Citation
Land Use and Land Cover in Inapari, Peru, and Assis Brazil, Brazil, Southwest Amazonia

Material Information

Title:
Land Use and Land Cover in Inapari, Peru, and Assis Brazil, Brazil, Southwest Amazonia
Creator:
ZAMBRANO, ANGELICA MARIA ALMEYDA
Copyright Date:
2008

Subjects

Subjects / Keywords:
Cattle ( jstor )
Crops ( jstor )
Farm forestry ( jstor )
Forest growth ( jstor )
Forests ( jstor )
Infrastructure ( jstor )
Land cover ( jstor )
Land use ( jstor )
Old growth forests ( jstor )
Secondary forests ( jstor )
City of Gainesville ( local )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Angelica Maria Almeyda Zambrano. 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:
12/18/2004

Downloads

This item has the following downloads:

almeyda_a ( .pdf )

almeyda_a_Page_042.txt

almeyda_a_Page_045.txt

almeyda_a_Page_116.txt

almeyda_a_Page_071.txt

almeyda_a_Page_095.txt

almeyda_a_Page_050.txt

almeyda_a_Page_074.txt

almeyda_a_Page_065.txt

almeyda_a_Page_078.txt

almeyda_a_Page_105.txt

almeyda_a_Page_099.txt

almeyda_a_Page_019.txt

almeyda_a_Page_022.txt

almeyda_a_Page_059.txt

almeyda_a_Page_120.txt

almeyda_a_Page_132.txt

almeyda_a_Page_094.txt

almeyda_a_Page_093.txt

almeyda_a_Page_129.txt

almeyda_a_Page_101.txt

almeyda_a_Page_086.txt

almeyda_a_Page_007.txt

almeyda_a_Page_013.txt

almeyda_a_Page_117.txt

almeyda_a_Page_115.txt

almeyda_a_Page_055.txt

almeyda_a_Page_006.txt

almeyda_a_Page_073.txt

almeyda_a_Page_089.txt

almeyda_a_Page_030.txt

almeyda_a_Page_011.txt

almeyda_a_Page_123.txt

almeyda_a_Page_126.txt

almeyda_a_Page_001.txt

almeyda_a_Page_087.txt

almeyda_a_Page_021.txt

almeyda_a_Page_080.txt

almeyda_a_Page_029.txt

almeyda_a_Page_076.txt

almeyda_a_Page_079.txt

almeyda_a_Page_090.txt

almeyda_a_Page_057.txt

almeyda_a_Page_103.txt

almeyda_a_Page_058.txt

almeyda_a_Page_097.txt

almeyda_a_Page_032.txt

almeyda_a_Page_051.txt

almeyda_a_Page_047.txt

almeyda_a_Page_085.txt

almeyda_a_Page_127.txt

almeyda_a_Page_066.txt

almeyda_a_Page_004.txt

almeyda_a_Page_025.txt

almeyda_a_Page_018.txt

almeyda_a_Page_108.txt

almeyda_a_Page_060.txt

almeyda_a_Page_121.txt

almeyda_a_Page_026.txt

almeyda_a_Page_052.txt

almeyda_a_Page_131.txt

almeyda_a_Page_044.txt

almeyda_a_Page_024.txt

almeyda_a_Page_122.txt

almeyda_a_Page_003.txt

almeyda_a_Page_031.txt

almeyda_a_Page_096.txt

almeyda_a_Page_100.txt

almeyda_a_Page_020.txt

almeyda_a_Page_033.txt

almeyda_a_Page_083.txt

almeyda_a_Page_112.txt

almeyda_a_Page_072.txt

almeyda_a_Page_109.txt

almeyda_a_Page_098.txt

almeyda_a_Page_023.txt

almeyda_a_Page_111.txt

almeyda_a_Page_028.txt

almeyda_a_Page_008.txt

almeyda_a_Page_064.txt

almeyda_a_Page_015.txt

almeyda_a_Page_104.txt

almeyda_a_Page_088.txt

almeyda_a_Page_077.txt

almeyda_a_Page_068.txt

almeyda_a_Page_113.txt

almeyda_a_Page_082.txt

almeyda_a_Page_027.txt

almeyda_a_Page_035.txt

almeyda_a_Page_046.txt

almeyda_a_Page_034.txt

almeyda_a_Page_118.txt

almeyda_a_Page_010.txt

almeyda_a_Page_133.txt

almeyda_a_Page_069.txt

almeyda_a_Page_091.txt

