multivariate regression analysis using the SASÂ® software package. Four models we re tested, using the four dependent variables noted above. Each model included the independent variables for net wealth per capita and market integration. For the market integration, proportion of income from offfarm labor, the variable was categorized into four rank categorical variables. Dummy variables were then used for market rank categ ories 2, 3 and 4, while category 1 served as the intercept base, representi ng no income from off-farm la bor. This was done, as the dependent variable in model 3, percent of productive income from extraction, must by definition go to 0 as percent of productive in come earned from off-farm labor goes to 1.0. Ranking and categorizing the variable eliminates this problem. This same variable in its dummy form was used in all four models for consistency. In addition to the explanatory variables, control variables were also incl uded. These are presented and operationalized in Chapter 6. Hypotheses H7 and H8 were tested by calculating the non-parametric statistic gamma using SPSSÂ® software package. Gamma is used to measure directional differences in the relationship of two ranked variables. As noted above, variables for net wealth per capita and market integration, as well as the measur e of cultural knowledge, individual consensus analysis scores, were ranked and categ orized. Hypotheses H9 and H10 employed QAP analysis using the AnthropacÂ® software package (Borgatti 1996a). This software was used for all the cognitiv e methods analysis, including analyzing the
66 free-list and pile sort data and the subseque nt use of hierarchical cluster analysis and multi-dimensional scaling. Finally, all values in this dissertation ar e reported in Brazilian Reais, rather than converted to U. S. Dollars. The conversion ra te when the economic data was collected in the fall of 2001 was approximately R$2.55 per U.S.$1.00. Conclusion Many of the rubber tapper households in this study participated in the empates to prevent ranchers from cutting the trees on thei r landholdings. Their fi ght to protect their livelihoods in the forest led to the establishm ent of the Chico Mendes Extractive Reserve. With the recent election of the Â“forest governmentÂ” the rubbe r tappers now have an ally at the state level. The rebuilding of the feeder road into the reserve and the Chico Mendes Law are examples of a government po licy attempting to assist rural families, providing better access to markets and build the economy on the regionÂ’s natural resources. Now, in collaboration with the Â“forest governmentÂ” they are trying to identify new production activities to diversify and raise incomes. This study will analyze the economic activ ities of 46 rubber tapper households in the Chico Mendes Reserve. Through the use of microeconomic and cognitive anthropological methods, I examine how econo mic forces may be shaping rubber tapper livelihoods in the reserve. In the following two chapters, Chapters 3 and 4, I analyze the rubber tapper economy, examining how wealth and integration into markets may be shaping household income, and more specificall y income from extractive activities. In Chapters 5 and 6, I then turn to examine how wealth and market in tegration may affect rubber tapper cultural knowledge of extractiv e resources. In Chapter 7, I conclude the
67 discussion by examining the implications of these findings for conservation and development in the reserve.
68 CHAPTER 3 THE RUBBER TAPPER HOUSEHOLD ECONOMY It must have been about 4:00 a.m. when Francisco awoke. He lit the wood-burning mud stove and filled a pan with water, and placed it on the stovetop to heat it for coffee. My research assistant and I twisted in our hammocks and attempted to fall back asleep. The radio went on. Music blared through the h eavy crackle of static as Francisco tuned the radio: the signal faded in and out as it tr ied desperately to reach the twisted piece of wire that served as an antenna that ran from roof of the house to the radio on the dining room table. It seemed loud enough to be heard for miles across the forest, maybe even at the neighborÂ’s landholding, about two or thre e miles away. We were sleeping in the kitchen as the thatch roof of the large front porch wher e we would normally sleep was filled with holes and slowly falling apart. It had looked like rain the night before and Francisco and his wife Renata ha d insisted that we sleep in the kitchen. They had put off the much needed re-thatching of the roof b ecause a new house was to be built. The wood to make the house had already been cut, the sawn timber planks stacked orderly in the clearing. Wooden shingles, or carvaco , for roofing, had also been cut and were ready. Nonetheless, it would be a few months be fore it would be built. Francisco was counting on help from his three sons to build the house, one a carpenter with housebuilding experience. However, it was not so easy to get them together. His oldest son still lived in the forest, in fact he lived on the same landholding which was now divided, but he now worked as a technical assist ant for SEFE, assisting in the recently implemented copaiba oil extraction, part of the govern mentÂ’s efforts to diversify the
69 extractive economy. As a result, he was often traveling for days at a time. A second son was living in the city of Xapuri with his wife and child, studying for his high school degree and working for a small cooperative of para-florestais , rubber tappers trained as extension agents, in which he was a founding member. A third son, while still living in the forest, often worked for day wages at an uncleÂ’s small farm on the Petropolis Road. He also worked on the labor crew that was bui lding the feeder road that stretched from Xapuri to the Community of Rio Branco a nd was now being extended to the community center of SÃ£o JoÃ£o de Guarani in the Boa Vi sta forest area. Eventually the house was built, months after the boards were cut and stacked. Francisco was up early each day, eith er to head out to work in the roÃ§ado , or manioc field, to pull manioc for making farinha , or manioc flour, to hunt, or some combination of the two; manioc fi elds are a prime hunting spot for paca , a large forest rodent that in most rubber tapp er homes is a staple part of the diet. The sale of manioc flour was now one of numerous activities in the householdÂ’s diverse income basket, which also included the occasional sale of a young steer or cow, or transport animal, the collection of Brazil nuts, and wage labor. Francisco had tapped rubber for years, even while neighboring households had given up this activity. However, he had recently stopped tapping rubber due to health problems. None of his sons tapped rubber (although all had when they were younger) and none had an y plans to return to this activity, though his middle son now living in Xapuri sometimes talked about it when work was hard to find. However, now wages, both in the forest and in the city, were attracting their labor. The household economy in the forest is ch anging: a wage labor opportunity appears with new development projects, the price of an agricultural product improves, savings
70 allow for the purchase of a cow, a son stops tapping rubber to work at a nearby farm or ranch. Together these changes are re shaping the rubber tapper household economy. This chapter and the following explore the economic changes taking place in the Chico Mendes Extractive Reserve by examin ing the rubber tapper household economy. More specifically, I analyze th e diverse wealth holdings and market-oriented activitiesÂ— including product trade and off-farm laborÂ—of rubber tapper households and how they are shaping household income, and more specifica lly income from extractive activities. I begin by briefly examining a few previous studies on the rubber tapper household economy conducted in the reserve. This is followed by general discussion of household demography, to introduce the study families. I provide this information as recent research in the reserve (Campbell 1996, Gomes 2001, Weigand 1997, Souza 2001) has focused less on the demographic makeup of r ubber tapper families, which also reveals changes in the forest household. I then turn to examining in detail the diverse wealth holdings of rubber tapper households, analyzing not only the varied ways that households inve st capital in their landholdings in the forest, but also in the city. This includes exploring investments oriented toward both production as well as consumer goods. The variation in household wealth holdings is much greater than meets the eye as one initially enters reserve landholdings. Among the 46 households in th is study, wealth holdings varied from approximately R$740.00 to R$41,000.00, the wealthiest household holding over 50 times the wealth of the poorest household. Although th is highest figure is not typical in the reserve, it demonstrates the ability of forest households to generate substantial wealth in the forest. Further, and more importantly, I consider how household investments change
71 as household wealth grows. I argue that whil e households with great er wealth holdings invest a greater proport ion of wealth in consumer items , they also undertake investments that are more associated with deforestat ion, including cattle production and construction of landholding structures that support this act ivity, such as fences and corrals. Thus, household accumulation of wealth in the reserve will likely pose challenges to the current reserve management plan that places limits on land clearing for agricultural activities, and more specifically pasture for cattle grazing. The second half of this chapter examines rubber tapper income earning activities. I examine the diverse sources of income bot h from on-landholding production as well as from off-landholding labor activities. On-l andholding, I consider the various production activities undertaken for both c onsumption and trade. Here I concentrate in particular on the role of extraction in bot h household consumption and trad e, considering not only the value that traditional extr active products, such as r ubber and Brazil nuts, add to household income, but also how medicinal pr oducts, vines, wild fruits and game contribute to household consumption. I argue that extraction remains an important activity for all reserve house holds, even among the most wealthy. Yet, the importance of specific income-earning activitie s, including extracti on, changes as wealth grows. Indeed the diversity in annual productive inco me among study households, ranging from approximately R$1,650.00 to R$11,200.00 suggests th at households are undertaking very different livelihood strategies. Within this discussion I analyze the role that off-farm labor activities play in hous ehold income, laying out the di verse off-farm income earning activities that rubber ta ppers undertake. These include bot h salaried positions with the
72 state, as teachers and health agents, as well as both skilled and unskilled wage and piecemeal positions. I also explore very briefly the trading pa tterns of rubber tapper households, with a particular focus on the trade of rubber and how trade patterns have changed with the implementation of the state rubber subsidy. In comparing the findings of this study with data collected with many of the same house holds in 1996, I find that the subsidy has had a tremendous effect on trading patterns in the forest. Yet, the subs idy has not translated into increased rubber production, or the reta king of this activity by households that had abandoned it by 1996. Indeed, rubber produc tion among 28 households fell dramatically over this five-year period. In the following chapter I bring together th e discussion of rubber tapper wealth and income by responding to the key questions th at drive this study: how do wealth and integration into markets, both through th e undertaking of trade and off-farm labor activities, affect household income, and mo re specifically income from extractive activities? Recent Household Economy Studies in the Chico Mendes Extractive Reserve Despite the importance the concept of ex tractive reserves holds for conservation and development in the Amazon region, we st ill know little about th e livelihoods of the rubber tappers who live there and how they are changing. This lack of research leads us to cling to the images of the rubber tapper household economy found in historical accounts of the rubber boom period (Dean 1987, Weinstein 1983) and the post-World War II period (Allegretti 1979), with the r ubber tapper plying his trade in the forest, tapping rubber trees and se lling his production to his patrÃ£o , or patron, itinerant traders or commercial establishments in the city (alway s being exploited), grow ing rice, beans, and
73 manioc, and hunting wild animals. These imag es themselves are of ten found in the local media in Acre, with local newspapers pub lishing photos of a rubber tapper using a poronga , a headlamp, made of metal with a ke rosene flame serving as a light. The poronga used to be standard equipment for rubber tappers, who would head out before sunrise to begin tapping rubber, a f lickering flame lighting the way. The poronga is now long retired from use by most rubber tapper hous eholds. Among the 46 households in my study, only three households reported that they own one, and only two said they actually used it. Some residents smiled and chuckled softly when I asked them if they owned one.22 Although this traditional image of the r ubber tapper remains strong, a few recent studies have begun to shed light on the cha nges taking place in ex tractive reserves in Acre, and more specifically the role that th e extraction of forest products has played in the household economy over the last decade. Campbell (1996) conducted research in two communities (approximately 25 households total) located in the Chico Mendes Extractive Reserve and the Chico Mendes Agroextractiv e Settlement (commonly referred to as Cachoeira) in 1991 and 1994. She found that in both 1991 and 1994, Brazil nuts contributed, on average, more to househol d income than rubber and the sale of agricultural crops. Mean inco me from the sale of rubber fell considerably over this 3year period, from approximately $292.00 to $93.00. For households where the male or female head of household was employed by the CAEX Brazil nut project, either by the in-forest mini-Brazil nut processing factor y, or in a household level processing unit, 22 Porongas ultimately were not included in the wealth measure for equipment. Despite searching at numerous businesses, including CAEX, in Xapuri, a poronga could not be found to a value it. Assuming we could find one in Xapuri, we did not ask any of the three households that owned one the purchase price.
74 employment contributed on average $361.00 to income in 1994 (Campbell 1996: 124). For these same households, she found a si gnificant difference in average household income from 1991 to 1994, growing from $656.00 to $1,363.00. But while CampbellÂ’s research showed the generous contributi on that Brazil nuts can make to household income (for those who have them), through collection and selling, as well as the creation of employment opportuni ties, it also suggested a declining interest in rubber production.23 From 1991 to 1994, rubberÂ’s contribution to household cash income fell from 45.0 % to 17.0 %. The numbe r of households that did not tap rubber increased from 2 of 23 to 6 of 21 households (Campbell 1996: 111). Hence, while CampbellÂ’s research documented the substantial impact of the Brazil nut project on the income of some rubber ta pper households, principa lly in the form of creating wage labor opportunitie s for forest households, but al so from collection and sale of the nuts themselves, it also highlighted th e effects of the falli ng price of rubber. Rubber tappers were beginning to seek othe r income sources, incl uding the production and sale of other extractive or agricultur al products, and off-farm wage and salaried labor. In a more recent study carried out under the UFAC project, AnÃ¡lise EconÃ´mica dos Sistemas BÃ¡sicos de ProduÃ§Ã£o Familiar (ASPF) (Economic Analysis of Basic Systems of Family Production) in the Chico Mendes Extractive Reserve in 1996, Castelo (2000) examined reserve households through an economic lens, including wealth, and in particular, production income and costs, and compared these figures with urban-based 23 Campbell conducted her research in two communities wh ere Brazil nut trees are found. In particular, one community in the Chico Mendes Agroextractive Settlement is known for its large Brazil nut production. There are a number of areas within the Chico Me ndes Reserve where Brazil nut trees are not found.
75 wages. Among the 67 households in th e study, average househol d wealth totaled approximately R$4,203.00, the largest portion of wealth, R$1,990.00 or 47.3 %, coming from production animals (i.e., cattle, pigs, ch ickens, etc). Also important to household wealth status were landholding structures , valued on average, R$784.00, 18.6 % of total wealth, and products in stock, R $752.00, at 17.9 %. (Castelo 2000: 61). In terms of income, mark et-based income before subtracting production costs averaged approximately R$1,912.00 per year, 8.2% greater than th e minimum salary (including the 13th salary) of urban workers (Caste lo 2000: 62). Subtracting production costs (including labor), income falls to R$138.00 per year, a figure equal to approximately 8.0% of the minimum wage figure. Only 51.0 % of households had a positive net income (Castelo 2000: 61). The value of goods produced for household consumption was, on average, R$2,449 per year with a total income in monetary terms of R$4,163. Using these figures, which totale d the equivalent of 1.4 and 2.5 times, respectively, of the minimum wage, he argued that the income of rubber tappers in the forest (including consumption values) compares favorably to that of urban workers living in the periphery of the city. In another study drawn from the ASPF proj ect, Souza (2001) calcu lated the median gross annual income of extractivists in the Chico Mendes Reserve at R$1,600.00. (SouzaÂ’s model does not include consumption as part of income). However, total production costs amounted to R$1,980.00, making reserve households, on average, incomelosing ventures (Souza 2001: 6). Souza found that 63.0 % of extractivist households were operating at a loss. These households, unlike those that maintained a profit, gained a higher percenta ge of their income from extractive activities, 48.0 %, than
76 agriculture and animal husbandry. Households that had a positive income maintained a more even distribution of income across various production activities, which include extractive activities, totali ng 35.0 %, agriculture, at 37.0 %, and animal husbandry at 28.0 %. He argued that for extractivists to in crease their income, they will have to pursue more environmentally destructive agricultural and cattle raising activities. This trend was reported in the Projeto Resex Report (2000, cited in Souza 20 01: 8), with the portion of household income from animal husbandry in th e reserve growing from 7.5 % in 1992 to 27.0 % in 1998. Each of these studies has contributed to our understanding of the socio-economic changes now taking place the Chico Mendes Ex tractive Reserve. They all point to a decreasing dependence on rubber as a source of cash income. SouzaÂ’s study argued that agriculture and cattle are becoming more im portant to household income, and households that gain their greatest share of income from extraction are operating at a loss, and thus are the least efficient. Yet, CampbellÂ’s study revealed the important role that Brazil nuts can play in raising incomes through both ex traction and employment. Missing, however, is an understanding of the consumptive role that extractivism plays in the household, principally in the form of hunting and medicina l plant use. Further, by testing specific hypotheses regarding wealth, markets, and in come, we can better understand the forces that may be driving changes in rubber ta pper households, and in particular, forest resource use, among poor and wealthy families, and households more and less integrated into product and labor markets. This and th e following chapter help fill this information gap, examining in detail the role of w ealth and markets in shaping rubber tapper
77 productive activities, and in particular, extrac tive activities. I now turn to presenting a model of the rubber tapper household economy A Model of the Rubber Tapper Household Economy Although much has been written about the r ubber tapper economy, and in particular the aviamento system that reigned over this economy at the turn of the 20th Century, lacking is a more recent and detailed pictur e of the rubber tapper household economy. Figure 3-1 below is a diagram of the rubber tapper household production system. Household demographic characteristics, incl uding number of household members, age, sex, health and education le vel (formal and informal)24 of both children and adult members determine the level of household labor available. These factors influence both the quantity and type of household labor availabl e, i.e., skilled or unskilled. The resource base, access to credit, house hold debt, technical and extens ion service available, and wages for off-farm income will influence how households allocate labor. For example, if a landholding does not have Brazil nut trees, th en labor would not be allocated to the collection of Brazil nuts, at le ast as an on-farm activity. However, household labor could be sold off-farm to work in the collec tion and cracking open of Brazil nuts on another landholding. This model focuses primarily on the economic activities that households may undertake in the forest; it does not lay ou t the various social relations, such as religious affiliation, family networks or social organization membership (CAEX, AMOREX, STR (Rural WorkersÂ’ Union)) th at might potentially influence household economic activities. However, data on me mbership in social organizations were collected and will be introduced into the statis tical analysis as a control factor in the 24 Informal education includes short tr aining courses periodically offered both in the reserve and in the city of Xapuri.
78 regression models at the end of the chapter. In addition to productive income activities, twelve rubber tapper households received monthly government social payments through retirement programs such as: FUNRURAL, a sp ecial program for rural workers; soldado do borracha, a special retirement program fo r rubber tappers who arrived in the Amazon region to tap rubber during or shortly afte r WWII; and, benefits for physically and mentally handicapped persons in the househol d. Each of these programs provides the beneficiary one minimum salary each mont h of the year, except for the soldado do borracha program, which provides two minimu m salaries per year. To collect this benefit, the recipient must travel to Xapuri on a designa ted day each month. All the beneficiaries, except for those who receive the soldado do borracha retirement salary, also receive what is termed a Â“13thÂ” salary, which is paid at the end of the year. In addition, five households in the study noted that they received the salÃ¡ rio maternidade, or maternity salary, benefit. This benefit, often provided through three or four monthly installments, but in some cases as a lump su m, is paid to women who give birth. Total maternity payments in th is study ranged from R$514.00 to R$724.00. Many households indicated that this benefit wa s particularly difficult to ob tain due to administrative processes and documentation required. As portrayed in the diagram, households ma y allocate available labor to on-farm or off-farm activities. Salaries and wages paid for both skilled and unskilled labor will influence whether households supply labor to the off-farm salaried and wage labor market. On-farm labor has been broken dow n into five main production activities: Basic food crops (rice, beans, manioc, corn), farm animals (i.e., chickens, pigs, ducks, goats, sheep, cows), extractive resources (rubber, Brazil nuts, artisan-craft production,
79 Household MembersAge SexEducation HealthWealth Household Labor Available Hired Labor Credit Debt On-Farm Labor Resource Base Landholding Size Rubber Tress Brazil Nut Trees Other Resources Off-Farm Labor Wages Skilled Labor Unskilled Labor Income from OffFarm Labor Basic Food Crops Animals Extractive Resources Processed Goods Plants Barter Income From On-Farm Activities Consumption Gift to Neighbors Market Price Income from OnFarm Activities Household Productive Income Figure 3-1. Model of Rubber Tapper Household Economy.
80 hunting and other gathering activ ities), agricultural processi ng activities (such as making goma , a think gummy liquid that is a by-produc t of making manioc flour), and other planted crops, (i.e. coffee, banana, tob acco). Households can consume production, donate production to neighbors, or sell or bart er output with a number of market outlets. In the forest, these trade outlets include other rubber tapper househol ds, itinerant traders, and cooperative trading posts. In the city of Xapuri, they may sell to CAEX, commercial trading houses, itinerant trader commercial houses,25 restaurants, private homes, and government or non-government organizations. Together, both on-farm productive activities and off-farm labor activities contribute to total household productive income. I will discuss the trading system in greater detail below. I now turn to laying out basic demogra phic data of the 46 reserve households included in this study. I begin by provid ing basic demographic data on household members, also considering from where resi dents arrived and the length of time household heads have been on landholdings. I also prov ide a brief discussion of education in the reserve and household membership in social orga nizations that provide social as well as economic support to rubber tappers. I cau tion that these figures are averages, and averages tend to obscure the diversity a nd uniqueness of each household. As noted above, this is not a random sample of house holds. Where appropriate, I include data on range and standard deviation to show th e variation among households often hidden among averages. This said, based on my ow n experience working in the reserve and knowledge of rubber tapper households both in the study region as well as other areas I 25 Itinerant traders buy and sell both in the forest an d at trading houses in the city of Xapuri. For transactions in the city, they pay a higher price for products and sell supp ies at lower prices in comparison to transactions at the landholdings of rubber tappers in the forest.
81 have visited, there is no reason to believe that the demographic data are not, in general, similar to what might be found through a random identification of households. The Rubber Tapper Household: A Demographic Portrait Rubber tapper households are diverse. De mographically, they differ in terms of family size, age of household heads, and edu cation level. And although a great majority of study household heads has lived in the region since birth, they differ in terms of where they arrived from and how long they have lived on their landholding. They also vary in terms of how large a landholding they mainta in, and even how they use their land. Differences extend to the social or ganizations to which they belong. In the following pages I present a basic demographic portrait of the 46 households in the study. The purpose of this discussion is to provide a general idea of whom exactly I am talking aboutÂ—literallyÂ—when I refer to a rubber tapper household. And while I provide averages in summarizing the data, range s have also been in cluded, and should be considered as important as the averages. Indeed, with more children (and adults) studying, some young adults leaving for the c ity to study or work, landholdings being divided to accommodate the many who stay, what is a rubber tapper household seems far from a statistical average, but rather a di verse mix of men and women, boys and girls, most sharing a similar history, although e ngaging with the forest and urban world according to individual means, as well as hopes and dreams. Age and Sex of Household Members Table 3-1 presents data on the age of household heads, sex and number of household members among the 46 households in the study. I consider both the oldest male and female as household heads (except in one case where the oldest female was 91 and nearly bedridden) as both carry out critical duties for the functioning of the
82 household and landholding. Two households had no females living permanently in the family home. Meieros , or hired hands that worked at the landholding tapping rubber, receiving 50.0 % of the value of the rubber they collect, we re not included as household members, even though they ate and slept at the house for at least part of the year. They were not included, as they do not share any of the return that they receive from collecting rubber with the rest of the household. Table 3-1. Age of household heads, number of household members, sex, and years of residence on landholding for 46 rubber tapper households. All Households Mean Range Standard deviation Age of male household head 42.9 years 18 to 74 years 14.6 Age of female household heada 35.6 years 16 to 61 years 12.1 Total household members 5.7 members 1 to 15 members 2.6 Male household members 3.2 members 1 to 8 males 1.7 Female household members 2.5 members 0 to 8 females 1.6 Years male household head on landholding 14.1 years 0 to 37 years 10.5 Years female household head on landholdingb 11.3 Years 1 to 37 years 7.9 Notes: aOnly 43 households had a female head of household; bData missing for total of 6 households, not including 3 households with no female head. The average age of the male head of household was 42.9 years, while the average age of the female household head was 35.6 years. Age of the male household head ranged from 18 to 74 years of age, while that for the female head was 16 to 61 years, demonstrating the diversity of lifecycle st ages among forest households. The average number of household members was 5.7 memb ers, although variation was considerable, with a range of 1 to 15 persons. The househol d with the greatest number of members was an extended family, with grandchildren maki ng up part of the household. However, two nuclear family households had 11 members, each made up of two adult parents and nine children. The average number of male household members (i ncluding the male head of household) was 3.2 members and the average nu mber of females was 2.5 members. The
83 number of male members ranged from 1 to 8 members, while number of females ranged from 0 to 8 members. The average number of years of reside nce of the male household head on the landholding was 14.1 years, with a range of 1 to 37 years. For females, residency averaged 11.3 years with a range from 1 to 37 years. Table 3-2 below shows the previous residence of the male and female heads of household. The most frequent prior location was another landholding within the same seringal , or forest area. Two-thirds of all male and female household heads had a prior locatio n of residence in a rural area within the municipality of Xapuri (the fi rst three categories in Table 32), indicating that movement within the forest was local and also rural to rural. Six male heads of household have resided at the same landhol ding since birth. Only two females were born on the landholding where they currently reside. Seve n male and eight female heads came to the landholding from an urban area. Table 3-2. Previous residence of ma le and female heads of household. Same landholding Other landholding in Forest Area Other landholding in Xapuri Other landholding outside Xapuri Urban area Farm in colonist settlement area Male household heada 6 (13.0%) 18 (39.1%) 7 (15.2%) 3 (6.5%) 7 (15.2%) 3 (6.5%) Female household head b 2 (13%) 20 (43.5%) 5 (10.9%) 3 (6.5%) 8 (17.4%) 3 (6.5%) Notes : aTwo cases missing. bThree households did not have a female head of households and two cases are missing. Just as many families have lived on the same landholding for years, households also maintain strong family ties in forest areas. Of the 46 study households, 19, or 41.0 % have a second family living in a separa te structure on the same landholding. This figure does not include landholdi ngs that are essentially one landholding that has been
84 subdivided, with children now residi ng on these subdivided holding, and where household members are only a few minutes ap art. I consider these as separate landholdings, having been subdivided many y ears ago and each with its own rubber trails.26 Thirty-nine, or 85.0 % of households ha ve another family member living in the same forest area. Fifteen la ndholdings had patrilineal relati onships, where a son lives on the divided landholding of his parents. A matrilineal relationship, where the daughter resides on the divided landhol ding of her parents, wa s found on two landholdings.27 Education in the Forest Education in the forest centers on the primary schools found in each of the three study communities. Formal education was la rgely unavailable for many rubber tappers of previous generations, although many fo rest communities now have elementary education.28 Students begin schooling in what is referred to as alfabetizaÃ§Ã£o , or beginning literacy, similar to kindergarten. Upon completion of basic literacy, students begin primeiro grau , or primary education. In each of the three res earch communities, students can study up until the quarta serie , or fourth grade. Education beyond the fourth grade must be completed in the city of Xapuri, requiring students to move to the city. In the forest, literacy and grade school are taught separately. However, first through fourth 26 Subdivision of landholdings is becoming more co mmon in the three communities in the reserve where the research took place. As children of households be gin their own families and either cannot afford to purchase a landholding or there are no landholdings empty, or for sale, near their parentÂ’s landholding, parents allow a child to build a home and allocate a set number of rubber trails for the new household to tap. In some cases, the new household grows its own basic crops while in other cases the households share labor duties for one crop area. Some landholding were divided years ago and are now essentially separate landholdings even though households from the same family may occupy these adjoined landholdings. However, others have only been divided only in recent years. In these cases, I consider the households to live on the same landholding. 27 In a third case, the daughter arrived first and her parents arrived a couple of months later. 28 It is important to note that not all communities or forest areas in the reserve have schools.
85 grade are taught together in one classroom. St udents are required to pass an exam to complete each level of schooling. Each of the study communities also has begun an adult literacy program where enrolled adults study two days per week. This is held on the weekends to facilitate adult participation. Additional teachers were recruited from within the communities to teach a dult literacy classes. Professors in the schools are generally from the community where the school is located. However, recently the community of Rio Branco has the assistance of a young woman professor who travels from Xapuri on Sunday and remains in the community until Thursday afternoon. The number of students in this community grew to over 60 students, too large for a single professor to conduct class, thus a second professor was needed. The construction of the new feeder road to the Rio Branco community center provides excellent access during the dry season when school is in se ssion. Residents of Rio Branco noted during the study that they hope to gain approval for primary education through the eighth grade, or completion of primeiro grau . Some suggested that with the improved feeder road, the municipality or state could provide daily transportation serviceÂ—picking up students in the morn ing and dropping them off in the eveningÂ— which would allow students to study in the c ity of Xapuri, eliminating the need for youth to live in the city. The cost of sending a child to the city to study can be substantial, not only in terms of lost labor, but also due to the costs of maintaining them in the city. Parents also indicated that ma ny who do study in the city do not want to return to the forest as they become accustomed to an urban lifestyle. Education level among household members va ries considerably across households. Labor needs at home and distance to school pr ohibit some students from studying. In the
86 community of Rio Branco, school is held f our days per week, while in SÃ£o JoÃ£o de Guarani and Terra Alta school is held three da ys per week. Table 3-3 details literacy and formal education levels among study households.29 Over half of both male and female heads of household indicated that they were able to read and write or had completed formal literacy training. Approximately one-t hird of male and female household heads have received formal schooling beyond literac y. Sixteen households have no adults (16 or older) with literacy or fo rmal schooling. All households that have children 8 years or older indicated that they have at least one child who is studying or has completed one year of formal schooling. Table 3-3. Literacy and formal schooli ng of male and female household heads. Proportion of All Households Male household head is literate (self-stated) 26/46 Female household head is literate (self-stated)a 27/43 Male household head has formal education beyond one year of literacy training 15/46 Female household head has formal education beyond one year of literacy traininga 14/43 At least one child 8 or older studies or has studiedb 33/33 Households with no adults with education 16/46 Notes : aThree households have no female head of household. b Only 33 households have children 8 years of age or older, a ge neral age when children begin to study. Table 3-4 on the following page provide s a breakdown of the maximum level of education completed by at least one househol d member. Two households do not have a literate household member, while basic literacy is the maximum level of education for 11 households. Nineteen households have a me mber who has complete d the 4th grade, the maximum one can complete in the forest, wh ile only seven households have a current member that has completed at least one year beyond 4th grade. This suggests that 29 Each household member was asked the highest level of education that they had completed. Although some members have not received formal training, th ey are able to read and write. Conversely, some household members who have attended schooling for literacy may be unable to read and write. No further questioning or testing was done to confirm the validity of each household memberÂ’s response.
87 households would likely benefit from the provision of additional years of schooling in the forest. The average number of years of e ducation beyond literacy per household member 8 years or older for all 46 households was 1.59 years. Table 3-4. Maximum level of educ ation completed by study households. Maximum Level of Education Completed by Current Household Member Proportion of All Households Non-literate 2/46 Basic literacy 11/46 1st Grade 2/46 2nd Grade 2/46 3rd Grade 3/46 4th Grade 19/46 5th Grade 1/46 7th Grade 2/46 Jr. High 3/46 High School 1/46 Total 46/46 Landholding Size and Land Use in the Forest Table 3-5 on the following page detail s landholding size and use, in hectares, among study households. The size of househol d landholdings is based on the number of rubber trails it contains. Rubber tappers noted that each rubber trail is loosely equivalent to 100 hectares of land area, and I have used this measurement in calculating landholding size.30 Among study households, landholding si ze ranged from 200 to 1,400 hectares. The mode landholding size was 300 hectares, wi th 18 households indicating that they have three rubber trails. Four principal categories of land use in clude basic crops, pasture, plantation/ agroforestry perennial production, and fallow. The basic crops category includes area planted for rice and beans, as we ll as other crops such as corn and manioc root. Only one household did not plant basic food crops in the 12 months prior to data collection, which took place in September 2001. This household was a single male who had spent 30 One hectare is equivalent to approximately 2.5 acres.
88 considerable time during the year in the city of Xapuri. Two households noted that they did not have pasture on their landholdings. Fifteen households noted that they did not have an area designated for perennials. Table 3-5. Landholding size and land use type in hectares in 2001. All Households Mean Mode Range Landholding size 545.65 ha 300 ha 200-1400 ha Land Use Basic cropsa 1.75 ha 1.0 ha and 2.0 hab 0.00 Â– 9.00 ha Pasture 4.30 ha 2.0 ha 0.00 Â– 19.20 ha Fallow 8.37 ha 2.0 ha and 10.0 hac 0.00 Â– 30.00 ha Plantation/ Agroforestry .40 ha .35 ha 0.00 Â– 3.00 ha Notes : aData for two households have been drop ped from this calculation. One household indicated it had 72 hectares of crops planted, while another in dicated 17 hectares of crops planted. bNine households stated 1 hectares and 2 hectares. c Five households stated 2 hectares and 10 hectares. It is important to note that the planti ng of perennials in th e reserve has been encouraged by two important pr ojects, the Islands of High Productivity Project (IAP) and the Vai Quem Querzinho Project. Some hous eholds in the communities of Rio Branco and SÃ£o JoÃ£o de Guarani have participated in both of these projects, receiving technical assistance as well as equipment from pr oject sponsors. The IAP project was implemented in a number of forest areas in the Chico Mendes Reserve, including the Floresta forest area where nearly all hous eholds from the community of Rio Branco reside. This project, init ially implemented by agronomist s at the UFAC Zoobotanical Park and subsequently adopted by the stat e government, involves the planting of rubber tree saplings produced from seeds selected from high yielding adults in a consortium with other tree species in high-den sity plots (Kageyama 1991). Th e objective of the project is to increase rubber production through high-density plantings , thus increasing production (and income) of forest households, while at the same time dramatically reducing labor costs. Consortium plantings are designed to reduce attacks from disease and insects.
89 A second project, the Vai Quem Querzinho Project, funded by the Norwegian Rainforest Foundation, is based in the re serve at a landholding called Vai Quem Querzinho, centered among households in the communities of Rio Branco and SÃ£o JoÃ£o de Guarani. Households from both commun ities have participated in the project, a sustainable development initiative focu sed on encouraging households to adopt agricultural technologies, such as the use of agroforestry, to reduce slash and burn agriculture in the reserve (Evaluation Report, n.d. ). As with the IAP project, participating households have received technical assistan ce and equipment through participation. Thus households in the communities of Rio Branco and SÃ£o JoÃ£o de Guarani have received outside encouragement and incentives for th e planting of perennial crops, as well as equipment, such as machetes, hammers, post-hole diggers and saws. Social Organizations in the Forest In Chapter 2, I noted the role that social organizations have play ed, and continue to play, in supporting rubber tapp er livelihoods both socially and economically. In Xapuri, the three principal organizat ions that support rubber ta ppers are AMOREX, CAEX and STR. Although these organizations, in genera l, work toward improving the livelihoods of all rubber tappers in the re gion, they also provide membersh ip services. For example, membership in CAEX provides members a s lightly higher selling price for rubber and Brazil nuts. Members in all three organiza tions have voting rights for choosing leaders and, in the cases of AMOREX, for establ ishing management practices for reserve residents. The oldest of these organizations is th e STR, begun in the 1970s to ally rubber tappers against ranchers encroaching on thei r lands (Costa Sobrinho 1992). CAEX was founded in 1988 to help rubber tappers earn a be tter price for rubber and Brazil nuts, as
90 well as provide basic supplies at lower prices than those o ffered by itinerant traders and trading houses. AMOREX was founded with the establishment of the Chico Mendes Extractive Reserve in the early 1990s, providing a voice for reserve residents living in the municipality of Xapuri and serving as a liaison between the federal government and rubber tappers. Membership in an associati on or cooperative is required in order to receive the state rubber subs idy. Membership in AMOREX and STR require annual payment of dues, R$12.00 and R$24.00 (R $2.00 per month) respectively, while membership in CAEX requires a one-time paym ent of R$100.00 or equi valent in kilos of rubber, to be returned to the member s hould they decide to leave the organization.31 Forty-two of the 46 study households are member s of at least one of these organizations: 35 are members of AMOREX, 26 are members of CAEX, and 34 are members of STR. The above discussion has provided a gene ral demographic picture of the rubber tapper households in the study. I now turn to discuss the central questions of this study, how wealth and markets influence income-ear ning activities of rubber tapper households. I begin with a description and analysis of household wealth holdings, followed by a discussion of income. I conclude the chap ter by bringing the two topics together to respond to the hypotheses posed earlier in th e chapter. The Diversity of Rubber Tapper Wealth Rubber tapper households in the Chico Me ndes Reserve hold wealth in a diverse set of assets. These are found on their landholdings in the forest, in urban areas, and in a few cases, other rural areas. To examine mo re closely how households of different wealth levels hold their wealth, I have broke n household wealth into three asset groupsÂ— 31 The annual cost of membership in the STR rose to R$36.00, or R$3.00 per month in 2004.
91 on-farm productive wealth, onfarm non-productive wealth, a nd off-farm wealth. Onfarm productive wealth includes equipment, stocks of inputs and goods, structures, and animals, both livestock and transport animals, wealth that produces or aids in the production of income. On-farm non-productive wealth includes the family house and consumer items. Arguably, the family residence does produce income through processing activities, such as cooking, that take place there. In addition, for some households, the house structure is used for st ocking or drying (in the case of tobacco) production, and carrying out producti on activities such as weavi ng artisan items or rolling tobacco. However, I contend that the house structure is very different from other productive structures, as ma ny households are now investi ng substantial capital for comfort, durability, and esthetic purposes, rather than to enhance production. Lastly, offfarm wealth includes land, houses and consumer items held in the city, rural land other than the landholding (one case), and bank accounts (two cases). In this section I examine how study househol ds hold wealth, first by examining the wealth holdings of all study households toge ther, and then examining how different households categorized by net wealth per capita invest their wealt h. I demonstrate how rubber tappers hold wealth in a diverse group of items, and how household investment in wealth varies as per capital wealth holdings grow. My analysis also reveals the great disparity in wealth holdings across study house holds. That there is disparity among study households is not surprising, as households were selected based on differences in wealth holdings. However, the sheer differences in household ability to accumulate capital are somewhat startling. The valuation techni ques for each wealth category have been explained in Chapter 2.
92 Table 3-6 on the following page presents the mean total and net wealth values for all 46 study households. It also includes a breakdown of th e different asset groups that households maintain. I first consider the to tal and net wealth valu es, and then examine more closely the different cat egories of wealth holdings. The mean total wealth holdings of al l households was R$7,484.00, with a mean net wealth slightly lower at R$6,981.00. As noted in the introduc tion of this chapter, there is a gaping divide between the poorest and wealth iest households, with total wealth ranging from R$740.00 to R$40,949.00 and net wealth ranging from R$(305.00) to R$35,380.00Â—one household had more debt than wealth. A few of the wealthiest households maintain over 50 times the wealth of the poorest house holds, a substantial difference for a forest population historical ly considered to be a very homogeneous group. The median figures (not stated on the table) of R$5,337.00 and R$4,914.00, for total wealth and net wealth, respectivel y, remove the distorting effects of a few households with the greatest total and net w ealth holdings and provi de a better gauge of wealth holdings among study households. These fi gures are more closely in line with the mean reserve household wealth figure of R$4,203.00 noted by Castelo (2000). Mean and range figures for total and net wealth per capita help level the wealth playing field by adjusting wealth to th e number of household members. For all households, the mean total wealth per cap ita was R$1,585.00, with a range of R$93.00 to a high of R$6,371.00. Again the wide gap in wea lth holdings is clear. Mean net wealth per capita was slightly lo wer at R$1,465.00, with a range of R$ (305.00) to R$6,233.00. The household with negative net wealth has only one member; therefore net wealth and
93Table 3-6. Total and net wealth per capita for 46 rubber tapp er households in the Chico Me ndes Extractive Reserve in 2001. Wealth Category Mean Net Wealth Range (R$) Standard Deviation (R$) Value (R$) Pct. of Total Wealth On-Farm Productive Wealth Equipment 1,349.00 18.0 192.00 Â– 4,459.00 889.00 Animals 2,691.00 36.0 10.00 Â– 15,829.00 2,955.00 Fowl 204.00 (2.7) 10.00 Â– 580.00 140.00 Small animals 149.00 (2.0) 0.00 Â– 768.00 192.00 Cattle 1,567.00 (20.9) 0.00 Â– 12,806.00 2,289.00 Transport Animals 771.00 (10.3) 0.00 Â– 4,369.00 883.00 Structures 1,009.00 13.5 28.00 Â– 5,681.00 1,205.00 Stock of Inputs 98.00 1.3 8.00 Â– 628.00 145.00 Stock of Goods 174.00 2.3 0.00 Â– 816.00 204.00 Total On-Farm Productive Wealth 5,321.00 71.1 443.00 Â– 22,739.00 4,457.00 Other On-Farm Wealth Household Structure 796.00 10.6 58.00 Â– 3,200.00 647.00 Household Consumer Goods 271.00 3.6 0.00 Â– 4,889.00 787.00 Total Other On-Farm Wealth 1,067.00 14.3 113.00 Â– 6,639.00 1,173.00 Non-Farm Wealth Urban Wealth 1,004.00 13.4 0.00 Â– 10,072.00 2,115.00 Other Rural Land 54.00 0.7 0.00 Â– 2,500.00 369.00 Bank Account 38.00 0.5 0.00 Â– 1,500.00 223.00 Total Non-Farm Wealth 1,096.00 14.6 0.00 Â– 11,572.00 2,268.00 Total Wealth 7,484.00 100.0 740.00 Â– 40,949.00 7,330.00 Debt Bank Debt 417.00 5.6 0.00 Â– 5,500.00 969.00 Other Debt 87.00 1.2 0.00 Â– 1,100.00 190.00 Total Debt 503.00 6.7 0.00 Â– 5,570.00 969.00 Net Wealth 6,981.00 (305.00) Â– 35,380.00 6,840.00 Total Wealth Per Capita 1,585.00 93.00 Â– 6,371.00 1,602.00 New Wealth Per Capita 1,465.00 (305.00) Â– 6,233.00 1,558.00
94 net wealth per capita are identical. Thus, ev en with wealth calculated on a per capita basis, the data demonstrate the great di fference in wealth found in the forest. Household Wealth Holdings by Asset Category A breakdown of wealth holdings by asse t group finds that on-farm productive wealth accounts for the great majority of rubber tapper hous ehold wealth, with 71.1 % of total wealth held in this form. Other onfarm wealth accounted for 14.3 % and non-farm wealth held 14.6 % of total wealth. Looking more specifically at on-farm pr oductive wealth, animal wealth accounted for the largest share of productive assets, 36.0 % of the total, followed by equipment at 18.0 %, landholding structures at 13.5 %, stocks of goods at 2.3 %, and inputs at 1.3 %. Again these findings differ moderately from the findings of Castelo (2000), who found that production animals accounted for 47.3 % of total wealth, and structures 18.6 % of total wealth. However a much larger diffe rence was found in terms of stock of goods, with Castelo noting that hous eholds held 17.9 % of wealth in goods, compared to the much lower 2.3 % noted in this study.32 Cattle and transport animals, at 20.9 %, and 10.3 % of total wealth, respectively, made up the largest share of animal wealth, with cattle being the single largest asset category. One household held nearly R$13,000.00 in cattle. Of the 46 households, 31 held at least one cow and/or steer, and 28 held a cow. A co w is often one of the first large animals acquired by househol ds (in addition to transport animals), as it provides the household with a steady supply of milk for consumption. Thirty-three households maintained an animal for transportÂ—ox, horse, or mule. The value of animal wealth held 32 Castelo (2000) does not state how assets were valued. It is possible that households were holding much larger quantities of rubber and Brazil nuts.
95 in fowl and small animalsÂ—mainly pigsÂ—was much lower. Every household held chickens, while fewer held ducks (30) or capotes (6), also known as Portuguese hens. Thirty households raised pigs, sheep or goats. Equipment is obtained through diverse means, including purchase, barter, as gifts, either from relatives or neighbors, acqui red by development or community assistance organizations (such as AMOREX), and for part icipation in various projects, principally the IAP and Vai Quem Querzinho projects not ed above, but others as well. Some equipment was simply found. Equipment items (43) for which data was collected were held in diverse quantities, some held by n early every household, others by few. For example, all 46 households held machetes and a hatchet, while 44 households held a hammer, and 42 had shotguns. Twenty-eight households owned steel ovens used for making manioc flour, while 33 owned post-hole diggers. These figures contrast with the holding of equipment of much greater value. For example, only 11 households owned a chainsaw, while nine held a rifle, and eight owned a gas-powered motor used primarily for processing manioc roots into flour. On ly three households owned a canoe and motor. Canoes and motors require a high capital inve stment, and this is one reason that many households do not hold them. In addition, as only households in the Terra Alta community travel by river, only resident s of this community would need them. Like equipment, the presence of productiv e structures on landhol dings varies, with a few basic structures found on nearly all la ndholdings, while other structures were found on only a few. Forty-one households had a paiol , or storage building, while 28 had a casa da farinha , an open-air hut for processing mani oc root into flour. Thirty-four households had some type of fencing on their landholding, while 28 had small fenced
96 gardens, 27 had chicken-coops, and 25 had pi gpens. Of the 27 households that owned chicken coops, 13 benefited from a project sponsored by AMOREX that provided all the materials for their construction, including payment for the extraction and sawing of timber planks. Rubber tapper households provi ded the labor for construction. Only ten households owned corrals and nine had small structures to store saplings and produce seedlings. Only four households had defumadores , the small huts used by all rubber tappers in years past to smoke latex over fire to produce rubber. The value of investments in structures ranged from a low of R$28.00 to a high of R$5,681.00. Households held a very small portion of to tal wealth in stocks of inputs and goods. The most commonly held inputs were hunting materials, including shotgun shell casings (35 households), gunpowder (30), lead shot (3 0), and firing caps (26). Rope and worm medicine for animals were also widely held, carried by 32 and 20 households, respectively. Less common were stocks of rock salt (16), barbed wire (14), and nails (14). Few households held vaccines (3) and vitamins (4) for animals, and no household held a stock of insecticides. Stocks of goods were he ld by numerous households in the form of agricultural goods, such as basic food crops (34), coffee (3), and tobacco (2). Six households were holdi ng rubber in stock. Although wealth held in income producing a ssets accounted for th e largest share of wealth on the landholding, other on-farm asse ts accounted for 14.3 % of total wealth. The home structure was the principal holder of other on-farm wealth, valued at 10.6 % of total wealth. All households maintained a house on their land holding, although these houses differed in material and construction costsÂ—the value of the homes ranged from R$58.00 to R$3,200.00. The highest value houses were made of sawn woodÂ—in nearly
97 all cases requiring the hiring of carpenter to cut boards and build the structureÂ—with roofs made of aluminum or Brasilit, a pressed-fiber board material, which must be purchased. Homes made of palms are almost always built with household laborÂ—both for extraction of mate rials and constructionÂ—and only basi c construction materials, such as nails. Of the 46 study houses, 22 were constructed of sawn wood, while 24 were constructed from palms, principally the paxiÃºba ( Iriartea deltodea Ruiz & Pav.) and aÃ§aÃ ( Euterpe precatoria Mart.) palms. The principal r oofing material of households was palm thatch, most commonly jarina ( Phytelephas macropcarpa Ruiz & Pav.) and ouricuri ( Attalea phalerata Mart. ex Spreng.). Twenty-four households used palm thatching for roofing, while nine house holds used wood shingles, which require considerable household labor for preparation and placement. More expensive roof materials, in the form of aluminum sheets and Brasilit, were used by nine and three households, respectively. Other on-farm wealth also included hous ehold consumer goods, which accounted for, on average, only 3.6% of total househol d wealth. However, the variation in the holding of consumer goods was great, rang ing from zero to approximately R$4,900.00. Clothes and pots and pans would likely be th e most commonly held consumer items, but the difficulty in obtaining an accurate a nd reliable measure of these items without conducting an extensive inventory led me to leave them out of the study. I focused on a short list of items that could be more readily recalled and counted by household members. Radios were the most commonly held consumer item, held by 31 households. Water filters were held by 16 households, while gas ovens and natural gas tanks were owned by eight and 11 households, respectiv ely. Two households had solar panels,
98 rechargeable batteries and televisions. Four households owned revolvers, carried primarily for safety from predatory animals, mainly jaguars, when traveling in the forest. Non-farm wealth, at 14.6 % of total wealth value, was principally held in urban assets, including land, houses and consumer items, which accounted for 13.4 % of total wealth. Thirteen households owned both land and a house in the city. One household owned only land, while another had partia l ownership of a house. Among major consumer items held by households in the city, 14 owned gas ovens or counter top ranges, and 15 owned natural gas containers. Te levisions (7), radios (6), stereos (6) and refrigerators (4) were less commonly owned. One household he ld chickens in the city. Again, the range in value of assets held in the city among study households was great. Thirty-three households held no assets outside the forest, while one household held urban assets valued at nearly R$12,000.00. Non-farm wealth was also held in the form of other rural landholdings and financial holdings. Other rural landholdings accounted fo r 0.7 % of total wealth. Four households noted that they owned land, or shared inher itance in rural area land other than their landholding. As noted in Chapter 2, the value of only one of these la nd areas is included in this figure. Financial holdings in banks totaled 0.5 % of total wealth. Only two households held bank accounts. Finally, debt, both to banks and trading hous es, the cooperative, and individuals in region, amounted to 6.7 % of total wealth. Debt to banks, in the form of both loans for production and custeios , or lines of credit for land holding improvement, accounted for most of this total. Eight households held outstanding bank debt for loans used for the planting of perennials, principally co ffee, but also peach palm fruit ( Bactris gasipaes
99 Kunth). Nine households had outstanding debt for custeios . Twenty-four households had debt with trading partners or indi viduals in the fore st or urban area.33 This snapshot of wealth holdings acro ss the 46 study households finds that rubber tapper households hold a diverse set of assets, both on and off-farm. The great majority of wealth is held in the forest on lan dholdings, invested principally in productive activities. Animals were the largest single a sset category, with cattle in particular, holding the greatest wealth value, although e quipment and investments in landholding infrastructure were also important asset hol dings. Conversely, othe r on-farm wealth and non-farm wealth together amounted to less than half of total household wealth, with the household structure maintaining the greatest share of non-productive wealth. While all study households held some form of producti ve wealth, the numb er of households holding assets in other categories differed gr eatly. For example, in measuring household consumer items, eight households did not mainta in any of the nine items included in this measure, and an additional 21 held only one itemÂ—principally radios. While the above description is useful in providing a basic unde rstanding of rubber tapper wealth holdings, a more important quest ion emerges regarding how asset holdings change as households become wealthier. More importantly, where are wealthier households investing their wealth, and what ar e the implications of these investments for future production activities in the reserve? And further, what are the implications for conservation? The following section will respond to these questions. 33 Data on outstanding debt other than bank debt are missing for one household.
