Citation
Effectiveness of Extractive Reserves, Agro-Extractive Settlements, and Colonist Settlements in Southwestern Amazonia: An Economic and Land-Cover Comparison of Three Land Tenure Types in Acre, Brazil

Material Information

Title:
Effectiveness of Extractive Reserves, Agro-Extractive Settlements, and Colonist Settlements in Southwestern Amazonia: An Economic and Land-Cover Comparison of Three Land Tenure Types in Acre, Brazil
Creator:
ARAUJO DE SOUZA, FRANCISCO KENNEDY ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Community forestry ( jstor )
Crops ( jstor )
Deforestation ( jstor )
Economics ( jstor )
Forests ( jstor )
Land cover ( jstor )
Land tenure ( jstor )
Land use ( jstor )
Rubber ( jstor )
Timber ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Francisco Kennedy Araujo De Souza. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
7/24/2006
Resource Identifier:
496181277 ( OCLC )

Downloads

This item has the following downloads:

souza_f ( .pdf )

souza_f_Page_127.txt

souza_f_Page_130.txt

souza_f_Page_112.txt

souza_f_Page_129.txt

souza_f_Page_059.txt

souza_f_Page_150.txt

souza_f_Page_086.txt

souza_f_Page_102.txt

souza_f_Page_147.txt

souza_f_pdf.txt

souza_f_Page_088.txt

souza_f_Page_071.txt

souza_f_Page_047.txt

souza_f_Page_114.txt

souza_f_Page_089.txt

souza_f_Page_050.txt

souza_f_Page_002.txt

souza_f_Page_042.txt

souza_f_Page_097.txt

souza_f_Page_004.txt

souza_f_Page_090.txt

souza_f_Page_019.txt

souza_f_Page_029.txt

souza_f_Page_079.txt

souza_f_Page_173.txt

souza_f_Page_020.txt

souza_f_Page_132.txt

souza_f_Page_110.txt

souza_f_Page_170.txt

souza_f_Page_109.txt

souza_f_Page_035.txt

souza_f_Page_068.txt

souza_f_Page_123.txt

souza_f_Page_163.txt

souza_f_Page_080.txt

souza_f_Page_054.txt

souza_f_Page_049.txt

souza_f_Page_074.txt

souza_f_Page_062.txt

souza_f_Page_105.txt

souza_f_Page_146.txt

souza_f_Page_065.txt

souza_f_Page_027.txt

souza_f_Page_104.txt

souza_f_Page_169.txt

souza_f_Page_144.txt

souza_f_Page_159.txt

souza_f_Page_126.txt

souza_f_Page_094.txt

souza_f_Page_026.txt

souza_f_Page_044.txt

souza_f_Page_171.txt

souza_f_Page_166.txt

souza_f_Page_106.txt

souza_f_Page_082.txt

souza_f_Page_076.txt

souza_f_Page_060.txt

souza_f_Page_096.txt

souza_f_Page_120.txt

souza_f_Page_145.txt

souza_f_Page_087.txt

souza_f_Page_151.txt

souza_f_Page_167.txt

souza_f_Page_133.txt

souza_f_Page_015.txt

souza_f_Page_041.txt

souza_f_Page_075.txt

souza_f_Page_025.txt

souza_f_Page_073.txt

souza_f_Page_005.txt

souza_f_Page_007.txt

souza_f_Page_010.txt

souza_f_Page_023.txt

souza_f_Page_085.txt

souza_f_Page_034.txt

souza_f_Page_061.txt

souza_f_Page_140.txt

souza_f_Page_028.txt

souza_f_Page_009.txt

souza_f_Page_174.txt

souza_f_Page_067.txt

souza_f_Page_103.txt

souza_f_Page_168.txt

souza_f_Page_125.txt

souza_f_Page_149.txt

souza_f_Page_128.txt

souza_f_Page_155.txt

souza_f_Page_022.txt

souza_f_Page_055.txt

souza_f_Page_113.txt

souza_f_Page_134.txt

souza_f_Page_164.txt

souza_f_Page_153.txt

souza_f_Page_046.txt

souza_f_Page_063.txt

souza_f_Page_161.txt

souza_f_Page_099.txt

souza_f_Page_001.txt

souza_f_Page_016.txt

souza_f_Page_013.txt

souza_f_Page_136.txt

souza_f_Page_039.txt

souza_f_Page_158.txt

souza_f_Page_018.txt

souza_f_Page_156.txt

souza_f_Page_066.txt

souza_f_Page_135.txt

souza_f_Page_162.txt

souza_f_Page_058.txt

souza_f_Page_057.txt

souza_f_Page_101.txt

souza_f_Page_084.txt

souza_f_Page_003.txt

souza_f_Page_051.txt

souza_f_Page_137.txt

souza_f_Page_098.txt

souza_f_Page_014.txt

souza_f_Page_064.txt

souza_f_Page_108.txt

souza_f_Page_083.txt

souza_f_Page_165.txt

souza_f_Page_118.txt

souza_f_Page_056.txt

souza_f_Page_154.txt

souza_f_Page_078.txt

souza_f_Page_031.txt

souza_f_Page_052.txt

souza_f_Page_011.txt

souza_f_Page_017.txt

souza_f_Page_070.txt

souza_f_Page_093.txt

souza_f_Page_142.txt

souza_f_Page_143.txt

souza_f_Page_117.txt

souza_f_Page_032.txt

souza_f_Page_119.txt

souza_f_Page_033.txt

souza_f_Page_038.txt

souza_f_Page_012.txt

souza_f_Page_111.txt

souza_f_Page_072.txt

souza_f_Page_115.txt

souza_f_Page_008.txt

souza_f_Page_053.txt

souza_f_Page_048.txt

souza_f_Page_077.txt

souza_f_Page_124.txt

souza_f_Page_043.txt

souza_f_Page_157.txt

souza_f_Page_172.txt

souza_f_Page_131.txt

souza_f_Page_116.txt

souza_f_Page_081.txt

souza_f_Page_152.txt

souza_f_Page_024.txt

souza_f_Page_138.txt

souza_f_Page_040.txt

souza_f_Page_021.txt

souza_f_Page_030.txt

souza_f_Page_148.txt

souza_f_Page_006.txt

souza_f_Page_091.txt

souza_f_Page_092.txt

souza_f_Page_107.txt

souza_f_Page_160.txt

souza_f_Page_036.txt

souza_f_Page_139.txt

souza_f_Page_141.txt

souza_f_Page_122.txt

souza_f_Page_037.txt

souza_f_Page_069.txt

souza_f_Page_121.txt

souza_f_Page_100.txt

souza_f_Page_045.txt

souza_f_Page_095.txt


Full Text











EFFECTIVENESS OF EXTRACTIVE RESERVES, AGRO-EXTRACTIVE
RESERVES AND COLONIST SETTLEMENTS IN SOUTHWESTERN AMAZONIA:
AN ECONOMIC AND LAND COVER COMPARISON OF THREE LAND TENURE
TYPES IN ACRE, BRAZIL














By

FRANCISCO KENNEDY ARAUjJO DE SOUZA


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS

UNIVERSITY OF FLORIDA


2006
































Copyright 2006

by

Francisco Kennedy Arauj o de Souza



































To: my Apurinan indigenous people, my deceased grandmother Nadira Valuah, my
mother Mada, and my daughter Yara.
















ACKNOWLEDGMENTS

This thesis was the result of many supporters, co-workers, friends, and community

partnerships. First, I would like to express my sincere gratitude to the rubber tapper and

colonist families with whom I have collaboratively worked over the past eight years in

the State of Acre, Brazil. Drivers, researchers, and friends at the Federal University of

Acre, Centro dos Trabalhadores da Amaz8nia (CTA), Group for Agroforestry System

Research (PESACRE), and Acre government were fundamental to generate the results

presented in this thesis. Especially I thank my research assistants, Tatiana, Marcio,

Emilton, Luciano, and Dermerson.

I am also grateful for the financial support provided by many organizations. First, I

thank the Ford Foundation International Fellowship Program for its funding my graduate

studies at University of Florida. In Brazil Maria Luiza and Marcia, at the Fundagio

Carlos Chagas, were fundamental in supporting my first steps in the USA. I thank

Gregory Marino for his institutional support at the Institute of Intemnational Education in

New York. I sincerely express my appreciation to the Federal University of Acre for

permitting me to pursue my graduate study. I sincerely thank the Tropical Conservation

Development Program for its financial support during my last semester at the University

of Florida. I also extend special thanks to the Woods Hole Research Center for

supporting my fieldwork in Acre in 2004. Two government agencies in Acre, SEPROAF

and IMAC, contributed with transport and Landsat images.










My supervisory committee was fundamental in my effort to complete this graduate

course. First, I sincerely express my gratitude to my chair, Dr. Stephen Perz, for his

constant support, encouragement, patience, and friendship. I thank Dr. Charles Wood for

understanding my English limitation during my first semester at UF. Moreover Dr. Wood

was essential in advancing my statistical data analyses. I sincerely thank Dr. Marianne

Schmink for her friendship, patience, encouragement, and commitment in collaborating

with the Amazonian people. I am especially indebted to Dr. Schmink for her support in

one of the most critical moments of my life.

Innumerable friends and co-workers supported me before and during my studies,

fieldwork, and the writing process of my thesis. Valerio has been a unique friend who

since my first steps in Gainesville has supported me with friendship, encouragement,

ideas, and professional experience. Geraldo, like Marianne, became my kariuk~~~~kkkkk~~~~kkkk brother and

sister after their important support during my critical phase in Gainesville. I thank Foster

Brown, Daniel Nepstad, and Eduardo Brondizio for their letters supporting my

application to the Ford Foundation. I am grateful to Richard Wallace, Christiane

Ehringhaus, Christie Klimas, Joanna Tucker, Valerio, and Jeff for their support in

reading, editing, and suggesting ideas in my thesis. I express my thanks to Wendy Lin for

her support with my papers. I thank many other friends: Diogo, Rodrigo, Charle

Crisostomo, Silvia Brilhante, Nivia, Pedrio, Magna, Roger, Savio Maia, Edson Carvalho,

Lira, Hulk, Marilene, Sumaia, Fadel, Ronei, Elaine, Andrea, and Leonardo.

Finally, I express my gratitude to my parents and my brother and sisters for their

examples of vigor and humanity. I am indebted to Yara, my daughter, for putting up with

my long time far away from her.


















TABLE OF CONTENTS


page

ACKNOWLEDGMENT S .............. .................... iv


LI ST OF T ABLE S ........._._.. ........ ............... ix...

LIST OF FIGURES .............. ...............x.....


AB STRAC T ................ .............. xi

1 INTRODUCTION ................. ...............1.......... ......


Research Objectives............... ...............
Overview of the Research Sites ................. ...............3............ ...
Theoretical Framework............... ...............7
M ethodology Summary .............. ......... .. ...... .. .......1
Study Significance: Local Research Needs on Land Tenure Analyses ................... ...11
Structure of the Thesis............... ...............12.


2 LAND USE MANAGEMENT AND LAND TENURE DIVERSITIES:
ECONOMIC BOOMS, State POLICIES, AND LAND RIGHTS INT ACRE............15

Introducti on ............... .. ... ..._. .... ...............15
Land Tenure Regimes in the Amazon Region ........___......... .. .. ..... .........._._....16
The State of Acre: History, Social Movements, and Land Tenure Diversity ............19
Geographic Context............... ...............19
Acre: Before the Ranchers................... ..... .................2
Agriculture Modernization: Large Cattle Ranchers, Rubber Tappers, and
C olonists ................ ........... .. ......... ... .......2
After the 1990s: No Economic Boom, but a Rubber Tapper Victory .................23
Acre: Social Movements, Community Land Tenures, and Forest-Based
Development Policies .............. .. .. ......... .. ... .......2
Neoextrativismo: A Long-Term and Systemic Approach ................ ...............27
Florestania: Citizenship Based on Regional Integration, Short-Term
Economic Growth, and Conservation .......__............ ._. ... ......_.._.. .....29
Community Land Tenure: An Emergent Issue for Assessment in Acre ....................31
Colonization Settlements (PC) .............. ...............33....
Extractive Reserves (Re sex) ................. .......... ......... ....... .....
Agro-extractive Settlements (PAE) ................. ...............36........... ....
The Study Sites ............... .. ..... .... ...... ........... .....................3
Peixoto Colonization Settlement: The Agrosilvopastoral Model ................... .....37












Porto Dias Agro-extractive Settlement: The Timber Forest Management
M odel ............... .. .......... .. ..... .. ... .......4
Chico Mendes Extractive Reserve: The Non-timber Forest Management
M odel .............. ...............42....
Conclusion ................ ...............45.................


3 METHODS AND TECHNIQUES FOR ECONOMIC AND REMOTE SENSING
ANALY SE S .............. ...............47....


Introducti on ............ ..... ._ ...............47....
The Socioeconomic Survey .............. ...............49....
Household Sampling .............. ...............50....
The Questionnaire .............. ...............51....
The Household Fieldwork ............ ......_ ...............52..
The Price Survey .............. .... .. ....... .. ... .. ....................5
Participatory Findings Assessment: Attempts for Calibration and Validation ...53
Land Cover Data Collection and Fieldwork ......____ ..... ... .__ ........._.......5
The Post-Survey: Data Processing and analysis............... ...............55
Household Economic Approach .....__.....___ ..........._ ............5
Economic Analysis ............ ..... ._ ...............58....
Land Cover Data Processing .............. ...............59....
Economic and Land Cover Integration............... ..............6
Conclusion ............ ..... ._ ...............62...


4 ECONOMIC AND LAND USE COMPARISON OF SMALLHOLDER LAND
TENURE MODELS INT THE STATE OF ACRE ................. .......... ...............64


Introducti on .................. ... ......... ...............64.......
Revising the Research Questions............... ...............6
Land Cover Distribution among Land Tenures ................ ................ ......... .67
Accuracy Assessment of Land Cover. ................ ............ ...................67
Comparing Land Cover among Land Tenures .............. ...............68....
Economic Analysis for Land Tenures in the State of Acre .............. ....................74
Gross Income ...._ ................. ...............74.......
General Cost Analysis ................ .......... ...............78......
Cost Analysis at Community Level ................. ...............82........... ...
Profitability Analysis............... .. ... .... .... ...............8
Integrating Land Cover and Economic Analyses for Understanding Land Tenure ...89
Conclusion ................ ...............96.................


5 CONCLUSIONS: DILEMMAS, IMPLICATIONS, AND FUTURE RESEARCH..99


Summarizing Key Findings ............ ....._ ............... 100.
Policy Implications ............_..........__ .. .. ..._ ... ..........10
Emergent Questions for Community Land Tenure in Acre............... ..................0











APPENDIX


A QUESTIONNAIRE USED INT THE CHICO MENDES EXTRACTIVE
RE SERVE ................. ...............107.............


B QUESTIONNAIRE USED INT THE PORTO DIAS AGRO-EXTRACTIVE
SETTLEMENT ............ ..... ._ ...............122...

C QUESTIONNAIRE USED INT THE PEIXOTO COLONIZATION
SETTLEMENT ................. ...............137................

D CIPEC TRAININ\G SAMPLE PROTOCOL ................. ............... ......... ...151

LIST OF REFERENCES ................. ...............155................

BIOGRAPHICAL SKETCH ................. ...............162......... ......


















LIST OF TABLES


Table pg

4-1. Accuracy assessment for 2004 supervised classification for the study sites in
Acre, Brazil. ............. ...............68.....

4-2: Land cover categories at household level, all sites (n=88), Acre, Brazil. ..................70

4-3. Descriptive statistics for land cover categories, by land tenure type, Acre, Brazil,
2004............... ...............72..

4-4. Descriptive statistics for annual income, by land tenure type, Acre, Brazil, 2004.....75

4-5. Descriptive statistics for annual cost categories, all sites (n=88), Acre, Brazil,
2004............... ...............79..

4-6. Annual mean for cost categories, by land tenure type, Acre, Brazil. 2004 ................83

4-7. Annual mean for profit for economic categories, by land tenure type, Acre, Brazil,
2004............... ...............86..

4-8. Site model at land cover types, all sites (n=88), Acre, Brazil, 2004. .........................91

4-9. Life cycle model at land cover types, all sites (n=88), Acre, Brazil, 2004. ...............92

4-10. Market integration model at land cover types, all sites (n=88), Acre, Brazil,
2004............... ...............93..

4-11. Investment model at land cover types, all sites (n=88), Acre, Brazil, 2004.............94

4-12. Assets model at land cover types, all sites (n=88), Acre, Brazil, 2004. ...................96


















LIST OF FIGURES


Figure pg

1-1. State of Acre Brazil, identifying colonizazation, agro-extractive and conservation
areas. ............. ...............4.....

1-2. Deforestation in the study sites, Acre, Brazil. ............. ...............6.....

3-1. Diagram of land cover processing in the study area, Acre, Brazil. ............................60
















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Arts

EFFECTIVENESS OF EXTRACTIVE RESERVES, AGRO-EXTRACTIVE
RESERVES AND COLONIST SETTLEMENTS IN SOUTHWESTERN AMAZONIA:
AN ECONOMIC AND LAND COVER COMPARISON OF THREE LAND TENURE
TYPES IN ACRE, BRAZIL

By

Francisco Kennedy Arauj o de Souza

May 2006

Chair: Stephen G. Perz
Major Department: Latin American Studies

Linking economic and remote sensing techniques, a comparison of an extractive

reserve (Resex), an agro-extractive settlement (PAE), and a colonist settlement (PC) were

conducted in the state of Acre, Brazil. These land tenure categories were represented by

Chico Mendes Resex, Porto Dias PAE, and Peixoto PC respectively. The datasets used in

this research were constituted of three components. First, socioeconomic data were

gathered from 88 households in the three sites in 2004. Second, 442 GPS land cover

samples were collected in 2004 and 2005. Third, I used ancillary data and Landsat TM 5,

bands 1 through 5, scenes 67-001 and 67-002 for performing a supervised classification

at community and household level within the three land tenures.

Economic findings indicated that while forest management was the more important

activity among rubber tappers in the Resex and PAE, in Peixoto agrosilvopastoral

strategies had little economic importance. However, high costs were related to all










economic activities, especially timber forest management. Moreover, non-timber forest

products were more effective than timber forest management.

The land cover analysis indicated that extractive reserves and agro-extractive

settlements are most effective to address conservation. In comparison to PC lots that

showed a mean of 36% of deforestation in 2004, selected colocapies in the Resex and

PAE had a mean of 5% and 12% of deforestation respectively. However, while pasture

was the category that most contributed to deforestation in Peixoto and Porto Dias, crops

were the most prominent cause of forest clearing in the Chico Mendes Resex.

An integrated analysis of economic and land cover data produced two main results.

First, the multivariate models indicated that the establishment of extractive reserves and

agro-extractive reserves may be an efficient strategy to achieve conservation goals.

However, this policy should be followed by state enforcement initiatives because the life

cycle model addressed in this study suggested that more recent inhabitants tend to have a

high propensity to deforest. This tendency may be increased if investment in roads is

carried out.

While this study is an initial step in linking economic and land cover analysis at the

household level, it may provide insights for the formulation of conservation policies in

Acre where a comprehensive and innovative policy has been implemented since 1999.

Core components of that policy include community forest management and sustainable

land uses based on forest management and agrosilvopastoral production. Findings from

this thesis will hopefully help establish new strategies for consolidating extractive

reserves, agro-extractive reserves, and colonist settlements based on a new development

model in Amazon.















CHAPTER 1
INTTRODUCTION

Community land tenures and their land uses in Amazonia have been the obj ect of

intensive debate, with supporters (e.g., Schwartzman 1989, Allegretti 1990, Anderson

1990) and opponents (e.g., Homma 1989, Browder 1990). While public policies in this

region have focused on regional development models based on private control of large

tracts of land (e.g., ranchers, miners, and loggers), community land regimes have been

considered economically unattractive. By the same token, forest and sustainable land uses

have been criticized as an impediment to economic growth (Hecht 1985). In contrast with

these perspectives, over the past seven years, the government in the Brazilian state of

Acre has made efforts to consolidate a policy based on community based land tenure and

sustainable land use strategies.

In this context, it has been crucial to understand and empirically assess the role that

timber-forest management, non-timber forest management, and agrosilvopastoral systems

have in providing subsistence revenue and income accumulation, within extractivist,

agro-extractivist, and colonist communities. The potential effects of these land use

strategies for household livelihood systems and their implications for land cover change

are not well-understood. Kaimowitz and Angelsen (1998) argue that trained surveyors

and time-consuming data collection are a main constraint for farm-level analyses.

However, focusing on this type of study, some scholars (Carpentier et al. 2000, Tomich

et al. 1998) have observed that rising cash-incomes among small landholders in

Amazonia has generated changes in their behavior.










The economic efficiency of sustainable initiatives, then, is expected to promote

changes in natural resource use. Among smallholders in Acre I have observed that

increased income has contributed to the substitution of subsistence production by

manufactured products. Families have focused their interests on higher priced products to

meet their new needs. As a result, this commonly increases pressure on the forest to meet

these demands.

Although a broad range of issues are related to community land use and land cover

change in Amazonia, a question that emerged as central issue of this thesis is: which

community land tenure is most effective in balancing economic and forest conservation

goals? Almost 30 years after the most intense and violent phase of land conflicts in

Amazonia, this region has an opportunity to merge national and state policies into a

common agenda for addressing economic growth guided by environmental conservation.

This provides a rare opportunity to determine whether areas managed by communities,

such as extractive reserves, agro-extractive settlements and colonization settlements, can

be a potential development model for the Amazon region.

Research Objectives

The goal of this thesis is to respond to the need for an empirical assessment of

community land tenures as economic and environmental conservation strategies in the

Brazilian Amazon. This study responds to the emergent demand to understand the effects

non-timber and timber forest management as well as agrosilvopastoral strategies in

extractive reserves, agro-extractive settlements, and colonist settlements in the state of

Acre, Brazil. To address this issue, I used data sets from three community land tenure

categories located in this state. The research obj ectives were threefold: 1) to compare

economic and environmental conservation in three distinct land tenure categories










designed to achieve different policy obj ectives; 2) to economically evaluate non-timber

and timber forest products in terms of their cost and benefits; and 3) to identify factors

most important for forest clearing in three distinct land tenure types.

Overview of the Research Sites

The state of Acre is located in southwestern Amazonia and its 153,000 km2 Of area

was occupied by over 550,000 people in 1999 (Govemno do Estado do Acre 2000a). It

borders the Brazilian states of Rond8nia and Amazonas and the tri-national frontier

known since 1999 as the MAP region, which also includes the Bolivian State of Pando

and the Peruvian state of Madre de Dios (Figure 1-1). Its territory still has almost 90%

intact rainforest which contains diverse human and ecological mosaics with complex

interactions at local and regional levels (Governo do Estado do Acre 2000a, Silveira et al.

1997). However, there is the possibility that this landscape will be drastically altered over

the next few years. A coalition of Brazilian, Peruvian, and international organizations

such as Inter-American Development Bank (IDB) is in the process of constructing and

paying the Pacific Highway which would be one of the most important roads for

transportation of commodities such as soybeans in South America. It may also increase

the probability of environmental degradation as a result of logging, soybean expansion,

and migratory movements (Carvalho et al. 2002).

The state of Acre, located between a degraded area (arc of deforestation) and a

region with less deforestation (Andean mountains), exhibits important community land

tenure diversity. Rubber tappers, indigenous people, and colonists, who have been the

priority social groups for governance policies implemented since 1999, control almost

44% of the territory (Figure 1-1) (Govemno do Estado do Acre 2000a). The future amount










of deforestation as well as potential negative effects of the Pacific Highway will be

strongly determined by land use decisions taking place in these communities.

Three study areas located in Acre's eastern region were selected for this research

based on three conditions: 1) they represented a community land tenure category:

extractive reserve, agro-extractive settlements, and colonist settlements; 2) their

inhabitants are addressing sustainable land uses based on timber management, non-timber

forest management, or agrosilpastoral initiatives; and 3) they reflect public policies

initiatives for sustainable land use. As a result of these criteria three land tenure types

were selected.

The Chico Mendes Reserve, with almost one million hectares, was the largest

extractive reserve in Brazil until 2004. It is sub-divided into 45 rubber estates (seringais)










AMAZONAS





-RQNDONIA


N Mad eE d~eUDios
I~ Pacific Highway
AgPR) a ro-exrctv stleet
Pandoacie etlmet
100 0 100 Miles \(BOLIVIA) I Conservation areas
colonization Settlements

Figure 1-1. State of Acre Brazil, identifying colonizazation, agro-extractive and
conservation areas.









and each has individual rubber tapper landholdings (colocapi~es). Three seringais, Sho

Pedro, Palmari, and Floresta (Figure 1-2, A), were selected. In this area, a coalition of

organizations led by state government agencies have sought to amplify the range of non-

timber forest products explored by local residents over the past 7 years. The National

Center for the Sustainable Development of Traditional Populations (CNPT), a bureau

under supervision of the Brazilian Institute for Environmental and Renewable Natural

Resources (IBAMA), is the governmental agency responsible for Resex' administration

in collaboration with rubber tapper associations.

Although the Porto Dias Agro-extractive Settlement' s 22,000 hectares are occupied

predominantly by rubber tapper families, it is administrated by the Brazilian Agency for

Agrarian Reform (INTCRA). There is no evident geographic division of seringais in this

area. Rubber tapper landholdings are grouped into three associations. The Porto Dias

Rubber Tapper Association area (Figure 1-2, B) was selected for sampling. This area is

one of the most important initiatives for timber management in Acre and has been

supported by a pool of organizations led by the Center for Amazonian Workers (CTA) to

consolidate this activity as an economic alternative for rubber tappers.

The third area is the Peixoto Colonization Settlement, the largest and oldest human

settlement area established by INTCRA at the end of the 1970s (Figure 1-2, C). As with

other colonist areas in the Amazonia, agriculture and cattle are the main land use

strategies in Peixoto. However, innovative experiments emerged at the end of the 1990s

to address environmental degradation and economic inefficiency. Since 1994, the Group

for Agroforestry System Research (PESACRE), a local NGO, has collaborated with

members of the Grupo Novo Ideal rural association, a community organization located in


















d re:r*~
:i:
r d i-'
* ~' a
~-r~I
~p~ii


Peixoto, to combine agriculture, livestock, agroforestry, and forest management into an

integrated agrosilvopastoral system. This community is the colonist sample for this

thesis.


Deforestation 2004

Water

Selected area

Land tenure limited

Chico Mendes Resex
Porto Dias PAE
Peixoto PC


(A)
(B)
(C)


Figure 1-2. Deforestation in the study sites, Acre, Brazil.









Theoretical Framework

The theoretical foundation of this work is guided by three bodies of literature. I

draw on political ecology as a framework to integrate land reform in the Amazonia region

and community land uses strategies as a dynamic process taking place at multiple scales.

This study seeks to understand local land use practices in the Amazon region as a result

of events established in larger, extra-local political and economic settings (Peet and Watts

1996, Bryant and Bailey 1997). Linking political, economic, and cultural perspectives,

this framework attempts to address a range of factors affecting use of natural resources as

related to local, national, and global political economies and ecosystems (Blaikie and

Brookfield 1987, Schmink and Wood 1987). Economic booms in Amazonia were a key

aspect to take into consideration in this perspective. As a result of international demand,

national and regional policies encouraged human migration, landownership regimes, and

environmental depletion. These processes are examples of how underlying factors and

political economies at multiple levels may affect socioeconomic and political dynamics at

the community level in Amazonian region.

Political ecology also constitutes a powerful framework for integrating natural and

social dynamics (Peterson 2000). This context is fundamental for understanding how

community land tenure regimes in Amazonia were a result of underlying social, cultural,

environmental, economic, and political factors from a local to global scale. In the case of

Acre under the Forest Government, political processes yielded land tenure categories

specially designed to produce more favorable economic as well as environmental

outcomes via more sustainable land use practices. Consequently, political ecology

provides a basis for understanding the historical context and political foundations for the

land tenure categories on which I focus in this thesis.









However, to empirically assess economic and conservation effectiveness of land

use strategies in different communities, I decided to add two other bodies of literature:

household economic analysis of land use management via a livelihoods approach; and

explanations at various scales for land cover change.

Economic growth based on sustainable land use has commonly been held up as a

blueprint for both conservation and human wellbeing. However, this research accepts

Wollenberg et al. (2001)' s view that those income-generating activities may alter the

social context, conservation behavior, and livelihood of local communities. This study

attempts to comprehend rural smallholders as a complex of diverse livelihood activities

(Ellis 2000). To explore this perspective for the land tenure categories selected in this

work, two levels of analysis were conducted. First, the sustainable land uses practiced in

the selected communities, have been under debate for over 30 years. Researchers of land

use and forest economics in the 1980s and 1990s (Godoy et al. 1993, Peters et al. 1989,

Pinedo-Vasquez et al. 1990, Nepstad and Schwartzman 1992, Padoch and Jong 1989)

identified potential challenges and opportunities for these strategies. Second, in addition

to their institutional, socio-cultural, and economic diversities, inhabitants of extractive

reserves, agro-extractive settlements, and colonist settlements are typified by the diversity

of their livelihood systems. To achieve an economic and land use analysis, understanding

the livelihood of each land tenure category emerged as a key component to evaluate the

effectiveness of community-based development in Amazonia.

Although household farming, as argued by Ellis (2000), attempts to diversify the

range of family activities as a strategy for maintaining their wellbeing, in the Amazon

this decision may be also influenced by cultural behavior (Rego 1999). Household









diversity, in this context, is problematic for empirical economic and social analysis.

Hildebrand and Schmink (2004) argue that economic, social, and institutional factors in

each livelihood system affect household economics, which have multiples consequences

for land use allocation.

This study uses livelihood analysis to understand how land cover and economic

dynamics are affected by different land use regimes. Based on differences in livelihood

strategies among land tenure types, the instruments utilized during the fieldwork

(questionnaire, participatory mapping, participant observation etc) were adapted to each

community. The economic findings discussed in this thesis were summarized in terms of

inputs and outputs; however, for visualizing the household diversity in their livelihood

systems, those findings were aggregated by categories of land use (agriculture, livestock,

forest management) that represent spatial land allocation (agriculture area, pasture, and

forest), and are the result of allocation of available resources (labor, resources, and

capital).

The third body of literature on which I draw is that addressing the causes and

consequences of land use and land cover changes at multiple scales. Theories focusing on

Amazonian deforestation have been dominated by neoclassical economics and political

ecology analyses; other studies have pointed to a combination of proximate causes and

underlying drivers in a geographical and historical context (Geist and Lambin 2002).

Yet, at the local level, individual and social responses to land cover dynamics are

affected by economic change led by institutional factors, in which new land use strategies

are created by market and policies increasingly affected by regional and global factors

(Lambin et al. 2001). Land cover and land use change, while affected by global and










regional drivers, is particularly influenced by individual micro-economic decision taking

place at community level (Anselin 2001). Mertens et al. (2000) argued that integration of

household survey and remote sensing information are an appropriate strategy to

empirically understand the linkages and feedbacks occurring at micro and regional levels.

Methodology Summary

This research involved multidisciplinary inquiry at the three sites and multiple data

collection methodologies including ground truthing for classification of land cover

classes, and quantitative analyses linking economic outcomes and forest cover dynamics

at household level. Principal research methods included semi-structured interviews,

participant observation, vegetation sampling, and remote sensing analysis.

The economic survey at household level was carried out in 2004. I conducted

interviews with 88 households in three land tenures located in Acre's southeast region. In

order to evaluate the economic and land cover effects of sustainable land use strategies

implemented by families, such as forest management and agroforestry systems, the

sample was distributed as follows. In the Chico Mendes Resex I selected 34 families

involved in management of various non-timber forest products. In the Porto Dias agro-

extractive settlement the sample was 27 households. Peixoto colonization settlement was

the third area, and had 27 small farmers selected. The set of interviews with those

families was conducted from May through August of 2004 in collaboration with CTA,

PESACRE, and governmental agencies.

The land cover sampling took place during two time periods. In 2004 when I

carried out the household surveys, I also collected 345 land cover GPS points in the

selected household areas with four research assistants. Inaccuracy and confusion of the

classification results in bamboo areas required additional ground-truthing in the summer









of 2005 during which 97 additional land cover samples were gathered. These data were

then utilized to perform a supervised classification using Landsat images scenes 67-001

and 67-002 of 2004.

Study Significance: Local Research Needs on Land Tenure Analyses

This research is important for four main reasons. First, it provides a comparison of

economic and land cover outcomes across multiple land tenure categories. Second, it will

be crucial for implementation of policies in Amazonia, specifically in Acre where

community land regimes are a foundation for alternative development based on

sustainable land uses. Third, although land use and land cover change literature in the

1980-1990s focused on Amazonia, there is a lack of comparisons assessing the economic

and conservation effectiveness of community initiatives based on timber and non-timber

forest management, and agrosilvopastoral systems. Fourth, this study will contribute to

the land use and land cover change literature by linking socioeconomic and satellite data

analysis at household level.

The first implication may be the most important for local stakeholders in Acre.

Although the state government has been guided by efforts to promote forest based

development since 1999, it has altered its policy approach over time. Policymakers have

concentrated their efforts on establishing economic alternatives for community areas of

the Rio Acre watershed, a region with social, cultural, and ecological specificities, in

which this study was conducted. In this territory, the governmental initiatives have

addressed a combination of timber and non-timber forest management, agroforestry

systems, as well as agriculture and cattle ranching zones as a strategy to increase the local

welfare while emphasizing environmental conservation. Over the next years, these









community models will likely be replicated in other areas of the state (i.e. the Jurua and

Purus regions) which accounted for over 65% of the intact forests in the State in 1999.

However, there is a lack of empirical studies addressing community land use.

During various moments in Acre, policymakers and social movements have called for

research programs assessing these emergent issues. Cross comparison among community

land use strategies should show whether the forest-based initiatives addressed are

effectively balancing economic growth and conservation goals. Specifically this type of

research will demonstrate the land cover and economic effects of state policy planning

within extractive reserves, agro-extractive settlements, and colonist settlements.

Structure of the Thesis

This thesis is organized in five chapters. Chapter 2 gives an overview of the

history, context, and status of community-based development in Amazonia and

specifically in the state of Acre. It introduces how economic booms, human migration,

economic processes, and policy drivers affected land ownership and resource use at a

regional scale. The chapter attempts to demonstrate that while community land tenure

regimes were a result of regional and national drivers, they were also produced by

political and economic events conducted at global scale. Based on a political ecology

perspective, this chapter argues that natural resource appropriation in Amazonia,

including land, has been determined by social confrontations and alliances at multiple

levels. While the 1970s included the most crucial rupture in the Amazonian economic

system, during which a jump from forest-based development to an agricultural economy

took place, this period was also notable for the establishment of innovative land reform

led by rubber tapper movements. In the 1980s and 1990s new land tenure categories

emerged as a result of social conflicts, economic changes, and policy implementation.










They were also characterized by their institutional and social differences, as well as the

remarkable cultural distinction of their inhabitants.

Chapter 2 also focuses on the study area and discusses why the State of Acre may

be considered a space of rare opportunity to assess the effectiveness of community land

tenure strategies based on sustainable land use initiatives. The forest government efforts

to build up an innovative regional model of public policy are also discussed in this

chapter. The maj or obj ective of this chapter is to clarify the multiple land use strategies

conducted by rural communities in this region and how their land rights were established

over time. The historical, cultural, and socioeconomic difference among these

populations is addressed via a comparative analysis. A brief background of the study

areas, which include the Chico Mendes Extractive Reserve, the Porto Dias Agro-

extractive Reserve, and the Peixoto Colonization Settlement, is provided.

Chapter 3 concentrates on the research design, field surveys, and remote sensing

techniques as well as the economic framework utilized to perform the data analysis. This

discussion addresses three key aspects. First, it shows how remote sensing and household

surveys are combined for economic analysis of resource management, particularly land

cover change. Second, it describes the research design used in this study, in terms of

economic and land cover comparisons between the three land tenure categories. Last but

not least, it emphasizes the research innovation achieved in this study by using satellite

and socioeconomic data at a household level to compare multiple land tenure types in

terms of their land uses.

Additionally, Chapter 3 discusses how the economic and land cover fieldwork was

carried out. This chapter a) summarizes how the communities and households were









selected; b) discusses the spatial and socioeconomic techniques undertaken through the

survey; and c) reviews the methodological approach for the economic and land cover data

analyses. This is followed by a discussion of the participatory tools used, prior to

economic analysis, to define indicators, field techniques, and sampling.

Chapters 4 and 5 detail the results of this study and compare the three community-

based strategies. The economic outcomes at a household level are linked to their

associated costs, and I evaluate their environmental performance for addressing forest-

conservation. The analysis is divided into three parts. The first discusses the land cover

distribution across the three land tenure categories. The second focuses on economic

analysis of costs and revenue to identify the most profitable land tenure and livelihood

activity. Finally, the third part presents five multivariate regression models to evaluate the

effects of economic and non-economic variables on deforestation among smallholders of

the study sites.















CHAPTER 2
LAND USE MANAGEMENT AND LAND TENURE DIVERSITIES: ECONOMIC
BOOMS, STATE POLICIES, AND LAND RIGHTS IN ACRE

Introduction

The purpose of this chapter is to summarize different land-tenure types in the State

of Acre and differentiate the land use strategies associated with each type. This

discussion starts by examining the larger context in which land tenure regimes developed

in the Amazon region, in particular, how the economic booms affected human migratory

movement, livelihoods of rubber tapper populations, and by extension, land rights.

Additionally, it introduces a few details of the policy-land tenure regime interface that

took place during three maj or landmark periods in Acre's history: the rubber booms

(1870-1950), the era of cattle expansion (1970-1980s), and the post-victory of the rubber

tappers' movement (1990s to today). This discussion introduces the main categories of

land tenure in Acre and points out central differences in their attempts to address

economic and conservation goals. An overview of the study sites and description of their

significance will also be provided.

In contrast with many other studies in which an economic-environmental tradeoff

analysis is undertaken, this chapter emphasizes why public policies are a key driver for

land cover and land use change in Amazonia via a comparison of distinct land tenure

types. This chapter concentrates on forest policy approaches implemented in Acre since

1999 in which rubber tappers and rural communities have been a key social actor. In

addition, this chapter introduces the overall discussion of the circumstances in which









different land tenure categories and resource management strategies were established.

Pointing out diverse contexts on how land tenure types emerged in Acre and their

remarkable dissimilarities, I argue that extractive reserves, agro-extractive settlement, and

colonist settlement proj ects played an important role in achieving an alternative

development model.

To begin the comparative analysis of different land tenure regimes in this state, this

chapter offers an overview of the historical and regional context in which land tenure

regimes are situated. The discussion is organized as follows. First, a summary of the

economic boom periods that took place in Amaz8nia will be provided. Second, the

historical and geographical context in which community tenures emerged in the state of

Acre will be identified. This discussion focuses on two issues: 1) the struggle for land

reform from the late 1970s to early 1990s; and 2) the government policies implemented

in Acre since 1999 related to land tenure communities. Third, differences among

extractive reserves, agro-extractive settlements, and colonist settlements will be

discussed. This section will also include a description of the three study sites, each

representing one of these land tenure types.

Land Tenure Regimes in the Amazon Region

Land rights and land tenure in the Amazon have historically been influenced by

complex connections among agents at global, national, and local scales. Simmons (2004)

emphasizes that while these dynamics have been defined by multiple agents, such as

indigenous people, miners, loggers, ranchers, and small farmers, a political-economic

framework for land-tenure in the Amazon requires an understanding of past processes. As

a part of this discussion, this section offers an historical review of development policies

of Amazonia and their implications for land distribution and land tenure implementation.










Specifically it focuses on the cycles of economic booms that took place in this region

from the 19th century to the late 1980s. Three historical moments illustrate the interface

between economic events and land tenure regimes.

The first phase, from 1870s though the 1950s, was the rubber boomer period which

took place in Amazonia in the middle 19th century and again in the 1940s and early

1950s. Demand from pneumatic tire factories in Europe and the U.S. motivated the

Brazilian government to move thousands of rural workers from the northeast region to

the north of Brazil to meet labor demand. The arrival of northeastern people in Acre was

an important starting point for establishment of the local society. Rubber tappers emerged

in this period as a social group forced to depend on forest resources for their subsistence.

The rubber booms, however, also fostered crucial socioeconomic changes in this

region. Thousands of indigenous people died or were expelled from their land by large

rubber estate owners (seringalista~s), the new landowners in the Amazon region (Melatti

1993). While during the first rubber boom (1870-1910) the economy was based on the

semi-slavery aviamnento system (Weinstein 1983), during the second boom (1940-1950)

the state assumed economic and social control of the rubber-based economy (Martinello

1988).

During a brief rubber boom in the Second World War when Malaysian areas were

controlled by the Japanese military, blocking America' s access to rubber, the category of

autonomous rubber tappers2 became more evident (Oliveira 1985, Bakx 1988). In spite its


i The semi-slavery aviamento system was a coalition of seringalistas, Brazilian state, and national and
international companies to implement the rubber economy in the 19th century in the Amazon region.
2 Autonomous rubber tappers refer to circumstance of relative freedom of rubber tappers in managing their
colocapaes. This situation was launched in the late 1910s, when external interference over the resources in
the colocagaes from state or seringalista became reduced. As a result, agriculture and livestock emerged as
new activities worked by rubber tappers for their subsistence.










shortness, the four years (1942-1945) of rubber demand stimulated some economic

recovery. A new migratory movement was then stimulated by the Brazilian government

to Amazonia, particularly to its Western region, the area with the largest population of

rubber trees.