almeyda_a_Page_062.txt

almeyda_a_Page_075.txt

almeyda_a_Page_014.txt

almeyda_a_Page_061.txt

almeyda_a_Page_125.txt

almeyda_a_Page_124.txt

almeyda_a_Page_036.txt

almeyda_a_Page_106.txt

almeyda_a_Page_048.txt

almeyda_a_Page_009.txt

almeyda_a_Page_114.txt

almeyda_a_Page_038.txt

almeyda_a_Page_063.txt

almeyda_a_Page_040.txt

almeyda_a_Page_043.txt

almeyda_a_Page_134.txt

almeyda_a_Page_053.txt

almeyda_a_Page_016.txt

almeyda_a_Page_081.txt

almeyda_a_Page_037.txt

almeyda_a_Page_041.txt

almeyda_a_Page_039.txt

almeyda_a_Page_049.txt

almeyda_a_Page_070.txt

almeyda_a_Page_119.txt

almeyda_a_Page_054.txt

almeyda_a_Page_110.txt

almeyda_a_Page_107.txt

almeyda_a_Page_092.txt

almeyda_a_Page_130.txt

almeyda_a_Page_056.txt

almeyda_a_Page_017.txt

almeyda_a_Page_002.txt

almeyda_a_Page_067.txt

almeyda_a_Page_102.txt

almeyda_a_Page_084.txt

almeyda_a_Page_005.txt

almeyda_a_Page_012.txt

almeyda_a_pdf.txt

almeyda_a_Page_128.txt


Full Text












LAND USE AND LAND COVER IN INAPARI, PERU, AND ASSIS BRAZIL,
BRAZIL, SOUTHWEST AMAZONIA
















By

ANGELICA MARIA ALMEYDA ZAMBRANO


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


2004

































Copyright 2004

by

Angelica Maria Almeyda Zambrano

































To the people of the MAP region.















ACKNOWLEDGMENTS

I would like to thank the people of Assis Brazil and Ifiapari, especially the 90

families that allowed me to interview them in order to gather the data for this research. I

also would like to thank Manuel Batista de Araujo, Prefeito of the Municipio of Assis

Brazil, and Mario Enrique Montes Le6n, Alcalde of the Province of Ifiapari.

I would like to thank the different funding sources that allowed me to attend the

Center for Latin American Studies at the University of Florida and to those who funded

the field research: the Organization of American States, the Tropical Conservation and

Development program at University of Florida, the Tinker Foundation and the Setor de

Estudos do Uso da Terra e Mudancas Globais at the Federal University of Acre.

I would like to thank the members of my committee, for their help and support

throughout the research process and especially for their comments on my various final

drafts-Michael Binford, Stephen Perz, and especially to Marianne Schmink, my chair.

I would like to thank I. Foster Brown for the opportunity to visit Acre for the first

time in 2001, and for his support and comments on my research, as well as for all his

logistical support during my field work.

I would like to thank my field assistants, Mercedes in Ifiapari and Marco in Assis

Brazil, for all their hard work that made my field research possible. I also would like to

thank a number of institutional offices that helped me in the field: INRENA, INADE,

MADERIJA and MADERACRE, the health post in Ifiapari, SEATER, IBAMA, and

SETEM.









I would like to thank my fiance, Eben Broadbent for all his support during the


writing process.
















TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S .................................................................... ......... .............. iv

L IST O F T A B L E S ........ ....................................................................... .. viii

LIST OF FIGURES ......... ......................... ...... ........ ............ xi

ABSTRACT ........ .............. ............. .. ...... .......... .......... xii

1 IN TR OD U CTION ............................................... .. ......................... ..

Land U se and Land Cover Change...................................... ..................... .. ........... 1
Proximate Sources and Driving Forces of Change.....................................................3
A ssis Brazil and Ifiapari ....................................................... ................ .4
Site D description ................................................... .................... 5
A ssis Brazil and A cre in Context ........................................ ....... ............... 6
P population com position ............................................ ........... ............... 8
N natural protected areas............................................................ ............... 10
Transportation and highways ........... ................................ ..................11
Ifiapari and Madre de Dios in Context ....................................... ............... 12
Natural protected areas and indigenous communities...............................14
Transportation and high ays................................ ........................ 14
The Case for Comparing Ifiapari and Assis Brazil ............................................. 15
The Role of Roads .................. ................................ .. ..... ................. 15
The Im portance of M markets .................................. ............................................ 16
Existence and Inexistence of Credit .............. ............................................. 19
Roads, Markets and Credits as Land Use Drivers.............................................19

2 THEORETICAL APPROACH ............................................................................21

Introdu action ........................................................ ............. ................. 2 1
Land U se D rivers and D eforestation ........................................ ....... ............... 21
M markets and C redit ......................... .. .................... ......... ........... 2 1
Roads D riving Land U se .............................................................................. 23
Approaches to Explain Deforestation..................... ... ....................... 26
Integrated Theories ........................ .................... ... .... ........ ......... 30
P political E ecology ................................................. ................ 32
H household D em ography ......................................................... ................ ..... 36
P an arch y T h eo ry ............. .......................................................... .. .... .. .. .. ..3 7









Integrated Fram ew orks .............. ..... ............................................................... 39
The Three-Tired Hierarchical Approach...........................................................39
Household Transformations Land Use And Environmental Change ................40
The Adaptive Cycle ....................... ...................... .. .... ................. 43
Potential, connectedness and resilience ....................................... .......... 43
H ierarchies and panarchies...................................... ......................... 45
Fram ew ork Integration ................ ... .... .... .................... ........ ............... 47
Looking at the household and its activities as adaptive cycles ....................49
N onlinear effect of land use drivers .................................. ............... 52
Considerations Regarding the Framework ...................................................53

3 LAND USE AND LAND COVER ........................................ ........................ 55

Introduction................. ...... ... .. ....... .. .... ... .......... ......... ............ 55
Fieldwork M ethods in Ifiapari and Assis Brazil ............................... ............... 56
Fieldw ork in Ifiapari ....................... .... ................ ... ...... .. .... ...........56
F ieldw ork in A ssis B razil .................... .. .......................... ............... .... 59
The Differences in Methodology and Their Implications .................................60
Operationalization of Variables ......... .......... ..... .. ................... 61
Comparing Variable Means for Assis Brazil and Ifiapari ........................... ........68
Correlations Between Independent and Dependent Variables .................................73
M ultivariate M odels........................ ........................ .. ............. .........79
Land Use Models...................... ..............................79
Land Cover M odels ..................... ........ ....... .... ... .... ........ .............. 86
Land Use and Land Cover Final M odels.................................. ............... 92

4 CONCLUSIONS ................................... .. .. ........ .. ............101

APPENDIX

A QUESTIONNAIRE APPLIED IN INAPARI, PERU.............................................104

B QUESTIONNAIRE APPLIED IN ASSIS BRAZIL, BRAZIL.............................109

L IST O F R E F E R E N C E S ...................................................................... ..................... 114

BIOGRAPHICAL SKETCH ............................................................. ............... 121
















LIST OF TABLES


Table p

1-1 Comparisons of land area and population from the country level to the
Municipio of Assis Brazil and the District of Ifiapari .............................................5

1-2 Comparing historical processes for the Municipio of Assis Brazil and the
D district of Ifiapari ......................... ..... ............................................... .... ...............