100 A Portrait of Changing Wealth Investments in the Reserve If you entered just about any rubber tapper landholding as recently as 20 to 25 years ago you probably would have encountered more or less the same picture. Houses were likely made of palm slat s and palm thatch, and the fron t of most houses would have a large open-sided porch with a wooden rai ling. The cleared area surrounding the house would be small, and you would lik ely see a storage shed and a defumador , to the side of the house, and a casa de farinha , where manioc flour is made, nearby. You might find a pigpen as well, but probably not a chicken coop or a corral. Mo st fencing would be limited to closing the entrance of the forest path to the landholding, principally to keep the few large animals, a Â“ reserva de recursos ,Â” to be sold in case of emergency, rather than as part of a larger production strategy, from wandering away (C NS 1992: 14). Most landholdings probably did not even have thes e blockades, as many would not have to worry about animals escaping, as they did not have any.34 There are still a few houses in the forest that are just like this. Yet, the situation for other landholdings is very diffe rent now. Houses are made of different materials Â– some even painted and many without porchesÂ—and an imals are frequently seen roaming across landholdings, with clearings fo r pasture stretching, in some cases, hundreds of meters. Fencing cuts across open fields, separating different pasture ar eas. These changes are not yet the norm, but they are app earing in diverse forms and they are clear signs that wealth holdings across rubber tapper hous eholds vary considerably. In this section, I examine how rubber tapper households of different wealth have invested their holdings both on their landholdi ngs and in the city. To do this, I have 34 See Allegretti (1979) for a de tailed picture of life in the seringal in Acre in the 1970s.
101 divided the households into four differ ent wealth rank groups. The division of households into ranked categories was determ ined by first producing a histogram that graphed the net wealth per capita of each househ old ranked from lowest to highest value. I then examined the histogram for natural breaks in the net wea lth per capita values, resulting in the creation of four wealth rank groups. Table 3-7 below provides a summary of households ranked by net wealth per capita. Although th e analysis will not focus on wealth differences across communitie s, I have also provided a breakdown of households in each wealth category by community. Table 3-7. Rubber tapper households ra nked by net wealth per capita in 2001. Wealth 1 Wealth 2 Wealth 3 Wealth 4 Total Households by Community Less than R$500 R$500 Â– R$999 R$1,000 to R$1,999 R$2,000 or greater Rio Branco 4 3 4 7 18 Guarani 2 5 2 4 13 Terra Alta 5 8 2 0 15 Total 11 16 8 11 46 Table 3-8 on the following page displays the mean wealth holding figures for each of the four wealth rank groups. The mean fi gures are included primarily to provide a general idea of the volume of wealth held by each of the wealth groups. The discussion will focus on how investments in wealth hold ings change as wealth holdings increase. Table 3-8 demonstrates the shift in wealth investments that takes place as wealth holdings increase. The accompanying Figure 32 provides a graphic display of this shift across the four wealth rank gr oups. For wealth rank 1 househol ds, all wealth is held onfarm: 75.8 % of wealth is held in on-farm productive wealth and 24.3% of wealth is held in other on-farm wealth. Wealth rank 1 hous eholds maintain no off-farm wealth. This contrasts with wealth rank 4 households , who hold 65.1 % of wealth in on-farm
102Table 3-8. Percentage of mean wealth from diverse wealth holdings for house holds ranked by net wealth per capita. Wealth Rank 1 NWP Less than R$500 Wealth Rank 2 NWP R$500 to R$999 Wealth Rank 3 NWP R$1,000 to R$1,999 Wealth Rank 4 NWP Greater than R$2,000 Mean (R$) Pct. of Total Wealth Mean (R$) Pct. of Total Wealth Mean (R$) Pct. of Total Wealth Mean (R$) Pct. of Total Wealth On-Farm Productive Wealth Equipment 751.00 38.3 1,122.00 21.6 1,688.00 25.7 2,032.00 11.9 Animals1 411.00 20.9 2,154.00 41.5 2,134.00 32.5 6,159.00 36.2 Fowl (124.00) (6.3) (239.00) (4.6) (237.00) (3.6) (207.00) (1.2) Small animals (72.00) (3.7) (172.00) (3.3) (146.00) (2.2) (192.00) (1.1) Cattle (0.00) (0.0) (1,146.00) ( 22.1) (1,097.00) (16.7) (4, 089.00) (24.0) Transport Animals (214.00) (10.9) (596.00) (11.5) (652.00) (9.9) (1,670.00) (9.8) Production structures 177.00 9.0 800.00 15.4 729.00 11.1 2,346.00 13.8 Stock of Inputs 36.00 1.8 86.00 1.7 69.00 1.1 197.00 1.2 Stock of Goods 111.00 5.7 94.00 1.8 176.00 2.7 350.00 2.1 Total On-Farm Productive Wealth 1,487.00 75.8 4,255.00 82.0 4,796.00 73.1 11,083.00 65.1 Other On-Farm Wealth Household Structure 415.00 21.2 595.00 11.5 802.00 12.2 1,467.00 8.6 Household Consumer Goods 60. 00 3.1 70.00 1.3 118.00 1.8 887.00 5.2 Total Other On-Farm Wealth 475.00 24.3 665.00 12.8 920.00 14.0 2,353.00 13.8 Non-Farm Wealth Urban Wealth 0.00 0.0 267.00 5.2 815.00 12.4 3,218.00 18.9 Other Rural Land 0.00 0.0 0.00 0.0 0.00 0.0 227.00 1.3 Bank Account 0.00 0.0 0.00 0.0 31.00 0.5 136.00 0.8 Total Non-Farm Wealth 0.00 0.0 267.00 5.2 846.00 12.9 3,582.00 21.0 Total Wealth 1,962.00 100.0 5,187.00 100.0 6,563.00 100.0 17,018.00 100.0 Debt Bank Debt 375.00 277.00 237.00 791.00 Other Debt 179.00 72.00 44.00 45.00 Total Debt 555.00 28.3 350.00 6.7 282.00 4.3 836.00 4.9 Net Wealth 1,407.00 4,837.00 6,312.00 16,182.00 Total Household Wealth Per Cap ita 370.00 777.00 1,479.00 4,051.00 Net Household Wealth Per Capita 187.00 717.00 1,402.00 3,879.00
103 productive wealth, 13.8 % in other on-farm wealth, and 21.0 of wealth in non-farm wealth. Examining all four wealth rank gr oups, general trends across the categories emerge. First, there is a decrease of on-fa rm wealth as a percentage of all wealth holdings, as wealth holdings increase, and a corresponding percentage increase in nonfarm wealth holdings. Looking more specifically at on-farm productive wealth, there is a general fall (except for wealth rank 2) in on-farm productive wealth holdings as a percentage of total wealth. Other on-farm wealth holdings, as a percentage of total wealth are greater for wealth rank 1, due to the high value of the physical house structure in comparison to other wealth holdings. Howe ver, as household wealth increases, the value of the house falls as a pe rcentage of total w ealth (from 21.2 % to 8.5 %), even as its value as a structure increases. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Wealth 1 Wealth 2 Wealth 3 Wealth 4 Wealth RankPercent of Wealth Holdings Total Non-Farm Wealth Total Other On-Farm Wealth Total On-Farm Productive Wealth Figure 3-2. Wealth holdings for househol ds by wealth rank group. Chico Mendes Extractive Reserve, 2001. Thus, after an initial increas e (see wealth rank 2) in the value of on-farm productive wealth, a lower percentage of total wealth is held in produc tive activities as households
104 allocate more wealth to off-farm investments, principally in the form of investments in land, houses and consumer goods in the city of Xapuri. Other on-farm wealth, held in the form of household structures and consumer ite ms, falls as a percentage of total wealth, leveling off across wealth ranks 2, 3 and 4, with households investing a similar percentage Â– around 13.0 % Â– of household wealth in this category. However, while the percentage of wealth invested in other onfarm wealth is maintained relatively steady across wealth rank groups 2 through 4, for w ealth rank 4 households, the value of the physical structure as a percen tage of wealth falls, and correspondingly, th e percentage value of wealth held in consumer goods increases. The data suggest that at some point, the house structure is no longer improved upon, and investment shifts to the increased acquisition of consumer items. The growing interest in motorcycles among th e wealthiest rubber tapper households in the community of Rio Branco, brought on by the improved feeder road, reflects this change. Between the first and second field visits, two househol ds in the studyÂ—in wealth ranks 3 and 4Â— acquired motorcycles. Table 3-8 also reveals important shifts in investments within the on-farm productive asset category. These shifts are s hown graphically in Figur es 3-3 and 3-4. In wealth rank group 1, the greates t percentage of wealth is invested in equipment, accounting for 38.3 % of total wealth, and ove r half (as seen in Figure 3-3 on the following page) of all productive wealth. Production structures on the landholding accounted for 9.0 % of total wealth, principall y in the form of storage buildings and casas da farinha .35 Animals account for 20.9 % of total wealth, with transport animals 35 Only households in the communities of Rio Branco an d Terra Alta indicated that they benefited from the AMOREX chicken coop project. I am unaware of how households were selected for inclusion in this
105 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Wealth 1 Wealth 2 Wealth 3 Wealth 4 Wealth RankPercent of On-Farm Productive Wealth Stock of Goods Stock of Inputs Production structures Animals Equipment Figure 3-3. Productive wealth holdings of rubber tapper households by wealth rank category. Chico Mendes Extractive Reserve, 2001. 0% 20% 40% 60% 80% 100% Wealth 1Wealth 2Wealth 3Wealth 4 Wealth RankPercent of Animal Wealth Transport Animals Cattle Small animals Fowl Figure 3-4. Animal wealth hol dings of rubber tapper househol ds by wealth rank category. Chico Mendes Extractive Reserve, 2001. project. If it is assumed that all households in these communities were eligible for this project, then nine of the 11 households in wealth group 1 would potentially be eligible. Of these nine households, only one household in this lowest wealth category indicated that they benefited from this project.
106 accounting for 10.9 %, or over half of animal wealth, and fowl holding 6.3 % of total wealth. Most notably, wealth rank 1 households did not own any cattle. In contrast, the three greater wealth ra nk groups show considerable differences. First, the value of equipment holdings as a pe rcent of total wealth falls substantially from wealth rank 1 to wealth rank 4 households. Although the monetary value of investments in equipment increases as total wealth rises, equipment accounts for a reduced percentage of total wealth, falling from 38.3 % for wealth rank 1 to a low of 11.9% for wealth rank 4 households. Conversely, investment in animal wealth is much great er as a percentage among wealth ranks 2, 3 and 4 compared to wealth rank 1. For wealth rank 2 households, animals accounted for 41.5 % of total wealth, w ith cattle making up the largest proportion of animal wealth and 22.1 % of total wealt h. Wealth rank groups 3 and 4 had similar percentages, 32.5 and 36.2, respec tively, of animal wealth ho ldings. Wealth rank 4 had the greatest percent of wealth holdings in cattle among all wealth groups, with 24.0 %, or nearly one-quarter of total wealth, held in cattle. Interestingly, the monetary value of wealth holdings in fowl and small animals did not change dramatically across wealth rank groups. Therefore, as a percentage of to tal wealth, the investme nt in these animals fell. However, investment in transport anim als increases in value as wealth rises, and thus remains relatively constant as a percentage of total w ealth across the four wealth rank groups. An examination of investments in on-farm productive structures shows a general trend towards increased investments in st ructures as wealth rises, although as a percentage of total wealth this figure does not change dramatically. These investments result from the need for fencing and a corral with growth of the cat tle herd. Although the
107 percentage of total wealth invested in pr oduction structures remains relatively stable across wealth groups, the monetary value of in vestment, particularly for wealth group 4, indicates that more fencing, and sturdier a nd larger corrals are be ing built. On-farm productive wealth holdings in stocks of inputs and stocks of goods as a percentage of total wealth were low for all four wealth rank groups and varied little among them. Finally, Table 3-8 reveals differences in rubber tapper debt holdings across the wealth rank groups. The value of debt is gr eatest for the wealthiest households, although debt, as a percentage of total wealth, is much greater among wealth rank 1 households, reaching 28.3 % of total wealth. Wealth rank 1 households also hold a greater percentage of non-bank debt, owed principally to trading pa rtners and individuals both in the forest and urban area. This suggests that low wea lth households rely more on short-term credit to fund purchases, while the wealthiest hous eholds rely on longer-term bank debt to invest in production activities. The above discussion argues that as house hold wealth grows, households invest this wealth in different ways. Although on-fa rm productive wealth is where all wealth groups hold the majority of their wealth, th e rubber tapper wealth portfolio changes as wealth grows. On-farm, productive investme nts shift from equipment to animals, with cattle an important asset for all but the lo west wealth households. Investment in productive landholding structures also grows, although not as dramatically as animal wealth. As wealth grows, households contin ue to invest in their homes, although as a proportion of wealth this investment falls . Consumer goods become particularly important for the wealthiest households. Wealthier households also diversify their holdings to include urban assets, through investments in land, houses and consumer
108 goods. This is most evident among the wealth iest households, which hold assets in the city of Xapuri that account for over 20.0% of total wealth. Thus, as households become wealthier, they continue to invest in pr oductive activities, but also invest a growing proportion of wealth off the farm in the city of Xapuri. The implications of changing wealth inve stments with a rise in household wealth have fairly clear consequences for resource conservation in the reserve. Most notably, the findings show that as households become wealthier, they inve st a greater proportion of their wealth in cattle. A growing cattle herd requires incr eased pasture, which in turn requires the increased cutting of trees. Not surprisingly, positive and statistically significant correlation was found between past ure area and househol d cattle wealth (PearsonÂ’s r = 0.469, p-value = 0.001) and household net wealth per capita (PearsonÂ’s r = 0.465, p-value = 0.001). As the wealthiest house holds confront the reserve management plan limits for agricultural production and past ure, they will have to find new ways to invest their wealth. Initia lly, this will include the opening of bank accountsÂ—only two households hold bank accounts, including the wealthiest household that is pushing the limits of pasture areaÂ—and increased investment in consumer items, such as motorcycles. In a very short time it is likely that the wealthiest households will own small pick-up trucks, potentially earn ing income by transporting other reserve residents to the city and back. Households will like ly increase investments in both urban property and urban consumer items. They will probably seek out investments in other rural areas to expand productive activities that the reserve management plan limits. It seems possible that they may look to purchase other landholdings in the reserve to increase cattle production. As roads move deeper into the reserve, even distant locations become market accessible.
109 There may be a call for amending the reserve management plan that limits agricultural activities, a proposal that would likely provoke an emotional debate among reserve residents. Having examined how rubber tappers hold w ealth in the reserve, including the changing investment strategies of households of different wealth rank, I now move the discussion to a second critical component of the rubber tapper economyÂ—income. Following this in-depth analysis of both on a nd off-farm income activities as well as an examination of how income changes, both in value, and in source, as wealth grows, I conclude with a statistical analysis of th e role of wealth and market integration on household income, responding to the hypotheses st ated at the outset of this chapter. Rubber Tapper Income in a Changing Forest Economy Â“ Eu vou pra rua pra vender o meu borracha Â” (I am going to Xapuri to sell my rubber). Those were the words of six-year ol d TomÃ¡s as his family was preparing to head to Xapuri. I had just completed my second vi sit to the household in the Boa Vista forest area, nearly 2 days travel by animal to the city. Just about ever yone in the household was planning on heading to the city the next day. The family was making its monthly trek to deliver 120 kilos of farinha under a contract esta blished with a local merchant. The trade value of the farinha Â—approximately R$60.00Â—would be used to pay off the householdsÂ’ debt for purchases made the previous mont h. TomÃ¡s wanted to go along to sell the few kilos of rubber that he had tapped over the past months. Although Paulo and Juliana sold farinha every month, they still relied principall y on the sale of extractive items, Brazil nuts and rubber, for cash, which accounted for approximately 38.0 % of all trade income. But a quick glance across the large clearing on the landholding hinted that this would be changing in the coming years. The househol d had only recently built a new corral, and
110 the sale of two steers trained to carry cargo had fetched a goo d price over the past yearÂ— R$500.00 eachÂ—and contributed handsomely to income. This was more income than an entire year of rubber tapping had produced. Yes, TomÃ¡s was heading to the city to sell his rubber now. But would he be doing the same in 5 or 10 years? In this section, I discuss the income ear ning activities of households, examining onfarm income producing activitiesÂ—both for c onsumption and tradeÂ—as well as off-farm income earning activities, including an anal ysis of the trading patterns of reserve households. I begin by examining the findings for all 46 study households, discussing in some detail the various categories of activ ities which produce household income in the forest. I focus in particular on the role that extractive activities play in income generation. Building on this discussion, I then bring in the above discussion of household wealth, and examine how household income vari es, both in terms of value and source, as wealth increases. The rubber tapper household economy m odel presented above (Figure 3-1) illustrates the diverse productive activities, both on and off-farm, which contribute to household income. On-farm activities produ ce agricultural and ex tractive goods for both consumption and trade. Off-farm activi ties provide cash income to households. Although nearly every household plants basic food crops for consumption, as well as for sale in many cases, not all do. Among th e 46 households in this study, three did not produce rice or beans in the 12 months prior to the first field visit. These households used income earned through off-farm labor to purchase basic goods for consumption. Table 3-9 on the following page summari zes the income earning activities for all 46 study households. Both mean and range values are provided to demonstrate the diversity
111Table 3-9. Mean income of 46 rubber tapper households in the Chico Mendes Extractive Reserve for 12-month period in 2001 Consumption Trade Total Range (R$) St. Deviation (R$) Mean Mean Mean Pct. of Pr od. Inc Pct. of Tot. Inc Productive Income On-farm Basic crops 620.00 122.00 742.00 17.4 14.2 0.00 2,327.00 466.00 Animals Fowl 183.00 89.00 272.00 6.4 5.2 65.00 Â– 1,562.00 275.00 Small 65.00 65.00 130.00 3.1 2.5 0.00 Â– 1,030.00 218.00 Cattle 15.00 354.00 369.00 8.7 7.1 0.00 2,510.00 619.00 Transport 89.00 89.00 2.1 1.7 0.00 Â– 800.00 206.00 Extraction 878.00 765.00 1,643.00 38.6 31.4 38.00 4,198.00 1,021.00 Plants 145.00 9.00 155.00 3.6 3.0 0.00 1,825.00 290.00 Processing 35.00 3.00 38.00 0.9 0.7 0.00 Â– 210.00 59.00 Other 6.00 0.1 0.1 0.00 Â– 200.00 32.00 Total On-farm Inc. 1,941.00 1,496.00 3,442.00 80.9 65.8 247.00 Â– 8,188.00 1,888.00 Off-farm Unskilled labor 51.00 1.2 1.0 0.00 Â– 400.00 109.00 Skilled labor 760.00 17.9 14.5 0.00 Â– 7,260.00 1450.00 Total Off-farm Inc. 812.00 19.1 15.5 0.00 Â– 7,260.00 1,449.00 Total Productive Inc. 4,253.00 100.00 81.3 1,659.00 Â– 11,217.00 2,241.00 Social Income 950.00 18.2 0.00 Â– 6,498.00 1,729.00 Gift Income 29.00 0.6 0.00 Â– 1,185.00 175.00 Total Income 5,232.00 100.00 1,728.00 Â– 13,383.00 2,711.00 Production Costs 40.00 0.00 Â– 300.00 73.00 Net Productive Inc. 4,214.00 1,617.00 Â– 11,016.00 2,223.00 Net Total Income 5,192.00 1,728.00 13,182.00 Net Prod. Income per day labor 5.98 1.7112.21 2.5316 .
112 (and disparity) of income earned overall, and al so within each categ ory. On-farm income activities are broken down into consumption and trade. Mean household productive income, includi ng both on and off-farm activities was R$4,253.00, with a range from R$1,659.00 to R$11,217.00. The mean net productive income of R$4,214.00 was nearly identical to that noted by the ASPF study of R$4,163.00, conducted in the reserve in 1996 (C astelo 2000). Mean total income, including income earned through various reti rement programs, healthcare payments due to physical or mental handicapped househol d members, or maternity benefits was R$5,232.00, ranging from R$1,728.00 to R$13,383.00. Productive income as a percentage of total income was 81.3 %, w ith government benefits and gift income accounting for 18.7 % of total income. On-far m income provided the greatest share of productive income, accounting for 80.9 % of production. Consumption accounted for more than half of the value of on-farm productive income. At the bottom of the Table 3-9 is the figure for the re turn on labor, calculated by dividing net productive income by total labor available. For all households, the mean return on labor was approximately R$6.00, approx imately the value of a day of off-farm manual labor working in the forest, which ranged from R$5.00 to R$8.00, as reported by individuals in study households. Notable is the range of re turn on labor that households experienced, from R$1.71 to R$12.21 per day labor available: one household earned over seven times the return on labor than the lowe st return households. In comparison to the minimum wage earned by urban worker s during the same 12 month period, approximately R$2,166.00, including a 13th minimum monthly salary that employees are provided, an average rubber tapper net productive income, including consumption, of
113 R$4,214.00 compares favorably, nearly two time s the minimum salary. However, it is important to note that the work of an enti re rural household is being compared to the earnings of an indivi dual in the city. Thus, a quick examination of household income data suggests that the majority of rubber tapper income is earned on-farm. It al so finds that there is a great disparity in income across households as well as return on labor. Below, an analysis of income variation across wealth group categories suggest s that wealth holdings shape the income earning activities of households, an d that it is the high wealth households that are able to earn the greatest return on labor. However, before analyzing how income varies across wealth rank groups, I want to examine mo re closely both on and off-farm production activities, including th e sources of income for all hous eholds. As this study is particularly interested in the role of extraction in th e household economy, I will present the findings on extractive activ ities in greater detail. On-Farm Productive Income: Consumption and Trade in the Forest On-farm production activities accounted fo r 80.9 % of productive income, with offfarm activities accounting for 19.1 % of productive income. Slightly more than half, or 56.4%, of on-farm income came through consumption, with 43.5 % earned through market transactions, both barter and sale. A small fraction of productive income was related to one household renti ng pasture for grazing animals. Figure 3-5 on the following page shows the percentage of income earned fr om different on-farm production activities. Extraction provided the greates t percentage of on-farm pr oductive income, followed by
114 animal production, and basic crops. Pere nnials and other planted food crops and processed goods accounted for a very small percentage of on-farm production.36 Basic Crops 22% Extraction 48% Animals 24% Other 0% Plants 5% Processed Goods 1% Figure 3-5. Percent of rubber tapper on-farm income earne d from different production activities. Chico Mendes Extractive Reserve, 2001. Consumptive income in the forest: the important role of extraction All households carried out on-farm production for consumption. Among all households, the mean value of consum ption was R$1,941.00, ranging from R$197.00 to R$5,040. Consumption accounted for, on average, 56.4 % of income, ranging from a low of 28.0 % to a high of 100.0 %Â—one household did not sell any production, earning all its non-consumptive income from off-fa rm labor. Annual per capita household consumption averaged R$367.00, with a ra nge from R$67.00 to R$860.00. This low value is related to a household that purchase d most of its food, as both adult heads work off-farm. The second lowest figure of R$142.00 is more appropriate in considering the 36 The value of income from plan t production would have been high er if an accurate and reliable measurement could have been made. The one-shot cu ed recall method was ill suited to accurately and reliably capture the consumption of tree fruits, including citrus fruits, oranges and limes, as well as papaya and mango, and food crops, such as gourds. Thus, they have not been included in the income figure.
115 low range of per capital house hold consumption for a house hold producing its own basic food crops. Thus household income varied greatly, not only in th e value of production consumed, but also in per capita consumption. The low per capita consumption figure for a household producing its own basic cr ops is particularly startling. Figure 3-6 shows that extr action and production of ba sic food crops were the principal sources of household consump tion, although animals, plant products and processed goods also made contributions to consumptive income. Nearly all households reported consumption of extractive items, basi c crops (produced on-farm), planted items, and at least one farm animal. Chickens we re the most frequently consumed animals, consumed by 45 households: in contrast, only two households consumed cattle. Only 20 households indicated that they had produced and consumed one of three processed goodsÂ— goma , rapadura ( including alfamin , a similar sweet), or cheese.37 Although not Processed Goods 2% Animals 14% Basic Crops 32% Plants 7% Extraction 45% Figure 3-6. Percent of on-farm household cons umption income contributed by different on-farm production activities. Chico Mendes Extractive Reserve, 2001. 37 One household also produced yogurt, although this item was not included in the valuation of processed goods.
116 valued due to problems with accuracy and re liability, 29 households noted that they had collected eggs on their landholding over the past year, while 26 households produced milk for consumption. Extraction provided the larges t single contribution to consumption, accounting for 45.2 % of consumptive income. As the role of extraction in the forest economy is a principal focus of this study, I will go into so me detail here to lay out more specifically the important role that different extrac tive items play in household consumption. Table 3-10 on the following page displays the contribution that different extractive activities make to consumption and trade in come. Wild game contributed, on average, R$807.00 to household consumptive income, equi valent to 90.0 % of all consumption from extractive activities. Brazil nuts, wild fruits, artisan producti on from plant fibers, and medicinal plants made only very small c ontributions to household income. Here I briefly discuss the contribution of each of these categories. Wild game . All households reported at least a po rtion of income from game. Table 3-11 provides a list of the animals that were hunted over the past year as well as the number of animals taken and value per unit. Paca was the most frequently killed animal among households, followed by capelÃ£o (also locally referred to as guariba ), howler monkeys, and jabuti, turtles. However veado , deer, provided the gr eatest consumption value to households, to taling R$10,058.00, followed by paca at R$7,947.00, and porco da mata , collared peccaries, at R$4,741.00. All hous eholds hunted at least one animal, although the contribution of this activity to household consumption varied greatly, from
117Table 3-10. Income from extractive activit ies of 46 rubber tapper households in the Chico Mendes Extractive Reserve for 12-mont h period in 2001. Consumption (R$) Trade (R$) Total Consumption and Trade Income (R$) Range (R$) St. Deviation (R$) Mean Mean Mean Pct. of Ex t. Inc Pct. of Prod. Inc Rubber 0.00 316.00 316.00 19.2 7.4 0.00 Â– 2,350.00 431.00 Brazil nuts 42.00 425.00 467.00 28.4 11.0 0.00 Â– 1,700.00 408.00 Fruits1 12.00 1.00 13.00 0.8 0.3 0.00 Â– 59.00 13.00 Artisan 8.00 4.00 11.00 0.7 0.3 0.00 Â– 155.00 29.00 Medicinal PlantsA 1.00 1.00 0.1 0.00 0.00 Â– 7.00 2.00 Copaiba/Honey 7.50 19.00 27.00 1.6 0.6 0.00 Â– 210.00 58.00 Game1 807.00 0.00 807.00 49.2 19.0 28.00 Â– 3,187.00 696.00 Total Extraction Income 878.00 765.00 1,642.00 100.00 38.6 38.00 Â– 4,198.00 1.021.00 Note : aConsumption figure is based on focus groups with rubber tappers to value non-market extractive resources. For comparative purposes, if game consumption was valued at the black market pri ce for game meat that is purcha sed, combined with rubber tapper values for game meat for which there is no black market pric e, the value of game consumption would be R$484.00, 60.0% of the rubber tapper value.
118 R$28.00 to R$3,187.00. Game animals are for household consumption rather than for tradeÂ—the reserve management plan stipulates that trade in game is prohibited.38 Table 3-11. Number and value of game anim als and birds hunted by rubber tapper households over a 12-month period. Chic o Mendes Extractive Reserve, 2001. Common Name in Portuguese a Common Name in English Scientific Name or Family Number killed Unit Valueb (R$) Total Value (R$) Paca Paca Agouti paca 27 7 28.69 7, 947.13 CapelÃ£o/guariba Red howler monkey Alouatta spp. 182 20.00 3,640.00 JabutÃ Tortoise Geochelene spp. 162 12.50 2,025.00 NambÃº Tinamou Tinamidae 149 4.13 1,862.50 Porco da mata Collared Peccary Ta yassu tajacu 129 36.75 4.740.75 Quatipuru Red-tailed squirrel Sciurus spp. 108 2.00 216.00 Quexada White-lipped peccary Taya ssu pecari 103 38 .75 3,991.25 Veado Deer Mozana spp. 99 101.60 10,058.40 JacÃº Guan Cracidae 99 4.50 445.50 CutÃa Agouti Dasyprocta spp. 72 5.50 396.00 Aracua Chachalaca Cracidae 43 1.50 64.50 Jacamim Trumpeter Pso phiidae 42 4.00 168.00 Tatu Armadillo Dasyphus spp. 27 42.00 1,134.00 Papagaio Parrot Psittacidae 22 1.75 38.50 Titi Unknown Unknown 11 1.25 13.75 Macaco prego Capuchin Ce bus spp. 10 3.75 37.50 Tucano Toucan Ramphastidae 10 1.63 16.30 Juriti Dove Columbidae 9 0.75 6.75 Cutiari Macaw Columbidae 6 1.50 9.00 Anta Tapir Tapirus terrestris 3 511.88 1,535.64 AraÃ§ari AraÃ§ari Ramphastidae 3 1.50 4.50 Capivara Capybara Hydrochaeris hydrochaeris 1 66.64 66.64 Maracana Macaw Psittacidae 1 0.75 .75 Nambu macucal Tinamou Tinamidae 1 2.50 2.50 Total 1,569 37,120.27 Notes : a For birds, only the family has been noted. bValue based on two focus groupsÂ— total of 12 rubber tappers. Sources : Sick (1993) and Andrade ( 1985) were used to identify the families for birds. Emmons and Feer(1997) was used to identify the mammal species. Medicinal goods. Wild collected medicinal it emsÂ—primarily plant barks and leavesÂ—also contributed to household consumpti on income. Prior research in the reserve has documented the diverse group of forest plan ts used for home remedies. Kainer and 38 During the years I have conducted research in the reserve, I have witnessed only the clandestine sale of jabutÃ, a turtle with highly valued meat, both in the reserve and in the city. No households indicated that they had sold any game over the past 12 months. Even though I have known many of these households for years, I do not think they would tell me if they did sell game. In 1996, I observed one household that participated in this study sell two jabuti during a canoe trip to the city. Although it is possible that households in this study did sell game during the research period, I would argue that this is unlikely.
119 Duryea (1992) conducting research with 14 rubb er tapper women in one forest area in the Chico Mendes Reserve and nearby Cachoeir a Reserve, as well as approximately 30 additional informants, identified 145 species re presenting 60 plant families that were used for medicinal purposes. Of these 145, the au thors note that approximately 35.0 % were collected from the forest, thus women liste d approximately 51 forest plants. Ming and Amaral Junior (n.d.), in a more recent study of medicinal plant us e in the reserve found similar results. Conducting inte rviews with 53 individuals in 18 forest areas, they identified 158 species in 61 plant families used for medicinal purposes. Of these 158, approximately 65.0 % were native to the Am azon region, with approximately 53.0 %, or 84, plants collected from the forest, and 12.0 % domesticated. Approximately 35.0 % of the species recorded were not native to the Amazon. Among the 46 study households, 28 househol ds consumed a portion of income through the extraction of 23 medicinal items.39 These items included 21 native plants, the tooth of a collared peccary, and cupim , the encrusted dirt-like termite-produced termite housing material found in the forest as well as in clearings. Table 3-12 on the following page presents a list of the items, includi ng the scientific name and the number of households that indicated they extr acted the item over the past year. The 21 plant species noted by households represented a minimum of 11 plant families. The most commonly used plant for medicinal purposes was jatobÃ¡ , the bark of which was extracted by 17 study households. Fifteen households extracted the bark of copaÃba . The bark of cumaru was noted eight times, although it is possible that this plant 39 During the household interviews, 33 households noted that they had extracted and used forest plants for medicinal purposes over the past year. However, two focus groups conducted with rubber tappers found that some plants that households had noted as forest extracted were domesticated plants, grown in the Â“ quintal Â” or in the clearing near the hou se. Only forest-collected plants are included in this figure.
120 item was referred to a total of 12 times, as at least one species referred to as cumaru is also referred to as cerejeira .40 Table 3-12. Extractive medicinal items used by households in the Chico Mendes Reserve. Extractive Medicinal Item Genus and/or Species Family Part used Number of Households Unha-de-gato Uncaria spp. Rubiaceae Bark 4 Pau dÂ’arco rocho Tabebuia spp. Bignoniaceae Bark 1 PicÃ£o Unknown Unknown Leaf 1 Cerejeira Amburana cearensis (Allemao) A.C. Sm Fabaceae Bark 4 JatobÃ¡ Hymenaea courbaril L. Caesalpiniaceae Bark 17 JutaÃ Hymenaea spp. or Dialium spp. Caesalpiniaceae Bark 5 Cumaru Unknown Fabaceae Bark 8 CopaÃba Copaifera spp. Caesalpiniaceae Bark and oila 15 Cumaruzinho Unknown Unknown Bark 3 Angico Parkia spp. Mimosaceae Bark 2 Bacuri Unknown Clusiaceae Bark 1 Pimenta longa Piper spp. Piperaceae Leaf 3 CipÃ³ escada Dalbergia spp or Bauh inia spp. Caesalpiniaceae Bark 1 Tatajuba Maclura tinctoria (L.) Steud. subsp. Tinctoria Moraceae Latex, sap 1 FumaÃ§a (TucumÃ£) Unko wn Aracaceae Unknown 1 Cedro Cedrela spp. Meliaceae Bark 3 Mangirioba Cassia spp or Senna spp. Caesalpiniaceae Leaf 1 CajÃº Anacardium spp. Anacardiaceae Bark 1 Castanha Bertholletia excelsa H.B.K. Lecythidaceae Nut covering 1 Cuaite Unknown Lecythidaceae Bark 1 Tipi Petiveria spp. Phytolaccaceae Root and bark 1 Porquinho Tayassu tajacu n.a. Tooth 1 Cupim Nest n.a. n.a. Nest 3 Note : aCopaÃba oil is valued under the c opaÃba oil and honey category. Unknown indicates that species, and in some cases th e plant family, and in one case the part used, could not be identified. N.A. indicates not applicable. The bark of plants was the most co mmonly used item among forest households, although leaves, leite , or tree sap, and roots were also noted. Plant collection was generally carried out as needed, i.e., when an illness arose, rather than on a predetermined schedule. Extraction of the re quired plant material can take from a few minutes, if the 40I have noted them separately in the table, as only one household indicated the full common nameÂ— cumaru de cheiro Â—that refers to this same plant species. The common name cumaru might refer to one of three different genera for cumaru that have been collected and identified in Acre (NYBG, n.d.).
121 plant or plants are near the family dwelli ng, to a few hours, when plants are scattered across the landholding. This activity is often ca rried out during routin e work activities in the forest, such as tapping rubber, collecti ng Brazil nuts, or hunting. When collecting the plant material, the extractor may collect more than required for the pa rticular use. For example, when collecting bark, the collector might extract a chunk of tree bark that can serve for more than one use. Although one household noted that it had collected unha-de-gato on 10 separate occasions, more frequently households collec ted individual medicinal items only a few times a year. For example, jatobÃ¡ and copaÃba , the plant items most commonly used among study householdsÂ—17 and 15 households respectivelyÂ—were ex tracted by these households from one to five times, with a single extraction the most common response for both items. Notwithstanding the importance of fore st medicinal items to rubber tapper households, the contribution of these items to extractive inco me was minimal due to the low use value that rubber tappers assigned to these items during focus group interviews. Consumption of medicinal plants contri buted, on average, only R$1.00 to household income, approximately 0.1 % of extractive income. Honey and copaÃba oil ( Copaifera spp ) . Honey and copaiba oil have been separated from other medicinals as they are both consumed and sold. Nineteen households collected honey for consumption, while two households indicated that they had collected copaÃba oil for home medicinal use. That few households noted that they extracted copaÃba oil for household use was some what surprising, as it is recognized to be a natural cicatriz , a natural healer that is quick to close open sores and cuts. The
122 frequency with which I have seen open cu ts, wounds and sores over the years would argue that every household should be stocking it . It is quite possibl e that they do, having extracted the item in years past, and storing th e oil in a jar or bottle, and using it when needed. Yet, as noted, only two households indi cated that they had us ed copaÃba oil, and households were provided two opportunities to note it Â– one at the beginning of the interview when asked to list illnesses and treatments, and a second time later in the interview when they were asked specifically about copaÃba oil. As with other medicinal items, Table 3-10 reveals that honey and copaÃba oil contributed only marginally to extractive cons umption income, less than 1.0 %. Further, although 19 households earned a portion of cons umptive income from these two items, three households accounted for over half of the total consumption. Brazil nuts and tree fruits . In addition to medicinal use, extracted plant items also contributed to consumption in other importa nt ways. Brazil nuts, although principally destined for the market, were consumed by 37 households. Nuts are eaten in raw form, but the oil of the Brazil nut is also used for cooking, tenderizing and flavoring game meat. Wild tree fruits, including aÃ§aÃ ( Euterpe precatoria Mart.), bacaba ( Oenocarpus mapora H. Karst), patauÃ¡ ( Oenocarpus bataua Mart.), cajÃ¡ ( Spondias mombim L.), cajarana ( Spondias spp. ), apuruÃ ( Alibertia spp. ) and cagaÃ§a (Family Sapotaceae) were collected and used to make vinho , or fruit drinks, often cons umed with sugar and manioc flour, by 34 households. AÃ§aÃ was the most commonly extracted fruit item, with 23 households collecting aÃ§aÃ and making vinho . Among these households, aÃ§aÃ was collected from one to five times during th e fruiting season, and processed into 10 to 15
123 liters of vinho .41 PatauÃ¡ was collected and processed into vinho by 14 households and bacaba by 11 households, with the other fruits extracted and processed much less frequently. The difficulty in extrac ting the palm fruits (as opposed to cajÃ¡, cajarana , and cagaÃ§a which are collected from the ground) in hibits some households from consuming them. In particular, the fruit bunches of aÃ§aÃ and patauÃ¡ can be difficult to extractÂ— aÃ§aÃ palms are very thin and tall, while patauÃ¡ , has a very thick and rough trunk surface making them hard to climb. Households must either have an individual who can climb them, construct ladders from nearby tr ees to aid in climbing the shorter patauÃ¡ trees, or cut them down, destroying the plant and eliminating future production. Although extracted and used by many hous eholds, like medicinal items, these forest fruit items, while importa nt to individual households, particularly those that enjoy the grainy, black raspberry taste of the aÃ§aÃ fruit, or the nutty (a nd even chocolate-like) flavor of bacaba and patauÃ¡ , they contribute little in pu re economic value to household income. However, the excitement that swirls around the hous eholds as children anticipate eati ng a bowl of aÃ§aÃ filled with manioc flour as their mother prepares the vinho suggests a greater cultural value that is poorly suited to measurement by economic valuation techniques. Plant fibers for artisan goods. Lastly, plant resources are also used for producing what are locally referred to as artisan items . I would argue that these goods are more utilitarian than artisan, due to their importa nce for practical use rather than artisan 41 See Strudwick and Sobel (1988) for a comprehensive discussion of acaÃ extraction in BelÃ©m, including photographs. Although their re search was conducted on a different species of acaÃ ( Euterpe oleracea ), the process for extraction and in-forest processing is very similar. Brondizio and Siqueira (1997) provide an informative analysis of the marketing system for aÃ§aÃ in Eastern Amazon, an in particular the city of BelÃ©m, ParÃ¡.
124 quality. Seventeen households earned a por tion of income from the extraction and manufacture of brooms, baskets, sifting baskets, or r ubber shoes. The most commonly produced item was a simple broom, produced by 14 households, most often made of cipÃ³ timbÃ³ ( Thoracacarpus or Evodianthus spp. ). Other items were much less commonly produced, as they require great er skill. Only five househol ds made large strong carrying baskets constructed of cipÃ³ ambÃ© (Family Araceae ) used for hauling manioc, corn, Brazil nuts, from harvest or collection locations back to the house. Sifting baskets, woven from cipÃ³ timbÃ³ , arumÃ£ ( Ischnosiphon spp.) or fibers extracted from the palm fronds of bacaba were made by four households. Interest ingly, only two households manufactured rubber shoes, made from Hevea latex, once staple foot wear for rubber tappers. In summary, a diverse group of extractiv e items contribute to household income. But while extractive re sources made, on average, the la rgest contribution to household consumptive income, the importance of this act ivity to households varied greatly, ranging from a low of just 4.0 % to a high of 85.0 % of consumptive income. Households participate in extractive activities to vary ing degrees, depending on skills (such as climbing), the presence of resources on th eir landholdings, preferences for goods, and knowledge of medicinal remedies, among other f actors. In the following chapter, I will also argue that economic factors affect extractive resource use. Thus, rubber tapper households plant and extract a diverse group of products on their landholdings for consumptive use. Ex traction and basic food crops play the most important role in consumption, with game m eat preferred over farm animal meat. Crops such as tobacco and coffee play a small but important role within individual households, particularly where cash income is low. Less important are processed goods, such as
125 goma, sweets and dairy products. Milk, although not valued in this study, also plays an important role, providing a nutritious additi on to the household diet and eliminating the need for the purchase of expensive powdered substitutes. The value of consumption income among study households varied greatl y, as did per capita household consumption. The findings reflect the divers e livelihood strategies that households follow, including reliance on on-farm production ve rsus off-farm labor for consumption. But the findings also certainly suggest that there is great disp arity in consumptive pract ices in the reserve. Income from product trade Table 3-9 above also identifies the major sources of household trade income while Figure 3-7 below highlights the different cont ributions that product categories make to productive income. As with consumpti on, extractive activitie s made the largest contribution to trade income, accounting for just over half of trade income. Farm animal trade also played an important role, with the sa le or barter of cattle, from newborns to Extraction 51% Animals 40% Basic Crops 8% Processed Goods 0% Plants 1% Figure 3-7. Percent of househol d on-farm production income from trade. Chico Mendes Extractive Reserve, 2001.
126 mature cows and steers, making the largest c ontribution to animal trade. Cattle trade provided nearly four times the income value of the trade of fowl or transport animals. Although basic food crops played an important role in consumption, their contribution to trade income was much smaller, accounting fo r 8.1 % of trade income. Income from the trade of plant and processed food items played only a very minor role in rubber tapper trade, together accoun ting for less than 1.0 % of trade income. Trade income resulted from transactions w ith a number of different trade partners. Table 3-13 identifies the various trading partners with whom households traded. Trading partners are found both in the forest, and in urban areas. In the forest, they are other rubber tapper households, the CAEX trading post, itinerant traders and cattle buyers, some based in colony areas near Xapuri, who will travel to the forest to purchase animals. One itinerant trader, a former rubber tapper, liv es in the community of Rio Branco. He buys and sells only animals. Table 3-13. Trading partners of rubber tapper households in the Chico Mendes Reserve. In-Forest Trading Partners Urban and Other Trading Partners Rubber tapper CAEX cooperative in Xapuri CAEX trading post Merchants in Xapuri COOPERACRE (through SEFE) Households, restaurants, and other in Xapuri Itinerant traders Colony areas near Xapuri Visitors to forest Rubber tappers also sell copaÃba oil to a local cooperative, COOPERACRE, established with the assistan ce of SEFE. However, every household, when asked to whom they sold copaÃba oil, responded simply Â“SEFE,Â” never mentioning the cooperative.42 One household in the community of SÃ£o JoÃ£o de Guarani sold rapadura , a 42 The role of COOPERACRE in the sale of copaÃb a oil was brought to my attention by SEFE. See Wallace (2002). See also Kainer et al . (2003) for a brief discussion of the role of CO OPERACRE in the marketing of products in Acre.
127 hard sweet food item made from sugar cane juice, to romeiros , individuals from the city who flock to the bricks and mo rtar chapel in the community of SÃ£o JoÃ£o de Guarani at the end of June to commemorate the fe stival of the patron Saint John and Â“ pagar promessas, Â” paying their thanks by making the long hike to the community for having their prayers answered. In the city, b uyers of goods include CAEX, merchants, households and restaurants, a nd colonists in the region who purchase cattle. CAEX and merchants purchase principally rubber and Br azil nuts, but also buy agricultural goods, such as manioc flour. Households and restaurants purchase small animals as well as processed goods. Rubber tappers as a group traded a variet y of goods, including items from each of the productive income categories. Table 3-14 on the following page provides a list of these categories and the items sold under eac h. The table suggests that the rubber tapper trade economy is limited to a relatively small number of agricultural and extractive products. Only 27 products were traded among all households.43 However, the trade of goods certainly goes beyond the trade of production items, and shotguns, radios, among other consumer goods are commonly bartered and sold in the forest. Indeed, I was informed that some individuals in the fore st continually trade items, always seeking a higher valued item for a series of objects or animals. For this study, however, I have focused on those items produced on the landholding rather than the trade of consumer goods. 43 After one interview was conducted, I was informed that a household had sold a homemade hot sauce in the city of Xapuri on a recent visit trip. This has not been included in th e results. It is possible that a few other items were sold in very small quantities that were not noted in the interview.