The second phase, from the 1960s through 1970s came with the decline of

international demand for Amazonian rubber. The forest economy, which had been

represented by the rubber tree, was substituted by a new development model based on

cattle and crops. This economic change was followed by large-scale state investments in

infrastructure such as roads, dams, and mines, and subsidy programs to modernize the

Amazonian economy (Cardoso and Miidler 1977). On the other hand, at a micro-scale, the

land tenure regimes were also transformed from seringais to large cattle ranches. At the

same time, rubber tappers engaged in defending their landholdings (colocapi~es), even as

some rubber tappers migrated to urban areas.

The fall in the price of rubber resulted in the bankruptcy of many seringalista~s and

contributed to important changes in the rubber tapper' s livelihood system. As a result of

more flexible social control, hundreds of rubber tappers became more independent and

were able to introduce new economic activities into their land use strategies such as slash

and burn agriculture, livestock, hunting, and harvesting of forest products. It is notable,

then, that the rubber tappers' relative freedom fundamentally altered their level of




3 This strategy was addressed by Brazilian government to face the lack of a labor force to exploit rubber
trees. State advertising was carried out to stimulate a new migratory movement of people from the
Northwest, this time called rubber soldiers (soldados da borracha), to the Amazon. Oliveira (1985) noted
that over 50,000 migrants from that poor Brazilian region were engaged on the rubber economy in Acre's
jungle. Yet, as pointed out by Sobrinho (1992), different from previous cycles, the rubber economic system
was controlled by the state which monopolized the commercialization process and the control of rubber
tappers in the seringais.









subsistence, even as they continued to focus on rubber as their main cash income activity

(Oliveira 1985).

These changes prompted rubber tappers to mobilize to defend their land rights

Consciousness-rai sing, facilitated by worker unions, was a key element influencing the

land struggle that took place through an economic transition from forest-based land use to

cattle ranching launched in the 1960s. In this context, the struggle for land of during this

period was characterized by social coalitions between workers and regional organizations

in order to pressure the State for land reform in the region.

Land tenure policies implemented during this period were complicated by the

governmental plan to integrate the Amazon by roads to other Brazilian territories. Human

migration was encouraged though colonization settlements which were established in

areas surrounding regional highways, such as Transamaz8nica, Cuiaba-Santarem, Belem-

Brasilia, and Rio Branco-Brasilia. As a result, land conflicts between rubber tappers,

colonists, and large cattle ranchers became more visible in Amazonia. However, intensity

of land conflicts varied among the Amazonian states as a consequence of local policies,

social coalitions, and the influence of national policies. In the following section I will

focus on the case of Acre. In particular, I will concentrate on the importance of social

movement, led by rubber tappers, to foster land reforms.

The State of Acre: History, Social Movements, and Land Tenure Diversity

Geographic Context

Acre's 153,000 km2 is occupied by over 550,000 people (Governo do Estado do

Acre 2000a) and is bordered by the Brazilian states of Rond8nia and Amazonas and by a

trans-national frontier constituted by the Bolivian state of Pando, and the Peruvian state

of Madre de Dios (Figure 1-1). Its territory is still constituted by almost 90% intact










rainforest with diverse human and ecological mosaics with complex interactions at local

and regional levels. Although this state has been cited as a rich region of palms (Daly and

Silveira 2001) and important diversity of other species of plants, habitats, and soil

(Governo do Estado do Acre 2000b), due to the enormous lack of scientific knowledge of

its ecology, the degree of its biodiversity is likely underestimated (MMA 2001a, Silveira

et. al 1997).

Besides its biological diversity, approximately 44% of its territory is occupied by

smallholders such as extractivist inhabitants, indigenous people, and colonists (Governo

do Estado do Acre 2000a, Figure 1-2) who have implemented a range of land use

strategies such as timber and non-timber forest management, annual and perennial crops,

and cattle and other livestock. In 1996, less than 30% of Acre's total deforestation was

located within community areas.4 Therefore, Acre's future conservation will be affected

by land use decisions of these communities.

Community land tenure regimes in Acre were the result of intensive social conflicts

that took place in the past. Social movements and rubber tapper strategies, such as the

"empate movements" were fundamental to influence land reforms in this State. These

rural workers assisted by various organizations from local to international level

contributed to making this region renowned worldwide in the 1980s. In order to provide


SThis assessment was performed using data available through the MacapB Seminar, a meeting organized by
the Brazilian government in 1999, in which Brazilian hotspots for conservation and sustainable uses were
identified. I utilized ArcView to process raster and rector files and Erdas Imagine to merge raster and
rector files. I then calculated deforestation within each land tenure type (see MMA 2001 for details about
the data).
5 The enyate movements were a form of collective resistance conducted by rubber tappers against
deforestation. The name was conceived based on the Portuguese word "enyatar" (to prevent). These
actions were usually carried out in the summer season, the moment in which large cattle ranchers cut the
forest. During these initiatives rubber tapper leaders firstly tried to convince farm laborers to desist cutting
down the forest. If they were not successful to stop the deforestation, then they organized themselves in line
(children, women, and men) to protect the forest.









an overall context in which those community land tenures were created, in the following

section I move to Acre's history. Specifically I will focus on how social movements

emerged and demanded land reform and the establishment of new categories of land

tenure types such as extractive reserves (Resex) and agro-extractive settlements (PAE).

This discussion will be divided into three parts. First, I summarize the period prior to

cattle raising in the 1970s. Second, I discuss the period of large-scale cattle ranching and

land conflicts from the 1970s to the 1980s. Third, I focus on the rubber tapper victory in

the 1990s when extractive reserves and agro-extractives settlements were created.

Acre: Before the Ranchers

The end of the rubber boom in 1945 contributed to the collapse of the Acrean

economy, and also resulted in the emergence of new and important social actors. Oliveira

(1985) and ELI (1994) point out that the crisis of rubber had two related consequences

for land property regimes in Acre. On the one hand, some of the seringalista~s went

bankrupt and abandoned their rubber estates. On the other hand, rubber tappers reacted

differently to this collapse, though in diverse ways.

One group of rubber tappers, like some seringalista~s, grew discouraged with rubber

tapping and migrated to urban areas in Acre or to other Brazilian states. Another large

number of rubber tappers remained under the seringalista~s' oppressive control that used

jagngos (mercenaries) to prohibit agriculture and other subsistence activities in the

colocapies (landholdings). Within these areas, then, no change was observed in terms of

social control, landownership, and rubber tappers' livelihood. And third, a more

significant transformation, which has also been noted by Bakx (1988), took place among

rubber tappers, who, as a consequence of the abandonment of rubber estates by









seringalista~s, were able to manage their land and their resources according to their

particular interests.

These socioeconomic changes took place among the third group of rubber tappers

contributed to the establishment of new land use strategies within seringais while also

providing the social context for the emergence of a new category of rubber tapper.

Conscious of their land rights, for the first time since their arrival in this territory, they

organized themselves against ranchers, the state, and the judicial system to demand rights

to their lands. This period was an important landmark in Acre's history in which an

economic rupture emerged while also bringing intensive land conflicts.

Agriculture Modernization: Large Cattle Ranchers, Rubber Tappers, and Colonists

The period from the 1970s through the 1990s was the most crucial phase of Acre' s

history in terms of social conflicts, land tenure regimes, policies, and economic strategies.

These striking social and political changes and their implications for development and

land use have been noted by various scholars (e.g., Oliveira 1985, Esteves 1999, and

Paula 2003) as the moment of substitution of forestry harvesting by agriculture and

pasture economies. This time period was marked by the polarization between forest

resource use by rubber tapper communities and traditional models based on crops and

cattle expansion.

The crops and cattle economic model implemented in Acre in the early 1970s was

led by the Wanderley Dantas. At that moment the governmental slogan was to "produce

in Acre, invest in Acre, and export via the Pacific" (Silva 1990) which meant to integrate

this state into the emergent Brazilian agribusiness economy. Based on cattle, a ranching

coalition constituted by local and national politicians, the board of judges, state police,










and large ranchers emerged to establish a new land tenure policy to benefit large

landowners.6

In terms of the ownership configuration of Acre' s territory, this period represented

a transition from seringais to large scale cattle farms. In addition, this phase was

characterized by social polarization between traditional populations (rubber tappers and

indigenous people) and large cattle ranchers. The str-uggle became more intense when the

"paulista~s", which included ranchers from the South, Central and Southwest of Brazil,

arrived in Acrean territory and encountered a large resident population of posseiros, or

autonomous seringueiros, that acquired their rights over the rubber estates when they

were abandoned by seringalistas. The "empate movement" was the main strategy utilized

by rubber tappers to prevent deforestation as well as to demand their land rights.

After the 1990s: No Economic Boom, but a Rubber Tapper Victory

The transition from forest-based production to a cattle ranching and land

speculation economy was followed by a violent str-uggle within Acre. Sobrinho (1992)

points out that from 1975 through 1988 in Brasileia and Xapuri7 400 rubber tappers were

imprisoned and 40 others killed or tortured. Both rubber tappers and ranchers organized

themselves as social movements. In mid-1985, large cattle ranchers created the 'Rural

Democratic Union (UDR)' with two main goals: 1) to influence national policies in the

interests of agri-business, for instance, by strengthening subsidies, credit, and road





6 MCSquita (1977) estimated that from 1971 through 1975 over five million hectares, that is, almost one
third of Acre's territory, were transferred to new owners. Commonly, emphasizes this author, the land was
occupied by rubber tappers who were forced to migrate to urban poor areas.
SXapuri and Brasil~ia are two municipalities located in South Acre region. Since the economic transition to
cattle ranching in the 1970s they became one the most important areas of both land conflicts and rubber-
tappers movement against roads and deforestation (see FASE 1989 and Allegretti 1989).









building; 2) to lead legal and many times violent efforts against smallholders, untitled

occupants, and indigenous people in Amazonia.

As a response, at the end of 1985, rubber-tappers organized their first national

meeting in Brasilia, the capital of Brazil. This event was supported by the National

Institute for Socioeconomic Studies (INTESC), the Brazilian Department of Culture, and

Oxfam, an international organization. At that moment, they created the National Rubber

Tappers' Council (CNS). With national and international supporters they also advocated

the creation of Extractive Reserves (Resex) as a pioneering land-reform measure within

Amazonia intended to maintain its natural resources and also providing rubber tappers

with long-term rights to land. Further strengthening this movement, by late 1988,

indigenous people, rubber tappers, and ribeirinhos j oined their voices in one of the

strongest collective actions in the Amazon. In Acre they built a coalition under the name

of the 'Forest Peoples Alliance'.

The large ranchers reaction, however, was extremely violent with the shooting of

Chico Mendes, one the most prominent rubber tapper leaders in Amazonia, in the late

1988. Although the 1980s period was one the most violent periods of Acre' s history, it

also created an opportunity for new leaders who argued for a new development plan

based on forest use and smallholders' land rights.

As a consequence, land tenure policies took their first steps toward land reforms. In

1989 the Brazilian government created four agro-extractive reserves in areas with

intensive land-conflicts in the State of Acre: Sho Luiz do Remanso, Santa Quiteria,

Macaui, and Cachoeira. In 1990, the Alto Jurua Extractive Reserve was established, the

first one in Amazonia, with more than 506,000 hectares. The same year, the Chico










Mendes Reserve, the largest Brazilian extractive reserve until 2004 with almost 1

million hectares, was formed.

Chico Mendes' murder led to the creation of seven other extractive reserves in five

Brazilian states by 1992. Chico Mendes' legacy was revitalized in 1998, when the

Workers Party won Acre's gubernatorial election, establishing a bridge among policy,

social movements, and agents managing diverse land tenures. In the following section I

discuss the innovative public policies implemented in the State of Acre since 1999.

Specifically I focus on the effects of governmental initiatives for community land

tenures.

Acre: Social Movements, Community Land Tenures, and Forest-Based
Development Policies

In this section I will summarize recent public policies implemented in Acre, which

have attempted to establish a forest-based development model, and their implication for

land tenure regimes. Particularly, I will concentrate on the efforts of the Acre government

to foster alternative policies in community land tenures areas since 1999.

The triumph of the forest-based development led by Jorge Viana, a forester who

worked as a technical adviser for rural communities in the 1980s, led to the establishment

of a development model based on sustainable use of natural resources. The "forest

government" slogan became a key concept which suggests the Acre territory could be

developed based on land tenure rights and land uses strategies of extractivists, colonists,

and indigenous people.

In addition to the fact that many of these policymakers worked closely with local

communities over the past 10 years, two factors help explain why the forest government

SThe largest Brazilian extractive reserve is the Verde para Sempre with 1.3 million hectare. It was created
in 2004 and is located in the State of Para.









concentrated much of its attention on the spatial distribution of land tenure regimes. First,

although the legal status of a large portion of land was not precisely known, the type of

landowners and their location and land use practices were recognized in Acre's territory.

Second, different from the previous governor, Viana attempted to implement a policy

taking into consideration the economic vocation of the population of each land tenure

category. Thus, economic and ecological zoning (ZEE)9 emerged as a strategic

instrument for state planning, territorial management, and policy implementation.

Based on the ZEE' s results, four maj or categories of land tenures types were

identified in the state territory (Figure 1-2). The first one, covering almost 3 5% of the

territory, is characterized by colonization settlements, agro-extractive settlements,

extractive reserves, and indigenous reserves. The second accounts for approximately 15%

of Acre' s area and is comprised of national and state parks, and other conservation areas.

Third, areas under INTCRA' s control, which should be utilized for human settlements,

correspond to approximately 7% of the state' s area. Lastly, the largest area of the state is

in the hands of private landowners (mainly large cattle ranchers) who control almost four

million hectares or 26% of Acre' s territory. There is, however, a large extent of land

without any immediate land tenure discrimination, covering over 22% of the territory.

Although these community land tenures have been a key to the forest government,

its policies have been altered from Viana' s first term of office (1999-2002) to the second

(2003-2006). Based on my personal experiences speaking with diverse local stakeholders,

I have identified four maj or differences between the first and the second of Viana' s terms

of office which could affect economic and land cover dynamics within land tenure


9 The ZEE is an important policy instrument for territorial management in which the state integrates
economic, social, and environmental studies as elements for subsidizing society's decisions.










regimes in Acre. First, while the first term of office emphasized non-timber forest

strategies, during the second mandate, timber management has been a priority. Second,

through 1999-2002 the government attempted to establish a forest model based on a

diversity of land use strategies addressed by forest inhabitants (rubber tappers, indigenous

and colonist). Third, the relationship between state and federal government was altered

when Luis Inacio Lula da Silva was elected president of Brazil in 2003. Since this period

considerable infra-structure investments have been carried out in Acre to integrate it to

the Peruvian region. These initiatives may have unknown effects for forest conservation

and community land tenures.

Lastly, as policymakers have implemented a model oriented by market demands

(e.g., timber and cattle) since 2003, strategies for increasing the competitiveness of forest

strategies have become a key economic priority in the second term. During this period,

macro-investments (e.g., road building, energy, and logging company incentives) were

more encouraged than community-based initiatives (e.g., cooperatives and rural

association subsidies). Overall, while the first mandate was based on the neoextrativismo

approach, the current term of office has been called florestania. In the following section I

will summarize each of these approaches.

Neoextrativismo: A Long-Term and Systemic Approach

Neoextrativismo was conceived by Rego (1999), an agricultural economist at the

Department of Economics at the Federal University of Acre, almost four years prior to

the election of the Acre forest government. This concept emerged primarily as a response

to critics of community extractivism implemented in Amazonia. Homma (1993), who

was one of the most ardent critics, argued that extractivism was not feasible as a regional

development model. After the Workers Party (PT) victory, Rego became the Secretary of










Production and one of the most important policymakers in the forest government from

1999 through 2002. Thus, neoextrativismo, during this period, was consolidated as an

element for integrating economic, social, and environmental sustainability in land tenure

policies in Acre.

Neoextrativismo incorporated five inseparable aspects into community land tenure

development. First, neoextrativismo took into consideration the cultural behavior of local

residents. Second, it was conceived to consider community livelihood systems and their

implications for land use and land cover change. Third, it addressed community as a

means of balancing economic and ecological goals for policy. Fourth, it argued that any

technology for sustainable land use should be adapted to local inhabitants. Finally, the

neoextrativist approach emphasizes that policies addressing sustainability will only be

successful when smallholders are able to diversify their systems, for instance, combining

agroforestry, forest management, agriculture, and livestock.

Despite its original and systemic overview of the socio-ecological diversities and

its suggestions for establishing innovative public policies in the Amazonia, few studies

have assessed this approach. Kainer et. al (2003) point out that for the first time in the

Amazonia, an integrated policy framework regarding the local ecological, cultural,

economic, and social diversities was articulated via neoextrativismo. Maciel (2003), an

economist at the Federal University of Acre, assessing the Islands of High Productivity

(IAPs)lo as a component in the neoextrativist strategy, estimated that it should increase


"' Islands of High Produtividade (IAPs) were conceived by Kageyama (2002) as a strategy to increase
rubber production in the rubber tappers livelihood system in Amazonia. This experiment was implemented
in Acre, through which rubber tappers were stimulated to plant rubber trees within deforested areas and
bordered by neighbour forestry diversity. In some ways, IAPs were very similar to agroforestry systems.
The IAPs, while conceived as a neoextrativist model incorporated three dimensions: economic (economic
growth with rubber productivity): social (rubber tappers usually worked collectively to implant those
experiments); cultural (rubber threes have been the main activity for rubber tappers over the past 30 years);










rubber tappers gross income by 400% as a consequence of a reduction in labor

requirements. These results indicate that the neoextrativismo approach attempts to

incorporate livelihoods into a public policy agenda, while also pointing to the fact that the

transition from traditional development to forest based development could require a long-

term process. However, while alternative land uses suggested by this approach such as

the implementation of agroforestry systems with brazil-nuts require almost 30 years to

yield, the government mandate in Brazil is limited to four years. This fact contributed to a

new public policy approach after 2002 in Acre which I will discuss in the next section.

Florestania: Citizenship Based on Regional Integration, Short-Term Economic
Growth, and Conservation

In 2002 Viana again won Acre's election in a new national context. When Lula da

Silva, a close friend of Chico Mendes in the 1980s, took power as president of Brazil,

Marina Silva, who was a rubber tapper in Acre during her childhood, was selected by

Lula to take the position of Secretary of the Ministry of the Environment. Moreover, after

almost three years of negotiations, the Inter-American Development Bank (IDB) financed

the Acre Sustainable Development Program through which the state received US$ 250

million in loans for supporting its forest-based initiatives.

These new macro factors were fundamental in changing state planning in Acre.

While during its first phase, the forest government was characterized by attempting to

consolidate a development model based on community land tenures and land uses,

Viana's second term has attempted to establish Acre as a regional model for the Amazon.

Neoextrativismo, which focused on micro-scale and long-term socioeconomic outcomes,

was substituted by florestania as a new agenda for the state policies. The florestania

enviromnental (it should to contribute for forest regeneration); and political (Acre's govermnent in 1999
incorporated this initiative into the local public policies).










approach moved to publicize Acre as an economically efficient Amazonian development

model that has attempted to integrate both urban and rural populations as beneficiaries of

sustainable forest use.

Florestania has incorporated two perspectives into forest-based development

implemented in Acre over the past six years. First, this concept was founded on the

principle of universal human rights, including the citizenship rights of forest peoples.

Second, Viana's policymakers also included an environmental component within

citizenship. In this perspective, forest-based development could produce economic

growth to overcome social inequalities via environmental conservation.

The context of these efforts is manifested in the IDB Proj ect, the largest program

implemented by the government during its second mandate. The IDB proj ect' s short-term

goals are the promotion of economic growth, environmental sustainability, and local

economic diversification (IDB 2002). To implement its actions, this program was divided

into three sub-programs: a) sustainable forest management, conservation, and state

enforcement; b) employment, technology dissemination, and community-based subsidies;

and c) infra-structure investments in roads, rivers, and energy generation (BID Proj ect

2002). The IDB program inaugurates a new phase in Acre in which forest based

initiatives were incorporated into a context of regional integration.

A range of complementary endowments has been mobilized by the forest

government in order to implement its macro-planning. Besides IDB, Acre has also been

supported by the Brazilian government and the Corporacion Andina de Fomento

(Andean Corporation for Development) with approximately US$ 640 million for










implementing investments in infra-structure, public services, economic subsidies, and

other economic initiatives by 2006 (SEPLANDS 2005).

While addressing regional integration, the Acre government has also continued to

encourage land use strategies employed in community land tenure regimes. In this new

phase, however, emergent demand for forest products and the need to show the success of

this development model have forced policymakers, along with national and international

organizations, to place greater focus on economic rather than conservation goals within

communities. Although the long-term implications of this policy approach for

socioeconomic development and land cover is still not clear, some economic growth-

deforestation tradeoffs such as in cattle raising and logging have been observed.

Initiatives for regional integration implemented during the florestania period have

generated new concerns about the continued viability of community land tenures models

designed in the early 1990s to achieve economic, social, and environmental goals.

I will now move toward the third part of this chapter, which focuses on three land

tenure types present in the state of Acre: extractive reserves, agro-extractive settlements,

and colonist settlements. It will be divided into two parts. First, an overall description of

those land tenures types will be provided with an emphasis on differences in their

economic and environmental outcomes. Second, three specific study sites focused on this

thesis are described.

Community Land Tenure: An Emergent Issue for Assessment in Acre

The largest portion of the Acre government' s actions has been concentrated in the

Rio Acre watershed region. Both economic integration through the Pacific, and

community-initiatives such as within the Chico Mendes Reserve, colonist settlements,

and agro-extractive settlements are located in this area. In addition, this region holds










approximately 71% of Acre's total population and almost 56% of the total rural

population (Governo do Estado do Acre 2000a). The Rio Acre region also accounts for

83% of the total deforestation in the State (FUNTAC 1999). Although only 37% of this

region remained as forested area as of 1996, the largest percent of these forest areas

(87%)" was located within community land tenures of extractive reserves, agro-

extractive settlements, indigenous land, and colonization settlements. Almost 70% of all

colonist and rubber tapper families of the state of Acre are living in the Rio Acre

watershed region. These characteristics make the Rio Acre watershed in eastern Acre a

crucial area to assess the effectiveness of community land tenures models.

While agricultural smallholders have been pointed out as key agents for forest

clearance (Angelsen 1995, Godoy, Jacobson, and Wilkie 1998), in Acre, they are

regarded as agents for promoting conservation. On the one hand, extractive reserves and

agro-extractives reserves have been advocated as an ideal model to address human needs

through the harvesting of forest products (Allegretti 1990, Murrieta and Rueda 1995).

New land use strategies, such as agroforestry and forest management, have also been

argued as strategies for colonist settlements to reduce deforestation and bring economic

gains (Menezes 2004). On the other hand, while the long-term effectiveness of

extractivist models has been questioned (Homma 1993), recent studies have identified the

tendency for increasing deforestation within these land categories (Sassagawa 1999,

Gomes 2001, Lorena 2003).

Although community land tenures and their associated land use strategies have had

diverse evaluations, they became key components in Acre's policies. Understanding land


11 This estimation was performed through vector files processing with ArcView. The datasets were
available by ecological and economic zoning of the Secretary of Environment of the State of Acre.










ownership in this state as well as key factors associated with resource use management

activities are fundamental steps to evaluate the effectiveness of community managed

areas to bring economic growth with environmental conservation. I will therefore

summarize the main differences among three major categories of community land tenures

existent in the state of Acre.

Colonization Settlements (PC)

Estimates regarding the number of colonist farmers in the State of Acre range from

15,000 to 23,000 families (Governo do Acre 2000a, IBGE 1998) and their territories

cover between 1.4 and 1.6 millions hectares (INCRA 1999). The creation of these

settlements had three maj or goals: 1) to stimulate human occupation of Acre' s territory;

2) to establish agriculture and cattle as a hegemonic model for economic growth; and 3)

to reduce social demand through agrarian reform from small farmers of south, central and

southwestern Brazilian regions (Cavalcanti 1994). The largest number of colonists,

known as sulista~s, settled in Acre beginning in the late 1970s, and had extensive

experience working with annual and perennial crops as well as livestock. These economic

strategies contributed to their relatively good socioeconomic conditions (e.g., Cavalcanti

1994, Menezes 2004), but this has been counterbalanced by the large amount of

deforestation, landowner turnover, and soil degradation associated with colonization

areas.

Colonists were stimulated by INCRA to deforest their lands to implement

agriculture and cattle pasture areas. In fact, this was facilitated by the livelihood

background of these families, largely based on shifting cultivation systems. Although

accelerated urbanization created a local market for crops there has also been

landownership turnover among households of colonization areas (Cronkleton 1998).










As pointed out by Rego (2003), cattle are the main economic activity for colonist

farmers located in the Acre River watershed. His analyses show that in 1997, 63% of the

households in colonist areas commercialized cattle contributing over 30% of total

income. Cattle expansion has been a common characteristic among colonist families

(Menezes 2004) and may stimulate landowners to obtain new land for pasture

establishment. Land concentration and migratory movements have been related to this

process. Moreover, negative social consequences (such as migration and land conflicts)

and negative environmental outcomes (such as continuing deforestation) are anticipated.

Among community land tenure regimes, colonist farmers are key actors

contributing to deforestation in Acre. Using data available from the Brazilian Ministry of

Environment, I estimated that by 1996, approximately 30% of the total deforested area in

Acre was located within colonization settlements, while extractive reserves, state, and

national parks contributed less than 5%12. Based on the same dataset, I estimated that

37% of colonization areas were deforested by 1996, which was superior to the 20% of

deforestation permitted by Brazilian law to all rural properties in Amazonial3. Taking

into consideration that cattle and crops has been the main economies for colonists in

Acre, deforestation could still be increased in this areas.











12 See footnote 4 for details of how this amount was estimated.
13 The Federal Decree No. 2166/01 that altered the Law No. 4771/1965 (Forestry Law) establishes that
small and large farmers in Amazon region are permitted to cut down until 20% of forest within their land
plots.










Extractive Reserves (Resex)

Currently, there are two Resex in Acre covering almost 10% of the state' s

territoryl4 (Governo do Acre 2000a). Rueda (1995) argues that the demand for Resex in

Acre emerged as ecological, economic, and socio-cultural dimensions of land rights. The

ecological perspective is explained by the importance of the forest in the rubber tapper

livelihood, without which there could be no extractivism. Under that circumstance the

possession of land would not be important for rubber tappers. The economic dimension

of the Resex model implies that land concession was a mechanism for guaranteeing long-

term survival of rubber tappers and their families. Lastly, the Resex' s socio-cultural

dimension argued for the rubber tappers' right to be extractivists.

The incorporation of these dimensions into the Resex proposal was a result of an

intensive struggle led by the rubber tapper movement to consolidate a model of land

reform in the Amazon based on the livelihood system of local inhabitants. Rubber tappers

developed a forest-based livelihood system, including consumption from the gathering of

wild foods, hunting, medicinal plants, and cash income from non-timber forest products.

As a land tenure regime, Resex are characterized by being occupied by families who

combine indigenous agriculture, livestock, and forest activities in their land use

strategies.

Since their creation in the early 1990s, Resex resource management has

concentrated on non-timber forest products. Although rubber has not had an attractive

market, rubber tappers have continued to harvest this product. Over the past seven years,


14 The state government may create three additional extractive reserves which may increase the total area to
over 2,000,000 ha or approximate 15% of the territory (Governo do Acre 2000b). These new planned
Resexs would be located in the JuruB River watershed.










however, stimulated by local policies, many other non-timber forest products were added

to the range of commercial products explored by rubber tappers families, such as copaiba

oil (Copaifera ssp), murmur-u nuts (Astrocaryum murmumur), unha de gator (Uncaria

tomentosa), and agai (Enterpe oleraceal).

Agro-extractive Settlements (PAE)

Agro-extractive Settlements (PAE)lS comprise approximately 194,000 hectares and

have an estimated population of 1,050 families in Acre (Governo do Acre 2000a). This

category of land tenure was also a response of the Brazilian government to reduce land

conflicts, as well as deforestation, in Amazonia. The agro-extractive settlement (PAE) is

an intermediate category of land tenure between Resex and colonization settlements.

PAEs were initially conceived in 1987 as an alternative solution for Resex demands and

also as an immediate response from the Brazilian state for reducing land conflicts in the

Amazon region. Different from the extractive reserves, PAEs, like colonization

settlements, are regulated by INCRA. However, their population is largely constituted by

rubber tappers who have implemented a land use system very similar to Resex' families.

Indeed, as a consequence of their institutional flexibility and proximity of their

population to colonist practices, the timber management proj ects, which had strong

opposition from rubber tappers, were more easily launched in PAEs.

Even though the PAEs are administered by INCRA, their regulations and

inhabitants differ from those of colonist areas, such as Peixoto. Decree No. 268/1996

legislates that PAEs may be used for forest management through sustainable practices

(INCRA 1999). Kageyama et. al (2004) point out important differences between colonist

'5 Earlier the INCRA'S Decree 627/1987 called this form of land-tenure extractive settlement projects.
However, later in 1996, it was altered to agro-extractive settlement. Surely this change characterizes the
PAE itself as a combination of colonist and extractive land-use practices (see Kageyama et. al 2004).










and agroextrativist settlements: a) PAEs' settlers are more sensitive to resource

conservation; b) the plot sizes allocated to PAE families have been around 300 ha, while

in colonist areas this amount varied between 25 ha and 100 ha; c) the size of agricultural

and pasture areas among PAE families are less than those found among colonist farmers,

and cleared areas have often returned to secondary forest. Moreover, similar to extractive

reserves, PAEs have been defined as possessions of the federal government. Based on the

same regulation, PAEs are characterized as community land concessions that must be

managed by an association, cooperative or other variety of collective organization.

Currently in Acre there are eight PAEs, but INCRA and the State government are

planning to create 30 others over the next few years, which may raise the PAEs' land area

to approximately 1,900,000 hectares, or, 12.3% of the Acrean territory (Governo do Acre

2000b). As pointed out by Rego (2003), PAE settlers prior to 1997 relied upon a

combination of agriculture, livestock, and non-timber forest product-based land use

systems. However, after 1999, stimulated by forest-policies initiatives, timber forest

management came up as a new opportunity for rubber-tappers of these types of

settlements.

In the following section I will discuss the study sites in this thesis. I focus on how

the three areas of this research were selected as a sampling of extractive reserve, agro-

extractive reserves, and colonist areas located in the Rio Acre watershed region.

The Study Sites

Peixoto Colonization Settlement: The Agrosilvopastoral Model

The Grupo Novo Ideal (GPNI) rural association, which is located in the Peixoto

settlement, was selected as a colonization sample in the Rio Acre watershed. The

Colonization Settlement, in which this community is located, has around 318,000










hectares, and was created in 1977 by INCRA. This settlement, facilitated by the presence

of the BR-364 highway, emerged in the 1980s as the most important area in the Rio Acre

watershed region for food production (Menezes 2004). Situated around 150 kilometers

from the city of Rio Branco, Acre' s capital, GNPI is currently comprised of 27

smallholders.

As colonists, their land use strategies include a range of slash and burn agriculture

as well as livestock activities, but as a result of policy incentives since 1999, cattle- and

coffee have emerged as their main economic activities. In 1994, however, they had

launched experiments with agroforestry as an alternative to deforestation and means to

diversify their cash income. However, the agrosilvopastoral strategies of those families

have been strongly influenced by dynamics of land use that took place in the land tenure

regime in which they are located.

The land use history within this community, as in other colonist areas located in

Acre, has been characterized by a drastic transformation, from forest uses (e.g., logging,

rubber, and Brazil nuts) to agriculture and cattle production. Over the past few years,

however, soil degradation has been a constraint for colonists in maintaining their

agriculture and pasture areas. Arauj o and Silva (2000) show that soil fertility in this

settlement is not appropriated for agricultural activities. Yet, despite the area' s limited

shifting cultivation potential, Lorena (2003) observed that in Peixoto's lots, the annual

and perennial crops area as well as pasture increased from ~17% in 1990 to ~3 5% of the

total area in 1999. These expanding activities explain why the forest cover within this

settlement declined from almost 90% in 1984 to 54% in 199816. Both ecological and


16 I USed Landsat TM imageries of 1984 and 1998 to perform a forest-non forest unsupervised classification
to estimate the amount of deforestation within this settlement.









socioeconomic factors stimulate small farmers to clear more forest in order to

compensate for pasture degradation (Valentim et al. 2000), as well as to implement

alternative land uses to avoid deforestation.

To address the problem of pasture degradation, some initiatives have been launched

in Peixoto and other colonization areas. Over the past few years a coalition of

governmental and non-governmental organizations has attempted to introduce alternative

land use practices such as agroforestry systems (Menezes 2004). The GPNI community

was among the first areas in which these initiatives have been implemented. While until

the early 1990s the families in GNPI concentrated their land use strategies on slash and

burn crops and livestock activities with emphasis on cattle, in 1994 they were stimulated

to introduce agroforestry systems.

In collaboration with the Research Group for Agroforestry Systems and Extension

(PESACRE), a local NGO in Acre, three new components were incorporated into

livelihood systems of GNPI households. First, for reducing environmental degradation,

PESACRE encouraged community members to introduce agroforestry systems into their

land use practices. Second, for enhancing the competitiveness of their products as well as

decreasing the risk of health problems as a consequence of pesticide use, organic

agriculture initiatives were also established among GPNI' members. The last proposal

attempted to address the need to focus on long-term market sustainability through

leadership training and commercialization of agroforestry products.

Although the GPNI community may be characterized as a colonization proj ect, its

members have adopted innovative strategies for changing the conventional land use

practices within this category of land tenure. These families have combined traditional










agriculture, agroforestry, cattle, livestock, coffee, and fruit processing. Further, the forest

government, in its spatial economic zoning of 2003, established the area in which the

GPNI is located- as having potential for coffee, pupunha, and dairy cattle. As a result,

governmental planning may stimulate community members to shift their production

systems from agrosilvopastoral strategies to these activities. Additionally, other drivers

commonly affecting land cover change in Amazonia have also been observed, such as

timber extraction and the presence of unofficial dirt roads. In this dynamic context, it

becomes important to assess economic and land cover outcomes in a changing

colonization proj ect such as GNPI.

Porto Dias Agro-extractive Settlement: The Timber Forest Management Model

In the late 1980s, guided by Decree No. 627/87 that introduced the agro-extractive

settlement (PAE) as a new category of land tenure, INCRA initiated the discussion to

create the PAE Porto Dias. In 1989 this area was established with approximately 22,000

hectares, and in 2002, Porto Dias had a population of almost 90 families (Governo do

Estado do Acre 2000a). While PAEs are supervised by INCRA, an agency created for

implementing colonization settlement in Amazonia, Porto Dias was planned to settle

rubber tapper families. It is also characterized by having a combination of agricultural

colonists and forest extractivists. This combination has been present since its initial

moments of discussion for launching its creation in the early 1980s (Stone 2003).

Moreover, when the Brazilian government launched colonization settlements in the early

1970s and 1980s, some colonist producers established their lots within Porto Dias

(Michelloti 1993).

Porto Dias is divided into three geographic zones. Zone 1 is comprised of 27

members who are represented by the Porto Dias Rubber Tappers Association. Zone 2 is










managed by Sho Jose Association, called Mossor6, represented by 11 of the PAE's

residents. Zone 3 is under the control of the ASPOMACRE, an association that has its

headquarters located in the state of Rond8nia and has 103 members of which only 11 are

inhabitants of PAE Porto Dias.

The households of zones 2 and 3 do not consider themselves as rubber tappers and

prefer to be categorized as colonists. When interviews were conducted in these areas

these inhabitants declared that they had a preference of having their landholdings

(colocapies) outside of PAE territory, but INTCRA did not approve their proposal. The

range of economic activities addressed by these families reflects their self-categorization

as colonists. Cattle and logging combined with agriculture and small-scale livestock

production are their major economic activities. Forest products, such as rubber,

medicinal plants, and Brazil nuts have no importance for their subsistence or

commercialization.

In this study I have selected zone 1 to assess community timber forest management

in the Porto Dias PAE. This is the area where rubber tappers have implemented initiatives

combining forest management, as well as other land use for agriculture, cattle, and other

livestock. The Porto Dias Association was created in 1988 and its number of members

has varied over time. For the purpose of this study, all of the 27 households in Porto Dias

PAE Zone I were selected.

Timber management was implemented by the Porto Dias Association in 1994 with

the support of the Centro dos Trabalhadores da Amazdnia (CTA) (Center for Amazonian

Workers). The first timber harvesting, however, took place two years later in 1996.

Besides timber products, through support from the Federal Ministry of the Environment










(MMA), the forest management plan included also rubber, Brazil nuts and medicinal oils

such as copaiba (Copaifera spp). Since that time period, however, association members

have predominantly oriented their efforts toward timber production (Rodrigues 2004).

In 2003 the PDA became the second community-managed project to achieve Forest

Stewardship Council (FSC) certification in Amazonia. This process is considered a

difficult task for any organization, given the bureaucratic demands and costs which are

particularly onerous for communities facing challenges in organization and physical

infrastructure. Technical support from outside organizations, such as CTA, was

fundamental for PDA Porto Dias to be successful in gaining certification status.

Although FSC certification was conceded to the association, the number of rubber

tappers engaged in timber management has varied from 7 to 10 families. Political

disagreements, lack of financial transparency, and limited financial and technical support

for community leaders, as well as absence of economic returns, have been the main

impediments to the timber management initiative.

Chico Mendes Extractive Reserve: The Non-timber Forest Management Model

The Chico Mendes Extractive Reserve is one of the premier communities in which

forest diversification has been pursued over the past years in eastern Acre. It is bordered

by large cattle ranchers that were the main group antagonistic to its creation. With almost

1 million hectares and an estimated current population of 3,000 families, the Chico

Mendes Reserve was the second extractive reserve created in the state of Acre in 1990,

established by Federal Decree 99, 144/90. Its inhabitants have concentrated their efforts

on non-timber forest products, small plots of agriculture, and livestock as their activities

for subsistence and cash income generation. Until 2004, it was the largest Brazilian

extractive reserve, stretching across parts of five Acre municipalities: Rio Branco,










Xapuri, Brasileia, Sena Madureira, and Assis Brasil. In the late 1970s and 1980s these

municipalities were where the land conflicts became most intense in Acre.

The first landmark for the creation of Chico Mendes occurred in 1980 when Wilson

Pinheiro, president of the rural workers union of Brasileia, was killed by large-scale cattle

ranchers. Almost ten years later in 1988, the assassination of Chico Mendes, the rubber

tappers' most high-profile leader, was the final step for the creation of this reserve.

Despite its importance as a land tenure category that has attempted to balance

economic, cultural, and environmental goals, some studies have identified the tendency

for accelerating forest loss in the Chico Mendes Reserve. Wallace (2004) noted that

rubber tappers have invested in livestock, including cattle, as a strategy for increasing

their total on-farm income. Sassagawa (1999) estimated that the percentage of deforested

area in this Resex increased from 0.72% in 1986 to 2.9% in 1998. However, it has been

an important barrier for forest clearing in this region. In 2001, monitoring using Satellite

GOES indicated that fire occurrence has been most intensive outside of this reserve

(Souza 2001).

Indeed, although the long-term viability of the Resex as a land tenure strategy has

been a concern among scholars, various initiatives conducted in collaboration with local

inhabitants have indicated its long-term economic and environmental sustainability (see

Maciel 2003 and Kageyama 2002). Over the past ten years, a coalition of governmental

and non-govemnmental organizations and universities has collaborated with Resex

communities to amplify their range of non-timber forest products as an alternative to

deforestation and intensive land use. For instance, since 1999 the forest government has

supported market-based initiatives though credit, subsidies, community training, and the










encouragement of 15 non-timber forest products (Kainer et al, 2003). In addition, the

Federal University of Acre (UFAC), supported by Brazilian and international

organizations, has been a key institutions collaborating with rubber tappers and their

organizations.

A few months after Jorge Viana won the 1998 election, a multi-institutional

partnership was established for implementing and monitoring the forest management

within three rubber tapper estates (seringal): Sho Pedro, Floresta, and Palmari. A group

of 34 rubber tapper landholdings (colocapi~es) were selected from those three seringais in

which the non-timber forest program has been implemented since 1999. These families

were selected in my study as cases of a non-timber forest model taking place within

Resexe in Acre.

The three seringais selected cover an area of almost 37,000 hectares and they are

located in the municipality of Xapuri. Some studies such as Esteves (1999) emphasize

that these three areas are among the places in the Chico Mendes Resex where, over the

past 10 years, intensive investments in Brazil-nut production has taken place, as well as

community organization activities. Esteves shows that economic and social benefits, such

as dirt roads, credit, and subsidies as well as education and health programs were

distributed based on kinship ties. These facts suggest that the socioeconomic benefits

were not distributed equally among rubber tappers of the Chico Mendes Resex. However,

since the forest government took power in 1999, the strategy employed by families of

these three seringais has been a reference point for establishing a common development

plan throughout the Chico Mendes Resex. These characteristics offer us a unique










opportunity to assess the community forest management model in terms of its economic

performance and capacity for forest conservation.

Land use strategies in the Chico Mendes Resex have been implemented in

accordance with its Utilization Plans (Planzos de Utilizagdo), which are defined by

common agreement between rubber tappers through their associations and IBAMA and

CNPT. Although many of those economic initiatives have been focused on non-timber

forest products, timber extraction has been discussed as an economic opportunity. Even

though timber management is not permitted within Resex areas, the Acre government

supported by Marina Silva, the current Brazilian Minister of the Environment, has

attempted to change the Resex law (Federal Decree No. 98,897) to incorporate timber

extraction for commercialization into the rubber tapper economy. Yet, the question of

timber versus non-timber models of forest management remains an important topic of

discussion among communities, their organizations, and policymakers.