2-1 Models of deforestation showing predicted effect of key variables....................28

2-2 Household demographic variables used in land use modeling for Amazonian
areas ................. ..................................... ...........................29

3-1 Descriptive statistics for land use and land cover outcome variables.
Ifiapari and Assis Brazil, 2003 ......... ....................................... 63

3-2 Descriptive statistics for household background information variables.
Ifiapari and A ssis B razil, 2003 ........................................ ........................... 65

3-3 Descriptive statistics for place, markets, credit and road infrastructure variables.
Ifiapari and A ssis B razil, 2003 ........................................ ........................... 66

3-4 Descriptive statistics for household life cycle variables.
Ifiapari and A ssis B razil, 2003 ................................... ............................................ 67

3-5 T-test of means for land use outcomes, land cover outcomes, background
information, markets and credit, road infrastructure and household life cycle
variables according to location in Ifiapari or Assis Brazil, 2003 ...........................69

3-6 Correlations between land use outcomes and land cover outcomes variables.
Ifiapari and A ssis Brazil, 2003 ............................................................................. 73

3-7 Correlations between land use outcomes and background information.
Ifiapari and A ssis B razil, 2003 ........................................ ........................... 74

3-8 Correlations between land use outcomes market and credit and road
infrastructure. Ifiapari and Assis Brazil, 2003 ................... ......................... 75

3-9 Correlations between land use outcomes and household cycles.
Ifiapari and A ssis B razil, 2003 ........................................ ........................... 76









3-10 Correlations between land use outcomes and land cover outcomes.
Ifiapari and A ssis B razil, 2003 ........................................ ........................... 76

3-11 Correlations between land cover outcomes and background information.
Ifiapari and A ssis B razil, 2003 ........................................ ........................... 77

3-12 Correlations between land cover outcomes, markets and credit and road
infrastructure. Ifiapari and Assis Brazil, 2003 ................... ......................... 78

3-13 Correlations between land cover outcomes and household cycles.
Ifiapari and A ssis B razil, 2003 ........................................ ........................... 78

3-14 Models of farm area in annual crops outcome regressed on background
information, market & credit road infrastructure, and household life cycle in
A ssis B razil and Ifiapari, 2003. .......................... ................ ......... ......................80

3-15 Models of farm area in perennial crops outcome regressed on background
information, market & credit road infrastructure, and household life cycle in
A ssis B razil and Ifiapari, 2003 ........................................ ........................... 82

3-16 Models of farm area in pasture outcome regressed on background information,
market & credit, road infrastructure, and household life cycle in Assis Brazil
and Inapari, 2003 ....................................... ............ .... .. ............ 84

3-17 Models of head of cattle outcome regressed on background information, market
and credit, road infrastructure, and household life cycle in Assis Brazil and
Ifi ap ari, 2 003 ...................... .... ......... .... .......... .................................85

3-18 Models of area of old growth forest outcome regressed on background
information, market and credit, road infrastructure, and household life cycle in
A ssis B razil and Ifiapari, 2003 ............................................. .... ..... .. ........ 87

3-19 Models of area of secondary forest outcome regressed on background
information, market and credit, road infrastructure, and household life cycle in
Assis Brazil and Inapari, 2003 ..................................... ..... .......... 88

3-20 Models of area (ha) deforested outcome regressed on background information,
market and credit, road infrastructure, and household life cycle in Assis Brazil
and Ifiapari, 2003 .................................................. ........ ............. ......... 90

3-21 Models of percentage of initial forest deforested since arrival outcome regressed
on background information, market and credit, road infrastructure, and
household life cycle in Assis Brazil and Ifiapari, 2003................................. 91

3-22 Final models for land use outcomes showing all the independent variables that
were significant in final multivariate land use and land cover models.
A ssis B razil and Ifiapari, 2003. .......................... ............................................. 93









3-23 Final models for land cover outcomes showing all the independent variables
that were significant in final multivariate land use and land cover models.
Assis Brazil and Ifiapari, 2003 .............................................................. ........ 94
















LIST OF FIGURES


Figure p

1-1 Study area: Tri-national border Ifiapari (Peru), Assis Brazil (Brazil) and
B olp eb ra (B oliv ia) ............................................... ................ 7

2-1 The three-tiered hierarchical approach ...... ......... ..... ................................ 41

2-2 Household transformations, land use and environmental change..........................44