128 The number of products traded by individua l households ranged from 0 to 11 items (mean = 5.1, mode = 7, median = 5). I have considered Â“cattleÂ” an all-encompassing item representing the different sizes and produc tion purposes (meat or dairy) of cattle. Only one household did not trade any products while two households traded 10 or more items. Forty-two households trad ed extractive items, and the sa me number traded at least one farm animal as well. Approximately half of the study households traded basic crops. Only a small number of households trad ed plant products or processed goods. Table 3-14. Agricultural and extractive trade it ems and number of households involved in the trade of product. Chico Mendes Extractive Reserve, 2001. Basic Crops (23) Animals (42) Extraction (42) Plants (5) Processed Items (2) Rice (11) Cattle (17) Rubber (3 3) Tobacco (3) Rapadura (1) Beans (10) Horse (7) Brazil nuts (36) Pineapples (1) Goma (2) Corn (8) Mule (3) CopaÃba oil (6) Bananas (1) Manioc flour (13) Chickens (33) AÃ§aÃ juice (2) Limes (2) Capote (1) Honey (2) JambÃº (2) Duck (19) Brooms (4) Pig (11) Sifting baskets (2) Goat (2) Sheep (3) Brazil nuts were the most commonly tr aded item among households, sold by 36 households, followed closely by rubber a nd chickens, both traded by 33 households. Seventeen households traded cattle. The table does not list coffee, as no households sold this item. However, three households were hol ding fairly sizeable st ocks of coffee (from 400 to 720 kilograms), which they intended to sell in the near future. All were hoping that the selling price would in crease in the coming months. The low number of items traded by households reflects the importance that a few items hold as cash (or barter) income earners. Table 3-15 be low lists nine products that played a key role in providing trade income. Clearly Brazil nuts and rubber continue to play an important role in trade income. Ni ne households gained 50.0 % or more of trade
129 income from Brazil nuts, and the same was tr ue for rubber. Five households gained 50.0 % or more of trade income from cattle, while two households earned more than half of their trade income fr om selling chickens. One househol d earned 100.0 % of trade income from the sale of a single mule, while anot her household earned 96.0 % of household trade income from the sale of a horse. Both of these households relied heavily on salaried income with one member of each household holding a position as a professor; each sold only one animal accounting for all, or nearly all, of trade income. Thus, 28 of 46 households earned 50.0 % or more of househol d trade income from only one product, demonstrating the high dependency househol ds have on individual products. The continuing importance of extraction in gene rating trade income is also clear: 18 households earned at least 50.0 % of trade income from either rubber or Brazil nuts. In contrast, five households earned at least half of trade income from cattle, three from the sale of small farm animals, and two from the sale of transport animal s. Further, although not noted in Table 3-15, 28 hous eholds earned over 50.0 % of trade income from the trade of extractive products, and 39 households earned at least 25.0 % of trade income from extraction. Table 3-15. Principal trade income sources as a percent of trade income for 46 rubber tapper households. Chico Mendes Extractive Reserve, 2001. Percent of Trade Income (Number of Households) Product 25.0 Â– 49.0 % 50.0 Â– 74.0 % 75.0 Â– 100.0 % Total Brazil nuts 12 5 4 21 Rubber 11 5 4 20 Cattle 7 5 0 12 Horse 1 0 1 2 Mule 0 0 1 1 Manioc flour 3 0 0 3 Tobacco 3 0 0 3 Chicken 1 2 0 3 Pig 1 0 0 1 Total 37/46 17/46 11/46
130 Yet, the above analysis presents a some what distorted view of household reliance on the sale of a few products, as many house holds do not rely prin cipally on trade to obtain income to purchase goods. Of the 46 study households, only 26 households earned the largest share of nonconsumptive incomeÂ—income from trade, off-farm labor, or social benefitsÂ—from trade. Twelve hous eholds earned the greate st share of incoming cash through social benefits, while eight relied principally on off-farm labor. Looking only at the 26 households who rely primarily on trade for generating cash, all earned 50.0 % or more of their trade in come from one or two products. Sixteen households earned 50.0 % of more of their trad e income from extract ive activities, and 15 of these earned 50.0 % or more of trade income from rubber and/or Brazil nuts. Thus, 15 households, one-third of all study households, still rely principally on rubber and Brazil nut trade to obtain cash. This argues that rubber and Brazil nuts remain extremely important to many households. However, it also suggests that many households have identified other income sources to obtain cash income for maintaining the household. The role of extraction in trade activities As extractive activities made the largest contribution to trade income I want to briefly consider the different sources of extractive income. In addition, a closer look at trade patterns for these two items reveals th e continuing role of ba rter in the forest. Returning to Table 3-10 above, Brazi l nuts and rubber account for a great majorityÂ—nearly 97.0 %Â—of trad e income from extraction. Th e sale of copaiba oil and honey made modest contributions to trade income, as did the production of artisan itemsÂ—brooms and sifting baskets made from forest fibersÂ—and aÃ§aÃ juice. Yet, very few households engaged in the sale of extract ive items other than Brazil nuts and rubber;
131 six households sold copaÃba oil, three hous eholds sold honey, four households sold artisan objects, and two sold aÃ§aÃ juice. So, although market-o riented extraction plays an important role in household income, few house holds have diversified extraction for the market beyond rubber and Brazil nuts. Rubber and Brazil nuts are traded with five principal trading partners: CAEX trading posts in the forest, CAEX in Xapuri, merchants in Xapuri, itinerant traders, and other rubber tapper households in the forest. Each of these partners, including in one case, the rubber tapper household in the fo rest, provides the house hold not only with an outlet for their production, but also serves as an important supplier of basic household goods. Of the 37 households that sold Brazil nuts, 22 received a portion of the value in supplies, while 11 applied a portion of the value to debt held with the trading partner. Approximately 35.0 % of the value of Brazil nut transactions was r eceived in supplies rather than cash. The findings were simila r for rubber. Of the 33 households that sold rubber, 22 households received a portion of the transaction value in goods, and 14 applied a portion of the value to debt. Nearly 45.0 % of the value of th e sale of rubber was received in supplies. Hence, the barter system, if not the aviamento system described in Chapter 1, remains strong in the forest, even as trading part ners have changed. Table 3-16 and the accompanying Figure 3-8 display the trade patterns for rubber among the 33 households that sold this product. I have chosen to focus on rubber, rather than Brazil nuts (although a si milar pattern was found) as the findings demonstrate the impact that the State rubber subsidy, disc ussed in Chapter 2, has played in shaping trading patterns in the forest. Table 3-16 and Figure 3-8 both show that in 2001, households traded the majority of rubber to the CAEX in-forest trading posts in the
132 communities of Rio Branco and SÃ£o JoÃ£o de Guarani, with the CAEX office in Xapuri the second most used trade outlet. As not ed above, to obtain th e state rubber subsidy payment of R$0.40 (rising to R$0.60 in Octobe r 2002), households must be associated with AMOREX or CAEX, and sell their pr oduct through CAEX, e ither its in-forest trading post or at th eir Xapuri operations. Table 3-16. Trading patterns for rubber by 33 households in the Chico Mendes Extractive Reserve, 2001. Trade Partner Householdsa Value (R$) Kilos Pct. of Rubber Sold Average Price per Kilo CAEX Trading Post in Forest 24 8,530.00 7,180 59.3 1.19 CAEX in Xapuri 13 5,063.00 4,009 33.1 1.26 Urban Merchant in Xapuri 5 533.00 533 4.4 1.00 Itinerant Trader in Forest 2 152.00 140 1.2 1.09 Rubber Tapper in Forest 2 260.00 260 2.1 1.00 Total 14,537.00 12,122 100.0 Note : aA few households traded with more than one partner. CAEX Trading Post 60% CAEX in Xapuri 33% Rubber Tapper 2% Itinerant Trader 1% Urban Merchant 4% Figure 3-8. Location and percent of r ubber production traded by rubber tapper households. Chico Mendes Extractive Reserve, 2001. Twenty-four households sold rubber to one of the two in-forest trading posts (there are CAEX trading posts in other forest areas as well) while 13 househol ds sold directly to CAEX in Xapuri, where they obtained a sli ghtly higher price per kilo sold. A few
133 households sold small amounts of rubber to itin erant traders, urban merchants or other rubber tappers in the forest. These trade channe ls are used as a mean s to obtain supplies or cash quickly. However, in the case of one transaction between two rubber tapper households, the seller did not have transportation to deliver hi s rubber to the side of the feeder road or to the CAEX tr ading post. He sold his rubbe r to another ru bber tapper for a slightly lower value than the full subsidy pric e. Interestingly, he did not receive the full value of this transaction in cash, but was provided supplies for half the value of the transaction. The household that purch ased the rubber (not among the 46 study households) viewed this as a favor to the other household rather than a money making business venture. The effects of the subsidy on rubber trading patterns are more clearly identified in Figure 3-9. This figure compar es the trade patterns for rubber sales of the same 28 households in the years 1996 and 2001. It s hows that in 1996, these households sold principally to urban merchants and itiner ant traders in the forest. With the implementation of the State rubber subsidy in 1999, a dramatic shift in trading partners took place; in 2001, these same households sold nearly all of their rubber production to the CAEX trading posts and to CAEX operations in Xapuri. Yet, while trading patterns have cha nged, the impact of the subsidy on rubber production was quite different. Although households received a higher price for rubber in 2001, and thus a higher return for their labor , fewer of these 28 households collected rubber in 2001. In 1996, 24 households sold r ubber, while in 2001, only 18 sold rubber. Of the four households among the 28 in 1996 th at did not tap rubber, none returned to rubber tapping in 2001. Further, rubber production among these households fell
134 dramatically from 1996 to 2001 as demonstrated in Figure 3-9. In 1996, total production among these 28 households was 11,697 kilos. In 2001, rubber sales fell to 6,877 kilos, a 41.2 % drop in production. Only four househol ds produced more rubber for sale in 2001 than was recorded in 1996. 0 1000 2000 3000 4000 5000 6000 CAEX Trading Post CAEX in Xapuri Urban Merchant Itinerant Trader Rubber Tapper Trading PartnerKilos of Rubber Traded 1996 2001 Figure 3-9. Comparison of trade patterns for rubber (kilos) for 28 households in the Chico Mendes Extractive Reserve in 1996 and 2001. Thus, the rubber subsidy has helped househol ds gain a higher price per kilo for rubber. And, concomitantly it has helped CAEX purchase a grea ter share of rubber production in the region, thus increasing its negotiating abilities with outside buyers. Yet, among households in this study, the r ubber subsidy is not stimulating production, nor encouraging households to return to ta p rubber. Indeed, the number of households who collected and sold rubber fell over this five-year period, as did the quantity of rubber they produced. The above discussion on trade income argues that extraction maintains an important role in forest trade. Extractive ac tivities, in particular the collecting of Brazil nuts and tapping of rubber, provide over half of the cash income to 15 of 26 households
135 that rely principally on trade for cash in come. Over 70.0 % of households tap rubber while over 75.0 % collect Brazil nuts, attesting to the prevalence of th ese activities. The barter system remains very strong in the re serve, with many hous eholds trading rubber and Brazil nuts for supplies and paying de bt, rather then receiving cash. The implementation of the State rubber subsidy has helped households earn a greater return for their labor, eliminating the middlemen that many households traded with only a few years past. Yet, the data also suggest that the extractive tr ade economy is highly specialized, limited to only a few products, and that rubber and Br azil nuts dominate trade. Further, the analysis of trade income indicates that cattle trade is also an important income earner for rubber tapper households. Se venteen households ea rned income from the sale of cattle, and five earned more that half of trade income from cattle production. The following chapter will examine how income varies across wealth categories, and consider the effects of wealth and inte gration into product markets on extractive activities. First, however, I want to exam ine the third and final contributor to rubber tapper household productive incomeÂ—the role of off-farm labor in the rubber tapper economy. Off-farm Labor Income Although off-farm income from wage labor and salaried positions accounted for a lower percentage of household productive in come than on-farm activities, it is an important source of income for many forest households . Among study households, productive income from off-farm activ ities ranged from R$0.00 to R$7,260.00, and accounted for 19.1 % of total produc tive income. A total of 33 households earned at least part of total productive income from off-fa rm labor activities. Of these 33 households,
136 off-farm labor accounted for nearly 40.0 % of all trade and off-farm income, or earned cash income, which comes into the household. Table 3-17 lists the different salaried and wage labor activities undertaken by study households. Ten households received at least a portion of total income from a salaried position while 26 households partic ipated in wage labor activit ies. Men were the primary earners of salaries, with only two women ear ning a fixed monthly salary for all or a portion of the year, one as a teacher and one as a health agent. Salaried employees of the governmentÂ—teachers and health agentsÂ—als o benefit from an extra, or Â“13thÂ”, monthly salary that comes as a benefit of empl oyment, further increasing household income.44 Table 3-17. List of off-farm skilled and uns killed labor activities. (The number of households participating in act ivity is in parentheses). Off-farm skilled labor activities Off-farm unskilled labor activities Chainsaw operator (3) Laborer on other landholding (6) Carpentry (3) Laborer on nearby ranch (2) Teacher (5)a Laborer in the city (3) Community health agent (2) a Laborer to clear river (1) Agroforestry extensionist (paraflorestal) (1) Transport animals (1) Trading post manager (2) a Brazil nut collector/cracker (3) Cooperative factory manager (2) a Operate rice thresher (1) Association administration (2) b Other labor (4) Association Â– other (2) Municipal government (1) Brazil nut factory worker (nut sheller) (5) Forest Guide (1) River transport (1) SEFE extensionist (1) Nanny (1) Notes : aPosition is salary based alth ough the number of months of pay for the year varies. b One position at AMOREX was based on a monthly salary for only 3 months. Of the 26 households that earned at least a portion of income fr om wage labor (not salaried labor), 18 earned wage s from skilled labor, while 13 carried out unskilled wage 44 Teachers who work in the program for adult literacy are not contracted for th e year and thus do not receive the 13th salary that other teachers receive.
137 labor. Five households earned income from bot h skilled and unskilled wage labor. Men were the primary wage earners among study households, although females also earned wage labor income. Wage labor among females was limited to four women who earned income working at the in-forest Brazil nut pr ocessing factory located in the community of Rio Branco. For 13 households, off-farm labor (either sa laried or wage) accounted for 50.0 % or more of total trade and off-farm income. In other words, 13 households earned at least half of their cash income (excluding social benefits) from off-farm labor. Of the 23 households that earned income from skilled la bor (either salary or wage), skilled labor contributed on average 31.0 % of productive income and accounted for nearly 50.0% of earned cash entering the household. One hous ehold earned 94.2 % of productive income from off-farm labor, while another earned 86.5 %. For two households, income from skilled labor accounted for al l of the householdÂ’s trade a nd off-farm activities. In contrast, off-farm income from unskille d labor accounted for only 4.0 % of productive income. These figures suggest that entering skilled labor markets, either salaried or wage, is likely to have a greater impact on on-farm production activities. Thus, income earned off-farm, principally in the forest, but also in the city, whether from a salaried position or from wage labor , played an important income role in the reserve, and households undertook a varied lis t of activities to br ing cash into the household. For salaried workers, off-farm la bor provided a steady flow of cash. Indeed, for two households, participation in off-farm salaried labor, combined in one case with skilled wage labor, resulted in almost no in come from on-farm production or extraction. This is less true for wage labor, but the skil led wage labor market did provide households
138 substantial earnings. Unskilled wage labor pl ayed only a supplemental role in productive income, serving as an opportunity to bring qui ck cash into the household rather than as a major source of income around which on-farm production is planned. Conclusion In this chapter I have explored in detail the rubber tapper household economy, providing an in-depth an alysis of rubber tapper wealth holdings and income. In doing so, I have argued that rubber ta ppers are not a financially or economically homogenous group, but rather hold wealth in diverse asse ts, varying as wealth increases, and earn income from a variety of sources, both th rough on-farm production and off-farm labor. I introduced this chapter with brie f demographic portrait of the 46 study households. The data showed that households vary in size as well as stage in their life cycles and the time that household heads ha ve lived on their landholdings. However, most household heads grew up in the forest areas near where they now live, with half of male and female household heads indicating that they lived in the same forest area prior to their current residence, while two-thirds noted that they came from rural Xapuri. Family ties in the forest are strong: almost all households, 85.0 %, have another family member living in the same forest area wh ere they reside, while 41.0 % have another family member living in another house on the same landholding. A moderately strong patrilineal relationship was found among hous eholds, with 15 landholdings exhibiting this father-son relationship. St rong social ties complement fa mily ties: a great majority (42 of 46) of households are also a member of at least one of three social organizations that support reserve households, or rural familie s in general. The data also showed that households are investing in education, with all but two households having at least one literate member. Nineteen households have at least one member that has completed the
139 fourth grade, the highest grade achievable in the forest, suggesting a demand for a higher level of educational services in the forest. A brief discussion of land use among hous eholds indicated th at households produce basic food crops mainly for subsistence, with the area for this activ ity normally one or two hectares, and also maintain small areas fo r the planting of fruit trees for consumption. Notable is the area for pasture. Although th e average pasture area held was 4.3 hectares, the range of zero to 19.2 hectares suggests th at rubber tapper house holds have different on-farm production strategies. This variation was documented in the analysis of wealth holdings and income. An in-depth examination of rubber tapper wealth holdings demonstrated that study households hold their wealth in diverse ways, and as wealth increases , households invest their wealth in different ways. As househol d net wealth per capita increases, households shift investments into on-farm non-productive assets, such as consumer goods, but also into urban assets, including land, homes and co nsumer goods in the city of Xapuri. Thus on-farm productive wealth as a proportion of to tal wealth actually declines as net wealth per capita increases. However, investments in on-farm productive asse ts shift as wealth increases, with increased investments in cat tle, and landholding stru ctures that support this activity. This indicate s that as rubber ta pper household wealth increases, they undertake more destructive la nd-use practices, and indeed, a moderately strong positive correlation was found between net wealth per capita and pasture area. The diverse ways that rubbe r tapper households invest capital on their landholdings suggests that their production ac tivities vary, and in the seco nd half of this chapter I examined in detail the various on and offfarm production activitie s that rubber tappers
140 carry out to earn income. On-farm, producti on is dominated by extraction, growing basic food crops and raising both small and large fa rm animals. For all households, production is destined for both consumption and trade: every household engaged in some type of market activity. The extraction of non-timber fo rest resources provide d the greatest share of both consumptive and market-oriented inco me; basic food crops provided the second largest contribution to consumption while fa rm animal trade made the second largest contribution to trade income. The value of extraction for consumption consisted largely of eating wild game. Although most house holds consume a variety of forest goods, including medicinal plants, tree fruits, and forest fibers for making artisan items, in general, these items are not highly valued by the market-economy, nor, in the case of medicinal plants, by rubber tappers themselves, at least economically. Market-oriented extraction is dominated by the trade of rubber and Brazil nuts. Thirty-three households extrac ted and sold rubber and 36 hou seholds collected and sold Brazil nuts. Eighteen households earned over ha lf of their income either from Brazil nuts or rubber. The barter role these two items continue to play suggests that an evolving forest economy still values services that trading partners provide as well as better prices. In contrast to extraction, onl y seventeen households sold ca ttle, and only five households earned over half their income from this activity. Thus data argue that on-farm production is largely dominated by extraction. However, despite the importance of this activity to many rubber tapper households, the State rubber subsidy, while bringing higher returns to those tapping rubber, is not initiating a retu rn to rubber extraction. On the contrary, a comparison of income from rubber tapping from 1996 finds that income from rubber tapping has fallen and production ha s dropped despite higher prices.
141 The analysis of off-farm income earning act ivities demonstrates that salaried labor and skilled wage labor contribu te substantially to household in come. This is less true for unskilled wage labor that contributes little to overall productive income although I would argue that even this small cash infusion is nonetheless important for those households that earn it. Maybe more important is the general finding that 33 households undertook some type of off-farm labor activity. Thus the question em erges regarding how integration into labor markets might influe nce productive activities . In the following chapter, I respond to this ques tion in examining the role of both wealth and integration into market on rubber tapper income.
142 CHAPTER 4 WEALTH, MARKETS, INCOME AND EXTRACTION In the previous chapter, I provided both a detailed description and analysis of the wealth holdings and income earning activities of rubber tapper households. The analysis of household wealth demonstrated that rubbe r tapper wealth holdings change as wealth increases, with increased investments in animal s, particularly cattle, but also a shift in investments into consumer goods and urban assets, including land, houses and consumer items. An examination of income earni ng activities of rubber tapper households has demonstrated that extraction plays an impor tant role in the household economy, through both consumption and trade: extraction, among all households, makes the largest single contribution to both consumption and trade. Ye t, even as extraction retains an important role, many households are entering into labor markets, and off-farm income earnings are considerable, and are clearly shaping on-farm pr oduction decisions. This leads us back to the principal questions asked by this study: how do wealth and integration into product and labor markets, as well as access to ur ban markets, affect rubber tapper household income, and in particular, income from extractive activities? In this chapter, I return to the hypothese s that have directed this study and bring together the discussion on rubber tapper wealt h, trade and labor. To anticipate my main findings, I found that an increase in household wealth has statistically significant positive effects on net household income and net producti ve income. I also found that wealth had a positive effect on extractive income, alt hough this relationship wa s not significant. However, I found that wealth has a negative and statistically signi ficant effect on the
143 percent of income earned from extraction, in dicating that although extraction income may increase as wealth grows, it contributes less as a percent to productive income. Yet, this relationship was very weak, indicating that ev en the wealthiest hous eholds rely heavily on the forest for income. In examining the effects of integration in to the market economy and market access on income, I found a statistically significant a nd positive relationship between travel time to the city and net income, opposite of that hypothesized, indicating that as market access improved, income fell. I also found a statis tically significant and negative relationship between integration into off-farm labor and pr oduct markets as well as travel time to the city and extractive income. The finding for travel time is opposite of that hypothesized, indicating that households with poorer market access earn less inco me from extractive resources. Only integration into off-farm la bor markets had a statis tically significant and negative affect on the percent of productive income earned fr om extraction. A simulation of the effects of labor markets on extraction demonstrates that the most negative effects on extraction income are found among households that undertake skil led labor activities (salaried or wage) rather than occasional unskilled wage labor jobs. Yet, these households did not maintain more land in past ure or hold greater cattle or animal wealth, suggesting that while integration into la bor markets may reduce on-farm extraction, it does not lead to more destructive land use practices. To begin the analysis, I briefly review th e six hypotheses stated in Chapter 1 that test the effects of wealth and markets on in come. I then build on the discussion in the previous chapter to examine how household in come varies across th e four wealth rank
144 categories. This leads to the final section of the chapter, where I present the results of the multivariate regression analysis that tests the six hypotheses. Statement of Hypotheses This study examines factors that may be influencing household income earning activities, and in particular, income fro m extractive activities (destined for both consumption and the market). More specifically, as introduced in Chapter 1, I am interested in how two specifi c variablesÂ—wealth and integration into market activitiesÂ— may be driving changes in household ex traction of non-timber forest products. My own observations through prior resear ch in the Chico Mendes Reserve found that there are considerable wealth differences among resident rubber tapper households. Previous studies have noted that families that have greater household wealth are able to invest both capital and labor in riskier, high er return activities, leading to higher incomes (Dercon 1997, Reardon and Vo sti 1995). Families poorly en dowed in wealth might be expected to invest in low risk, low return activities to insure against unforeseen income shocks and smooth their income stream, lead ing to lower income. Therefore, my first hypothesis (H1) states, as household wealth increases, household income will rise. Building on the H1, households that have great er household wealth are able to invest in riskier, higher return activitie s. For these households, the opportunity cost of employing household labor in low return or subsistence ac tivities will increase, a nd therefore they will leave these activities to invest in higher return activities. Conversely, wealth-poor families will invest labor in activities that require lower capital investment, such as extractive activities. Therefore, my second hypothesis (H2) states, as ho usehold wealth increases, the value of household net productive income (incl uding consumption) from non-timber forest resource extraction will fall.
145 Two previous studies in Asia (Gunatille ke, Senaratne, and Abeygunaward ena 1993, Appasamy 1993) found that ex traction of plants (including fuelwood) accounted for a greater share of household in come among poor families. Familie s that have greater wealth, and therefore the ability to invest capital and labor in riskier inco me generating activities (that may require increased capita l investment for entry), will not only receive less income from extractive activities, but the proporti on of total income from extraction should decrease. Therefore, my third hypothesis (H 3) states, as househol d wealth increases, household income (including consumption) from non-timber forest resource extraction as a proportion of household produc tive income will fall. Recent studies have examined the marketÂ’s effects on tropical lands and people, such as deforestation (Bedoya Garland 1996, Go doy 2001, Godoy, Wilkie and Franks 1997), health (Byron 2003, Godoy and CÃ¡rdenas 2000) and indi genous knowledge of plants (Reyes-Garcia 2001, Godoy, Brokaw and Wilkie 1998). I measured market integration as: 1) proportion of net productive income earne d through off-farm labor; 2) proportion of net productive income earned through cash sales or barter of extractive or agricultural products; and, 3) travel time from the househol d to the city of Xapuri, the closest major market center. Participation in the market for wage la bor also may influenc e on-farm productive activities (Godoy 2001). My previous research in the reserve revealed that many rubber tapper households supple ment income from extractive activ ities through off-farm salaried and wage labor activities, contract ed both in Xapuri and in-fores t. As off-farm wage labor opportunities appear, the opportunity cost of low return on-farm labor activities increases. Income from cash sales and barter from agricu ltural and extractive products also provides a
146 meaningful measure of market integration (Godoy, Wilkie an d Franks 1997, Godoy 2001). My previous research in the reserve found that households in the reserve sell and barter a variety of extractive and agricultural products wi th diverse trading partners, located both in the forest and in the city of Xapuri, and households vary in terms of their integration into product markets. Finally, househ olds living closer to the city of Xapuri will have greater access to markets and therefore should have greater ease in selling products. They may also have lower costs to reach off-farm labor opportunities (Murphy, Bilsborrow and PinchÃ³n 1997). Thus, my fourth hypothesis (H4) states, that for each indicator of market integration, as market integration increases , household income and productive income will rise. Building on this hypothesis, the same ar gument is made for income from nontimber forest products. As income from wa ge labor increases, households will have fewer resources to allocate toward traditional extractive activities and thus income from this activity should fall. As income from cash sales and barter increases, they will leave less risky but low return extractive activities. And finally, as travel time to the city decreases, families will have greater acce ss to markets, which may facilitate and encourage diversification of market-oriente d activities, and therefore decrease income from non-timber forest resource activities. Therefore, my fifth hypot hesis (H5) states, as market integration increases, household income (including consumption) from nontimber forest resource extraction will fall. Further, following this same line of argument, as market integration increases, not only w ill income from non-timber forest resources fall, but the proportion of hous ehold productive income (inc luding consumption) from this activity will also fall. Thus my sixth hypothesis (H6) states, as market integration
147 increases, household income from non-timbe r forest resources, as a proportion of household productive income, will decrease. Having presented the six hypotheses to be te sted, I now turn to present the findings of my analysis of how household income varies across the four wealth rank categories. I conclude with the results and discussion of the statistical analysis of hypotheses H1 through H6. Rubber Tapper Income and Wealth Tables 4-1 and 4-2 on the following two pages summarize the income sources of rubber tapper households categorized by net wealth per capita rank as noted above. Figure 4-1, which follows Tables 4-1 and 4-2, di splays the percent of income earned from different production categories by wealth ra nk. The percent of income from on-farm activities was higher fo r wealth rank 1 than the three w ealthier categories. Over 90.0 % of productive income of the poorest wealth group came from on-farm production, with the other three wea lthier groups ranging from 75.6 % to 80.4 %. Thus wealthier households are gaining a greate r share of productive income from off-farm activities. Looking more specifically at the contribu tion of different categories of productive income, animal income makes a smaller cont ribution, only 11.8 %, to productive income among wealth rank 1 households, compared to the three higher wealth rank groups. Wealth rank groups 2, 3, and 4, at 18.4% , 17.4%, and 29.1%, respectively, earned a greater proportion of income from farm animal production. Regarding income from cattle trade, no households in the lowest w ealth rank category earned income from the sale of cattle (as noted earlie r in the chapter, they did not own any cattle either), while cattle production accounted for 14.7 % of productive income of the wealthiest households.
148Table 4-1. 12-month income summary for r ubber tapper households ranked by net wealth per capita. Chico Mendes Extractive Reserve, 2001. Wealth Rank 1 NWPC Less than R$500 Wealth Rank 2 NWPC R$500 to R$999 Consumption Trade Total (R$) Pct. of Productive Income Consumption Trade Total (R$) Pct. of Productive Income Productive Income On-farm Basic crops 509.00 86.00 595.00 19.2 638.00 62.00 700.00 15.7 Animals 164.00 202.00 366.00 11.8 330.00 491.00 821.00 18.4 Fowl 125.00 42.00 167.00 5.4 228.00 36.00 264.00 5.9 Small 39.00 91.00 131.00 4.2 59.00 5.00 65.00 1.5 Cattle 0.00 0.00 0.00 0.0 43.00 366.00 408.00 9.2 Transport 68.00 68.00 2.2 84.00 84.00 1.9 Extraction 853.00 725.00 1,578.00 50.8 952.00 876.00 1,828.00 41.0 Plants 276.00 10.00 286.00 9.2 103.00 3.00 106.00 2.4 Processing 15.00 0.00 15.00 0.5 44.00 6.00 51.00 1.1 Other 0.00 0.00 0.00 0.00 0.00 0.0 Total On-farm Inc. 1,817.00 1,023.00 2,840.00 91.4 2,067.00 1,438.00 3,505.00 78.8 Off-farm Unskilled labor 54.00 1.7 63.00 1.4 Skilled labor 213.00 6.9 882.00 19.8 Total Off-farm Inc. 267.00 8.6 945.00 21.2 Total Productive Inc. 3,107.00 100.00 4,450.00 100.00 Social Income 132.00 1,818.00 Gift Income 112.00 0.00 Total Income 3,351.00 5,624.00 Production Costs 30.00 38.00 Net Productive Inc. 3,077.00 4,412.00 Net Income 3,321.00 5,586.00 Net Prod. Income per day labor 5.30 5.17 Net Income per day labor 5.90 6.42
149Table 4-2. 12-month income summary for r ubber tapper households ranked by net wea lth per capita. Chico Mendes Extractive Reserve, 2001. Wealth Rank 3 NWPC R$1,000 to R$1,999 Wealth Rank 4 NWPC R$2,000 or greater Consumption Trade Total (R$) Pct. of Productive Income Consumption Trade Total (R$) Pct. of Productive Income Productive Income On-farm Basic crops 573.00 218.00 791.00 19.0 739.00 174.00 914.00 17.6 Animals 220.00 504.00 724.00 17.4 295.00 1,213.00 1,508.00 29.1 Fowl 183.00 119.00 301.00 7.2 175.00 191.00 366.00 7.0 Small 37.00 100.00 137.00 3.3 119.00 99.00 219.00 4.2 Cattle 0.00 261.00 261.00 6.3 0.00 760.00 760.00 14.7 Transport 25.00 25.00 0.6 164.00 3.2 Extraction 697.00 732.00 1,429.00 34.3 926.00 667.00 1,592.00 30.7 Plants 126.00 2.00 128.00 3.1 89.00 23.00 112.00 2.2 Processing 47.00 0.00 47.00 1.1 32.00 3.00 35.00 0.7 Other 25.00 0.6 7.00 0.1 Total On-farm Inc. 1,663.00 1,456.00 3,144.00 75.6 2,080.00 2,081.00 4,167.00 80.4 Off-farm Unskilled labor 81.00 1.9 11.00 0.2 Skilled labor 936. 00 22.5 1,003.00 19.4 Total Off-farm Inc. 1,017.00 24.4 1,014.00 19.6 Total Productive Inc. 4,161.00 100.00 5,181.00 100.00 Social Income 0.00 2,133.00 Gift Income 14.00 0.00 Total Income 4,175.00 7,315.00 Production Costs 42.00 88.00 Net Productive Income 4,119.00 5,093.00 Net Income 4,133.00 7,227.00 Net Prod. Income per day labor 7.00 7.12 Net Income per day labor 7.12 11.41
150 Tables 4-1 and 4-2 also display the return the different wealth groups received on labor. Wealth rank groups 3 and 4 received more than R$1.70 more per day labor available, an approximate 33.0 % greater retu rn per day labor, than wealth rank groups 1 and 2, arguing that households with greater wealth are able to employ their labor resources with greater efficiency. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Wealth 1Wealth 2Wealth 3Wealth 4 Wealth RankPercent of Productive Income Off-farm Other on-farm Processed goods Plants Extraction Farm Animals Basic crops Figure 4-1. Percent of rubber tapper household productive income from on and off-farm activities by wealth rank category, Ch ico Mendes Extractive Reserve, 2001. Income from extraction made the larges t categorical contribution to productive income among all wealth rank groups, although the proportion of income from extraction fell from just over 50.0 % of productive in come for wealth rank 1 households to approximately 30.0 % for wealth rank 4 househol ds. Interestingly, and clearly shown in Tables 4-1 and 4-2, the Real value of extractive activitie s, both for consumption and trade, remained relatively st able across the four wealth rank groups. Figure 4-2 on the following page details the s ources of extractive income across the four wealth rank groups. Game consumption contributes the gr eatest value to extr active income for all wealth rank categories, although gameÂ’s propo rtional contribution to productive income
151 falls across the wealth groups. Brazil nuts a nd rubber are clearly important to productive income for all wealth rank groups. However, the roles of these tw o products diverge as wealth holdings grow. Brazil nuts remain a strong contributor to income across wealth categories while the contribution of rubber to productive income fa lls in wealth rank groups 3 and 4. As discussed earlier in the ch apter, extractive activit ies other than game, Brazil nuts and rubber contribu ted very limited value to extr active income, and this held true for all wealth rank groups. 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Wealth 1Wealth 2Wealth 3Wealth 4 Wealth Rank Income from Extraction (R$) (Consumption and Trade) Game Copaiba oil/Honey Medicinal Plants Artisan Fruits Brazil nuts Rubber Figure 4-2. Value of extraction from cons umption and trade by wealth rank group for rubber tapper households. Chico Me ndes Extractive Reserve, 2001. Figure 4-3 displays the percentage contribution of different extractive activities to extractive income. This figure mirrors the findings displayed in Figure 4-2. Game accounts for the greatest percentage of inco me from extraction, followed by Brazil nuts and rubber. The importance of game a nd Brazil nuts to prod uctive income and concomitantly, the decreasing importance of r ubber, as wealth increases, are clearly displayed in this figure. Although not noted in Figure 4-3, also notable is the fall in importance of extractive activit ies to trade income across the wealth rank groups. As
152 noted in the previous chapter, income from trade in extractive products provides at least 50.0 % or more of trade income to 28 study hous eholds. However, this includes only six of 19 households in the two highest wealth rank groups. Further, nine of the 11 households in the greatest wealth category are found among the 18 households that earn less than 50.0 % of trade income from extraction. This sugge sts that while extraction is important across wealth categorie s, it does not maintain the same level of importance to trade income. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Wealth 1Wealth 2Wealth 3Wealth 4 Wealth RankPercent of Income from Extraction (R$) Game Copaiba oil/Honey Medicinal Plants Artisan Fruits Brazil nuts Rubber Figure 4-3. Percentage contri bution of extractive activities to rubbe r tapper household extractive income by wealth gr oup rank. Chico Mendes Reserve, 2001. Thus, the above discussion suggests that as household wealth increases, productive activities on-farm, and more specifically, extractive activities are also undergoing changes. To analyze how wealth may be reshaping productive income, as well as how integration into product and labor markets may be redirecting production activities, I conclude the chapter by employing a statisti cal model to measure the strength and significance of the rela tionship between these economic variables.
153 The Effects of Wealth and Market Integration on Household Income To measure the effects of wealth and mark et integration on different measures of household income, I included the key test variab les, as defined in Chapter 2, as well as a number of control variables. Table 4-3 disp lays the dependent and independent variables included in the four multivariate regression m odels. For the regressions, the measure of integration into labor marketsÂ—percent of productive income earned from off-farm laborÂ—has been broken into three dummy variables. This was necessary, as the dependent variable in model 4, percent of income form extraction (piext), must, by definition, fall to 0.0 % when the continuous va riable, percent of labor from off-farm labor, increases to 100.0 %. Th e use of dummy variables elim inates this problem. The dummy variables also allow for a non-linear relationship between off-farm income and the four different measures of income and extractive income. The models can be found in Appendix A. Table 4-3. Multivariate regression model variables. Variables Definition Dependent Inctvrt Total net household income (R$) Princtvrt Net productive household income (R$) Extinct Income from on-farm extractive activities (R$) Piext Percent of household productive income from on-farm extraction (Pct) Independent Nwper Net wealth per capita (R$) Pisv Percent of productive income from barter or trade of products (Pct) Mkt1rnk2 Percent of productive income from off-farm labor (1-10%)a Mkt1rnk3 Percent of productive income from off-farm labor (11-45%)a Mkt1rnk4 Percent of productive income from off-farm labor (> 45%)a Travtime Travel time to city (minutes) Totlabor Total labor available including family and hired labor (days) Landsize Landholding size (hectares) Yredp8up Education per capita 8 years of age and older (years) Yrshead Years male household head on landholding (years)b Socsal Household member received a monthly social paymenta Nutsell Household sold Brazil nuts over the past 12 monthsa Caex Household member holds a membership in CAEXa Notes: a Dummy variable; b Where three is no male head of household, the data for the female head was used.
154 I employed the same method described above in dividing households by net wealth per capita to place households into different levels of market integration, creating a histogram of households, ranking households from lowest to hi ghest levels of percent of income from off-farm labor, then dividing th e households into groups by natural breaks in the histogram. Table 4-4 indicates the four labor market inte gration rank groups and the number of households in each. The intercept for the model serves as the base for interpreting the dummy variab les and represents no income from off-farm labor. Table 4-4. Rubber tapper households ranked by level of household integration into offfarm labor markets. Chico Mendes Extractive Reserve, 2001. Community Group 1 No Off-Farm Labor Group 2 1-10% Income OffFarm Group 3 11-45% Income OffFarm Group 4 Greater than 45% Income Off-Farm Total Rio Branco 2 8 7 1 18 Guarani 6 3 2 2 13 Terra Alta 5 5 3 2 15 Total 13 16 12 5 46 In addition to the key test variables, th e following control variables were also included in the regression model: household labor available, house hold education level, landholding size, years the household head has lived on the landholding, whether the household sold Brazil nuts (dummy to c ontrol for whether the landholding held a marketable quantity of Brazil nuts), whet her the household was a member of CAEX (dummy to control for membership in CAEX), and whether the household received a monthly government social payment (dummy to control for receipt of monthly retirement or physical/mental disability income).45 Households that received only a maternity 45 Years of education per household member eight years of age and older (yredp8up) was moderately positively correlated (r=0.64, p <0.001) with the variable, percent of productive income from off-farm labor (ppirtlab). I regressed yredp8up onto ppirtlab to further examine the strength of this relationship. The results of this regression were significant (R2=0.41, F=30.25) with ppirtlab having a coefficent of 0.10, indicating that a one year increase in the number of years of education per household member eight years or older would lead to a 0.10 increase in percent of in come from off-farm labor, i.e., for each additional year of study per member, off-farm income increases 10.0%. Thus, one could argue that education was
155 payment during the prior 12 months were not coded as a Â“1Â” for this last independent variable, as a maternity payment is not a regu lar and long-term benef it that would affect production planning. Table 4-5 on the following page presents the results of the f our regression models. The top row of the table contains the di fferent dependent variables tested. The independent variables are found on the left column. For mode l 2, a natural log transformation of household ne t productive income was used to normalize residuals. The first three models were run with only 45 households. One case was removed from the analysis on these three regression m odels as an examination of the influence (dfbetas) of observations indi cated that this one househol d was exerting a particularly large influence on at least one of the key test variable coe fficients. I have included the results of the regression models w ith all 46 observations in Appendix B.46 endogenous with off-farm labor. Ultimately I chose to use dummy variables to represent different levels of market integration in each of the four models. E liminating yredp8up did not substantially alter the coefficients or t-values of the off-farm dummy variables (mkt1rnk2, mkt1rnk3 and mkt1rnk4). The variable yredp8up was retained in the models to avoid bias in the coefficients. An examination of the conditional index for each model did not indicate multicollinearity among independent variables. 46 Christenson (1997) suggests that for small samples, dfbetas above 1.0 should be examined closely. An examination of the influence statistics found that for on e case the dfbetas were greater than 1.0 for at least one key test variable in models 1, 2 and 3. For model 3, in particular, dfbetas were greater than 1.0 for four of the six test variables and greater than 2.0 for on e variable. Thus, this case was heavily influencing the regression coefficients, including the si gn and statistical signifi cance of variables in the model. As a result, I eliminated this case from the model. A number of factors may help explain this: the household had relatively low wealth (wealth group 2) likely due to the selling off of approximately 25.0 % of transport animal wealth and nearly all cattle wealth over the pa st year; the household earned substantial income from the sale of rubber and Brazil nuts and the sale of animal assets; the household gained a very high share of production income from trade; it earned a share of pr oductive income from off-farm skilled labor, and; it received a monthly income th roughout the year for the mental disability of a da ughter. An examination of the results of the analysis with 45 households showed that one case had dfbetas 1 for one key test variable in model 2 and two key test variables in model 3. An examination of this case found nothing particularly odd regarding the test variables than having very high net wealth per capita holdings. Thus, as the influence of this case on the variables was only moderately higher than that recommended, and an examination of the case revealed nothing particularly odd, the case was kept in the model. Running these two models without this case altered the results slightly. In model 2, the R2 dropped slightly and the coefficients for the variables mkt1rnk2, mkt1rnk3 , pisv and yrshead became positive, but remained statistically nonsignificant. In model 3, R2 increased slightly and the variable nwper became statistically significant (p 0.05).
156 Table 4-5. Multivariate regression models re lating wealth and market integration to measures of household income and income from extraction Model 1: Household Net Income Model 2: Household Net Productive Incomea Model 3: Household Extractive Income Model 4: Percent of Productive Income from Extraction Variable Coefficient Coeffici ent Coefficient Coefficient Net wealth per capita (R$) 0.72*** 0.00017*** 0.12 -0.00004** Pct. of productive income from barter or trade -1,665.88 -0.29296 -2,238.27** -0.18352 Percent productive income from off-farm labor (1-10%)c 303.27 -0.00718 -532.08* -0.16628** Percent productive income from off-farm labor (11-45%)c -34.95 -0.07847 -1,276.50*** -0.24914*** Percent productive income from off-farm labor (> 45%)c 528.26 0.13518 -1,483.32*** -0.41360*** Travel time to city (minutes) 3.91* -0.00008 -1.59* -0.00023 Total Labor available (days) 3.07*** 0.00067*** 0.41 0.00009 Landholding Size (hectares) 2.48* 0.00085*** 3.02*** 0.00023* Education per capita 8 yrs and older (years) 193.44 0.01093 -.32.74 -0.00670 Years Head on Landholding 13.52 -0.00411 2.51 -0.00068 Monthly social paymentb 1,055.90 -0.68948*** -1,488.33*** -0.03129 Sold Brazil nutsb 130.43 -0.02635 212.03 0.06929 CAEX membershipb 1,346.68** 0.15520 529.23** 0.04151 Observations 45 45 45 46 Adjusted R-squared 0.68 0.54 0.65 0.50 Pr > F <0.0001 0.0001 <0.0001 0.0003 Notes : aUsed natural log transformation. bDummy variable. Membership in AMOREX and the STR were also tested for possible incl usion in the model, but were not significant. CAEX better captured the economic influence of social organizations working with rubber tapper households in the reserve. c Dummy variable. The coefficient represents the change compared to the intercept base. The base is no income from off-farm labor. *Significant at 0.10; ** significant at 0.05; *** significant at 0.01. Models 1 and 2 respond to hypothesis H1 and H4, examining the effects of wealth and market integration on two measurements of household income. I argued that an increase in wealth or integrat ion into markets, or greater ma rket access, would lead to an increase in household income. In Model 1, with the dependent variable net household income, all variables showed a positive rela tionship with the depe ndent variable except for two of the key test variables representi ng trade income and the dummy variable for households most integrated into labor ma rkets. The model showed a good fit in explaining the dependent variable (Adjusted R2 = 0.68, F = 8.16, Pr > F < 0.0001). Of
157 the key independent variables, net wealth per capita (p 0.01) was highly statistically significant and travel time (p 0.10) weakly significant. The coefficient of 0.72 for the net wealth variable indicates that as net wealth increases by R$1.00 per capita, total net household income increases by R$0.72, controlling fo r all other variables. For the travel time variable, the positive coefficient of 3.91 is opposite the relationship hypothesized. Thus, total income rises by nearly R$3.91 for each minute increase in travel time. The negative coefficient for percent of income fr om trade and one of the dummy variables for integration into off-farm labor markets, al so represent relationships opposite of those hypothesized, although these variables were not statistically significan t. Three control variables, including total hous ehold labor available (p 0.01), the dummy variable for membership in CAEX (p 0.05), and landholding size (p 0.10), were also statistically significant and positively related to total household income. Model 2 examines the relationship between the dependent variableÂ—net productive incomeÂ—and the same independent variables. I hypothesized that an increase in wealth and level of market integration or market a ccess would lead to an increase in productive income. The model provided a moderately good fit in explaining the dependent variable (Adjusted R2 = 0.54, F = 5.01, Pr > F = 0.0001). However, among the key test variables, only net wealth per capita was statistically significant (p 0.01). Although not statistically significant, coefficients for th e market integration variables percent of income from trade as well as the two lowest levels of integration into labor markets showed a negative association with net productive income, opposite of those hypothesized, indicating that linking into pr oduct and labor markets led to lower
158 productive income. The coefficient for tr avel time had a nega tive association as hypothesized. Among the control variables, landholding size , household labor available, and the dummy variable for receiving a monthly soci al benefit were all highly significant (p 0.01). The negative and high value of the coeffi cient for the receipt of a social benefit reflects the non-inclusion of monthly social benefits in the dependent variable net productive income. Figure 4-4 on the following page simulates th e effects of an increase in net wealth per capita on net productive income, controlling for all other variables in model 2. For the simulation, I used the mean figure as the default value for the continuous independent variables. For the dummy variables, I appl ied the following codes: as the majority of households sold Brazil nuts, nutsell is coded Â“1Â”; as the majority of households do not receive a monthly social salary, socsal is coded Â“0Â”; as more than half of the households are members of CAEX, caex is coded Â“1.Â” Fo r the off-farm labor dummy variables, all have been coded Â“0.Â” To simulate the eff ects of higher or lower net wealth per capita holdings on productive income, I have used th e mean figure for net wealth per capita and different fractions and multiples of the mean. Starting at the mean net wealth per ca pita, or R$1,466.00, the model simulates the effects of both a decrease and increase in net wealth per capita. At the mean, the model produces a net productive income of R$4,653.00. A doubling of net wealth per capita, while holding all other variables constant, resu lts in an increase in productive income of approximately 30.0 % to R$6,011.00. Concomitantl y, a net wealth per capita of 25.0 % of the mean, or R$367.00, produces producti ve income of approximately R$3,841.00.