Conclusion

While economic booms were historically a key driver for stimulating human

settlement in the Amazon region, policy instruments have more recently been utilized

there to facilitate economic development. In addition, local, national, and international

forces were arranged based on social coalitions among actors at various levels. For

instance, the social movement led by rubber tappers in the 1970s and 1980s with support

of organizations was successful to influence changes in the land tenure policies for

creating extractive reserves and agro-extractive settlements.

Although the policies of the forest government have been one of the most important

factors influencing land tenure after the rubber tapper movement of the late 1980s,

remarkable differences are visible in development policy since the forest government










entered in 1999. While neoextrativismo emphasized a policy based on sociocultural use

of natural resources, florestania has called for polices based on products with

comparative economic advantage at the regional level. The effectiveness of these policies

in terms of land cover and economic results will be evaluated through empirical data for

2004.

This study concentrates on three community land tenure systems for assessing the

economic and forest cover effects of these polices. The Chico Mendes Extractive

Reserve, Porto Dias Agro-extractive Settlement, and the Peixoto Colonization settlement

were selected as areas for achieving this analysis. These land tenure systems are a

representation of community land tenures located in the Rio Acre Watershed Region,

where both neoextrativista and florestania initiatives have been implemented.

Remarkable differences, such as socio-cultural, institutional, and livelihood system, are

evident among these land tenure types.

In the following chapter, I address methodological issues arising from the distinct

requirements for conducting economic and environmental research in each of these three

land tenure categories I divide the discussion into three maj or sections. The first will

concentrate on how the fieldwork was conducted. Secondly, I will explain how the land

cover and economic data were processed, as well as the key dependent and independent

variables in this study. And finally, I will discuss how economic and land cover were

linked in order to conduct the analysis.















CHAPTER 3
METHODS AND TECHNIQUES FOR ECONOMIC AND REMOTE SENSING
ANALYSES

Introduction

This chapter explains how I used economic and remote sensing approaches to

evaluate three land tenure types located in the state of Acre, Brazil. To identify the land

use and economic dynamics in the study areas, I combined survey fieldwork with remote

sensing data from an extractive reserve, an agro-extractive settlement, and a colonist

settlement. I compared land tenure categories in order to understand the linkages between

land use strategies and land cover in each community area selected.

Various studies have concentrated efforts on linking socioeconomic and land cover

factors at multiple scales. While some analyses have focused on the macro-scale (e.g.,

Cattaneo 2001, Carvalho et al. 2002), other studies have evaluated the dynamics and

drivers at the micro-scale, particularly at the household level. For example, Vosti et al.

(2003) elaborated a farm-level bioeconomic linear programming model to simulate

consequences of product and technology choices for deforestation. Walker (2003),

seeking an economic explanation of forest cover change at the household level in the

Amazon region, provided a smallholder model for colonist settlements based on

Chayanov household theory in which demographic and market variable are linked. Perz

(2001) also focused on economic and land use analysis at household level in colonist

areas of Amazon region using multivariate regression.










While these and other studies evaluate deforestation at the family level in the

Brazilian Amazon, I have identified virtually no work that compares multiple land tenure

types. A notable exception is Mertens et al. (2002), who focus on cattle economy and

deforestation in the state of Para, Brazil, and published a remarkable study in which

socioeconomic and satellite data were linked to understand land cover change over a

three year period. While these authors compared three land tenure types, the comparison

is made at the municipal level, which leaves out many important explanations found only

at a household level.

I focus on land use processes at the household level within three land tenure

categories in southwestern Amazonia: extractive reserve, agro-extractive settlement, and

colonist settlement. Consequently, the methodology involves both economic and land

cover analyses and was performed using three methods. First, I conducted Hieldwork,

which was divided into two parts: 1) a household economic survey in 2004; and 2) the

land cover GPS data collection during the summers of 2004 and 2005. Second, I

processed the data from the economic survey and GPS and satellite images. Households

were analyzed in terms of their economic results, that is, income and cost values

aggregated by economic activities (crops, timber and non-timber products, as well as

livestock). I utilized the GPS points and Landsat data to perform a supervised

classification for each community area. In order to correlate these results to the economic

outcomes, my classification distinguished four land cover types: crops, forest, pasture,

and forest regrowth. Third, I linked economic and land cover data for the three study

sites, in order to conduct a multivariate analysis to test my hypotheses about land tenure,

the economics of land use, and land cover outcomes.









In the next sections I discuss these three phases in more detail. I begin by

describing how the socioeconomic survey was carried out. In this part I discuss the

questionnaire, planning for fieldwork, a price survey, and the fieldwork itself. Second, I

focus on how the land cover GPS points were collected. The third part provides details of

how a calibration and validation of satellite data were carried out. The fourth section I

explains how economic and land cover data were analyzed.

The Socioeconomic Survey

In this section, I discuss how the socioeconomic data were collected. It is important

to note at the outset that this study resulted from multi-institutional collaborations with

various organizations in Acre, so I recognize my collaborators as co-authors of the

database on which I will draw. The 2004 survey resulted from collaborative efforts

among PESACRE, CTA, UFAC, and the State Government.

Besides research assistants, community members were also directly involved in the

field data collection. At the Porto Dias Agro-extractive Settlement, the residents

association selected five community members to take part in the research group. In the

Peixoto colonization site, the Novo Ideal Association's board director designated three

resident members to help us during our fieldwork. In the Chico Mendes Resex, the

Association of the Residents in Xapuri (AMOREX) mobilized five rubber tappers to

guide the group within the forest. They also helped applying the semi-structured

questionnaire. The Chico Mendes' fieldwork also counted on the participation of one

board director from the Rural Workers Union of Xapuri who had past experience with

socioeconomic surveys.

Prior to the fieldwork, additional planning was required given the distinct

characteristics of the study sites. First, helped by local stakeholders, I selected my study









sites within each land tenure category. Second, in order to grasp the land use specificity

of each land tenure type, I used meetings with community and institutions as an

opportunity to elaborate my questionnaire. After these two stages I carried out two

surveys. The first one was the household fieldwork in the three selected sites. The second

one was a price survey to make possible the estimation of income and costs. When I

concluded my two surveys, I organized the data in a spreadsheet program and calibrated

and validated the land cover classifications during the summer of 2005.

Household Sampling

The household sampling was carried out following three sequential phases. First,

helped by IBAMA, CNPT, CNS, and community organizations, I identified in the Rio

Acre watershed region community land tenures representing extractivist, agro-

extractivist, and colonist areas. For each community land-tenure types present in this

region, I distinguished communities by degree of socioeconomic development, taking

into consideration access (roads), presence of public services (school, hospital, etc),

community organization (association and cooperative), technical assistance, and level of

production. As a result, the Chico Mendes Resex, Porto Dias PAE, and Peixoto PC were

selected. Second, within these areas, I distinguished families working with

agrosilvopastoral and timber as well as non-timber forest management. This criterion was

adopted in order to evaluate the interface with public polices taking place in Acre since

1999, which have emphasized those land use strategies. Third, I also considered areas

with prior governmental investments, such as road construction, subsidies, governmental

support for commercialization, and certification. Consequently, one rubber tapper

association area, one rural association area, and three rubber estates were identified for

conducting this study. In the Chico Mendes Resex, the sample size was constituted by 34









families from the Floresta, Palmari, and Sho Pedro rubber estates. In Porto Dias, 27

households located in the area under the control of the Porto Dias Rubber Tapper

Association were selected. Lastly, within the Peixoto Settlement, the household sample

included 27 families organized under the Novo Ideal Rural Associationl6

The Questionnaire

The questionnaire was derived from participatory efforts with various stakeholders

and community members. This strategy was implemented to ensure that the livelihood

system of each land tenure type was captured by the survey. Overall, the questionnaire

focused on land cover, socioeconomic aspects, and resource management practices

targeted by forest government policies in Acre. As the livelihood and land use strategies

across land tenure are not similar, I elaborated one questionnaire for each study site (see

Appendix A, B, and C). For instance, in the Chico Mendes Resex communities, timber

forest management was not included in the questionnaire, because this activity is not

permitted in that tenure type.

The questionnaire was divided into eight sections. The first included

socioeconomic topics, such as household composition, migration, accessibility, and land

use distribution. Section two focused on crop production, livestock, and machines and

equipment. The third section quantifies costs and yield per type of activity conducted by

households, such as crops, livestock, and forest management. This section also included

family expenses on purchased goods and services (such as rice, beans, oil, salt, clothes,

health etc), as well as family and outside labor force. The fourth part evaluated the

amount of goods cultivated and/or harvested that were used for family consumption, such


SThe number of families does not correspond to the current number of members in this association. That is
because the sample included families that were not affiliated with the organizations.










as annual and perennial crops, livestock, herbs, forest products, among others. The fifth

section of the questionnaire considered cash income from commercialized production.

The sixth included off-farm income (e.g., bolsa escolar, salaries, retirement income). To

assess institutional support for land use strategies, the seventh section included questions

on rural technical assistance from governmental or non-governmental organizations.

Finally, the last section had open questions about public policies implemented in the

community area.

The Household Fieldwork

The field survey was carried out from May to August 2004. I selected five research

assistants with prior experience in socioeconomic surveys in rural communities in Acre,

each with different backgrounds (economics, geography, history, agronomic engineering,

and forestry engineering). On average, we spent one day with each household, in order to

conduct a semi-structured interview, participatory land use mapping, and GPS training

sampling.

Prior to the fieldwork, the assistants attended a one-week training course focusing

on the livelihood systems of smallholders, semi-structured questionnaires, and the use of

remote sensing and GIS techniques in the field. In the last two days of training, they

practiced these techniques in the the Humaita Colonization Settlement, located

approximately 30 kilometers from Rio Branco, in the municipality of Porto Acre.

Moreover, community members were included as part of the research team.

The Price Survey

To estimate values for all household costs (expenses) and income sources

(revenues), I conducted a price survey in the municipalities where families

commercialized their products and bought goods: Xapuri, Brasileia, Rio Branco,









Acrelindia, Epitaciolindia, and Placido de Castro. Moreover, cooperatives and

associations of producers were contacted, as well as the government offices to determine

the prices for timber and non-timber forest products. To correct for inflation, I used the

IGP (General Price Index), the most common index in Brazil for that purpose, to correct

all prices to February 2005. I used a mean price for each product or service to estimate

the income and costs of each family, which were corrected by IGP. These values were

converted to the mean U.S. dollar Brazilian real (R$) exchange for February 20052

Participatory Findings Assessment: Attempts for Calibration and Validation

Household analysis, especially when focused on economic analysis, has often been

an obstacle for socioeconomic researchers. Ellis (2000) emphasized the complexity of

empirical appraisal at the household level as a result of livelihood diversity as well as

inter-temporal social processes continuously taking place. Schmink (1984) argued that

household economics are determined by social processes which are historically defined.

In this context, I have understood that rural policies, as in Acre, have incessantly

failed as a consequence of lack of knowledge about smallholder livelihoods and

alternative strategies. In order to address these challenges, Hildebrand and Schmink

(2004) suggest a framework combining participatory methods and ethnographic linear

programming fostered by a constant participative process of simulation, calibration, and

validation of household models.

The survey methodology used in this study employ this approach. I involved local

residents and decision makers of local, regional, and national organizations during the

sampling, fieldwork planning, and the fieldwork itself. Calibration and validation of data


2 This study utilized the mean US$ to R$ exchange for February 2005 of US$ 1.00 to R$ 2.597.










were carried out in May 2005 when several stakeholders attended seminars in order to

evaluate the datasets.

These seminars were conducted in four phases. First, after the household and price

surveys, the data underwent processing and preliminary analyses. Results were then

discussed with researchers, policymakers, NGOs, and community organizations in Rio

Branco. Next, the data related to the amount of production and prices for timber and non-

timber forest products were corrected. Third, findings were submitted to community

members for evaluation of in each study site via participatory dialogue we discussed the

forest and agrosilvopastoral seasonal calendar, as well as the range of main products

extracted and their prices, and income and costs distribution. After these meetings, I

identified discrepancies, and the data were adjusted and re-processed.

Land Cover Data Collection and Fieldwork

In this study, I used Landsat 5 TM+ images from the scenes 67/001 and 67/002 of

about the same time of the year. The 2004 images were taken between July 29 and

August 26. For the two scenes, I used bands 1 through 5. In addition, I utilized ancillary

data including polygon shape files containing land tenure types applied for the study

sites. Polygons delimiting the Chico Mendes Resex, Porto Dias Agro-extractive

Settlement, and Peixoto Colonization Settlement were selected for this analysis.

While doing household interviews, the research team collected training samples of

land cover types (crops, pasture, forest regrowth, and forest) at the three study sites.

Additionally, a printed sub-set image of 2004 Landsat image3 was provided to each team

member for each area to identify likely land cover types in the household area. All GPS


3 The Landsat images were adjusted for RGB= 5, 4, 3 and georefereced for UTM, datum SAD 69, zone 19
South.









were adjusted to UTM, datum SAD 69, and zone number 19 South. To increase accuracy

of these data, each land cover point was collected at least three times. Training samples

collection was carried out using a modified version of the CIPEC training sample

protocol (Appendix D), which includes details such as soil, vegetation type, and presence

of human disturbance.

In this study, we collected 442 training samples and 366 CIPEC forms. This

fieldwork was carried out in two phases. First, during the fieldwork in 2004, we collected

345 training samples and completed approximately 300 land cover forms. Although the

field team was intensively trained for the land cover GPS point collections, I encountered

some difficulties in performing a supervised classification for the study areas during the

fall 2004. As Daly and Silveira (2001) pointed out, over 50% of Acre' s territory is

covered by bamboo forest. On the first supervised classification performed at the end of

2004, this vegetation dominance created confusion between forest regrowth, crops, and

bamboo-dominated vegetation. Consequently (second), from May to June 2005, I visited

all three community areas to collect new land cover samples (principally within bamboo

and secondary forest areas). During this period, I gathered 97 samples and 66 land cover

protocols. In the summer of 2005, I was able to complete a new supervised classification

and verify its accuracy at each study site.

The Post-Survey: Data Processing and analysis

Household Economic Approach

The economic analysis performed here was based initially on-Brazilian literature on

micro-economics and agricultural economics (e.g., Gastal 1980, Mafra 1988, Barros and

Estacio 1972, Guerreiro 1994, and Martin 1995) that elaborated economic parameters for

assessing input and outputs among smallholders. In addition, I drew on studies focusing









on the Amazonian smallholders (Vosti et al. 2003, Walker 2003) which have evaluated

households as economically rational decision makers who pursue profit maximization,

that is, they continuously plan their cost-income balances. This traditional model of

analysis ignores the aspects highlighted in peasant economy theory (Thorner et al. 1986).

This theory establishes that household economics (e.g., extractivists, agro-extractivists,

and small farmer colonists) aim to balance both market (sold yield) and consumption

(subsistence yield) goals. In this context, family labor as a cost component and

subsistence as an income element represented two important factors in the analysis.

For identifying the land tenure differences, an ex-post4 eCOnomic analysis was

performed in which inputs (factors of production) and outputs (production or economic

results) were estimated. Three aspects advocated by Gastal (1980), Mafra (1988), and

Barros and Estacio (1972) were utilized as guidelines for my economic assessment: 1) the

community land tenures as a system constituted by social, economic, and ecological

components; 2) an understanding of the extractive reserves, agro-extractive settlements,

and colonist settlements as systems of production defined by biophysical and social

dimensions; and 3) a view of the household as an economic unit within each land tenure

system and constituted by production and management goals. These considerations

required that inputs and outputs should be measured to capture the uniqueness of each

household within each community and land tenure type. The first step of the economic

analysis was based on income, cost, and profitability by type of product at the household

level. In the following section, I discuss these economic measures.



4 By ex post economic analysis I mean to say that household economic assessment was performed based on
household economic activities carried out prior to the survey. The 2004 household survey covered
economic activities carried out from June 2003 through July 2004.










Gross Income (GI) is defined here as the result of selling a product in the market,

and acts as an indicator of the scale of production. It also includes family consumption

from crops, livestock, and forest products. In addition, it reflects the land use strategy of

each family. Crops and cattle are expected to represent the maj or components for income

of colonist residents (Peixoto community), while timber and non-timber products are

expected to be most important for Porto Dias and Chico Mendes respectively. Gross

Income is calculated as follows:

GI= Qm. pp, (..

Where,

Qm = qv + qe + q, (3.2.)

Qm=quantity of the product destined for the market

qv= quantity of the product sold

qe=quantity produced in previous years and not sold

qc=quantity produced or harvested consumed by family

qp=unit price of the product

Total Cost (Te) is defined here as the total monetary value spent by the household

to produce or harvest specific products. It is derived by summing two components: the

variable costs (Ve) and the Eixed costs (F,). In the household analysis addressed by this

study, labor is the main element in V,. The total Eixed cost is distributed between values

that correspond to the determined product i, and is made up of the specific Eixed cost for a

determined product i (EFe) and the common cost of i products, called the fixed common

cost (CFe).


To = (Ye), + (e,) c 7F (3.3.)
=1- =1-









Where,

T, = Total Production Cost

V, = Variable Production Cost of product i

e,, = Specific Fixed Cost of product i

Ce = COmmon Fixed Cost of product i.

To evaluate the economic viability of the household production system, this study

also measured the profitability of rubber tappers and colonists economies categorized by

products. I therefore generated a profit (P) analysis to compare categories of products and

to correlate their profitability with deforestation.

Profitability also allows verification of the possibility of a financial gain, and

consequently accumulation at the family level. In contrast to some other studies which

used a type of analysis based only on gross income and total cost (Vosti et al. 2003), this

work included family subsistence into the gross income. The calculated formula for

profitability is,

R = GI Tc (3.4)

According to the results, I may observe three specific situations:

R > 0, where there is a gain

R < 1, where there is a loss

R= 1, break-even

Economic Analysis

The economic analysis of the study was based on primary data from 2004 and

focused on economic outcomes of land use strategies utilized by families in each land-

tenure type. For comparative purposes, I divided the data analysis into three parts. First, I

performed an economic analysis at the community level. During this phase I divided the










data into income and cost outcomes aggregated by products. In order to evaluate these

results I utilized mean, standard deviation, and skew. Second, I carried out a similar

analysis by land tenure types aggregated by crops, livestock, non-timber, and timber

activities. During this phase I used F-tests results to evaluate whether differences of

means among sites were significant. And third, I also performed a profitability analysis

by land tenure types.

Land Cover Data Processing

In addition to the socioeconomic survey, remote sensing techniques were utilized in

order to evaluate the land cover distribution within each study site. The land cover

processing performed in this study is presented in Figure 3-1. Using georeferenced land

cover samples collected in 2004 and 2005, I produced a land cover signature based on

2004 images for four classes: forest, crops, pasture, and regrowth. I then utilized this

signature to perform a supervised classification on a mosaic of 67-001 and 67-002 scenes.

Two hundred and twenty-one training samples of 442 collected in 2004 and 2005 were

selected through stratified random sampling for image classification. The remaining 221

training samples were used to evaluate the accuracy of my classification. All analyses

were performed using Erdas Imagine software version 8.4, ArcView GIS version 3.2, and

SPSS version 11.0.

Landsat images processing was carried out following various steps (Figure 3-1).

The first procedure to assess land cover was to calibrate the images (Green, 1999). I also

performed band-by-band atmospheric correction of the 2004 images as discussed in

Jensen (2005). I used a layer-stack to merge the five bands for analysis of each scene in

separate .img (Erdas) files. Next, I performed pre-processing tasks, and image-to-image

geo-referencing was done based on a 1995 image as a reference (Jensen, 2005). Once I












obtained the georeferenced, calibrated, atmospherically-corrected, and merged images for


67-001 and 67-002 scenes, I performed a supervised classification of the 2004 image.


Acre s
| ~1996 Image Image Training
Individu I Merged Georeferenced cairto albr td etm sape
2004 Lan sat LI0 ea fLa 0


Atmospheric
correction Sprie


boundaries Training
samples

smallholders
2004 Image Classified
land cover subset Accuracy 2004
types sssmnImage


SPSS Images
Processing assessment



Figure 3-1. Diagram of land cover processing in the study area, Acre, Brazil.


The classification for 2004 was submitted to two types of assessment. The accuracy


assessment was conducted based on 221 training samples collected during 2004 and


2005. Furthermore, a field appraisal of this classification was carried out in June 2005.


Over two weeks I visited each of the three study sites and assessed the accuracy of the


2004 classification, comparing on-the-ground observations with a printed map of the land


cover results in each community areas. With a GPS and via conversations with local


residents, I was able to validate the accuracy of each land cover class within each


community area. After the accuracy assessment and field appraisal of the classification


results, I identified the amount of area covered by each class at the household level.


These results were incorporated in the SPSS economic database.


Since this study addresses a socioeconomic and land cover analysis at household


level, I created a border for each family's landholding and analyzed it separately. In










Peixoto, I utilized the georeferenced plot boundaries collected by INCRA for each local

resident. Based on these digitalized lots I identified the GPNI community boundaries as

well as the properties selected in this study on the classified Landsat image (Figure 3-1).

Land cover analyses were performed on those selected residents.

In the Chico Mendes and Porto Dias communities, the limits of each household's

landholdings were not as easy to identify. Commonly, rubber tappers utilize rubber-trails

(estrada~s de seringa) as their natural boundaries. CNPT/IBAMA and INCRA, the

institutions responsible for those land tenure types, have had difficulty establishing

boundaries among smallholders of extractive and agro-extractive reserves. After an

intensive literature search for studies with similar problems in delineating household

property boundaries, I decided to utilize the Thiessen polygon algorithms based on fuzzy

set theory to produce boundaries for each colocagdo on these rubber estates.

Using the algorithm discussed above, I was able to define the boundaries among

each household property following three procedures. First, I used the location of the

household's home within their colocagdo, available from CNPT/IBAMA and INCRA, to

identify each family area. In the Chico Mendes Resex, I only selected colocapies within

the Sho Pedro, Palmari, and Floresta seringais. In the Porto Dias case, there were no

borders among seringais, and I made use of all georeferenced colocagdo points. Second, I



5 This is a remote sensing technique in which an individual area of influence around each set of points is
defined. Based on fuzzy theory, the polygons' boundaries are the area that is closest to each point relative
to all other points. They are mathematically calculated through two sequential steps. First, the entire area is
divided into polygons in which one polygon per point is established. Second, the distance among points
indicating colocagaes area is minimized. However, in the broad sense, the fuzzy logic utilized in this
approach to estimate boundaries of colocagaes, is a process of inferring imprecise conclusions from
imprecise premises. As pointed out by Dragi~evid (2005) this deduction is approximate rather than exact
because data described and generated under the circumstance of lack of information and accuracy
assessment implies a high cost. Since my summer time in Acre was not sufficient to assess the accuracy of
this technique within each family area, I was only able to evaluate the boundaries of two smallholders
within of each land tenure type.










produced colocagdo polygons using the Thiessen algorithm in ArcView based on fuzzy

set theory. Third, I selected polygons of sampled colocapies, which were georeferenced

during the 2004 fieldwork, in both Chico Mendes and Porto Dias totaling 61 family areas.

Economic and Land Cover Integration

Economic data from the household surveys were integrated spatially in a GIS with

satellite data for the land cover classes for each corresponding property polygon in the

three study sites. To identify the main factors related to deforestation in the study sites, a

multivariate regression analysis was carried out. Taking into consideration the limited

sample size in this research, this analysis was performed for all sites (88 households).

During this part of my analysis, I constructed five regression models based on

different theoretical foundations. The first model evaluated the effect of land tenure

category for deforestation. The second was a life cycle model in which I assessed the

effects of family dynamics on deforestation in the study sites. The third was a market

integration model in which I evaluated the influence of market access and family cash

purchases on land cover dynamics. The fourth model assessed the effects of cost

composition for deforestation among extractivist, agro-extractivist, and colonist areas.

Finally, in the last model I estimated the importance of assets for land cover dynamics in

the study sites.

Conclusion

This thesis seeks to establish linkages between economic and forest cover datasets

of three community land tenures types in southwestern Amazonia. I utilized a

combination of methods to link socioeconomic and satellite data. First, to generate an

understanding of distinct land tenure categories, a questionnaire was generated via a

collaborative participatory process. Second, I conducted the economic field survey of










households in the three study sites, recognizing the differences in the rules of each land

tenure type concerning resource management. Third, the classification of satellite images

for each study site was based on the four basic land cover types existent within of the

communities: forest, forest regrowth, pasture, and crops. Fourth, the data integration was

carried out by defining a common unit of analysis for both economic and remote sensing

approaches, in order to unify the data spatially to allow for a multivariate regression

analy si s.















CHAPTER 4
ECONOMIC AND LAND USE COMPARISON OF SMALLHOLDER LAND
TENURE MODELS IN THE STATE OF ACRE

Introduction

This chapter responds to the need for an empirical assessment of how community

land tenure models in Acre differ in terms of their economic and land cover outcomes.

Specifically, the chapter concentrates on how economic variables affect land cover

change among households within Resex, agro-extractive settlements, and colonist

settlements.

In order to address these issues, this chapter has three obj ectives. The first is to

examine crop, pasture, regrowth, and forest areas at the household level and to determine

the extent to which timber and non-timber forest management as well as

agrosilvopastoral strategies lead to different amounts of deforestation. The second

obj ective is to estimate the economic efficiency of each land use strategy implemented by

inhabitants of each study area. Finally, the third objective is to evaluate how micro-

economic factors are related to land cover distribution. Other non-economic factors, such

as family composition, age, distance to market, and length of residence are also

considered to explain land cover at the community level.

This chapter is organized into three parts. Part one offers an overview of land cover

distributions in each study site, in terms of forest, crops, pasture, and forest regrowth.

Part two introduces the economic analysis by evaluating household income, costs, and

profitability among tenure types and income sources, particularly pointing out the










differences between timber and non-timber forest management. Finally, part three

presents a multivariate statistical analysis of the effects of land tenure, economic

variables, and non-economic factors on land cover change. The chapter concludes with a

discussion of the main findings.

Revising the Research Questions

The first research question posed by this study is: how do economic and land cover

outcomes compare among the three types of land tenures? The land cover change

literature includes studies of land tenure, but studies comparing land cover outcomes

among various land tenure types in a given region remain very rare. The importance of

designing land tenure institutions as an influence on economic and environmental

outcomes therefore remains poorly understood. This limitation in the land cover change

literature is all the more significant because the design of land tenure institutions is

something that governments can control, and the diversity of land tenures in Acre, Brazil,

reflects many policy experiments to yield better economic and environmental outcomes

from land use. For precisely this reason, there is great interest in Acre, in Brazil and

elsewhere, and this interest motivates my first research question in this thesis.

The second question addresses an economic assessment of non-timber and timber

forest products among small producers. Since the creation of Resex and agro-extractive

settlements, their inhabitants and a series of development programs have concentrated on

the economic use of a range of non-timber forest products. More recently, timber forest

management proj ects have been introduced in several of these areas. The shift from

neoextrativismo to florestania is significant not only for the potential consequences on

land and forest resource use in different tenure types in Acre, but also for the profitability

of timber and non-timber forest products. That said, over the past years an intensive










debate has emerged focusing on potential economic risks associated with timber and non-

timber resource use within community areas in Acre. Furthermore, economic outcomes

from the recent emphasis on timber extraction are a key factor influencing smallholder

land use decisions. Therefore, my second research question focuses on the profitability of

timber and non-timber resource use.

The third question focuses on the factors best explain forest clearing among

households in Resex, agro-extrative settlements, and colonization settlements. Although

economic goals have been considered a key component in influencing land use strategies

within community land tenure areas, various income-deforestation tradeoffs can emerge.

For instance, smallholders can use profits to invest in pasture for cattle, which is a labor-

saving strategy utilized by families, but ranching results in more forest clearing.

However, by the same token, if forest management is economically viable (profitable), it

can also stimulate local inhabitants to use the net income to invest in activities that imply

to increase deforestation.

In working with each of these three questions I have developed one or more

specific expectations (hypotheses):

* H1: Extractive reserves such as the Chico Mendes Resex will have the most favorable

environmental outcome (lowest proportion of land deforested per household).

Colonization will have the best economic outcome (highest gross household income).

Agro-extractive settlement will have the best overall outcome (deforestation nearly as

low as Resex, and income almost as high as colonization settlement).

* H2: Non-timber forest products will be more profitable, because of their lower costs,

than timber extraction.










*H3: Higher profitability leads to greater deforestation, though land tenure categories

will also be important.

Land Cover Distribution among Land Tenures

This section presents a land cover assessment of household landholdings in the

three study sites. These results are particularly important to evaluate the policy

experiments concerning these different land tenure models in terms of land cover

outcomes. This study thereby provides a comparative understanding of how land cover

varies among the Resex, PAE and PC I visited. In this section I provide the results for my

supervised classification of the three land tenure types. First, I discuss the accuracy

assessment. Second, I compare three study sites in terms of their land cover distribution.

Accuracy Assessment of Land Cover

The accuracy assessment evaluates the precision of the supervised classification in

identifying land cover categories among the study sites. Table 4-1 presents the results in

the error matrix for the classes of forest, cropland, pasture, and forest regrowth from the

supervised classification. The matrix illustrates both errors of exclusion (omission errors,

i.e. degree of Producer' s accuracy), which measures how much of the land in each

category was classified correctly, as well as errors of inclusion (commission errors, i.e.

degree of User' s accuracy) that measures the percentage of correctly classified pixels in

each land cover category of the images (Jensen 2005). The overall accuracy is expressed

as a combined percentage of the test-pixels successfully assigned to the correct classes.

The results indicate that the classification had a high degree of accuracy for the

land cover classes 'crops', 'regrowth', and 'pasture,' with 89-96% producer's accuracy.

However, the accuracy for correctly identifying forest areas was lower, with a 74%

producer' s accuracy, which likely resulted from the difficulty in classifying bamboo-










dominated forested areas in the study region, which can easily be confused with

regrowth. Nevertheless, the user' s accuracy results for pasture and crops indicate that my

2004 classification was conservative, implying a potential bias toward classifying these

land cover categories as forest, resulting in an underestimation of the extent of

deforestation. On the whole, one might conclude that because the overall accuracy of the

entire classification was quite high (87%) and the producer' s accuracy varied from 90 to

96%, this classification is adequate for analyzing forest cover in the study areas.

Table 4-1. Accuracy assessment for 2004 supervised classification for the study sites in
Acre, Brazil.
Reference Classified Number Producer' s User' s
Class name
totals totals correct accuracy accuracy

Unclassified 9 7 0
Forest 39 33 29 74.4% 87.9%
Pasture 36 42 32 88.9% 76.2%
Crops 49 58 47 95.9% 81.0%
Forest regrowth 88 81 80 90.9% 98.8%

Totals 221 221 193
Overall Classification Accuracy = 87.3%

Comparing Land Cover among Land Tenures

Land cover analysis was performed for the households sampled in the three study

sites, each corresponding to a different category of community land tenure. To perform

the analysis, it was necessary to identify or estimate the boundaries of land claimed by

the sampled households. While boundary data regarding the division among household

lots in the Peixoto colonization proj ect were available from INCRA, the geographic

limits of the colocagdes in both the Chico Mendes Resex and the Porto Dias PAE had to

be estimated using the Thiessen polygon algorithm based on fuzzy set theory.









Before applying this estimation method to the whole data set, the methodology was

tested in 5 colocagdes in the Porto Dias community. During that test, rubber tappers of

the Porto Dias indicated that their areas had between 3-5 rubber trails This results in

overall estimates of 300-500 hectares per household landholding. Using a GPS unit and

guided by rubber tapper household members and their neighbors, I measured colocagdo

boundaries and on the basis of these measurements estimated the total colocagdo areas to

be between 230-560 hectares, indicating a wider range of area than conventional

estimates. The colocagdes' shapes resulting from using the Thiessen polygons based on

fuzzy set theory were compared to the Hield data and indicated that they were similar to

the limits identified with the GPS tracking points.

An important implication that can be derived from these results is that deforestation

levels for each rubber tapper area in the Chico Mendes and Porto Dias, if based on the

conventional 100 hectare/trail assumption, are likely to be underestimated for some

families and overestimated for others. Hence, most previous deforestation estimates and

deforestation controls at household levels have likely not been very accurate. This study

therefore presents a first attempt to overcome this problem.

Using land cover classifications and areal estimation Eigures for all three study

sites, I obtained the results presented in Table 4-2, which shows descriptive statistics for

each land cover category. In order to identify the most appropriate metrics to evaluate

land cover distributions, I present results for: raw values, percentages, and natural log

values. Land cover was categorized in terms of crops, pasture, and regrowth. In addition,

I calculated deforestation category summing crops, pasture, and regrowth.


SRubber trails have been conventionally estimated as covering roughly 100 ha each










Table 4-2: Land cover categories at household level, all sites (n=88), Acre, Brazil.
Land cover category Hectares Percentage Log value

Deforestation
Mean 29.94 16.59 3.11
Standard Deviation 22.45 14.46 .82
Skew 1.55 .68 -.49

Crops
Mean 8.46 4.55 1.92
Standard Deviation 6.31 3.87 .69
Skew 2.35 1.22 -.23

Pasture
Mean 20.20 11.39 2.48
Standard Deviation 17.32 11.21 1.21
Skew 1.00 .79 -.704

Regrowth
Mean 1.28 .65 -.38
Standard Deviation 1.68 .87 1.27
Skew 2.62 2.24 -.07




While no land cover types presented a high standard deviation value for all types of

measurements, the skewness indicated some deviation from normal-di stributions. Skew

values close to "O" suggest a normal distribution. On the one hand, the skew for mean

raw and percentage values was positive, which means that their distributions tend to have

extreme high-values. On the other hand, skew values for logged values were all negative,

which indicates that this distribution tends to have extreme low-values. Based on the

findings in Table 4-2, for the remainder of the analysis I focus on the percentages. This is

because they have low standard deviations and skew values, something lacking in the raw

values. Moreover percentages are easier to understand and compare than raw or logged

values.










Crops and pasture are the most important land cover categories deforestation in the

three study sites. Deforestation overall exhibits a mean of around 30 hectares (~ 17% of

total area) per household. This result might indicate a total level of deforestation above

the 10% authorized by IBAMA and INCRA within Resex and agro-extractive

settlements. However, this interpretation ignores potential differences in the extent of

deforestation among land tenure types.

The lowest values are for the regrowth class, with a mean of less than 1% of the

total area. This result can be explained by the importance of crops and pasture to

smallholders. Commonly smallholders temporarily abandon productive areas to regrowth

for 2-5 years, called the pousio period, then they clear, burn, and reuse the regrowth for

crops or pasture (Amaral 1998).

The most important results in Table 4-2 are twofold. First, percentages are the most

appropriate measure to evaluate the distribution of land cover in the study sites. Besides

the low standard deviations and skew values, percentages allow for direct comparisons

among the land tenure categories and are easier to interpret. Second, the results show that

pasture is the main land use category that contributes to overall deforestation. However,

the importance of crop cultivation and cattle ranching may vary among land tenure

categories. To assess this possibility, the next step in this analysis compares the

percentage of land covers among the three land tenure types.

Table 4-3 provides a comparison of land covers among land tenure types in the

three study sites. I present F-tests for the significance of differences among the three

tenure types. Deforestation percentages are significantly different across land tenure

types (p<.01). Although colonist families in Peixoto PC have made efforts to establish






























Deforestation (%) 4.99 12.27 35.52 181.543**
Crops (%) 2.32 3.13 8.75 48.555**
Pasture (%) 1.99 8.87 25.76 169.111**
Regrowth (%) .68 .25 1.00 5.560*

p<.05, **p<.01

In addition, the percentage areas under crops and pasture are also significantly

different among the three tenure types represented in the study sites (p<.01). On the one

hand, pasture is the land cover that contributes most to deforestation in both the Porto

Dias and Peixoto sites. In the Chico Mendes Resex, on the other hand, most forest

clearing was for crop production. This indicates that while rubber tappers of Chico

Mendes are focusing their efforts on agriculture, possibly for their subsistence and

production of cassava sale, cattle raising emerges in Porto Dias and Peixoto as the most

important activity contributing to deforestation.

The F-Tests result indicated that the differences among mean percentage for

regrowth across land tenures are significant (p < .05), the differences are not as

pronounced as for the other land cover categories. Porto Dias PAE seems to be the


sustainable agrosilvopastoral systems, this area exhibited the highest level of

deforestation (~ 36%). In contrast to Peixoto, the three rubber estates of the Chico

Mendes Resex selected in this study exhibited the lowest level deforestation percentage

(~5%). And in the Porto Dias PAE, the households sampled had deforestation

percentages in between the other two land tenures (~12%), but closer to the percentage in

the Resex than the PC.

Table 4-3. Descriptive statistics for land cover categories, by land tenure type, Acre,
Brazil, 2004
Chico Mendes Porto Dias
Land cover Peixoto PC
Re sex PAE F -te sts
variables (N3 N2)(N=27)









community with the lowest level of this land cover category (0.25%). In the Chico

Mendes Resex, this mean percent value is 0.68%. As regrowth is an area of reserve for

potential future use, its extent can be interpreted as an indicator that other areas within

landholdings might be been utilized for crops or pasture purposes. In her research in

Acre, Ehringhaus (2005) observed that the overall percent of regrowth area is decreasing

due to two parallel processes: 1) as residents invest more in ranching, they are converting

young regrowth into pastures, and 2) they are converting agricultural fields directly into

pasture and not letting them lay fallow for the development of new regrowth. Moreover,

she noted that residents are also increasingly establishing new fields in primary forests

and less in regrowth areas.

These results provide important insights to evaluate the effectiveness of policies

underlying each land tenure category with regard to its environmental conservation goals.

While the Chico Mendes Resex and the Porto Dias PAE were created to establish land

concessions for sustainable forest use, the Peixoto PC emerged as an area for agricultural

land use, and not forest conservation. The policy goals of colonization proj ects, agro-

extractive settlements, and extractive reserves are thus very evident in the higher

percentage of deforestation in Peixoto than in Porto Dias and Chico Mendes. This

suggests that the land use regulations institutionalized in agro-extractive settlements and

extractive reserves, at least as practiced in Porto Dias and Chico Mendes, are effective in

reducing deforestation relative to that seen in colonization areas such as Peixoto.

This section evaluated land cover in the Chico Mendes Resex, Porto Dias PAE, and

Peixoto PC. Findings from remote sensing indicated differences between the three sites,

especially with respect to deforestation for pasture and agriculture. The different land










cover profiles of the three tenure categories raise questions about the conservation value

of these areas and the household economic decisions being made in the context of

different land tenure regulations concerning resource management. In the next section, I

therefore address such questions with an analysis of income, costs, and the profitability of

different economic activities in the three study sites.

Economic Analysis for Land Tenures in the State of Acre

This section offers insights into smallholder land use strategies based on three

economic criteria: income, cost, and profitability. First, I will describe the data regarding

cash income generation across the three land tenure types. Second, I will compare costs

of production for different products. Third, I will concentrate on a profitability analysis

for all product types by category of land tenure. These results will allow a comparison

across land tenure types as well as categories of products. As forest policies and

institutional goals within these land tenure models have sought to enhance economic

welfare among households, this analysis can provide an evaluation of the effectiveness of

the strategies behind policies constituting different land tenure types. In particular, this

section will compare the three types of alternative land use strategies implemented in the

three land tenure models: timber, non-timber, and agrosilvopastoral production. I will

evaluate these strategies economically in terms of their gross income generation,

production costs, and profitability, may be quite dissimilar. In addition, I will address the

issue of whether these economic outcomes may vary so as to reveal tradeoffs with respect

to land cover outcomes.

Gross Income

Table 4-4 compares gross household income among the three land tenure type.

Standard deviations were low in most land tenure types. In addition, except for gross









income from livestock within the Chico Mendes Resex and the total income for Porto

Dias PAE, all skew values were larger than 1.0, which suggests a few "outliers" with

high values. Logging is not authorized within the Chico Mendes Resex by IBAMA and

CNPT; therefore I report no income data from logging in this area.

Table 4-4. Descriptive statistics for annual income, by land tenure type, Acre, Brazil,
2004
Chico Mendes Resex Porto Dias PAE Peixoto PC
Ecnoicacivty(n=34) (n=27) (n=27)

Crops (US$)
Mean 181.00 311.00 1,763.00
Standard Deviation 140.00 150.00 1,433.00
Skew .33 1.78 .77

Livestock (US$)
Mean 268.00 979.00 1,931.00
Standard Deviation 176.00 723.00 1,879.00
Skew -. 11 1.84 1.19

Non-timber (US$)
Mean 681.00 805.00 28.00
Standard Deviation 427.00 467.00 61.00
Skew .63 .32 3.00

Timber (US$)
Mean -- 1,385.00 153.00
Standard Deviation -- 1,840.00 736.00
Skew -- .57 5.15

Total (US$)
Mean 1,130.00 3,567.00 4,109.00
Standard Deviation 495.00 1,497.00 2,759.00
Skew .32 -.12 .71



Three noteworthy differences emerge in Table 4-4. First, as expected, Peixoto had

the highest mean total gross income (~ US$ 4,100.00 per year). Livestock as well as

crops are the most important products to the households of this area. Given the high









deforestation percentage in Peixoto, this result indicates an economic-environmental

tradeoff, which also casts doubt on the commitment to sustainability on which

agrosilvopastoral systems were based. In fact, during the Hieldwork in 2004, no small

farmers reported any income from agroforestry or from a combined crops-forestry

system. Second, households in the Chico Mendes Resex had the lowest mean total gross

revenue (US$ 1,130.00 per year). Given the low deforestation percentage in the Resex,

this finding also implies an economic-environmental tradeoff. But third, Porto Dias

stands out, because residents of this area generated a mean gross income of US$ 3,570.00

per year. This is a key finding, because resource use in Porto Dias reflects a combined

forestry and agrosilvopastoral model, which yields a mean gross household income

nearly as high as in the Peixoto colonization proj ect, with a deforestation percentage

nearly as low as in Chico Mendes Resex, via a diversified production system.