2-3 The adaptive cycle, the four ecosystem functions (r, K, Q,ao) and the flow of
events among them............. ........... ......... ................ 46

2-4 Framework integration, the three-tiered hierarchical approach, household
transformations approach and the panarchy approach. .........................................48

2-5 Levels of interaction in a nested set of adaptive cycles......................... .......... 49

2-6 Looking at the household and its land use activities as adaptive cycles ................51

3-1 Household farms visited in the Municipio of Assis Brazil and in the District of
Ifi ap ari ................ ..................................... ........................... 58















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

LAND USE AND LAND COVER IN INAPARI, PERU, AND ASSIS BRAZIL,
BRAZIL, SOUTHWEST AMAZONIA

By

Angelica Maria Almeyda Zambrano

December 2004

Chair: Marianne Schmink
Major Department: Center for Latin American Studies

The present thesis research has its roots in the growing field of land use and land

cover change. It compares land use and land cover across space in two bordering areas:

the district of Ifiapari in Madre de Dios, Peru, and the municipio of Assis Brazil in Acre,

Brazil. This study case has a micro-level approach that focuses on small farm households

in this area know as the tri-national border Peru-Brazil-Bolivia.

The research analyzes the differences in land use and land cover in Assis Brazil and

Ifiapari and proposes an innovative integrative framework with roots on Political

Ecology, Household Demography and Panarchy theories. The research also has an

empirical component, the assessment of the proposed framework is done by modeling

land use and land cover in both tows. The findings reveal the relevance of road

infrastructure variables, market variables and background variables in explaining

differences in land use and similarities in land cover outcomes. It also suggests that









further development of the proposed integrative framework may contribute to a better

understanding of different level process that drive land use and land cover change.














CHAPTER 1
INTRODUCTION

The subject matter of this research has its roots in the growing field of land use and

land cover change; however, differently from most studies it does not deal with changes

across time. It compares land use and land cover across space in two bordering areas: the

district of Ifiapari in Madre de Dios, Peru and the municipio of Assis Brazil in Acre,

Brazil.

The present document has been organized in four chapters. The chapter 1 briefly

reviews the bases of the field of land use and land cover change, and presents the study

area from a historical and current perspective observing its similarities and differences.

The chapter 2 goes more in depth on relevant theories used to explain land use and land

cover changes, and it ends by proposing an integrated theoretical framework for the

present research. Chapter 3 describes the methods used to gather and analyze the data. It

also presents and interprets the results obtained form the analysis. Chapter 4 presents a

general discussion of the findings from Chapter 3 and elaborates general conclusions for

the present research.

Land Use and Land Cover Change

It is only relatively recently that humans have taken a large role in modifying

landscapes across the globe; although the process of massive change began in temperate

areas, it is currently centered in the tropics (Ojima et al. 1994). Changes are so severe that

when globally added they notably impinge on important Earth System functions (Lambin

et al. 2001). Traditionally, research on human dimensions of global change has focused









on two broad fields: industrial metabolism and land use and land cover change. The

present research is framed within the latter. Land use and land cover change is recognized

as an interdisciplinary research area. The land use component refers to the utilization of

land, and it has been traditionally studied by social scientists, while the land cover

component refers to the biotic and physical components of land surface, and has been

traditionally studied by natural scientists (Meyer and Turner 1992).

Land cover changes occur in two ways: conversion and modification. Conversion

implies change in land category and modification implies change within a category

(Meyer and Turner 1992). Land cover conversion is usually related to land use changes in

area, and land cover modification is usually related to land use changes in intensity. The

majority of existing literature deals with land cover conversion, although land cover

modifications are also widespread and probably as important as the former but more

difficult to asses (Meyer and Turner 1992; Lambin et al. 2001).

Houghton (1994) explains that changes in land use have both intended and

unintended consequences. The intended consequence is to increase area or productivity of

a certain type of product, although some land uses have the opposite effect. The

unintended consequence is to have a negative effect on global climate. Land use change

has contributed to the enhancement of the greenhouse effect, 25% of human-caused

emissions. Land use change rapidly changes ecosystems properties, regular inputs and

exchanges of energy; water and nutrients in ecosystems are being severely changed and it

may also create greater opportunities for exotic species invasion (Ojima et al. 1994).