159 Thus, holding all variables constant, hous ehold net wealth per capita maintains a moderately strong positive e ffect on productive income. 0 1000 2000 3000 4000 5000 6000 7000 .25 mean.5 meanmean (R$1,466) 1.5 mean2.0 mean Net Wealth per Capita (R$) Net Productive Income (R$) Figure 4-4. Simulation of the effects of a change in net wealth per capita on net productive income. Chico Mendes Extractive Reserve, 2001. Model 3 examines the effects of house hold wealth and market integration on income from extractive activities, responding to hypotheses H2 and H5. I argued that an increase in wealth or market integration, or market access, would lead to a decline in income from extraction. Again, the model exhibited a strong fit for explaining the dependent variable (Adjusted R2 = 0.65, F = 7.27, Pr > F = < 0.0001). All of the key test variables except for the wealth variable were st atistically significant in model 3. Each of the three dummy variables measuring integr ation into off-farm labor markets were negative, as hypothesized, and sta tistically significant, with the variables for percent of productive income from off-farm labor fr om 11-45 % and greater than 45% highly significant (p 0.01). The increasing value of the co efficients indicates a decline in household extractive income as the level of inte gration into labor markets increases. The two other key test variables, pe rcent of income from trade (p 0.05) and travel time (p
160 0.10) were also statistically si gnificant. The variable for integration into product trade was negative as hypothesized, i ndicating a decline in extrac tive income as households enter into product markets. However, the trav el time variable coefficient was negative, opposite from the hypothesized direction, indicati ng that as travel tim e increases, income from extraction falls. The coefficient for net wealth per capita was slightly positive, indicating a positive association between w ealth and extractive income, opposite that hypothesized, although as noted ab ove, it was not significant. Control variables for landholding size (p 0.01), monthly social benefits (p 0.01), and CAEX membership (p 0.05) were also statisticall y significant. The values of the regression coefficients for both CA EX membership and landholding size were positive, indicating a positive association w ith extractive income. Moving from nonmembership in CAEX to membership incr eased extractive income R$529.00, holding all other variables constant, while an increase in land area of 100 hectares results in an increase in extractive income of R$302.00. Conversely, the receipt of monthly social benefits carries a negative coefficient, indi cating that moving from no monthly retirement or health benefit to receiving this benefit leads to a R$1,488.00 fall in extractive income. This likely reflects the aging of the hous ehold and concomitant decline in on-farm production. Figures 4-5 and 4-6 on the following pages si mulate the effects of integrating into labor and product markets.47 Both simulations use the mean value for continuous variables, with the dummy variables coded as noted in the above simulation. Figure 4-5 simulates the effects of integrating into off-farm labor markets on extractive income. The 47 A simulation of the effects of travel time on extractive income has not been included as it was only significant at the p 0.10 level.
161 simulation demonstrates the negative effect th at integration into labor markets maintains on household extractive activit ies. For an Â“averageÂ” household, holding all other variables constant, as the house hold gains a greater share of in come from off-farm labor, extractive income falls from an initial level of R$2,914.00, representing no income from off-farm labor, to a low of R$1,431.00, wh en a household earns 45.0 % or more of income off-farm. 0.00 500.00 1000.00 1500.00 2000.00 2500.00 3000.00 3500.00 No Off-farm Labor 1-10% 11-45%Greater than 45% Percent of Income from Off-Farm LaborIncome from Extraction (R$) Figure 4-5. Simulation of the effects of integr ation into off-farm labor markets on income from extractive activities. Chico Mendes Extractive Reserve, 2001. Figure 4-5 shows that the largest decrease in extractive income comes as a household moves from greater than 1.0 10.0 % of income fro m off-farm labor, to 11.0 45.0 % of income from off-farm labor. This would suggest that shifts in productive activities leading to a reduction in extractiv e income are greater as labor moves from occasional unskilled wage labor activities to skilled off-farm labor activities. An examination of income earned from off-farm labor confirms this. Of the 16 households that earned between 1.0 10.0 % of producti ve income off-farm, only six provided
162 skilled labor service. In contrast, of th e 12 households that earned between 11.0 45.0 % of productive income off-farm, all provided skilled labor services. Figure 4-6 simulates the eff ects of integrating into product markets on extractive income. With all other variab les held constant, an increase in the percent of productive income earned from trade leads to a fall in income from extraction. For a household that earns only 10.0 % of productive income off-farm (i.e., consumption accounts for 90.0 % of productive income), extr active income totals R$3,433.00. With 33.0 % of productive income earned through product trade, extr active income falls to R$2,930.00. Earning 66.0 % of income from product trade, extr active income decreases to R$2184.00. For each 10.0 % increase in income from pr oduct trade, extractive income falls approximately R$220.00. 0.00 500.00 1000.00 1500.00 2000.00 2500.00 3000.00 3500.00 4000.00 10%25%33%50%66% Percent of Income from Product TradeIncome from Extraction (R$) Figure 4-6. Simulation of the effects of integr ation into product markets on income from extractive activities. Chico Mendes Extractive Reserve, 2001. Model 4 examines the effects of wea lth and markets on the percentage of productive income from on-farm extractive act ivities, responding to Hypotheses H3 and H6. I hypothesized that an increase in hous ehold wealth and market integration, and
163 market access, would lead to a decrease in the percent of pro ductive income from extraction. The model displayed a moderate ly good fit in explaining the dependent variable (Adjusted R2 = 0.50, F = 4.50, Pr > F = 0.0003). Among the key test variables, net wealth per capita and the three dummy va riables for percent of income earned from off-farm labor were statistically significant. The coefficients for each variable are as hypothesized, indicating a decline in the percentage of income from extraction as net wealth per capita and integrat ion into off-farm labor markets increase. Other key test variables, percent of income from trade and travel time to the city were not significant in this model. However, the coefficient for pe rcent of income from trade was negative, as hypothesized. This was not true for the tr avel time variable, which indicated an association with percent of income from extraction opposite of that hypothesized. Among the control variables, only landholding size was statistically significant. The coefficient for this variable was positive, indicating that a 100 hectares increase in landholding size brings a 2.3 % increase in the percent of inco me from extraction, controlling for all other variables. Figure 4-7 on the following page simulates the individual and combined effects of rising net wealth per capita and increased in tegration into off-farm labor markets on the percent of productive income from extraction. Again, the simulation employs the mean value for all continuous variable s held constant, with the du mmy variables coded as noted above. On the far left of the figure, a hous ehold with 25.0 % of the mean net wealth per capita, or approximately R$367.00 per house hold member, would earn 67.0 % of its productive income from extraction if it earn ed no income from off-farm labor. An increase in net wealth per capita to the mean value of R$1,466.00 would result in very
164 slight decrease in the per cent of productive income fr om extraction to 63.0 % of productive income, controlling for all other variables. A household holding twice the 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8.25 mean.5 meanmean (R$1,466) 1.5 mean2.0 meanNet Wealth per CapitaPercent of Productive Income from Extraction No Off-Farm 1-10% 11-45% Greater than 45% Figure 4-7. Simulation of the effects of wealth and integration into off-farm labor markets on percent of household productive income earned from extractive activities. Chico Mendes Extractive Reserve, 2001. mean net capital wealth woul d earn 44.0 % of productive income from extraction. Thus, the simulation demonstrates that even as ne t wealth per capita increases dramatically, households still retain a s ubstantial proportion of producti ve income from extractive activities. Concomitantly, Figure 4-7 also displays the e ffects of an increase in the percent of income earned from off-farm labor on the percent of income fr om extraction. For a household at the mean net wealth per capita, or R$1,466.00, and receiving no income from off-farm labor, extractive income w ould account for 63.0 % of productive income. At the lowest level of integr ation into labor markets, or 1.0 10.0 % of income earned off-farm, a household at the mean net wea lth per capita level would earn 46.0 % of income from extraction, holdi ng all other variables constant . Greater reliance on off-farm
165 income leads to a further drop in the propor tion of productive income from extraction: a household earning 11.0 45.0 % of income off -farm would earn 38.0 % of income from extraction, while a household ea rning more than 45.0 % of in come would earn just 21.0 % from extraction. Simulating the combined effects of a rise in net wealth per capita and increased integration into off-farm labor markets finds that a household with twice the mean net wealth per capita and earni ng greater than 45.0 % of produc tive income off-farm would earn only 3.0 % of productive in come from extraction. This suggests that for wealthy households that are highly integrated into labor markets, hunting as well as marketoriented extractive activities, such as th e collection of Brazil nuts and rubber, fall dramatically. The results of the regression models s uggest mixed findings regarding the six hypotheses tested and the key test variables. Net wealth per capita was positive and statistically significant in explaining th e two measures of ho usehold income, net household income and net productive income (models 1 and 2), confirming hypothesis H1. This argues that as household wealth increases, household income and productive income increase. The simulation of the eff ect of wealth on net pr oductive income (model 2) showed a moderately strong positive re lationship between wealth and productive income. Net wealth was negative and statistica lly significant in explaining the percent of income from extraction (model 4), confir ming hypothesis H3; with an increase in household wealth, the percent of productive income from extr action falls. However, the simulation showed that wealth has only a moderate negative effect on the dependent variable, suggesting that othe r factors can better explain a decline in the percent of
166 income households earn from extraction. Finally, wealth was not significant in explaining extractive income (mode l 3), thus rejecting hypothesis H2 that an increase in wealth leads to a fall in income from extrac tion. This finding conf irms statistically the data displayed in Tables 4-6 and 4-7. Hous eholds across wealth ca tegories earn sizeable income (if not proportional to productiv e income) from extractive activities. Statistical tests measuring the effects of integration into markets on household income and income from extraction also prov ided mixed findings. In testing the three measures of market integration, or mark et access, on the two m easures of household income (models 1 and 2), travel time showed a weak statistically significant relationship with household net income. Ho wever, the coefficient indi cated a positive relationship between travel time and household income , opposite the hypothesized directionÂ—as travel time to the city increases (indicati ng poorer market access), household income also increases. This argues, counter intuitively, that households with better access to markets earn lower incomes. The results of this measur e of market integrati on might be attributed to the dynamic changes taking place in the reserv e and the disparity in access to transport. Road building and access to motorized transport, such as access to the municipal truck that services the communities of Rio Bran co and SÃ£o JÃµao de Guarani during the dry season, has decreased travel time to the ci ty of Xapuri for many households that are geographically distant from the city. Ho wever, not all households reported that motorized transport is their principal means of transport. Thus, travel time to the city varied widely, with some households geographi cally closer to the city reporting longer travel times to the city than more distant hous eholds that use motori zed transport. This
167 might help explain the unexpected results of this measure of market integration for each of the regression models. Neither the level of integration into produc t markets or labor markets of the other two measures was statistically significant in their relationship with the two measures of household income. Thus, the statistical anal ysis rejects hypothesis H2 and finds that higher levels of market integration or better market access do not lead to higher income. The findings on the effects of market inte gration on extractive income and percent of income from extraction (models 3 and 4) al so varied. Market in tegration measured by percent of productive income from product tr ade showed a negative relationship with both extractive income and percent of income from extraction. However, this relationship was statistically significant only with the dependent variable extractive income. The simulation showed a moderately strong eff ect of increasing inte gration into product markets on extractive income. The variable travel time was negative and statistically significant only for the dependent variable ex tractive income (model 3). However, this relationship was opposite the hypothesized directi on, indicating that as travel time to the city increases, income from extraction falls . I expected households with longer travel times to the city to use the forest more intensively. This findi ng indicates that poor access to markets negatively affects the ability of households to commercialize extractive products. As noted above, the results of the regression for this measure of market integration might be partially explained by th e variation in travel time across households with similar geographic distances to the city. Finally, market integration measured by the three dummy variables for percent of producti ve income from off-farm labor showed negative and statistically significant relati onships with both ex tractive income and
168 percent of productive income from extraction. In simulating the effects of integration into labor markets for models 3 and 4, the du mmy variables showed a particularly strong and negative effect on the percent of income from extractive income. Thus, the analysis both accepts and rejects hypotheses H4 and H6, depending on the measure of market integration. The anal ysis accepts the hypothe ses that integration into off-farm labor markets negatively aff ects both extractive income and percent of income from extraction. It also accepts th e hypothesis that incr eases in product trade negatively affect extractive income, but rejects the hypothesis th at increasing product trade negatively affects the percent of inco me from extraction. Finally, the hypotheses that greater market access, measured by trav el time to the city, negatively affects both extractive income and percent of income from extraction are rejected. Further, it seems important to consider the implication of the statistically significant findings for landholding size and me mbership in CAEX. First, landholding size was positively related to both measures of household income (models 1 and 2) and extractive income (model 3). With landholdings being subdivided in the reserve, this finding suggests that subdividing landholdi ngs could lead to falling income and extractive income. Second, CAEX membership was also positively related to both net household income measures and extractive in come. Here it is important to note that CAEX membership is found principally where it maintains an in-for est trading post, in this case, in the communities of Rio Branco and Terra Alta. Only 3 of 13 households in the community of SÃ£o JoÃ£o de Guarani are members of CAEX, compared to 13 of 18 households in Rio Branco and 10 of 15 househol ds in Terra Alta. Thus this positive relationship between CAEX membership a nd both productive income and extractive
169 income may reflect broader community develo pment issues at work within these two communities, as well as the economic benefits of CAEX membership. These in turn, have brought economic gains to members.48 The detailed analysis of the wealth hold ings and income earnings from both on and off-farm labor activities, of the 46 study hous eholds has provided both a broad and indepth view of the rubber tappe r economy. The description a nd analysis of how rubber tapper households invest in w ealth, revealed not only the diverse wealth holdings of rubber tappers in the Chico Mendes Extractiv e Reserve, but also how asset holdings change as wealth holdings grow. For the least wealthy households, all wealth was held on-farmÂ—they had no non-farm wealth. For this group, productive wealth was held primarily in equipment and animals, with a slightly lower percenta ge held in productive structures. Animal wealth was dominated by transport animals, fowl and small animals; the least wealthy househ olds had no cattle. The data also showed how wealth holdings shift as wealth increases. As wealth grows, on-farm investments shift into an imal production, particularly cattle, and landholding structures that support livestock production, such as fencing and corrals, while the proportion of wealth he ld in equipment declines. In addition, as wealth grows, households increase their on-farm holdings in consumer goods, as well as their off-farm holdings in the city, including investments in land, houses, and consumer goods, such as televisions, stoves and refriger ators. Households with grea ter wealth actually hold a lower percent of total wealth in on-farm pr oductive assets. The mo st notable shift in productive wealth is the higher investment in cattle found in the three wealthiest groups, 48 Dummy variables for each community were tested for significance in all regression models. None were significant so they were not included in the models.
170 with the most wealthy households holding the greatest percent of wealth in cattle. This shift in investments into livestock is reflect ed in income earnings; the three wealthiest groups earn a greater share (although not increasingly acros s wealth categories) of income from cattle, with the wealthiest group earning the greatest percent of income from cattle sales among the wealth groups. But even as asset investments shift as wealth holdings rise, the data also demonstrated that extractive ac tivities continue to play an important role in household production. Across all wealth categories, extr action is the major source of on-farm productive income, with even the wealthiest households earning incomes from extractive activities at values very si milar to the poorest households. Although rubber production falls among the wealthier households, and extraction makes a proportionally smaller contribution to trade income among the weal thiest households, game hunting and Brazil nut production remain important to produc tive income for all wealth groups. These findings were reflected in the stat istical analysis. Wealth had a weak positive association (although not st atistically significant) with the value of income from extraction. However, the analysis also found a modest negative and statistically significant relationship between wealth and th e percent of income from extraction: as household wealth increases, extr action plays a declining role as a proportion of household productive income. This is a reflection of the growing income contribution of animal production in the higher wealth categories. This study corroborates the varied findings of other studies examining the role of wealth on extractive activities . My findings suggest that extraction may help smooth consumptive activities of both poor and wealt hy forest households (P attanayak and Sills
171 2001) and that households with greater wealth spend more time in agriculture, and concomitantly, earn income from diverse so urces (Demmer and Overton 2001), and are able to invest in riskier activities su ch as cattle production (Dercon 1998). The implications of increasing invest ment in cattle for the environment are clear. As rubber tappers invest in cattle production, this will require concomitant investments in structures, such as fencing and corrals, as well as transforming forestland to pasture. Just as wealth is shaping household produc tion activities, this chapter has also argued that integration into markets is bringing changes to rubber tapper household production. The statistical models demonstrat ed that as households integrate into offfarm labor markets and product markets, th e value of income earned from extraction declines. Further, as integration into labor markets increases, the percent of productive income from extraction also falls. A closer look at the data revealed that skilled wage and salaried labor, rather than occasional unski lled wage labor, are the principal off-farm activities that are shapin g productive activities. The implications of these findings for c onservation are less cl ear. Less extraction, or a falling contribution of extraction to income does not mean that households are carrying out other destructive land use practices. A quick ex amination of the association between measures of market integration and variables that proxy la nd conversion, such as land held in pasture, cattle w ealth holdings, and the value of the production of basic food crops, provides some guidance. Table 46 on the following page suggests that only integration into product markets is positively co rrelated to cattle wea lth and the value of basic crops, indicating a potenti ally weak negative impact on the environment. Travel
172 time and integration into off-farm labor mark ets showed slightly positive environmental effects for the variables cattle w ealth and crop production, respectively. Table 4-6. Correlations between measures of market integration and three proxies for deforestation. Land in Pasture Cattle Wealth Value of Basic Food Crop Production Percent of productive income from off-farm labor 0.03 0.02 -0.34 Percent of productive income from barter and trade 0.07 0.25* 0.27* Travel time to Xapuri 0.04 -0.22 0.04 *Significant at the .10 level. A closer examination of the association between off-farm labor Â– the indicator which showed the largest drop in the percentage of productiv e income from extraction at higher levels of market integrationÂ—and cat tle wealth is also informative. The 21 households that held over R$1000.00 in cattle wealth (a total of 31 households held wealth in cattle), included onl y eight of the 17 households in the two highest off-farm labor market categories. In contrast, 13 of these 21 households were from the two highest product trade categories. Thus, even as households integrate into o ff-farm labor markets, and the value and percent of income earned from extraction falls, households, as of yet, are not investing off-farm earnings into environmentally destru ctive activities. My findings corroborate previous research that found th at integration into off-farm labor markets does not lead to increased deforestation (Godoy, Wilkie and Fr anks 1997) and may help stabilize forest stocks (Bluffstone 1995). This is an impor tant finding, as a growing number of studies are documenting the increasing importance of off-farm labor, both agricultural and nonagricultural, on rural inco mes (Deininger and Olinto 2001, Escobal 2001, de Janvry and Sadoulet 2001, Reardon, BerdeguÃ©, and Escoba r 2001). However, the medium and long-
173 term effects of integration into off-farm labor markets will need to be monitored, as households earning substantial in come from off-farm activities will be searching for ways to productively invest accumula ted earnings on their landholdings. Although a decline in ex tractive activities rela ted to increasing integration into offfarm labor markets is not associated with greater investments in environmentally destructive activities, the conservation implications for increasing integration into product markets are less clear. Table 4-6 suggests a weak correlation between integration into product markets and cattle wealth and crop production. This suggests support for other studies that have argued that integration into product trade can lead to increased deforestation (Bedoya Garland 1995, Godoy, Wilkie and Franks 1997) and increased crop cultivation (Henrich 1997). Conclusion Thus, this chapter finds that wealth a nd markets are bringing changes to rubber tapper productive activities and in particular, extractive activities. In the regression models, wealth appears to ha ve a relatively moderate e ffect on household extractive activities. Even the wealthiest households produce substantial income from extractive activities. However, a br eakdown of households into wealth rank groups indicates a growing importance of income from livestock production as wealth grows. In the shortterm, household wealth is likely to grow slow ly, thus the implicati ons for conservation Â– large scale land conversion for pasture areasÂ—may not be imme diately seen. Indeed, the conversion of land will take place over years. However, in the medium and long-term, increasing wealth accumulation among households will bring greater need for pasture and thus, deforestation.
174 The regression models also found a strong, ne gative relationship of integration into labor markets on extraction. As noted above , the implications of this finding for conservation are encouraging, despite the str ong negative relationship. Higher levels of integration into labor markets are related to falling income from extraction and a lower percent of income from extr action, but less income from th e forest was not associated with more destructive land us e practices. Yet, the story of Cleilson, and Daniela, does suggest concern. Cleilson is a carpenter and his wife Dona Daniela is a schoolteacher. Together they earn a substantial income, and this is reflected in the new home they recently built and its furnishings. The house is beautifully construc ted with sawn timber planks, thanks to CleilsonÂ’s skill as a carpenter, and a Brasilit r oof finishes off the structure. Inside they have a gas oven and steel sink. Gas ovens ar e still scarce in the fo rest, but are slowly growing in number. Sinks are rarely found. These acquisitions clearly came as a result of their high off-farm incomes. But while off-farm labor has brought new acquisitions, it has also affected the householdÂ’s production strategy. Cleilson and Dona Daniela di d not plant any crops the year prior to the research peri od and Cleilson rarely went hun ting; this past year he had killed only one peccary, two pacas, one deer , two jacÃºs, and one aracua. Animal production was also relatively low; the selling of a mule made up most of the householdsÂ’ trade income. After my first round of inte rviews, the household had one steer and a few small animals. But this will soon change. As I was leaving their landholding at the end of my second visit, I noticed that a large area had recently been cut near their house. A fence would soon be in place. Cleilson and Dona Daniela were going to start investing
175 more in livestock production. When Cleilson to ld me about his plans for cattle during my visit, I could sense that he was not entirely comfortable with the idea of a growing cattle herd. Cleilson made it clear that they would respect the limits imposed by the reserve management plan. But while they were concer ned about the increase in cattle raising in the forest, they also believed that cattle wa s a good investment, both in terms of their financial situation, and in te rms of potential consumption. I found that the larges t decline in extractiv e activities occurs at only the highest levels of integration into labor markets. In the short-term, and even medium term, the demand for rural labor in these sectors will be limitedÂ—the forest economy will only support a few carpenters in a region and the need for teachers and cooperative managers is unlikely to grow substan tially. But other off-farm la bor opportunities are likely to emerge, especially with the continued building of feeder roads which provide better access to off-farm labor opportunities. An in crease in development projects, such as community timber management projects now be ing considered for implementation in the reserve, may bring not only new forms of on-farm income that compete with extractivism, but also new opportunities for sa laries and wages for community members. This study found that it is sala ried and skilled wage positi ons, such as those held by Cleilson and Dona Daniella, which have the strongest negative effect on extractivism. Yet, using the forest less for extaction wa s not accompanied by the investment of earned income in more destructive activities. Th is is good news for conservation. But my conversation with Cleilson and Dona Daniela suggests that as hous eholds earn higher incomes, they eventually must make choices about how they invest their wealth. Their
176 decision mirrors the findings I have laid out in this chapter: as households accumulate wealth, they will invest in on-farm consumer goods, but also in cattle. This chapter also briefly examined the e ffects of the Lei Chico Mendes, the state rubber subsidy designed to increase rubber ta pper household incomes and concomitantly keep stem rural-urban migration. Al though rubber tapper house holds are earning a higher price for rubber, a comparison of th e production of 28 househol ds both before and after the implementation of the subsidy found that the s ubsidy is not stimulating production, nor encouraging households to re turn to rubber tapping. Rubber production among these households fell dramatically from 1996 to 2001 and the number of households tapping rubber fell as well. Thus, at least among this small group of households in the reserve, the rubber subsi dy is not stimulating a return to rubber production. Having responded to the first set of questi ons directing this study, and examined the effects of wealth and market integration on rubber tapper household income, and more specifically, income from extractive activit ies, I now turn to respond to the second set of questions posed by this study. As hous eholds accumulate wealth, and integrate into markets, how might these changes affect th e cultural knowledge of rubber tappers? In the following two chapters, I combine the resu lts of cognitive research methods with data presented in this chapter to answer these questions.
177 CHAPTER 5 RUBBER TAPPER CULTURA L KNOWLEDGE OF EXTRACTIVE RESOURCES After my research assistant and I finished up our second round of interviews with Roberto, Cibele and their family, we gathered wi th them for a few pictures in front of the house. This usually involved taking photogra phs of diverse combinations of parents, brothers, sisters and friends, sometimes with pe ts (particularly parrots and in one instance a peccary, but rarely the family dog) and farm animals. Household pets and farm animals I have photographed over the years include mo nkeys (one in its small hammock), ducks, pigs, parakeets, parrots and la rge animals, mainly cows and horses. However, one picture included the family muleÂ—referred to affectionately as the Ministro de Transporte , the Minster of Transportation. Occasionally, me mbers of the families would ask to be photographed sitting on top of a horse or mule. In one instance, I was asked to give up my slip-on boots (a style often worn by lo cal ranch hands) so that the young man could be photographed in partial, if not full, cowboy regalia. At Roberto and CibeleÂ’s house, we sna pped only a couple of pictures of their family. However, as I was packing my camera into my backpack for the hike back to a neighboring landholding, I was asked if I wanted to take a look at the small corral built just behind the house. In the year since my previous visit, the family had acquired a young steer and constructed a small corral to house it. The young men of the family had it penned up and wanted me to take a look. I grabbed my camera and walked back to find a single steer in a corral that measured appr oximately 9 feet by 9 feet. With everyone grinning widely, I was asked if I wanted to ri de it. I declined, a lthough indicated I would
178 be happy to take a picture of the others if th ey dared. I readied my camera as one of the brothers slid onto the animalÂ’s back and th e others held the anim al steady against the corral fence. He secured his hand under a rope wrapped around the steerÂ’s torso. The animal clearly did not like what was ha ppening, or at least was excited about the possibility of throwing off the rider. I clim bed up onto the fence to get a better view of the steer and rider and steadied myself, camera in hand. Â“ Pronto ,Â” I shouted, when I was focused on the rider and animal and ready fo r the shot. Released, the animal bucked wildly, trying to throw off the rider, and cras hed into the side of the fence. The ride lasted a few seconds before the rider, fa lling to the side, was off the animal and clambering for safety on the fence. Eyes gleaming with laughter and excitement, the steer, snorting foamy saliva from its nostrils, was secured once again for another brother to take a short ride. In many ways, Roberto, Cibeli and their fa mily typify the household that one envisions when thinking about rubber tappers . Roberto and Cibele have eight children ranging in age from four to seventeen. R oberto and Cibele are both the children of rubber tappers; the family now lives on the landholding where Roberto grew up. Roberto has a brother who lives on a landholding nearby, w ho in turn is married to the daughter of another rubber tapper from the same forest area. The family just recently built a new house. Although the new structure is built of sawn wood planks substituting the more traditional planks of the paxiÃºba (Aracaeae) and aÃ§aÃ ( Euterpe precatoria ) palms, the roof remains composed of palm thatch rather than aluminum or Brasilit , that, as noted in Chapter 3, is becoming increasingly popular in new home construction among rubber tapper families.
179 In addition, the household s till gains the majority of income from sales of extractive productsÂ—principally rubber and Braz il nuts. Small animal production is also important to income. The family only recently had the resources to invest in livestock. There are no chainsaws, no motorcycles, no motors for making manioc flour, no solar panels, no televisions, no houses in the cit y. The family has a canoe and motor to facilitate travel to Xapuri, not unl ike a number of families that live along Riozinho . The sons continue to tap rubber, an o ccupation that many youth have given up. Yet, the excitement at the small corral is a sign that the livelihood of the rubber tapper and the dreams of rubber tapper youth, ev en among those households that still rely heavily on rubber trade for income, are changi ng. Indeed, as noted in Chapter 3, many households no longer tap rubber, and for t hose who do, many are producing less than in previous years. This finding meshes well w ith the results of CampbellÂ’s (1996) research in the Chico Mendes and nearby Cachoeira Ex tractive Reserves in the early 1990s. She found that from 1991 to 1994, the number of fam ilies that did not want their children to be rubber tappers increased from 31.0 % to 51.0 %, and some households referred to those who still tapped rubber with pityÂ—Â“ coitado do seringueiro Â” (Campbell 1996:172). How then, might the cultural identity of the rubber tapper be changing? Can we identify variations in culture among rubber ta pper households? What are the implications of culture change for the ex tractive reserve concept and, more generally, rainforest conservation in the Amazon? Over the next two chapters I analyze the results of cognitive anthropological tests conducted with forest households to exam ine one aspect of ru bber tapper cultural identity: cultural knowledge. In this chapter, I test for vari ation in cultural knowledge of
180 non-timber forest resources across rubber ta pper households, comparing the knowledge of household heads, as well as the knowledge of individuals in di fferent sub-groups of study participants. In the fo llowing chapter, I test for va riation in cultural knowledge across households segmented by levels of w ealth and market integration, as well as examining differences in knowledge among sub-groups of the study participants, including males and females, and youth, young a dults and adults. Cognitive tests provide a unique window to view how rubber tapper s Â“thinkÂ” about the domain of non-timber forest resourcesÂ—how they categorize items, a nd more specifically, th e level of cultural consensus on this domain. They can help us understand how rubber tapper cultural knowledge of forest resources may be changi ng, or varying, across individuals, and subgroups of individuals. In this chapter I argue that despite differi ng livelihoods in the forest, rubber tappers maintain a high degree of cultural cons ensus on the domain of non-timber forest resources. Further, sub-groups of individuals , categorized by sex and age, also maintain a high degree of shared cultural knowledge on th is domain. In the following chapter, I demonstrate that this high de gree of shared knowledge is ma intained, in general, across different levels of wealth and market inte gration. Thus, even as household wealth and levels of market integration increase, rubbe r tappersÂ—both males and females and of all agesÂ—widely share cultural knowledg e of the forest resources th at they use, and in turn, that shape their lives. However, two exceptions will be noted in the following chapter. The next question is, what ar e the implications of this high degree of consensus for conservation? I argue that this bodes well for conservation for two reasons. First, the maintenance of cultural knowledge across wealth levels and age groups should facilitate
181 the continued use of forest resources despite diverse livelihood activities that will emerge as wealth increases. This maintains rubber tapper household ties to the forest, as a potential resource for consumptive (and comm ercial) use, should the household financial status change (cf Pattanayak and Sills 2001). This is particularly true for cultural knowledge of the medicinal properties of pl ants. Second, cognitive tests can help us understand the subtle changes taking place in how people think about non-timber forest resources. Understanding even subtle variat ions in how individuals think about forest resources can help predict future resource use patterns and precipitate the need to develop management plans for resources that may be moving from consumptive use to market oriented use. They may also help us to identify who are the most appropriate candidates for development interventions. Concomitantl y, we can also better understand the effects of new development activities in the reserv e, and how they may be reshaping knowledge of these resources. In the following chapte r, I will explore some of these subtle knowledge changes and consider their implications. This chapter will begin with an analysis of the results of the free-list activity conducted during the first field visit in 2001. I then analyze the result s of the pile-sorting exercise, including a close look at the different type s of item groups created by participants. This discussion folds into an examination of the multi-dimensional scaling (MDS) and hierarchical cluster analysis carri ed out with pile sort data. I will then examine the results of the consensus analysis procedure, again gene rated by the pile-sort data. Finally, QAP is employed to examine th e correlation of different sub-groups of the rubber tapper population. The following chapter will respond to hypotheses H7 through
182 H10, presented in Chapter 1, testing the eff ects of wealth and market integration on rubber tapper cultural knowledge of non-timber forest resources. Free-Listing of Non-Timber Forest Resources Free listing involves asking respondents to li st all of the items they can think of with regard to a particular cultural do main (Bernard 1995). I asked households the following question: Â“Can you make a list of thin gs of the forest that you use, other than animals and wood?Â”49 For this cognitive test, I worked with household members as a group, including adults and child ren, not individuals. I deci ded on this method based on my previous experience working with rubber ta pper families in the reserve. I believed that working with all household members togeth er, rather than indivi duals, would provide a healthy forum to elicit free list items: as one household member would list an item, another member would be encouraged to list items as well. This would also help bring younger household members into the activity, es pecially younger wome n and adolescent girls, who are sometimes reluctan t to participate in research.50 Further, I planned to conduct pile sort tests with all teenage and olde r household members individually, on my second field visit. This would provide the opportunity to understa nd how individuals of different sex and age think about the do main of non-timber forest resources. Before I begin the analysis, the question I asked to elicit freelists, Â“Can you make a list all the things from the forest that you use, other than animals and wood?Â” deserves a brief comment. As noted above, I was inte rested in non-timber forest resources, so I 49 The question phrased in Portuguese was Â“Pode fazer uma lista das coisas da floresta que usa, menos animais e madeireiros?Â” 50 It is possible that with some households, family members might have chosen to remain silent during the group interview, allowing the household head(s) to respond to questions. In conducting the free-lists, when a prolonged silence appeared to signal an end to the householdÂ’s free-list, I asked if anyone had any other items they would like to add to the list, addressing the questions to all household members.
183 specifically wanted to avoid the listing of forest resources used for timber. Asking households to not list wood resources ( madeireiras ) did not preclude them from noting the use of a particular part of a species, other than its use fo r timber, such as the tree bark for medicinal purposes. I also wanted to elimin ate the listing of game animals. As noted in Chapter 3, households hunt a number of fo rest animals, and they are important for household consumption as well as providing m eat, through sharing, to other families in the area. For nearly all households, game animals contributed subs tantially to productive income. However, I was interested primar ily in plant resource use and knowledge, including items both consumed and traded, ther efore these resources would take priority over the inclusion of other forest resources. Pre-testing of the pile sort test led me to limit the number of items to 24, and I felt this number was too low to include game animals. I also asked households to list things of the forest that they Â“use.Â” This gave the household a more concrete framework for res ponding to the question rather than a more open question asking them to list Â“things,Â” or Â“things that th ey know.Â” However, it also put limits on what households might respond. For example, only 23 of 44 households listed either Â“ seringa Â” or Â“ borracha ,Â” both referring to rubber. Given a more open question, more households may have listed rubb er, an item that all households know and most likely extracted at some point in their past. Eleven households did not tap rubber, although four still placed it on their free-list. More interesting, only 19 of 33 households that tapped rubber in the 12 months prior to the research placed Â“ seringa Â” or Â“ borracha Â” on their free-list. This in itself is notable, as well as surprising, as rubber was a major source of income for many of these households.
184 One more clarification regarding the freelists is important to make. In 1996 and 1997 I conducted research in the reserve with many of the same rubber tapper households in this study. During this period I carried ou t a study on three particular forest fruits, all palm species: aÃ§aÃ , bacab , and patauÃ¡ . This likely affected the response of households to the free-list exercise in two ways. First, my presence when asking the question almost certainly cued the response of some househol ds, some of which immediately rattled off aÃ§aÃ , bacaba and patauÃ¡ as free list items. Second, these items were often the first items stated by households, thus ta inting any analysis of the or der in which the items were listed. As evidence supporting this asserti on, of all items listed more than two times among all households (29 items were on two or more household free-lists), only rubber and Brazil nuts maintained a highe r average rank on the lists than aÃ§aÃ , bacaba and patauÃ¡ . The free-lists provided by study househol ds open a unique window to understand rubber tapper use and knowledge of non-timber forest resource. Through analysis of these lists, we can begin to s ee not only the diversity of the plant (and non-plant) forest resources that rubber tapper households use, including, in some cases, the particular part of the plant utilized, but also the different Linnaean plan t families that are important to forest households. Table 5-1 on the following page contains a complete list of the 83 free-list items provided by 44 rubber tapper households. All but two items were common plant names or plant parts: households also listed two non-plant itemsÂ—honey and water. Of the 81 plant-related items, all but two referred to a sp ecific plant. In tw o cases, generic plant parts, rather than common plant names, were listedÂ— palha , or palm leaf or frond
185 (literally, straw), and palmito , or heart of palm.51 In numerous cases, the common plant name was accompanied with the name of the specific plant part used. For example, copaÃba was listed, as well as copaÃba oil and copaÃba bark. These are counted as three different items (of the total 83), but all refer to the same plant genus, Copiafera .52 Other plant parts noted along with common plant na mes included palm leaves, coconut, fruit, vines, and seeds. Table 5-1. Extractive resource free-list items and frequenc y of occurrence for 45 rubber tapper households. Chico Mendes Extractive Reserve, 2001. Free-list Items AÃ§aÃ (33) CipÃ³ arumÃ£ (3) CajÃº (1) Marupa (1) Castanha (29) PaxiÃºba (3) Casca de canelÃ£o (1) Murumuru (1) PatauÃ¡ (28) ArumÃ£ (3) CarapanaÃºba (1) CÃ´co de ouricuri (1) Bacaba (24) IngÃ¡ (2) Carm elitana (1) Ouricuri (1) Bacuri (17) Cajarana (2 ) CatuÃ¡ba (1) Pariri (1) Cacau (17) Palha de ouricuri (2) Casca de Cumaru de Cheiro (1) "leo de patauÃ¡ (1) Borracha (16) MamÃ£o (2) Cumaru de dheiro (1) Vinho de patauÃ¡ (1) JabobÃ¡ (10) Casca de jatobÃ¡ (2) Ch ichuacha (1) Casca de pau d'arco (1) CipÃ³ timbÃ³ (10) Casca de cerejeira (1) TimbÃ³ (1) Pau d'arco (1) CipÃ³ ambÃ© (8) Cerejeira (1) Gameleira (1) Pupunha (1) Seringa (7) Pau dÂ’arco (1) Goiaba (1) Sucuuba (1) JutaÃ (6) Vinho de aÃ§aÃ (1) Gu ariuba (1) Tachi preto (1) CopaÃba (5) Sementes de aÃ§aÃ (1 ) IngÃ¡-da-mata (1) Tatajuba (1) Pama (5) Casca de angico (1) JarÃna (1) Toari (1) Mel de abelha (5) Apurui (1) JataÃ (1) Palha de ubim (1) Palha (4) CÃ´co de arumÃ£ (1) Fruta de jatobÃ¡ (1) CipÃ³ unha fe gato (1) Quina-quina (4) AmbÃ© (1) Casca de jutaÃ (1) Unha de gato (1) "leo de copaÃba (4) Ata-da-mat a (1) Laranjeira (1) Ãgua (1) Palha de jarina (4) Banana (1) Maca (1) Palmito (1) Buriti (3) Breu (1) MacÃ£o (1) Sementes de faveira (1) Casca de copaÃba (3) CajÃ¡ (1) Mamui (1) In two cases, the same plant species was re ferred to through the use of two different common names or in different formsÂ—proce ssed and unprocessed. In the first case, rubber tappers used the common names cerejeira and cumaru de cheiro in referring to the 51 In Acre, heart of palm is extracted principally from two palm species, aÃ§aÃ ( Euterpe precatoria ), a single stemmed palm and peach palm, locally referred to as pupunha ( Bactris gisaepes ), a multi-stemmed palm planted in agroforestry projects. 52 There are seven species of the genus Copiafera that occur in Acre.
186 single species, Amburana cearensis (Allemao) A.C. Sm. Timber from A. cearensis is used for construction while the bark is used for medicinal purposes. In the second case, rubber ( Hevea brasiliensis ) was listed in two different forms, seringa and borracha . Seringa refers to the latex that is extracted from the tree (no households listed seringueira , which would refer to the tree itself), while borracha refers to rubber, the product that the rubber tappers produce with seringa through the use of simple in-forest processing technologies. For pile-sorting purpos es, in the former case, I chose to use the term cerejeira Â—neither common name ( cerejeira and cumaru de cheiro) was used more often than the other. In the latter case I used the term borracha , as this term was freelisted more often than seringa . Table 5-2 on the following page presen ts free-list items categorized by plant family. The number of genera, as well as the number of free-list items in each family, is included. Free-list items represent approxima tely 29 plant families. As some common plant names can be attributed to different genera and families, the exact number of plant families represented could not be determine d. Arecaceae, the palm family, was the most common plant family found among free-list items. In this family, 19 different items were listed, representing 10 different genera. No othe r plant family came close to this number. It represents nearly 25.0 % of all plant items listed and approximately 20.0 % of all genera. Clearly palms are an important plan t species to rubber ta pper households. This is not surprising; scholars ha ve christened the palm plant th e Â“tree of lifeÂ” (Balick 1988), due to the various uses that it serves , including food, construction, medicine, and handicraft production, among others.
187 Table 5-2. Free-list items categorized by plant family and genus. Family Number of Generab Number of Free List Items in Family Anacardiaceae 2 3 Annonaceae 1 1 Apocynaceae 2 2 Arecaceae 10 19 Araceae 1 2 Bignoniaceae 3 4 Burseraceae 1 1 Caesalpiniaceae 3 9 Caricaceae 1 or 2 2 Celastraceae/Hippocrateaceaea 1 1 Clusiaceae 1 1 Cyclanthaceae 1 2 Euphorbiaceae 1 2 Fabaceae 1 4 Fabaceae/Caesalpiniaceae/Bignoniaceaea 1 1 Flacourtiaceae/Caesalpiniaceaea 1 1 Lauraceae 1 1 Lecythidaceae 2 2 Marantaceae 1 3 Mimosaceae 2 3 Mimosaceae/ Caesalpiniaceae/ Bignoniaceaea 1 1 Moracaea 3-4 4 Musaceae 1 1 Myrtaceae 1 1 Rosaceae 1 1 Rubiaceae 2 3 Rubiaceae/Apocynaceaea 1 1 Rutaceae 1 1 Sapotaceae 1 1 Sterculiaceae 1 1 Verbenaceae 1 1 Vochysiaceae/Menispermaceae/ Melastomataceae/ Flacourtiaceae/Zaiaceaea 1 1 Total 52-54 81 Notes : aMultiple families indicate that the family could not be determined from Daly et al. (n.d.) and consultation with a botanist in Acre, Brazil. Without a botanical identification, the species could be attributed to more than one family. bGenera were determined by previous plant collections in Ac re, Daly et al. (n.d.) and consultation with a botanist in Acre, Brazil. In some cases, the genus for the free list item could not be determined. How do these data compare to previous ethno-botanical research among rubber tapper households in Acre? As noted in Ch apter 3, Kainer and Du ryea (1992) identified 145 plant species among 60 plant families that women in the Chico Mendes and
188 Cachoeira Extractive Reserves use.53 They carried out struct ured interviews with 14 women, although noted that they gathered in formation from key informants during the first phase of their study and carried out inform al interviews with more than 30 additional informants. Kainer and Duryea (1992) noted that 35.0 % of these plants were located in the forest, thus we can estimate that wo men listed approximately 51 forest plants.54 In another study, Ming and Amaral Junior (n.d.) interviewed 53 indivi duals who identified 158 species in 61 plant families for medicinal purposes. Of these 158, approximately 65.0% were native to the Amazon region, w ith approximately 53.0%, or 84, plants collected from the forest , and 11.8% domesticated. My study included the particip ation of all household memb ers, so the number of total participants involved in providing the fr ee-lists was much greater than both of these studies. The free-list activity elicited approximately 47 fore st plant species, 37 of which are native to Acre, below the 51 estimat ed for Kainer and Duryea (1992) and 84 identified by Ming and Amaral Junior (n.d.).55 Thus, in comparison, it appears that the free-list activity with 44 households captured at least some of the variety of native forest plants used by rubber tappers. The number of items listed by households va ried from 2 to 18 items. The average free-list length of all house holds was 7.2 items (s.d. = 3.23). Table 5-1 above displayed all of the items listed, incl uding frequency. The most frequently listed item was aÃ§aÃ ( E. precatoria ) noted by 34 of 44 households. Second was castanha , or Brazil nut ( B. 53 Kainer and Duryea (1992) broke down plant use into eight categories: food, beverages, spices, medicines, animal feed, firewood, cons truction materials, and miscellaneous. 54 Other plant sites included compound, crop field, fallow field and garden/raised bed. 55 I have assumed that all of the estimated 51 plant sp ecies for Kainer and DuryeaÂ’s (1992) study that were collected in the forest are native.
189 excelsa ) listed by 29 households), followed by patauÃ¡ ( O. bataua ) by 28 households and bacaba ( O. mapora ) by 24 households. As noted above, my prior research in the Chico Mendes reserve on the three palm species ma y have influenced the listing of these species. Only four items were listed by at least 50.0 % of households while seven items were listed by 25.0 % of the households. Tw enty-nine items appeared on two or more free-lists, while 54 items were listed only once. Table 5-3 on the following page combin es all items relate d to a single common name. Most notable, copaiba ( Copiafera spp. ) listed previously as copaÃba , copaÃba oil and copaÃba bark was listed in some form by 11 households. Combining common name items results in 28 items listed by two or more households and 31 items listed by only one household. The results of the free list activity found that rubber tapper households use a diverse range of plants from many plant families. The frequency of listing of palm species reveals that pa lms are a particularly important plant family to rubber tapper households. Yet, households, as a whole, listed many items only once or twice, indicating that there may be considerable use differences across study households. And rubber (listed as borracha and seringa ), an important resource to generations of rubber tappers, was listed by just over half of th e households. These results suggest use differences among study households and this suggests that th ere may be differences in knowledge as well. This, in turn leads us to ask one of the centra l questions of this dissertation: do rubber tappers vary in their cu ltural knowledge of extractive resources? The remainder of this chapter and th e following respond to this question.