I suggest two possible explanations for differences in cash income generation

across land tenure types. First, a household's ability to mix multiple income generation

strategies contributes considerably to the access to various market niches. Second, a

combination of prices incentives, distance to market, and local demand may also be key

factors affecting these communities. These two interpretations were confirmed during the

Hieldwork. Rubber tappers of the Chico Mendes Resex concentrated their efforts on the

extraction of rubber, a product with lower prices, and on harvesting Brazil nuts that bring

high prices. In Porto Dias PAE, families did not explore rubber and preferred to devote

their efforts to products with high prices, such as Brazil nuts, copaiba oil, forest seeds,

and medicinal plants. Peixoto is an area with higher crop and cattle demands from large










ranchers and urban areas, which may have contributed to a higher gross income from

crops and livestock activities.

With the exception of Peixoto, the results confirmed the economic importance of

land use strategies based on non-timber and timber production. In the Chico Mendes

Resex, non-timber forest management was the land use that provided the largest share of

mean gross income (~ US$ 690.00 per year). Timber (~ US$ 1,385.00 per year), on the

other hand, was the product with the largest share of mean income among rubber tappers

of Porto Dias. Further, the mean income obtained from non-timber forest products (~

US$ 800.00 per year) among residents of Porto Dias was higher than income obtained by

families in Chico Mendes.

Although sustainable forest uses were confirmed as the most important activity for

rubber tapper communities in Chico Mendes and Porto Dias, the results also indicated

livestock as an important activity for all areas. Among residents of Chico Mendes and

Porto Dias, this was the second most important activity, with approximately US$ 270.00

and US$ 980.00 of mean gross income per year, respectively. In Peixoto, however,

livestock was the most important source of income.

These results indicate that the residents of 'sustainable' land tenure models such as

Resex and agro-extractive settlements rely to a higher extent on forest products for their

income. While the average income of these families is lower than those living in the

colonization proj ects, the importance of forest-based production is clear. Another

important conclusion from these results is that although colonist households of Peixoto

have implemented an agrosilvopastoral model in their areas, livestock production remains

the most important source of income. At the same time, cattle ranching is also gaining










importance among rubber tapper households, raising new questions regarding the long-

term ecological sustainability of Resex and agro-extractive land tenure.

General Cost Analysis

While gross income is an important indicator of economic success of a particular

land use, the cost involved among the production types is also a key economic variable

driving affecting land use decisions in the different land tenure models. However,

economic analyses of land cover change have rarely considered production costs as well

as gross incomes generated by land use at the household level. In the following section I

will first focus on cost composition, by providing first an overall cost assessment for

households in my sample to identify the overall cost patterns. Second, I will concentrate

on a comparative cost assessment among the land tenure types. This discussion will also

compare production costs among types of activities carried out by households. This

information is important for identifying the level of economic viability for each type of

product, in the context of specific land tenure types. For instance, it may be useful to

assess the viability of implementing agroforestry system in the Chico Mendes Resex as

well as in the Porto Dias PAE.

Table 4-5 summarizes the composition of annual mean costs by economic activity

for all three communities taken together. Three types of costs are presented in this table.

The first column in the table refers to the total cost for each economic activity. The

variable costs are expenses that vary according to the volume and intensity of each type

of input, such as quantity of labor time, salt for cattle, seeds for pasture or crops, oil for

vehicles, etc. Fixed costs, on the other hand, are costs that arise with or without

production such as the depreciation of machines and equipment, and other investments in

the property. Table 4-5 presents total, Eixed, and variable costs for different economic










activities, namely crops, livestock, non-timber forest products, timber, and all of these

products together.

Table 4-5. Descriptive statistics for annual cost categories, all sites (n=88), Acre, Brazil,
2004.
Economic activity Total Fixed Variable

Crops (US$)
Mean 497.00 122.50 359.00
Standard Deviation 373.00 166.00 248.00
Skew 1.83 2.90 1.00

Livestock (US$)
Mean 593.00 137.50 455.00
Standard Deviation 622.00 194.00 533.00
Skew 1.86 4.23 2.00

Non-timber (US$)
Mean 453.00 65.00 388.00
Standard Deviation 442.00 78.00 396.00
Skew .78 2.54 .86

Timber (US$)
Mean 731.00 106.00 626.00
Standard Deviation 2,047.00 293.00 1,755.00
Skew 2.49 2.50 2.50

Total (US$)
Mean 2,316.00 437.00 1,878.00
Standard Deviation 2,099.00 374.00 1,859.00
Skew 1.86 2.51 1.87


As can be seen in Table 4-5, all economic activities had costs with relatively small

standard deviations, except the total and variable costs for timber. The result for timber

was strongly influenced by timber production costs in Porto Dias. The skew value for the

fixed costs for livestock were the highest among all categories (4.23), which suggests that

the cost values for this category tend to be higher rather than lower. The non-timber' s

skew value for total costs, on the contrary, was the lowest value (0.78). Overall, skew










values indicate that the results for all categories of products across all cost types tend to

be high, indicating "outlier" households that spend considerably more on certain

economic activities.

All cost types for timber showed high standard deviations, which reflect the high

costs of timber management. Certification, transport, road repairs, depreciation of

equipment, and a skilled labor force are the expenses that have contributed most to higher

costs for timber extraction.

Three main insights can be drawn from Table 4-5. First, variable costs are the main

category of costs for all productive activities. Among smallholders in Acre, family labor

is the main component of variable costs, a reflection of the importance of household labor

inputs and the labor-intensity of many of the activities in Table 4-5. Families in the three

study sites reported that all family members, including children, adults, and the elderly

are engaged in all production activities.

Second, timber extraction is very costly and not as economically attractive as its

high gross incomes seen earlier might suggest. The mean total cost (~ US$ 730.00 per

year) for timber extraction was higher than for livestock (~ US$590 per year), or any

other activity. As a result, when rubber tappers compare timber with other economic

activities, they are more likely to invest in non-timber forest products or cattle raising

instead of timber harvesting. Nevertheless, this may depend on the price incentive from

each economic activity. For instance, while the mean farm gate price for cattle in these

areas was US$ 143.00/head, during the same period the price for timber was US$

115.00/m3. Later in this chapter I will undertake a profitability analysis, which will

further explore this issue.









Third, although timber exhibited the highest total and variable costs, crops and

livestock were the activities with highest fixed costs. Two reasons account for this

finding. In recent years many households, especially in Peixoto, have invested in

equipment and assets for cattle and agriculture such as fences and equipment, supported

by BASA's (the Amazonian Bank) credit program. This contributed to high depreciation

costs for Peixoto households. Rubber tappers participating in the timber forest

management, on the other hand, rented skidders and other machines from third party

companies, thus reducing the costs of depreciation.

Three other conclusions may be drawn from the costs analysis in light of household

financial well-being as well as the earlier analysis of land cover. First, during the survey

the families pointed out that their systems are highly dependent and limited by family

labor availability. Second, approximately 80% of the total cost was constituted by

variable costs (US$ 1,878.00), which reveals the weak capacity of families to make

investments in technologies such as machines and other apparatus for reducing labor

costs. This situation has been partially overcome with funds and subsidies from

organizations such as BASA or the Acre government. However, this assistance is also

likely to increase the fixed costs as well as increasing the demands on family labor that

might have to be allocated to generate sufficient income to repay the credit. The forest

government has established programs to donate equipment for rubbers tappers and

colonist families to avoid this cycle of smallholder debts. The third conclusion relates to

the implications of costs for household land use decisions and land cover changes. Both

rubber tappers and colonists consider total costs and labor requirements in their land use

activities. In the absence of a family labor reserve and in the presence of high production









costs, many households may move toward activities that are less labor-intensive, such as

cattle raising. As result, this situation may contribute to increase the number of families

in debt with Einancial organizations.

The issues that emerged from the cost analysis are key aspects for any evaluation of

the effect of subsidy policies on household economic welfare as well as land use and land

cover change in the state of Acre. Economic and conservation strategies, then, should

take into consideration total costs and especially labor costs as key constraints for local

producers. However, it is important to take into consideration the variability of costs

among land tenure types designed to encourage different land use strategies.

Cost Analysis at Community Level

In this part I provide the second part of my cost analysis, examining costs

differences across land tenure types. It compares costs for different economic strategies

across households in extrativist, agro-extractivist, and colonist areas. Special attention

will be given to the costs of timber and non-timber forest production. In addition, this

section aims to briefly discuss why the cost categories for each productive activity differ

among land tenure types. These results will provide insight on how costs for each group

of products affect household land use decisions and land cover change.

Table 4-6 provides mean Eixed, variable, and total costs by land tenure type and

type of product. As anticipated, there are noticeable differences in the production costs

among the three land tenure types. F-tests results indicate that differences of costs for all

groups of activities are significant across sites. While Porto Dias was the community with

the largest mean total cost (~ US$ 3,900.00 per year), total costs in Peixoto were nearly

as high, whereas the families of Chico Mendes Resex had the lowest relative costs.

Although the mean total value for fixed cost was comparable for Peixoto and Porto Dias,









there are differences in the cost composition among their productive activities, especially

for timber. Moreover, it is apparent that the variable costs comprise the greatest portion

of total costs of all economic activities in all three areas. These results again confirm the

importance of family labor for all land uses.

Table 4-6. Annual mean for cost categories, by land tenure type, Acre, Brazil. 2004
Chico Mendes Porto Dias
Peixoto PC
Economic activity Resex PAE F-tests
(N= 27)
(N= 34) (N= 27)

Crops (US$)
Total Cost 401.00 303.00 812.00 21.054**
Fixed Cost 86.00 36.00 255.00 18.089**
Variable Cost 315.00 267.00 505.00 8.290*

Livestock (US$)
Total Cost 140.00 689.00 1,066.00 27.676**
Fixed Cost 64.00 104.00 263.00 10.226**
Variable Cost 76.00 585.00 803.00 22.795**

Non-timber (US$)
Total Cost 714.00 549.00 29.00 32.706**
Fixed Cost 117.00 54.00 9.00 22.325**
Variable Cost 594.00 495.00 21.00 27.633**

Timber (US$)
Total Cost .00 2,371.00 12.00 32.706**
Fixed Cost .00 339.00 5.00 22.325**
Variable Cost .00 2,033.00 7.00 27.633**

Total (US$)
Total Cost 1,252.00 3,914.00 2,057.00 16.929**
Fixed Cost 268.00 534.00 554.00 6.423*
Variable Cost 985.00 3,380.00 1,503.00 18.674**

*p< .05; **p< .01

Table 4-6 indicates that livestock production costs varied considerably across sites.

While the lowest total cost for livestock production was in the Chico Mendes Resex (US$

140.00 per year), the costs were higher in the Porto Dias PAE (US$ 689.00 per year) and

especially in the Peixoto PC (US$ 1,066.00 per year). These increasing costs can be










explained by the relative importance of this category of production for each community.

Notably, pasture degradation has contributed to an increase in the annual cost of livestock

production for families in Peixoto PC.

Table 4-6 also shows that variable costs are the highest costs for all products across

all sites. However, soil degradation has been related by colonist families as a key

ecological factor negatively affecting the costs of livestock production within the Peixoto

PC. This is an important as it may indicate the potential long-term economic inefficiency

of cattle ranching among small farmers in this region.

There are two significant differences between the Chico Mendes Resex and the

Porto Dias PAE. Although crop production was the second highest cost category among

households of the Chico Mendes after non-timber forest products, for the Porto Dias

families the livestock production costs were the second highest after timber production.

There are two likely explanations for these results. During the period of this research

(2004), the maj ority of families of Porto Dias were clearing regrowth for pasture, and this

task was carried out with family labor, which explains the relatively high mean variable

cost for livestock. In contrast to this, many colocagdes of the Chico Mendes Resex

cleared forest for the production of annual crops, which required intensive use of family

labor as well.

For the rubber tapper families, the highest production costs were associated with

forest products. In the Porto Dias PAE, the highest production costs stemmed from

timber production and to a lesser extent from livestock production. The forest

management proj ect implemented by the NGO CTA in this area since 1994 has

introduced an entrepreneurial management plan that has involved high economic costs.









Although over the past two years, rubber tappers participating in the timber initiative in

Porto Dias have attempted to reduce costs, expenses related to transport, depreciation,

and hired employees remain high, and represent a great part of the total costs. This

suggests the long-term economic challenged faced by households participating in timber

extraction. Non-timber forest products were the products with the highest costs in the

Chico Mendes Resex. Rubber remains a key activity for many families in this area, but

the low concentration of rubber trees per hectare as well as intensive labor inputs required

contributed to high variable costs.

Although cost may be a key variable affecting household decisions, economic

viability is the primary goal of all families. Thus, in the following section, as a third

component in the economic assessment I analyze the profitability of the different

productive activities in different tenure areas.

Profitability Analysis

After analyzing both gross income and costs of different land use strategies within

the three land tenure models, here I compare both income and costs in a profitability

analysis for all categories of products across land tenure types. This represents the most

important step of the economic analysis because it measures the economic effectiveness

of each production strategy implemented by rural households in the study sites. The

following analysis not only compares the economic viability of different types of

activities, it also identifies which land tenure type is most effective in yielding economic

welfare. The results of the profit analysis will also allow us to link profitability to land

cover dynamics among land tenure categories.

Table 4-7 compares the profitability of crops, livestock, and timber and non-timber

forest products among the three tenure types. I calculated profit as gross income minus









total costs for a given activity for each tenure type. Profit thus indicates the monetary

gain or loss to households as a result of their economic decision-making in the context of

their land tenure category. Negative values indicate lost revenue in 2004, whereas

positive values constitute profits accumulated by households. F-tests indicate that

differences across sites were significant.

Table 4-7. Annual mean for profit for economic categories, by land tenure type, Acre,
Brazil, 2004.
All Chico Mendes Porto Dias Peixoto PC
Economic activity F-tests
(n=88) Resex (N=34) PAE (N=27) (N=27)

Crops (US$) -125.14 -219.87 -17.26 -102.17 12.383**
Livestock (US$) -96.45 -132.48 289.96 -213.31 3.597*
Non-timber (US$) -10.22 -33.36 255.64 -41.70 4.355*
Timber (US$) -870.51 --917.21 140.56 4.250+
Total (US$) -86.71 -122.54 56.52 71.83 12.729**

+p <.10; p<.05; ** p<.01

The results indicate negative incomes for all economic activities when considering

the three study sites as a whole. The total mean profit value was US$ 86.71, which

suggests that overall the households sampled incurred an economic loss in 2004. While

crops were the category of product with the weakest economic efficiency (-US$ 125.14),

non-timber forest products almost covered the costs of production (-US$ 10.22). Timber,

however, was the product with largest economic loss. These results indicate that

agricultural crops and timber emerge as the products with the lowest economic viability,

while non-timber forest products, although also showing a loss, nearly covered the costs

of production.

Nevertheless, when we consider profitability among the three study sites, the

Endings are more mixed. In examining the results by study area, the Peixoto PC and

Porto Dias exhibit net economic gains overall. Their total mean profit value suggests









some economic viability (US$ 71.83 and US$ 56.52 respectively). That said, the

activities that proved profitable varied among land tenure categories. While in Peixoto,

timber was the only profitable product, in Porto Dias, livestock and non-timber forest

products were the profitable activities. Crops and livestock were the categories with the

poorest economic performance among colonists, and livestock production was the second

least effective production within all the sites. Soil degradation may be a main factor

contributing to increased costs of these activities in this area. The excellent profitability

for timber extraction (US$ 140.56) in Peixoto in comparison to other activities is

explained by its low expenses, in which the costs are reduced via the frequent

involvement of local logging companies that pay for a certain amount of timber and

provide their skidder to harvest and transport the timber.

In contrast, the Chico Mendes Resex has the lowest economic viability in terms of

profitability. In great part, this poor economic performance is due to agricultural crop

production, which yielded considerable financial losses (-US$ 219.87). This might be

explained by the very high costs for forest clearing to establish swidden agriculture.

Rubber tappers still utilize very simple technologies in the form of machetes for clearing

forests, resulting in very high labor costs. Second, the costs of transport are an important

component of the total costs among rubber tappers, who on average reside in colocagdes

around 30 km from the city of Xapuri.

Production activities in Porto Dias PAE were also profitable overall. Although its

mean profit was inferior to that of the Peixoto PC, livestock and non-timber forest

products demonstrated their economic viability for rubber tappers in this area. However,

the total costs of timber management, as presented earlier, are so high that they greatly









exceed the total gross income from timber. Therefore timber management in this study

site compromises the overall economic success of these families.

Three groups of factors help explain this performance for Porto Dias. First,

although the average price per head of cattle (US$ 150.00) in this area was similar to the

price offered to colonist farmers of Peixoto, the rubber tappers of Porto Dias have not had

the problems with soil degradation found in Peixoto, thus reducing total costs. Second,

for non-timber forest products, rubber tappers in Porto Dias concentrated their efforts on

Brazil nuts, copaiba, and forest seeds, which are products with relatively high prices and

significantly lower production costs than rubber. Third, as pointed out earlier, the type of

timber forest management implemented in Porto Dias, which follows an entrepreneurial

management model, has resulted in high operational and implementation costs.

This section concludes my economic analysis in which I first compared gross

income across land tenure types and land uses. I also estimated how fixed, variable, and

total costs were distributed among products and across communities. Finally, I integrated

cost and income data into a profit analysis to evaluate the economic viability of each

productive activity across sites. While public polices favoring extrativist, agro-

extractivist, and colonist households have focused on intervention strategies to generate

cash income, the results of this study indicate that the high costs associated with many

activities severely limit their economic viability. Not only has the timber forest

management proj ect in Porto Dias shown inferior results due to high costs, but the

proj ect-driven agrosilvopastoral system in Peixoto has not yielded favorable economic

outcomes for colonists. In this context, non-timber forest products do marginally better,

almost reaching break-even, however, these activities are also unprofitable.




Full Text

PAGE 1

EFFECTIVENESS OF EXTRACTIVE RESERVES, AGRO-EXTRACTIVE RESERVES AND COLONIST SETTLEMENTS IN SOUTHWESTERN AMAZONIA: AN ECONOMIC AND LAND COVER COMPARISON OF THREE LAND TENURE TYPES IN ACRE, BRAZIL By FRANCISCO KENNEDY ARAJO DE SOUZA A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Francisco Kennedy Arajo de Souza

PAGE 3

To: my Apurinan indigenous people, my deceased grandmother Nadira Valuah, my mother Mada, and my daughter Yara.

PAGE 4

iv ACKNOWLEDGMENTS This thesis was the result of many suppor ters, co-workers, friends, and community partnerships. First, I would lik e to express my sincere gratitu de to the rubber tapper and colonist families with whom I have collabora tively worked over the past eight years in the State of Acre, Brazil. Drivers, researcher s, and friends at the Federal University of Acre, Centro dos Trabalhadores da Amazni a (CTA), Group for Agroforestry System Research (PESACRE), and Acre government we re fundamental to generate the results presented in this thesis. Especially I tha nk my research assistants, Tatiana, Mrcio, Ernilton, Luciano, and Dermerson. I am also grateful for the financial suppor t provided by many organizations. First, I thank the Ford Foundation International Fellow ship Program for its funding my graduate studies at University of Florida. In Br azil Maria Luiza and Marcia, at the Fundao Carlos Chagas, were fundamental in suppor ting my first steps in the USA. I thank Gregory Marino for his institutional support at th e Institute of International Education in New York. I sincerely express my appreciati on to the Federal University of Acre for permitting me to pursue my graduate study. I sincerely thank the Tropical Conservation Development Program for its financial support du ring my last semester at the University of Florida. I also extend sp ecial thanks to the Woods Hole Research Center for supporting my fieldwork in Acre in 2004. Two government agencies in Acre, SEPROAF and IMAC, contributed with transport and Landsat images.

PAGE 5

v My supervisory committee was fundamental in my effort to complete this graduate course. First, I sincerely express my gratit ude to my chair, Dr. Stephen Perz, for his constant support, encouragement, patience, and friendship. I thank Dr. Charles Wood for understanding my English limitation during my first semester at UF. Moreover Dr. Wood was essential in advancing my statistical da ta analyses. I sincerel y thank Dr. Marianne Schmink for her friendship, patience, encour agement, and commitment in collaborating with the Amazonian people. I am especially indebted to Dr. Schmink for her support in one of the most critical moments of my life. Innumerable friends and co-workers suppor ted me before and during my studies, fieldwork, and the writing pro cess of my thesis. Valrio has been a unique friend who since my first steps in Gain esville has supported me with friendship, encouragement, ideas, and professional experience. Geraldo, like Marianne, became my kariu brother and sister after their important support during my cr itical phase in Gainesville. I thank Foster Brown, Daniel Nepstad, and Eduardo Br ondizio for their letters supporting my application to the Ford Foundation. I am gr ateful to Richard Wallace, Christiane Ehringhaus, Christie Klimas, Joanna Tuck er, Valrio, and Jeff for their support in reading, editing, and suggesting ideas in my th esis. I express my thanks to Wendy Lin for her support with my papers. I thank ma ny other friends: Diogo, Rodrigo, Charle Crisostomo, Silvia Brilhante, Nivia, Pedr o, Magna, Roger, Savio Maia, Edson Carvalho, Lira, Hulk, Marilene, Sumaia, Fadel, Ronei, Elaine, Andrea, and Leonardo. Finally, I express my gratitude to my pare nts and my brother and sisters for their examples of vigor and humanity. I am indebt ed to Yara, my daughter, for putting up with my long time far away from her.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES.............................................................................................................x ABSTRACT....................................................................................................................... xi 1 INTRODUCTION........................................................................................................1 Research Objectives......................................................................................................2 Overview of the Research Sites....................................................................................3 Theoretical Framework.................................................................................................7 Methodology Summary..............................................................................................10 Study Significance: Local Research Needs on Land Tenure Analyses......................11 Structure of the Thesis................................................................................................12 2 LAND USE MANAGEMENT AND LAND TENURE DIVERSITIES: ECONOMIC BOOMS, State POLICI ES, AND LAND RIGHTS IN ACRE............15 Introduction.................................................................................................................15 Land Tenure Regimes in the Amazon Region............................................................16 The State of Acre: History, Social M ovements, and Land Tenure Diversity.............19 Geographic Context.............................................................................................19 Acre: Before the Ranchers...................................................................................21 Agriculture Modernization: Large Ca ttle Ranchers, Rubber Tappers, and Colonists..........................................................................................................22 After the 1990s: No Economic Boom but a Rubber Tapper Victory.................23 Acre: Social Movements, Community Land Tenures, and Forest-Based Development Policies............................................................................................25 Neoextrativismo : A Long-Term and Systemic Approach..................................27 Florestania : Citizenship Based on Regiona l Integration, Short-Term Economic Growth, and Conservation..............................................................29 Community Land Tenure: An Emergent Issue for Assessment in Acre....................31 Colonization Settlements (PC)............................................................................33 Extractive Reserves (Resex)................................................................................35 Agro-extractive Settlements (PAE).....................................................................36 The Study Sites...........................................................................................................37 Peixoto Colonization Settlement: The Agrosilvopastoral Model........................37

PAGE 7

vii Porto Dias Agro-extractive Settleme nt: The Timber Forest Management Model...............................................................................................................40 Chico Mendes Extractive Reserve: The Non-timber Forest Management Model...............................................................................................................42 Conclusion..................................................................................................................45 3 METHODS AND TECNIQUES FOR ECONOMIC AND REMOTE SENSING ANALYSES...............................................................................................................47 Introduction.................................................................................................................47 The Socioeconomic Survey........................................................................................49 Household Sampling...........................................................................................50 The Questionnaire...............................................................................................51 The Household Fieldwork...................................................................................52 The Price Survey.................................................................................................52 Participatory Findings Assessment: Attemp ts for Calibration and Validation...53 Land Cover Data Collection and Fieldwork...............................................................54 The Post-Survey: Data Processing and analysis.........................................................55 Household Economic Approach..........................................................................55 Economic Analysis..............................................................................................58 Land Cover Data Processing...............................................................................59 Economic and Land Cover Integration................................................................62 Conclusion..................................................................................................................62 4 ECONOMIC AND LAND USE COMPARISON OF SMALLHOLDER LAND TENURE MODELS IN THE STATE OF ACRE......................................................64 Introduction.................................................................................................................64 Revising the Research Questions................................................................................65 Land Cover Distributi on among Land Tenures..........................................................67 Accuracy Assessment of Land Cover..................................................................67 Comparing Land Cover among Land Tenures....................................................68 Economic Analysis for Land Tenures in the State of Acre........................................74 Gross Income.......................................................................................................74 General Cost Analysis.........................................................................................78 Cost Analysis at Community Level.....................................................................82 Profitability Analysis...........................................................................................85 Integrating Land Cover and Economic An alyses for Understanding Land Tenure...89 Conclusion..................................................................................................................96 5 CONCLUSIONS: DILEMMAS, IMPLIC ATIONS, AND FUTURE RESEARCH..99 Summarizing Key Findings......................................................................................100 Policy Implications...................................................................................................102 Emergent Questions for Community Land Tenure in Acre......................................105

PAGE 8

viii APPENDIX A QUESTIONNAIRE USED IN THE CHICO MENDES EXTRACTIVE RESERVE.................................................................................................................107 B QUESTIONNAIRE USED IN THE PORTO DIAS AGRO-EXTRACTIVE SETTLEMENT.........................................................................................................122 C QUESTIONNAIRE USED IN THE PEIXOTO COLONIZATION SETTLEMENT.........................................................................................................137 D CIPEC TRAINING SAMPLE PROTOCOL............................................................151 LIST OF REFERENCES.................................................................................................155 BIOGRAPHICAL SKETCH...........................................................................................162

PAGE 9

ix LIST OF TABLES Table page 4-1. Accuracy assessment for 2004 supervised classification for the study sites in Acre, Brazil...............................................................................................................68 4-2: Land cover categories at household le vel, all sites (n=88), Acre, Brazil...................70 4-3. Descriptive statistics for land cover ca tegories, by land tenure type, Acre, Brazil, 2004........................................................................................................................... 72 4-4. Descriptive statistics for annual inco me, by land tenure type, Acre, Brazil, 2004.....75 4-5. Descriptive statistics for annual cost categories, all sites (n=88), Acre, Brazil, 2004........................................................................................................................... 79 4-6. Annual mean for cost categories, by land tenure type, Acre, Brazil. 2004................83 4-7. Annual mean for profit for economic categ ories, by land tenure type, Acre, Brazil, 2004........................................................................................................................... 86 4-8. Site model at land cover types, all sites (n=88), Acre, Brazil, 2004..........................91 4-9. Life cycle model at land cover types, all sites (n=88), Acre, Brazil, 2004................92 4-10. Market integration model at land cove r types, all sites (n=88), Acre, Brazil, 2004........................................................................................................................... 93 4-11. Investment model at land cover type s, all sites (n=88), Acre, Brazil, 2004.............94 4-12. Assets model at land cover types, all sites (n=88), Acre, Brazil, 2004....................96

PAGE 10

x LIST OF FIGURES Figure page 1-1. State of Acre Brazil, identifying col onizazation, agro-extractive and conservation areas.......................................................................................................................... ..4 1-2. Deforestation in the study sites, Acre, Brazil...............................................................6 3-1. Diagram of land cover processing in the study area, Acre, Brazil.............................60

PAGE 11

xi Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts EFFECTIVENESS OF EXTRACTIVE RESERVES, AGRO-EXTRACTIVE RESERVES AND COLONIST SETTLEMENTS IN SOUTHWESTERN AMAZONIA: AN ECONOMIC AND LAND COVER COMPARISON OF THREE LAND TENURE TYPES IN ACRE, BRAZIL By Francisco Kennedy Arajo de Souza May 2006 Chair: Stephen G. Perz Major Department: Latin American Studies Linking economic and remote sensing tec hniques, a comparison of an extractive reserve (Resex), an agro-extractive settlement (PAE), and a colonist settlement (PC) were conducted in the state of Acre, Brazil. Thes e land tenure categories were represented by Chico Mendes Resex, Porto Dias PAE, and Peix oto PC respectively. The datasets used in this research were constituted of three co mponents. First, socioeconomic data were gathered from 88 households in the th ree sites in 2004. Second, 442 GPS land cover samples were collected in 2004 and 2005. Third, I used ancillary data and Landsat TM 5, bands 1 through 5, scenes 67-001 and 67-002 for performing a supervised classification at community and household level within the three land tenures. Economic findings indicated that while fore st management was the more important activity among rubber tappers in the Resex and PAE, in Peixoto agrosilvopastoral strategies had little economic importance. However, high costs were related to all

PAGE 12

xii economic activities, especially timber fore st management. Moreover, non-timber forest products were more effective th an timber forest management. The land cover analysis indicated that extractive reserves and agro-extractive settlements are most effective to address c onservation. In comparison to PC lots that showed a mean of 36% of de forestation in 2004, selected colocaes in the Resex and PAE had a mean of 5% and 12% of deforest ation respectively. However, while pasture was the category that most contributed to de forestation in Peixoto and Porto Dias, crops were the most prominent cause of fore st clearing in the Chico Mendes Resex. An integrated analysis of economic and la nd cover data produced two main results. First, the multivariate models indicated that the establishment of extractive reserves and agro-extractive reserves may be an effici ent strategy to achieve conservation goals. However, this policy should be followed by st ate enforcement initia tives because the life cycle model addressed in this study suggested that more recent inhabitants tend to have a high propensity to deforest. This tendency ma y be increased if investment in roads is carried out. While this study is an initia l step in linking economic and land cover analysis at the household level, it may provide insights for the formulation of conservation policies in Acre where a comprehensive and innovative policy has been implemented since 1999. Core components of that policy include co mmunity forest management and sustainable land uses based on forest management and agrosilvopastoral pr oduction. Findings from this thesis will hopefully help establish new strategies for consolidating extractive reserves, agro-extractive reserves, and col onist settlements based on a new development model in Amazon.

PAGE 13

1 CHAPTER 1 INTRODUCTION Community land tenures and their land uses in Amazonia have been the object of intensive debate, with su pporters (e.g., Schwartzman 1989, Allegretti 1990, Anderson 1990) and opponents (e.g., Homma 1989, Browde r 1990). While public policies in this region have focused on regional development mo dels based on private control of large tracts of land (e.g., ranchers, miners, and loggers), community land regimes have been considered economically unattractive. By the same token, forest and sustainable land uses have been criticized as an impediment to ec onomic growth (Hecht 1985). In contrast with these perspectives, over the past seven years, the government in the Brazilian state of Acre has made efforts to consolidate a po licy based on community based land tenure and sustainable land use strategies. In this context, it has been crucial to unde rstand and empirically assess the role that timber-forest management, non-timber forest management, and agrosilvopastoral systems have in providing subsistence revenue a nd income accumulation, within extractivist, agro-extractivist, and colonist communities The potential effects of these land use strategies for household livelihood systems and their implications for land cover change are not well-understood. Kaimowitz and Angels en (1998) argue that trained surveyors and time-consuming data collection are a ma in constraint for farm-level analyses. However, focusing on this type of study, some scholars (Carpentier et al. 2000, Tomich et al. 1998) have observed that rising cas h-incomes among small landholders in Amazonia has generated cha nges in their behavior.

PAGE 14

2 The economic efficiency of sustainable in itiatives, then, is expected to promote changes in natural resource use. Among sma llholders in Acre I have observed that increased income has contributed to the substitution of subsistence production by manufactured products. Families have focused their interests on higher priced products to meet their new needs. As a result, this comm only increases pressure on the forest to meet these demands. Although a broad range of issues are rela ted to community land use and land cover change in Amazonia, a question that emerged as central issue of this thesis is: which community land tenure is most effective in balancing economic and forest conservation goals? Almost 30 years after the most intens e and violent phase of land conflicts in Amazonia, this region has an opportunity to merge national and state policies into a common agenda for addressi ng economic growth guided by environmental conservation. This provides a rare opportunity to dete rmine whether areas managed by communities, such as extractive reserves, agro-extractive settlements and colonization settlements, can be a potential development model for the Amazon region. Research Objectives The goal of this thesis is to respond to the need for an empirical assessment of community land tenures as economic and envi ronmental conservation strategies in the Brazilian Amazon. This study responds to the emergent demand to understand the effects non-timber and timber forest management as well as agrosilvopastoral strategies in extractive reserves, agro-extractive settlements and colonist settlements in the state of Acre, Brazil. To address this issue, I used data sets from three community land tenure categories located in this stat e. The research objectives were threefold: 1) to compare economic and environmental conservation in three distinct land tenure categories

PAGE 15

3 designed to achieve different policy objectives ; 2) to economically evaluate non-timber and timber forest products in terms of their co st and benefits; and 3) to identify factors most important for forest clearing in three distinct land tenure types. Overview of the Research Sites The state of Acre is located in southwestern Amazonia and its 153,000 km2 of area was occupied by over 550,000 people in 1999 (Governo do Estado do Acre 2000a). It borders the Brazilian states of Rondnia and Amazonas a nd the tri-nati onal frontier known since 1999 as the MAP region, which al so includes the Bolivian State of Pando and the Peruvian state of Madre de Dios (Fig ure 1-1). Its territory still has almost 90% intact rainforest which contains diverse human and ecological mosaics with complex interactions at local and regional levels (Governo do Estado do Acre 2000a, Silveira et al. 1997). However, there is the possibility that th is landscape will be dr astically altered over the next few years. A coalition of Brazilia n, Peruvian, and intern ational organizations such as Inter-American Development Bank (ID B) is in the process of constructing and paving the Pacific Highway which would be one of the most important roads for transportation of commodities such as soybean s in South America. It may also increase the probability of environmental degradation as a result of logging, soybean expansion, and migratory movements (Carvalho et al. 2002). The state of Acre, located between a degraded area (arc of deforestation) and a region with less deforestation (Andean mount ains), exhibits impor tant community land tenure diversity. Rubber tappers, indigenous pe ople, and colonists, who have been the priority social groups for governance polic ies implemented since 1999, control almost 44% of the territory (Figure 1-1) (Governo do Estado do Acre 2000a). The future amount

PAGE 16

4 of deforestation as well as potential negative effects of the Pacific Highway will be strongly determined by land use decisi ons taking place in these communities. Three study areas located in Ac reÂ’s eastern region were se lected for this research based on three conditions: 1) they repres ented a community land tenure category: extractive reserve, agro-extractive settlem ents, and colonist settlements; 2) their inhabitants are addressing su stainable land uses based on timber management, non-timber forest management, or agrosilpastoral initi atives; and 3) they reflect public policies initiatives for sustainable land use. As a re sult of these criteria three land tenure types were selected. The Chico Mendes Reserve with almost one million hectares was the largest extractive reserve in Brazil until 2004. It is sub-divided into 45 rubber estates ( seringais ) Figure 1-1. State of Acre Brazil, iden tifying colonizazation, agro-extractive and conservation areas. BR-317 Pacific Highway Agro-extractive SettlementsConservation areas Colonization Settlements A MAZONAS ROND NIA Madre de Dios (PERU) Pando (BOLIVIA)

PAGE 17

5 and each has individual rubber tapper landholdings ( colocaes ). Three seringais, So Pedro, Palmari, and Floresta (Figure 1-2, A), we re selected. In this area, a coalition of organizations led by state government agencies have sought to amplify the range of nontimber forest products explored by local resi dents over the past 7 years. The National Center for the Sustainable Development of Traditional Populations (CNPT), a bureau under supervision of the Brazilian Institute for Environmental and Renewable Natural Resources (IBAMA), is the governmental agen cy responsible for Re sexÂ’ administration in collaboration with rubb er tapper associations. Although the Porto Dias Agro-extractive Se ttlementÂ’s 22,000 hectares are occupied predominantly by rubber tapper families, it is administrated by the Brazilian Agency for Agrarian Reform (INCRA). There is no evident geographic division of seringais in this area. Rubber tapper landholdings are grouped in to three associati ons. The Porto Dias Rubber Tapper Association area (Figure 1-2, B) was selected for sampling. This area is one of the most important initiatives for timber management in Acre and has been supported by a pool of organizations led by th e Center for Amazonian Workers (CTA) to consolidate this activity as an ec onomic alternative for rubber tappers. The third area is the Peixoto Colonization Settlement, the largest and oldest human settlement area established by INCRA at the end of the 1970s (Figure 1-2, C). As with other colonist areas in the Amazonia, ag riculture and cattle are the main land use strategies in Peixoto. However, innovative experiments emerged at the end of the 1990s to address environmental de gradation and economic ineffi ciency. Since 1994, the Group for Agroforestry System Research (PESAC RE), a local NGO, has collaborated with members of the Grupo Novo Ideal rural association, a commun ity organization located in

PAGE 18

6 Peixoto, to combine agriculture, livestock, agro forestry, and forest management into an integrated agrosilvopastoral system. This co mmunity is the colonist sample for this thesis. Figure 1-2. Deforestation in th e study sites, Acre, Brazil. Deforestation 2004 Water Selected area Land tenure limite (A) Chico Mendes Resex (B) Porto Dias PAE (C) Peixoto PC (A) (B) (C) A C D

PAGE 19

7 Theoretical Framework The theoretical foundation of this work is guided by three bodies of literature. I draw on political ecology as a framework to integrate land reform in the Amazonia region and community land uses strategies as a dyna mic process taking place at multiple scales. This study seeks to understand local land use practices in the Amazon region as a result of events established in larger, extra-local political and economic se ttings (Peet and Watts 1996, Bryant and Bailey 1997). Linking politic al, economic, and cultural perspectives, this framework attempts to address a range of factors affecting use of natural resources as related to local, national, a nd global political economies and ecosystems (Blaikie and Brookfield 1987, Schmink and Wood 1987). Economic booms in Amazonia were a key aspect to take into consideration in this pe rspective. As a result of international demand, national and regional policies encouraged hum an migration, landownership regimes, and environmental depletion. These processes ar e examples of how underlying factors and political economies at multiple levels may affect socioeconomic and political dynamics at the community level in Amazonian region. Political ecology also constitutes a powerful framework for integrating natural and social dynamics (Peterson 2000). This cont ext is fundamental for understanding how community land tenure regimes in Amazonia were a result of underlyi ng social, cultural, environmental, economic, and political factors fr om a local to global s cale. In the case of Acre under the Forest Government, political processes yielded land tenure categories specially designed to produce more favorab le economic as well as environmental outcomes via more sustainable land use pr actices. Consequently, political ecology provides a basis for understanding the historic al context and political foundations for the land tenure categories on which I focus in this thesis.