Well-recognized changes are trace-gas emissions, detriment of water quality, changes in

water flows, and soil alteration and erosion.









Proximate Sources and Driving Forces of Change

Meyer and Turner (1992) differentiate between the proximate sources of change

and the driving forces of change. Proximate sources are the human actions that alter the

land cover, while driving forces are the underlying causes of proximate sources. At the

global level, much research on driving forces has focused on human population pressure,

but different changes may involve different driving forces, and the same changes may

involve different driving forces in different areas of the world (Meyer and Turner 1992).

According to Lambin et al. (2001) poverty and population growth are not the major

causes of land cover change at the global scale. Significant correlations between land

cover conversion and population were found only when the research area possessed

similar social and environmental characteristics (Meyer and Turner 1992). The role of

political, cultural and other demographic factors in land use decisions are gradually

taking more relevance in the effort to understand global change (Ojima et al. 1994).

Relationships are complex; the current higher rates of change in the so called developing

areas may be explained by the demand from developed countries, since international

trade is often an important land use change driver (Houghton 1994).

Tropical deforestation has been a central concern among other land cover

conversions. In general agricultural expansion is considered to be the main proximate

source (Barbier 2001) and has been found to be driven by changing economic factors that

are associated to institutional factors like social and political changes (Hecht 1985;

Lambin et al. 2001). On average, 50% of the forest area lost in the tropics per year is used

to replace agricultural areas that are no longer productive and were abandoned. Therefore

only 50% goes to actually increase the area in agriculture (Houghton 1994). Data on past

land use and land cover are not enough to improve current models of land use and land









cover change. Data should be accompanied by a better understanding of the causes of

land use and land cover change (Lambin et al. 2001).

Barbier (2001) reported a list of factors that play an important role in tropical

deforestation. Among those listed at the cross-country and country level are factors like

income, population growth and density, agricultural prices and returns, agricultural

yields, logging prices and returns, roads and road building and institutional factors. In

general changes in markets, credit and roads are associated with changes in land cover,

and therefore we will review them in more detail in the following chapter.

Assis Brazil and Ifiapari

Comparison studies in land use and land cover change are rare. One of the main

constraints is to adequately address differences at the land use driver level.

Socioeconomic and biophysical drivers may interact in ways that are difficult to

understand when trying to make comparisons. To assess the role of biophysical land use

drivers one would like to compare populations with very similar socioeconomic

characteristics that are located in different landscapes. In a similar way, to assess the role

of socioeconomic land use drivers one would like to compare populations that are in the

same biophysical landscape that have different socioeconomic characteristics. The latter

is precisely the case I expect to make for Assis Brazil and Ifiapari in this section.

Both towns, due to their proximity, are located in very similar biophysical

landscapes. This fact will allow us to compare the effect of socioeconomic land use

drivers. Socioeconomic land use drivers are especially interesting in this bi-national

context since they largely result from different development policies applied by Peruvian

and Brazilian governments since this area was first reached by the white man in the early

1900s.









Site Description

General differences between Brazil and Peru are very evident at national levels.

Brazil is larger in size and population (Table 1-1). However when we look at the lower

administration levels, obvious differences diminish. The municipio of Assis Brazil has

2884.2 km2 while the district of Ifiapari has 3,793.9 km2. In the year 2000, Assis had a

population of 3,490 persons. In 1993 Ifiapari had a population of 841; the population of

Ifiapari had changed little by the year 2000, due to internal migration to bigger cities in

Peru.