190 Table 5-3. Free list results with common name items combined and listed by frequency. Common Name Family Frequency AÃ§aÃ Aracaceae 34 Castanha Lecythidaceae 29 PatauÃ¡ Aracaceae 29 Bacaba Aracaceae 24 Borracha Euphorbiaceae 23 Bacuri Clusiaceae 17 Cacau Sterculiaceae 17 JatobÃ¡ Caesalpiniaceae 13 CipÃ³ timbÃ³ Cyclanthaceae 11 CopaÃba Caesalpiniaceae 11 CipÃ³ ambÃ© Araceae 9 JutaÃ Caesalpiniaceae 7 ArumÃ£ Marantaceae 7 Pama Moraceae 5 Mel de abelha NA 5 Jarina Aracaceae 5 Palhaa Aracaceae 4 Ouricuri Aracaceae 4 Quina-quina Rubiaceae/Apocynaceae 4 Cerejeira Fabaceae 4 IngÃ¡ Mimosaceae 3 BuritÃ Aracaceae 3 PaxiÃºba Aracaceae 3 MamÃ£o Caricaceae 2 Pau dÂ’Arco Bignoniaceae 2 Unha de gato Rubiaceae 2 Cajarana Anacardiaceae 2 Angico Mimosaceae 2 Note : aPalha (palm leaf), listed by 4 households is not directly related to a specific species. It is included under the palm family, Arecaceae. The Pile Sort Test The pile sort test tells us how i ndividual rubber tapper household members categorize items found in the domain of extr active forest resources. Rather than categorize items into pre-iden tified categories, the pile so rt test gives respondents the opportunity to make categories that match thei r individual understandi ng of the resources. This is very different from more typical ethno-botanical research where researchers typically break out various categories of pl ant use, (Kainer and Duryea 1992 as noted above) or slight variations which might incl ude identifying more sp ecific use categories such as dyes and tannins a nd poisons (Coe and Anderson 1996: 71-107) or the lumping
191 of categories, such as identifying a Â“technol ogicalÂ” category to describe species with general local use, including t ools, varnishes, poisons, resi ns and crafts (Johnston and Colquhoun 1996: 183). The pile sort test allows individuals to crea te categories not tied to a general theme, such as Â“use.Â” For example, one group of items may be sorted by location in the forest, while another group might represent the utility of the items, while another group might represent plants that fl ower during the same season. However, the discussion to follow reveals that use is ex tremely important in understanding how rubber tappers categorize non-tim ber forest resources. The Selection of Pile Sort Items In Chapter 2, I described in detail ho w the pile sort test was implemented. However, before discussing the results of th is activity I will briefly describe how I selected the items included in the pile-sort test. As noted above, the free-lis t activity produced a list of 83 items. Of these, I selected 24 items for the pile sort test. In general, after consolidating items that were actually the same item (not of the same common name) I chose those items most frequently listed. This meant that I consolidated casca de cerejeira with casca de cumaru de cheiro , as they are the same item, but considered copaÃba , Ã³leo de copaÃba , and casca de copaÃba as individual items. Of the 30 ite ms listed two or more times, 22 items were selected for the pile-sort test. I included all items listed three or more times except as follows: I eliminated palha as it did not refer to a specific plant; I did not include copaÃba , as I chose to include the more specific items, Ã³leo de copaÃba and casca de copaÃba ; as noted above, I chose to use only borracha , and not seringa , as these I believed were too similar to in clude both; I did not include buritÃ as previous research in the area found that it is scarce if not completely absent in the three study communities; I
192 decided to leave out paxiÃºba , a palm of which the trunk is commonly used for construction, as I had already decided to incl ude other palms. Five items were selected from the remaining items ( cajarana , palha de ouricuri , casca de cerejeira , unha de gato , and cajÃ¡ ). Thus, all items selected were listed at least two times except for unha de gato and cajÃ¡ . I decided to include unha de gato (also listed as cipÃ³ unha de gato ), a vine species with medicinal properties, to see how it was sorted with other vine items. CajÃ¡ was included as the fruit of this tree is a sister fruit of cajarana and is processe d into a juice and sold in the local market. I was inte rested in learning ho w rubber tappers would categorize this item with regard to other items that produce fruits, as well as income items. Results of the Pile Sort Test The pile sort data include the results of 118 individua l respondents. The large number of respondents provided the opportunity to analyze the pile sort results of subgroups of the sample to examine potential differences in how sub-groupsÂ—male and female, and youth, young adult and adult respon dentsÂ—categorize items in the domain of non-timber forest resources. In th is chapter I report th e results of the pile sort tests for all individuals and individuals categorized by sub-groups. In the following chapter I summarize the results of the pile sort test for individuals and subgroups across different levels of household wealth and market inte gration, thus responding to hypotheses H7 through H10. To examine the results of the pile sorts, I employed a number of analytical tools, including consensus analysis, multidimensiona l scaling (MDS), JohnsonÂ’s Hierarchical Clustering (hereon referred to as cluster an alysis), and quadratic assignment procedure
193 (QAP). Consensus analysis, discussed in Chapter 1, examines how knowledge is shared among individuals within a culture and in particular, the degree of knowledge each individual maintains (Romney, Weller and Batchelder 1986). Both MDS and cluster analysis output a figure that represents the cu ltural relationship of each item to all other items. MDS provides a multi-dimensional view of this relationship (Johnson and Griffith 1998) while cluster analysis tells us what groups cluster together as well as the strength of these clusters, i.e., the group relationshi ps. Finally, I employ QAP to examine the correlation, or similarity, of cultural knowledge of two s ub-groups within a culture. In the discussion that follows, in particular with regard to the analysis of the MDS and cluster analysis results, I will refer to all items in Port uguese rather than English, i.e., I will refer to borracha , rather than rubber, castanha rather than Brazil nut, Ã³leo and casca de copaÃba rather than copaÃba oil and bark , mel de abelha rather than honey, etc. I do this as all of the MDS scat terplots and cluster analyses results contain the Portuguese terms for the 24 pile sort items. This w ill hopefully make it easier for the reader to follow my interpretation of the figures. Initial Coding of Pile Sort Test Results The 118 individuals who completed the p ile-sort test represent forty-five households. Within these 45 households, the nu mber of individuals conducting pile sorts ranged from one to seven. Participants in cluded 73 males and 45 females, with ages ranging from 13 to 71 years. As noted in Chapter 2, two visits were made to each household, the first to carry out a structured interview, the second to implement the pile sort test. During this second vi sit, in some cases, an indivi dual 13 years or older declined to participate or was not at home. Ultimat ely, 32 individuals (of the 45 study households) who might have participated in the pile sort activity did not.
194 The pile-sort tests took place at the resi dence of respondents except in one case where the individual was working as a cook in the community center, so the test was conducted there. Although we did not note th e time each individual needed to complete the test, the time ranged from approximately 5 to 30 minutes. When the respondent indicated that they had finished the test, I asked them if they had any final changes to make. I then asked them to explain the logic behind their sorting. The number of Â“pilesÂ” or groups identified by individual respondents ranged from 2 to 19 groups. The average number of groups formed by respondents was 8.1, the mode 5.0, and the median 7.5. Sixty respondents or just over 50 % include d at least one item on its own, as a unique group. The number of unique items Â“sortedÂ” by these 60 respondents ranged from 1 to 16, meaning that one individual left 16 items without a partner. For those who created unique, or Â“singleton pilesÂ” (Bernard 1995: 250), this group was referred to in various ways, such as Â“separate,Â” Â“doesnÂ’t have a group,Â” Â“doesnÂ’t have a partner,Â” and Â“doesnÂ’t know wh ere it fits,Â” or Â“doesnÂ’t know what it is used for,Â” the latter indicating that it was se parated because of lack of knowledge of the item rather than lack of a partner. If th e respondent sorted more than one unique group, they would sometimes refer to these piles as Â“ todos separados Â” or Â“all separate.Â” In some cases, this single item group was given a name such as Â“medicine,Â” Â“good for remedy for diabetes,Â”56 Â“juice,Â” Â“not used at all, or Â“f or eating by monkeys.Â” In one case, the respondent placed four items into their own unique groups (i.e., sorted each separately) but called two of these unique gr oups Â“juiceÂ” and the ot her two Â“only serves to eat.Â” So, unique items were Â“uniqueÂ” for va rious reasons, not just because they didnÂ’t 56 The wife of this respondent was living in the c ity of Xapuri receiving treatment for diabetes.
195 have a partner. The item sorted alone most often was quina-quina , placed in its own unique group by 32 respondents, or over 25.0 % of pile sort participants. Unha de gato , arumÃ£ and borracha were each placed in their own unique group by 22 respondents. In contrast, Ã³leo de copaÃba was sorted alone only once, while castanha , jatobÃ¡ , jutaÃ and patauÃ¡ were placed in their own unique group three times (See Table 5-4 below). Asking respondents the names of the piles they created, i. e., the reason they sorted items together, was useful for understanding the item relationships that were identified in the results of the MDS and cluster analyses, to be presented below. By doing this, I was able to gain a much greater depth of unders tanding of the cognitive processes behind the sorted cards (cf. Bernard 1995: 252) and iden tify Â“themesÂ” emerging from this process (cf. Ryan and Bernard 2003). This process of discovering Â“themesÂ” was simplified as the respondents gave very short responses in naming their groups, ofte n only one or two words. However, some respondents gave multiple reasons for sorting item s together, such as Â“eat fruit, bark is medicinalÂ” and Â“we eat and serves as remedy.Â” In general, however, responses were very short and clear. To identify themes, I looke d for repetitive words or phrases in the pile sort names and gave each a unique code (Ryan and Bernard 2003). I began by establishing a few key codes a priori based on previous experience working and living with rubber tapper househol ds. These were: palm, vinho , eat, medicinal, income/sell, artisan and vine. However, as I began the c oding process I soon real ized that these codes were insufficient to capture the diversity of reasoning used to sort the 24 items. Nonetheless, these initial c odes ultimately became the higher-level themes that emerged from the coding process.
196 The coding process resulted in establishi ng 131 codes. I constructed a 118 by 24 respondent by item matrix in an ExcelÂ© spreadsheet and coded th e individualÂ’s response for each item. In a number of instances, multip le codes were inserted in the matrix to reflect the multiple meanings given to the item . For example, a pile could be named both Â“vinesÂ” and Â“medicinalÂ” by the respondent. I then recoded items to higher-level codes in which the original codes fell. For example, Â“remedy,Â” Â“syrupÂ” and Â“ lambedor Â” (i.e., home remedy) were recoded to the medicina l category; Â“makes brooms and baskets,Â” was recoded to the artisan category; Â“extractiv e product most soldÂ” was recoded to the sell/income/market category; Â“better known fru its,Â” Â“bunch of fruits,Â” and Â“nutsÂ” were recoded to the fruit category. Some items how ever, that one might infer are similar, I decided to keep separate. For example, a higher level Â“incomeÂ” code, did not include Â“things we work withÂ” and Â“productsÂ” or Â“p roductiveÂ” but did include codes such as, Â“give money,Â” Â“sell,Â” Â“goldÂ” or Â“support cost of life.Â” To recode responses in the ExcelÂ© matrix, I ran the find and replace command. By tallying up the most frequent responses of the 24 item columns, I calcu lated the mode code, i.e., mode response, for each item. Table 5-4 provides a summary of the results, showing each item, its mode code, the second most common code and the number of times the items was sorted as unique.57 Interestingly, for only three itemsÂ— casca de cerejeira , casca de copaÃba and cipÃ³ timbÃ³ Â—did the mode code represent over 50.0 % of the respondents. Thus, respondents 57 Many of the initial codes did not easily collapse into a higher-level code that I thought was on a comparable level to themes, or higher level codes noted in Table 5-4, such as eat, medicinal or artisan. Therefore, a number of initial codes were not recoded for calculating the mode response. Recoding initial codes such as Â“found in lowlands, stream edgeÂ” and Â“found in terra firm,Â” two very different forest environments, into a higher Â“forest locationÂ” code would not have changed the mode code for any of the 24 items.
197 did not overwhelmingly agree on what an item should be Â“named.Â” This does not mean that they did not sort items similarly, just that they have given them different names. Table 5-4. Frequency of higher-level code s for 24 pile sort items. (Number of respondents of 118 total in parentheses) Item Mode Code 2nd Most Frequent Code Unique "leo de copaÃba Medicinal (58) sell (26) 1 Cacau eat (54) fruit (20) 12 JatobÃ¡ Medicinal (54) eat (21) 3 Bacuri eat (52) fruit (23) 9 Unha de gato Medicinal (47) vine (14) 22 Casca de copaÃba Medicinal (68) sell (8) 6 Cajarana Juice (42) eat (36) 6 Quina quina Medicinal (55) eat (7)a 32 ArumÃ£ artisan (43) vine (13) 22 Bacaba Juice (44) eat (22) 5 CipÃ³ ambe artisan (56) vine (27) 5 Caja Juice (41) eat (36) 6 Palha de jarina cover house (49) palm leaf (36) 5 Castanha sell (50) eat (17) 3 AÃ§aÃ Juice (44) eat (22) 4 Casca de cerejeira Medicinal (76) work with (5)b 4 Mel de abelha Medicinal (51) sell (16) 9 Borracha sell (55) work with (9) 22 JutaÃ Medicinal (48) eat (23) 3 IngÃ¡ eat (53) fruit (19) 16 PatauÃ¡ Juice (44) palm leaf (21) 3 CipÃ³ timbÃ³ artisan (59) vine (26) 5 Palha de ouricuri cover house (47) palm leaf (31) 4 Pama eat (52) fruit (22) 19 Notes : aThe second most frequent code was Â“unnamed,Â” as the group in which the item was placed was not named by nine respondents. I have placed the next most frequent code in its place. bThe second most frequent code wa s a tie between Â“un-namedÂ” and Â“work with.Â” Throughout the rest of this chapter, as I analyze the results of the MDS and cluster analyses procedures, I will ofte n refer back to the raw pile sort data, i.e., how respondents identified items, to better understand what re spondents were Â“thinkingÂ” when they sorted items together. This will be particularly important when looking at results of sub-groups of individuals categorized by different levels of wealth and market integration. This
198 analysis will be presented in the following ch apter. In this analysis, I went beyond the conclusion that the results of sub-groups were similar to explore further if different subgroups were thinking the same thing when they sorted items similarly. This was especially useful in exploring the relations hip of rubber and Brazil nuts, two items that have a strong economic (and historical, in the case of rubber) im portance to reserve households. As noted in Chapter 3, the ec onomic importance of rubber is changing for many households, and understanding how i ndividuals within s ub-groups of items identified items will help reveal subtle di fferences in rubber tapper cultural knowledge. The analysis of the MDS and cluster anal yses results of sub-groups will have a particularly strong focus on these two items. I now turn to examining the results of the MDS, cluster analysis, QAP and consensus analysis procedures. I begin by an alyzing the results of all 118 respondents, followed by the results of males and females, and then the three generations of rubber tappers. Pile Sort Test Results For All Respondents To begin my analysis, I want to focus on th e aggregate proximity matrix that results from inputting the raw pile sort data. F ound in Table 5-5 on the following page, this matrix contains an item-by-item (24 x 24) matr ix that details the corr elation of each item with every other item. In other words, it tells us the strength of each item-to-item relationships based on the pile-s orts of all 118 individuals. I very briefly describe the strongest and weakest item correlations. Table 5-5 reveals that the strongest correlations are found among palm items. The strongest relationship is between aÃ§aÃ and bacaba , at 0.87, followed by palha de ouricuri
199Table 5-5. Aggregate proximity matrix genera ted from pile sort data of 118 respondents. OLEO CACAUJATOBBACURUNHA-CASCACAJARQUINAARUMABACABCIPO CAJAPALHACASTA ACAICASCAMEL-DBORRAJUTAI INGAPATA UCIPO PALHA PAMA -----------------------------------------------------------------------1 OLEO DE COPAIBA 1.00 0.08 0.29 0.08 0.19 0.53 0.06 0.21 0.05 0.06 0.08 0.03 0.02 0.30 0.06 0.45 0.58 0.31 0.25 0.05 0.0 6 0.05 0.03 0.04 2 CACAU 0.08 1.00 0.16 0.53 0.05 0.09 0.37 0.10 0.08 0.09 0.07 0.40 0.04 0.18 0.07 0.08 0.12 0.08 0.21 0.49 0.1 0 0.07 0.03 0.42 3 JATOBA 0.29 0.16 1.00 0.17 0.16 0.31 0.16 0.25 0.06 0.07 0.08 0.11 0.02 0.26 0.07 0.38 0.28 0.16 0.71 0.12 0.0 8 0.06 0.04 0.13 4 BACURI 0.08 0.53 0.17 1.00 0.08 0.13 0.39 0.09 0.05 0.10 0.08 0.42 0.04 0.14 0.08 0.10 0.12 0.08 0.22 0.41 0.0 8 0.07 0.04 0.53 5 UNHA-DE-GATO 0.19 0.05 0.16 0.08 1.00 0.24 0.05 0.30 0.21 0.01 0.31 0.08 0.04 0.04 0.04 0.30 0.19 0.06 0.17 0.14 0.0 3 0.28 0.03 0.06 6 CASCA DE COPAIBA 0.53 0.09 0.31 0.13 0.24 1.00 0.10 0.40 0.05 0.06 0.09 0.06 0.05 0.14 0.05 0.67 0.38 0.14 0.30 0.04 0.0 5 0.06 0.04 0.06 7 CAJARANA 0.06 0.37 0.16 0.39 0.05 0.10 1.00 0.08 0.05 0.09 0.10 0.71 0.04 0.21 0.08 0.08 0.12 0.08 0.17 0.39 0.1 3 0.06 0.05 0.35 8 QUINA-QUINA 0.21 0.10 0.25 0.09 0.30 0.40 0.08 0.99 0.08 0.01 0.09 0.05 0.04 0.08 0.01 0.38 0.17 0.07 0.26 0.10 0.0 2 0.08 0.04 0.09 9 ARUMA 0.05 0.08 0.06 0.05 0.21 0.05 0.05 0.08 1.00 0.08 0.58 0.06 0.14 0.05 0.07 0.03 0.03 0.09 0.05 0.08 0.0 8 0.58 0.13 0.03 10 BACABA 0.06 0.09 0.07 0.10 0.01 0.06 0.09 0.01 0.08 1.00 0.03 0.11 0.39 0.08 0.87 0.03 0.05 0.08 0.03 0.05 0.8 1 0.02 0.41 0.06 11 CIPO AMBE 0.08 0.07 0.08 0.08 0.31 0.09 0.10 0.09 0.58 0.03 1.00 0.05 0.08 0.09 0.03 0.06 0.08 0.10 0.09 0.09 0.0 3 0.82 0.06 0.08 12 CAJA 0.03 0.40 0.11 0.42 0.08 0.06 0.71 0.05 0.06 0.11 0.05 0.98 0.03 0.14 0.08 0.06 0.10 0.06 0.14 0.36 0.1 1 0.05 0.03 0.36 13 PALHA DE JARINA 0.02 0.04 0.02 0.04 0.04 0.05 0.04 0.04 0.14 0.39 0.08 0.03 1.00 0.03 0.41 0.03 0.03 0.02 0.03 0.05 0.4 2 0.07 0.85 0.02 14 CASTANHA 0.30 0.18 0.26 0.14 0.04 0.14 0.21 0.08 0.05 0.08 0.09 0.14 0.03 1.00 0.08 0.09 0.22 0.63 0.19 0.13 0.0 8 0.08 0.03 0.09 15 ACAI 0.06 0.07 0.07 0.08 0.04 0.05 0.08 0.01 0.07 0.87 0.03 0.08 0.41 0.08 1.00 0.05 0.06 0.07 0.03 0.03 0.8 2 0.03 0.41 0.05 16 CASCA DE CEREJEIRA 0.45 0.08 0.38 0.10 0.30 0.67 0.08 0.38 0.03 0.03 0.06 0.06 0.03 0.09 0.05 0.98 0.44 0.11 0.31 0.04 0.0 4 0.06 0.03 0.06 17 MEL-DE-ABELHA 0.58 0.12 0.28 0.12 0.19 0.38 0.12 0.17 0.03 0.05 0.08 0.10 0.03 0.22 0.06 0.44 0.98 0.21 0.21 0.11 0.0 5 0.08 0.04 0.08 18 BORRACHA 0.31 0.08 0.16 0.08 0.06 0.14 0.08 0.07 0.09 0.08 0.10 0.06 0.02 0.63 0.07 0.11 0.21 1.00 0.10 0.06 0.0 6 0.12 0.02 0.05 19 JUTAI 0.25 0.21 0.71 0.22 0.17 0.30 0.17 0.26 0.05 0.03 0.09 0.14 0.03 0.19 0.03 0.31 0.21 0.10 1.00 0.15 0.0 5 0.05 0.05 0.23 20 INGA 0.05 0.49 0.12 0.41 0.14 0.04 0.39 0.10 0.08 0.05 0.09 0.36 0.05 0.13 0.03 0.04 0.11 0.06 0.15 1.00 0.0 7 0.08 0.06 0.42 21 PATAUA 0.06 0.10 0.08 0.08 0.03 0.05 0.13 0.02 0.08 0.81 0.03 0.11 0.42 0.08 0.82 0.04 0.05 0.06 0.05 0.07 1.0 0 0.03 0.44 0.07 22 CIPO TIMBO 0.05 0.07 0.06 0.07 0.28 0.06 0.06 0.08 0.58 0.02 0.82 0.05 0.07 0.08 0.03 0.06 0.08 0.12 0.05 0.08 0.0 3 1.00 0.06 0.03 23 PALHA DE OURICURI 0.03 0.03 0.04 0.04 0.03 0.04 0.05 0.04 0.13 0.41 0.06 0.03 0.85 0.03 0.41 0.03 0.04 0.02 0.05 0.06 0.4 4 0.06 1.00 0.03 24 PAMA 0.04 0.42 0.13 0.53 0.06 0.06 0.35 0.09 0.03 0.06 0.08 0.36 0.02 0.09 0.05 0.06 0.08 0.05 0.23 0.42 0.0 7 0.03 0.03 1.00
200 and palha de jarinÃ¡ at 0.85, aÃ§aÃ and patauÃ¡ at 0.82, cipÃ³ timbÃ³ and cipÃ³ ambÃ© at 0.82 and bacaba and patauÃ¡ at 0.81. As noted a bove in Table 5-3, aÃ§aÃ , bacaba , palha de ouricuri , palha de jarinÃ¡ and patauÃ¡ are from the palm family, Araceceae. As noted in Chapter 3, the fruit of aÃ§aÃ , bacaba and patauÃ¡ can be used for making vinho , a thick juice drink. The palm leaves of ouricuri and jarina are used for thatch. CipÃ³ timbÃ³ and cipÃ³ ambÃ© are vines, both used for the production of artisan (and utilita rian) items such as baskets, sifters and brooms (t he brush head, not the handle). Other strong relationships include: jutaÃ and jatobÃ¡ (0.71), both from the Caesalpiniaceae family, and eaten and used for medicinal purposes; cajÃ¡ and cajarana (0.71), both of the genus, Spondias of the Anacardiaceae family, the fruits of which are used to make vinho ; casca de copaÃba and casca de cerejeira (0.67), both tree barks and used for medicinal purposes, and: castanha (Brazil nuts) and borracha (rubber) (0.63), items traditionally extracted and so ld by rubber tapper households, although castanha is also consumed. The items with the weak est correlations with other items were unha de gato , which correlated with cipÃ³ ambe at 0.31, and quina quina , which correlated with casca de copaÃba at 0.40. I now turn to examining the results of th e MDS and cluster analyses procedures. I spend considerable time describing and anal yzing these two initial figures for all 118 respondents, as they lay out the general pa tterns that are seen in subsequent MDS scatterplots and cluster analyses fi gures for sub-groups of participants. Figure 5-1 presents the two-dimensional MD S scatterplot for all 118 pile sort test participants. The MDS scatterplot is us eful for identifying not only item-to-item relationships, but also the gene ral grouping of items . In fact, Borgatti (1996b) cautions
201 not to read too much into the strength of item-to-item relationships among a group of items that are clumped together. The MDS pr ocedure also calculates the stress level related to the scatterp lot. Bernard (1995: 502) defines stress as the Â“measure of how far off the graph is from one that is perfectly proportional,Â” or what might be called a Â“goodness of fitÂ” indicator. In other words, th e lower the stress, the better the scatterplot fits the relationships found am ong all items. Bernard (personal comm.) notes that 0.30 should be considered the upper limit for a good fi t, while Borgatti ( 1996b) advises that an upper limit of 0.15 should be considered. All of the MDS scatterplots discussed in this and the following chapter had a stress level of less than 0.20, and most less than 0.15, so in general, they portray a pret ty good fit of ite m relationships. Dim 2 1.20 CIPO TIMBO ARUMA CIPO A MBE UNHA DE GATO 0.64 OLEO DE COPAIBA BORRACHA CASCA CASCA DE CEREJEIRA QUINA QUIN MEL DE ABELHA 0.08 PALHA DE OURICURI JATOBA CASTANHA JUTAI ACAI BACPATAUA 0.48 CACAU BACURI CAJARANA INGA CAJA 1.04 PAMA 1.10 0.60 0.10 0.41 0.91 Dim 1 Figure 5-1. MDS scatterplot for pile -sort test of 118 respondents. There is one other important idea to remember when analyzing the MDS scatterplots. In addition to looking for groups of items clumped together, dimensions, or what Borgatti (1996b: 35) terms Â“item attributes that seem to order the items in a map
202 along a continuum . . . that are thought to Â“explainÂ” the perceived similarity between itemsÂ” might also be found. Thus, two ite ms may be spatially spread across the scatterplot but still have a relationship along a dimension.58 Of course, items found at the far ends of one dimension, or opposite sides of the scatterplot, woul d not be considered strongly similar on this dimension, though they mi ght be related in some way, i.e., one is not considered medicinal, and one is cons idered highly medicinal. I will consider dimensions in greater detail below. I have chosen to display the items in two dimensions, as this is the simplest form for interpretation. A third dimension could be added with the output providing a third Â“depthÂ” dimension, thus giving a better idea of the similarity of items. I use a twodimensional scatterplot, as a third dimens ion, although Â“fittingÂ” the items better, would provide limited assistance to the reader in viewing the resu lts (Clark et al. 1998) while cluttering the plots with the c oordinates for a third dimension. The stress level for Figure 5-1 is 0.159, s uggesting a reasonable fit for the 24 items. Because of space limitations, some items ar e partially, and in some cases wholly, concealed by another item.59 As the MDS scatterplot is generated from the aggregate proximity matrix (Table 5-5), item relationships reflect these correlations. On the far left of the scatterplot figure is palha de ouricuri , underneath which are aÃ§aÃ , bacaba and patauÃ¡ . Palha de jarina is hidden beneath palha de ouricuri . Thus we can see a very 58 Dimensions do not necessarily run vertically and horizontally, but could also be diagonal. 59 It would be possible to display the names of all items in full by altering the size of the output scatterplot. This can be done when inputting the aggregate proximity matrix file into the AnthropacÂ® MDS command. However, this would result in a very large scatterplot, which, when reduced for placing in this document, would be very difficult to read. The scatterplots have been adjusted to a size that I thought best for presentation. The MDS procedure outputs the coordina tes (not included here) of the exact location of partially or fully hidden items.
203 strong relationship among all palm family items, and in particular the two palm subgroups: the palm leaf sub-group, and the remain ing three palms, the fruits of which are used for making vinho . The palms are clearly separate d from the rest of the items indicating they have weak ties with other items. At the top of the figure is the vine/artisan group: arumÃ£ , cipo timbÃ³ and cipÃ³ ambÃ©; cipo timbÃ³ and cipÃ³ ambÃ© . On the right side of the figure things are a bit messier. A total of eight items, including unha de gato , Ã³leo de copaÃba , casca de cerejeira , casca de copaÃba (partially hidden), quina quina , mel de abelha , j atobÃ¡ and jutaÃ appear to be part of one group. However, within this group there are clearly sub-groups of stronger relationships: jatobÃ¡ and jutaÃ are close together, as are Ã³leo de copaÃba , casca de cerejeira , casca de copaÃba , quina quina , and mel de abelha . Unha de gato , slightly above and to the left of this group appears to be pulled by its re lationship (although weak) with other vine species, cipÃ³ timbÃ³ and cipÃ³ ambÃ© . All of the items on the right have medicinal properties and this was the modal code for each of these items (See Table 5-4 above). At the bottom of the scatterplot, ju st to the right of the center are cajÃ¡ , cajarana , cacau , bacuri , ingÃ¡ and pama . All of these trees produce fruits that are consumed. CajÃ¡ and cajaranÃ¡ are processed into vinho , while the others are eaten mainly in their raw harvest form.60 Finally, in the center-right of the scatterplot are castanha and borracha . Castanha appears close to jatobÃ¡ and jutaÃ ; jatobÃ¡ and jutaÃ both produce a hard-shelled fruit, or nut, thus their association with castanha . Borracha , on the other hand sits somewhat alone, though its relationship with castanha is clear. 60 Cacau is also processed into a fruit drink and the fruitÂ’s seeds are used to make chocolate.
204 Figure 5-2 displays this same image with items grouped according to these relationships, but also based loosel y on how respondents identified items.61 Groups that can be identified are palms, vines/artisan, me dicinal, fruits, and in come, or products sold. Dim 2 1.20 CIPO TIMBO ARUMA CIPO AMBE UNHA-DE-GATO 0.64 OLEO DE COPAIBA BORRACHA CASCA CASCA DE CEREJEIRA QUINA-QUIN MEL-DE-ABELHA 0.08 PALHA DE OURICURI JATOBA CASTANHA JUTAI ACAI BACPATAUA -0.48 CACAU BACURI CAJARANA INGA CAJA -1.04 PAMA -1.10 -0.60 -0.10 0.41 0.91 Dim 1 Figure 5-2. MDS scatterplot for pile-sort te st of 118 respondents with items grouped. In addition to the clumping of items, I iden tified what appear to be two dimensions in the MDS scatterplot, which I will ca ll the Â“medicinalÂ” and Â“eatÂ” dimensions (See Figure 5-3 below). The medicinal dimension be gins on the left side of the scatterplot, with the palm cluster, items identified as leas t medicinal, and runs horizontally across the scatterplot, passing thro ugh the cluster of medicinal items on the right side, including, casca de cerejeira and quina quina among others. Items become Â“moreÂ” medicinal as you move from left to right. 61 I state Â“looselyÂ” as items can be sorted together fo r a number of reasons. For example, I have called one group Â“fruitsÂ” in Figure 5-2, although these items were identified and placed together for numerous reasons, such as Â“fruits,Â” Â“eaten,Â” and Â“juice,Â” among others. As noted in Table 5-4, the mode codes for these six Â“fruitÂ” items were not the same.
205 A second dimension, Â“eaten,Â” can also be identified. The Â“eatÂ” dimension starts at the top of the scatterplot a nd runs vertically down to th e bottom of the plot. Moving downward, items were increasingly identified as things eaten. In the figure, the vine items at the top of the scatterp lot, would be the least edible items, while fruit items, such as cajÃ¡ , pama and ingÃ¡ , found on the bottom, and would be the most edible. Items such as castanha , jatobÃ¡ and jutaÃ sit in the center of the dime nsion, and palm items on the far left side of scatterplot, including aÃ§aÃ , bacaba and patauÃ¡ , are also edible, although identified less so by respondents, than fruit items. PROFIT (PROperty FITting) analysis provid es a tool to test for the existence and strength of these dimensions across all item s (Borgatti 1996b). To do this, I gave each item a Â“ratingÂ” which was the number of times an item was identified as Â“medicinalÂ” or Â“eatenÂ” by the 118 respondents. I then car ried out the PROFIT analysis, using Anthropac.Â© PROFIT analysis, in short, conducts a regression analysis, regressing the coordinates of each of the MDS scatterplo t items (independent variables) onto the attribute ratings (dependent variable) (B orgatti 1996b). The output includes the calculation of the R2 for the model, which indicates how good a fit the map did in relation to the attribute score. The MDS scatterplot output with the PROFIT analysis arrays, which show the best fit of the attribute data, are presented on the following page in Figure 5-3. Both arrays run in the directions predicte d, the Â“medicinalÂ” horizontally, left to right and the Â“eatenÂ” vertically, top to bottom. The distance from the array is not important, nor on which side of the array an item lies. What is important is where each item lies along the array if connected to the array by a pe rpendicular line. Thus, as you move across the Â“medicinalÂ”
206 array, or down the Â“eatenÂ” array, each item is more Â“medicinalÂ” or more Â“eaten.Â” The R2 for the medicinal and eaten arrays, re spectively, are 0.66 and 0.80, indicating that coordinates for each item is a fairly good fit for the item ratings for each of these Â“attributes.Â” Borgatti (1996b) notes that an R2 of 0.80 is required to conclude that the data supports your hypothesis with less than 20 items. Hence, with 24 items, the PROFIT analysis arguably supports the hypotheses that Â“medicinalÂ” and, and to a lesser extent, Â“eaten,Â” attributes were shaping Â“perceive d similarity between itemsÂ” (Borgatti 1996b: 35). These dimensions are found in the MDS scatterplots for the male and female subgroups, although they become less clear when respondents are divided into age groups. I will discuss this further when presenting thes e figures in the following sections of the chapter. Dim 2 1.20 CIPO TIMBO ARUMA CIPO AMBE UNHA-DE-GATO 0.64 OLEO DE COPAIBA BORRACHA CASCAMEDASCA DECEREJEIRA QUINA-QUIN MEL-DE-ABELHA 0.08 PALHA DE OURICURI + JATOBA CASTANHA JUTAI ACAI BAPATAUA -0.48 CACAU BACURI CAJARANA INGA CAJA EAT -1.04 PAMA -1.10 -0.61 -0.12 0.37 0.86 Dim 1 Less Medicinal More Medicinal Less Eaten More Eaten Figure 5-3. PROFIT Analysis for all 118 re spondents testing medicinal and eaten attributes.
207 Figure 5-4 presents the cluster analysis for all 118 respondents. This figure complements the MDS scatterplot by providing both a visual and numerical display of the strength of the item relationships . The 24 item names (and corresponding identification number that was written on the re verse side of pile sort cards) are listed across the top of the figure. The Â“xÂ” wedge d into the space between columns of items notes when two items cluster together. The num ber on the left side of the figure specifies C P A A C S P L A C O A H S A L L A C E M H U A D O E A D N Q E L E C H U D D D C I A I E C E D E O I P C B -N E E C U P O A O D A C R C -A B P J R O S R E -O E O A J J B A A A I A T T R -Q P J P B A J A C A C A T R C R A I A A G U A E A E T U R C I A C P A C A I U U M M N C A I I I I L O T A A N C U A B A U N R M B B H H T N B R B H B A N J G A R M A I A A I A E O A A O A A A A A A I A A A U I A 1 1 2 1 2 1 2 1 1 1 1 1 1 2 2 Level 0 5 1 3 3 9 1 2 4 8 5 8 6 6 1 7 3 9 7 2 0 2 4 4 -----------------------------0.8729 XXX . . . . . . . . . . . . . . . . . . . . . . 0.8475 XXX . XXX . . . . . . . . . . . . . . . . . . . 0.8220 XXX . XXX . XXX . . . . . . . . . . . . . . . . 0.8192 XXXXX XXX . XXX . . . . . . . . . . . . . . . . 0.7119 XXXXX XXX . XXX . . . . . . . . XXX XXX . . . . 0.6695 XXXXX XXX . XXX . . . . XXX . . XXX XXX . . . . 0.6271 XXXXX XXX . XXX XXX . . XXX . . XXX XXX . . . . 0.5847 XXXXX XXX XXXXX XXX . . XXX XXX XXX XXX . . . . 0.5254 XXXXX XXX XXXXX XXX . . XXX XXX XXX XXX . XXX . 0.4887 XXXXX XXX XXXXX XXX . . XXX XXX XXX XXX . XXXXX 0.4388 XXXXX XXX XXXXX XXX . . XXXXXXX XXX XXX . XXXXX 0.4282 XXXXX XXX XXXXX XXX . . XXXXXXX XXX XXX XXXXXXX 0.4230 XXXXXXXXX XXXXX XXX . . XXXXXXX XXX XXX XXXXXXX 0.3763 XXXXXXXXX XXXXX XXX . . XXXXXXX XXX XXXXXXXXXXX 0.2966 XXXXXXXXX XXXXX XXX XXX XXXXXXX XXX XXXXXXXXXXX 0.2687 XXXXXXXXX XXXXX XXX XXX XXXXXXXXXXX XXXXXXXXXXX 0.2328 XXXXXXXXX XXXXX XXX XXXXXXXXXXXXXXX XXXXXXXXXXX 0.1515 XXXXXXXXX XXXXX XXXXXXXXXXXXXXXXXXX XXXXXXXXXXX 0.1191 XXXXXXXXX XXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0678 XXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0500 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX Figure 5-4. Hierarchical cluster an alysis for all 118 respondents. the strength of the relatio nship when two items Â“meet.Â” This level, in general, reflects the correlation value found in the aggregate prox imity matrix, although when groups of items are joining together, the co rrelation level between some items may be higher than
208 indicated. Examining where items come toge ther at the top of the figure reveals how strongly items correlate, i.e., the strength of the item in the cluster. Examining where groups of items come together at the bottom of the figure reveals the strength of the relationship of groups of items to other groups of items. 62 As seen in the MDS scatterplot, palms have the strongest relationship among all items. Bacaba and aÃ§aÃ cluster together at 0.8729 with patauÃ¡ joining shortly after. Palha de ouricuri and palha de jarina also have a strong re lationship, although they do not join the other palms until the 0.4230 level. The group of palm items does not join the other 19 items until the 0.05 level, noted by the sole Â“xÂ” connecting palha de ouricuri with arumÃ£ . This reveals the strong group relations hip among palms as well as the weak relationship the palms maintain with other items. The artisan/vine group is also tightly bound; cipÃ³ timbÃ³ and cipÃ³ ambÃ© have a strong relationship with arumÃ£ joining this pair at the 0.5847 level. Together as a cluster they join the palms at the 0.05 level and castanha at the 0.0678 level, indicating weak ties with these items. The six items on the right si de of the figure, all of which produce fruits and are consumed, also form a strong cluster. CajÃ¡ and cajarana display a particularly strong relationship, joining at the 0.7119 level and c onnecting to the other fruits at the 0.3763 level. This group only joins the group of medicinal items (see below) at the 0.1191 level. In the center of the figure sits a large cluster, bordered by castanha and jutaÃ , which includes strong sub-clusters of items. Borracha and castanha have a strong relationship, 62 It is important to remember that the cluster analys is procedure provides the be st fit to demonstrate how items cluster together and the strengths of the items within the clusters; however, this does not mean that items in a cluster do not correlate with other items outside the cluster.
209 joining together at the 0. 6271 level and connecting with unha de gato and other medicinal items only at the 0.1515 level. Thei r joining of the medicinal group before any other items (i.e., non-medicinal items) appe ars to be a product of its ties with Ã³leo de copaÃba and mel de abelha , which were also identified as income items (though less frequently than as medicinal ite ms) during the pile sort test. JutaÃ and jatobÃ¡ also form a pair, joining other items in the medicinal clus ter near the bottom of the figure. A strong sub-cluster also exists among Ã³leo de copaÃba , mel de abelha, casca de cerejeira and casca de copaÃba. Unha de gato and quina quina have a weak relationship with each other as well as all other items. The cluster analysis reveals four major it em groups Â– fruits consumed, medicinal, vines/artisan and palms; however borracha and castanha maintain a very strong independent cluster within the medicinal group. Examination of the cluster analyses of sub-groups will reveal that borracha and castanha form their own cluster in some subgroups, or cluster with a group of ite ms other than the medicinal group. Together, the tight grouping, or clusteri ng, of items in the MDS scatterplot and cluster analysis reveals that rubber tapper re spondents think about these items in very similar ways and suggests a high level of agreement among respondents regarding Â“what goes with what.Â” But just how strong is the agreement among rubber tappers? Consensus analysis helps us better unders tand the uniformity of rubber tapper cultural knowledge on this domain. Consensus Analysis As noted in Chapter 1, consensus analys is examines the degree of agreement among the respondents, and calculates the know ledge of each participant based on what are the Â“culturally correctÂ” answers (Weller 1987: 179). Ro mney, Weller and Batchelder
210 (1986: 316) explain that if we assume there is a high degree of shared information on a domain, and that knowledge on the domain vari es, then we can identify how competent each individual is in terms of their knowle dge. By knowing these competency levels, the answer key can be developed Â“by weighting each informantÂ’s input and aggregating to the most likely answer,Â” with those most knowledgeable carrying more weight in determining what is the Â“culturally correctÂ” response. The results of the consensus analysis pro cedures with all 118 respondents revealed that there is strong consensu s on the domain of non-timber forest resources among study participants. Table 5-6, cont aining the procedure output, indicates that three factors influenced rubber tappers in carrying out the pile-sorting activity. Ho wever, as factor 1 had a value over three times (in this case near ly 19 times) the value of factor 2 (see the ratio column), there is strong Â“cultural leve l of agreementÂ” on this domain (Caulkins 1998: 179), and therefore, the respon dents represent a single culture. Table 5-6. Consensus analysis ei genvalues for all 118 respondents Factor Value Percent Ratio 1 76.185 91.5 18.824 2 4.047 4.9 1.341 3 3.017 3.6 83.250 100.0 I will use the results of consensus analysis in the following chapter in calculating gamma to test how wealth and market integra tion influence knowledge scores. But first I will analyze the pile sort test results of all respondents categorized by sex (males and females) and age (young-young adult-adult). Male and Female Cultural Knowledge Prior research in the Chico Mendes Re serve found that men and women rubber tappers carry out different roles in th e forest household (Campbell 1996, Kainer and
211 Duryea 1992) and work with different plan t materials (Kainer and Duryea 1992). Do these different roles and experiences with plants lead to differences in cultural knowledge? By separating the respondents by sex, I test whether these different roles and contact with plants reveal themselves in the cultura l knowledge of these items. Table 5-7 provides a breakdown of the 118 re spondents by sex and age. Forty-five females and 73 males participated in the pile sort test. In addition to examining the separate results of MDS and cl uster analysis proce dures, I also ran the QAP analysis to statistically test for knowledge differences between these sub-groups. The QAP analysis showed a very high and statistically significant correlation (0.957, p-value = 0.00) between male and female respondents. Thus males and females maintain a very high degree of shared knowledge on this domai n. This is not surprising given the high consensus found among all 118 respondents. Yet, a closer look at the results of MDS and cluster analysis reveals subtle differences in regards to how the different sexes think about extractive resources. Table 5-7. Categorization of 118 respondents by sex and age. Age 1 (< 25) Age 2 (25 Â– 49) Age 3 ( > 50) Total Males 24 31 18 73 Females 12 25 8 45 Total 36 56 26 118 Figures 5-5 and 5-6 display the result s of the MDS procedure for males and females.63 Although the two figures initially appear very different with regard to item relationships, spatially the items are similarl y placed. Both scatterplots show a strong grouping of the five palm items, with sub-groups of palm leaves and the three palms that produce a juice drink. For both males and fe males, the six fruit items are clumped 63 The stress values for the male and female MDS scatterplots were 0.159 and 0.152, respectively.
212 together, as are the vine/artisan items. Again, unha de gato is pulled between the vine/artisan group and th e medicinal group, and j atobÃ¡ , jutaÃ and castanha form a Â“nutÂ” sub-group. One notable differe nce is the separation of borracha from castanha . In the female MDS scatterplot, borracha is more distant from castanha . However, an examination of the aggregate proximity matri ces of both males and females reveals that both sexes have a correlation of 0.62 between these two items. The Â“medicinalÂ” and Â“eatÂ” dimensions are found in both scatterp lots: the medicinal dimension moves horizontally from right to left for females, and left to right for males, while the Â“eatÂ” dimension moves vertically, from top to bottom for females, and bottom to top for males. The PROFIT analysis calculated robust R2 for both females and males for both the medicinal and eaten attributes. For males, the R2 for the medicinal attribute was 0.64 while that for the eaten attr ibute was 0.79. For females, the R2 for the medicinal attribute was 0.61 and for the eaten attribute, 0.70. The findings suggest that among both males and females the eaten attribute was infl uencing the sorting of items more than the medicinal attribute. Figures 5-7 and 5-8 on the following pages present the cluster analyses for female and male respondents. The clusters of items identified in the MDS s catterplots can easily be found in the cluster analys es, although subtle differences regarding the strength of relationships can be found. First, comparing th e palm clusters for males and females, the female cluster displays a part icularly strong relationship am ong the three palms used to make vinho , while the males shows a stronger relati onship (albeit slight) between the two palm leaves. This reflects the findings of Kainer and Duryea (1992) that women do the
213 food processing in the household, while me n carry out different activities including building of landholding structures, which woul d include the use of palms for thatching. Dim 2 1.46 PAMA 0.90 EAT CAJA INGA CACAU CAJARANABACURI 0.33 BPATAUA ACAI CASTANHA JUTAI PALHA DE OURICURI + JATOBA PALHA DE JARINA -0.23 QUINA-QUINA BORRACHA MEL-DE-ABELHA CASMEDSCA DE CEREJEIRA OLEO DE COPAIBA -0.80 ARUMA UNHA-DE-GATO CIPO AMBE CIPO TIMBO -1.12 -0.57 -0.02 0.53 1.08 Dim 1 More Eaten Less Eaten Less Medicinal More Medicinal Figure 5-5. MDS and PROFIT analysis for 73 male respondents with medicinal and eat attributes. Dim 2 1.46 CIPO TIMBO 0.90 ARUMA BORRACHA CIPO AMBE UNHA-DE-GATO CASCA DE CEREJEIRA QUINA-QUINA 0.33 MED CASCA DE COPAIBA PALHA DE JARI OLEO DE COPAIBA PALHA DE O MEL-DE-ABELHA + JATOBA JUTAI -0.23 ACAI CASTANHA PATAUA BACABA BACURI PAMA CACAU -0.80 CAJARANA INGA EAT CAJA -0.86 -0.37 0.13 0.63 1.12 Dim 1 Less Medicinal More Medicinal Less Eaten More Eaten Figure 5-6. MDS and PROFIT analysis 45 fema le respondents with medicinal and eat attributes.
214 The cluster analyses figure also supports Kainer and DuryeaÂ’s (1992) argument that women, in particular, maintain important knowledge of medicinal plants. Females clustered numerous items in the medicinal group stronger than male s: pairings between Ã³leo de copaÃba and mel de abelha , and casca de copaba and casca de cerejeiraÂ¸ as well C P A A C S P L A C O A H S A L L A C E M U H A D O E N A D Q E L H C E U D D A C I D I E C E D I P E O C B N E E C D P O U A O A C R C A E O B P J R S R O E O A J J B A T A A A I T R Q P J P B A J A C A G R A I C A T R C A A U A E A E T U R C A C I P A U M M A C A I U N C I I I I L O T A A C U N A T M B B B A U N R H H N B R B H B A N J A R G M O A E O A I A A I A A A A A A A A I A A U I A A 1 2 1 1 2 1 2 1 1 1 1 1 1 2 2 Level 5 9 1 2 0 5 1 3 3 4 8 8 6 6 1 7 3 9 7 2 2 4 0 4 -----0.8889 . . . . XXX . . . . . . . . . . . . . . . . . . 0.8296 . . . . XXXXX . . . . . . . . . . . . . . . . . 0.8000 . . XXX XXXXX . . . . . . . . . . . . . . . . . 0.7778 . . XXX XXXXX XXX . . . . . . . . . . . . . . . 0.7111 . . XXX XXXXX XXX . . . . . . . . . XXX . . . . 0.6889 . . XXX XXXXX XXX . . . XXX XXX . . XXX . . . . 0.6667 . . XXX XXXXX XXX . . . XXX XXX XXX XXX . . . . 0.6222 . . XXX XXXXX XXX XXX . XXX XXX XXX XXX . . . . 0.6000 . . XXX XXXXX XXX XXX . XXX XXX XXX XXX XXX . . 0.5926 . XXXXX XXXXX XXX XXX . XXX XXX XXX XXX XXX . . 0.5111 . XXXXX XXXXX XXX XXX . XXX XXX XXX XXX XXX XXX 0.4741 . XXXXX XXXXX XXX XXX . XXX XXX XXX XXX XXXXXXX 0.4741 . XXXXX XXXXX XXX XXX XXXXX XXX XXX XXX XXXXXXX 0.3815 . XXXXX XXXXX XXX XXX XXXXXXXXX XXX XXX XXXXXXX 0.3723 . XXXXX XXXXXXXXX XXX XXXXXXXXX XXX XXX XXXXXXX 0.3695 . XXXXX XXXXXXXXX XXX XXXXXXXXX XXX XXXXXXXXXXX 0.2589 . XXXXX XXXXXXXXX XXX XXXXXXXXXXXXX XXXXXXXXXXX 0.2556 XXXXXXX XXXXXXXXX XXX XXXXXXXXXXXXX XXXXXXXXXXX 0.1510 XXXXXXX XXXXXXXXX XXX XXXXXXXXXXXXXXXXXXXXXXXXX 0.1011 XXXXXXX XXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0806 XXXXXXXXXXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0587 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX Figure 5-7. Hierarchical cluster an alysis for 45 females respondents. as the single item quina quina demonstrated stronger clus tering within this group. JatobÃ¡ and jutaÃ showed a slightly str onger cluster among male resp ondents, although they were similar.