PAGE 20

8 However, to empirically assess economic and conservation effectiveness of land use strategies in different communities, I decided to add two other bodies of literature: household economic analysis of land use ma nagement via a livelihoods approach; and explanations at various scal es for land cover change. Economic growth based on sustainable la nd use has commonly been held up as a blueprint for both conservation and human we llbeing. However, this research accepts Wollenberg et al. (2001)Â’s view that those income-g enerating activities may alter the social context, conservation behavior, and livelihood of local comm unities. This study attempts to comprehend rural smallholders as a complex of divers e livelihood activities (Ellis 2000). To explore this perspective for the land tenure categorie s selected in this work, two levels of analysis were conducted. First, the sustainable land uses practiced in the selected communities, have been under debate for over 30 years. Researchers of land use and forest economics in the 1980s and 1990s (Godoy et al. 1993, Peters et al. 1989, Pinedo-Vasquez et al. 1990, Nepstad and Schwartzman 1992, Padoch and Jong 1989) identified potential challenges and opportuni ties for these strategies. Second, in addition to their institutional, socio-cultural, and ec onomic diversities, inha bitants of extractive reserves, agro-extractive settlements, and col onist settlements are ty pified by the diversity of their livelihood systems. To achieve an economic and land use analysis, understanding the livelihood of each land tenure category emer ged as a key component to evaluate the effectiveness of community-based development in Amazonia. Although household farming, as argued by Elli s (2000), attempts to diversify the range of family activities as a strategy for maintaining their wellbeing, in the Amazon this decision may be also influenced by cultural behavior (Rego 1999). Household

PAGE 21

9 diversity, in this context, is problematic for empirical economic and social analysis. Hildebrand and Schmink (2004) ar gue that economic, social, and institutiona l factors in each livelihood system affect household economics, which have multiples consequences for land use allocation. This study uses livelihood analysis to understand how land cover and economic dynamics are affected by different land use re gimes. Based on differences in livelihood strategies among land tenure types, the in struments utilized during the fieldwork (questionnaire, participatory mapping, particip ant observation etc) we re adapted to each community. The economic findings discussed in th is thesis were summ arized in terms of inputs and outputs; however, for visualizing the household diversity in their livelihood systems, those findings were aggregated by ca tegories of land use (agriculture, livestock, forest management) that represent spatial la nd allocation (agriculture area, pasture, and forest), and are the result of allocation of available re sources (labor, resources, and capital). The third body of literature on which I draw is that addressing the causes and consequences of land use and land cover changes at multiple scales. Theories focusing on Amazonian deforestation have been dominate d by neoclassical economics and political ecology analyses; other studies have pointed to a combina tion of proximate causes and underlying drivers in a geographical and hist orical context (Geist and Lambin 2002). Yet, at the local level, i ndividual and social responses to land cover dynamics are affected by economic change led by institutional factors, in which new land use strategies are created by market and policies increasi ngly affected by regional and global factors (Lambin et al. 2001). Land cover and land use change, while affected by global and

PAGE 22

10 regional drivers, is particularly influen ced by individual micro-economic decision taking place at community level (Anselin 2001). Mertens et al. (2000) argued that integration of household survey and remote sensing information are an appropriate strategy to empirically understand the linkages and feedback s occurring at micro and regional levels. Methodology Summary This research involved multidisciplinary inqu iry at the three site s and multiple data collection methodologies including ground trut hing for classification of land cover classes, and quantitative analyses linking economic outcomes and forest cover dynamics at household level. Principal research me thods included semi-structured interviews, participant observation, vegetation samp ling, and remote sensing analysis. The economic survey at household leve l was carried out in 2004. I conducted interviews with 88 households in three land te nures located in AcreÂ’s southeast region. In order to evaluate the economic and land cover effects of sustainable land use strategies implemented by families, such as forest management and agroforestry systems, the sample was distributed as follows. In the Chico Mendes Resex I selected 34 families involved in management of various non timber forest products. In the Porto Dias agroextractive settlement the sample was 27 house holds. Peixoto colonization settlement was the third area, and had 27 small farmers selected. The set of interviews with those families was conducted from May through August of 2004 in collaboration with CTA, PESACRE, and governmental agencies. The land cover sampling took place during two time periods. In 2004 when I carried out the household surveys, I also collected 345 land cover GPS points in the selected household areas with f our research assistants. Ina ccuracy and confusion of the classification results in bamboo areas requi red additional ground-truthing in the summer

PAGE 23

11 of 2005 during which 97 additional land cover sa mples were gathered. These data were then utilized to perform a s upervised classification using Landsat images scenes 67-001 and 67-002 of 2004. Study Significance: Local Research Needs on Land Tenure Analyses This research is important for four main reasons. First, it provides a comparison of economic and land cover outcomes across multip le land tenure categor ies. Second, it will be crucial for implementation of policies in Amazonia, specifically in Acre where community land regimes are a foundation for alternative development based on sustainable land uses. Third, although land use and land cover change literature in the 1980-1990s focused on Amazonia, there is a lack of comparisons assessing the economic and conservation effectiveness of community initiatives based on timber and non-timber forest management, and agrosilvopastoral syst ems. Fourth, this study will contribute to the land use and land cover change literature by linking socioeconomic and satellite data analysis at household level. The first implication may be the most im portant for local stakeholders in Acre. Although the state government has been guide d by efforts to promote forest based development since 1999, it has al tered its policy approach over time. Policymakers have concentrated their efforts on establishing ec onomic alternatives for community areas of the Rio Acre watershed, a region with social, cultural, and ecological specificities, in which this study was conducted. In this terr itory, the governmental initiatives have addressed a combination of timber and nontimber forest management, agroforestry systems, as well as agriculture and cattle ranc hing zones as a strategy to increase the local welfare while emphasizing environmental c onservation. Over the next years, these

PAGE 24

12 community models will likely be replicated in other areas of the state (i.e. the Juru and Purus regions) which accounted for over 65% of the intact forests in the State in 1999. However, there is a lack of empirical studies addressing community land use. During various moments in Acre, policymaker s and social movements have called for research programs assessing these emergent issues. Cross comparison among community land use strategies should show whether th e forest-based initiatives addressed are effectively balancing economic growth and c onservation goals. Specifically this type of research will demonstrate the land cover a nd economic effects of state policy planning within extractive reserves, agro-extractive settlements, and colonist settlements. Structure of the Thesis This thesis is organized in five chapte rs. Chapter 2 gives an overview of the history, context, and status of commun ity-based development in Amazonia and specifically in the state of Acre. It intr oduces how economic booms, human migration, economic processes, and policy drivers aff ected land ownership and resource use at a regional scale. The chapter attempts to de monstrate that while community land tenure regimes were a result of regional and nationa l drivers, they were also produced by political and economic events conducted at global scale. Based on a political ecology perspective, this chapter argues that na tural resource appropri ation in Amazonia, including land, has been determined by social confrontations and alliances at multiple levels. While the 1970s included the most cr ucial rupture in the Amazonian economic system, during which a jump from forest-based development to an agricultural economy took place, this period was also notable fo r the establishment of innovative land reform led by rubber tapper movements. In the 1980s and 1990s new land tenure categories emerged as a result of social conflicts, economic changes, and policy implementation.

PAGE 25

13 They were also characterized by their institut ional and social differe nces, as well as the remarkable cultural distinction of their inhabitants. Chapter 2 also focuses on the study area a nd discusses why the State of Acre may be considered a space of rare opportunity to assess the effectiveness of community land tenure strategies based on sustainable land us e initiatives. The forest government efforts to build up an innovative regional model of public policy are also discussed in this chapter. The major objective of this chapter is to clarify the multiple land use strategies conducted by rural communities in this region and how their land rights were established over time. The historical, cultural, an d socioeconomic difference among these populations is addressed via a comparativ e analysis. A brief background of the study areas, which include the Chico Mendes Ex tractive Reserve, the Porto Dias Agroextractive Reserve, and the Peixoto Co lonization Settlement, is provided. Chapter 3 concentrates on the research design, field surveys, and remote sensing techniques as well as the economic framework ut ilized to perform the data analysis. This discussion addresses three key aspects. First, it shows how remote sensing and household surveys are combined for economic analysis of resource management, particularly land cover change. Second, it describes the resear ch design used in this study, in terms of economic and land cover comparisons between the three land tenure cat egories. Last but not least, it emphasizes the research innovati on achieved in this study by using satellite and socioeconomic data at a household leve l to compare multiple land tenure types in terms of their land uses. Additionally, Chapter 3 disc usses how the economic and land cover fieldwork was carried out. This chapter a) summarizes how the communities and households were

PAGE 26

14 selected; b) discusses the spatial and so cioeconomic techniques undertaken through the survey; and c) reviews the methodological appr oach for the economic and land cover data analyses. This is followed by a discussion of the participatory tools used, prior to economic analysis, to define indicat ors, field techniques, and sampling. Chapters 4 and 5 detail the results of this study and compare the three communitybased strategies. The economic outcomes at a household level are linked to their associated costs, and I evaluate their envi ronmental performance for addressing forestconservation. The analysis is divided into th ree parts. The first di scusses the land cover distribution across the three land tenure categories. The second focuses on economic analysis of costs and revenue to identify the most prof itable land tenur e and livelihood activity. Finally, the third part presents five multivariate regression models to evaluate the effects of economic and non economic variables on deforest ation among smallholders of the study sites.

PAGE 27

15 CHAPTER 2 LAND USE MANAGEMENT AND LAND TENURE DIVERSITIES: ECONOMIC BOOMS, STATE POLICIES, AN D LAND RIGHTS IN ACRE Introduction The purpose of this chapter is to summarize different land-tenure types in the State of Acre and differentiate the land use stra tegies associated with each type. This discussion starts by examining the larger context in which land tenure regimes developed in the Amazon region, in particular, how th e economic booms affected human migratory movement, livelihoods of rubber tapper populations, and by extension, land rights. Additionally, it introduces a few details of th e policy-land tenure regime interface that took place during three major landmark periods in AcreÂ’s history: the rubber booms (1870-1950), the era of cattle expansion (1970-19 80s), and the post-victory of the rubber tappersÂ’ movement (1990s to today). This di scussion introduces the main categories of land tenure in Acre and point s out central differences in their attempts to address economic and conservation goals. An overview of the study sites and description of their significance will also be provided. In contrast with many other studies in which an economic-environmental tradeoff analysis is undertaken, this chapter emphasizes why public policies are a key driver for land cover and land use change in Amazonia via a comparison of distinct land tenure types. This chapter concentrates on forest policy approaches impleme nted in Acre since 1999 in which rubber tappers and rural commun ities have been a key social actor. In addition, this chapter introduces the overall discussion of the circumstances in which

PAGE 28

16 different land tenure categorie s and resource management st rategies were established. Pointing out diverse contexts on how land te nure types emerged in Acre and their remarkable dissimilarities, I argue that extrac tive reserves, agro-extractive settlement, and colonist settlement projects played an im portant role in achieving an alternative development model. To begin the comparative analysis of differe nt land tenure regimes in this state, this chapter offers an overview of the historical and regional context in which land tenure regimes are situated. The discussion is orga nized as follows. First, a summary of the economic boom periods that took place in Amaznia will be provided. Second, the historical and geographical context in which community tenures emerged in the state of Acre will be identified. This discussion focu ses on two issues: 1) the struggle for land reform from the late 1970s to early 1990s; and 2) the gove rnment policies implemented in Acre since 1999 related to land tenur e communities. Third, differences among extractive reserves, agro-extractive settlements, and colonist settlements will be discussed. This section will also include a description of the three study sites, each representing one of th ese land tenure types. Land Tenure Regimes in the Amazon Region Land rights and land tenure in the Amazon have historically been influenced by complex connections among agents at global, national, and local s cales. Simmons (2004) emphasizes that while these dynamics have been defined by multiple agents, such as indigenous people, miners, loggers, ranchers and small farmers, a political-economic framework for land-tenure in the Amazon requires an understa nding of past processes. As a part of this discussion, this section offers an historical review of development policies of Amazonia and their implications for land distribution and land tenure implementation.

PAGE 29

17 Specifically it focuses on the cycles of ec onomic booms that took place in this region from the 19th century to the late 1980s. Three hist orical moments illustrate the interface between economic events and land tenure regimes. The first phase, from 1870s though the 1950s, was the rubber boom s period which took place in Amazonia in the middle 19th century and again in the 1940s and early 1950s. Demand from pneumatic tire factories in Europe and the U.S. motivated the Brazilian government to move thousands of rural workers from the northeast region to the north of Brazil to meet labor demand. The arrival of northeastern people in Acre was an important starting point for establishment of the local society. Rubber tappers emerged in this period as a social group forced to de pend on forest resources for their subsistence. The rubber booms, however, also fostered crucial socioeconomi c changes in this region. Thousands of indigenous people died or were expelled from their land by large rubber estate owners ( seringalistas ), the new landowners in the Amazon region (Melatti 1993). While during the first rubber boom (1870-1910) the economy was based on the semi-slavery aviamento system1 (Weinstein 1983), duri ng the second boom (1940-1950) the state assumed economic and social contro l of the rubber-based economy (Martinello 1988). During a brief rubber boom in the Second World War when Malaysian areas were controlled by the Japanese military, blocking AmericaÂ’s access to rubber, the category of autonomous rubber tappers2 became more evident (Oliveira 1985, Bakx 1988). In spite its 1 The semi-slavery aviamento system was a coalition of seringalistas Brazilian state, and national and international companies to implement the rubber economy in the 19th century in the Amazon region. 2 Autonomous rubber tappers refer to circumstance of relative freedom of rubber tappers in managing their colocaes This situation was launched in the late 1910s, when external interference over the resources in the colocaes from state or seringalista became reduced. As a result, agriculture and livestock emerged as new activities worked by rubber tappers for their subsistence.

PAGE 30

18 shortness, the four year s (1942-1945) of rubber demand stimulated some economic recovery. A new migratory movement3 was then stimulated by the Brazilian government to Amazonia, particularly to its Western region, the area with th e largest population of rubber trees. The second phase, from the 1960s through 1970s came with the decline of international demand for Amazonian rubber. The forest economy, which had been represented by the rubber tree, was substituted by a new development model based on cattle and crops. This economic change was fo llowed by large-scale state investments in infrastructure such as roads, dams, and mi nes, and subsidy progr ams to modernize the Amazonian economy (Cardoso and Mller 1977). On the other hand, at a micro-scale, the land tenure regimes were also transformed from seringais to large cattle ranches. At the same time, rubber tappers engage d in defending their landholdings ( colocaes ), even as some rubber tappers migrated to urban areas. The fall in the price of rubber re sulted in the bankruptcy of many seringalistas and contributed to important ch anges in the rubber tapper Â’ s livelihood system. As a result of more flexible social control, hundreds of rubber tappers became more independent and were able to introduce new econom ic activities into their land use strategies such as slash and burn agriculture, livestock, hunting, and harv esting of forest products. It is notable, then, that the rubber tappersÂ’ relative free dom fundamentally altered their level of 3 This strategy was addressed by Br azilian government to face the lack of a labor force to exploit rubber trees. State advertising was carried out to stimulate a new migratory movement of people from the Northwest, this time called rubber soldiers ( soldados da borracha ), to the Amazon. Oliveira (1985) noted that over 50,000 migrants from that poor Brazilian region were engaged on the rubber economy in AcreÂ’s jungle. Yet, as pointed out by Sobrinho (1992), different from previous cycles, the rubber economic system was controlled by the state which monopolized the co mmercialization process and the control of rubber tappers in the seringais

PAGE 31

19 subsistence, even as they continued to focus on rubber as their main cash income activity (Oliveira 1985). These changes prompted rubber tappers to mobilize to defend their land rights Consciousness-raising, facilita ted by worker unions, was a key element influencing the land struggle that took place through an economic transition from forest-based land use to cattle ranching launched in the 1960s. In th is context, the st ruggle for land of during this period was characterized by social coalitions between workers and regional organizations in order to pressure the State for land reform in the region. Land tenure policies implemented during th is period were complicated by the governmental plan to integrate the Amazon by roads to other Brazili an territories. Human migration was encouraged though colonization settlements which were established in areas surrounding regional highw ays, such as Transamaznica, Cuiaba-Santarem, BelmBraslia, and Rio Branco-Braslia. As a result land conflicts between rubber tappers, colonists, and large cattle ranchers became mo re visible in Amazonia. However, intensity of land conflicts varied among the Amazonian st ates as a consequenc e of local policies, social coalitions, and the influence of nati onal policies. In the following section I will focus on the case of Acre. In particular, I wi ll concentrate on the importance of social movement, led by rubber tappers to foster land reforms. The State of Acre: History, Social M ovements, and Land Tenure Diversity Geographic Context AcreÂ’s 153,000 km2 is occupied by over 550,000 people (Governo do Estado do Acre 2000a) and is bordered by the Brazilian states of Rondnia and Amazonas and by a trans-national frontier constituted by the Bolivian state of Pando, and the Peruvian state of Madre de Dios (Figure 1-1). Its territory is still constituted by almost 90% intact

PAGE 32

20 rainforest with diverse human and ecological mosaics with complex interactions at local and regional levels. Although this state has been cited as a rich region of palms (Daly and Silveira 2001) and important diversity of other species of plan ts, habitats, and soil (Governo do Estado do Acre 2000b), due to the e normous lack of scientific knowledge of its ecology, the degree of its biodiversity is likely underestimated (MMA 2001a, Silveira et. al 1997). Besides its biological divers ity, approximately 44% of its territory is occupied by smallholders such as extrac tivist inhabitants, indigenous people, and colo nists (Governo do Estado do Acre 2000a, Figure 1-2) who have implemented a range of land use strategies such as timber and non-timber fo rest management, annual and perennial crops, and cattle and other livestock. In 1996, less than 30% of Acre ’s total deforestation was located within community areas.4 Therefore, Acre’s future conservation will be affected by land use decisions of these communities. Community land tenure regimes in Acre were the result of intensive social conflicts that took place in the past. Social movement s and rubber tapper strategies, such as the “ empate movements5 were fundamental to influence la nd reforms in this State. These rural workers assisted by various organi zations from local to international level contributed to making this re gion renowned worldwide in the 1980s. In order to provide 4 This assessment was performed using data available through the Macap Seminar, a meeting organized by the Brazilian government in 1999, in which Brazilian ho tspots for conservation and sustainable uses were identified. I utilized ArcView to process raster and vector files and Erdas Imagine to merge raster and vector files. I then calculated deforestation within each land tenure type (see MMA 2001 for details about the data). 5 The empate movements were a form of collective resist ance conducted by rubber tappers against deforestation. The name was conceived based on the Portuguese word “ empatar ” (to prevent). These actions were usually carried out in the summer season, the moment in which large cattle ranchers cut the forest. During these initiatives rubber tapper leaders fi rstly tried to convince farm laborers to desist cutting down the forest. If they were not su ccessful to stop the deforestation, th en they organized themselves in line (children, women, and men) to protect the forest.

PAGE 33

21 an overall context in which t hose community land tenures we re created, in the following section I move to AcreÂ’s history. Specifi cally I will focus on how social movements emerged and demanded land reform and the establishment of new categories of land tenure types such as extractive reserves (Res ex) and agro-extractive settlements (PAE). This discussion will be divided into three parts. First, I summarize the period prior to cattle raising in the 1970s. S econd, I discuss the period of la rge-scale cattle ranching and land conflicts from the 1970s to the 1980s. Thir d, I focus on the rubber tapper victory in the 1990s when extractive reserves and agro -extractives settlements were created. Acre: Before the Ranchers The end of the rubber boom in 1945 contri buted to the collapse of the Acrean economy, and also resulted in the emergence of new and important soci al actors. Oliveira (1985) and ELI (1994) poi nt out that the crisis of rubbe r had two related consequences for land property regimes in Acre. On the one hand, some of the seringalistas went bankrupt and abandoned their rubber estates. On the other hand, rubber tappers reacted differently to this collapse, though in diverse ways. One group of rubber tappers, like some seringalistas, grew discouraged with rubber tapping and migrated to urban areas in Acre or to other Brazilian states. Another large number of rubber tappers remained under the seringalistas Â’ oppressive control that used jagunos (mercenaries) to prohibit agriculture and other subsistence activities in the colocaes (landholdings). Within these areas, the n, no change was observed in terms of social control, landownership, and rubber tappersÂ’ livelihood. And third, a more significant transformation, which has also been noted by Bakx (1988), took place among rubber tappers, who, as a consequence of the abandonment of rubber estates by

PAGE 34

22 seringalistas were able to manage their land a nd their resources according to their particular interests. These socioeconomic changes took place among the third group of rubber tappers contributed to the establishment of new land use strategies within seringais while also providing the social context for the emerge nce of a new category of rubber tapper. Conscious of their land rights, for the first time since their arrival in this territory, they organized themselves against ranchers, the stat e, and the judicial system to demand rights to their lands. This period was an important landmark in Acre’s history in which an economic rupture emerged while also bringing intensive land conflicts. Agriculture Modernization: Large Cattle Ra nchers, Rubber Tappers, and Colonists The period from the 1970s through the 1990s was the most crucial phase of Acre’s history in terms of social c onflicts, land tenure regimes, polic ies, and economic strategies. These striking social and political changes and their implications for development and land use have been noted by various schol ars (e.g., Oliveira 1985, Esteves 1999, and Paula 2003) as the moment of substitution of forestry harvesting by agriculture and pasture economies. This time period was ma rked by the polarization between forest resource use by rubber tapper communities a nd traditional models based on crops and cattle expansion. The crops and cattle economic model implem ented in Acre in the early 1970s was led by the Wanderley Dantas. At that moment the governmental slogan was to “produce in Acre, invest in Acre, and export via the Pa cific” (Silva 1990) which meant to integrate this state into the emergent Brazilian ag ribusiness economy. Based on cattle, a ranching coalition constituted by local and national polit icians, the board of judges, state police,

PAGE 35

23 and large ranchers emerged to establish a new land tenure policy to benefit large landowners.6 In terms of the ownership configuration of Acre’s territory, this period represented a transition from seringais to large scale cattle farms. In addition, this phase was characterised by social polarization between traditional populations (rubber tappers and indigenous people) and large cattle ranchers. The struggle became more intense when the “ paulistas ”, which included ranchers from the South, Central and Southwest of Brazil, arrived in Acrean territory and encount ered a large resident population of posseiros, or autonomous seringueiros, that acquired their rights over the rubber estates when they were abandoned by seringalistas. The “ empate movement” was the main strategy utilized by rubber tappers to prevent deforestation as well as to demand their land rights. After the 1990s: No Economic Boom, but a Rubber Tapper Victory The transition from forest-based prod uction to a cattle ranching and land speculation economy was followed by a violen t struggle within Acre. Sobrinho (1992) points out that from 1975 thr ough 1988 in Brasilia and Xapuri7 400 rubber tappers were imprisoned and 40 others killed or tortured. Both rubber tappers a nd ranchers organized themselves as social movements. In mid1985, large cattle ranchers created the ‘Rural Democratic Union (UDR)’ with two main goals : 1) to influence national policies in the interests of agri-business, for instance, by strengthening subsidie s, credit, and road 6 Mesquita (1977) estimated that from 1971 through 1975 over five million hectares, that is, almost one third of Acre’s territory, were tran sferred to new owners. Commonly, emphasizes this author, the land was occupied by rubber tappers who were forced to migrate to urban poor areas. 7 Xapuri and Brasilia are two municipalities located in South Acre region. Since the economic transition to cattle ranching in the 1970s they became one the most important areas of both land conflicts and rubbertappers movement against roads and defore station (see FASE 1989 and Allegretti 1989).

PAGE 36

24 building; 2) to lead legal and many times vi olent efforts against smallholders, untitled occupants, and indigenous people in Amazonia. As a response, at the end of 1985, rubber-t appers organized their first national meeting in Braslia, the capital of Brazil. This event was supported by the National Institute for Socioeconomic Studies (INESC) the Brazilian Department of Culture, and Oxfam, an international organization. At th at moment, they created the National Rubber Tappers’ Council (CNS). With na tional and international suppor ters they also advocated the creation of Extractive Rese rves (Resex) as a pioneering land-reform measure within Amazonia intended to maintain its natural re sources and also provi ding rubber tappers with long-term rights to land. Further strengthening this movement, by late 1988, indigenous people, rubber tappers, and ribeirinhos joined their voices in one of the strongest collective actions in the Amazon. In Acre they built a co alition under the name of the ‘Forest Peoples Alliance’. The large ranchers reaction, however, was extremely violent with the shooting of Chico Mendes, one the most prominent rubber tapper leaders in Amaz onia, in the late 1988. Although the 1980s period was one the most violent periods of Acre’s history, it also created an opportunity for new leader s who argued for a new development plan based on forest use and sm allholders’ land rights. As a consequence, land tenure policies took th eir first steps toward land reforms. In 1989 the Brazilian government created four ag ro-extractive reserv es in areas with intensive land-conflicts in the State of Acre: So Luiz do Remanso, Santa Quitria, Macau, and Cachoeira. In 1990, the Alto Ju ru Extractive Reserve was established, the first one in Amazonia, with more than 506,000 hectares. The same year, the Chico

PAGE 37

25 Mendes Reserve, the largest Braz ilian extractive reserve until 20048, with almost 1 million hectares, was formed. Chico Mendes’ murder led to the creation of seven other extractive reserves in five Brazilian states by 1992. Chico Mendes’ le gacy was revitalized in 1998, when the Workers Party won Acre’s gubernatorial elec tion, establishing a bridge among policy, social movements, and agents managing dive rse land tenures. In the following section I discuss the innovative public policies implemented in th e State of Acre since 1999. Specifically I focus on the effects of gove rnmental initiatives for community land tenures. Acre: Social Movements, Community Land Tenures, and Forest-Based Development Policies In this section I will summarize recent publ ic policies implemented in Acre, which have attempted to establish a forest-based development model, and their implication for land tenure regimes. Particularly, I will con centrate on the efforts of the Acre government to foster alternative policies in community land tenures areas since 1999. The triumph of the forest-based development led by Jorge Viana, a forester who worked as a technical adviser for rural comm unities in the 1980s, led to the establishment of a development model based on sustainabl e use of natural resources. The “forest government” slogan became a key concept whic h suggests the Acre territory could be developed based on land te nure rights and land uses strategies of extrac tivists, colonists, and indigenous people. In addition to the fact that many of thes e policymakers worked closely with local communities over the past 10 years, two factor s help explain why the forest government 8 The largest Brazilian extractive reserve is the Verde para Sempre with 1.3 million hect are. It was created in 2004 and is located in the State of Para.

PAGE 38

26 concentrated much of its atten tion on the spatial dist ribution of land tenure regimes. First, although the legal status of a large portion of land was not precisely known, the type of landowners and their location and land use prac tices were recognized in AcreÂ’s territory. Second, different from the previous governor Viana attempted to implement a policy taking into consideration the economic voca tion of the population of each land tenure category. Thus, economic and ecological zoning (ZEE)9 emerged as a strategic instrument for state planning, territorial management, and policy implementation. Based on the ZEEÂ’s results, four major ca tegories of land tenures types were identified in the state territory (Figure 12). The first one, covering almost 35% of the territory, is characterized by colonization settlements, agro-extractive settlements, extractive reserves, and indigenous reserves The second accounts for approximately 15% of AcreÂ’s area and is comprise d of national and state parks, and other conservation areas. Third, areas under INCRAÂ’s control, which s hould be utilized for human settlements, correspond to approximately 7% of the stateÂ’s ar ea. Lastly, the largest area of the state is in the hands of privat e landowners (mainly large cattle ra nchers) who control almost four million hectares or 26% of AcreÂ’s territory. There is, however, a large extent of land without any immediate land tenure discrimi nation, covering over 22% of the territory. Although these community land tenures have been a key to the forest government, its policies have been altered from VianaÂ’s first term of office ( 1999-2002) to the second (2003-2006). Based on my personal experiences sp eaking with diverse local stakeholders, I have identified four major differences between the first and the second of VianaÂ’s term s of office which could affect economic a nd land cover dynamics within land tenure 9 The ZEE is an important policy instrument for terr itorial management in which the state integrates economic, social, and environmental studies as el ements for subsidizing societyÂ’s decisions.

PAGE 39

27 regimes in Acre. First, while the first te rm of office emphasized non-timber forest strategies, during the second mandate, timber management has been a priority. Second, through 1999-2002 the government attempted to establish a forest model based on a diversity of land use strategies addressed by fo rest inhabitants (rubber tappers, indigenous and colonist). Third, the rela tionship between state and fe deral government was altered when Luis Incio Lula da Silva was elected president of Brazil in 2003. Since this period considerable infra-structure investments have been carried out in Acre to integrate it to the Peruvian region. These in itiatives may have unknown effe cts for forest conservation and community land tenures. Lastly, as policymakers have implemente d a model oriented by market demands (e.g., timber and cattle) since 2003, strategies for increasing the competitiveness of forest strategies have become a key economic priori ty in the second term. During this period, macro-investments (e.g., road building, ener gy, and logging company incentives) were more encouraged than community-based initiatives (e.g., coop eratives and rural association subsidies). Overall, while the first mandate was based on the neoextrativismo approach, the current term of office has been called florestania In the following section I will summarize each of these approaches. Neoextrativismo : A Long-Term and Systemic Approach Neoextrativismo was conceived by Rego (1999), an agricultural economist at the Department of Economics at the Federal Univer sity of Acre, almost four years prior to the election of the Acre forest government. This concept emerged primarily as a response to critics of community extractivism im plemented in Amazonia. Homma (1993), who was one of the most ardent critics, argued that extractivism was not feasible as a regional development model. After the Workers Party (PT) victory, Rego beca me the Secretary of

PAGE 40

28 Production and one of the most important pol icymakers in the forest government from 1999 through 2002. Thus, neoextrativismo during this period, wa s consolidated as an element for integrating economic, social, and en vironmental sustainability in land tenure policies in Acre. Neoextrativismo incorporated five inseparable aspects into community land tenure development. First, neoextrativismo took into consideration the cultural behavior of local residents. Second, it was conceived to cons ider community liveli hood systems and their implications for land use and land cover ch ange. Third, it addressed community as a means of balancing economic and ecological goa ls for policy. Fourt h, it argued that any technology for sustainable land use should be ad apted to local inhabitants. Finally, the neoextrativist approach emphasizes that policies addressing sustainability will only be successful when smallholders are able to dive rsify their systems, for instance, combining agroforestry, forest manageme nt, agriculture, and livestock. Despite its original and systemic overview of the socio-ecologi cal diversities and its suggestions for establishing innovative pub lic policies in the Amazonia, few studies have assessed this approach. Kainer et. al (2003) point out that fo r the first time in the Amazonia, an integrated policy framework regarding the local ecological, cultural, economic, and social diversities was articulated via neoextrativismo Maciel (2003), an economist at the Federal University of Acre assessing the Islands of High Productivity (IAPs)10 as a component in the neoextrativist strategy, estimated that it should increase 10 Islands of High Produtividade (IAPs) were conceived by Kageyama (2002) as a strategy to increase rubber production in the rubber tappers livelihood sy stem in Amazonia. This experiment was implemented in Acre, through which rubber tappers were stimulated to plant rubber trees within deforested areas and bordered by neighbour forestry diversity. In some ways IAPs were very similar to agroforestry systems. The IAPs, while conceived as a neoextrativist model incorporated three di mensions: economic (economic growth with rubber productivity); social (rubber tappers usually worked collectively to implant those experiments); cultural (rubber threes have been the main activity for rubber tappers over the past 30 years);

PAGE 41

29 rubber tappers gross income by 400% as a consequence of a reduction in labor requirements. These results indicate that the neoextrativismo approach attempts to incorporate livelihoods into a publ ic policy agenda, while also poi nting to the fact that the transition from traditional development to fo rest based development could require a longterm process. However, while alternative la nd uses suggested by this approach such as the implementation of agroforestry systems w ith brazil-nuts require almost 30 years to yield, the government mandate in Brazil is limite d to four years. This fact contributed to a new public policy approach after 2002 in Acre which I will discuss in the next section. Florestania : Citizenship Based on Regional Integration, Short-Term Economic Growth, and Conservation In 2002 Viana again won AcreÂ’s election in a new national context. When Lula da Silva, a close friend of Chico Mendes in th e 1980s, took power as president of Brazil, Marina Silva, who was a rubber tapper in Acre during her chil dhood, was selected by Lula to take the position of S ecretary of the Ministry of th e Environment. Moreover, after almost three years of negotiations, the Inte r-American Development Bank (IDB) financed the Acre Sustainable Development Program through which the state received US$ 250 million in loans for supporting its forest-based initiatives. These new macro factors were fundamental in changing state planning in Acre. While during its first phase, the forest gove rnment was characterized by attempting to consolidate a development model based on community land tenures and land uses, VianaÂ’s second term has attempted to estab lish Acre as a regional model for the Amazon. Neoextrativismo which focused on micro-scale and long-term socioeconomic outcomes, was substituted by florestania as a new agenda for the state policies. The florestania environmental (it should to contribute for forest re generation); and political (AcreÂ’s government in 1999 incorporated this initiative into the local public policies).

PAGE 42

30 approach moved to publicize Acre as an economically efficient Amazonian development model that has attempted to integrate both urba n and rural populations as beneficiaries of sustainable forest use. Florestania has incorporated two perspectives into forest-based development implemented in Acre over the past six y ears. First, this concept was founded on the principle of universal human rights, including the citizenship rights of forest peoples. Second, VianaÂ’s policymakers also include d an environmental component within citizenship. In this perspective, forest -based development could produce economic growth to overcome social inequali ties via environmental conservation. The context of these efforts is manifested in the IDB Project, the largest program implemented by the government during its seco nd mandate. The IDB projectÂ’s short-term goals are the promotion of economic growt h, environmental sustainability, and local economic diversification (IDB 2002). To implem ent its actions, this program was divided into three sub-programs: a) sustainable fo rest management, conservation, and state enforcement; b) employment, technology dissemination, and community-based subsidies; and c) infra-structure investments in roads, rivers, and energy ge neration (BID Project 2002). The IDB program inaugurates a new pha se in Acre in which forest based initiatives were incorporated into a context of regional integration. A range of complementary endowments has been mobilized by the forest government in order to implement its macroplanning. Besides IDB, Acre has also been supported by the Brazilian government and the Corporacin Andina de Fomento (Andean Corporation for Development) with approximately US$ 640 million for

PAGE 43

31 implementing investments in infra-structure public services, economic subsidies, and other economic initiatives by 2006 (SEPLANDS 2005). While addressing regional in tegration, the Acre government has also continued to encourage land use strategies employed in co mmunity land tenure regimes. In this new phase, however, emergent demand for forest pro ducts and the need to show the success of this development model have forced policym akers, along with nati onal and international organizations, to place greater focus on econom ic rather than cons ervation goals within communities. Although the long-term impli cations of this policy approach for socioeconomic development and land cover is still not clear, so me economic growthdeforestation tradeoffs such as in cattle raising and logging have been observed. Initiatives for regional inte gration implemented during the florestania period have generated new concerns about the conti nued viability of community land tenures models designed in the early 1990s to achieve economic, social, and environmental goals. I will now move toward the third part of this chapter, which focuses on three land tenure types present in the state of Acre: extr active reserves, agro-extractive settlements, and colonist settlements. It will be divided into two parts. Fi rst, an overall description of those land tenures types will be provided w ith an emphasis on differences in their economic and environmental outcomes. Second, three specific study sites focused on this thesis are described. Community Land Tenure: An Emergen t Issue for Assess ment in Acre The largest portion of the Acre governmentÂ’s actions has been concentrated in the Rio Acre watershed region. Both economic integration through the Pacific, and community-initiatives such as within the Ch ico Mendes Reserve, colonist settlements, and agro-extractive settlement s are located in this area. In addition, this region holds

PAGE 44

32 approximately 71% of AcreÂ’s total populat ion and almost 56% of the total rural population (Governo do Estado do Acre 2000a). The Rio Acre region also accounts for 83% of the total deforestation in the St ate (FUNTAC 1999). Although only 37% of this region remained as forested area as of 1996, the largest percent of these forest areas (87%)11 was located within community land tenures of extractive reserves, agroextractive settlements, indigenous land, and co lonization settlements. Almost 70% of all colonist and rubber tapper families of the st ate of Acre are living in the Rio Acre watershed region. These characteristics make the Rio Acre watershed in eastern Acre a crucial area to assess the effectivenes s of community land tenures models. While agricultural smallholders have been pointed out as key agents for forest clearance (Angelsen 1995, Godoy, Jacobson, and Wilkie 1998), in Acre, they are regarded as agents for promoting conserva tion. On the one hand, extractive reserves and agro-extractives reserves have been advocated as an ideal model to address human needs through the harvesting of forest products (Allegretti 1990, Murrieta and Rueda 1995). New land use strategies, such as agroforestry and forest management, have also been argued as strategies for col onist settlements to reduce deforestation and bring economic gains (Menezes 2004). On the other hand, while the long-term effectiveness of extractivist models has been questioned (Homma 1993), recent studies have identified the tendency for increasing deforestation with in these land categor ies (Sassagawa 1999, Gomes 2001, Lorena 2003). Although community land tenures and their associated land use strategies have had diverse evaluations, they became key component s in AcreÂ’s policies. Understanding land 11 This estimation was performed through vector file s processing with ArcView. The datasets were available by ecological and econom ic zoning of the Secretary of En vironment of the State of Acre.

PAGE 45

33 ownership in this state as well as key factor s associated with resource use management activities are fundamental step s to evaluate the effectiv eness of community managed areas to bring economic growth with e nvironmental conservation. I will therefore summarize the main differences among three major categories of community land tenures existent in the state of Acre. Colonization Settlements (PC) Estimates regarding the number of colonist farmers in the State of Acre range from 15,000 to 23,000 families (Governo do Acre 2000a, IBGE 1998) and their territories cover between 1.4 and 1.6 millions hectares (INCRA 1999). The creation of these settlements had three major goals: 1) to stim ulate human occupation of AcreÂ’s territory; 2) to establish agricu lture and cattle as a hegemonic model for economic growth; and 3) to reduce social demand through agrarian refo rm from small farmers of south, central and southwestern Brazilian regions (Cavalcanti 1994). The largest num ber of colonists, known as sulistas settled in Acre beginning in the late 1970s, and had extensive experience working with annual and perennial crops as well as livestock. These economic strategies contributed to th eir relatively good socioeconomic conditions (e.g., Cavalcanti 1994, Menezes 2004), but this has been c ounterbalanced by the large amount of deforestation, landowner turnove r, and soil degradation asso ciated with colonization areas. Colonists were stimulated by INCRA to deforest their lands to implement agriculture and cattle pasture areas. In f act, this was facilitated by the livelihood background of these families, largely base d on shifting cultivation systems. Although accelerated urbanization created a local ma rket for crops there has also been landownership turnover among households of colonization areas (Cronkleton 1998).

PAGE 46

34 As pointed out by Rego (2003), cattle are th e main economic activity for colonist farmers located in the Acre River watershed. His analyses show that in 1997, 63% of the households in colonist areas commerciali zed cattle contributing over 30% of total income. Cattle expansion has been a common characteristic among colonist families (Menezes 2004) and may stimulate landowners to obtain new land for pasture establishment. Land concentra tion and migratory movements ha ve been related to this process. Moreover, negative social conseque nces (such as migration and land conflicts) and negative environmental outcomes (such as continuing deforestation) are anticipated. Among community land tenure regimes, colonist farmers are key actors contributing to deforestation in Acre. Using da ta available from the Brazilian Ministry of Environment, I estimated that by 1996, approxima tely 30% of the total deforested area in Acre was located within colonization settlements, while extractive reserves, state, and national parks contributed less than 5%12. Based on the same dataset, I estimated that 37% of colonization areas were deforested by 1996 which was superior to the 20% of deforestation permitted by Brazilian law to all rural properties in Amazonia13. Taking into consideration that cattle and crops ha s been the main economies for colonists in Acre, deforestation could still be increased in this areas. 12 See footnote 4 for details of how this amount was estimated. 13 The Federal Decree No. 2166/01 that altered the La w No. 4771/1965 (Forestry Law) establishes that small and large farmers in Amazon region are permitted to cut down until 20% of forest within their land plots.

PAGE 47

35 Extractive Reserves (Resex) Currently, there are two Resex in Acre covering almost 10% of the stateÂ’s territory14 (Governo do Acre 2000a). Rueda (1995) argues that the demand for Resex in Acre emerged as ecological, economic, and soci o-cultural dimensions of land rights. The ecological perspective is explained by the im portance of the forest in the rubber tapper livelihood, without which there could be no ex tractivism. Under th at circumstance the possession of land would not be important fo r rubber tappers. The economic dimension of the Resex model implies that land con cession was a mechanism for guaranteeing longterm survival of rubber tappers and their fa milies. Lastly, the ResexÂ’s socio-cultural dimension argued for the rubber tapp ersÂ’ right to be extractivists. The incorporation of these dimensions in to the Resex proposal was a result of an intensive struggle led by the rubber tapper movement to consolidate a model of land reform in the Amazon based on the livelihood sy stem of local inhabitants. Rubber tappers developed a forest-based livelihood system, in cluding consumption from the gathering of wild foods, hunting, medicinal plants, and cas h income from non-timber forest products. As a land tenure regime, Resex are characterized by being occupied by families who combine indigenous agriculture, livestock, and forest activities in their land use strategies. Since their creation in the early 1990 s, Resex resource management has concentrated on non-timber forest products. Although rubber has not had an attractive market, rubber tappers have continued to harves t this product. Over the past seven years, 14 The state government may create thr ee additional extractive reserves which may increase the total area to over 2,000,000 ha or approximate 15% of the territory (Governo do Acre 2000b). These new planned Resexs would be located in the Juru River watershed.

PAGE 48

36 however, stimulated by local policies, many ot her non-timber forest products were added to the range of commercial pr oducts explored by rubber tappers families, such as copaiba oil ( Copaifera ssp ), murmuru nuts ( Astrocaryum murumuru), unha de gato ( Uncaria tomentosa ), and aa ( Enterpe oleraceal ). Agro-extractive Settlements (PAE) Agro-extractive Settlements (PAE)15 comprise approximately 194,000 hectares and have an estimated population of 1,050 families in Acre (Governo do Acre 2000a). This category of land tenure was also a response of the Brazilian government to reduce land conflicts, as well as deforesta tion, in Amazonia. The agro-ext ractive settlement (PAE) is an intermediate category of land tenure between Resex a nd colonization settlements. PAEs were initially conceived in 1987 as an alternative solution for Resex demands and also as an immediate response from the Brazili an state for reducing land conflicts in the Amazon region. Different from the extrac tive reserves, PAEs, like colonization settlements, are regulated by INCRA. However, their population is largely constituted by rubber tappers who have implemented a land use system very similar to ResexÂ’ families. Indeed, as a consequence of their institu tional flexibility and proximity of their population to colonist practices, the timber management projects, which had strong opposition from rubber tappers, were more easily launched in PAEs. Even though the PAEs are administer ed by INCRA, their regulations and inhabitants differ from those of colonist areas, such as Peixoto. Decree No. 268/1996 legislates that PAEs may be used for fore st management through sustainable practices (INCRA 1999). Kageyama et. al (2004) point out important di fferences between colonist 15 Earlier the INCRAÂ’S Decree 627/1987 called this form of land-tenure extractive settlement projects. However, later in 1996, it was altered to agro-extract ive settlement. Surely this change characterizes the PAE itself as a combination of colonist and extractive land-use practices (see Kageyama et. al 2004).