Table 1-1. Comparisons of land area and population from the country level to the
Municipio of Assis Brazil and the District of Ifiapari
Variable Brazil Peru
Federative Republic Constitutional Republic
Government 26 states 24 departments
5 regions 12 regions
Land area km2 8.5 million 1.3 million
Population 172.6 million 26.1 million
State: Acre Department: Madre de Dios
Land area km2 153,149.91 85,182.62
Administration 5 Development regions 3 Provinces
Municipio: Assis Brasil District: Ifiapari
Land area km2 2,884.23 3,793.94
Population 34906 (2000) 841' (1993)
Municipal Seat: Assis Brasil District capital: Ifiapari
1 Governo do Estado do Acre GEA (2000a), 2Insituto Nacional de Estadistica e
Informatica INEI (1997), 3 GEA (2000a), 4Instituto Nacional de Recursos Naturales -
INRENA (1998), 5INEI (2002), 6Instituto Brasileiro de Geografia e Estatistica IBGE
(2002).

The study sites are located in what is called Southwest Amazonia. Assis Brazil is a

municipal seat in the state of Acre in Brazil and Ifiapari is a district capital in the

department of Madre de Dios in Peru. The area of study consists of the Municipio of

Assis Brazil and the District of Ifiapari (Figure 1-1).

In geomorphologic terms, the area is quite similar on both sides of the border, and

results from the interaction of tectonic, climatic, and erosive factors that have shaped its









surface. The climate is the same: hot and tropical, seasonally humid, and with abundant

rains and a short dry season that usually lasts from June to August.

There are some initiatives that cross the international border between Peru and

Brazil such as the "Development Program for the Peruvian-Brazilian Border

Communities", a bi-national program led by the ministers of foreign affairs of Peru and

Brazil with the assistance of the General Secretariat of the Organization of American

States (GS/OAS). The Brazilian area covers the Municipio of Assis Brazil and the

Peruvian side covers the Province of Tahuamanu; the district of Ifiapari is located within

this province (SUDAM and INADE 1998). Since 2000 there have been annual meetings

known as MAP (Madre de Dios-Acre-Pando) that bring together academic institutions,

international cooperation agencies, non-governmental organizations, and local, state and

national governments committed to sustainable development and conservation in the

MAP area.

According to the document elaborated by the SUDAM and INADE (1998), cities in

the area are mainly rural, work on farms being the principal economic activities. Assis

Brazil and Ifiapari are the main centers with urban characteristics. Both are small towns

cut by a road. Table 1-2 presents a general time line for both towns making a comparison

of the different events that should explain land cover differences.

Assis Brazil and Acre in Context

Acre was the traditional territory of different indigenous groups that fled, were

killed or were forced to move by the "correrias" when the white man arrived looking for

rubber (Hevea brasiliensis). No missions or indigenous slavery existed in the area during

the colonial period (GEA 2000b).
























































r -



LU


* 111 -|


.2 -
1(1 1 .g


r-
I--


a CPIO a












Table 1-2. Comparing historical processes for the Municipio of Assis Brazil and the
District of Iiapari
Year Ifiapari Assis Brazil
1900s Indigenous territory
1900-13 Rubber boom until plantations in Asia took over rubber production
1914-50 Migration to Puerto Maldonado and Tire industry and World War II
Cuzco maintain rubber tapping in Acre
Seringa
1950s Banco de Fomento Agropecuario
1960s Unpaved main roads are built
1970s Agrarian Bank is established Operation Amazonia
1980s Directed Settlement Conflicts over land
1990s Agrarian Bank is closed Secondary roads are built
Main road is paved
2000s Main road is improved Main road is paved
Credit programs are established

Acre was settled when rubber extraction expanded beyond Belem and Manaus;

between 1850 and 1870, suppliers spread their network westward to the Madeira and

Purus rivers (Schmink and Wood 1992). This was Bolivian territory, incorporated to

Brazil in 1903 after a war, with the signing of the Petropolis Treaty (GEA 2000a).

Population composition

Household productive structures in Acre are classified in the following categories

(GEA 2000b):

Ribeirinhos. The first settlements in Acre were reached by river; along their banks

are found most of the municipal seats. The riparian populations established communities

based on family productive activities. They had a diversified subsistence production,

adapted to the Amazonian environment, without large-scale shifting cultivation practices

(GEA 2000b).