215 Close inspection of the borracha/castanha cl uster for the different sexes also reveals a slight difference. For females, the casta nha/borracha has separa ted itself from the medicinal group, creating its own group. For male s, the borracha/castanha cluster joins the medicinal group before joining the fruit group. C P A A C S L P A C O H A S A L A L C E M H U A D O E D A N Q E L E H C U D D D A C I I E C E D O E I P C B N E E C U D P O A O A C R C A B P R J E O S R O E O A J J B A A I A A T T R Q P J P B A J A C A C A T C R G R A I A A U A E A E T U R C A I C P A C A U I A U M M N C I I I I L O T A A C N U A B A U R N T M B B H H N B R B H B A N J A G R M A I A I A O A E O A A A A A A A A I A A U A I A 1 1 2 2 1 1 2 1 1 1 1 1 1 2 2 Level 0 5 1 3 3 5 9 1 2 4 8 8 6 6 1 7 3 9 7 2 2 0 4 4 -----0.8767 . . . XXX . . . . . . . . . . . . . . . . . . . 0.8493 XXX . XXX . . . . . . . . . . . . . . . . . . . 0.8219 XXX . XXX . . XXX . . . . . . . . . . . . . . . 0.7991 XXXXX XXX . . XXX . . . . . . . . . . . . . . . 0.7397 XXXXX XXX . . XXX . . . . . . . XXX . . . . . . 0.6986 XXXXX XXX . . XXX . . . . . . . XXX XXX . . . . 0.6438 XXXXX XXX . . XXX . . . XXX . . XXX XXX . . . . 0.6164 XXXXX XXX . . XXX XXX . XXX . . XXX XXX . . . . 0.5799 XXXXX XXX . XXXXX XXX . XXX . . XXX XXX . . . . 0.5342 XXXXX XXX . XXXXX XXX . XXX . . XXX XXX . . XXX 0.5068 XXXXX XXX . XXXXX XXX . XXX XXX XXX XXX XXX XXX 0.4542 XXXXXXXXX . XXXXX XXX . XXX XXX XXX XXX XXX XXX 0.4277 XXXXXXXXX . XXXXX XXX . XXXXXXX XXX XXX XXX XXX 0.3866 XXXXXXXXX . XXXXX XXX . XXXXXXX XXX XXX XXXXXXX 0.3744 XXXXXXXXX . XXXXX XXX . XXXXXXX XXX XXXXXXXXXXX 0.2808 XXXXXXXXX XXXXXXX XXX . XXXXXXX XXX XXXXXXXXXXX 0.2730 XXXXXXXXX XXXXXXX XXX . XXXXXXXXXXX XXXXXXXXXXX 0.2213 XXXXXXXXX XXXXXXX XXX XXXXXXXXXXXXX XXXXXXXXXXX 0.1622 XXXXXXXXX XXXXXXX XXXXXXXXXXXXXXXXX XXXXXXXXXXX 0.1086 XXXXXXXXX XXXXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0737 XXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0421 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX Figure 5-8. Hierarchical cluster an alyses for 73 male respondents. Two points may help explain this. First, data from females exhibit a stronger relationship between Ã³leo de copaÃba and mel de abelha than males and this may be due to their more frequent use of these items fo r medicinal purposes. Examining the pile-sort data finds that a greater pr oportion of females identified Ã³leo de copaÃba (26 of 45) and
216 mel de abelha (24 of 45) as medicinal items than male respondents (32 of 73 and 27 of 73, respectively). Conversely, a grea ter proportion of males identified Ã³leo de copaiba (19 of 73) and mel de abelha (11 of 73) as income items than females (7 of 45 and 5 of 45, respectively). Second, a grea ter proportion of males identified castanha (33 of 73) and borracha (36 of 72) as income items than females (17 of 45 and 19 of 45, respectively). Thus, a strong income associ ation among these items by males would help create stronger ties among Ã³leo de copaiba , mel de abelha , borracha and castanha , and this may explain why they are found in the same cluster. Notwithstanding these differences, males and females maintain a high degree of shared knowledge of extractive resources. This is shown statistically in the results of the QAP procedure, and in the similarities of the MDS scatterplots and cluster analyses figures. The PROFIT analysis found that the Â“medicinalÂ” a nd Â“eatenÂ” qualities of items helped shape how both female and male res pondents sorted items, with the medicinal dimension displaying a particularly good fit for both sexes. But although the results for females and ma les are similar, slight variations in knowledge were identified through a closer ex amination of how individuals identified items. Underneath a similar visual display of items, males identified not only traditional extractive itemsÂ—rubber and Brazil nutsÂ—as in come items in greater proportion than females, and also items considered to be medicinalÂ—copaÃba oil and honeyÂ—as income items. These findings reflect different gender roles in resource use in the forest (i.e., Kainer and Duryea 1992), but also suggest that government policies, such as the state governmentÂ’s efforts to promote the extraction and sale of copaÃba oil, may be reshaping cultural knowledge in different ways for different groups.
217 Age and Cultural Knowledge As noted above, Campbell (1996) found that a growing number of mothers do not want their children to tap rubber, while Es teves (n.d.) argued that a new generation of rubber tappers holds different ideas about th e forest and its resources. To test for differences in cultural knowledge across generations of rubber tappers, I divided respondents into three age categor ies: less than 25 years, 25 to 49 years, and 50 years and older (See Table 5-7 above). These cate gories were identified in the same manner employed to divide households by wealth and ma rket integration, as noted in Chapters 3 and 4. I produced a histogram of all 118 respondents by age, ranked from youngest to oldest, and then examined the figure for na tural breaks in the age of respondents to capture the generational changes, at the same time attempting to keep a minimum number of respondents in each age category.64 Despite age differences, the results s howed a high degree of shared knowledge across the three generations of rubber tappers. The QAP analysis found strong and statistically significant corre lations among all three age group combinations: age 1 and age 2 correlated at 0.92, age 1 and age 3 at 0.86, and age 2 and age 3 at 0.93 (p-value = 0.00 for all three QAP procedur es). The MDS scatterplots65 and cluster analysis figures for the three age categories, found in Figures 5-9 through 5-14, also displayed the high degree of shared knowledge across generations. As found in the above MDS scatterplots for all respondents, and those for males a nd females, a general grouping of itemsÂ—palms, 64 The last age category, 50 and older, has only 26 respondents. This is a small number in comparison to the other two categories. However, it was important to establish a category that would capture an older generation of rubber tappers. Thus I have sacrificed an equally divided population to better capture three generations of rubber tappers. 65 Stress values for the two-dimensional MDS scatterplots for age categories 1, 2 and 3, were 0.166, 0.142, and 0.139, respectively.
218 Dim 2 1.22 EAT INGA 0.73 CACAUBACURI CAJA CIPO AMBE CIPO TIMBO CAJARANA PAMA UNHA-DE-GATO ARUMA 0.24 JUTAI + JATOBA CASTANHA QUINA-QUINA -0.26 PALHA DE OU MEL-DE-ABELHA CASCA MEDCEREJEIRA PATAUAABA -0.75 ACAI CASCA DE COPAIBA OLEO DE COPAIBA BORRACHA -0.81 -0.37 0.06 0.49 0.93 Dim 1 Figure 5-9. MDS and PROFIT analysis for 36 respondents in the youngest age group (age 1) with medicinal and eat attributes. Dim 2 1.30 PAMA CAJA EAT INGA 0.77 CACAU CAJARANA BACURI PATAUA BACABA 0.24 JUTAI ACAI JATOBA CAST+NHA PALHA DE OURICURI PALHA DE JARINA MEL-DE-ABELHA -0.29 CACASCA DE CEREJEIRA OLEO DE COPAIBA MED QUINA-QUINA BORRACHA -0.82 UNHA-DE-GATO CIPO AMBE ARUMA CIPO TIMBO -1.15 -0.65 -0.14 0.37 0.87 Dim 1 Figure 5-10. MDS and PROFIT analysis fo r 56 respondents in the young adult age group (age 2) with medicina l and eat attributes.
219 Dim 2 1.30 BORRACHA 0.75 BACABA CASTANHA ACAI EAT JATOBA OLEO DE COPAIBA MEL-DE-ABE PATAUA BACURI 0.20 CASCA PAMA CAJARANA CASCA DMEDOPAIBA CAJA + JUTAI PALHA DE OURICURI PALHA DE JARINA INGA -0.35 CACAU UNHA-DE-GATO QUINA-QUINA -0.91 ARUMA CIPO AMBE CIPO TIMBO -1.15 -0.65 -0.14 0.37 0.87 Dim 1 Figure 5-11. MDS and PROFIT analysis for 26 respondents in the adult age group (age 3) with medicinal and eat attributes. vine/artisan, medicinal, and fruitsÂ—is present in all age categories. Yet, a closer examination reveals subtle, though important, distinctions across the three age groups. First, in the youngest age groupÂ’s scatte rplot and cluster analysis figures, jutaÃ and jatobÃ¡ have moved from the medicinal group and are now found closer (or clustered with) the fruit group. An examination of how respondents identified these items finds that moving from the youngest age group to the oldest age group, respondents proportionally decrease their identi fication of these items as fru its or things to eat, and increase their identification of them as medi cinal items. This would corroborate findings by Kainer and Duryea (1992); younger household members are more likely to be eating these items in the forest, and arguably, less li kely to be familiar with their medicinal properties.
220 C P A A C S P L A C O A H S A L L A C E M U H A D O E N A D E L Q H C E D D -U A C I D E C E D I -I P E O E E C B N C D P O U C R C -A O A A E O B P J R O E O A S R -J J B -A T A A A I P J P B T R Q A J A C A G R A I C A T R C A E A E A A U T U R C I A C P A U M M A C A I U I I I L N C I O T A A N C U A T M B B B A U N R B R B H H H N B A N J G A R M O A E O A I A A I A A A A A A A A I A A A U I A 1 2 1 1 2 1 2 1 1 1 1 1 1 2 2 Level 5 9 1 2 0 5 1 3 3 6 6 1 7 4 8 8 3 9 7 2 0 2 4 4 ----------------------------0.8333 . . . . XXX . XXX . . . . . . . . . . . . . . . 0.8056 . . XXX XXX . XXX . . . . . . . . . . . . . . . 0.7963 . . XXX XXXXX XXX . . . . . . . . . . . . . . . 0.7222 . . XXX XXXXX XXX . . . . . . . XXX XXX . . . . 0.5833 . . XXX XXXXX XXX XXX . . . . . XXX XXX . . . . 0.5556 . . XXX XXXXX XXX XXX . . . . . XXX XXX . XXX . 0.5000 . . XXX XXXXX XXX XXX . . XXX . XXX XXX . XXX . 0.4722 . . XXX XXXXX XXX XXX XXX XXX . XXX XXX . XXXXX 0.4444 . XXXXX XXXXX XXX XXX XXX XXX . XXX XXX . XXXXX 0.4185 . XXXXX XXXXX XXX XXX XXX XXX . XXX XXX XXXXXXX 0.4074 . XXXXX XXXXX XXX XXXXXXX XXX . XXX XXX XXXXXXX 0.3958 XXXXXXX XXXXX XXX XXXXXXX XXX . XXX XXX XXXXXXX 0.3846 XXXXXXX XXXXXXXXX XXXXXXX XXX . XXX XXX XXXXXXX 0.2906 XXXXXXX XXXXXXXXX XXXXXXX XXX . XXX XXXXXXXXXXX 0.2037 XXXXXXX XXXXXXXXX XXXXXXX XXX XXXXX XXXXXXXXXXX 0.1728 XXXXXXX XXXXXXXXX XXXXXXX XXX XXXXXXXXXXXXXXXXX 0.1595 XXXXXXX XXXXXXXXX XXXXXXXXXXX XXXXXXXXXXXXXXXXX 0.0940 XXXXXXX XXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0862 XXXXXXXXXXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0425 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX Figure 5-12. Hierarchical cl uster analysis for 36 responde nts in youngest age group (age 1). Second, unha de gato has also moved from a close a ssociation with the vine/artisan cluster among the youngest respondents (age 1) , to firmly part of the medicinal group among the oldest respondents (age 3). Agai n, how they identified these items helps explain this movement; a greater pro portion of age 1 respondents identified unha de gato as a vine (a morphological char acteristic) or artisan item than age 2 and 3 respondents.
221 C P A A C S P L A C O A H S A L L A C E M H U A D O E A D N Q E L E H C U D D D A C I I E C E D E O -I P C B N E E C U D P O A O A C R C -A B P J R E O S R -O E O A J J B A A A I -A T T R Q P J P B A J A C A C A T R C G R A I A A U A E A E T U R C I A C P A C A I U A U M M N C I I I I L O T A A N C U A B A U N R T M B B H H N B R B H B A N J G A R M A I A A I O A E O A A A A A A A A I A A A U I A 1 1 2 1 2 1 2 1 1 1 1 1 1 2 2 Level 0 5 1 3 3 5 9 1 2 4 8 8 6 6 1 7 3 9 7 2 0 2 4 4 ----------------------------0.9107 XXX . . . . . . . . . . . . . . . . . . . . . . 0.8571 XXX . XXX . . . . . . . . . . . . . . . . . . . 0.8393 XXXXX XXX . . XXX . . . . . . . . . . . . . . . 0.7679 XXXXX XXX . . XXX . . . XXX . . XXX . . . . . . 0.7143 XXXXX XXX . . XXX . . . XXX . . XXX XXX . . . . 0.6607 XXXXX XXX . . XXX XXX . XXX XXX XXX XXX . . . . 0.6190 XXXXX XXX . XXXXX XXX . XXX XXX XXX XXX . . . . 0.5714 XXXXX XXX . XXXXX XXX . XXX XXX XXX XXX . XXX . 0.4536 XXXXXXXXX . XXXXX XXX . XXX XXX XXX XXX . XXX . 0.4524 XXXXXXXXX . XXXXX XXX XXXXX XXX XXX XXX . XXXXX 0.4119 XXXXXXXXX . XXXXX XXX XXXXX XXX XXX XXX XXXXXXX 0.3929 XXXXXXXXX . XXXXX XXX XXXXXXXXX XXX XXX XXXXXXX 0.3820 XXXXXXXXX . XXXXX XXX XXXXXXXXX XXX XXXXXXXXXXX 0.2883 XXXXXXXXX . XXXXX XXX XXXXXXXXXXXXX XXXXXXXXXXX 0.2679 XXXXXXXXX XXXXXXX XXX XXXXXXXXXXXXX XXXXXXXXXXX 0.1717 XXXXXXXXX XXXXXXX XXXXXXXXXXXXXXXXX XXXXXXXXXXX 0.1394 XXXXXXXXX XXXXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0804 XXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0476 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX Figure 5-13. Hierarchical clus ter analysis for 56 respondents in adult group (age 2). Conversely, a greater proportion of age 3 respondents identified this item for its medicinal properties than age 1 and 2 respondents. A third difference among age categories, regarding the borracha / castanha relationship, may be the most interesting. Both the MDS scatterplots and cluster analyses figures display a strengthening of this re lationship moving from youngest (age 1) to oldest (age 3) respondents. Thus, younger re spondents found other items with which to
222 C P A A C S L P O A C H A L S A A L E C M H U O A D E D A N Q E L E C H U D D D C I A I E E C D O E I P C B -N E E C U P O A O D A C C R -A B P R J O S R E -O O E A J J B A A I A A T T R -Q P P J B A J C A A C A T C R R A I A A G U A A E E T U A I R C C P A C A U I U M M N C A I I I I L O T C N A A U A B A U R N M B B H H T N B B R H B A A G N J R M A I A I A A E O A A O A A A A A A I U A A A I A 1 1 2 2 1 1 2 1 1 1 1 1 2 1 2 Level 0 5 1 3 3 9 1 2 4 8 5 8 1 6 6 7 3 9 2 0 7 2 4 4 -----------------------------0.8462 XXX . XXX . . . . . . . . . . . . . . . . . . . 0.8077 XXXXX XXX . XXX . . . . . . . . . . . . . . . . 0.7308 XXXXX XXX . XXX XXX . . . . . . . . . . . . . . 0.7051 XXXXX XXX XXXXX XXX . . . . . . . . . . . . . . 0.6923 XXXXX XXX XXXXX XXX . . . . . . . . . . XXX . . 0.6538 XXXXX XXX XXXXX XXX . . . . . . . . XXX XXX XXX 0.6154 XXXXX XXX XXXXX XXX . . XXX . . . . XXX XXX XXX 0.5769 XXXXX XXX XXXXX XXX . . XXX . . XXX XXX XXX XXX 0.5513 XXXXX XXX XXXXX XXX . . XXXXX . XXX XXX XXX XXX 0.5085 XXXXX XXX XXXXX XXX . . XXXXX . XXX XXX XXXXXXX 0.5000 XXXXX XXX XXXXX XXX XXX XXXXX . XXX XXX XXXXXXX 0.4436 XXXXX XXX XXXXX XXX XXX XXXXXXX XXX XXX XXXXXXX 0.4387 XXXXX XXX XXXXX XXX XXX XXXXXXX XXX XXXXXXXXXXX 0.4103 XXXXXXXXX XXXXX XXX XXX XXXXXXX XXX XXXXXXXXXXX 0.3568 XXXXXXXXX XXXXX XXX XXX XXXXXXXXXXX XXXXXXXXXXX 0.3037 XXXXXXXXX XXXXX XXX XXXXXXXXXXXXXXX XXXXXXXXXXX 0.1594 XXXXXXXXX XXXXX XXXXXXXXXXXXXXXXXXX XXXXXXXXXXX 0.0891 XXXXXXXXX XXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0697 XXXXXXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX 0.0598 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX Figure 5-14. Hierarchical clus ter analysis for 26 respondents in adult group (age 3). identify borracha . An examination of how responde nts identified these items reveals important differences. First, not only did a smaller proportion of age 1 respondents sort these items together, but, when they placed them together, a smaller proportion tied them together as income items. Of age 1 responde nts, 12 of 36 identified them together as income items, compared to 23 of 56 of age 2 respondents, and 12 of 26 age 3
223 respondents.66 Second, and concomitantly, the pro portion of respondents who identified borracha as an income item fell (albeit very slightly) moving from older to younger respondents, from 13 of 26 (50.0 %) age 3 respondents, to 27 of 56 (48.2 %) age 2 respondents, to 15 of 36 (41.7 %) age 1 respondents. Finally, a greater proportion of age 1 respondents (8 of 36) identified borracha as a unique item compared to age 2 respondents (10 of 56) and age 3 respondents (4 of 26). The findings suggest that the youngest generation of rubber tapp ers may hold different ideas a bout rubber, in particular its role as an income-ear ning product for the household. Results of the PROFIT analysis were le ss conclusive for the three age categories than seen above among all respondents, as well as males and females. Looking first at the youngest group, age 1, the R2 for the medicinal and eate n attributes, respectively, were 0.56 and 0.46. Neither attribute showed a good fit, indicating that perceived similarities among items for the youngest re spondents were not driven by these two attributes. For young adults, age 2, the R2 for the medicinal attribute was 0.81 and for the eaten attribute, 0.57. For the young adults, th e medicinal dimension shows a robust fit with the item coordinates, indicating that this attribute helped shape the sorting of items. Finally, for the oldest group, age 3, the R2 for the medicinal dimens ion was a fairly robust 0.70, while the eaten attribute was a very poor fit, with an R2 of only 0.18. Thus, as I expected, for the two older generations the medicinal dimension showed a good fit with 66 A stronger relationship between castanha and some of the eaten fruits, such as pama and ingÃ¡ (and cacau in comparison to only age 3 respondents) might also help explain this difference, with younger respondents viewing castanha as a fruit to be eaten rather than as an inco me item. A greater percentage of respondents in age category 1 identified castanha as a Â“fruitÂ” or Â“to eatÂ” item than age categories 2 and 3. This would fit with the observation made by Kainer and Duryea (1992) regarding the fruit eating habits of younger children. However, age 2 and age 3 categories showed similar relationships found in age 1 with regard to correlations between castanha and cajÃ¡ and cajarana . Indeed, adults were more likely to identify castanha as a Â“vinhoÂ” item, or as leite , or milk, as the oil from Brazil nuts is commonly referred, which is mixed with juice from cajÃ¡ and cajarana to make a fruit drink.
224 the MDS coordinates. However, surprisi ngly, for the youngest generation, the eaten dimension was not a good fit for the MDS coordinates. Although rubber tappers of different genera tions have very similar knowledge of the domain of extractive resources, subtle di fferences across age categories were noted. In the MDS and cluster analysis procedures, di fferences across age categories with regard to the identification of items as medicinal items or as items to be eaten support the findings of Kainer and Duryea (1992). C oncomitantly, differences in how rubber was identified suggest support for the arguments of Campbell (1996) and Esteves (n.d.), that new generations of rubber tappers may have di fferent ideas about the source of future income. Results of PROFIT analysis argue that the perception of Â“medicinalÂ” and Â“eatenÂ” qualities of all 24 items was not shaping item identification similarly across age categories. For example, for each age category, the Â“eatenÂ” attribute was not a robust fit, suggesting that this attribut e was not shaping rubber tapp er thinking among all items. However, the Â“medicinalÂ” attribute found a good fit for the young adult group, and was fairly strong for oldest respondents as well. This does not contra dict the finding that groups of items were strongly identified for their medicinal or eat properties by these age groups, but indicates that these item attributes were not shared across items. The poor fit of the Â“eatenÂ” attribute among age 1 respondents was somewhat surprising. I believed the eaten attribute would provide a str onger fit among the youngest generation, based on the findings of Kainer and Duryea (1992), as well as my own experience working with forest households. Conclusion In this chapter I examined how rubber tappers think about the domain of nontimber forest resources. By employing cogniti ve anthropological methods, including free
225 lists and pile sorts, and analyzing the result s of respondents categorized by sex and age, a picture has emerged regarding how rubber ta ppers understand this group of extractive resources. The first part of this chapter examined th e results of the free list activity. Simply asking rubber tapper households to list extractive items found that they use a range of forest plants from a number of plant families. Households listed 83 items that fell into approximately 29 plant families. Plants in the palm family, Arecaceae, were listed most often, attesting to their importance for a va riety of consumptive uses, such as food, construction and medicine. Indeed, three of th e top four items listed were palm species. Although this may be partially attributed to my previous research in the reserve, it nonetheless indicates that palms have a diverse role in househol d subsistence. Brazil nuts and rubber were also noted frequently. But curiously, rubber was noticeably absent from many lists. This was true even among househol ds that still tap rubbe r. In fact, rubber was listed by only 23 households, just over ha lf of all households. That a number of households that collect and sell rubber, and gain a sizeable po rtion of their income from it, did not list this item suggests a differe nt understanding of this resource. It seems possible that households did not list rubber because they donÂ’t Â“useÂ” it, but rather extract and sell it. Regardless of why, its absence is notable. One other finding is particularly inte resting. Although households listed many items from a number of different plant families, a great proportion of these items, 54 of 83, approximately 65 %, were listed only once or twice. This suggests that there are considerable differences in the use of ex tractive items, even as households might hold similar knowledge about medicinal remedies or food preparation.
226 The analysis of the results of the cognitive tests run with the pile sort data comprised the second portion of this chapter. Most striking about th e results of the pile sort activity was how similar a ll respondents, as well as respondents divided into different sub-groups, sorted the 24 extractive items. The MDS and hierarchical cluster analysis figures for the sub-groups of the sampleÂ—m ale, female and three age groupsÂ—displayed a strong similarity in thinki ng about this domain. Cultura l consensus analysis found a high degree of shared knowledge. Further, the results of the QAP analysis de monstrate that cultur al knowledge is also shared across sub-groups of the rubber tapper population; strong and significant correlations were found in comparing the result s of males and females, as well as three different generations of rubber tappers. This high correlation across the three generations seems to contradict EsteveÂ’s (n.d.) contenti on that a new generation of rubber tappers is placing new values on the forest. Yet, a closer examination of the results of the different tests lends support to her argument. Although the cluster analysis figur es across age groups were very similar, greater scrutiny of the relationship between r ubber and Brazil nuts found that as age fell, the strength of this pair weakened, and conc omitantly, the youngest generation associated these items less than the ot her age groups with income. The above discussion leads us to two di fferent conclusions, though they are not necessarily contradictory. On one hand, a high degree of agreement of all respondents, as seen in the high level of consensus, showed that rubber tappers widely share cultural knowledge of extractive resources. Indeed, re spondents of different sexes and different age groups held very similar ideas of how these 24 extractive items should be grouped.
227 On the other hand, a closer examination of how male and female respondents, and respondents across generations, suggests subtle differences. Most interesting are the findings across age categories. Here we be gin to see changes across generations, with implications for future market-oriented resource use in the reserve. Most notable is the falling identification of rubber as an inco me item among younger rubber tappers, and the concomitant weakening of its relationship to Brazil nuts, another item traditionally extracted and sold in the reserve. This sugge sts that this new gene ration of rubber tappers may be looking for new Â“itemsÂ” to bring inco me into the household. But these items will also include non-extractive items, such as low-impact logging, increased crop production, and the steer that the children of Roberto and Cibele rode during my last visit. Having explored changes in cultural knowle dge of rubber tappers by sex and age, I now return to one of the central questions of this dissertation Â—how might wealth and markets be shaping rubber tapper livelihoods and knowledge? In the following chapter, I examine rubber tapper knowledge of extractiv e resources across levels of wealth and market integration. This will include analyz ing the results among all respondents, as well as separating individuals into sub-groups of the sampleÂ—males and females, and three generations.
228 CHAPTER 6 WEALTH, MARKETS AND CULTURAL KNOWLEDGE In the previous chapter, I found that rubbe r tappers have a high degree of shared knowledge on the domain of non-timber forest re sources. This was demonstrated in the results of the QAP analysis and the strong cu ltural consensus on this domain. However, a close examination of the data also reveal ed subtle variations in knowledge among subgroups of rubber tappers divided by sex and age. The variations were slight, but they suggest that individuals of different sex or generations ho ld varying ideas about particular extractive resources. In part icular, the data indicated th at rubber tapper youth hold different ideas about the role of rubber in the household econo my, and this hints that they may have different ideas about future production activities in the reserve. In this chapter, I build on this di scussion, now examining how wealth and integration into the market economy may affect cultural knowledge of extractive resources. I test for differences in cultur al knowledge of all indi viduals, categorized by level of household wealth and market integra tion, as well as subgroups of individuals, also categorized by leve l of household wealth and market integration. To do this, I use the results of the cognitive tests in two differe nt ways. First, I use the non-parametric statistic gamma to test if there is an association between level of wealth and market integration with individual consensus analys is scores. Second, I employ QAP to test the degree of fit of the pile sort results fo r individuals, and sub-groups of individuals, categorized by level of wealth and market integration. In addition, I will examine in
229 some detail the results of the MDS and cluste r analysis procedures for individuals and sub-groups of individuals categorized by le vel of wealth and market integration. In this chapter I demonstrate that cultu ral knowledge of extractive resources is shared widely among rubber tappers from househ olds of different le vels of wealth and market integration. Results of the statis tical test gamma and QAP analysis largely confirm that rubber tappers in this study, fi rst as a group, and also divided into subgroups, maintain a high degree of shared knowledge on the domain of extractive resources. However, despite the strong c onsensus on this domain, the gamma test found two instances of a negative re lationship between level of ma rket integration and cultural knowledge. Further, a close examination of the pile sort data suggests that economic factors may be creating subtle knowledge diffe rences. Together, these findings suggest that while cultural knowledge on this domain is widely shared, subtle differences associated with wealth and market integration are emerging, with implications for conservation and development in the reserve. Statement of Hypotheses In Chapters 3 and 4, I explored the re lationship between household wealth and market integration on income, and in particul ar income from extr action. I argued that greater levels of material wealth and market integration would result in an increase in income, but a fall in income from extraction, and the percent of inco me from extraction. In this chapter, I build on these hypotheses in testing how wealth and market integration affect cultural knowledge of extractive resources. For hypotheses H7 and H8, I argue that higher levels of wealth and greater in tegration into markets (or better access to markets), will lead to lower knowledge levels. I hypothesize that: H7, as the level of household wealth increases, the cultural knowledge of non-timber forest resources of all
230 respondents, the household head , and sub-groups of individual s within the household (men, women, youth, young adults, ad ults) will fall, and; H8: as the level of household market integration increases, the cultural knowledge of non-timber forest resources of all respondents, household head, a nd knowledge of sub-groups of individuals within the household (men, women, yo uth, young adults, adults) will fall . To test these hypotheses, I calculate the non-parametric st atistic gamma and measure th e association of level of cultural knowledge with levels of wealth and market integration. I conduct similar tests in hypotheses H9 and H10, employing QAP to examine the similarity, or degree of fit, of cultural knowledge of different s ub-groups, categorized by levels of wealth and market integration. I hypothesize that: (H9) the knowledge of subgroups of individuals (men, women, youth, young adults, adults ) in wealthier households will be significantly diff erent from the knowledge of sub-gr oups of individuals in poorer households, and; (H10) the knowledge of sub-gr oups of individuals in households (men, women, youth, young adults, adults) with a high er level of market integration will be significantly different from the knowledge of sub-groups of individuals in households with a lower level of market integration. I will begin by presenting the results of hypotheses H7 and H8. This will be followed by a discussion of the results of the p ile sort test for indi viduals and sub-groups of individuals, categorized by level of wealth an d market integration. This would ordinarily follow with a discussion of the resu lts of the QAP analyses for individuals, and sub-groups of individuals, ca tegorized by level of wealth and market integration. However, the results of the QAP analysis f ound that for all individua ls and sub-groups of individuals across levels of wealth and the different indi cators of market integration were
231 highly correlated and statistically significant. This is important, as it means that hypothesis H9 and H10 are rejected. I argued that the QAP test would show significant differences in cultural knowledge among all respondents categorized by level of wealth and market integration, and among sub-groups of respondents categorized by level of wealth and market integration. This was not the case: all were highly correlated and statistically significant. I ha ve placed the result s of the QAP analysis in Appendix C. In Chapters 3 and 4, I explained how I categorized households by level of net wealth per capita and level of market integrat ion into labor markets. I created histograms of households ranked from lowest to highest net wealth per cap ita (and percent of productive income from off-farm labor) and identified natural breaks in the plots and established rank categories for each measure. I categorized households by the other two market integration variables, percent of productive income from barter and trade, and travel time to the city using this same method. Tables 6-1 through 6-4 summarize the different categories and numb er of respondents that fall into each category for each variable. These categories were used for calculating gamma as well as conducting the QAP analysis. The tables include only 45 households, as one household moved before I conducted the cognitive tests in my last field visit. Finally, as gamma measures the level of association of two ordinal ranked variables, I also categorized individual cultural knowle dge scores obtained through the consensus analysis procedure. To do this, I first generated a cultu ral knowledge score for each individual, which was calculated by running the consensus analysis procedure with all 118 study participants. I then separate d individuals into groups, by household heads, and by sub-groups of individuals by sex and ag e, and categorized indi viduals within these
232 Table 6-1. Households and individual respondents grouped by sex and age, and categorized by net wealth per capita. Wealth Category 1 Wealth Category 2 Wealth Category 3 Wealth Category 4 Total Less than R$500 R$500 Â– R$999 R$1000 to R$1999 Greater than R$2,000 Households 11 15 8 11 45 Individuals 22 46 21 29 118 Females 9 17 10 9 45 Males 13 29 11 20 73 Age 1 4 14 11 7 36 Age 2 15 22 10 9 56 Age 3 3 10 0 13 26 Table 6-2. Households and individual re spondents, grouped by sex and age, and categorized by percent of household produc tive income from off-farm labor. Pct. Off-farm Category 1 0 % Pct. Off-farm Category 2 0-10 % Pct. Off-farm Category 3 11 Â– 45 % Pct. Off-farm Category 4 Greater than 45 % Total Households 13 15 12 5 45 Individuals 32 40 36 10 118 Females 11 17 13 4 45 Males 21 23 23 6 73 Age 1 7 11 15 3 36 Age 2 12 23 14 7 56 Age 3 13 6 7 0 26 Table 6-3. Households and individual re spondents, grouped by sex and age, and categorized by percent of productive in come from barter and trade of agriculture and ex tractive products. Pct. Trade Category 1 0 Â– 20 % Pct. Trade Category 2 21 35 % Pct. Trade Category 3 36 Â– 49 % Pct. Trade Category 4 50 % or greater Total Households 14 11 9 11 45 Individuals 35 31 24 28 118 Females 11 12 10 12 45 Males 24 19 14 16 73 Age 1 11 11 8 6 36 Age 2 16 14 13 13 56 Age 3 8 6 3 9 26 sub-groups by their knowledge scores into tw o equally divided rank categories using the Â“categorize variable sÂ” command in SPSSÂ®. Thus, one category re presents respondents with the highest knowledge scores, and a seco nd category represents respondents with the lowest knowledge scores. I now turn to re spond to hypotheses H7 and H8, which will be
233 Table 6-4. Households and individual re spondents, grouped by sex and age, and categorized by travel time from household to city of Xapuri. Travel Time Category 1 2 hours or less Travel Time Category 2 Greater than 2, but less than 4 Â½ hours Travel Time Category 3 4 Â½ hours or more Total Households 11 18 16 45 Individuals 35 46 37 118 Females 14 15 16 45 Males 21 31 21 73 Age 1 11 14 11 36 Age 2 18 21 17 56 Age 3 6 11 9 26 followed by a discussion of the pile sort tests of individuals and subgroups of individuals categorized by level of wea lth and market integration. Wealth, Market Integration and Cultural Consensus I have argued that wealth and integrati on into markets not only affect household production activities, and in particular extractiv e activities, but can also lead to changes in how individuals think about extractive resources. In this secti on I respond to hypotheses H7 and H8 to test how wealth and markets may be influencing cultural knowledge. I do this by calculating the gamma st atistic to test for associations between level of household net wealth per capita and di fferent indicators of market integration, and level of cultural knowledge, measured by th e individualÂ’s consen sus analysis score. The gamma statistic is between -1.0 and 1.0, wi th -1.0 being a strong negative association and 1.0 being a strong positive association. A scor e of 0 indicates there is no association. Table 6.5 on the following page presents the gamma statistic for all respondents, heads of households,67 and sub-groups of individuals, in cluding males and females, and the three age groups. Among the tests run, only three statistica lly significant and 67 For heads of household I used the score of the male head of household in all but two cases. For one household, the female was the owner of the landholding while for another, the male household head did not complete the pile sort test so the female household headÂ’s score was used. In another household, the maleÂ’s score was negative, so it is not included in the analysis.
234 negative relationships were found. For all res pondents, a weak, nega tive, association was found between percent of inco me from product trade, and cultural knowledge. For the youngest respondents (age 1), a moderately strong negative association was found between percent of productive income from product trade, and cultural knowledge. For young adults (age 2), a moderately strong negative association was found between percent of productive income from off-fa rm labor, and cultural knowledge. These findings confirm, in part, hypothesis H8, s uggesting that greater levels of market integration lead to lower cultural knowledge scores among these variables and respondent categories. However, the results of gamma, in genera l, do not confirm hypotheses H7 and H8. For wealth, very slight nega tive associations with cultu ral knowledge were found for household heads, males and the youngest ag e group, although the gamma statistic was near zero for each and not st atistically significan t. Therefore, hypothesis H7 must be rejected. For percent of producti ve income from off-farm labo r, other than the results of for the age 2 sub-group which confirmed hypothe sis H8, the results reject the hypothesis for other sub-groups. Household heads did show a weak negativ e association with cultural knowledge, but it was not significant. For percent of income from trade, in addition to a statistically si gnificant and weak negative asso ciation for all respondents, and significant and moderately strong negativ e association noted for rubber tapper youth, weak negative associations were also found for household heads, males, females and the oldest age group. However, these associations were not significant, nor strong enough to confirm hypothesis H8. Intere stingly, travel time had a negative associationÂ—opposite of that hypothesizedÂ—with cultural knowledge in all groups exce pt for the oldest adults,
235 which showed a slight positive association for this factor. Therefore, the findings for this variable also reject hypothesis H8. Table 6-5. Gamma statistic measuring the associ ation between level of wealth and market integration and level of cultural knowle dge for all respondents and sub-groups of rubber tappers. Respondent Net Wealth per Capita Pct. Income from Off-farm labor Pct. Income from Trade Travel Time to Xapuri All respondents 0.077 -0.123 -0.229* -0.175 Household head 0.086 -0.179 -0.152 -0.244 Males 0.026 0.050 -0.278 -0.118 Females 0.057 -0.018 -0.188 -0.272 Age 1 0.00 0.220 -0.474** -0.33 Age 2 0.037 -0.413** 0.052 -0.164 Age 3 0.180 -0.153 0.00 -0.169 Notes: ** Significant at the 0.05 level. * Significant at the 0.10 level. The results of the statistical tests to measure the association between household wealth and market integration with rubbe r tapper cultural knowl edge of extractive resources, in general, were not strong e nough to confirm hypotheses H7 and H8. Only three statistically significan t associations were found: for all respondents and age 1 respondents, percent of producti ve income from trade did show a weak and moderately strong and statistically signifi cant association, respectively, with cultural knowledge, and a negative and moderately strong associat ion was found for age 2 respondents between percent of productive income from off-farm trade and knowledge. Yet, while hypotheses H7 and H8 are generally rejected, it is im portant to note that gamma showed weak negative associations (as hypothesized) between the market integration variables for offfarm labor and trade and cultural knowledge fo r a majority of the statistical tests. Although these associations ar e very weak, the consistenc y of findings for these two variables across the majority of sub-groups suggests that changes in cultural knowledge may be emerging. I explore this in the remainder of this chapter.
236 Wealth, Market Integration and Cultural Knowledge In the previous chapter I di scussed in detail the results of the pile sort activity, analyzing first how all respondents as a group identified items, and then examining in detail how sub-groups of respondents identified items, pointing out subtle differences in cultural knowledge. Respondents across sex an d age categories maintained a high degree of shared knowledge on this domain, as demons trated by the consensus analysis and QAP procedures. Yet, subtle differences in how individuals understand these items emerged through a closer analysis of the pile sort data. In this section, I ex amine the results of the pile sort activity by individuals and sub-gr oups of individuals cat egorized by level of wealth and market integration, follo wing a similar line of research. As noted above, QAP analysis rejected both hypotheses H9 and H10. The testing of cultural knowledge of indivi duals and sub-groups of indivi duals across different levels of household wealth and market integration showed strong and statistically significant correlations: there was no statistical differe nce in the responses among any of the groups of individuals that are compared in the remainder of the chap ter. Therefore, the reader should bear in mind throughout the discussion in the following pages, that even as I identify subtle differences in knowledg e across sub-groups of individuals, the respondents maintain a high degree of shared knowledge on this domain. In the discussion that follows, I summar ize the subtle changes across different wealth and market integration categories drawn from the MDS and cluster analyses figures, often referring back to the raw pile sort data to examine how respondents identified items.68 In this analysis, I focus almo st exclusively on two itemsÂ—rubber and 68 I have not included the MDS scatterplots and cluster an alysis figures, as there are nearly 200 total figures across the sub-groups.
237 Brazil nutsÂ—extractive items with which rubber tappers have strong economic and cultural (and historical, in th e case of rubber) ties, and the two principal income-earning products among these items. I examine the strength of this tie across le vels of wealth and market integration, and how groups identified these items at the completion of the pile sort test. The discussion of copaÃba oil, another product emerging as an income-earning item for rubber tapper households, as well as other items, are also introduced into the discussion where relevant. Cultural Knowledge and Wealth In Chapter 4, in testing the effects of w ealth on income from extractive resources, I found a positive relationship be tween wealth and income from extractionÂ—opposite of that hypothesizedÂ— although this relationship wa s not statistically si gnificant. However, I found a negative and statistic ally significant relationship between household wealth and the percent of income from ex traction. In this se ction, I examine how different levels of household wealth may lead to variations in rubber tapper cultural knowledge of extractive forest resources. I begin by examining the results for all res pondents categorized into the four wealth rank categories. I then separa tely examine the results of female and male respondents, and respondents grouped by age, with each su b-group categorized by wealth level as per Table 6-1 above. When the division of subgroups by wealth category resulted in a small number of respondents in a wealth categor y (less than 9 individuals), I combined adjoining wealth categories. I summar ize the principal findings for each group. Wealth and cultural kn owledge: all respondents . The four wealth categories exhibited very similar groupings of items. However, an init ial examination of the pile sort results led to a closer analysis of th e rubber and Brazil nut pa iring, and the copaÃba
238 oil and honey pairing. As noted in Chapter 3, each of these products was commercialized by study households. Contrary to what I ex pected, respondents with the greatest wealth demonstrated the strongest pairing of rubbe r and Brazil nuts, while the lowest wealth group showed the weakest pairing of these tw o items. I expected the least wealthy respondents to show a stronger tie between th ese two items and wealthier respondents to show weaker ties, reflecting a decline in the importance of these items to household production and income as wealth increased. However, a closer examination of how respondents identified rubber a nd Brazil nuts revealed slight ly different results across wealth categories. Although th e least wealthy respondents show ed the weakest pairing of these items, nearly all of those who did pair them identified them as income items. This proportion fell through the middle w ealth categories, although in creased in highest wealth category, although not to the level found in th e lowest wealth group. A similar pattern was found in examining rubber individually, wi th the lowest wealth group identifying this item in greater proporti on to all other wealth groups. Findings for copaÃba oil and honey also suggested subtle knowledge differences across wealth levels, again indicating a higher association of these two items with income among less wealthy respondents. As noted in Ta ble 5-4 in the previous chapter, the mode code for these two items was medicinal, a nd this was true for each of the wealth categories. However, a closer examination of this pair found that the two least wealthy categories not only placed these items together more often, but they also identified them with income more frequently than wealthier re spondents (with the exception of honey in the wealthiest category).
239 Thus, least wealthy responde nts identified traditional extractive itemsÂ—rubber and Brazil nutsÂ—as well as othe r itemsÂ—copaÃba oil and honeyÂ— with income most often among the wealth categories. The findings sugg est that the lowest wealth category ties these items more tightly to production and income. However, lack of a pattern across wealth categories indicates that wealth is a weak predictor of even subtle variation in cultural knowledge of extractive resources. Cultural knowledge and wealth: females and males . The results of the pile sort activity revealed strong similarities for fema le and male rubber ta ppers across the four wealth categories. For females, I began by examining the medicinal grouping, an area in which females have been noted to have part icular expertise (Kainer and Duryea 1992). However, little variation was found across wealth categories, with the exception of the pairing of copaiba oil and honey. Although fema les in both the least and most wealthy households had similarly strong associations between these two items, those in less wealthy households associated them with income in greater proportion. An examination of the tie between rubbe r and Brazil nuts also revealed minor differences across wealth categories. Intere stingly, the wealthiest category formed the strongest bond of these two items while those least wealthy showed the weakest tie. As noted above, I expected least wealthy househol ds to show a stronger tie between these items. Near equal proportions of the least wealthy and most wealthy respondents placed this pair together as income items. Howeve r, examining the results for only rubber found that females in the two least wealthy categories together asso ciated rubber with income in greater proportion than the two highest wealth categories.
240 The findings for males across wealth categ ories mirrored those found for females. Although the least and most wea lthy males showed similar st rengths in the tie between rubber and Brazil nuts, the le ast wealthy males more str ongly associated rubber and Brazil nuts with income. Further, copa Ãba oil and honey were also more strongly associated with income among the least weal thy male respondents in comparison to other groups. This strong tie of rubber, Brazil nut s, copaÃba oil and honey with income was most notably seen in the cluster analysis figure for the least wea lthy respondents, with these items forming a small cluster before join ing other medicinal items, a result in part due to their association with income. Thus, for both males and females, despite generally similar patterns of sorting across wealth categories, subtle differences emerged, particularly with regard to how extractive items were identifie d. Least wealthy households associated rubber, Brazil nuts, copaÃba oil and honey with income in greater proportion than other wealth categories. However, the data did not yield distinct trends for males and females across wealth groups. Cultural knowledge and wealth: youth, young adults and the older generation . The examination of the results of the th ree age groups found that only the young adult and adult groups, ages 2 and 3, displayed sli ght variations in know ledge of extractive resources by wealth category. For the age 2 group, although a ll wealth categories showed similar strengths in the relationship between rubber and Brazil nuts, respondents in the least wealthy category identifie d this pair proportionally more frequently with income than greater wealth categories. Further, th e two least wealthy categories demonstrated a much stronger identification of copaÃba oil and honey with income than the two higher
241 wealth groups, suggesting that the poores t households view these items not only as medicinal items, but also as potential income earners. An examination of the results for the ol dest group of respondents also revealed slight differences in knowle dge, but opposite of what I expe cted. In combining the two least and most wealthy categories and comp aring them, both combined groups showed a strong pairing of rubber and Brazil nuts. Ho wever, the wealthiest category not only showed the strongest tie, but also a slightly greater propor tion of this group tied these items together as income. Fu rther the wealthiest respondents also identified rubber as an income item in slightly greater proportion. The discussion in Chapter 3 demonstrated that rubber tapper households range widely in wealth holdings, and the type of assets they holds varies as wealth increases. Wealthier households hold a gr eater percent of their productiv e wealth in cattle, but also invest more in off-farm assets. Despite th ese wealth differences, the results of the QAP procedure showed that r ubber tappers widely share knowledge on the domain of extractive resources. However, a closer examinat ion of the results of the pile sort activity among sub-groups of individuals categorized by w ealth found subtle variations in cultural knowledge, particularly in their understanding of rubber and Brazil nu ts. Most notable, across nearly all sub-groups (except age 3), the least wea lthy respondents more strongly identified rubber and Brazil nuts as income items. But curiousl y, across many of these same groups, it was also the wealthiest respond ents who were most si milar to the poorest group. Further, among all respondents, as we ll as males, females and the age 2 subgroup, the least wealthy responde nts identified copaiba oil with income in greater proportions than the wea lthiest respondents.
242 Although the lack of a clear pa ttern across categories sugges ts that wealth is a weak predictor of variation in r ubber tapper knowledge on this domain, these findings hint at subtle differences in the cultural knowledge of extractive resources th at may be useful to policymakers. For copaÃba oil, the findings lightly imply that the poorest respondents may see greater market potential in this item than wealthier respondents of the same sex and age, while the findings for rubber and Brazil nuts suggest that the least wealthy households hold stronger economic ties with these items. Thes e variations hint that the poorest households may be more receptive to pr ojects that promote extractive activities in the reserve. Cultural Knowledge and Market Integration In Chapters 3 and 4, I discussed the rubbe r tapper economy in detail, examining the diverse trade and off-farm labor activities th at households undert ake, noting how these activities vary in importance to household inco me. I then tested how integration into trade and labor markets, as well as travel ti me to the city, affect household income and income from extraction. Here I examine how differences in level of market integration may affect rubber tapper cultura l knowledge of extractive res ources. As above, I look for subtle variations in patterns of knowledge acro ss different levels of market integration by examining more closely the result s of the pile sort activity. I begin by analyzing how knowledge may vary due to household integration into off-farm labor activities, and then follow with a similar discussion on how knowledge may vary by integration into product markets and travel time. As in the analysis of wealth above, for each measure of market integrati on, I examine results of all respondents, and then sub-groups of respondents. As noted a bove, when a market in tegration category had few individuals, I consolidated respondents into adjacent categories.