PAGE 49

37 and agroextrativist settlements: a) PAEsÂ’ settlers are more sensitive to resource conservation; b) the plot sizes allocated to PAE families have been around 300 ha, while in colonist areas this amount varied between 25 ha and 100 ha; c) the size of agricultural and pasture areas among PAE families are le ss than those found among colonist farmers, and cleared areas have often returned to sec ondary forest. Moreover, similar to extractive reserves, PAEs have been defined as posse ssions of the federal government. Based on the same regulation, PAEs are characterized as community land concessions that must be managed by an association, cooperative or other variety of collective organization. Currently in Acre there are eight PAEs but INCRA and the State government are planning to create 30 others ove r the next few years, which ma y raise the PAEsÂ’ land area to approximately 1,900,000 hectar es, or, 12.3% of the Acrean territory (Governo do Acre 2000b). As pointed out by Rego (2003), PA E settlers prior to 1997 relied upon a combination of agriculture, livestock, and non-timber forest product-based land use systems. However, after 1999, stimulated by forest-policies initiatives, timber forest management came up as a new opportunity for rubber-tappers of these types of settlements. In the following section I will discuss the study sites in this thesis. I focus on how the three areas of this research were select ed as a sampling of extractive reserve, agroextractive reserves, and col onist areas located in the Rio Acre watershed region. The Study Sites Peixoto Colonization Settlement: The Agrosilvopastoral Model The Grupo Novo Ideal (GPNI) rural association, whic h is located in the Peixoto settlement, was selected as a colonizati on sample in the Rio Acre watershed. The Colonization Settlement, in which this community is located, has around 318,000

PAGE 50

38 hectares, and was created in 1977 by INCRA. Th is settlement, facilitated by the presence of the BR-364 highway, emerged in the 1980s as the most important area in the Rio Acre watershed region for food production (Menez es 2004). Situated around 150 kilometers from the city of Rio Branco, AcreÂ’s cap ital, GNPI is currently comprised of 27 smallholders. As colonists, their land use strategies in clude a range of slas h and burn agriculture as well as livestock activities, but as a result of policy incentives since 1999, cattle and coffee have emerged as their main econo mic activities. In 1994, however, they had launched experiments with agroforestry as an alternative to defore station and means to diversify their cash income. However, the agro silvopastoral strategies of those families have been strongly influenced by dynamics of land use that took place in the land tenure regime in which they are located. The land use history within this community, as in other colonist areas located in Acre, has been characterized by a drastic tr ansformation, from forest uses (e.g., logging, rubber, and Brazil nuts) to agriculture and cattle production. Over th e past few years, however, soil degradation has been a constr aint for colonists in maintaining their agriculture and pasture areas. Ar ajo and Silva (2000) show that soil fertility in this settlement is not appropriated for agricultural activities. Yet, despite the areaÂ’s limited shifting cultivation pote ntial, Lorena (2003) observed that in PeixotoÂ’s lots, the annual and perennial crops area as we ll as pasture increased from ~17% in 1990 to ~35% of the total area in 1999. These expanding activities ex plain why the forest cover within this settlement declined from almost 90% in 1984 to 54% in 199816. Both ecological and 16 I used Landsat TM imageries of 1984 and 1998 to perform a forest-non forest unsupervised classification to estimate the amount of defore station within this settlement.

PAGE 51

39 socioeconomic factors stimulate small farm ers to clear more forest in order to compensate for pasture degradation (Valentim et al. 2000), as well as to implement alternative land uses to avoid deforestation. To address the problem of pasture degrada tion, some initiatives have been launched in Peixoto and other coloni zation areas. Over the past few years a coalition of governmental and non-governmental organizations has attempted to in troduce alternative land use practices such as agroforestry systems (Menezes 2004). The GPNI community was among the first areas in which these initi atives have been implemented. While until the early 1990s the families in GNPI concentr ated their land use strategies on slash and burn crops and livestock activit ies with emphasis on cattle, in 1994 they were stimulated to introduce agroforestry systems. In collaboration with the Research Group for Agroforestry Systems and Extension (PESACRE), a local NGO in Acre, three ne w components were in corporated into livelihood systems of GNPI households. First, for reducing environmental degradation, PESACRE encouraged community members to introduce agroforestry systems into their land use practices. Second, for enhancing the co mpetitiveness of their products as well as decreasing the risk of health problems as a consequence of pesticide use, organic agriculture initiatives were also established among GP NIÂ’ members. The last proposal attempted to address the n eed to focus on long-term ma rket sustainability through leadership training and commercialization of ag roforestry products. Although the GPNI community may be charac terized as a coloni zation project, its members have adopted innovative strategies for changing the conventional land use practices within this category of land tenur e. These families have combined traditional

PAGE 52

40 agriculture, agroforestry cattle, livestock, coffee, and fru it processing. Further, the forest government, in its spatial economic zoning of 2003, established th e area in which the GPNI is located, as having potential for coffee, pupunha and dairy cattle. As a result, governmental planning may stimulate comm unity members to shift their production systems from agrosilvopastoral strategies to these activities. Add itionally, other drivers commonly affecting land cover change in Am azonia have also been observed, such as timber extraction and the presence of unofficial dirt roads. In this dynamic context, it becomes important to assess economic and land cover outcomes in a changing colonization project such as GNPI. Porto Dias Agro-extractive Settlemen t: The Timber Forest Management Model In the late 1980s, guided by Decree No. 627/ 87 that introduced th e agro-extractive settlement (PAE) as a new category of land tenure, INCRA initiated the discussion to create the PAE Porto Dias. In 1989 this ar ea was established with approximately 22,000 hectares, and in 2002, Porto Dias had a popul ation of almost 90 families (Governo do Estado do Acre 2000a). While PAEs are superv ised by INCRA, an agency created for implementing colonization settlement in Amazo nia, Porto Dias was planned to settle rubber tapper families. It is also characteri zed by having a combination of agricultural colonists and forest extractivists. This comb ination has been present since its initial moments of discussion for launching its creation in the early 1980s (Stone 2003). Moreover, when the Brazilian government launc hed colonization settlements in the early 1970s and 1980s, some colonist producers esta blished their lots within Porto Dias (Michelloti 1993). Porto Dias is divided into three geogra phic zones. Zone 1 is comprised of 27 members who are represented by the Porto Di as Rubber Tappers Asso ciation. Zone 2 is

PAGE 53

41 managed by So Jos Association, called Mossor, represented by 11 of the PAEÂ’s residents. Zone 3 is under the control of the ASPOMACRE, an asso ciation that has its headquarters located in the state of Rondni a and has 103 members of which only 11 are inhabitants of PAE Porto Dias. The households of zones 2 and 3 do not cons ider themselves as rubber tappers and prefer to be categorized as colonists. When interviews were conducted in these areas these inhabitants declared that they had a preference of having their landholdings ( colocaes ) outside of PAE territory, but INCRA did not approve their proposal. The range of economic activities addressed by these families reflects their self categorization as colonists. Cattle and logging combined w ith agriculture and small-scale livestock production are their major economic activitie s. Forest products, such as rubber, medicinal plants, and Brazil nuts have no importance for their subsistence or commercialization. In this study I have selected zone 1 to assess community timber forest management in the Porto Dias PAE. This is the area wher e rubber tappers have implemented initiatives combining forest management, as well as othe r land use for agriculture, cattle, and other livestock. The Porto Dias Association was created in 1988 and its number of members has varied over time. For the purpose of this study, all of the 27 households in Porto Dias PAE Zone 1 were selected. Timber management was implemented by th e Porto Dias Association in 1994 with the support of the Centro dos Trabalhadores da Amaznia (CTA) (Center for Amazonian Workers). The first timber harvesting, how ever, took place two years later in 1996. Besides timber products, through support from the Federal Ministry of the Environment

PAGE 54

42 (MMA), the forest management plan included also rubber, Brazil nuts and medicinal oils such as copaiba ( Copaifera spp ). Since that time period, however, association members have predominantly oriented their effort s toward timber production (Rodrigues 2004). In 2003 the PDA became the second commun ity-managed project to achieve Forest Stewardship Council (FSC) certification in Am azonia. This process is considered a difficult task for any organization, given th e bureaucratic demands and costs which are particularly onerous for communities faci ng challenges in organization and physical infrastructure. Technical support from out side organizations, such as CTA, was fundamental for PDA Porto Dias to be succe ssful in gaining certification status. Although FSC certification was conceded to the association, the number of rubber tappers engaged in timber management has varied from 7 to 10 families. Political disagreements, lack of financial transparen cy, and limited financial and technical support for community leaders, as well as absence of economic returns, have been the main impediments to the timber management initiative. Chico Mendes Extractive Reserve: The Non-timber Forest Management Model The Chico Mendes Extractive Reserve is one of the premier communities in which forest diversification has been pursued over the past years in eastern Acre. It is bordered by large cattle ranchers that we re the main group antagonistic to its creation. With almost 1 million hectares and an estimated curre nt population of 3,000 families, the Chico Mendes Reserve was the second extractive reserv e created in the stat e of Acre in 1990, established by Federal Decree 99,144/90. Its inhabitants have concentrated their efforts on non-timber forest products, small plots of ag riculture, and livestock as their activities for subsistence and cash income generati on. Until 2004, it was the largest Brazilian extractive reserve, stretching across parts of five Acre municipalities: Rio Branco,

PAGE 55

43 Xapuri, Brasilia, Sena Madurei ra, and Assis Brasil. In the late 1970s and 1980s these municipalities were where the land conflic ts became most intense in Acre. The first landmark for the creation of Ch ico Mendes occurred in 1980 when Wilson Pinheiro, president of the rural workers union of Brasilia, was killed by large scale cattle ranchers. Almost ten years la ter in 1988, the assassination of Chico Mendes, the rubber tappersÂ’ most high-profile leader was the final step for the creation of this reserve. Despite its importance as a land tenure category that has attempted to balance economic, cultural, and environmental goals, so me studies have iden tified the tendency for accelerating forest loss in the Chico Me ndes Reserve. Wallace (2004) noted that rubber tappers have invested in livestock, in cluding cattle, as a strategy for increasing their total on-farm income. Sassagawa (1999) es timated that the percentage of deforested area in this Resex increased from 0.72% in 1986 to 2.9% in 1998. However, it has been an important barrier for forest clearing in this region. In 2001, monitoring using Satellite GOES indicated that fire occurrence has been most intensive outside of this reserve (Souza 2001). Indeed, although the long-term viability of the Resex as a land tenure strategy has been a concern among scholars, various initia tives conducted in coll aboration with local inhabitants have indicated its long-term ec onomic and environmental sustainability (see Maciel 2003 and Kageyama 2002). Over the past ten years, a coali tion of governmental and non-governmental organizations and uni versities has collaborated with Resex communities to amplify their range of non-timbe r forest products as an alternative to deforestation and intensive land use. For in stance, since 1999 the fo rest government has supported market-based initiatives though cred it, subsidies, commun ity training, and the

PAGE 56

44 encouragement of 15 non-timber forest produc ts (Kainer et al, 2003). In addition, the Federal University of Acre (UFAC), s upported by Brazilian and international organizations, has been a key institutions collaborating with ru bber tappers and their organizations. A few months after Jorge Viana won the 1998 election, a multi-institutional partnership was established for implementi ng and monitoring the forest management within three rubbe r tapper estates ( seringal ): So Pedro, Floresta, and Palmari. A group of 34 rubber tapper landholdings ( colocaes ) were selected from those three seringais in which the non-timber forest program has been implemented since 1999. These families were selected in my study as cases of a non-timber forest model taking place within Resexs in Acre. The three seringais selected cover an area of al most 37,000 hectares and they are located in the municipality of Xapuri. Some studies such as Esteves (1999) emphasize that these three areas are among the places in the Chico Mendes Resex where, over the past 10 years, intensive investments in Brazi l-nut production has taken place, as well as community organization activities. Esteves show s that economic and social benefits, such as dirt roads, credit, and subsidies as well as education and health programs were distributed based on kinship ties. These fact s suggest that the socioeconomic benefits were not distributed equally among rubber ta ppers of the Chico Mendes Resex. However, since the forest government took power in 1999, the strategy employed by families of these three seringais has been a reference point for establishing a common development plan throughout the Chico Mendes Resex. Th ese characteristics offer us a unique

PAGE 57

45 opportunity to assess the community forest management model in terms of its economic performance and capacity for forest conservation. Land use strategies in the Chico Mende s Resex have been implemented in accordance with its Utilization Plans ( Planos de Utilizao ), which are defined by common agreement between rubber tappers th rough their associat ions and IBAMA and CNPT. Although many of those economic initiatives have been focused on non-timber forest products, timber extraction has been discussed as an economic opportunity. Even though timber management is not permitted within Resex areas, the Acre government supported by Marina Silva, the current Br azilian Minister of the Environment, has attempted to change the Resex law (Feder al Decree No. 98,897) to incorporate timber extraction for commercialization into the rubber tapper economy. Yet, the question of timber versus non-timber models of forest management remains an important topic of discussion among communities, their or ganizations, and policymakers. Conclusion While economic booms were historically a key driver for stimulating human settlement in the Amazon region, policy inst ruments have more recently been utilized there to facilitate ec onomic development. In addition, lo cal, national, an d international forces were arranged based on social coalitions among act ors at various levels. For instance, the social movement led by rubber tappers in the 1970s and 1980s with support of organizations was successf ul to influence changes in the land tenure policies for creating extractive reserves and agro-extractive settlements. Although the policies of the forest government have been one of the most important factors influencing land tenure after the rubber tapper movement of the late 1980s, remarkable differences are visible in deve lopment policy since the forest government

PAGE 58

46 entered in 1999. While neoextrativismo emphasized a policy based on sociocultural use of natural resources, florestania has called for polices based on products with comparative economic advantage at the regional level. The effectiveness of these policies in terms of land cover and economic results wi ll be evaluated through empirical data for 2004. This study concentrates on three community land tenure systems for assessing the economic and forest cover effects of these polices. The Chico Mendes Extractive Reserve, Porto Dias Agro-extractive Settlemen t, and the Peixoto Colonization settlement were selected as areas for achieving this analysis. These land tenure systems are a representation of community land tenures located in the Rio Acre Watershed Region, where both neoextrativista and florestania initiatives have been implemented. Remarkable differences, such as socio-cultur al, institutional, and livelihood system, are evident among these land tenure types. In the following chapter, I address methodol ogical issues arising from the distinct requirements for conducting economic and environm ental research in each of these three land tenure categories I divide the discussi on into three major sections. The first will concentrate on how the fieldwork was conducte d. Secondly, I will explain how the land cover and economic data were processed, as well as the key dependent and independent variables in this study. And finally, I will discuss how economic and land cover were linked in order to conduct the analysis.

PAGE 59

47 CHAPTER 3 METHODS AND TECNIQUES FOR EC ONOMIC AND REMOTE SENSING ANALYSES Introduction This chapter explains how I used econo mic and remote sensing approaches to evaluate three land tenure types located in th e state of Acre, Brazil. To identify the land use and economic dynamics in the study areas, I combined survey fieldwork with remote sensing data from an extractive reserve, an agro-extractive settlement, and a colonist settlement. I compared land tenure categories in order to understa nd the linkages between land use strategies and land cover in each community area selected. Various studies have concentrated effo rts on linking socioeconomic and land cover factors at multiple scales. While some anal yses have focused on the macro-scale (e.g., Cattaneo 2001, Carvalho et al. 2002), other studies have evaluated the dynamics and drivers at the micro-scale, particularly at the household level. For example, Vosti et al. (2003) elaborated a farm-level bioecono mic linear programming model to simulate consequences of product and tec hnology choices for deforestation Walker (2003), seeking an economic explanation of forest cover change at the household level in the Amazon region, provided a smallholder mode l for colonist settlements based on Chayanov household theory in which demogra phic and market variable are linked. Perz (2001) also focused on economic and land use an alysis at household level in colonist areas of Amazon region using multivariate regression.

PAGE 60

48 While these and other studies evaluate de forestation at the family level in the Brazilian Amazon, I have identified virtually no work that compares multiple land tenure types. A notable exception is Mertens et al. (2002), who focus on cattle economy and deforestation in the state of Par, Brazil, and published a remarkable study in which socioeconomic and satellite data were li nked to understand land cover change over a three year period. While these authors compared three land tenure types, the comparison is made at the municipal level, which leav es out many important explanations found only at a household level. I focus on land use processes at the hous ehold level within three land tenure categories in southwestern Amazonia: extractiv e reserve, agro-extractive settlement, and colonist settlement. Consequently, th e methodology involves both economic and land cover analyses and was performed using th ree methods. First, I conducted fieldwork, which was divided into two pa rts: 1) a household economic survey in 2004; and 2) the land cover GPS data colle ction during the summers of 2004 and 2005. Second, I processed the data from the economic survey and GPS and satellite images. Households were analyzed in terms of their economic re sults, that is, income and cost values aggregated by economic activities (crops, ti mber and non-timber products, as well as livestock). I utilized the GPS points and Landsat data to perform a supervised classification for each community area. In order to correlate these results to the economic outcomes, my classification distinguished four land cover types: crop s, forest, pasture, and forest regrowth. Third, I linked economic and land cover data for the three study sites, in order to conduct a mu ltivariate analysis to test my hypotheses about land tenure, the economics of land use, and land cover outcomes.

PAGE 61

49 In the next sections I discuss these three phases in more detail. I begin by describing how the socioeconomic survey was carried out. In this part I discuss the questionnaire, planning for fieldwork, a pric e survey, and the fieldwork itself. Second, I focus on how the land cover GPS points were coll ected. The third part provides details of how a calibration and validation of satellite da ta were carried out. The fourth section I explains how economic and land cover data were analyzed. The Socioeconomic Survey In this section, I discuss how the socioeconomic data were collected. It is important to note at the outset that this study resulted from multi-institutional collaborations with various organizations in Acre, so I recogni ze my collaborators as co-authors of the database on which I will draw. The 2004 surv ey resulted from collaborative efforts among PESACRE, CTA, UFAC, a nd the State Government. Besides research assistants, community memb ers were also directly involved in the field data collection. At th e Porto Dias Agro-extractiv e Settlement, the residents association selected five community members to take part in the research group. In the Peixoto colonization site, the Novo Ideal AssociationÂ’s board director designated three resident members to help us during our fieldwork. In the Chico Mendes Resex, the Association of the Residents in Xapuri (A MOREX) mobilized five rubber tappers to guide the group within the forest. They al so helped applying the semi-structured questionnaire. The Chico MendesÂ’ fieldwork al so counted on the participation of one board director from the Rural Workers Uni on of Xapuri who had past experience with socioeconomic surveys. Prior to the fieldwork, additional planning was required given the distinct characteristics of the study sites. First, he lped by local stakeholde rs, I selected my study

PAGE 62

50 sites within each land tenure ca tegory. Second, in order to gr asp the land use specificity of each land tenure type, I used meetings with community and institutions as an opportunity to elaborate my questionnaire. After these two stages I carried out two surveys. The first one was the household fieldw ork in the three select ed sites. The second one was a price survey to ma ke possible the estimation of income and costs. When I concluded my two surveys, I organized the da ta in a spreadsheet program and calibrated and validated the land cover classifi cations during the summer of 2005. Household Sampling The household sampling was carried out following three sequential phases. First, helped by IBAMA, CNPT, CNS, and community organizations, I identified in the Rio Acre watershed region community land te nures representing extractivist, agroextractivist, and colonist areas. For each co mmunity land-tenure types present in this region, I distinguished communities by degr ee of socioeconomic development, taking into consideration access (roads), presence of public services (school, hospital, etc), community organization (association and coopera tive), technical assistance, and level of production. As a result, the Chico Mendes Resex, Porto Dias PAE, and Peixoto PC were selected. Second, within these areas, I distinguished families working with agrosilvopastoral and timber as well as non-timber forest ma nagement. This criterion was adopted in order to evaluate the interface w ith public polices taking place in Acre since 1999, which have emphasized those land use stra tegies. Third, I also considered areas with prior governmental investments, such as road construction, subsidies, governmental support for commercialization, and certif ication. Consequently, one rubber tapper association area, one rural association area, a nd three rubber estates were identified for conducting this study. In the Chico Mendes Re sex, the sample size was constituted by 34

PAGE 63

51 families from the Floresta, Palmari, and S o Pedro rubber estates. In Porto Dias, 27 households located in the area under the c ontrol of the Porto Dias Rubber Tapper Association1, were selected. Lastly, within the Pe ixoto Settlement, the household sample included 27 families organized under the Novo Ideal Rural Association16. The Questionnaire The questionnaire was derived from particip atory efforts with various stakeholders and community members. This strategy was implemented to ensure that the livelihood system of each land tenure type was capture d by the survey. Overall, the questionnaire focused on land cover, socioeconomic aspect s, and resource management practices targeted by forest government policies in Acre As the livelihood and land use strategies across land tenure are not similar, I elaborat ed one questionnaire fo r each study site (see Appendix A, B, and C). For instance, in the Chico Mendes Resex communities, timber forest management was not included in the questionnaire, because this activity is not permitted in that tenure type. The questionnaire was divided into eight sections. The first included socioeconomic topics, such as household co mposition, migration, accessibility, and land use distribution. Section two focused on cr op production, livestock, and machines and equipment. The third section quantifies cost s and yield per type of activity conducted by households, such as crops, livestock, and fore st management. This section also included family expenses on purchased goods and services (such as rice, beans, oil, salt, clothes, health etc), as well as family and outside labor force. The fourth part evaluated the amount of goods cultivated and/or harvested th at were used for family consumption, such 1 The number of families does not correspond to the current number of members in this association. That is because the sample included families that we re not affiliated with the organizations.

PAGE 64

52 as annual and perennial crops, livestock, herbs, forest products, among others. The fifth section of the questionnaire considered cash in come from commerc ialized production. The sixth included off-farm income (e.g., bolsa escola salaries, retirement income). To assess institutional support fo r land use strategies, the seve nth section included questions on rural technical assistance from govern mental or non-governmental organizations. Finally, the last section had open questions about public policies implemented in the community area. The Household Fieldwork The field survey was carried out from Ma y to August 2004. I selected five research assistants with prior experience in socioec onomic surveys in rural communities in Acre, each with different backgrounds (economics geography, history, agronomic engineering, and forestry engineering). On average, we spent one day with each household, in order to conduct a semi-structured interview, partic ipatory land use mapping, and GPS training sampling. Prior to the fieldwork, the assistants atte nded a one-week training course focusing on the livelihood systems of smallholders, semi -structured questionnai res, and the use of remote sensing and GIS techniques in the field. In the last tw o days of training they practiced these techniques in the the Humait Colonization Settlement, located approximately 30 kilometers from Rio Branco, in the municipality of Porto Acre. Moreover, community members were incl uded as part of th e research team. The Price Survey To estimate values for all household costs (expenses) and income sources (revenues), I conducted a price survey in the municipalities where families commercialized their products and bought goods: Xapuri, Brasilia, Rio Branco,

PAGE 65

53 Acrelndia, Epitaciolndia, and Plcido de Castro. Moreover, cooperatives and associations of producers were contacted, as well as the gove rnment offices to determine the prices for timber and non-timber forest produ cts. To correct for inflation, I used the IGP (General Price Index), the most common i ndex in Brazil for that purpose, to correct all prices to February 2005. I used a mean price for each product or service to estimate the income and costs of each family, which were corrected by IGP. These values were converted to the mean U.S. dollar – Brazi lian real (R$) exchange for February 20052. Participatory Findings Assessment: Atte mpts for Calibration and Validation Household analysis, especially when focu sed on economic analysis, has often been an obstacle for socioeconomic researchers. Ellis (2000) emphasized the complexity of empirical appraisal at the household level as a result of livelihood diversity as well as inter-temporal social processes continuously taking place. Schmink (1984) argued that household economics are determined by social processes which are historically defined. In this context, I have understood that ru ral policies, as in Acre, have incessantly failed as a consequence of lack of know ledge about smallholder livelihoods and alternative strategies. In order to addre ss these challenges, Hildebrand and Schmink (2004) suggest a framework combining part icipatory methods and ethnographic linear programming fostered by a constant partic ipative process of si mulation, calibration, and validation of household models. The survey methodology used in this study employ this approach. I involved local residents and decision makers of local, re gional, and national or ganizations during the sampling, fieldwork planning, and the fieldwork itself. Calibra tion and validation of data 2 This study utilized the mean US$ to R$ exchange for February 2005 of US$ 1.00 to R$ 2.597.

PAGE 66

54 were carried out in May 2005 when several st akeholders attended seminars in order to evaluate the datasets. These seminars were conducted in four pha ses. First, after th e household and price surveys, the data underwent processing and preliminary analyses. Results were then discussed with researchers, policymakers, NGOs, and community or ganizations in Rio Branco. Next, the data related to the amount of production and prices for timber and nontimber forest products were corrected. Thir d, findings were submitted to community members for evaluation of in each study site via particip atory dialogue we discussed the forest and agrosilvopastoral seasonal calenda r, as well as the range of main products extracted and their prices, and income and costs distribution. After these meetings, I identified discrepancies, and the da ta were adjusted and re-processed. Land Cover Data Collection and Fieldwork In this study, I used Landsat 5 TM+ imag es from the scenes 67/001 and 67/002 of about the same time of the year. The 2004 images were taken between July 29 and August 26. For the two scenes, I used bands 1 through 5. In addition, I utilized ancillary data including polygon shape files containi ng land tenure types applied for the study sites. Polygons delimiting the Chico Mendes Resex, Porto Dias Agro-extractive Settlement, and Peixoto Colonization Settlem ent were selected for this analysis. While doing household interviews, the resear ch team collected training samples of land cover types (crops, pasture, forest regr owth, and forest) at the three study sites. Additionally, a printe d sub-set image of 2004 Landsat image3 was provided to each team member for each area to identify likely land cover types in the household area. All GPS 3 The Landsat images were adjusted for RGB= 5, 4, 3 and georefereced for UTM, datum SAD 69, zone 19 South.

PAGE 67

55 were adjusted to UTM, datum SAD 69, and zone number 19 South. To increase accuracy of these data, each land cover point was coll ected at least three times. Training samples collection was carried out using a modified version of the CIPEC training sample protocol (Appendix D), which incl udes details such as soil, vegetation type, and presence of human disturbance. In this study, we collected 442 traini ng samples and 366 CIPEC forms. This fieldwork was carried out in two phases. Firs t, during the fieldwork in 2004, we collected 345 training samples and completed approxi mately 300 land cover forms. Although the field team was intensively trained fo r the land cover GPS point collections I encountered some difficulties in performing a supervised classification for th e study areas during the fall 2004. As Daly and Silveira (2001) pointed out, over 50% of Ac reÂ’s territory is covered by bamboo forest. On the first supervis ed classification perf ormed at the end of 2004, this vegetation dominance created confus ion between forest regrowth, crops, and bamboo-dominated vegetation. Consequently (second), from May to June 2005, I visited all three community areas to collect new land cover samples (principally within bamboo and secondary forest areas). During this pe riod, I gathered 97 samples and 66 land cover protocols. In the summer of 2005, I was able to complete a new supervised classification and verify its accuracy at each study site. The Post-Survey: Data Processing and analysis Household Economic Approach The economic analysis performed here was based initially on Brazilian literature on micro-economics and agricultural economics (e.g., Gastal 1980, Mafra 1988, Barros and Estcio 1972, Guerreiro 1994, and Martin 1995) th at elaborated economic parameters for assessing input and outputs among smallholde rs. In addition, I drew on studies focusing

PAGE 68

56 on the Amazonian smallholders (Vosti et al. 2003, Walker 2003) which have evaluated households as economically rational decisi on makers who pursue profit maximization, that is, they continuously plan their cost-i ncome balances. This traditional model of analysis ignores the asp ects highlighted in peasant economy theory (Thorner et al. 1986). This theory establishes that household economics (e.g., extrac tivists, agro-extractivists, and small farmer colonists) aim to balance both market (sold yi eld) and consumption (subsistence yield) goals. In this context, family labor as a cost component and subsistence as an income element represente d two important factors in the analysis. For identifying the land tenure differences, an ex-post4 economic analysis was performed in which inputs (factors of produc tion) and outputs (pr oduction or economic results) were estimated. Three aspects a dvocated by Gastal (1 980), Mafra (1988), and Barros and Estcio (1972) were utilized as guidelines for my economic assessment: 1) the community land tenures as a system consti tuted by social, economic, and ecological components; 2) an understanding of the extract ive reserves, agro-extractive settlements, and colonist settlements as systems of production defined by bi ophysical and social dimensions; and 3) a view of the household as an economic unit with in each land tenure system and constituted by production and management goals. These considerations required that inputs and outputs should be m easured to capture the uniqueness of each household within each community and land tenu re type. The first step of the economic analysis was based on income, cost, and prof itability by type of product at the household level. In the following section, I discuss these economic measures. 4 By ex post economic analysis I mean to say that household economic assessment was performed based on household economic activities carried out prior to the survey. The 2004 household survey covered economic activities carried out from June 2003 through July 2004.

PAGE 69

57 Gross Income ( GI) is defined here as the result of selling a product in the market, and acts as an indicator of the scale of pr oduction. It also includes family consumption from crops, livestock, and forest products. In addition, it reflects th e land use strategy of each family. Crops and cattle are expected to represent the major components for income of colonist residents (Peixoto community ), while timber and non-timber products are expected to be most important for Porto Dias and Chico Mendes respectively. Gross Income is calculated as follows: GI = Qm pp (3.1.) Where, Qm = qv + qe + qc (3.2.) Qm=quantity of the product destined for the market qv=quantity of the product sold qe=quantity produced in prev ious years and not sold qc=quantity produced or harv ested consumed by family qp=unit price of the product Total Cost (Tc) is defined here as the total m onetary value spent by the household to produce or harvest specific products. It is derived by summing two components: the variable costs (Vc) and the fixed costs (FC). In the household anal ysis addressed by this study, labor is the main element in Vc. The total fixed cost is distributed between values that correspond to the determined product i and is made up of the specific fixed cost for a determined product i (EFc) and the common cost of i products, called the fixed common cost (CFc). Fc n i i Fc n i i C CC e V T 1 1 (3.3.)

PAGE 70

58 Where, TC = Total Production Cost VC = Variable Production Cost of product i eFc = Specific Fixed Cost of product i CFc = Common Fixed Cost of product i. To evaluate the economic viability of the household production system, this study also measured the profitability of rubber tappers and colonists economies categorized by products. I therefore generated a profit (P) analysis to compare cat egories of products and to correlate their profitab ility with deforestation. Profitability also allows verification of the possibility of a financial gain, and consequently accumulation at the family level. In contrast to some other studies which used a type of analysis based only on gross income and total cost (Vosti et al. 2003), this work included family subsistence into the gross income. The calculated formula for profitability is, R = GI TC (3.4) According to the results, I may observe three specific situations: R > 0, where there is a gain R < 1, where there is a loss R = 1, break-even Economic Analysis The economic analysis of the study was based on primary data from 2004 and focused on economic outcomes of land use strategies utilized by families in each landtenure type. For comparative purpos es, I divided the data analysis into three parts. First, I performed an economic analysis at the commun ity level. During this phase I divided the

PAGE 71

59 data into income and cost outcomes aggregat ed by products. In order to evaluate these results I utilized mean, standard deviati on, and skew. Second, I car ried out a similar analysis by land tenure types aggregated by crops, livestock, non-timber, and timber activities. During this phase I used F-tests results to evaluate wh ether differences of means among sites were signifi cant. And third, I also perfor med a profitability analysis by land tenure types. Land Cover Data Processing In addition to the socioeconomic survey, re mote sensing techniques were utilized in order to evaluate the land cover distribu tion within each study site. The land cover processing performed in this study is presen ted in Figure 3-1. Usi ng georeferenced land cover samples collected in 2004 and 2005, I pr oduced a land cover signature based on 2004 images for four classes: forest, crops, pa sture, and regrowth. I then utilized this signature to perform a superv ised classification on a mosa ic of 67-001 and 67-002 scenes. Two hundred and twenty-one training sample s of 442 collected in 2004 and 2005 were selected through stra tified random sampling for image classification. The remaining 221 training samples were used to evaluate the accuracy of my classi fication. All analyses were performed using Erdas Imagine softwa re version 8.4, ArcView GIS version 3.2, and SPSS version 11.0. Landsat images processing was carried out following various steps (Figure 3-1). The first procedure to assess land cover was to calibrate th e images (Green, 1999). I also performed band by band atmospheric correction of the 2004 images as discussed in Jensen (2005). I used a layer-stack to merge the five bands for analysis of each scene in separate .img (Erdas) files. Next I performed pre-processing tasks and image-to-image geo-referencing was done based on a 1995 im age as a reference (Jensen, 2005). Once I

PAGE 72

60 obtained the georeferenced, calibrated, atmosphe rically-corrected, and merged images for 67-001 and 67-002 scenes, I performed a superv ised classification of the 2004 image. Figure 3-1. Diagram of land cover proce ssing in the study area, Acre, Brazil. The classification for 2004 was submitted to two types of assessment. The accuracy assessment was conducted based on 221 trai ning samples collected during 2004 and 2005. Furthermore, a field appraisal of this classification was carri ed out in June 2005. Over two weeks I visited each of the three study sites and assessed the accuracy of the 2004 classification, comparing onthe-ground observations with a printed map of the land cover results in each community areas With a GPS and via co nversations with local residents, I was able to validate the accu racy of each land cover class within each community area. After the accur acy assessment and field appr aisal of the classification results, I identified the amount of area cove red by each class at the household level. These results were incorporated in the SPSS economic database. Since this study addresses a socioeconomi c and land cover analysis at household level, I created a border for each familyÂ’s landholding and analyzed it separately. In Individual 2004 Landsat band files (1,2,3,4,5) Layer stack Merged 2004 Landsat Image Image to image georeferencing AcreÂ’s georeferenced 1996 image Georeferenced 1998 Landsat Image Georeferenced/ calibrated/atm. corrected 2004 Image Supervised classification Training samples Classified 2004 Image Accuracy assessment Training samples smallholders 2004 land cover types Image subset Landowners boundaries Images error assessment SPSS Processing Image calibration Atmospheric correction Individual 2004 Landsat band files (1,2,3,4,5) Individual 2004 Landsat band files (1,2,3,4,5) Layer stack Merged 2004 Landsat Image Image to image georeferencing AcreÂ’s georeferenced 1996 image Georeferenced 1998 Landsat Image Georeferenced/ calibrated/atm. corrected 2004 Image Supervised classification Training samples Classified 2004 Image Accuracy assessment Training samples Accuracy assessment Training samples smallholders 2004 land cover types Image subset Landowners boundaries Image subset Landowners boundaries Images error assessment SPSS Processing Image calibration Atmospheric correction Image calibration Atmospheric correction

PAGE 73

61 Peixoto, I utilized the georeferenced plot boundaries collected by INCRA for each local resident. Based on these digitalized lots I identified the GPNI community boundaries as well as the properties selected in this study on the classified Landsat image (Figure 3-1). Land cover analyses were performe d on those selected residents. In the Chico Mendes and Porto Dias comm unities, the limits of each householdÂ’s landholdings were not as easy to identify. Co mmonly, rubber tappers utilize rubber-trails (estradas de seringa) as their natural boundaries. CNPT/IBAMA and INCRA, the institutions responsible for those land te nure types, have had difficulty establishing boundaries among smallholders of extractive and agro-extractive reserves. After an intensive literature search for studies w ith similar problems in delineating household property boundaries, I decided to ut ilize the Thiessen polygon algorithm5 based on fuzzy set theory to produce boundaries for each colocao on these rubber estates. Using the algorithm discussed above, I wa s able to define the boundaries among each household property following three procedur es. First, I used the location of the householdÂ’s home within their colocao, available from CNPT /IBAMA and INCRA, to identify each family area. In the Chico Mendes Resex, I only selected colocaes within the So Pedro, Palmari, and Floresta seringais. In the Porto Dias case, there were no borders among seringais, and I made use of all georeferenced colocao points. Second, I 5 This is a remote sensing technique in which an indi vidual area of influence around each set of points is defined. Based on fuzzy theory, the polygonsÂ’ boundaries are the area that is closest to each point relative to all other points. They are mathematically calculated through two sequential steps. First, the entire area is divided into polygons in which one polygon per point is established. Second, the distance among points indicating colocaes area is minimized. However, in the broad sense, the fuzzy logic utilized in this approach to estimate boundaries of colocaes is a process of inferring imprecise conclusions from imprecise premises. As pointed out by Dragi evi (2005) this deduction is approximate rather than exact because data described and generated under the ci rcumstance of lack of information and accuracy assessment implies a high cost. Since my summer time in Acre was not sufficient to assess the accuracy of this technique within each family ar ea, I was only able to evaluate the boundaries of two smallholders within of each land tenure type.

PAGE 74

62 produced colocao polygons using the Thiessen algorithm in ArcView based on fuzzy set theory. Third, I sele cted polygons of sampled colocaes, which were georeferenced during the 2004 fieldwork, in both Chico Mende s and Porto Dias totaling 61 family areas. Economic and Land Cover Integration Economic data from the household surveys we re integrated spatially in a GIS with satellite data for the land cover classes for each corresponding property polygon in the three study sites. To identify th e main factors related to defo restation in the study sites, a multivariate regression analysis was carried out. Taking into consideration the limited sample size in this research, this analysis was performed for all sites (88 households). During this part of my analysis, I cons tructed five regression models based on different theoretical foundations. The first model evaluated the effect of land tenure category for deforestation. The second was a life cycle model in which I assessed the effects of family dynamics on deforestation in the study sites. The third was a market integration model in which I evaluated the in fluence of market access and family cash purchases on land cover dynamics. The fourth model assessed the effects of cost composition for deforestation among extractivist, agro-extractivist, and colonist areas. Finally, in the last model I estimated the im portance of assets for land cover dynamics in the study sites. Conclusion This thesis seeks to establish linkages be tween economic and fore st cover datasets of three community land tenures types in southwestern Amazonia. I utilized a combination of methods to link socioeconomic and satellite data. First, to generate an understanding of distinct land tenure categor ies, a questionnaire was generated via a collaborative participatory process. Second, I conducted th e economic field survey of

PAGE 75

63 households in the three study sites, recognizing the differences in the rules of each land tenure type concerning resource management. Third, the classifi cation of satellite images for each study site was based on the four basi c land cover types existent within of the communities: forest, forest regrowth, pasture, and crops. Fourth, the data integration was carried out by defining a common unit of anal ysis for both economic and remote sensing approaches, in order to unify the data spat ially to allow for a multivariate regression analysis.

PAGE 76

64 CHAPTER 4 ECONOMIC AND LAND USE COMPARISON OF SMALLHOLDER LAND TENURE MODELS IN TH E STATE OF ACRE Introduction This chapter responds to the need for an empirical assessment of how community land tenure models in Acre differ in term s of their economic and land cover outcomes. Specifically, the chapter concentrates on how economic variables affect land cover change among households within Resex, agro -extractive settlements, and colonist settlements. In order to address these issues, this ch apter has three objectives. The first is to examine crop, pasture, regrowth, and forest ar eas at the household le vel and to determine the extent to which timber and non-timber forest management as well as agrosilvopastoral strategies lead to different amounts of deforestation. The second objective is to estimate the economic efficien cy of each land use strategy implemented by inhabitants of each study area. Finally, the third objective is to evaluate how microeconomic factors are related to land cover di stribution. Other non-econom ic factors, such as family composition, age, distance to ma rket, and length of residence are also considered to explain land c over at the community level. This chapter is organized into three parts. Part one offers an overview of land cover distributions in each study site, in terms of forest, crops, pasture, and forest regrowth. Part two introduces the economic analysis by evaluating household income, costs, and profitability among tenure types and income sources, particularly pointing out the

PAGE 77

65 differences between timber and non-timber forest management. Finally, part three presents a multivariate statistical analysis of the effects of land tenure, economic variables, and non-economic factors on land c over change. The chapte r concludes with a discussion of the main findings. Revising the Research Questions The first research question posed by this study is: how do economic and land cover outcomes compare among the three types of land tenures? The land cover change literature includes studies of land tenure, but studies comparing land cover outcomes among various land tenure types in a given re gion remain very rare. The importance of designing land tenure institutions as an influence on economic and environmental outcomes therefore remains poorly understood. Th is limitation in the land cover change literature is all the more significant because the design of land tenure institutions is something that governments can control, and th e diversity of land tenures in Acre, Brazil, reflects many policy experiments to yield better economic and envi ronmental outcomes from land use. For precisely th is reason, there is great inte rest in Acre, in Brazil and elsewhere, and this interest motivates my first research question in this thesis. The second question addresses an economi c assessment of non-timber and timber forest products among small producers. Since the creation of Resex and agro-extractive settlements, their inhabitants and a series of development programs have concentrated on the economic use of a range of non-timber fore st products. More recently, timber forest management projects have been introduced in several of these areas. The shift from neoextrativismo to florestania is significant not only for the potential consequences on land and forest resource use in different tenure types in Acre, but also for the profitability of timber and non-timber forest products. That said, over the past years an intensive

PAGE 78

66 debate has emerged focusing on potential econo mic risks associated with timber and nontimber resource use within community areas in Acre. Furthermore, economic outcomes from the recent emphasis on timber extraction are a key factor influencing smallholder land use decisions. Therefore, my second resear ch question focuses on the profitability of timber and non-timber resource use. The third question focuses on the factor s best explain fore st clearing among households in Resex, agro-extrative settleme nts, and colonization settlements. Although economic goals have been considered a key co mponent in influencing land use strategies within community land tenure ar eas, various income-defores tation tradeoffs can emerge. For instance, smallholders can use profits to in vest in pasture for cattle, which is a laborsaving strategy utilized by families, but ranching results in more forest clearing. However, by the same token, if forest manageme nt is economically viab le (profitable), it can also stimulate local inhabitants to use the net income to invest in activities that imply to increase deforestation. In working with each of these three que stions I have developed one or more specific expectations (hypotheses): H1: Extractive reserves such as the Chico Mendes Resex will have the most favorable environmental outcome (lowest proporti on of land deforested per household). Colonization will have the best economic outcome (highest gross household income). Agro-extractive settlement will have the best overall outcome (deforestation nearly as low as Resex, and income almost as high as colonization settlement). H2: Non-timber forest products will be more profitable, because of their lower costs, than timber extraction.