243 Cultural knowledge and off-farm labor Income from off-farm labor makes an im portant contribution to the income of many reserve families. Off-farm income comes from diverse sources, from unskilled manual wage labor on nearby landholdings, to sk illed salaried positions as teachers and health agents at the community center. Fo r a few households, off-farm income comprises nearly all of the householdÂ’ s productive income. Convers ely, 13 households earned no income from off-farm labor. In Chapter 4, I te sted the effects of integration into off-farm labor markets on extractive income and f ound that as households gained a greater proportion of productive income from off-farm labor, both extractive income and percent of productive income from extraction fell. In this section I address how integration into off-farm labor markets may be subtly sh aping rubber tapper know ledge of extractive resources, again focusing particularly on thei r understanding of rubbe r and Brazil nuts, as well as copaÃba oil. Cultural knowledge and off-farm labor: all respondents . Respondents categorized by level of market integration s howed few visible differences in cultural knowledge on this domain. The most notab le difference across the four market integration categories was found in the rubbe r and Brazil nut pair ing in the cluster analyses: as the percent of in come from off-farm labor increased, the strength of this pairing declined. This suggested a rela tionship that I expected: respondents from households more integrated into labor mark ets would associate r ubber and Brazil nuts less with income. However, a closer examina tion of the pile sort data found that while the strength of this relationship declined as percent of income from off-farm labor increased, similar proportions of respondents ti ed them together as income items. An examination of the results for rubber separate ly found that a slightly greater proportion of
244 the least market integrated respondents iden tified rubber as an income item, and this proportion fell through the middle categories, although increased for the most marketintegrated respondents. A similar pattern was found for copaÃba oil. The results for respondents categorized by le vel of integration in to off-farm labor hint at subtle differences in cultural knowledge of extractive resources among the different market integration cat egories. However, they do no t reveal any clear patterns, and this suggests that among all respondents, level of integration into off-farm labor markets is a weak factor in predic ting cultural knowledge on this domain. Cultural knowledge and off-farm labor: females and males . The results of the pile sort activity revealed strong similari ties for female and males across the market integration categories. However, females, in particular, displayed slight differences in knowledge across categories, no t found among males, which sugge st that integration into off-farm labor markets may be varying their cultural knowledge of this domain. Most notable was the clustering of rubber, Braz il nuts, copaÃba oil, and honey by the least integrated female respondents, those from households with no off-farm income. The grouping of copaÃba oil and honey with rubbe r and Brazil nuts s uggested that they associated these two items with income rather than medicinal properties. The raw pile sort data showed that a higher proportion (though a very small number) of those least integrated into off-farm labor markets id entified copaÃba oil as an income item. A closer examination of the rubber and Brazil nut pairing also suggests that females from households less integrated into labor markets more strongly identify these items with income. First, the strength of this pa iring weakened as market integration increased. Second, as market integration increased, thes e two items together were identified less
245 with income. And third, focusing solely on rubber, a falling proportion of respondents identified rubber with income as income fr om off-farm labor increased. Further, a growing number of respondents ca tegorized rubber as a unique item as market integration increased. Thus, females at in creasingly higher levels of in tegration into off-farm labor markets not only associated rubber less w ith income, but increasingly did not know where to place it. Together, these findings for female res pondents suggest a sub tle variation in cultural knowledge of extractive resources across different leve ls of integration into offfarm labor markets. Females with weak ties to off-farm labor markets associated traditional extractive itemsÂ—rubber and BrazilÂ—as well as copaiba oil, more frequently with income, and conversely, females from hous eholds with stronger ties to off-farm labor markets associated these items less in income terms. Cultural knowledge and off-farm labor: youth, young adults and the older generation . Among the three age groups, only the results of the youngest and middle age groups suggested a variati on in knowledge across levels of market integration. To analyze the results for the youngest age gr oup, I combined the two lowest and two highest categories and compared them. Youth from households less integrated into offfarm labor markets displayed a slightly stronger pairing of rubber and Brazil nuts, and a slightly higher proportion also identified them as income items. This category also identified rubber alone as an income item in greater proportion. Curiously, few youth in either category identified copaiba oil with income. I expected the least market-oriented group to demonstrate a stronge r association of copaÃba oi l with income, reflecting a greater interest in increasing income through extractive sector trade.
246 The results for the middle age group were similar to those of the youngest group. Although the rubber and Brazil nut pairing was strong for all market integration categories, as level of market integration increa sed, a declining propor tion of respondents identified this pair with income. A si milar pattern was found in examining how respondents identified rubber alone, as well as with copaÃba oil and honey. Many of the households in this study ear ned a portion of in come from off-farm labor. However, household off-farm income varied widely. Over one-quarter of the households in this study earned no off-farm income, while others earned a few reais from day labor in the forest. A few households earned large sums off-farm through highly paid work as a carpenter and sala ried positions as teachers . The examination of how integration into off-farm labor markets ma y vary rubber tapper knowledge of extractive resources found that female respondents, as well as youth and young adults, displayed subtle knowledge variations in their understa nding of rubber and Br azil. For these subgroups of respondents, as integration into la bor markets increased, the tie between these items weakened, and they identified them less with income. This finding is particularly notable for the youngest generation of rubber tappers. As rural to urban linkages strengthen, more off-farm la bor opportunities are likely to emerge, suggesting that rubber and Brazil may have a declining role in th e landholding production activities for these young and future household heads. However, the findings also suggest that households poorly linked to off-farm labor markets may be more responsive to projects that promote extractive activities. Cultural knowledge and product trade Rubber tapper households engage in the bart er and trade of a variety of agricultural and extractive products with a diverse group of buyers, including itiner ant traders in the
247 forest, CAEXÂ—both in the forest and in the cityÂ—and merchants and restaurants in the city of Xapuri. For many households, product trade is the only source of income for buying essential supplies, such as kerosene , sugar and ammunition. In Chapter 4, in examining the effects of integrating into product markets on income, I found a negative and statistically significant relationship between percen t of productive income from product trade and income from extraction. I also found a negative relationship between percent of productive income from product trad e and percent of inco me from extraction, although this relationship was not significant. Here I examine the effects of integr ation into product markets on cultural knowledge of extractive resources. Table 6-3, presented above, displays the four categories that represent the different categ ories of integration into product markets.69 I begin with an analysis of th e results of all respondents, a nd then examine the results of females and males, followed by an analysis of the three age groups. Cultural knowledge and product trade: all respondents. Respondents across all levels of integration into product markets s howed very similar groupings of items. An initial analysis of the sor ting of traditionally extracte d and commercialized itemsÂ—rubber and Brazil nutsÂ—revealed a strong clustering of these two items across market integration categories. A closer examination of how respondents across level of product trade identified these items found that very simila r proportions of res pondents identified them with income, although those from households least integrated into product markets did 69 One important point to remember throughout the analys is is that households least integrated into product markets are not necessarily the lowest income households. A high inco me household that received a large proportion of its income from off-farm labor and did not extract or produce for the market would also be in the lowest market integration category, as would a low income household that produced mainly for consumption and sold only a small portion of total production.
248 this in a slightly higher proportion. This wa s also true for rubber alone, as well as copaiba oil. This suggests that households le ast integrated into product markets tie these items more closely with economic benefits. Ho wever, the analysis did not reveal a trend across the four market integration categories, and even as slight differences emerged, the data argue that cultural knowle dge of extractive resources vari es very little across level of product trade Cultural knowledge and product trade: females and males. The analysis of the pile sort data for males and females categ orized by level of in tegration into product markets revealed similar results, with neith er group displaying a pattern that would suggest that cultural knowledge varies with increasing market integration. For females, maybe most notable is how similar they we re in grouping items, both those which they are known to process, such as medicinal items, as well as production items, such as rubber and Brazil nuts. For males, results of the pile sort s showed slight differences across different levels of market integrat ion, but no particular pattern. Strong ties between rubber and Brazil nuts were found at a ll levels of market in tegration except for one middle category. Interestingly, a greater proportion of the leas t and most marketintegrated respondents identified this pair as income items. This was also found in examining the results for rubber alone. Thus for males, minor variations in cu ltural knowledge of r ubber and Brazil nuts were found among market integra tion categories, but they di d not present trends that would suggest variation in cultural knowle dge as households integrate into product markets. Females showed a very high degree of shared knowledge across market
249 integration categories, and even a closer examination of the da ta did not reveal differences. Cultural knowledge and product trade: youth, young adults and the older generation . Upon closer examination of the data of the three age sub-groups, only the youngest respondents displayed slight variati ons in cultural knowledge of rubber and Brazil nuts. For the youngest age group, although the least an d most market integrated categories (the most market integrated categ ory combines the two highest levels) showed similar strength in the tie between these it ems, a greater proporti on of respondents least integrated into product market s tied rubber and Brazil nuts to gether as income items, and this proportion fell as market integrati on increased. A similar trend was found in examining the results for rubber al one, as well as copaÃba oil. A closer examination of the results for th e two adult sub-groups (age groups 2 and 3) revealed that the least and most market integrated categories held the most similar cultural knowledge of rubber and Brazil nuts, although a patt ern of knowledge variation across market integratio n categories was not found. Many households in the reserve are divers ifying production and tr ade activities as a means to increase incomes. The improved feed er road and regular transportation to the city during the dry season, cu rrently highly subsidized by the local government, have facilitated access to markets in the city. As noted in Chapter 2, trade activities are also being stimulated by the government projects to support the expansi on of the extractive sector. In this section I ex amined more closely how inte gration into product markets might be creating subtle knowledge differe nces in the domain of non-timber forest resources. Only the youngest group displaye d a subtle pattern across categories market
250 integration that suggests that increasing integration into pr oduct markets may be varying cultural knowledge of the tw o principal extractive products , rubber and Brazil nuts. Females displayed a strong and similar re lationship for rubber and Brazil nuts, and medicinal items as well, across market integration categories, while the results for both adult age groups, as well as male respondent s did not show any clear patterns. The findings for youth hint that th e increased integration into product markets may bring increased interest in identifying new sour ces of income, beyond traditional extractive products. Concurrently, they suggest that you th from households poor ly integrated into trade markets may be more receptive to de velopment activities that promote income growth through extraction. Cultural knowledge and travel time to the city Distances in the forest are becoming clos er. Road building is forging new links between forest populations and urban areas. These new rural-urban linkages come in the form of market transactions, new profe ssors for schools, and better communication between forest settlements and they city. In Chapter 4, I ex amined how travel time to the city of Xapuri affects income from extr active activities. I f ound a negative and statistically significant rela tionship between travel time and extractive income; as household travel time to Xapuri increased, inco me from extractive activities fell. This relationship was opposite of th at hypothesized, indicating th at households with longer travel times to the city earned less inco me (combined consumption and trade income) from extraction. A negative relationship wa s also found between travel time to Xapuri and percent of income from extraction, although it was not statistically significant. In this section, I examine how market acce ss, measured by travel time from the landholding to the city of Xapuri, may be ch anging rubber tapper cultural knowledge of
251 extractive resources. Again, I begin by ex amining the results of all respondents categorized by the three market integration cat egories, then of each of the sub-groups of respondents, as carried out above. Cultural knowledge and travel time: all respondents . The results of the pile sort test suggested subtle differences in cultura l knowledge of extractive resources as travel time varies. Although all categor ies of travel time displayed a strong tie between rubber and Brazil nuts, this tie was strongest for res pondents with longer trav el times to Xapuri, and declined as travel time fell. In addition, an in creasing proportion (t hough slight) of respondents identified these items with in come as travel time increased. Thus, respondents with shorter travel times to the c ity not only sorted these items together less, but, when they did, they increasingly identifi ed non-income reasons for doing so. This pattern was not found for rubber alone, although approximately half of respondents in the two categories with greater travel times id entified rubber as an income item versus approximately one-third of thos e with shorter travel times. Interestingly, respondents with the shortest travel time to the city also showed a slightly stronger association of copaÃba oil wi th rubber and Brazil nuts. This category of respondents is comprised mainly of resident s of the community of Rio Branco, where copaÃba oil is now being extracted and sold with the support from the state government. Yet, a closer examination of how respondent s identified copaÃba o il revealed unexpected results: a slightly greater proportion of respondents with lo nger travel times identified copaÃba oil with income in comparison to the other two categories. This is likely explained by the fact that residents of a ll three communities are aware of the government-
252 sponsored project to promote the extraction of copaÃba oil, leading even individuals from households with longer travel times to associate this item with income. Although these findings are subtle, they hi nt that respondents with longer travel times to the city may hold slightly different ideas about rubber and Br azil nuts, as well as copaÃba oil. As travel times falls, residents may be rethinki ng their ideas about the role rubber and Brazil nuts play in the household economy, and be more open to new forms of production and income-earning activ ities not based on extraction. Cultural knowledge and travel time: females and males . A closer examination of the results for females and males categorized by travel time to the city found that only females displayed a slight variation in cu ltural knowledge of ex tractive resources as travel time increased. This variation was found in the results of the medicinal group as well as the rubber and Brazil nut pairing. Fi rst, female respondents with the greatest travel time grouped medicinal items more tightly, and the relationship among the grouping of medicinal items weakened as travel time to the city fell. This pattern held true for three different pairs of items found in the medicinal group. Second, as travel time increased, the proportion of respondents th at identified rubber and Brazil nuts with income increased. Together, these findings s uggest that as travel time varies, females have subtle differences in cultural know ledge of these extractive resources. Cultural knowledge and travel time: youth, young adults and the older generation. Among the three age categories, on ly the youngest and oldest age groups displayed subtle variations in cultural knowledge of rubber a nd Brazil nuts with a change in travel time. For youth, the pairing of rubber and Brazil nuts was weakest for respondents with the shortest travel time and strengthened as travel time increased.
253 Concomitantly, an increasing proportion, t hough slight, of responde nts identified these items with income as travel time grew. An examination of rubber alone also found this same pattern. A similar pattern was found fo r the oldest age group: individuals from households with greater travel times identifi ed rubber and Brazil nuts more frequently with income than those from households with sh orter travel times to the city. In addition, those with longer travel times also identified copaÃba oil with income in slightly greatly proportion than those with shorter travel. As feeder roads are improved and extended, distances between forest communities and urban areas are becoming increasingly cl oser. New modes of transportation are appearing, such as motorcycle taxis, whic h will strengthen the ties between forest households and urban areas. In testing how know ledge varies as travel time to the city changes, a more in-depth analysis of th e domain of non-timber forest resources, for which rubber tappers of all ages and sex share knowledge widely, revealed subtle differences in cultural knowle dge. In particular, a close examination of two extractive itemsÂ—rubber and Brazil nutsÂ—found that among all respondents, as well as females, and the youngest and oldest groups, as trav el time increased, a gr eater proportion of respondents identified these two items with income. Although these differences were slight, they hint that responde nts with better access to the city hold different ideas about these two items. This in turn suggests that households with better access to markets may be identifying new sources of income. This finding among rubber tapper youth seems particularly important, as they will be the household heads in the coming decades, prioritizing activities among different producti on alternatives, both in the extractive and non-extractive sectors. Youth with greater market access to the city may be more
254 inclined to pursue non-extractive productive alternatives, while youth with longer travel times to the city may be more responsive to development activities that promote extractive activities. Therefore households w ithout access to transpor tation, or deeper in the forest may be the best targets for government policies aimed at improving household incomes by fostering extractive resource use. Conclusion In this chapter, I employed cognitive anthr opological tools, such as free lists, pile sorts, and related tests, such as MDS and cl uster analysis, to analyze cultural change taking place part in the Chico Mendes Reserv e. More specifica lly, I examined how rubber tappers think about the cultural domain of non-timber extr active resources, an area of study of particular importance to unde rstanding change in the Chico Mendes Extractive Reserve as extraction has played an integral part in the lives of the rubber tapper population. Variations in cultural knowledge of non-timber forest resources among rubber tappers represen t different ideas regarding how individuals understand the forest in a changing economy. These variations , in turn can signal changes in the future role of extraction, and the path of developm ent, in the reserve. This in turn, has implications for the extractive reserv e concept, as well as conservation. The results of the cognitive tests presented in this chapter, as well as the previous chapter, show that rubber tappers maintain a high degree of shar ed cultural knowledge on the domain of non-timber extractive resources. This argues that despite the economic changes in the reserve, rubber tappers of both sexes, and of all ages, have very similar ideas about extractive items, and cultural bonds are strong despite accelerating change. However, the results also contend that rubbe r tappers demonstrate s ubtle variations in knowledge that suggest that sub-groups of the rubber tappers, in pa rticular rubber tapper
255 youth, the future household heads, hold differe nt ideas about some extractive items, in particular, rubber and Brazil nuts, and this ma y carry implications for future livelihood practices in the reserve. At the beginning of this chapter I laid out the four hypotheses I would test to examine the impact of wealth and market integration on rubber tapper knowledge of extractive resources. For the first set of hypothesis, I argued that as household wealth (H7) and market integration (H8) increased, the cultural knowledge of non-timber forest resources (measured by the cultural consensus score) of the household head, and knowledge of sub-groups of individuals with in the household (men, women, youth, young adults, adults) would fall. The statistical te st gamma found that hi gher levels of wealth were not related to higher levels of knowledg e, therefore H7 must be rejected. For H8, three statistically significant associations were found among the three market integration variables that confirm in part this hypothesis: for all re spondents and age 1 respondents, percent of productive income from trade ha d statistically significant and weak, and moderately strong, negative a ssociations, respectively, with cultural knowledge; and for age 2 respondents, percent of productive inco me from off-farm trade had a moderately strong and statistically signifi cant negative association with cultural knowledge. I will discuss the implications of these findings below. However, although no other variables were significant among th e sub-groups of individuals, the results of the gamma showed weak negati ve associations between offfarm labor and product trade indicators and cultural knowledge for a majority of the subgroups of respondents. This suggests that something is going on, and future research
256 should continue to explore and monitor how integration into labor and product markets may be shaping knowledge. For the second set of hypotheses, I employed QAP analysis to test for differences in knowledge of sub-groups, categorized by level of wealth (H9) and market integration (H10). I argued that knowledge of individua ls, as well as sub-groups of individuals, in households at lower levels of wealth or market integration would be significantly different from individuals in households at higher levels of market integration. As noted in the discussion above, both hypotheses H9 and H10 are rejected ; a comparison of individuals, as well as sub-gr oups of individuals, categori zed by levels of wealth and market integration found highl y correlated and sta tistically significant relationships across wealth and market integration categories. These results argue that even as rubber tapper households become wealthier, integrate into labor and product markets, or have better access to ur ban markets, rubber tappers still maintain a very high-degree of shared cultural knowledge of extractive resources. More specifically, males and fe males, as well as different generations of rubber tappers, continue to think very simila rly about extractive res ources as sub-groups of respondents, despite differences in wealth and market integration. Thus, even as the household economy may be undergoing changes, r ubber tappers share ties with the forest. Notwithstanding this great similarity in knowledge across leve ls of wealth and market integration, I carried out a more detail ed analysis to explor e minor differences and trends that might suggest s ubtle variation in knowledge within sub-groups. Although respondents categorized by level of net wealth per household member showed differences in knowledge among wealth catego ries, patterns across catego ries that would indicate
257 trends in changing knowledge as househol d wealth increased were not found. Interestingly, the least wealthy respondents (combined responses of the two least wealth categories) did associate tw o extractive itemsÂ—copaÃba oil and honeyÂ—with income with greater frequency than more wealthy responde nts, suggesting that the least wealthy may be more receptive to projects promoting new ex tractive items. However, the analysis of the sub-groups of respondents categorized by wealth revealed that wealth was a weak predictor of knowledge varia tion. In fact, in many cases , the findings in sub-groups revealed that as household wealth increase d, association of rubber with income would initially fall, then once again increase as weal th rose to the highest level. This would indicate that the respondents in the least and most wealthy households hold similar ideas about rubberÂ’s role in household income. A detailed examination of the pile sort data of the three indicators for market integration found mixed results regarding th e effects of increasing integration into markets on cultural knowledge of extractive reso urces. For the first indicator, market integration measured as percent of household income from off-farm labor, results differed across the sub-groups of responde nts. For males and the oldest generation of respondents categorized by level of integra tion into labor market s, no patterns emerged. However, for females and rubber tapper youth, a slight trend emerged across categories: for these subgroups, as integration into labor markets increased, sub-groups of females, youth and young adults demonstrated a weakening of th e tie between rubber a nd Brazil nuts. In addition, females, males, age 1 and age 2 respondents from more market-integrated households all associated this pair, and r ubber individually, less w ith income, with the results for age 1 and age 2 respondents showing a declining trend across categories. This
258 finding for age 2 respondents supports the results of the gamma test, which found a negative and statistically significant association between level of integration into labor markets and cultural knowledge. They also me sh well with the findings of the regression analyses conducted in Chapter 4, where I f ound that integration in to off-farm labor markets had a strong negative effect on inco me from extraction, and the percent of productive income earned from extraction. Toge ther the results of the cognitive tests and the economic analysis suggest th at integration into labor mark ets is shaping both cultural knowledge and productive activities of the middle generation. An examination of the second market inte gration indicator, percent of household productive income from the sale of agricu ltural and extractive pr oducts, showed similar results found above, but only for rubber ta pper youth. As households gained an increasing share of income from product trade, a smaller proportion of the youngest household members associated this pair, and in particular rubber, with income. Again, the finding for youth supports the results of the gamma test for youth respondents, which found that higher levels of integration into product markets was asso ciated with a lower level of cultural knowledge. The gamma test for all respondents, which showed a weak negative association between cultural knowledg e and level of integration into product markets, was not supported by a pattern across all market integrati on categories, although the least integrated cat egory did identify rubber, as well as other extractive items, with income in slightly greater propor tions than the other categories. These associations are further informed by the discussion of the trade data in Chapter 4, where I noted that households in the two highest rank pr oduct trade categories include 13 of the 21 households that hold cat tle wealth valued at R$1,000.00 or more.
259 This indicates that a majority of respondent s more integrated into product markets are also from households with the greatest cat tle wealth. Together, these findings suggest that as households integrate into product ma rkets, youth in particular, may be thinking differently Â– beyond rubber and Brazil nutsÂ—ab out future production activities in the reserve. The third indicator of market integration, or market acce ss in this case, measured by travel time to the city of Xapuri, also re vealed subtle variations in how rubber tappers think about extractive items. Among all 118 respondents, those from households with the shortest travel time displayed a much weaker association of Brazil nuts and rubber, as well as rubber individually, w ith income. In addition, for sub-groups of females and the three age groups, households with the shortest travel time to the city displayed the weakest association of these items with income . Further, youth with the shortest travel time displayed the weakest re lationship between Brazil nu ts and rubber among all subgroups of respondents. Nota bly, it was the weakest rela tionship found between rubber and Brazil nuts for all sub-groups of individu als categorized by leve l of wealth or the three measures of market integration. This falling tie between rubber and Braz il nuts, combined with a weakening association of these two items with income as rubber tappers integrate into markets, suggests that rubber tapper ideas about the role of extractive ac tivities in household livelihoods is evolving. It is not a drastic change of thinki ng, but rather one that hints that new employment opportunities, new opportu nities for product trade, and better access to urban areas may be altering how r ubber tappers understand these two traditional products. This change does not suggest a dimi nished role for these items in formation of
260 rubber tapper identity, but rather the changing ro le that extraction now plays in the forest economy. Among these findings, the subtle va riations in cultural knowledge of rubber and Brazil nuts as travel time to the city falls seems particularly important to highlight. Feeder road construction and more frequent and reliable transporta tion are reaching more distant populations in the reserve. This suggests that in the near future, residents deeper in the reserve may be rethi nking household production strategi es. The findings hint of a falling economic role for rubber, and the id entification of new production activities. The most suggestive finding he re is for rubber tapper youth. First, as noted in the previous chapter, youth displayed a slight di fference in thinking a bout rubber and Brazil nuts when compared to the two older ag e groups, one that revealed a weakening association of these items to income. This suggests that futu re household heads are already developing different ideas about the role rubber might play in future production, and the findings point toward a diminished ro le for rubber as an income earner. Second, as households integrated into off-farm labor or product markets, or had better market access, youth, as a sub-category of respondents, consistently showed th is same pattern of thought. This suggests not only that youth in ge neral, in comparison to other age groups, associated these items less with income, but also that among youth, increasing integration into markets may also be altering their ideas about extraction. This does not necessarily mean that youth will turn to more destructiv e land use practices as households become wealthier or integrate into markets. New extractive items, such as copaiba oil, may provide an income-earning alternative, a nd a number of youth, as well as young adults, associated this item with income. This s uggests that extractive items can move along the production path from consumption to market and have a continued role in household
261 market strategies. In particular, youth fr om households least integrated into product markets may be most responsive to initiatives that promote these activities. However, youth, in general, might be most interested in pursuing new extr active activities as a means to increase household income. Campbell (1996: 115) stated that there is a Â“changing cultural identity within the seringal.Â” The afternoon I sp ent with Roberto, Cibele and their children was one more experience in the reserve that points toward change. This and the previous chapter argued that if we think of culture as shared knowledge, then rubber ta pper culture, at least in terms of how people think about the extr active forest resources, is widely shared among diverse rubber tapper householdsÂ—poor and wealthy, those that participate in labor and product markets and those that donÂ’t, those with short and longer travel timesÂ— as well as among males and females, and acro ss generations. Yet, a closer examination of the data revealed subtle variations in cu ltural knowledge that s uggest that respondents do hold different ideas, if only s light, about extractive resources, in particular with regard to their role in the househol d economy. I do not think that these findings fully support the idea that cultural identity of the rubber tapper is changi ng; the cultural consensus on this domain was very high, and even the vari ations that I explored were subtle. I do think, nonetheless, that these findings, combined with the findings noted in the Chapters 3 and 4 on the rubber tapper household economy, have implications for future livelihood strategies in the reserve, and conservation and development. I have discussed these implications separately above and in Chap ters 3 and 4. I now bring these findings together in the concluding chapter, and consider them in the context of the site of this study, the Chico Mendes Extractive Reserve.
262 CHAPTER 7 CONCLUSION Â“Previously, people survived only by these; today it is different.Â” That was how Dona Leiticia, 31-year old daughter and wife of a rubber tapper, mother of four children, and living in the Chico Mendes Extractive Reserve described for me why she placed rubber and Brazil nuts togeth er in a group during the pile sort test. There might not be a better description of life in the Chico Mendes Extractive Reserve today. Livelihoods of the rubber tapper ar e changing, and I believe these changes are arriving at households at an accelerating pace. For the decades following World War II, the rubber tapper plied hi s trade in the forest, selling rubber to his patrÃ£o . In the 1970s, the organization of local unions of rural workers aided the ru bber tappers in their struggle for land rights, now challenged by ranchers de termined to convert natural rubber stands into pasture. This workersÂ’ movement le d to the establishment of CAEX, which opened in the late 1980s, helping to free rubber tappers from their debt bonds with itinerant traders and local trading houses. Through the 1990s and until today, CAEX continues to assist rubber tappers as they seek new ways to add value to their products alongside their fight for better prices. In 1990, the establishment of the Chico Mendes Extractive Reserve provided rubber tappers long-term land us e rights to maintain their livelihoods in the forest, still largely based on the extraction and sale of rubber and Brazil nuts. These changes were slow, and cost many lives, but to gether they provided rubber tappers with a more secure and dignified life in the forest , one that made rubber tappers proud of their history, one that was beginning to provi de a life beyond simply subsistence.
263 I believe that the days of slow change may be over. In just th e past few years, the Acre government has implemented new program s as well as construction projects that have dramatically altered life in the forest. Many of these programs have focused on the extractive sector, as discussed in Chapter 1, such as the state rubbe r subsidy, as well as programs to diversify incomes through the ma rketing of other extractive resources. But other programs are also bringi ng change to the households in the reserve. New projects focusing on the community management of timber are now nearing implementation. Feeder roads have been constructed providi ng greater access to urban areas and markets, and rubber tappers in more remote areas will almost certainly be demanding new roads as well. Together, these policies and programs ar e bringing dynamic change to life in the forest. This dissertation captu res part of this change. Wo rking with 46 households in the Chico Mendes Extractive Reserve, I employed microeconomic and cognitive anthropological methods to better understa nd the changes now taking place in reserve. I set out to document not only the how the rubber tapper economy is changing, but to examine how economic changes may be shaping rubber tapper cultural knowledge of the forest, and consider the implications of th ese changes for conservation and development in the reserve. To do this, I tested 10 hypot heses to examine how wealth and integration into markets affect rubber tapper use and know ledge of non-timber extractive resources. My hypotheses argued that as wealth increased, and households integrated into product or off-farm labor markets, or had better market access, measured by travel time to the city, household incomes would increase, but inco me from extraction would fall, and more specifically households would earn a lower proportion of income from extraction.
264 Concomitantly, I argued that an increase in we alth, or integration in to markets would lead to differences in cultural knowledge of non-tim ber forest resources. In this concluding chapter, I summarize the principal findings of this study, including the results of my hypotheses, and discuss their implications fo r forest livelihoods, and conservation and development in the reserve. Summary of Key Study Findings In the Introduction of this dissertation I argued that one only needs a short time in the forest to observe that rubber tappers are not an economically homogenous group. In Chapters 3 and 4, I explored the economic cha nges taking place in the reserve, examining in detail rubber tapper wealth holdings and production activities, and tested six hypotheses to examine the effects of wealth and market integrati on on household income and extractive income. In my analysis of how rubber tapper households invest in wealth, I found that as household wealth increases, di fferences emerge regarding how households invest in assets. As households accumulate w ealth, even as the valu e of wealth held in on-farm productive assets incr eases, a greater pro portion of wealth is held in on-farm non-productive assets and non-farm wealth. On-farm, households invest a greater proportion of wealth in consumer items, while o ff-farm, households invest in assets in the city, including land, homes and consumer items. However, most notable is the contrasting productive wealth holdings of the poorest and wealthiest households: poor households held a greater pr oportion of wealth in equipm ent and a lower proportion in animal wealth, while the wealthiest househol ds held a greater proportion of wealth in animals, and more specifically cattle, and concomitantly, the structures that support livestock production, such as fenc ing and corrals. For two of the three wealthiest groups, cattle held the greatest proportion of producti ve wealth while the lowest wealth group
265 held no wealth in cattle. This indicates that as hous ehold wealth increases (although there was not a clear pattern ac ross wealth categories), the wealthiest households invest a greater proportion of wealth in productive assets related to conversion of forest to pasture. This finding was further supported by positive correlation between wealth and pasture. This suggests that as household wealth increases, th e rainforest environment is negatively affected. Following my discussion of wealth, I examin ed in detail the produc tive activities of rubber tapper households, analyzing on-farm cons umption and trade activities, as well as off-farm skilled and non-skilled labor activities. Notable is how important extractive activities are to reserve households, playing a critical role in both consumption and trade activities. Among all study households, extr active activities made the single largest contribution to mean household income, a nd claimed the largest proportion of both household consumption and trade income. Rubber, Brazil nuts, and game accounted for the greatest proportion of produc tive income; medicinal plants , fibers for making artisan items, fruits, and copaÃba oil and honey contri buted very low values to income. Rubber and Brazil nuts were collect ed and sold by 33 and 36 households. Yet, few households sold extractive goods beyond these two items, de monstrating that that extractive trade is concentrated on these two products. This is a symptom of the rubber tapper economy in general, with households trading few produc ts: of those households that earned the majority of their cash income from trade, all earned more than 50.0 % of productive income from one or two products. Fift een households, nearly one-third of study households, earned at least half of their trade income from rubber or Brazil nuts. In contrast, only five households earned at leas t half of their trade income from cattle
266 production. Thus, extractive products still dom inate forest trade. Indeed, this study found that for many households these products still play a critical role in barter trade for basic supplies. Although on-farm production dominates in come earnings, I also found that offfarm income makes an important contributi on to productive income for many households. Thirty-three households earned at least a portion of income from off-farm labor, and 13 households earned at least 50.0 % of cash income (income from trade and off-farm labor) from off-farm labor. Skilled labor (salaried or wage) had by far the greatest impact on earnings, accounting for 31.0 % of productive income for households undertaking these activities. In contrast, unskilled wage labor accounted for only 4.0 % of productive income for households engaged in these activ ities. This suggeste d that income from skilled labor will have the gr eatest effects on on-farm produc tion activities, a finding that was confirmed by the results of the statistical analysis. In Chapter 4, I brought together the disc ussion of wealth and income to analyze how production activities varied across different wealth hol dings and then tested my hypotheses on the effects of w ealth and integration into markets on household income, and in particular extractive income. In my initial analysis of household income across wealth rank categories, I found that extractiv e activities played an important role in productive income across wealth categories, w ith comparable extractive income values in all wealth groups. Notable were the importa nce of Brazil nuts and game remain across wealth groups; however, there was also a cl ear fall in rubber production in the two highest wealth groups. And, while the value of extraction was maintained across wealth groups, extractive income as a proportion of pr oductive income fell. Other differences
267 between poor and wealthy households emerged: cattle sales were important to all wealth groups except for the lowest wealth group which sold no cattle, a reflection of the divergence in animal wealth holdings; the three highest wealth groups earned a much greater proportion of income from off-farm labor, suggesting that they are engaged in skilled labor activities; and finally, the two hi ghest wealth groups also earned a higher net productive income return per day labor th an the two lowest wealth groups. The wealthiest households earned more than 33.0 % more per day labor than the poorest group, suggesting a more effici ent use of labor resources. The above findings hint at the results of the stat istical analysis conducted to test my hypotheses on the effects of w ealth and market integration on household income, and extractive income. For hypotheses H1 through H3, I argued that as wealth increased, net income would also rise, but income from extraction and the proportion of proportion of income earned from extraction would fall. Bo th H1 and H3 are confirmed: I found that wealth had a positive and statistically significant effect on both net income and net productive income, and wealth had a negativ e and statistically significant effect on percent of income from extraction. However, I found that wealth has a slightly positive, although not statistically signifi cant effect on income from extraction, and thus H2 is rejected. For hypotheses H4 through H6, I argued that as households integrate into markets, earning a greater proportion of in come from product trade or off-farm labor, or gaining better market access through lower travel times to the city, household income would rise, but income from extractive activities, and ex tractive income as a proportion of productive income would fall. The results of the testi ng of these three measur es found mixed results
268 across the three hypotheses. For H4, none of the measures of market integration had a positive association and were statistically si gnificant with net income or net productive income. The travel time variable was statis tically significant in one of the regressions with net household income, but the coeffi cient was positive, opposite the direction hypothesized. Thus, being close to the market , or having access to transport to get you there faster, was not helping households earn higher incomes. As noted earlier, the findings for the travel time variable may be in part due to re cent road building and differential access to transportation by both ge ographically close and distant households. For H5, both integration into product ma rkets and off-farm labor markets were statistically significant and ne gatively associated with inco me from extraction, thereby confirming H5 for these two measures. The in creasing values of the coefficients of the three dummy variables representing increasing le vels of integration into off-farm labor markets argues that the greatest effects of mark et integration are felt at the highest levels of market integration. As noted above, the higher leve ls of integration represent households undertaking skilled la bor activities, rather than wage labor, suggesting that these activities have a much stronger negativ e effect on extractive income. The third measure, travel time to the city was stat istically significant, but again, in the opposite direction hypothesized. Thus, H5 is rejected for this measure. For H6, only integration into off-farm la bor markets was statistically significant and negatively associated with percent of income from extraction, and thus confirming H6 for only this variable. Again, the increasi ng values of the regression coefficients for these three dummy variables argues that integration into labor markets has an increasingly negative effect on the dependent variable. In tegration into product markets
269 was negatively associated with percent of income from extraction, although not statistically significant. Ag ain, the travel time variable , although not statistically significant, was opposite the hypothesized dire ction, suggesting that households with better market access earned a greater percent of inco me from extraction. In Chapters 5 and 6 I employed cognitive tests to examine rubber tapper cultural knowledge of extractive resources. I first exam ined the results of the free lists and pile sort tests for respondents, and then tested four hypotheses to ex amine the effects of wealth and market integration on rubber tapper cultural knowl edge of extractive resources. The palm family accounted for the largest number of items listed, reflecting the importance of palm species to forest households for construction and consumption, among other uses. Rubber and Brazil nuts were also listed frequently. However, many items were listed only once or twice, s uggesting that households use the forest in different ways. The results of the cognitive tests found that rubber tappers have a very high degree of shared knowledge on the domain of non-timber forest products. The cultural consensus analysis showed that rubber tapp er knowledge is widely shared, and the QAP analysis found that pile sort results of s ub-groups of study participants, divided by sex and age were highly correlated. However, a cl oser examination of the data did reveal subtle differences in knowledge. Most notable was the falling associ ation of rubber with income moving from the oldest to youngest age group. This suggests not only that the importance of rubber to income may be falling across the three generations, but also hints that the youngest generation may be identify ing, or open to, new productive activities in
270 the forest. However, their receptiveness to new projects may vary depending on their household economic situation, which I explore below. In Chapter 6 I tested four hypotheses to ex amine the effects of wealth and market integration on rubber tapper cultural knowledge of extractiv e resources. For hypotheses H7 and H8, employing the gamma statistic, I argued that higher levels of wealth (H7) and integration into markets (H8) would be associated with lower levels of cultural knowledge, as indicated by the indi vidualÂ’s cultural consensus anal ysis score. I tested all respondents, the household heads, and sub-gr oups of individuals sub-divided by sex and age. The results of gamma showed no stat istically significant associations between wealth and cultural knowledge for any of th e groups of respondents, therefore H7 is rejected. For H8, three statistically signifi cant and negative associ ations were found, one weak, the other two moderately strong. For all respondents, I found a weak negative relationship between level of product trade and level of cultural knowledge. For the youngest age category, level of integration in to product markets showed a moderately strong negative association with level of cultural knowledge. For the middle age category, level of integration into labor markets also showed a moderately strong negative association. Thus, for these three groups and measures of market integration, H8 is confirmed. Hypotheses H9 and H10 I employed QAP analysis to examine differences in knowledge across levels of wealth and mark et integration. I argued that cultural knowledge of individuals and sub-groups with gr eater wealth and higher levels of market integration would be different from individuals and sub-groups at lower levels of wealth and market integration. The results of th e QAP analysis showed strong correlations
271 across wealth and market integration categ ories, and thus both hypotheses H9 and H10 were soundly rejected. However, a closer examination of the raw p ile sort data did indicate differences in cultural knowledge across levels of wealth and market in tegration that suggest subtle differences in how individuals understand extrac tive resources. A number of findings are notable. First, females, youth and middle ag e respondents displayed a falling association of rubber with income as they integrated into off-farm labor markets. The finding for age 2 respondents fits well with the results of th e gamma test for this group which found that as young adults integrate into off-farm labor markets, they have a different understanding of extractive resources. Further, this me shes well with the re sults of the economic analysis, which found that integr ation into off-farm labor mark ets resulted in a negative and statistically significant relationship with income from extraction, and percent of productive income from extraction. Second, all sub-groups except for males displayed a falling association of rubber with income as travel time to the city fell. This suggests that as new roads are built and new methods of transportation faci litate access to the city, other households may be changing their ideas about the role of rubber in production activities. The findings for rubber tapper youth are most notable. Youth, categorized by the three measures of market integration, showed a falling association of rubber with income as market integration increased for all th ree measures. The gamma statistic for the market integration variable percent of income from trade also revealed a statistically significant and negative associ ation between cultural knowle dge and level of market integration, also suggesting knowledge varia tion on extractive resources. These findings
272 are particularly important, as youth will be directing household decision-making over the coming decades. Their lower association of ru bber with income with integration into markets suggests that markets may be reshap ing ideas about production activities in these households, and hints that trad itional activities su ch as rubber extraction may be replaced by other production activities. Implications of Findings for Conservation The implications of these findings for cons ervation in the reserve can be interpreted as both good and bad. On the positive side, although I found that in creasing integration into off-farm labor markets leads to falling in come from extraction, as well as percent of income from extraction, there was no correla tion between integrati on into labor markets and cattle wealth. Further, integration into labor markets showed a negative correlation with income from basic crop production. Th is indicates that hous eholds earning high incomes from skilled labor act ivities are not investing in come into cattle production activities and they ar e producing a lower value of basi c crops, suggesting less slash and burn agriculture. This argues that househol d integration into labor markets may be good for the conservation of tropical rainforests Â—households not only use the forest less, for trade and consumptive activities, such as hunting, but they are not converting it to pasture or crop areas. The findings for integration into trad e markets are less clear. I found that integration into product trade markets resulted in only a moderate fall in income from extraction, suggesting that extraction remains important to all house holds regardless of level of product trade. However, I also found that percent of income from trade showed a weak, but positive and statistically significan t correlation with both cattle wealth and income from crop production. This suggests th at income from integration into product
273 markets is being at least partially invest ed in cattle wealth and crop production, both involving land conversion. Thus, integration into product markets may not dramatically reduce extractive activities, but production activities that accompany integration into trade marketsÂ—investment in cattle and sl ash and burn agricultureÂ—do have negative implications for conservation. The findings for wealth present the la rgest challenge for conservation. Although households continue to use the forest as wealth increases, they concomitantly invest a greater percent of wealth in cattle production. This argues that as households emerge from poverty they invest a greater percentage of wealth in destructiv e land use practices. It also suggests that development projects focused on improving pr oduction activities to lift households out of poverty may, in turn, result in the investments of profits in environmentally destructive activities. Implications for Policy Development The findings of this study also provide guidance for policymakers and development practitioners. The results of the cogn itive tests found that cultural knowledge of extractive resources is widely shared, not only among all individuals, but also by subgroups of individuals, and by sub-groups of i ndividuals categorized by different levels of wealth and market integration. Howeve r, the subtle differences in knowledge, particularly with regard to how respondents id entified extractive items during the pile sort test, most notably a falling association of rubber with in come as market integration increased for some groups, indicate that cert ain individuals in households less integrated into off-farm labor markets and product markets, as well as with longer travel times to the city, may be more receptive to development pr ojects that promote extractive activities. This was especially true for rubber tapper youth. My findings s uggest that youth least
274 integrated into labor and produc t markets, or with longer trav el times to the city may be more inclined to enter projects that promot e income growth through development of nontimber extractive resources. In addition to youth, I found that that females and middle age adults least integrated in to labor markets and with longe r travel times may also be more receptive to projects that pr omote trade in extractive products. The findings regarding the e ffects of the rubber subsi dy on rubber production of 28 reserve households also inform policy devel opment in the region. First, the findings argue that the additional payment per kil ogram of production was not sufficient to stimulate rubber production among households th at have left rubber tapping. Second, the findings argue that a higher pr ice will not necessarily lead to production increases among households that do undertake this activity Â– onl y four households showed an increase in production in comparing 1996 data with data collected for this study. However, the findings do suggest that househol ds with the lowest wealth holdings may benefit more from this policy, as the two least wealthy household categories earne d the greatest share of income from rubber production. By this measure, the rubber subsidy has helped increase the incomes of the poorest of rubber tapper households in this study. Thus, my research argues that it is a very effective pol icy in targeting the poor est households in the forest. Finally, study findings indicate that govern ment policy can have a great effect on trading patterns in the forest. The subsidy ba sically eliminated rubbe r trade with itinerant traders in the region, and resu lted in the concentration of rubber sales to CAEX, which channels production into local processing facilities now managed by a private sector
275 entrepreneur. Thus, the policy brings the potential to concentr ate regional production into local processing facilities and facilitate economies of scale. Further, the study findings also inform th e debate on the potential of the extractive reserve concept as a conservation and deve lopment strategy. My findings demonstrate that households across levels of wealth, integration into product trade markets, and households with better access to the city a nd those with longer trav el times rely on the forest for their livelihoods. The importance of both rubbe r and Brazil nuts for trade income, and the heavy reliance on game fo r consumption, argue that households are using the large areas of land that they are provided under th e extractive reserve management regime. However, the findings of this study indicate that integration into off-farm labor markets leads to a decline in extractive resource use. Those households most integrated into skilled off-farm labor ac tivities, and in particular salaried positions, showed the most dramatic decline in extractiv e resource use. The construction of feeder roads will facilitate access to off-farm labor activities, and development projects, such as community timber management projects now nearing implementation in the reserve which will create employment opportunities, some of which will likely bring substantial returns to forest households. However, it seems unlikely that a large number of high paying salaried and skilled labor positions will appear even in the short or medium term. Thus, while HommaÂ’s (1992) contention that labor opportunities would siphon rural workers out of low-return extractive activitie s into better return off-farm labor markets has occurred, a parallel reduc tion in extractive income is found only among those households most integrated into labor mark ets. Households pursuing unskilled wage
276 labor activities, and even house holds gaining close to half of their income from wage or salaried labor still earn a si zeable portion of productive income from extractive activities. While one of the principal findings of th is study was the importance of extractive activities to household income, and that househol ds with diverse economic situations rely heavily on the forest, this study also found that reserve households invest a greater proportion of wealth in cattle production as wealth increases. While households may justify their large landholdings as critical to carrying out extractive activities that are fundamental to their livelihoods, increasi ng investments in cattle production and concomitant deforestation (cf. Gomes 2001) ar e likely to raise concerns regarding the long term sustainability of the overall pr oductive practices that rubber tappers employ, and the benefits of the extr active reserve concep t for conservation. As households continue to invest in cattle production, the benefits of provi ding large usufruct rights to rubber tapper households as a strategy to conserve tropical forests may face tough questions. Study Contribution to Research Methods This study also makes a contribution to soci al science research methodology. First, I found that the use of participatory methods for the collection of economic dataÂ—in this case diagrams of wealth items and productive activitiesÂ—can facilitate household recall of production and extraction ac tivities, a difficu lt activity, especially when asking households to recall activities undertaken ove r the past year. Further, the diagrams facilitated the participati on of the youngest households members, who at times, would correct the responses of household heads. The formal questionnaire lasted over two hours in a few cases, and the diagrams appeared to make the interview less tiresome for respondents.
277 On another scale, this study contributes to research methodology by demonstrating how microeconomic and cognitive anthropologi cal research methods can be combined to better understand how forest households both use and think about the forest. By employing diverse tools to ask similar questions , such as QAP analysis, and the use of the gamma statistic to test for the association of the cultural consensus scores and household data for wealth and market integration, the analysis was able to illuminate both the high degree of shared knowledge across these variables, but also isolate differences in cultural knowledge across these same variables. In addition, by collecti ng information on why individuals sorted items in certain ways, I was able to examine the data for subtle variations in cultural knowledge across levels of wealth and market integration. These findings provide insight into how members of a cultural group think about the extractive resources that have shaped their livelihoods for more than a century, and how this understanding may vary across indivi duals sub-divided by age and sex. Finally, for part of the analysis, this study measured cultural knowledge by the individualÂ’s consensus analys is score: the individualÂ’s know ledge in comparison to the norm. Therefore, individuals who have th e lower cultural consensus score may share more or less knowledge with their cultu ral peers (Reyes-Garcia 2001). So, while researchers need to be aware of the Â“sha man effectÂ”Â—individuals in the sample who maintain specialized knowledge, and thus fall out of the consensusÂ—they must also be aware that cultural consensus analysis doe s not measure the level of agreement on Â“traditionalÂ” knowledge, but what is the norm at that particular time. Reyes-Garcia (2001) suggested the need to construct a Â“benchma rkÂ” of knowledge from which change can be measured. This is a useful sugges tion, particularly for documenting changes in
278 ethnobotanical knowledge, which was a focus of her study. However, I would argue that it poses methodological challenges: it assumes th at we are able to identify a particular point in time to serve as a benchmark for shared knowledge. As cultural knowledge is a moving target, this may be difficult. Speak ing with shamans will reveal knowledge of medicinal plants and healing rituals, but this may not be the knowledge that was widely shared in years past. Thus, maybe we n eed to consider two types of knowledgeÂ—more specialized ethnobotanical knowledg e, and cultural knowledge in general. For the latter, I would suggest the need to document cha nges in knowledge over time, examining how peopleÂ’s understanding of the forest changes, and identify ing the varying richness of knowledge that passes across generations. This can be examined either qualitatively or quantitatively. Study Contribution to Anthropology In the Introduction, I noted that anthropologists have been studying how economic factors shape livelihood activities and culture for decades. Indeed one of the earliest studies examined how the introduction of rubber tapping in the Amazon region among Amerindians dramatically alte red the economy and cu ltural ties of the indigenous peoples and their leaders (Murphy and Steward 1956). My research follows in this trad ition, examining how economic factors may be shapin g resource use, but also how a changing economy affects cultural knowledge of the natural environment. Thus, this study makes a theoretical c ontribution to the fields of economic anthropology and cognitive anthropology. Th rough an analysis of both household wealth and productive activities, as well as rubber tapper cultural knowl edge of extractive resources, and subsequently linking these tw o lines of research, my study informs not only our understanding of the rubber tapper household economy and cultural knowledge
279 of extractive resources independently, but also how the economy may be shaping cultural knowledge. This study argues that wealth and integration into mark ets do shape use of forest resources. Yet, the QAP analysis found that cultural groups are able to maintain their understanding of natural resources desp ite changing livelihoods and stronger market ties. However, concomitantly, a different stat istical test revealed a moderate correlation between level of cultural knowledge and level of market integration for some sub-groups of the study respondents. Thus this st udy suggests that cultural knowledge may be widely shared, but variations in knowledge can be identified and measured by looking more closely at sub-groups of the population an d the role of specific factors that may be shaping cultural processes.
280 APPENDIX A MULTIPLE REGRESSION MODELS Model 1 i i i i i i i i i i i i i iCAEX BNUTS SOCSAL YRSHEAD UP YREDP LANDSIZE TOTLABOR TRAVTIME RNK MKT RNK MKT RNK MKT PISV NWPER Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) 8 ( Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) 4 1 ( Âˆ ) 3 1 ( Âˆ ) 2 1 ( Âˆ ) ( Âˆ ) ( Âˆ Âˆ6 12 11 10 9 8 7 6 5 4 3 2 1 0 1 Y1 = Net household income (R$) NWPER = Net wealth per capita (R$) PISV = Percent of income from barter or trade of products MKT1RNK2 = Percent of inco me from off-farm labor fr om greater than 0.0 to 10.0% (Base = 0.0 % Percent income from off-farm labor) MKT1RNK3 = Percent of income from off-farm labor from 11.0 to 45.0% MKT1RNK4 = Percent of income from off-farm labor greater than 45.0% TRAVTIME = Travel time to city of Xapuri (minutes) TOTLABOR = Total labor available includ ing family and hired labor (days) LANDSIZE = Landholding size (hectares) YREDP8UP = Education per capita 8 years of age and older (years) YRSHEAD = Years male househol d head on landholding (years) SOCSAL = Household received a monthly social payment (Base = Household did not receive a monthly social payment) BNUTS = Household sold Brazil nuts over past 12 months (Base = Household did not sell Brazil nuts) CAEX = Household member is a member of th e Agro-Extractive Cooperative of Xapuri (CAEX) (Base = No household member is a member of CAEX) See Tables 4-3 and 4-5 Model 2 i i i i i i i i i i i i i iCAEX BNUTS SOCSAL YRSHEAD UP YREDP LANDSIZE TOTLABOR TRAVTIME RNK MKT RNK MKT RNK MKT PISV NWPER Log Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) 8 ( Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) 4 1 ( Âˆ ) 3 1 ( Âˆ ) 2 1 ( Âˆ ) ( Âˆ ) ( Âˆ Âˆ6 12 11 10 9 8 7 6 5 4 3 2 1 0 2 LogY2 = Natural log of net productive household income (R$) All other variables as defined in Model 1
281 Model 3 i i i i i i i i i i i i i iCAEX BNUTS SOCSAL YRSHEAD UP YREDP LANDSIZE TOTLABOR TRAVTIME RNK MKT RNK MKT RNK MKT PISV NWPER Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) 8 ( Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) 4 1 ( Âˆ ) 3 1 ( Âˆ ) 2 1 ( Âˆ ) ( Âˆ ) ( Âˆ Âˆ6 12 11 10 9 8 7 6 5 4 3 2 1 0 3 Y3 = Income from on-farm extractive activities All other variables as defined in Model 1 Model 4 i i i i i i i i i i i i i iCAEX BNUTS SOCSAL YRSHEAD UP YREDP LANDSIZE TOTLABOR TRAVTIME RNK MKT RNK MKT RNK MKT PISV NWPER Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) 8 ( Âˆ ) ( Âˆ ) ( Âˆ ) ( Âˆ ) 4 1 ( Âˆ ) 3 1 ( Âˆ ) 2 1 ( Âˆ ) ( Âˆ ) ( Âˆ Âˆ6 12 11 10 9 8 7 6 5 4 3 2 1 0 4 Y4 = Percent of household productive income from on-farm extraction All other variables as defined in Model 1
282 APPENDIX B RESULTS OF MULTIVARIATE REGRESSION WITH ALL HOUSEHOLDS Table B-1 Multivariate regression model findi ngs for all 46 study households in Chico Mendes Extractive Reserve. Model 1: Household Net Income Model 2: Household Net Productive Incomea Model 3: Household Extractive Income Model 4: Percent of Productive Income from Extraction Variable Coefficient Coeffici ent Coefficient Coefficient Net wealth per capita (R$) 0.56*** 0.00014*** 0.013 -0.00004** Pct. of productive income from barter or trade 992.24 -0.29959 -479.60 -0.18352 Percent productive income from off-farm labor (1-10%)c 721.25 -0.08599 -255.68 -0.16628** Percent productive income from off-farm labor (11-45%)c 446.41 -0.02883 -959.35** -0.24914*** Percent productive income from off-farm labor (> 45%)c 1,203.04 0.28560 -1,038.73 -0.41360*** Travel time to city (minutes) 3.57 -0.00015 -1.81* -0.00023 Total Labor available (days) 3.21*** 0.00070*** 0.50 0.00009 Landholding Size (hectares) 2.28 0.00080** 2.87*** 0.00023* Education per capita 8 yrs and older (years) 110.94 -0.00746 -0.87.09 -0.00670 Years Head on Landholding -5.62 -0.00804 -10.10 -0.00068 Monthly social paymentb 2,139.10** -0.44801** -778.02** -0.03129 Sold Brazil nutsb -463.63 -0.15757 -175.80 0.06929 CAEX membershipb 1,190.36* 0.12035 426.24 0.04151 Observations 46 46 46 46 R-squared 0.65 0.49 0.51 0.50 Pr > F <0.0001 0.0004 0.0002 0.0003 Notes : a Used natural log transformation. b Dummy variable. Membership in AMOREX and the STR were also tested for possible incl usion in the model but were not significant and thus not included. CAEX better captu res the economic influence of social organizations working with rubber tapper househ olds in the reserve. The base is do not sell Brazil nuts, or not a member of CAEX. c Dummy variable. The coefficient represents the change compared to the intercept base. Th e base is no income from off-farm labor. *Significant at .010; ** significant at 0.05; *** significant at 0.01.
283 APPENDIX C QUADRATIC ASSIGNMENT PROCEDURE RESULTS Table C-1. Quadratic Assignment Procedure (QAP) correlations for wealth for all respondents and respondents s ub-divided by sex and age. Wealth Groups Compared Correlation P-value All Respondents Wealth 1 and Wealth 2 .911 .00 Wealth 1 and Wealth 3 .824 .00 Wealth 1 and Wealth 4 .898 .00 Wealth 2 and Wealth 3 .852 .00 Wealth 2 and Wealth 4 .920 .00 Wealth 3 and Wealth 4 .868 .00 Females Wealth 1 and Wealth 2 .820 .00 Wealth 1 and Wealth 3 .739 .00 Wealth 1 and Wealth 4 .758 .00 Wealth 2 and Wealth 3 .744 .00 Wealth 2 and Wealth 4 .783 .00 Wealth 3 and Wealth 4 .674 .00 Males Wealth 1 and Wealth 2 .835 .00 Wealth 1 and Wealth 3 .761 .00 Wealth 1 and Wealth 4 .801 .00 Wealth 2 and Wealth 3 .814 .00 Wealth 2 and Wealth 4 .894 .00 Wealth 3 and Wealth 4 .823 .00 Age 1 Wealth 1 & 2 and Wealth 3 & 4 .831 .00 Age 2 Wealth 1 and Wealth 2 .843 .00 Wealth 1 and Wealth 3 .726 .00 Wealth 1 and Wealth 4 .768 .00 Wealth 2 and Wealth 3 .762 .00 Wealth 2 and Wealth 4 .861 .00 Wealth 3 and Wealth 4 .685 .00 Age 3 Wealth 1 & 2 and Wealth 4 .843 .00
284 Table C-2. Quadratic Assignment Procedure ( QAP) correlations for integration into offfarm labor for all respondents and re spondents sub-divided by sex and age. Market Integration Groups Compared Correlation P-value All Respondents Category 1 and Category 2 .911 .00 Category 1 and Category 3 .900 .00 Category 1 and Category 4 .855 .00 Category 2 and Category 3 .928 .00 Category 2 and Category 4 .884 .00 Category 3 and Category 4 .886 .00 Females Category 1 and Category 2 .813 .00 Category 1 and Category 3 & 4 .733 .00 Category 2 and Category 3 & 4 .863 .00 Males Category 1 and Category 2 .855 .00 Category 1 and Category 3 & 4 .899 .00 Category 2 and Category 3 & 4 .893 .00 Age 1 Category 1 & 2 and Category 3 & 4 .854 .00 Age 2 Category 1 and Category 2 .900 .00 Category 1 and Category 3 & 4 .885 .00 Category 2 and Category 3 & 4 .890 .00 Age 3 Category 1 and Category 2 & 3 .833 .00
285 Table C-3. Quadratic Assignment Procedure (QAP) correlations for integration into product markets all respondents and res pondents sub-divided by sex and age. Market Integration Groups Co mpared Correlation P-value All Respondents Category 1 and Category 2 .937 .00 Category 1 and Category 3 .829 .00 Category 1 and Category 4 .936 .00 Category 2 and Category 3 .809 .00 Category 2 and Category 4 .895 .00 Category 3 and Category 4 .842 .00 Females Category 1 and Category 2 .857 .00 Category 1 and Category 3 .682 .00 Category 1 and Category 4 .834 .00 Category 2 and Category 3 .692 .00 Category 2 and Category 4 .783 .00 Category 3 and Category 4 .659 .00 Males Category 1 and Category 2 .900 .00 Category 1 and Category 3 .744 .00 Category 1 and Category 4 .912 .00 Category 2 and Category 3 .702 .00 Category 2 and Category 4 .860 .00 Category 3 and Category 4 .739 .00 Age 1 Category 1 and Category 2 .836 .00 Category 1 and Category 3 & 4 .782 .00 Category 2 and Category 3 & 4 .751 .00 Age 2 Category 1 and Category 2 .876 .00 Category 1 and Category 3 .783 .00 Category 1 and Category 4 .834 .00 Category 2 and Category 3 .742 .00 Category 2 and Category 4 .800 .00 Category 3 and Category 4 .769 .00 Age 3 Category 1 & 2 and Category 3 & 4 .808 .00
286 Table C-4. Quadratic Assignment Procedure (QAP) correlations for travel time to the city of Xapuri for all respondents and res pondents sub-divided by sex and age. Market Integration Groups Compared Correlation P-value All Respondents Category 1 and Category 2 .915 .00 Category 1 and Category 3 .908 .00 Category 2 and Category 3 .944 .00 Females Category 1 and Category 2 .785 .00 Category 1 and Category 3 .754 .00 Category 2 and Category 3 .852 .00 Males Category 1 and Category 2 .898 .00 Category 1 and Category 3 .877 .00 Category 2 and Category 3 .914 .00 Age 1 Category 1 and Category 2 .817 .00 Category 1 and Category 3 .800 .00 Category 2 and Category 3 .834 .00 Age 2 Category 1 and Category 2 .902 .00 Category 1 and Category 3 .837 .00 Category 2 and Category 3 .827 .00 Age 3 Category 1 & 2 and Category 3 .869 .00
287 LIST OF REFERENCES Allegretti, Mary Helena 1995 The Amazon and Extracting Activities. In Brazilian Perspectives of Sustainable Development of the Amaz on Region. M. Clusener-Godt and I. Sachs, eds. Pp. 157-173. Man and th e Biosphere Series; v. 15. Paris Carnforth, Lancs, UK; Pearl Ri ver, NY: Unesco: Parthenon. 1990 Extractive Reserves: An Alternativ e for Reconciling Development and Environmental Conservation in Amazonia. In Alternatives to Deforestation: Steps Toward Sustaina ble Use of the Amazon Rain Forest. A.B. Anderson, ed. Pp. 252-264. New York: Columbia University Press. 1979 Os Seringueiros. MasterÂ’s Thesis. Brasilia: Univers ity of Brasilia. Amorim RÃªgo, Patricia de, and MÃ¡ rcia Regina de Sousa Pereira n.d. A Efetividade do Direito Ambiental Brasileiro: Instrumentos Legais de ProteÃ§Ã£o do Meio Ambiente: Unidades de ConservaÃ§Ã£o e ImplementaÃ§Ã£o Legal: Um Estudo de Caso (O Estado do Acre). Retreived November 23, 2004 from h ttp://www.mp.ac.gov.br/4/files/tese19.pdf . Anderson, Anthony B., and Edviges Marta Ioris 1992 The Logic of Extraction: Resource Management and Income Generation by Extractive Producers in the Amazon. In Conservation of Neotropical Forests: Working from Traditional Resource Use. C. Padoch and K.H. Redford, eds. Pp. 175-199. New York: Columbia University Press. Andrade, Gabriel Augusto de 1985 Nomes Populares das Aves do Brasil. Belo Horizonte, Brazil: Sociedade OrnitolÃ³gica Mineira and the Instit uto Brasileiro de Desenvolvimento Florestal. Appasamy, Paul 1993 Role of Non-Timber Forest Products in a Subsistence Economy: The Case of a Joint Forestry Project in India. Economic Botany 47(3):258-267. Arnold, J. E. Michael, and M. Ruiz PÃ©rez 2001 Can Non-Timber Forest Products Ma tch Tropical Forest Conservation and Development Objectives? Ecol ogical Economics 39(3):437-447.
288 Bakx, Keith 1988a From Proletarian to Peasant: Rural Transformati on in the State of Acre, 1870-1986. Journal of Development Studies 24(2):141-160. 1988b The Shanty Town, Final St ages of Rural Development. In The Future of Amazonia: Destruction or Developmen t. D. Goodman and A. Hall., eds. Pp. 49-69. New York: St. MartinÂ’s Press. 1986 Peasant formation and Capitalist Development: The Case of Acre, SouthWest Amazonia. PhD Dissertation, Univ ersity of Liver pool, Great Britain. Balick, Michael J. (ed). 1988 The Palm Â– Tree of Life: Biol ogy, Utilization and Conservation: Proceedings of a Symposium of th e 1986 Annual Meeting of the Society for Economic Botany held at the New York Botanical Garden, Bronx, New York, 13-14 June. Bronx, New York : New York Botanical Garden. Barham, Bradford L., and Oliver T. Coomes 1996 Prosperity's Promise: The Amazon Rubber Boom and Distorted Economic Development. Boulder: Westview Press. Barham, Bradford.L., OliverT. Coomes, and Yoshito Takasaki 1999 Rain Forest Livelihoods: Income Generation, Household Wealth and Forest Use. Unasylva 50(198):32-42. Barlett, Peggy 1982 Agricultural Choice and Change: Decision Making in a Costa Rican Community. New Brunswick, N.J.: Rutgers University Press. Bedoya Garland, Eduardo 1995 The Social and Economic Causes of Deforestation in the Peruvian Amazon Basin: Natives and Colonists. In The Social Causes of Environmental Destruction in Latin America. M. Painter and W. H. Durham, eds. Pp. x, 274. Ann Arbor: University of Michigan Press. Bernard, H. Russell 1995 Research Methods in Anthropology: Qualitative and Quantitative Methods. Walnut Creek, CA: AltaMira Press. Bluffstone, Randall A. 1995 The Effect of Labor-Market Performa nce on Deforestation in Developing Countries under Open Access: An Exam ple from Rural Nepal. Journal of Environmental Economics and Management 29(1):42-63. Borgatti, Steven P. 1996a ANTHROPAC 4.0. Natick, MA: Analytic Technologies. 1996b ANTHROPAC 4.0 Methods Guide. Na tick, MA: Analytic Technologies.
289 Boster, James S. 1986 Exchange of Varieties and Information Between Aguaruna Manioc Cultivators. American Anthropologist 88(2):428-436. Boster, James S., and Jeffrey C. Johnson 1989 Form or Function: A Comparison of Expert and Novice Judgments of Similarity among Fish. Ameri can Anthropologist 91(4):866-889. Brondizio, Eduardo S. and AndrÃ©a D. Siqueira 1997 From Extractivists to Forest Farm ers: Changing Concepts of Caboclo Agroforestry in the Amazon Estuary. Research in Economic Anthropology. 18:233-279. Browder, John 1992a The Limits of Extractivism: Tropical Forest Strategies Beyond Extractive Reserves. Bioscience 42(3):174-183. 1992b Social and Economic Constraints on the Development of Market-Oriented Extractive Reserves in Amazon Rain Forests. In Non-Timber Products from Tropical Forests: Evaluation of a Conservation and Development Strategy. D.C. Nepstad and S. Schwartzman, eds. Pp. 33-41. Advances in Economic Botany, V. 9. Bronx, N.Y.: New York Botanical Garden. Brown, Irving Foster, Silvia H.C. Brilhante , Elsa Mendoza, and Ivanir Ribeiro de Oliveira. 2002 Estrada de Rio Branco, Acre, Bras il aos Portos do Pacifico: Como Maximizar os BenefÃcios e Mi nimizar os PrejuÃzos para o Desenvolvimento SustentÃ¡vel da AmazÃ´nia Sul-Ocidental. In La IntegraciÃ³n Regional entre BolÃvia , Brasil y Peru. Pp. 281-296. Lima, Peru: CEPEI (Centro Peruano de Estudos Internacionales). Burkhalter, S. Brian, and Robert F. Murphy 1989 Tappers and Sappers: Rubber, Gold and Money Among the MundurucÃº. American Ethnologist 16(1):100-116. Byron, Elizabeth 2003 Market Integration and Health: The Impact of Markets and Acculturation on the Self-Perceived Morbidity, Diet and Nutritional Status of the TismanÃ© Amerindians of Lowla nd Bolivia. Ph.D. Dissertation. Gainesville: University of Florida. Campbell, Constance E. 1996 Forest, Field and Factory: Cha nging Livelihood Strategies in Two Extractive Reserves in the Br azilian Amazon. Ph.D. Dissertation, University of Florida, Gainesville.
290 Castelo, Carlos EstevÃ£o Ferreira 2000 AvaliaÃ§Ã£o EconÃ´mica da ProduÃ§Ã£o Familiar na Reserva Extrativista Chico Mendes no Estado do Acre. Caderno de Pesquisas em AdministraÃ§Ã£o, SÃ£o Paulo 1 (11): 58-64. Caulkins, Douglas 1988 Consensus Analysis: Do Scottish Business Advisors Agree on Models of Success? In Using Methods in the Field: A Practical Introduction and Casebook. V.C. de Munck and E.J. Sobo, eds. Walnut Creek, CA: Alta Mira. Centro Nacional de Desenvolvimento Sustenta do das PopulaÃ§Ãµes Tradicionais (CNPT) 1995 Plano de UtilizaÃ§ao da Reserva Extr ativista Chico Mendes, Acre. Projeto para as Reservas Extrativistas. Brasilia: IBAMA. Chayanov, Alexander V. 1985 On the Theory of Peasant Ec onomy. Madison, WI: University of Wisconsin Press. Christensen, Leslie A. 1997 Introduction to Building a Linear Re gression Model. Proceedings of the Twenty-Second Annual Meeting of the SASÂ® UserÂ’s Group International Conference. Pp. 1234-1240. Cary, NC: SAS Institute. Also available at www2.sas.com/proceedings/sugi22/stats/paper267.pdf Clark, Lauren, Carol Vojir, Nancy O. Hest er, Roxie Foster, and Karen L. Miller 1998 MDS and QAP: How Do Children Rate Painful Experiences. In Using Methods in the Field: A Practical Introduction and Casebook. V.C. de. Munck and E.J. Sobo, eds. Pp. 229-247. Walnut Creek, CA: AltaMira Press. Clay, Jason W. 1993 Strategies for Enhancing Income Generation from the Tropical Forest. In Selected Species and Strategies to Enhance Income Generation from Amazonian Forests. J. Clay and C.R. Clement, eds. Rome: Food and Agriculture Organization of the United Nations. 1992 Some General Principles and Strate gies for Developing Markets in North America and Europe for Non-Timber Forests Products: Lessons from Cultural Survival Enterprises. In Non-Timber Products from Tropical Forests: Evaluation of a Conservati on and Development Strategy. D.C. Nepstad and S. Schwartzman, eds. Pp. 101-106, Advances in Economic Botany, v. 9. Bronx, NY: New York Botanical Garden. Coe, Felix. G., and Gregory. J. Anderson 1996 Ethnobotany of the Garifuna of Eastern Nicaragua. Economic Botany 50(1):71-107.
291 Collier, Richard 1968 The River that God Forgot: The St ory of the Amazon Rubber Boom. New York: Dutton. Conselho Nacional dos Serengueiros (CNS) 1992 RelatÃ³rio SÃ³cio-econÃ´mico e Cadastro da Reserva Extrativista Chico Mendes. Rio Branco, Acre: CNS. Mimeo. Coomes, Oliver T. 1996 Income Formation among Amazonian P easant Households in Northeastern Peru: Empirical Observations and Implications for Market-Oriented Conservation. Yearbook, Conference of Latin Americanist Geographers 22:51-64. Corry, Stephen 1993 The Rainforest Harvest: Who Reaps th e Benefit? The Ecologist 23(4):148153. Costa Sobrinho, Pedro Vicente 1992 Capital e Trabalho na AmazÃ´nia Ocid ental. Rio Branco and SÃ£o Paulo, Brazil: Universidade Federa l do Acre and Cortez Editora. Crook, Carolyn, and Roger Alex Clapp 1998 Is Market-Oriented Forest Conser vation a Contradiction in Terms? Environmental Conservation 25(2):131-145. Cunha, Euclides da, and Leandro Tocantins 1986 Um ParaÃso Perdido. Volume 203. Rio de Janeiro: J. Olympio Editora em co-ediÃ§Ã£o com a FundaÃ§Ã£o Desenvolvimento de Recursos Humanos da Cultura e do Desporto do Governo do Estado do Acre. Daly, Douglas C. 1992 The National Cancer Institute Plant Collections Program: Update and Implications for Tropical Forests. In Sustainable Harvest and Marketing of Rain Forest Products. M.J. Plotkin and L.M. Famolare, eds. Pp. 224-230. Washington, D.C.: Island Press. 1990 Extractive Reserves: A Great New Hope. Garden 14(6):14-15, 17-18, 2021, 32. Daly, D.C., M. Silveira, A.P. Kirschgessner, and A.S. Roberts n.d. Index to Common Names of Plants in Acre. In Floristics and Economic Botany of Acre. Bronx, NY: New York Bo tanical Garden. In Press. Also available at www.nybg.org/bscie/acre/names/htm . D'Andrade, Roy 1995 The Development of Cognitive Anthropology. Cambridge, UK: Cambridge University Press.
292 de Janvry, Alain., and Elisabeth Sadoulet 2001 Income Strategies Among Rural Hous eholds in Mexico: The Role of OffFarm Activities. World Development 29(3):467-480. Dean, Warren 1987 Brazil and the Struggle for Rubber: A Study in Environmental History. Cambridge, UK: Cambridge University Press. Deaton, Angus 1997 The Analysis of Household Surveys: A Microeconometric Approach to Development Policy. Baltimore, MD: Published for the World Bank by Johns Hopkins University Press. Deininger, Klaus, and Pedro Olinto 2001 Rural Nonfarm Employment and Inco me Diversification in Colombia. World Development 29(3):455-465. Demmer, Maria Josefien, and J ohannes Petrus Maria Overman 2001 Indigenous People Conserving the Rain Forest?: The Effect Of Wealth and Markets on the Economic Behaviour of Tawahka Amerindians in Honduras. Wageningen, The Netherla nds: Tropenbos International. Dercon, Stefan 1998 Wealth, Risk and Activity Choice: Catt le in Western Tanzania. Journal of Development Economics 55(1):1-42. Dove, Michael. R. 1993 A Revisionist View of Tropical Deforestation and Development. Environmental Conservation 20(1):17-24. Ellen, Roy. F. 1979 Omniscience and Ignorance Variation in Nuaulu Knowledge, Identification and Classification of Animals. Language in Society 8(3):337-359. Emmons, Louise, and Francois Feer 1997 Neotropical Rainforest Mammals: A Field Guide. Chicago: University of Chicago Press. Escobal, Javier 2001 The Determinants of Nonfarm Inco me Diversification in Rural Peru. World Development 29(3):497-508. Esteves, Benedita Maria Gomes n.d. MudanÃ§a Social: O Deslocamento do Fazer PolÃtico, o Estado e a RacionalizaÃ§Ã£o dos Recursos naturais na Reserva Extrativista Chico Mendes. Rio Branco, Acre. Unpublished manuscript.
293 1999 Do Â“MansoÂ” ao Â“GuardiÃ£oÂ” da Floresta: Estudo do Processo de TransformaÃ§Ã£o Social do Sistema Seri ngal, a Partir do Caso da Reserva Extrativista Chico Mendes. SeropÃ©di ca. Rio de Janeiro: Universidade Federal Rural do Rio de Janeiro. Evaluation Report Vai Quem Querzinho Proj ect: Sustainable Agriculture for Rubber Tappers n.d. Retreived December 6, 2004 from h ttp://www.agronor.org/gumtap.htm. Fearnside, Philip M. 1989 Extractive Reserves in Brazilia n Amazonia. Bioscience 39 (6):387-393. Fleisher, Mark S., and Jennifer A. Harrington 1998 Freelisting: Management at a Women's Federal Prison Camp. In Using Methods in the Field: A Practical Introduction and Casebook. V.C. De Munck and E.J. Sobo, eds. Pp. 69-83. Wa lnut Creek, CA: AltaMira Press. Foster, George 1967 Peasant Society and the Image of the Limited Good. In Peasant Society: A Reader. J.M. Potter, M.N. Diaz, and G.M. Foster, eds. Pp. 300-323. The Little Brown Series in Anthr opology. Boston: Little Brown. Freese, Curtis H. 1998 Wild species as Commodities: Managing Markets and Ecosystems for Sustainability. Washington, DC: Island Press. Furneaux, Robin 1969 The Amazon: The Story of a Great River. London: Hamilton. GalvÃ£o, Eduardo 1952 The Religion of an Amazon Community: A Study in Culture Change. Ph.D. Dissertation, Ann Arbor : University Michigan. Gardner, Peter M. 1976 Birds, Words and a Requiem for the Omniscient Informant. American Ethnologist 3:446-468. Garro, Linda. C. 1986 Intracultural Variation in Folk Medical Knowledge: A Comparison Between Curers and Noncurers. Amer ican Anthropologist 88(2):351-370. Godoy, Ricardo 2001 Indians, Markets, and Rainforests: Theory, Methods, Analysis. New York: Columbia University Press.
294 Godoy, Ricardo, Nicholas Broka w, and David Wilkie 1998 Of Trade and Cognition: Markets and the Loss of Folk Knowledge Among the Tawahka Indians of the Hondur an Rain Forest. Journal of Anthropological Research 54(2):219-233. Godoy, Ricardo, and Marina CÃ¡rdenas 2000 Markets and the Health of I ndigenous People: A Methodological Contribution. Human Organization 59(1):117-124. Godoy, Ricardo, Jeffrey R. Franks , and Mario Alvarado Claudio 1998 Adoption of Modern Agricultural Technologies by Lowland Indigenous Groups in Bolivia: the Role of H ouseholds, Villages, Ethnicity, and Markets. Human Ecology 26(3):351-369. Godoy, Ricardo, Ruben Lubowski, and Anil Markandya 1993 A Method for the Economic Valuation of Non-Timber Tropical Forest Products. Economic Botany 47(3):220-233. Godoy, Ricardo, David Wilkie, and Jeffrey Franks 1997 The Effects of Markets on Neotr opical Deforestation: A Comparative Study of Four Amerindian Societie s. Current Anthropology 38(5):875878. Gomes, Carlos Valerio 2001 Dynamics of Land Use in an Amazoni an Extractive Reserve: The Case of the Chico Mendes Extractive Reserve in Acre, Brazil. MasterÂ’s Thesis. Gainesville: University of Florida. Gram, SÃ¸ren, Lars P. Kvist, and Armando CÃ¡seres 2001 The Economic Importance of Products Extracted from Amazonian Flood Plain Forests. Ambio 30(6):365-368. Gray, Andrew 1990 Indigenous Peoples and the Marketi ng of the Rainforest. The Ecologist 20(6):223-227. Gunatilleke, H. M., D.M.A.H. Senaratne, and P. Abeygunawardena 1993 Role of Non-Timber Forest Produc ts in the Economy of Peripheral Communities of Knuckles National Wilderness Area of Sri-Lanka a Farming Systems-Approach. Economic Botany 47(3):275-281. Hays, Terence E. 1976 An Empirical Method for the Identif ication of Covert Categories in Ethnobiology. American Ethnologist 3:489-507. Hecht, Susanna B., and Alexander Cockburn 1989 The Fate of the Forest: Developers , Destroyers, and Defenders of the Amazon. London and New York: Verso.
295 Hegde, R., and T. Enters 2000 Forest products and household economy: a case study from Mudumalai wildlife sanctuary, southern India. Environmental Conservation 27(3):250259. Hegde, R., S. Suryaprakash, L. Achoth, and K. S. Bawa 1996 Extraction of non-timber forest products in the forests of Biligiri Rangan Hills, India: contribution to rural income. Economic Botany 50(3):243251. Henrich, Joseph. 1997 Market Incorporation, Agricultural Change, and Sustainability Among the Machigueuga Indians of the Peruvian Amazon. Human Ecology 25(2):319-351. Homma, Alfredo Kingo Oyama 1993 Extrativismo Vegetal na AmazÃ´nia. Limites e Oportunidades. BrasÃlia: EMBRAPA-SPI. 1992 The Dynamics of Extraction in Am azonia: A Historical Perspective. In Non-timber Products from Tropical Fore sts: Evaluation of a Conservation and Development Strategy. D.C. Nepstad and S. Schwartzman, eds. Pp. 23-32. Advances in Economic Botany, V. 9. Bronx, NY: New York Botanical Garden. Instituto Brasileiro de Ge ografia e EstatÃstica (IBGE) 2000 AnuÃ¡rio EstatÃstica do Brasil. V.60. Rio de Janeiro: IBGE. Johnson, Jeffrey C. And David C. Griffith 1998 Visual Data: Collection, An alysis, and Representation. In Using Methods in the Field: A Practical Introduc tion and Casebook. V.C. De Munck and E.J. Sobo, eds. Pp. 211-228. Walnut Creek, CA: AltaMira Press. Johnston, M., and A. Colquhoun 1996 Preliminary Ethnobotanical Survey of Kurupukari: An Amerindian Settlement of Central Guyana . Economic Botany 50(2):182-194. Kageyama, Paulo Y. 1991 Extractive Reserves in Brazilia n Amazonia and Genetic Resources Conservation. Paper Presented at the 10th World Forestry Congress, Paris, France. Kainer, Karen A., and Mary L. Duryea 1992 Tapping Womens Knowledge: Plant Re source Use in Extractive Reserves, Acre, Brazil. Economic Botany 46(4):408-425.
296 Kainer, Karen. A., Marianne Schmink, Arthur Cezar Pinheiro Leite, and MÃ¡rio Jorge da Silva Fadell 2003 Experiments in Forest-Based Development in Western Amazonia. Society and Natural Resources 16(10):869-886. Keck, Margaret E. 1995 Social Equity and Environmental Politics in Brazil: Lessons from the Rubber Tappers of Acre. Compar ative Politics 27(4):409-424. Mahar, Dennis J. 1989 Government Policies and Deforestation in Brazil's Amazon Region. Washington, D.C.: World Bank. Mendes, Chico 1989 Fight for the Forest: Chico Mendes in His Own Words. With additional material by Tony Gross. London: Latin America Bureau. Ming, Li Chau, and Ayrton Amaral Junior n.d. Aspectos Etnobotanicos de Plantas Me dicinais na Reserve Extrativista Â“Chico Mendes.Â” In Floristics and Economic Botany of Acre. Bronx, NY: New York Botanical Garden. In Press. Also available at www.nybg.org/bscie/acre/ medicinal/htm . Moran, Emilio 1996 Deforestation in the Brazilian Amazon. In Tropical Deforestation: The Human Dimension. L.E. Sponsel, T.N. Headland, and R.C. Bailey, eds. Pp. 149-164. Methods and Cases in Conservation Science. New York: Columbia University Press. Morduch, Jonathon 1995 Income Smoothing and Consumption Smoothing. Journal of Economic Perspectives 9(3):103-114. Moreno Black, G., and L. L. Price 1993 The Marketing of Gathered Food as an Economic-Strategy of Women in Northeastern Thailand. Human Organization 52(4):398-404. Murphy, Laura, Richard Bilsbo rrow, and Francisco PichÃ³n 1997 Poverty and Prosperity Among Migrant Settlers in the Amazon Rainforest Frontier of Ecuador. Journal of Development Studies 34(2):35-65. Murphy, Robert F. and Julian H. Steward 1956 Tappers and Trappers Parallel Process in Acculturation. Economic Development and Cultural Change 4(4):335-355. Netting, Robert McC. 1993 Smallholders, Householders: Farm Fa milies and the Ecology of Intensive, Sustainable Agriculture. Stanford, CA: Stanford University Press.
297 1989 Smallholders, Householders, Freeholders: Why the Family Farm Works Well Worldwide. In Household Economy: Reconsidering the Domestic Mode of Production. Richard Wilk, ed. Pp. 197-220. Boulder: Westview Press. Neumann, Roderick P., and Edward Hirsch 2000 Commercialization of Non-Timber Fo rest Products: Review and Analysis of Research. Bogor: Center for Inte rnational Forestry Research, and Rome: Food and Agriculture Organization. Nugent, Stephen 1991 The Limitations of Environmental Â‘M anagementÂ’: Forest Utilization in the Lower Amazon. In Environment and Development in Latin America: The Politics of Sustainability. D. Good man and M. Redclift, eds. Pp. 141-154. Manchester, United Kingdom: Ma nchester University Press. Pandit, Bishnu Hari, and Gopal B. Thapa 2003 A tragedy of non-timber forest resources in the mountain commons of Nepal. Environmental Conservation 30(3):283-292. Pattanayak, Subhrendu K., and Erin O. Sills 2001 Do Tropical Forests Provide Natural Insurance? The Microeconomics of Non-timber Forest Product Collection in the Brazilian Amazon. Land economics 77(4):595-613. Pendleton, Linwood H., and E. Lance Howe 2002 Market Integration, Development, and Smallholder Forest Clearance. Land Economics 78(1):1-19. Peters, Charles M., Michael J. Balick , Francis Kahn, and Anthony B. Anderson 1989 Oligarchic Forests of Economic Pl ants in Amazonia: Utilization and Conservation of an Important Tr opical Resource. Conservation Biology 3(4):341-349. Peters, Charles M., Alwyn H. Gentry, and Robert. O. Mendelsohn 1989 Valuation of an Amazonian Rainforest. Nature 339(6227):655-656. Phillips, Oliver 1993 The Potential for Harvesting Fruits in Tropical Rain-Forests: New Data from Amazonian Peru. Biodivers ity and Conservation 2(1):18-38. Plattner, Stuart 1989 Economic Anthropology. Stanford, CA: Stanford University Press. Plotkin, Mark J. 1993 Tales of a ShamanÂ’s Apprentice: An Ethnobotanist Searches for New Medicines in the Amazon Rain Forest. New York: Viking.
298 Plotkin, Mark J., and Lisa M. Famolare (Eds) 1992 Conclusions and Recommendations. In Sustainable Harvest and Marketing of Rain Forest Products. M.J. Plot kin and L.M. Famolare, eds. Pp. 310312. Washington, D.C.: Island Press. Projeto Resex 2000 RelatÃ³rio Final. BrasÃli a: MMA/IBAMA/PPG-7/CNPT. Rancy, Cleusa Maria Damo 1992 RaÃzes do Acre. Rio Branco, Acre: Falangola EditÃ´ra. Reardon, Thomas, Julio BerdeguÃ©, and GermÃ¡n Escobar 2001 Rural Nonfarm Employment and Incomes in Latin America: Overview and Policy Implications. Worl d Development 29(3):395-409. Reardon, Thomas and Stephen A. Vosti 1995 Links Between Rural Poverty a nd the Environment in Developing Countries: Asset Categori es and Investment Poverty. World Development 23(9):1495-1506. Rego, JosÃ© Fernandes do 1999 Amazonia: do Extrativismo ao Neoe xtrativismo. CiÃªnci a Hoje 25(147):6265. Reyes-Garcia, V., R. Godoy, V.Vadez, E. By ron, T. Huanca, W.R. Leonard, E. PÃ©rez, and D. Wilkie. 2003 Ethnobotanical Knowledge Shar ed Widely Among TsimaneÂ’ Amerindians, Bolivia. Science 299(5613):1707. Reyes-Garcia, Viki 2001 Indigenous People, Ethnobotanical Knowledge, and Market Economy: A Case Study of the Tsimane. Ph.D. Disse rtation. Gainesville: University of Florida. Robinson, Elizabeth. J. Z., Jeffrey C. Williams, and Heidi J. Albers 2002 The Influence of Markets and Polic y on Spatial Patterns of Non-Timber Forest Product Extraction. Land Economics 78(2):260-271. Romney, A. Kimball, John P. Boyd, Carmella C. Moore, William H. Batchelder, and Timothy J. Brazil 1996 Culture as Shared Cognitive Representations. Proceedings of the National Academy of Sciences of the Unite d States of America 93(10):4699-4705. Romney, A. Kimball, Susan C. We ller, and William H. Batchelder 1986 Culture as Consensus: A Theory of Culture and Informant Accuracy. American Anthropologist 88(2):313-338.
299 Roos, Gun 1998 Pile Sorting: Â“Kids Like CandyÂ”. In Using Methods in the Field: A Practical Introduction and Casebook. V.C. De Munck and E.J. Sobo, eds. Pp. 97-109. Walnut Creek, CA: AltaMira Press. Ross, Norbert 2002 Cognitive Aspects of Intergenerational Change: Mental Models, Cultural Change, and Environmental Behavior Among the Lacandon Maya of Southern Mexico. Human Organization 61(2):125-138. Ryan, Gery W., and H. Russell Bernard 2003 Techniques to Identify Themes . Field Methods 15(1):85-109. Sahlins, Marshall David 1972 Stone Age Economics. Chicago: Aldine-Atherton. Santos, Roberto 1980 HistÃ³ria EconÃ´mica da AmazÃ´nia ( 1800-1920). SÃ£o Paulo: T.A. Queiroz. Schmink, Marianne 1992 Amazonian Resistance Movements and the International Alliance. In Ecological Disorder in Amazonia: Soci al Aspects. L.A. Kosinski, ed. Rio de Janeiro: UNESCO/ISSC/Educam. 2003 No Longer Invisible, but Still Enig matic: Amazonian Peasant Identities and Cosmographies. Reviews in Anthropology 32 (3):223-237. Schmink, Marianne and Charles H. Wood 1992 Contested Frontiers in Amazonia. Ne w York: Colombia University Press. Schwartzman, Stephan 1992 Land Distribution and Social Costs of Frontier Development in Brazil: Social and Historical Context of Extractive Reserves. In Non-Timber Products from Tropical Forests: Evaluation of a Conservation and Development Strategy. D.C. Nepstad and S. Schwartzman, eds. Pp. 51-66. Advances in Economic Botany, V. 9. Bronx, NY: New York Botanical Garden. 1989 Extractive Reserves: The Rubber TappersÂ’ Strategy for Sustainable Use of the Amazon Rainforest. In Fragile Lands of Latin America: Strategies for Sustainable Development. J.O. Browder, ed. Pp. 150-163. Westview Special Studies in Social, Political, and Economic Development. Boulder: Westview Press. Secretaria Executiva de Flores tas e Extrativismo (SEFE) n.d. Unpublished data. Rio Branco, Acre: SEFE.
300 Shackleton, Charlie M. 2001 Re-Examining Local and Market-Orien tated Use of Wild Species for the Conservation of Biodiversity. Enviro nmental Conservation 28(3):270-278. Shanley, Patricia, Alan R. Pierce, Sarah A. Laird and Abraham Guillen 2003 Tapping the Green Market: The Cert ification and Management of NonTimber Forest Products. London and Sterling, VA: Earthscan Publications. Sick, Helmut 1993 Birds in Brazil: A Natural History. Il lustrated by Paul Barruel with Color Plates by Paul Barruel and John P. OÂ´Neill. Translated from the Portuguese by William Betton. Princeton, New Jersey: Princeton University Press. Silveira, M., N. M. C. de Paula, I. F. Brown, H. B. N. Borges, D. Daly and L. A. Ferreira 1997. Os Â“Buracos NegrosÂ” da Divers idade: Estudos no Acre Revelam Precariedade do Conhecimento sobre a Flora AmazÃ´nica. CiÃªncia Hoje 22(128):64-65 Smith, Nigel J. H. 1996 The Enchanted Amazon Rain Fores t: Stories from a Vanishing World. Gainesville: University Press of Florida. Souza, Francisco Kennedy 2001 Pasture Instead of Rubber? The Ra nching Tendencies of Family-Based Agriculture in Extractive Reserves and Colonization Projects in Acre, Brazil, Southwestern Amazonia. Pa per Presented at the 2001 Open Meeting of the Human Dimensions of Global Environmental Change.Research Community. October 6-8, Rio de Janeiro, Brazil Spicer, Edward H. 1971 Persistent Cultural Systems Comparative Study of Identity Systems That Can Adapt to Contrasting Envi ronments. Science 174(4011): 795-800. Sponsel, Leslie E., Thomas N. Headla nd, and Robert Converse Bailey (eds.) 1996 Tropical Deforestation: The Huma n Dimension. New York: Columbia University Press. Stanfield, Michael Edward 1998 Red Rubber, Bleeding Trees: Violence , Slavery, and Empire in Northwest Amazonia, 1850-1933. Albuquerque, N.M.: University of New Mexico Press. State of Acre 2000a Zoneamento EcolÃ³gico-EconÃ´mico do Acre. Indicativos para a GestÃ£o Territorial do Acre. Volume 3. Rio Branco, Acre, Brazil: Secretaria de Estada de CiÃªncia, Tecnologia e Meio Ambiente.
301 2000b Zoneamento EcolÃ³gico-EconÃ´mico do Acre. Aspectos SÃ´cio-EconÃ´micos e OcupaÃ§Ã£o Territorial. Volume 2. Rio Br anco, Acre, Brazil: Secretaria de Estada de CiÃªncia, Tecnologia e Meio Ambiente. 2000c Zoneamento EcolÃ³gico-EconÃ´mico do Acre. Recursos Naturais e Meio Ambiente. Volume 1. Rio Branco, Acre, Brazil: Secretaria de Estada de CiÃªncia, Tecnologia e Meio Ambiente. 2000d Zoneamento EcolÃ³gico-EconÃ´mico do Acre. Dados Digitais. Rio Branco, Acre, Brazil: Secretaria de Estada de CiÃªncia, Tecnologia e Meio Ambiente. 1999 Development That We Strive For. Acre, Brazil: Sate of Acre. Strudwick, Jeremy and Gail L. Sobel 1988 Uses of Euterpe oleracea Mart. in the Amazon Esturary, Brazil. In The Palm Tree of Life. Michael J. Balick, ed. Pp. 225-253. Advances in Economic Botany, V. 6. Bronx, NY: The New York Botanical Garden. Takasaki, Yoshito, Bradford L. Barham, and Oliver T. Coomes 2000 Wealth Accumulation and Activity Choice Evolution Among Amazonian Forest Peasant Households. Depart ment of Agricultural and Applied Economics, University of Wisconsin-Madison, Staff Paper, No. 434, September. Vasquez, Rodolfo, and Alwyn H. Gentry 1989 Use and Misuse of Forest-Harvest ed Fruits in the Iquitos Area. Conservation Biology 3(4):350-361. Wagley, Charles 1953 Amazon Town: A Study of Man in the Tropics. New York: Macmillan. Wallace, Richard H. 2002 ExploraÃ§Ã£o e Potencial de Produtos Florestais NÃ£o-MadarÃ¡veis na AmazÃ´nia. Paper presented at the II Congresso Ibero-Americano de Pesquisa e Desenvolvimento de Produt os Florestais, I SeminÃ¡rio em Tecnologia de Madeira e Produtos NÃ£ o-MaderÃ¡veis, September 9 to 13, Curitiba, Brazil. Weigand, Ronaldo 1996 Conservation Attitudes in Three Rubber Tapper Communities in Acre, Brazil. MasterÂ’s Thesis. Gainesville; University of Florida. Weinstein, Barbara 1983 The Amazon Rubber Boom, 1850-1920. Stanford, CA: Stanford University Press.
302 1980 Capital Penetration and Problems of Labor Control in the NineteenthCentury Amazon. Meeting of the Latin American Studies Association. Weller, Susan C. 1987 Shared Knowledge, Intracultural Vari ation, and Knowledge Aggregation. American Behavioral Scientist 31(2):178-193. Wolf, Eric R. 1966 Peasants. Englewood Cliffs, NJ: Prentice-Hall. Wolf, Howard, and Ralph Frank Wolf 1936 Rubber: A Story of Glory and Greed. New York: Covici Friede. Wollenberg, Eva and Andrew Ingles (eds.) 1998 Incomes from the Forest: Methods for the Development and Conservation of Forest Products for Local Commun ities. Bogor, Indonesia: Center for International Forestry Research, World Conservation Union. Wood, Charles H. 2002 Introduction: Land Use and Deforestation in the Amazon. In Deforestation and Land Use in the Amazon. Charles H. Wood and Roberto Porro, eds. Pp. 1-38. Gainesville: Univer sity Press of Florida. Ziffer, Karen 1992 The Tagau Initiative: Building the Market for a Rain Forest Product. In Sustainable Harvest and Marketing of Rain Forest Products. M.J. Plotkin and L.M. Famolare, eds. Pp. 274-279. Washington, D.C.: Island Press.
303 BIOGRAPHICAL SKETCH Richard Hood Wallace was raised in Lancas ter, Ohio. He received a B.S. in marketing with a minor in international busin ess from Miami University in Oxford, Ohio, in 1984, where he also attended the Miami University Luxembourg Program. After a brief international banking career with Lloyds Ba nk Plc. in Chicago, he received an M.A. in international development from The Amer ican University in Washington, D.C., in 1994. After finishing his masterÂ’s degree, he conducted research in Acre, Brazil, from 1995-1997 under the Rainforest Alliance Kleinha ns Fellowship. Upon completion of his doctoral studies in anthropology at the Univer sity of Florida, he will work as a postdoctoral research associ ate in Acre, Brazil.