PAGE 79

67 H3: Higher profitability lead s to greater deforestation, though land tenure categories will also be important. Land Cover Distribution among Land Tenures This section presents a land cover asse ssment of household landholdings in the three study sites. These results are partic ularly important to evaluate the policy experiments concerning these different la nd tenure models in terms of land cover outcomes. This study thereby provides a co mparative understanding of how land cover varies among the Resex, PAE and PC I visited. In this section I provide the results for my supervised classification of the three land tenure types. Fi rst, I discuss the accuracy assessment. Second, I compare th ree study sites in terms of their land cover distribution. Accuracy Assessment of Land Cover The accuracy assessment evaluates the precis ion of the supervised classification in identifying land cover categories among the study sites. Table 4-1 pres ents the results in the error matrix for the classes of forest, crop land, pasture, and fore st regrowth from the supervised classification. The matrix illustrates both errors of exclusion (omission errors, i.e. degree of Producer’s accuracy), which measures how much of the land in each category was classified correctly, as well as errors of inclusion (commission errors, i.e. degree of User’s accuracy) that measures the pe rcentage of correctly classified pixels in each land cover category of the images (Jense n 2005). The overall accuracy is expressed as a combined percentage of the test-pixels su ccessfully assigned to the correct classes. The results indicate that the classificati on had a high degree of accuracy for the land cover classes ‘crops’, ‘regrowth’, and ‘pasture,’ with 89-96% producer’s accuracy. However, the accuracy for correctly identify ing forest areas was lower, with a 74% producer’s accuracy, which likely resulted fr om the difficulty in classifying bamboo-

PAGE 80

68 dominated forested areas in the study re gion, which can easily be confused with regrowth. Nevertheless, the user Â’s accuracy results for pastur e and crops indicate that my 2004 classification was conservati ve, implying a potential bias toward classifying these land cover categories as forest, resulting in an underestimation of the extent of deforestation. On the whole, one might conclude that because the overall accuracy of the entire classification was quite high (87%) and the producerÂ’s accuracy varied from 90 to 96%, this classification is adequate for anal yzing forest cover in the study areas. Table 4-1. Accuracy assessment for 2004 superv ised classification for the study sites in Acre, Brazil. Class name Reference totals Classified totals Number correct ProducerÂ’s accuracy UserÂ’s accuracy Unclassified 9 70 Forest 39 3329 74.4% 87.9% Pasture 36 4232 88.9% 76.2% Crops 49 5847 95.9% 81.0% Forest regrowth 88 8180 90.9% 98.8% Totals 221 221193 Overall Classificati on Accuracy = 87.3% Comparing Land Cover among Land Tenures Land cover analysis was performed for th e households sampled in the three study sites, each corresponding to a different cate gory of community land tenure. To perform the analysis, it was necessary to identify or estimate the boundaries of land claimed by the sampled households. While boundary data regarding the division among household lots in the Peixoto colonization project we re available from INCRA, the geographic limits of the colocaes in both the Chico Mendes Resex and the Porto Dias PAE had to be estimated using the Thiessen polygon algorithm based on fuzzy set theory.

PAGE 81

69 Before applying this estimation method to the whole data set, the methodology was tested in 5 colocaes in the Porto Dias community. Duri ng that test, rubber tappers of the Porto Dias indicated that their areas had between 3-5 rubber trails1. This results in overall estimates of 300-500 hectares pe r household landholding. Using a GPS unit and guided by rubber tapper household member s and their neighbors, I measured colocao boundaries and on the basis of th ese measurements estimated the total colocao areas to be between 230-560 hectares, indicating a wider range of area than conventional estimates. The colocaesÂ’ shapes resulting from using the Thiessen polygons based on fuzzy set theory were compared to the field da ta and indicated that they were similar to the limits identified with the GPS tracking points. An important implication that can be derive d from these results is that deforestation levels for each rubber tapper area in the Chico Mendes and Porto Dias, if based on the conventional 100 hectare/trail assumption, are likely to be underestimated for some families and overestimated for others. Hence, most previous deforestation estimates and deforestation controls at household levels have likely not been very accurate. This study therefore presents a first attempt to overcome this problem. Using land cover classifications and area l estimation figures for all three study sites, I obtained the results presented in Tabl e 4-2, which shows descriptive statistics for each land cover category. In order to identify the most appropriate metrics to evaluate land cover distributions, I present results for: raw values, percentages, and natural log values. Land cover was categorized in terms of crops, pasture, and regrowth. In addition, I calculated deforestation category summ ing crops, pasture, and regrowth. 1 Rubber trails have been conventionally estimated as covering roughly 100 ha each

PAGE 82

70 Table 4-2: Land cover categorie s at household level, all si tes (n=88), Acre, Brazil. Land cover category Hectares Percentage Log value Deforestation Mean Standard Deviation Skew Crops Mean Standard Deviation Skew Pasture Mean Standard Deviation Skew Regrowth Mean Standard Deviation Skew 29.94 22.45 1.55 8.46 6.31 2.35 20.20 17.32 1.00 1.28 1.68 2.62 16.59 14.46 .68 4.55 3.87 1.22 11.39 11.21 .79 .65 .87 2.24 3.11 .82 -.49 1.92 .69 -.23 2.48 1.21 -.704 -.38 1.27 -.07 While no land cover types presented a high st andard deviation value for all types of measurements, the skewness indicated some deviation from normal-distributions. Skew values close to “0” suggest a normal distri bution. On the one hand, the skew for mean raw and percentage values was positive, which means that their distributions tend to have extreme high-values. On the other hand, skew va lues for logged values were all negative, which indicates that this di stribution tends to have extreme low-values. Based on the findings in Table 4-2, for the remainder of the analysis I focus on the percentages. This is because they have low standard deviations a nd skew values, something lacking in the raw values. Moreover percentages are easier to understand and compare than raw or logged values.

PAGE 83

71 Crops and pasture are the most important land cover categories de forestation in the three study sites. Deforestation overall exhib its a mean of around 30 hectares (~ 17% of total area) per household. This result might i ndicate a total level of deforestation above the 10% authorized by IBAMA and INCRA within Resex and agro-extractive settlements. However, this interpretation ignor es potential differences in the extent of deforestation among land tenure types. The lowest values are for the regrowth cla ss, with a mean of less than 1% of the total area. This result can be explained by the importance of crops and pasture to smallholders. Commonly smallholders tempor arily abandon productive areas to regrowth for 2-5 years, called the pousio period, then they clear, bur n, and reuse the regrowth for crops or pasture (Amaral 1998). The most important results in Table 4-2 ar e twofold. First, per centages are the most appropriate measure to evaluate the distribut ion of land cover in the study sites. Besides the low standard deviations and skew values percentages allow for direct comparisons among the land tenure categories and are easier to interpret. Second, the results show that pasture is the main land use category that co ntributes to overall deforestation. However, the importance of crop cultivation and cat tle ranching may vary among land tenure categories. To assess this possibility, the ne xt step in this analysis compares the percentage of land covers among the three land tenure types. Table 4-3 provides a comparison of land covers among land tenure types in the three study sites. I present Ftests for the significance of differences among the three tenure types. Deforestation percentages are significantly different across land tenure types (p<.01). Although colonist families in Peixoto PC have made efforts to establish

PAGE 84

72 sustainable agrosilvopastoral systems, th is area exhibited th e highest level of deforestation (~ 36%). In contrast to Pe ixoto, the three rubber estates of the Chico Mendes Resex selected in this study exhibited the lowest le vel deforestation percentage (~5%). And in the Porto Dias PAE, th e households sampled had deforestation percentages in between the othe r two land tenures (~12%), but cl oser to the percentage in the Resex than the PC. Table 4-3. Descriptive statis tics for land cover categories, by land tenure type, Acre, Brazil, 2004 Land cover variables Chico Mendes Resex (N=34) Porto Dias PAE (N=27) Peixoto PC (N=27) F-tests Deforestation (%) Crops (%) Pasture (%) Regrowth (%) 4.99 2.32 1.99 .68 12.27 3.13 8.87 .25 35.52 8.75 25.76 1.00 181.543** 48.555** 169.111** 5.560* p<.05, **p<.01 In addition, the percentage areas under crops and pastur e are also significantly different among the three tenure types represen ted in the study site s (p<.01). On the one hand, pasture is the land cover th at contributes most to defo restation in both the Porto Dias and Peixoto sites. In the Chico Me ndes Resex, on the other hand, most forest clearing was for crop production. This indica tes that while rubber tappers of Chico Mendes are focusing their efforts on agricu lture, possibly for th eir subsistence and production of cassava sale, cattle raising emerges in Porto Di as and Peixoto as the most important activity contributing to deforestation. The F-Tests result indicated that the differences among mean percentage for regrowth across land tenures are significant (p < .05), the differences are not as pronounced as for the other land cover categor ies. Porto Dias PAE seems to be the

PAGE 85

73 community with the lowest level of this land cover category (0.25%). In the Chico Mendes Resex, this mean percent value is 0.68%. As regrowth is an area of reserve for potential future use, its extent can be interp reted as an indicator that other areas within landholdings might be been utilized for crops or pasture purposes. In her research in Acre, Ehringhaus (2005) observe d that the overall percent of regrowth area is decreasing due to two parallel processes: 1) as residents invest more in ranching, they are converting young regrowth into pastures, and 2) they are co nverting agricultural fi elds directly into pasture and not letting them la y fallow for the development of new regrowth. Moreover, she noted that residents are also increasingl y establishing new fields in primary forests and less in regrowth areas. These results provide important insights to evaluate the effect iveness of policies underlying each land tenure category with regard to its environmenta l conservation goals. While the Chico Mendes Resex and the Porto Di as PAE were created to establish land concessions for sustainable forest use, the Peixoto PC emerged as an area for agricultural land use, and not forest conservation. The policy goals of coloni zation projects, agroextractive settlements, and extractive reserv es are thus very ev ident in the higher percentage of deforestation in Peixoto th an in Porto Dias and Chico Mendes. This suggests that the land use regul ations institutionalized in ag ro-extractive settlements and extractive reserves, at least as practiced in Porto Dias and Chico Mendes, are effective in reducing deforestation relative to that s een in colonization areas such as Peixoto. This section evaluated land cover in the Chico Mendes Re sex, Porto Dias PAE, and Peixoto PC. Findings from remote sensing indi cated differences betw een the three sites, especially with respect to deforestation fo r pasture and agriculture. The different land

PAGE 86

74 cover profiles of the three tenure categories raise questions about the conservation value of these areas and the household economic decisions being made in the context of different land tenure regulati ons concerning resource manageme nt. In the next section, I therefore address such questions with an analys is of income, costs, and the profitability of different economic activities in the three study sites. Economic Analysis for Land Te nures in the State of Acre This section offers insights into sma llholder land use strategies based on three economic criteria: income, cost, and profitability. First, I will describe the data regarding cash income generation across the three land tenure types. Second, I will compare costs of production for different products. Third, I wi ll concentrate on a prof itability analysis for all product types by category of land tenure. These results will allow a comparison across land tenure types as well as categor ies of products. As forest policies and institutional goals w ithin these land tenure models ha ve sought to enhance economic welfare among households, this analysis can pr ovide an evaluation of the effectiveness of the strategies behind policies constituting diffe rent land tenure types. In particular, this section will compare the three types of altern ative land use strategies implemented in the three land tenure models: timber, non-timber, and agrosilvopastora l production. I will evaluate these strategies economically in terms of their gross income generation, production costs, and profitability, may be quite dissimilar. In addition, I will address the issue of whether these economic outcomes may vary so as to reveal tradeoffs with respect to land cover outcomes. Gross Income Table 4-4 compares gross household inco me among the three land tenure type. Standard deviations were low in most land tenure types. In addition, except for gross

PAGE 87

75 income from livestock within the Chico Me ndes Resex and the total income for Porto Dias PAE, all skew values were larger th an 1.0, which suggests a few “outliers” with high values. Logging is not authorized with in the Chico Mendes Resex by IBAMA and CNPT; therefore I report no income da ta from logging in this area. Table 4-4. Descriptive statis tics for annual income, by land tenure type, Acre, Brazil, 2004 Economic activity Chico Mendes Resex (n=34) Porto Dias PAE (n=27) Peixoto PC (n=27) Crops (US$) Mean Standard Deviation Skew Livestock (US$) Mean Standard Deviation Skew Non-timber (US$) Mean Standard Deviation Skew Timber (US$) Mean Standard Deviation Skew Total (US$) Mean Standard Deviation Skew 181.00 140.00 .33 268.00 176.00 -.11 681.00 427.00 .63 ---1,130.00 495.00 .32 311.00 150.00 1.78 979.00 723.00 1.84 805.00 467.00 .32 1,385.00 1,840.00 .57 3,567.00 1,497.00 -.12 1,763.00 1,433.00 .77 1,931.00 1,879.00 1.19 28.00 61.00 3.00 153.00 736.00 5.15 4,109.00 2,759.00 .71 Three noteworthy differences emerge in Ta ble 4-4. First, as expected, Peixoto had the highest mean total gross income (~ US$ 4,100.00 per year). Livestock as well as crops are the most important products to th e households of this area. Given the high

PAGE 88

76 deforestation percentage in Peixoto, this result indicates an economic-environmental tradeoff, which also casts doubt on the co mmitment to sustainability on which agrosilvopastoral systems were based. In fact, during the fieldwork in 2004, no small farmers reported any income from agrofore stry or from a combined crops-forestry system. Second, households in the Chico Mende s Resex had the lowest mean total gross revenue (US$ 1,130.00 per year). Given the low deforestation percentage in the Resex, this finding also implies an economic-envir onmental tradeoff. But third, Porto Dias stands out, because residents of this area generated a mean gross income of US$ 3,570.00 per year. This is a key finding, because resour ce use in Porto Dias reflects a combined forestry and agrosilvopastoral model, wh ich yields a mean gross household income nearly as high as in the Peixoto colonizati on project, with a deforestation percentage nearly as low as in Chico Mendes Re sex, via a diversified production system. I suggest two possible explanations for differences in cash income generation across land tenure types. First, a householdÂ’s ability to mix multiple income generation strategies contribute s considerably to th e access to various market niches. Second, a combination of prices incentives, distance to market, and local demand may also be key factors affecting these communities. These two interpretations were confirmed during the fieldwork. Rubber tappers of the Chico Mende s Resex concentrated their efforts on the extraction of rubber, a product with lower pr ices, and on harvesting Brazil nuts that bring high prices. In Porto Dias PAE, families did not explore rubber and preferred to devote their efforts to products with hi gh prices, such as Brazil nuts, copaiba oil, forest seeds, and medicinal plants. Peixoto is an area with higher crop and cattle demands from large

PAGE 89

77 ranchers and urban areas, which may have c ontributed to a higher gross income from crops and livestock activities. With the exception of Peixoto, the results confirmed the economic importance of land use strategies based on non-timber and timber production. In the Chico Mendes Resex, non-timber forest management was the la nd use that provided the largest share of mean gross income (~ US$ 690.00 per year). Timber (~ US$ 1,385.00 per year), on the other hand, was the product with the largest share of mean income among rubber tappers of Porto Dias. Further, the mean income obtained from non-timber forest products (~ US$ 800.00 per year) among residents of Porto Di as was higher than income obtained by families in Chico Mendes. Although sustainable forest uses were confir med as the most important activity for rubber tapper communities in Chico Mendes and Porto Dias, the results also indicated livestock as an important activity for all areas. Among residents of Chico Mendes and Porto Dias, this was the second most importa nt activity, with approximately US$ 270.00 and US$ 980.00 of mean gross income per y ear, respectively. In Peixoto, however, livestock was the most impor tant source of income. These results indicate that the residents of ‘sustainable ’ land tenure models such as Resex and agro-extractive settlements rely to a higher extent on forest products for their income. While the average income of these fa milies is lower than those living in the colonization projects, the importance of fo rest-based production is clear. Another important conclusion from these results is that although colonist households of Peixoto have implemented an agrosilvopastoral model in their areas, livestock production remains the most important source of income. At the same time, cattle ranching is also gaining

PAGE 90

78 importance among rubber tapper households, raising new questions regarding the longterm ecological sustainability of Re sex and agro-extractive land tenure. General Cost Analysis While gross income is an important indi cator of economic success of a particular land use, the cost involved among the producti on types is also a key economic variable driving affecting land use decisions in th e different land tenure models. However, economic analyses of land cover change have rarely considered produc tion costs as well as gross incomes generated by land use at the household level. In th e following section I will first focus on cost composition, by providing first an overall cost assessment for households in my sample to identify the ove rall cost patterns. S econd, I will concentrate on a comparative cost assessment among the land tenure types. This discussion will also compare production costs among types of ac tivities carried out by households. This information is important for identifying the le vel of economic viability for each type of product, in the context of specific land tenure types. For instance, it may be useful to assess the viability of implementing agrofore stry system in the Chico Mendes Resex as well as in the Porto Dias PAE. Table 4-5 summarizes the composition of annual mean costs by economic activity for all three communities taken together. Three t ypes of costs are presented in this table. The first column in the table refers to the total cost for each economic activity. The variable costs are expenses that vary according to the volume and intensity of each type of input, such as quantity of labor time, salt for cattle, seeds for pa sture or crops, oil for vehicles, etc. Fixed costs, on the other ha nd, are costs that arise with or without production such as the depreciation of machines and equipment, and other investments in the property. Table 4-5 presents total, fixe d, and variable costs for different economic

PAGE 91

79 activities, namely crops, lives tock, non-timber forest products, timber, and all of these products together. Table 4-5. Descriptive st atistics for annual cost categories, all sites (n=88), Acre, Brazil, 2004. Economic activity Total Fixed Variable Crops (US$) Mean Standard Deviation Skew Livestock (US$) Mean Standard Deviation Skew Non-timber (US$) Mean Standard Deviation Skew Timber (US$) Mean Standard Deviation Skew Total (US$) Mean Standard Deviation Skew 497.00 373.00 1.83 593.00 622.00 1.86 453.00 442.00 .78 731.00 2,047.00 2.49 2,316.00 2,099.00 1.86 122.50 166.00 2.90 137.50 194.00 4.23 65.00 78.00 2.54 106.00 293.00 2.50 437.00 374.00 2.51 359.00 248.00 1.00 455.00 533.00 2.00 388.00 396.00 .86 626.00 1,755.00 2.50 1,878.00 1,859.00 1.87 As can be seen in Table 4-5, all economic activities had costs with relatively small standard deviations, except th e total and variable costs for timber. The result for timber was strongly influenced by timbe r production costs in Porto Dias. The skew value for the fixed costs for livestock were the highest am ong all categories (4.23), which suggests that the cost values for this category tend to be higher rather than lower. The non-timberÂ’s skew value for total costs, on the contrary, was the lowest value (0.78). Overall, skew

PAGE 92

80 values indicate that the results for all categ ories of products across all cost types tend to be high, indicating “outlier” households th at spend considerab ly more on certain economic activities. All cost types for timber showed high sta ndard deviations, which reflect the high costs of timber management. Certification, tr ansport, road repair s, depreciation of equipment, and a skilled labor force are the e xpenses that have contributed most to higher costs for timber extraction. Three main insights can be drawn from Table 4-5. First, variable costs are the main category of costs for all productive activities Among smallholders in Acre, family labor is the main component of variable costs, a reflection of the importa nce of household labor inputs and the labor-intensity of many of the activities in Table 4-5. Families in the three study sites reported that all family members, including children, adults, and the elderly are engaged in all production activities. Second, timber extraction is ve ry costly and not as econom ically attractive as its high gross incomes seen earlier might sugge st. The mean total cost (~ US$ 730.00 per year) for timber extraction was higher than for livestock (~ US$590 per year), or any other activity. As a result, when rubber tappers compare timber with other economic activities, they are more likely to invest in non-timber forest produc ts or cattle raising instead of timber harvesting. Nevertheless, this may depend on the price incentive from each economic activity. For instance, while the m ean farm gate price for cattle in these areas was US$ 143.00/head, during the same period the price for timber was US$ 115.00/m3. Later in this chapter I will undertak e a profitability analysis, which will further explore this issue.

PAGE 93

81 Third, although timber exhibited the highest total and variable costs, crops and livestock were the activities with highest fixed costs. Two reasons account for this finding. In recent years many households, espe cially in Peixoto, have invested in equipment and assets for cattle and agriculture such as fences and equipment, supported by BASAÂ’s (the Amazonian Bank) credit progr am. This contributed to high depreciation costs for Peixoto households. Rubber tappers participating in the timber forest management, on the other hand, rented skidde rs and other machines from third party companies, thus reducing the costs of depreciation. Three other conclusions may be drawn from the costs analysis in light of household financial well-being as well as the earlier anal ysis of land cover. Fi rst, during the survey the families pointed out that their systems are highly dependent and limited by family labor availability. Second, approximately 80% of the total cost was constituted by variable costs (US$ 1,878.00), which reveals the weak capacity of families to make investments in technologies such as mach ines and other apparatus for reducing labor costs. This situation has been partially overcome with funds and subsidies from organizations such as BASA or the Acre gove rnment. However, this assistance is also likely to increase the fixed costs as well as increasing the demands on family labor that might have to be allocated to generate suffi cient income to repay the credit. The forest government has established programs to dona te equipment for rubbers tappers and colonist families to avoid this cycle of sma llholder debts. The third conclusion relates to the implications of costs for household land use decisions and land cover changes. Both rubber tappers and colonists consider total co sts and labor requirements in their land use activities. In the absence of a family labor reserve and in the pres ence of high production

PAGE 94

82 costs, many households may move toward activi ties that are less labor -intensive, such as cattle raising. As result, this situation may contribute to increase the number of families in debt with fina ncial organizations. The issues that emerged from the cost anal ysis are key aspects for any evaluation of the effect of subsidy policies on household ec onomic welfare as well as land use and land cover change in the state of Acre. Economi c and conservation strategies, then, should take into consideration total costs and especi ally labor costs as key constraints for local producers. However, it is important to take into consideration the variability of costs among land tenure types designed to encour age different land use strategies. Cost Analysis at Community Level In this part I provide th e second part of my cost analysis, examining costs differences across land tenure types. It compar es costs for different economic strategies across households in extr ativist, agro-extractiv ist, and colonist areas. Special attention will be given to the costs of timber and nontimber forest production. In addition, this section aims to briefly discuss why the cost categories for each productive activity differ among land tenure types. These results will pr ovide insight on how costs for each group of products affect household land use decisions and land cover change. Table 4-6 provides mean fixed, variable, and total costs by la nd tenure type and type of product. As anticipated, there are not iceable differences in the production costs among the three land tenure types. F-tests result s indicate that differences of costs for all groups of activities are significant across sites. While Porto Dias was the community with the largest mean total cost (~ US$ 3,900.00 per year), total co sts in Peixoto were nearly as high, whereas the families of Chico Mende s Resex had the lowest relative costs. Although the mean total value for fixed cost was comparable for Peixoto and Porto Dias,

PAGE 95

83 there are differences in the cost compositi on among their productive act ivities, especially for timber. Moreover, it is apparent that th e variable costs comprise the greatest portion of total costs of all economic activities in all three areas. These results again confirm the importance of family labor for all land uses. Table 4-6. Annual mean for cost categorie s, by land tenure type, Acre, Brazil. 2004 Economic activity Chico Mendes Resex (N=34) Porto Dias PAE (N=27) Peixoto PC (N=27) F-tests Crops (US$) Total Cost Fixed Cost Variable Cost Livestock (US$) Total Cost Fixed Cost Variable Cost Non-timber (US$) Total Cost Fixed Cost Variable Cost Timber (US$) Total Cost Fixed Cost Variable Cost Total (US$) Total Cost Fixed Cost Variable Cost 401.00 86.00 315.00 140.00 64.00 76.00 714.00 117.00 594.00 .00 .00 .00 1,252.00 268.00 985.00 303.00 36.00 267.00 689.00 104.00 585.00 549.00 54.00 495.00 2,371.00 339.00 2,033.00 3,914.00 534.00 3,380.00 812.00 255.00 505.00 1,066.00 263.00 803.00 29.00 9.00 21.00 12.00 5.00 7.00 2,057.00 554.00 1,503.00 21.054** 18.089** 8.290* 27.676** 10.226** 22.795** 32.706** 22.325** 27.633** 32.706** 22.325** 27.633** 16.929** 6.423* 18.674** p< .05; ** p< .01 Table 4-6 indicates that live stock production costs varied considerably across sites. While the lowest total cost fo r livestock production was in the Chico Mendes Resex (US$ 140.00 per year), the costs were higher in the Porto Dias PAE (US$ 689.00 per year) and especially in the Peixoto PC (US$ 1,066.00 per year). These increasing costs can be

PAGE 96

84 explained by the relative importance of th is category of production for each community. Notably, pasture degradation has contributed to an increase in the a nnual cost of livestock production for families in Peixoto PC. Table 4-6 also shows that variable costs are the highest costs for all products across all sites. However, soil degradation has been related by colonist families as a key ecological factor negatively affecting the cost s of livestock production within the Peixoto PC. This is an important as it may indicate the potential long-term economic inefficiency of cattle ranching among small farmers in this region. There are two significant differences be tween the Chico Mendes Resex and the Porto Dias PAE. Although crop production wa s the second highest cost category among households of the Chico Mendes after non-timb er forest products, for the Porto Dias families the livestock production costs were the second highest after timber production. There are two likely explanations for these re sults. During the period of this research (2004), the majority of families of Porto Dias were clearing regrowth for pasture, and this task was carried out with family labor, which explains the relatively high mean variable cost for livestock. In contrast to this, many colocaes of the Chico Mendes Resex cleared forest for the production of annual crops which required inte nsive use of family labor as well. For the rubber tapper families, the highest production costs were associated with forest products. In the Porto Dias PAE, the highest production costs stemmed from timber production and to a lesser extent from livestock production. The forest management project implemented by the NGO CTA in this area since 1994 has introduced an entrepreneurial management plan that has involved high economic costs.

PAGE 97

85 Although over the past two years, rubber tappers part icipating in the tim ber initiative in Porto Dias have attempted to reduce costs, e xpenses related to tran sport, depreciation, and hired employees remain high, and represen t a great part of the total costs. This suggests the long-term economic challenged f aced by households participating in timber extraction. Non-timber forest products were th e products with the highest costs in the Chico Mendes Resex. Rubber remains a key activity for many families in this area, but the low concentration of rubber trees per hectare as well as intensive labor inputs required contributed to high variable costs. Although cost may be a key variable affecting household decisions, economic viability is the primary goal of all families. Thus, in the following section, as a third component in the economic assessment I anal yze the profitability of the different productive activities in different tenure areas. Profitability Analysis After analyzing both gross income and costs of different land use strategies within the three land tenure models, here I compar e both income and costs in a profitability analysis for all categories of products across land tenure types. This represents the most important step of the economic analysis b ecause it measures the economic effectiveness of each production strategy implemented by rural households in the study sites. The following analysis not only compares the economic viability of different types of activities, it also identifies which land tenur e type is most effec tive in yielding economic welfare. The results of the profit analysis will also allow us to link profitability to land cover dynamics among land tenure categories. Table 4-7 compares the profitability of crops, livestock, and timber and non-timber forest products among the three tenure types. I calculated profit as gross income minus

PAGE 98

86 total costs for a given activity for each tenur e type. Profit thus indicates the monetary gain or loss to households as a result of th eir economic decision-maki ng in the context of their land tenure category. Nega tive values indicate lost revenue in 2004, whereas positive values constitute profits accumula ted by households. F-tests indicate that differences across sites were significant. Table 4-7. Annual mean for profit for economic categories, by land tenure type, Acre, Brazil, 2004. Economic activity All (n=88) Chico Mendes Resex (N=34) Porto Dias PAE (N=27) Peixoto PC (N=27) F-tests Crops (US$) Livestock (US$) Non-timber (US$) Timber (US$) Total (US$) -125.14 -96.45 -10.22 -870.51 -86.71 -219.87 -132.48 -33.36 -122.54 -17.26 289.96 255.64 -917.21 56.52 -102.17 -213.31 -41.70 140.56 71.83 12.383** 3.597* 4.355* 4.250+ 12.729** + p <.10; p<.05; ** p<.01 The results indicate negative incomes for all economic activities when considering the three study sites as a whole. The total mean profit value was US$ 86.71, which suggests that overall the households sample d incurred an economic loss in 2004. While crops were the category of product with th e weakest economic efficiency (-US$ 125.14), non-timber forest products almost covered th e costs of producti on (-US$ 10.22). Timber, however, was the product with largest econom ic loss. These results indicate that agricultural crops and timber em erge as the products with the lowest economic viability, while non-timber forest products although also showing a loss nearly covered the costs of production. Nevertheless, when we consider profita bility among the three study sites, the findings are more mixed. In examining th e results by study area, the Peixoto PC and Porto Dias exhibit net economic gains overa ll. Their total mean profit value suggests

PAGE 99

87 some economic viability (US$ 71.83 and US$ 56.52 respectively). That said, the activities that proved profitabl e varied among land tenure cate gories. While in Peixoto, timber was the only profitable product, in Po rto Dias, livestock and non-timber forest products were the profitable activities. Crops and livestock were th e categories with the poorest economic performance among colonist s, and livestock production was the second least effective production within all the site s. Soil degradation may be a main factor contributing to increased costs of these activi ties in this area. The excellent profitability for timber extraction (US$ 140.56) in Peixot o in comparison to other activities is explained by its low expenses, in which the costs are reduced via the frequent involvement of local logging companies that pay for a certain amount of timber and provide their skidder to harv est and transport the timber. In contrast, the Chico Mendes Resex has th e lowest economic viability in terms of profitability. In great part, this poor econom ic performance is due to agricultural crop production, which yielded cons iderable financial losses (-US$ 219.87). This might be explained by the very high costs for forest clearing to establis h swidden agriculture. Rubber tappers still utili ze very simple technologies in th e form of machetes for clearing forests, resulting in very high labor costs. S econd, the costs of transport are an important component of the total costs among rubbe r tappers, who on average reside in colocaes around 30 km from the city of Xapuri. Production activities in Porto Dias PAE we re also profitable overall. Although its mean profit was inferior to that of the Peixoto PC, livestock and non-timber forest products demonstrated their ec onomic viability for rubber tappe rs in this area. However, the total costs of timber management, as pres ented earlier, are so hi gh that they greatly

PAGE 100

88 exceed the total gross income from timber. Therefore timber management in this study site compromises the overall economic success of these families. Three groups of factors help explain th is performance for Porto Dias. First, although the average price per head of cattle (US$ 150.00) in this area was similar to the price offered to colonist farmers of Peixoto, the rubber tappers of Po rto Dias have not had the problems with soil degradation found in Peixoto, thus reducing total costs. Second, for non-timber forest products, rubber tappers in Porto Dias concentrated their efforts on Brazil nuts, copaiba, and forest seeds, which are produc ts with relatively high prices and significantly lower produc tion costs than rubber. Third, as po inted out earlier, the type of timber forest management implemented in Po rto Dias, which follows an entrepreneurial management model, has resulted in high operational and implementation costs. This section concludes my economic anal ysis in which I first compared gross income across land tenure types and land uses. I also estimated how fixed, variable, and total costs were distributed among products a nd across communities. Finally, I integrated cost and income data into a profit analysis to evaluate the economic viability of each productive activity across sites. While pub lic polices favoring extrativist, agroextractivist, and colonist house holds have focused on interven tion strategies to generate cash income, the results of this study indicate that the high costs associated with many activities severely limit their economic vi ability. Not only has the timber forest management project in Porto Dias shown in ferior results due to high costs, but the project-driven agrosilvopasto ral system in Peixoto has not yielded favorable economic outcomes for colonists. In this context, non-ti mber forest products do marginally better, almost reaching break-even, however, thes e activities are also unprofitable.

PAGE 101

89 Policies that have been implemented in Resex, agro-extractives reserve, and colonist settlement projects in Acre over the past seven years have attempted to foster land use models that have a better econo mic performance and conserve the natural resource base. However, this study finds that crops and livestock have been an economic strategy employed by rubber tappers families. Moreover, colonist families in Peixoto who have implemented agrosilvopastoral system as an alternative to bring economic gains with lower deforestation have failed. Thes e findings seem to suggest that economic income-deforestation tradeoffs have emerged in these alternative land uses supported by the forest government. Integrating Land Cover and Economic An alyses for Understanding Land Tenure In this section, I evaluate economic a nd non-economic factors that may affect deforestation in the three study sites. Policymakers in Acre ha ve over the past seven years sought a comprehensive approach to balan ce both economic welfare and environmental conservation, but there may be tradeoffs between those goals. In particular, many economic policies have sought to stimulate sp ecific land uses asso ciated with higher levels of profitability as we ll as less deforestation. But th ere may also be important noneconomic factors, such as land tenure types and family labor availability dictated by household life cycles that may also be impor tant for understanding deforestation. In order to address this issue, I conducted a multivar iate regression analysis to examine the various factors that might be related to la nd cover. In the next section I focus on the results achieved from various regression models. In order to gain insights on how economi c and non-economic variables affect the land cover variables, I constructed several multivariate models. I employ several models rather than one due to limitation of sample size restricted the nu mber of independent

PAGE 102

90 variables that could be included in a given model. I developed five models, each derived from a different theoretical position, as a means of identifying key factors affecting deforestation among households in the study site s. I will concentrate my interpretation on policy implications about the significant drivers affecting defo restation in Acre. Table 4-8 presents the first model (i.e., M odel 1). It attempts to evaluate whether differences in land tenure regulations among the study sites account variation in deforestation. The Chico Mendes Resex and Po rto Dias PAE were compared to Peixoto PC, the reference group. As expected, the results pr ovided by this model suggest that deforestation is significantly influenced by t ype of land tenure. The R2 equals 0.81, so tenure type explains 81% of the variation in deforest ation among households in the sample, and the F-test is statistically significant. Coefficien ts for the Chico Mendes Resex and Porto Dias PAE were negative, which implies that in co mparison to Peixoto, those categories of land tenures exhibit lower percen tages of deforestation. Among all sites, however, Chico Mendes seems to be the area with the lowest deforestation percentage, and thus greater efficiency to address forest conservation. The second column of Table 4-8 presents the models for crops regressed on land tenure categories. The model is also very significant (R2 = 0.53) and the largest difference is between Peixoto PC and Porto Dias PAE, which had the smallest crop areas. The third column presents the pasture model in which the largest difference is between Peixoto PC and Chico Mendes Resex. This mo del is also very strong (R2 = 0.80), and suggests that most of the differences in deforested areas across land tenures types are due to differences in pasture areas.

PAGE 103

91 Table 4-8. Site model at land cover type s, all sites (n=88) Acre, Brazil, 2004. Dependent variables Independents variables Deforestation Crops Pasture Regrowth Constant Chico Mendes Resex Porto Dias PAE (Reference Peixoto PC) 35.515** -30.524** -23.245** 8.749** -.6423** -5.609** 25.759** -23.769** -16.883** 1.007** -.332 -.752* R2 .810 .533 .799 .116 F ratio 181.543** 48.555** 169.111** 5.560* p<.05; ** p<.01 The land use regulations in each land te nure is an important explanation of variability in land cover. Federal Decree 2.166 /01 and other Brazilian laws regulate that in colonization lots families are permitted to deforest up to 20% of their area, while in colocaes of Resexs and agro-extractive settleme nts, this percent is only 10% of the total area. Land use management plans a pproved for Chico Mendes established that families are authorized to limit pasture areas to 5% of their total area. INCRA, on the other hand, instituted that deforestation in Porto Dias is limited to 10%, and may not exceed 30 hectares per landholding. Table 4-8 suggests that policies attempti ng to address forest conservation through land tenure regulation may take into consider ation the category and institutional approach addressed in this process. Although various scholars (Wallace 2004, Gomes 2001, Sassagawa 1999) have pointed out the incr easing deforestation in the Chico Mendes Resex, it still remains an efficient land tenur e model to address conservation goals. Porto Dias Resex seems to be a similar success. Ho wever, I have speculate d that deforestation in these land tenures will possibly also be aff ected by family labor availability. In order to address this question I elaborated my second model in this analysis.

PAGE 104

92 Table 4-9 provides a life cycle model with le ngth of residence, age, and number of adults and children as independent variables. Although this model is weak in explaining deforestation (R2 < 0.07), length of residence seems to be a factor negatively affecting on the dependent variables defore station, crops and pasture. Th is suggests that long-time residents tend to deforest less than new arriva ls. This tendency also applies to crops and pasture. Table 4-9. Life cycle model at land cover types, all sites (n=8 8), Acre, Brazil, 2004. Dependent variables Independents variables Deforestation Crops Pasture Regrowth Constant Length of residence Mean age Number of adults Number of children 21.847 -.515* .111 -.181 -.053 6.151* -.128* .039 -.254 .008 14.677* -.389* .077 .106 .106 1.019* -.002 -.004 -.034 -.803 R2 .065 .068 .058 .201 F ratio 1.432 1.514 1.280 .421 p< .05 The life cycle model provides important in sights for forestry policies taking place in Acre over the past seven years. Since it s first mandate in 1999, the forest government has stimulated a return of rubber tappers a nd colonists to their or iginal rural lands, including areas of the Chico Mendes Resex as well as in agro-extractive settlements, and colonist projects located in the Rio Acre watershed. This strategy was established to generate employment and income in rural area s and lessen pressure in urban areas. This policy was also intended to reduce deforestati on. However, the recent arrivals appear to be clearing more forest than th e long-time forest residents. In addition, differences among families in te rms of market integration might be an important explanation for land cover change. I therefore ran a third model to evaluate the

PAGE 105

93 importance of market integration for the exte nt of deforestation. The third model shows the effects of market integration for land cover change. Table 4-10 provides a multivariate model for land cover variables regressed on subsistence income, distance from city, and expenses on purchased goods. Th e subsistence income is an annual value referring to consumption of crops, livest ock, and forest products by all household members. Distance from town refers to k ilometers from a smallholderÂ’s area to the closest town. The expenses va riable refers to the value of all reported goods bought by families for their consumption. The R2 values for all models indicate that market integration provides a weak explan ation for deforestation (0.09 < R2 < 0.29). Family expenses on purchased goods show no si gnificant effects on crops, pasture, or deforestation among smallholders in my sample. However, the F ratio value indicates this model is significant to explain deforesta tion as well as crops and pasture areas. Table 4-10. Market integration model at land cover types, al l sites (n=88), Acre, Brazil, 2004. Dependent variables Independents variables Deforestation Crops Pasture Regrowth Constant Subsistence shadow income Distance from town Expenses on purchased goods 34.901** -.306* -.003* .002 9.00** -.070* -.001* .000 24.209** -.216+ -.003* -.002 1.687* -.019* .000 .000 R2 .281 .225 .248 .087 F ratio 10.410** 7.720** 8.813 2.533+ + p <.10; p<.05; ** p<.01 Two variables, however, seem to be important for deforestation among smallholders selected in this study. First, s ubsistence coefficients indicate that families with higher levels of subsistence have less de forestation. The effect of this variable is more significant for crops than for pasture. Second, distance from city also negatively

PAGE 106

94 affects land cover dynamics among the study sites. This suggests that families close to the cities tend to have a higher percentage of area occupied by crops, pasture, and therefore greater deforestation. Overall, the model indicates that policy initiatives addressing forest conservation might take into consideration the importanc e of subsistence (“shadow”) income and access to markets for deforestation. While road s have been pointed out by many scholars as negatively affecting fore st areas, policy programs focu sing on food security through subsistence income may be an effective m echanism for contributing to environmental conservation. The fourth model (Table 4-11), denomina ted an investment model, assesses whether costs affect land cover outcomes. Specifically, this model estimates how investment priorities among categories of pr oducts can influence forest loss. Except for regrowth, this model shows a good fit fo r all land cover categories, with R2 varying from 0.44 to 0.61. F ratios were significant for all land cover categories. Table 4-11. Investment model at land cover types, all sites (n=88), Acre, Brazil, 2004. Dependent variables Independents variables Deforestation Crops Pasture Regrowth Constant Total Cost Crops Total Cost Livestock Total Cost Non-Timber Total Cost Timber 16.720** .004* .002* -.007** .000 5.036** .001* .000 -.002** -.000 10.410** .003* .002* -.005** .000 1.274** .000 .000* .000* .000+ R2 .609 .437 .597 .164 F ratio 32.329**16.112** 30.724** 4.058 + p < .10; p< .05; ** p< .01 Costs affect deforestation differently, depending on the economic activity for the costs in question. Total costs for non-timbe r forest products reduced deforestation

PAGE 107

95 because they reduced crops and pasture. This suggests that households with higher costs for non-timber products tend to have smaller areas of crops, pasture and by extension deforestation. Total costs for crops, on the ot her hand, result in positive influences on deforestation, crops, and pasture. The effect of livestock costs on defore station seems to co rroborate the earlier analysis on the forest management-forest clea ring tradeoff taking plac e in Porto Dias and Peixoto. The results for the livestock variable in dicate that areas with higher total cost for livestock tend to have a greater deforested ar ea due to larger pasture areas. Investments in pasture expansion may also foster forest de cline due to abandonment of old areas with soil degradation, which will possibly increase the area occupied by regrowth. The positive effect of total cost for livestock on regrowth confirmed this tendency. Finally, Table 4-12 presents an “assets” m odel to explain deforestation. This model attempts to theoretically evalua te the effect of assets on la nd cover. The results of this model although significant, had a low R2 value. Moreover, the F value of the assets in this model was significant for all land cover categories. Real estate was the only asset category that affected the percentage of deforestation among areas of this study. This effect was mo re significant for deforestation and pasture than for crops. Families with high-value real estate tend to have more pasture and by extension more deforestation. Surprisingly, livestock value and cattle head seems to indicate no effect on forest loss, which may s uggest that deforestation related to cattle will possibly be associated to cattle income and not to number of cattle heads, because families have sold their calves to large farmers. Moreover, pasture degradation in Peixoto, the area with the highest concentrati on of cattle among the la nd tenures selected,

PAGE 108

96 has forced families to sell cattle to avoid losses. In 2004 they re duced their assets of cattle and increased their total incomes. Table 4-12. Assets model at land cover t ypes, all sites (n=88), Acre, Brazil, 2004. Dependent variables Independents variables Deforestation Crops Pasture Regrowth Constant Real estate Livestock Cattle head Machine and equipment 4.3591 .0042** .0000 -.0002 -.5484 2.2523* .0009* -.0000 .0070 -.0003 1.4202 .0027** .0004 -.0053 .0006 .6875* .0000 .0000 -.0015 -.0001 R2 .392 .226 .390 .035 F ratio 13.363** 6.070** 14.935** .750 p<.05; ** p<.01 Conclusion Three hypotheses emerged for this study. Firs t, a land tenure hypothesis posits that Porto Dias PAE may be the most optimal mode l in terms of balancing income generation and forest conservation. Second, a land use hypothesis asserts that non-timber forest management would be more effective econom ically than timber management. Third, a deforestation hypothesis suggests that grea ter economic viability might imply more deforestation. Results provided from this analysis c onfirm all three hypotheses. In comparison with households in the other two land tenure types, rubber tappers of Porto Dias have been able to obtain favorable economic out comes (profit result of US$ 56.52) with a relatively low percentage of deforestation (12%). The second hypothesis was validated in the profitability analysis, in which nontimber forest products (economic loss US$ 10.22) were more economically viable th an timber (economic loss US$ 870.51). And the third hypothesis was confirmed by simu ltaneously examining the profitability of

PAGE 109

97 different land tenures (Table 4-7) and the investment model (Table 4-11), where Peixoto and Porto Dias proved profitable overall, but also exhibited the hi ghest deforestation percentages. That said, while crops and livestock costs were positively related to deforestation, forest management reduced forest clearing despite non-timber products being more economically viable than timber extraction. Other results arise from this analysis. W ith an accuracy of 87%, the land cover analysis identified a pronounced difference in the percent deforest ation in Peixoto PAE (36%) and in the Chico Mendes Resex (5%). While for the Chic o Mendes Resex, the findings indicated crops as the activity infl uencing forest cleari ng, in Peixoto most deforestation was rela ted to pasture. Economic-conservation tradeoffs were, howev er, already visible. While the Chico Mendes Resex is suggested as an efficien t model of forest conservation, its poor economic performance in terms of income gene ration and profitability raise concerns. By the same token, in Porto Dias, the high cost s associated with timb er management, a key component of forest government policies, hindered its otherwise favorable economic performance. In addition, the findings suggest that income may be used to increase deforestation, especially for cattle expansion. Three specific insights arise out of the results of multivariate regression models. First, the land tenure model indicates that land tenure type is a key factor affecting deforestation. Second, the investment model suggests that when smallholders pay high costs and reduce their short-term economic gains, they may re duce forest loss within their areas. This may indicate the potential for e nvironmental service payment programs such as Proambiente and the Chico Mendes Law to be highly effective in compensating

PAGE 110

98 families for the environmental benefits they provide to society. Although the assets model was relatively weak in explaining defo restation, it suggests that families with highvalue real estate have more deforestation. Im portantly, wealth is i ndirectly related to income generation; therefore, the assets model may indicate th at “wealth” may alter family interests in the forest so as to incr ease forest clearing, as noted by Wallace (2004). This conclusion is consistent with findings about subsistence shadow income in the market integration model where subsis tence income reduces deforestation. Three key issues emerge from the findings in this chapter, when taken as a whole. First, a lack of understanding of the live lihood strategies taking place within each land tenure type may contribute to failures of projects and acti ons aiming at “sustainable development.” Second, strategies only focused on economic viability might generate environmental tradeoffs, such as deforest ation for cattle expansion, which may be profitable but destructive and unsustainable Third, lack of ecological and market concerns in strategies implemented in the early 1990s also contributed to unsuccessful initiatives undertaken in these communiti es. These findings may contribute to the establishment of different feedbacks and linkages between research, community, and policymakers. In the following chapter I will conclude this thesis by discussing the implication of this study.

PAGE 111

99 CHAPTER 5 CONCLUSIONS: DILEMMAS, IMPLICATIONS, AND FUTURE RESEARCH “In the past we were poor, but happy a nd united rubber tappers, and harvested many products. Today we ar e rich, but unhappy and dis united managers, and yield timber.” Those were words spoken by a rubber tapper during the 2004 fieldwork. He refers to the current dilemmas faced by comm unities in balancing economic growth with their cultural, social, and environmental pract ices. While in the 1980s extractivists, agroextractivists, and colonists emerged as a common social movement to demand land reform in the Amazonia, after their victory in the 1990s, other challenges emerged. Public policies and NGO intervention have attempted to consolidate extr active reserves and agro-extractive settlements as land tenure models for achieving economic and environmental goals. These models have s ought to institutionalize alternative land use strategies focused on non-timber forest produc ts, agrosilvopastoral strategies, and more recently, timber products. Since 1999, the fore st government in Acre has integrated community land tenures and sustainable la nd use strategies into a comprehensive development approach. This study compared economic and land cover outcomes among three different community land tenure types: an extractive reserve, an agro -extractive settlement, and a colonization settlement. I used socioeconomic and remote sensing data for 88 households among these three land tenure types correspondi ng to three study sites in the State of Acre. The Chico Mendes was selected as model of extractive reserve and non-timber forest management. The agro-extractive settl ement model was Porto Dias in which rubber

PAGE 112

100 tappers have implemented an economic stra tegy based on timber management. Last but not least, small farmers of Peixoto were also selected as a model of colonist settlements and agrosilvopastoral management. Three research questions guided this study. First, what are the economic and deforestation outcomes of land uses implemen ted in the three land tenures types? Second, what is the most economically effective fore stry strategy: timber or non-timber? Third, does economic effectiveness all land use strate gies lead to increas ed deforestation, and thus an economic-environmental tradeoff? In addressing these questions I elaborated three hypotheses. First, I expected that Porto Dias PAE would be the most effective land tenure model in terms of generating relatively high gross incomes and favorable profit outcomes while also maintaining forest cover. Second, I expected that non-timber forest management would be more economically effi cient than timber extraction, due to the higher costs and lower profitability of tim ber extraction. Third, I anticipated that economic viability defined in terms of pr ofitability would impl y a higher level of deforestation. In this chapter, I summar ize the main findings and discuss their implications. I also call for a collaborativ e research effort to be implemented in southwestern Amazonia focusing on economic anal ysis of alternative land use strategies. Summarizing Key Findings In collaboration with community organi zations of the study sites, and with personnel in universitie s, NGOs, and governmental agencies, I performed this research in three phases. First, I carried out a colla borative participatory socioeconomic and land cover survey. Second, I classified satellite images of each study site in terms of four basic land cover types: forest, forest regrowth, past ure, and agriculture. Third, I conducted data analysis via descriptive statistics and multivariate regression analysis. Chapter 4

PAGE 113

101 presented these results and pr ovided three key insights regard ing the research questions I pose. First, the extent of defore station varied across land te nures. This variation followed the pattern I had expected. The pe rcentage of deforestation for colocaes of the Chico Mendes Resex was below 5%, Porto Dias ES ha d an intermediary level of deforestation (12%), and in Peixoto this percentage incr eased substantially, to 36%. These results indicated that extractive and agro-extractive m odels were very effective, in comparison to Peixoto, for addressing forest conservation. Second, all three land tenure types compared in this study showed themselves to be economically fragile in 2004. While the findi ngs for mean household profits indicated that Porto Dias and Peixoto were areas with low economic gains in 2004 (US$ 56.52 and US$ 71.83 respectively), among the Chico Mendes households, the situation worse yet, with economic losses (-US$ 122.04). These result s were surprising because previous to these results I had anticipated that forest gove rnment policies implemen ted in the state of Acre since 1999 would contribu te to improved economic viab ility of smallholders. In particular, I analyzed the pr ofitability of sustainable strategies based on forest management and agrosilvopastoral strategi es, but neither was economically successful. On average, the annual profit for non-timber forest products in the Chico Mendes and timber extraction in Porto Dias were –U S$ 33.36 and –US$ 987.21, respectively. Except for in Porto Dias, livestock was also economically ineffective. Third, integrated analysis of economic and land cover variables indicated that forest management implemented within PAEs may be an effective strategy for addressing both economic and conservation goals, since incomes in Porto Dias were nearly as high as

PAGE 114

102 those in the Peixoto colonizati on settlement, while deforestati on in Porto Dias was nearly as low as in the Chico Mendes Resex. That said, some tradeoffs were also evident; Peixoto had higher incomes but also much higher deforestation, and Chico Mendes had low deforestation but also lower incomes. Nonetheless, the non-timber forest product strategy proved more economically viable than timber extraction, even as non-timber forest products cause much less damage to fo rests. And households that were less marketintegrated, with lower costs, and farther fr om towns also exhibited lower deforestation percentages. Consequentl y, polices attempting to combine economic welfare and environmental conservation may be guided by instruments such as the agro-extractive reserve model, non-timber forest management, environmental service payments, and food security programs through subsistence shadow income. As a result, these findings may have various implications which I discuss in the next section of this concluding chapter. Policy Implications Over the past several years, policies impleme nted in the state of Acre have emerged as a new benchmark in Amazonia. Based on two different approaches, neoextrativismo and florestania, sustainable land use strategies implemented by extractivists, agroextractivists, and colonist families, such as forest management and agrosilvopastoral systems, have been core elements in these policies. As the defore station percentages in sampled colocaes of the Chico Mendes Resex and Porto Dias PAE were lower, I have recognized a relative success of these areas in terms of forest conservation. Yet, the land use strategies in all three sites do not app ear to be particularly economically viable. Moreover, I have observed that profitability, especially fo r timber extraction, may imply tradeoffs between forest and income in comm unity land tenure areas (Table 4-10). Based on these findings, I find three main imp lications for public policies in Acre.

PAGE 115

103 Regional integration has been a core component in the second forest government’s mandate (florestania period). The BR-317 road that runs through the Rio Acre watershed region, in which almost 50% of its territory is covered by extractive reserves, agroextractive settlements, and colonist settlemen ts, may become one of the most important highways in South America for commodity expo rts. My market integration model (Table 4-10) indicated that access could be a key driver of deforestation among community land tenures. However, as indicated by the land tenu re model (Table 4-9), the negative effects of roads might be reduced whether extractive reserves or agro-extractive settlements are established bordering the Pacific highway. This initiative might be also followed by state enforcement of tenure rules limiting forest clearing among those residing in Resexs and PAEs. This is because the life cycle model (Table 4-9) indica ted that newer arrivals in such land tenure types tend to de forest more than long-time resident households. In fact, the forest government has advocated the es tablishment of “green roads” along which zones of sustainable economic use might be created (Governo do Estado do Acre 2005). However, these green zones will be based on the timber forest management model already implemented in eastern Acre (R ego 2004), which based on my analysis is economically ineffective and may produce tradeoffs if local residents have to bear all of the costs (Table 4-7). Forest management policies implemented by the government of Acre were based on non-timber forest management in its first mandate and timber products in its second term of office. While non-timber products in Acre have had unstable demand and price fluctuations (Governo do Esta do do Acre 2001), the timber mark et has been characterized by increasing demand and high prices (Br ilhante 2001). Although timber products had a

PAGE 116

104 market advantage in comparison with non-ti mber, the timber forest management model implemented in Porto Dias also generated trad eoffs (Table 4-7) and involved very high costs to communities. With economic gains from this product, rubber tappers have invested in cattle as a strategy of cash ge neration to face their decline in subsistence shadow income. Meanwhile, other non-timber fore st products have actually proven to be more economically viable due to their lower co sts, as well as less damaging to the forest. Cost compensation has been a strategy utilized by the forest government since its first mandate to reimburse rubber tappers for their high costs in harvesting forest products. This initiative is known as the Ch ico Mendes Law or Rubber Subsidy and is constituted by paying a subsidized price for r ubber. In fact this st rategy is a type of environmental payment that compensates rubber tappers for their hi gh cost of harvesting rubber trees. The investment m odel (Table 4-11) analyzed in this study indicated that if cost compensation is expanded to all community land tenure types in this state, in which other products besides rubber are included, th is initiative might tend to reduce the tendency for forest clearing. That is because fa milies would not be stimulated to deforest in their attempts to reduce costs. By the same token, smallholders should not move to activities with low costs and high defo restation, such as cattle or crops. Although this study identified key gaps a ffecting economic viability and forest clearing across three selected land tenures types, how can these results proactively contribute to create alternatives for these communities? The findings themselves will not be sufficient to foster alternatives models with superior economic viability and effective for addressing conservation. These results, how ever, can be added to a collective and participatory effort to subsidize forest ma nagement and agrosilvopastoral initiatives

PAGE 117

105 implemented by community land tenures in the st ate of Acre. I recognize that the success of this coalition will depend on our ability to integrate public policies, participatory planning involving stakeholders, and research for collaborative environmental planning. Particularly, research and capacity build ing efforts addresse d by Brazilian and international organizations in southwestern Amazonia will be funda mental in supporting initiatives for alternative land use strategies that are economically viable and environmentally sustainable. In the followi ng section of this concluding chapter I will focus on research implications of these results. Emergent Questions for Community Land Tenure in Acre Based on these results, and whether new polic ies or other initiatives take place, I anticipate three scenarios for the land tenures studied in this thesis. First, taking into consideration the attractive le vel of prices for dairy produc ts and local demand for crops, small farmers, especially in Peixoto, will likely be stimulated to maintain their investments in cattle and crop production. However, they emphasized during the fieldwork that the high cost of production fo r crops and pasture as a result of soil degradation will emerge as a constraint fo r them. But stimulated by the market, they might nonetheless increase their deforested areas in order to establish new fields for crops and pasture. Second, although during the firs t three years of the timber forest management project, rubber tapper s in Porto Dias felt themselves at tracted by potential gross income gains, they have been disappointed with the high cost related to th is activity. Moreover, as this area is located in the same municipali ty in which Peixoto is located, and they have been influenced by crops and dairy production demand taking place in that area. All these factors may also stimulate forest clearing in Porto Dias. If costs for timber extraction are

PAGE 118

106 reduced, the profitability of timber will increase, but househol ds will tend to invest in cattle and will likely still deforest. The scenario for Chico Mendes Resex s eems more complex. The profitability analysis among the households of this area indicated that al l productive activities were economically ineffective. However, househol ds were able to obtain livelihoods with lower percentages of deforestation compared to Porto Dias and Peixoto. Although labor family is an important component in househol d costs, reducing costs of transportation from the Resex may be a key component to move rubber tappers toward a more profitable model of land tenur e. Households in the Chico Mendes Resex seem more concentrated on food security efforts, so I anti cipate that deforestati on will be affected by both crop and pasture expansion. Guided by these findings and my experience in this region, despite the land reforms after the 1980s and supportiv e policies of the forest government since 1999 households residing in all of the land te nure categories studied here still face important challenges in terms of their profitability as well as thei r environmental sustai nability. Is economic effectiveness compatible with traditional land uses of smallholders, which have implied forest conservation? What are the implica tions of alternative land uses for household livelihood systems? Can profitability be co mpatible with multiple use forestry? Can ecological constraints, such as soil degrad ation taking place in Pe ixoto, be addressed? How have economic and land cover dynamics, across community land tenures, varied over the forest government period in Acre ? These questions go beyond purely economic analyses, and require a larger collaborative effort invol ving contributions from many disciplines.

PAGE 119

107 APPENDIX A QUESTIONNAIRE USED IN THE CHICO MENDES EXTRACTIVE RESERVE

PAGE 120

108

PAGE 121

109

PAGE 122

110

PAGE 123

111

PAGE 124

112

PAGE 125

113

PAGE 126

114

PAGE 127

115

PAGE 128

116

PAGE 129

117

PAGE 130

118

PAGE 131

119

PAGE 132

120

PAGE 133

121

PAGE 134

122 APPENDIX B QUESTIONNAIRE USED IN THE PORTO DIAS AGRO-EXTRACTIVE SETTLEMENT

PAGE 135

123

PAGE 136

124

PAGE 137

125

PAGE 138

126

PAGE 139

127

PAGE 140

128

PAGE 141

129

PAGE 142

130

PAGE 143

131

PAGE 144

132

PAGE 145

133

PAGE 146

134

PAGE 147

135

PAGE 148

136

PAGE 149

137 APPENDIX C QUESTIONNAIRE USED IN THE PE IXOTO COLONIZA TION SETTLEMENT

PAGE 150

138

PAGE 151

139

PAGE 152

140

PAGE 153

141

PAGE 154

142

PAGE 155

143

PAGE 156

144

PAGE 157

145

PAGE 158

146

PAGE 159

147

PAGE 160

148

PAGE 161

149

PAGE 162

150

PAGE 163

151 APPENDIX D CIPEC TRAINING SAMPLE PROTOCOL Observation type: Within-site Edge Vantage RESEARCH ID: COUNTRY ID: SITE ID: RANDOM SAMPLE TS #: _____ OPPORTUNISTIC TS#_____ TODAY'S DATE (mm/dd/yr):__ /____/00 LOCAL TIME______ COLLECTORÂ’S NAME/EMAIL:________________________ TS AREA NAME / OWNER NAME (if applicable)______________________________ IMAGE PRODUCTS USED: Image ID/dates: Color Composite Used: R= 5, G=4 B=3 GEOGRAPHIC COORDINATES IN FIELD: UTM Northing (X )_________________ [m] UT M Easting (Y):_________________ [m] UTM Zone 16 Datum NAD27 GPS INFO: FILE NAME:______________________________ PDOP:_____________ Garmin Unit #:_________________ LOCATION OF PLOT TOPOGRAPHICALLY: Ridge_____ Slope______ Flat______ Steepness of Slope:______ (0-90 ) Azimuth (downhill direction of maximum slope in which water would naturally run)_____ (0-360 ) ELEVATION ____________meters above sea level (altimeter reading)

PAGE 164

152 DIAGRAMS OF GENERAL OBSERVATIONS: S how GPS points, North & training sample area in relationship to features. Aerial View | Profile Diagram ( parallel to maximum slope) | | | | | | | | | (land marks, north arrow and scale bar) | (dra w of vegetation and sl ope, vertical scale) LAND COVER TYPE (put a check mark next to land cover type or write in others): EXISTING VEGETATION TYPE DISTURBED: AGRICULTURE PLANTATION: Semi-deciduous broadleaf forest SS 1 (initial succession) Broadleaf crop Mixed semi-dec. forest (needle/broad) SS 2 (intermediate succession) Wood perennial fruit crop Mountain needleleaf forest SS 3 (advanced succession) Agroforestry/crops Cloud forest Disturbed forest (l ogging) Agroforestry/pasture Grassland Burned field Pasture Tall grasses and shrubs Quarry/Gravel pit Pasture wi shrubs/woody regrowth Other: Forest with cleared understory Bare soil Other: Stubble field Plowed field INFRASTRUCTURE: Crops Urban area Crops Rural settlement Crops Gravel Other: Other: If existing vegetation is secondary, give original vegetation if known:________________

PAGE 165

153 VEGETATION STRUCTURE ESTIMATES : [ N/A:_____ No vegetation in sample] Use ground cover estimate sheet to near est 5%: % he rbaceous ____ ; % litter, ____; % soil ____; % rock ____ Canopy closure: ______% cover Av erage canopy height:___________m, Height of emergent tree s:__________m No trees:_____ Average DBH of canopy trees: 2-10 cm___; 10-20 cm____; 20-30cm____; 30-50 cm____; 50-70 cm____; 70cm-1m____; > 1m ____ Average DBH of emergent trees: 210cm___; 10-20 cm____; 20-30 cm____; 30-50cm____; 50-70 cm____; 70cm-1m____; > 1m ___ Presence of Saplings: Absent _____, Few _____, Moderate_____, Abundant _____ Presence of Seedlings: Absent_____, Few _____, Moderate_____, Abundant _____ Presence of Epiphytes: Absent_____, Few _____, Moderate_____, Abundant _____ Presence of Succulents: Absent _____, Few _____, Moderate_____, Abundant _____ Presence of Others: ______ Absent _____, Few _____, Moderate_____, Abundant _____ DOMINANT SPECIES (Sci. names; common names) ________________________________________________________________________ __________________________________________________________________ ________________________________________________________________________ _______________________________________________________ PRESENCE OF MANAGED SPECIES (agricultu re, agroforestry, plantation): Number of managed species (inc. planted)_____ Sci. Name (Family/Genus/Species):_____________ Common name:__________________ Density: Few _____, Moderate_____, Abundant ________ Sci. Name (Family/Genus/Species):___________ Common Name:__________________ Density: Few _____, Moderate_____, Abundant _____ Other Observations:_______________________________________________________ ________________________________________________________________________

PAGE 166

154 LAND USE HISTORY (Fill out as far back in time as possible, recording dates of change to forest, pastur e, crop, plantation, etc.): Time period (mm/yr) Land Cover/Land Use Informant :______ ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________ ________________________________________________________________________ ESTIMATED AGE OF LAND COVER IF NO INFORMANT IS AVAILABLE:______ GENERAL OBSERVATIONS: Photos: Roll #:__________ Exposures # / direction (N, S, E, W, Sky, Ground):_______________________________ Seasonal change affects land use or la nd cover: No____ Yes____ If yes, explain:_____________________________________________ Training sample marked on image products: No___ Yes___ If no, explain:____________________________________________________________ Other Comments:_________________________________________________________

PAGE 167

155 LIST OF REFERENCES Allegretti, M. 1990. Extractive re serves: an alternative for reconciling development and environmental conservation in Amazonia. In: Alternatives to Deforestation: Steps toward Sustainable Use of the Amazonia Rain Forest. A.B. Anderson (Eds.). New York, NT, Columbia University Press, 252-264. Amaral, E.F. 1998. Levantamento de Solos no Sistema de Capacidade de Uso em Nvel de Pequena Propriedade Rural: o Caso do PED, Municpio de Senador Guimoard, Acre. Rio Branco, Embrapa-CPAF/AC. Anderson, A.B. 1990. Alternatives to Deforestation: St eps Toward Sustainable Use of the Amazon Rain Forest. New York, Columbia University Press. Angelsen, A. 1995. Shifiting cultivation and deforestation: study from Indonesia. World Development. 23(3), 1713-1729. Anselin, L. 2001. Spatial effects in econometric practices in environmental and resources economics. American Journal of Agricultural Economics. 83(3), 705-710. Arajo, H.J.B and L.G. Silva. 2000. Lista de espcies florestais do Acre: ocorrncia com base em inventrios florestais. Rio Branco-Acre, Embrapa-CPAFAC. Bakx, K. 1988. From proletatian to peasant: rural transforma tion in the State of Acre, 1870-1986. The Journal of Development Studies. 24(2), 141-160. Barros, H. and F. Estcio, 1972. Economia da Empresa Agrcola. Nova Lisboa, Universidade de Luanda. BID (Banco Internamericano de Desenvolvimento) Project. 2002. Program de Desenvolvimento Sustentvel do Acre: Nota Tcnica do Governo do Estado. Rio Branco-Acre, SEPLAN. (www.ac.gov.br). Accessed on June 2005. Blaikie, P. and H. Brookfield. 1987. Land Degradation and Society. London, Methunen. Browder, J.O. 1990. Extractive reserv es will not save the tropics. BioScience. 40(1), 626645. Bryant, R.L. and S. Bailey. 1997. Third World Political Ecology. London, Routledge. Cardoso, F.H. and G. Mller. 1977. Amaznia: Expanso do Capitalismo. So Paulo, Editora Brasiliense.

PAGE 168

156 Carpentier, C.L., S.A. Vosti, and J. W itcover. 2000. Intensified production systems on western Brazilian Amazon settlement fa rms: could they save the forest? Agricultural Ecosystem Environment. 82(3), 73-88. Carvalho, G.O., D. Nepstad, D. McGrath, M. del Carmem Vera Dias, M. Santilli, and A.C. Barros. 2002. Frontier expansion in the Amazon: balancing development and sustainability. Environment. 44(3), 34-42. Cattaneo, A. 2001. Deforestation in the Brazi lian Amazon: comparing the impacts of macroeconomic shocks, land tenu re, and technological change. Land Economics. 77 (2), 219-240. Cavalcanti, T.J.S. 1994. Colonizao no Acre: uma Scio-Econmica do Projeto de Assentamento do Assentamento Dirigido “Pedro Peixoto”. Fortaleza, Cear, Universidade Federal do Cear. (Master’s Thesis). Cronkleton, P. 1998. Landownership Turnover and Fam ily Farm Survival in an Amazon Resettlement Project. Gainesville, University of Florida (Ph.D. Dissertation). Daly, D. C. and M. Silveira. 2001. Affinities of the Acre Flora: A First Analysis. Rio Branco-Acre, UFAC-NYBG. (Unpublished work). Dragi evi S. 2005. Multi-dimensional interpola tions with fuzzy sets. In Fuzzy Modeling with Spatial Information for Geographic Problems. F.E. Petry, V.B. Robinson, M.A. Cobb (Eds.). New York, Springer. Ehringhaus, C. 2005. Pos-victory Dilemmas: Land Use, Development and Social Moviment in Amazonian Extractive Reserves. New Haven, USA, Yale University. (Ph.D. Dissertation) ELI (Environmental Law Institute). 1994. As Reservas Extrativistas do Brasil: Aspectos Fundamentais de sua Implantao. Washington, D.C, ELI. Ellis, F. 2000. Rural Livelihoods and Diversity in Developing Countries. Oxford, Oxford University Press. Esteves, B.M.G. 1999. Do Manso ao Guardio da Floresta: Estudo do Processo de Transformao Social do Sistema Seringal a partir do Caso da Reserva Extrativista Chico Mendes. Rio de Janeiro, Universidade Federal do Rio de Janeiro. (Ph.D. Dissertation) FUNTAC (Fundao de Technologia do Estado do Acre). 1999. Diagnstico do Setor Florestal do Estado do Acre: Monito ramento da Cobertura Florestal, Desmatamento e Uso da Terra. Rio Branco-Acre, FUNTAC. Gastal, E. 1980. Enfoque de Sistemas na Programao da Pesquisa Agropecuria. Braslia, UFCE.

PAGE 169

157 Geist, H.J. and E.F. Lambin, 2002. Proximate cause and underlying driving forces of tropical deforestation. BioScience. 52 (2), 143-150. Godoy, R., R. Lubowiski, A. Markandya. 1993. A method for the economic valuation of non-timber tropical forest products. Economic Botany. 47(2), 220-233. Godoy, R., M. Jacobson, and D. Wilkie. 1998. Strategies of rainforest dwellers against misfortunes: the Tsimane Indians of Bolivia. Ethnology. 37 (1), 55-70. Gomes, C.V.A. 2001. Dynamics of land use in an Amazoni an extractive reserve: the case of the Chico Mendes extractive reserve in Acre, Brazil. Gainesville, University of Florida. (MasterÂ’s thesis). Governo do Estado do Acre. 2000a. Zoneamento Ecolgico-Econmico: Aspectos Socioeconmicos e Ocupao Territorial. Rio Branco-Acre, SECTMA. Governo do Estado do Acre. 2000b. Zoneamento Ecolgico-Econmico: Indicativos Para a Gesto Territorial do Acre. Rio Branco-Acre, SECTMA. Governo do Estado do Acre. 2005. Potential I nvestments for Acre and its Boundaries. In Forum for Sustainable Development of the Acre. Rio Branco-Acre, SEPLANDS. Guerreiro, E. 1994. Caracterizao, Tipologia, e Di agnstico de Sistemas de Produo Predominantes em uma Comunidade Rural: o Caso de Cerro da Ponte Alta, Irati, PR. Londrina, IAPAR Hecht, S.B. 1985. Environment, development, and politics: capit al accumulation and livestock sector in eastern Amazonia. World Development. 13 (6), 663-684. Hildebrand, P.E. and M. Schmink. 2004. Agroforestry for improved livelihoods and food security for diverse smallholders in Latin America and the Caribbean. In 1st World Agroforestry Congress. Orlando, University of Florida. Homma, A.K.O. 1993. Extrativismo vegetal na Amaz nia: Limites e Oportunidades. Braslia, EMBRAPA/SPI. Homma, A.K.O. 1989. Reservas Extrativista: uma Op o Vivel de Desenvolvimento para Amaznia? Belm, Par Desenvolvimento. IDB (Inter-American Development Bank). 2002. Acre Sustainable Development Program. Loan Proposal. Washington, Inter-American Development Bank. (www.idb.org). Accessed on August 2005. IBGE (Instituto Brasileiro de Geografia e Estatstica).1998. Censo Agropecurio 19951996. Rio de Janeiro, IBGE. INCRA (Instituto Nacional de Colonizao e Reforma Agrria). 1999. Evoluo da Estrutura Agrria do Agrria do Brasil. Braslia, INCRA.

PAGE 170

158 Jensen, J.R. 2005. Introductory Digital Image Proce ssing: A Remote Sensing Approach. Saddle River, NJ, Prentice-Hall. Kageyama, P. 2002. Ganhos na seleo para a produtividade de ltex em populao natural de hevea brasiliensis na Reserva Chico Mendes: estudo de caso da IAPs (Ilhas de Alta Produtividade). Scientia Florestalis. 61(1), 79-85. Kageyama, P., D. Caron, F. Gandara, K. Mar tins, L.H. Oliveira, C.M.B. Lacerda, N. Teresinha Boufleuer, L. Arruda Riba s, A.M. Moreno, and E.M. Ferraz. 2004. Genetic and Ecological Aspects of Nonwood Forest Product Exploitation in two Western Amazonian Settlements. In IPGRI (International Plant Genetic Resource Institute). Challenge in Managing Forest Gene tic Resources for Livelihoods. Rome, Italy, IPGRI. 149-165. Kaimowitz, D. and A. Angelsen. 1998. Economic Models of Tropical Deforestation: a Review. Indonesia, CIFOR. Kainer, K., M. Schmink, A.C.P. Leite, and M. J. Fadel. 2003. Experiments in forest-based development in Western Amazonia. Society & Natural Resources. 16(1), 869-887. Lambin, E.F., B.L. Turner, H.J. Geist, S. B. Agbola, A.Angelsen, J.W. Bruce, O.T. Coomes, R. Dirzo, G. Fischer, C. Folke, P.S. George, K. Homewood, J. Imbernon, R. Leemans, X. Li. 2001. The causes of la nd-use and land-cove r change: moving behind myths. Global Environmental Change. 11 (1), 261-269. Lorena, R.B. 2003. Evoluo do Uso da Terra em Poro da Amaznia Ocidental (Acre), com Uso de Tcnicas de Deteco de Mudanas. So Jos dos Campos, INPE. (Master’s Thesis). Maciel, R.C.G. 2003. Ilhas de Alta Produtividade: Inovao Essencial para a Manuteno dos Seringueiros nas Reservas Extrativistas. Campinas, UNICAMP. (Master’s Thesis) Mafra, R.C. 1988. Agricultural Tropical: Curso de Especializao por Tutoria a Distncia. Braslia, UnB. Martin, N.B. 1995. Aplicativo “Custos”: Sistema de Custo de Produo Agrcola, Verso 1.1. So Paulo, IEA/EMBRAPA/FUNDEPAG. Martinello, P. 1988. A “batalha da borracha” na segunda guerra mundial e suas conseqncias para o vale amaznico. Rio Branco-Acre, Universidade Federal do Acre. Melatti, J.C. 1993. ndios do Brasil. So Paulo, Hucitec; Braslia, Editora UNB. Menezes, R.S. 2004. A Importncia da Reserva Legal na Gerao de Renda de Pequenos Produtores Rurais: Estudo de Caso no Estado do Acre, Amazonia. Curitiba, Universidade Federal do Paran. (Master’s Thesis).

PAGE 171

159 Mertens, B., W. D. Sunderlin, O. N doye, and E. F. Lambin. 2000. Impact of Macroeconomic Change on Deforestation in South Cameroon: Integration of Household Survey and Remotely Sensed Data. World Development. 28 (2), 983– 999. Mertens, B., R.Poccard-Chapuis, M.G. Pikett y, A.E. Lacques, and A. Venturieri. 2002. Crossing spatial analyses and livestock economics to understand deforestation processes in the Brazilian Amazon: the case of So Flix do Xing in south Par. Agricultural Economics. 27 (3), 269-294. Mesquita, G. 1977. Evidncias apresentada para a Comi sso de Agricultura da Cmera dos Deputados. Brasilia, Dirio do Congresso Nacional. Michelotti, F. 1993. Estratgias para o Desenvolvimento Comunitrio na Reserva Extrativista Porto Dias. So Paulo, Universidade de So Paulo. (Relatrio de Graduao) MMA (Ministrio do Meio Ambiente). 2001. Avaliao e Identificao de Aes Prioritrias para a Conservao, Ut ilizao Sustentvel e Repartio dos Benefcios da Biodiversidade na Amaznia Brasileira. Braslia, Ministrio do Meio Ambiente. MMA (Ministrio do Meio Ambiente). 2001a. Projeto RESEX – Realatrio 1995-2000. Braslia, Ministrio do Meio Ambiente. Murrieta, J.R. and R.P. Rueda. 1995. Reservas Extrativistas. Reino Unido. Cambridge. UICN, CCE, CNPT. Nepstad, D.C., and S. Schwartzman. 1992. In troduction: non-timber products from tropical forests: evaluati on of a conservaion and development strategy. In Nontimber products from tropi cal forests evaluation of a conservation and development strategy. Advances in Economic Botany. D.C. Nepstad, and S. Schwartzman, (Eds.). New York, The New York Botanical Garden. Oliveira. L.A.P. 1985. O Sertanejo, o Brabo e o Posse iro: cem anos de andanas da populao acreana. Rio Branco-Acre, Governo do estado do Acre. Padoch, D. and W. Jong. 1989. Production and prof it in agroforestry: an example from the Peruvian Amazon. In: Fragile Lands of Latin America: Strategies for Sustainable Development. Browder, J.G. (ed.). Boul der, CO, Westview Press Paula, E.A. 2003. Estado e desenvolvimento insusten tvel na Amaznia Ocidental: dos missionrios do progresso aos mercadores da natureza. Rio de Janeiro, Universidade Federal do Rio de Janeiro. (Ph.D. Dissertation) Peet, R. and M. Watts. 1996. Liberation Ecologies: Environment, Development, Social Movements. London, Routledge.

PAGE 172

160 Perz, S.G. 2001. From sustainable developmen t to “productive conservation:” forest conservation options and agriculture inco me and assets in the Brazilian Amazon. Rural Sociology. 66(1), 93-112. Peters, C.M., Gentry, A.H. and Mendelsohn, R.O. 1989. Valuation of an Amazonian rain forest. Nature. 339(1), 655-656. Peterson, G. 2000. Political ecology and ecologica l resilience: an integration of human and ecological dynamics. Ecological Economics. 35 (2), 323-336. Pinedo-Vasquez, M., Zarin, D., Jipp, P. 1990. Economic returns from forest conversion in the Peruvian Amazon. Ecological Economics. 6(1), 163-173. Rego, J.F. 1999. Amaznia: do extra tivismo ao neoextrativismo. Cincia Hoje. 25(2), 6265. Rego, J.F. 2003. Anlise Econmica dos Sistemas de Produo Familiar Rural da Regio do Vale do Acre – 1996/1997. Rio Branco-Acre, Universidade Federal do AcreSEBRAE-The Ford Foundation. Rego, J.F. 2004. Referencial do Zoneamento Ecolgico-Econmico do Acre – Fase II. Rio Branco-Acre. IMAC. (unpublished). Rodrigues, E. 2004. Vatangem Competitiva do Ecossistema na Amazia: o Cluster Florestal do Acre. Braslia Universidade de Bras lia. (Ph.D. Dissertation). Rueda, R.P. 1995. Evoluo Histrica do Extrativismo. In Reservas Extrativistas. J.R. Murrieta and R.P. Rueda. Cambridge, UICN, CNPT. Sassagawa. H.S. 1999. Desmatamento na Chico Mendes. So Paulo, INPE. (Master’s thesis). Schmink, M. 1984. Hosehold economic strate gies: review and research agenda. Latin American Research Review. 19(3), 87-101. Schmink, M. and C.J. Wood. 1987. The Political Ecology of Am azonia. Lands at Risk in the Third World: Local-Level Perspectives. Boulder, Westview Press. Schwartzman, S. 1989. Extractive reserves: the rubber tappers’ strategies for sustainable use of the Amazon rain forest. In: Fragile Lands In latin America: Strategies for Sustainable Development. J.O. Browder. Boulder, CO, Westview Press, 150-165. SEPLANDS (Secretaria de Planejamen to e Desenvolvimento Sustentvel). 2005. Forum de Desenvolvimento Suste ntvel do Estado do Acre: Investimentos Potenciais para o Acre e Regio. SEPLANDS. (www.ac.gov.br). Accessed on September 2005. Silva, A.F. 1990. Razes da ocupao recente das terras do Acre: movimento de capitais, especulao fundiria e disputa pela terra. Rio Branco-Acre, Casa da Amaznia.

PAGE 173

161 Silveira, M., N.M.C. Paula, I.F. Brown, H. Bruno, N. Borges, D. Daily, and L.A. Ferreira. 1997. Os ‘buracos negros’ da diversidade. Cincia Hoje. 22 (128), 64-65. Simmons, C.S. 2004. The political economy of land conflict in the eastern Brazilian Amazon. Annals of the Association of American Geographers. 94(1), 183-2006. Sobrinho, P.V.C. 1992. Capital e Trabalho na Amaznia Ocidental: Contribuio Histria Social e das Lutas Sindicais no Acre. So Paulo, Editora Cortez. Souza, F.K.A. 2001. Pasture instead of rubber? The ranching tendencies of family-based agriculture in extractive reserves and colonization projects in Acre, Brazil, Southwestern Amazonia. Annals of Open Meeting of the Human Dimensions of Global environmental Change Research Community. Rio de Janeiro, CIESIN. Thorner, D., B. Kerblay, and R.E.F. Smiths (eds.) 1986. Chayanov on the theory of peasant economy. Madison, University of Winsconsin Press. Tomich, T., M.V. Noordwijk,, S. Vosti, and J. Witcover. 1998. Agricultural development with rainforest conservation: methods for seeking best bet alternatives to slashburn, with application to Brazil and Indonesia. Agricultural Economics. 19(3), 159174. Valentim, J.F., E.F. Amaral, M. Cavalcan te, M. Fazolin, S.S.V. Caballero, R.M. Boddery, R.D. Sharma, A.W.F. Melo. 2000. Diagnosis and potential socioeconomic and environmental impacts of pasture death in the western Brazilian Amazonia. LBA Meeting. Manaus, LBA Program. Vosti, S.A., E.M. Braz, C.L. Carpentier, M.V.N. D’Oliveira, and J. Witcover. 2003. Rights to forest products, deforestation a nd smallholder income: evidence from the western Brazilian Amazon. World Development. 31(11), 1889-1901. Walker, R. 2003. Mapping process to pattern in the landscape change of the Amazonian frontier. Annals of the Association of American Geographers. 93(2), 376-398. Wallace, R.H. 2004. The effects of Wealth and Ma rkets on Rubber Tapper Use and Knowledge of Forest Resource in Acre, Brazil. Gainesville, University of Florida. (Ph.D. Dissertation). Weinstein, B. 1983. The Amazon Rubber Boom: 1850-1920. Stanford, Stanford University Press. Wollenberg, E., D. Edmunds, and J. Ande rson. 2001. Accomodation multiple stakeholder interests in local forest management. International Journal of Agricultural Resources, Governance, and Ecology. 1(2), 193-198.

PAGE 174

162 BIOGRAPHICAL SKETCH Francisco Kennedy Arajo de Souza was rais ed in Rio Branco, the capital of the state of Acre, Brazil. His pa rents are Apurinan, an indige nous people in the Amazon, and one year before he was born, in 1969, they migrated with many of his family members from Camicu, today an indigenous reserve in Amazonas, to Rio Branco. Since 1993 he has worked as a staff member at the Fede ral University of Acre (UFAC), where he received a B.S. in economics in 1998. After finishing his undergraduate degree, he was led to collaborate with research programs at UFAC. Since 1999 he has managed various research and capacity building programs in the Amazon. Over the past two years he has collaborated with the Large-Scale Biosphere Atmosphere in Amazonia (LBA) Program funded by NASA and the Brazilian government He has also collaborated with other programs funded by USAID, the Inter-Ame rican Development Bank, International Development Research Centre (IDRC), th e Ford Foundation, World Wildlife Fund, International Tropical Timber Organizati on, and the Brazilian government. In 2001, he was awarded a fellowship for masterÂ’s stud ies in the United States from the Ford Foundation International Fellowship Program (IFP). Upon completion of his MasterÂ’s degree in Latin American studies with a c oncentration in tropi cal conservation and development at the University of Florida, he will work as a researcher at UFAC.