Seringueiros. Migrants from the northeast, nordestinos, were the labor force for

rubber extraction. However, English plantations in Asia took over world production of









rubber after 1913, giving origin to the rubber crisis in Brazil. In the 1920s in areas

endowed with dense stands of the castanheira tree (Bertholletia excelsa), Brazil nuts

became an especially important export item (Schmink and Wood 1992). During World

War II the North American and Brazilian governments coordinated to stimulate rubber

extraction in Brazil (Batalha da Borracha) with a second nordestino migration to Acre.

After the armed forces took control of the Brazilian government in 1964, the Operation

Amazonia (Wood and Schmink 1993) began in 1966, and many landlords from the south

and Southeast moved into Acre stimulated by federal incentives on cattle, logging and

mining. This generated major land conflicts in the mid 1970s that resulted in the

institutionalization of Extractivist Settlement Projects in 1987 (Projetos de Assentamento

Extrativista PAEs under Instituto Nacional de Colonizacgo e Reforma Agraria INCRA

administration) and Extractive Reserves in 1990 (Reservas Extrativistas RESEX, under

Institute Brasileiro de Meio Ambiente e dos Recursos Naturais Renovaveis IBAMA

administration). Due to the decline in rubber price, many families now migrate to

agricultural settlements, ranches and urban peripheries. Families that remain in PAE and

RESEX areas move from extractive to farming, cattle and logging activities (GEA

2000b).

Colonos. Farming families are located in the Directed Settlement Projects (Proj eto

de Assentamento Dirigido PAD) and Colonization Projects of INCRA. Traditional state

models of rural settlement in Acre present many problems; for example, family plots may

be designed without considering soil quality, topography or water courses. Their main

farming activities are corn, rice and beans, coffee, papaya, passion fruit, banana and

pineapple. Extractive activities may be directed to wood, brazil nut, acai and game (GEA









2000b). There is also a strong tendency to establish cattle for milk and meat, with

consequences in the increase of deforested areas.

Pecuaristas. The first expansion phase of cattle raising occurred in the 1960s-70s,

when the military regime began the Amazonian Operation, offering incentives for cattle

raising. Prior to this they had cut financial help to rubber traders who were forced to sell

their lands at low prices to the landlords from the South. In the 1970s both the federal and

state governments wanted to convert Acre into a major meat producer. The second phase

(1979-1989) was marked by an increase in degraded areas, mainly because of a pest that

thrived in established pastures. At the same time new pasture species adapted to tropical

weather were introduced, as well as better cattle management practices. The third phase

was initiated in 1989 by the federal government prohibition on the use of official credit

for development activities that result in deforestation in the Amazonia. The development

of environmental laws, the increase in environmental control and the availability of new

technologies and pressures on tropical forest conservation made pecuaristas adopt

strategies to recover deforested areas. But it is clear that there is a strong tendency to

increase cattle raising areas in households of all kinds and size (GEA 2000b).

Natural protected areas

Acre has two Indirect Use Conservation Units (Unidade de Conservacao de Uso

Indireto UCUI), the Serra do Divisor National Park and the Acre River Ecological

Station, both are under the administration of IBAMA. The latter is located in the

southeast, distributed in the municipios of Assis Brasil and Sena Madureira. It was

created in 1981 by the federal government. The Ecological Station also borders two

Indigenous Lands (Terras Indigenas TIs under the administration of Funda9ao Nacional

do Indio FUNAI). The TI Cabeceira Do Rio Acre covers 78.513 ha, and has 134









persons, from the Jaminawa ethnic group. Environment conflicts in the Ecological

Station and TIs are caused by the proximity to Assis Brazil, which increases logging and

game activities by poachers (GEA 2000b).

Transportation and highways

The oldest transportation system in Acre is the rivers, the Madeira River being the

most important as it is the cheapest way to get to Manaus. The Acre River is also

important during the rainy season. There is aerial domestic transportation from Rio

Branco (flights from Rio Branco to Puerto Maldonado are offered in a less consistent

manner). Finally there is a network of federal and state highways, municipal roads and

INCRA's secondary roads.

The federal government during the military dictatorship, seekign to promote

Brazil's industrialization, gave priority to the construction of highways. In the case of

Acre they were: BR-364 (Rio Branco Peruvian border), BR-317 (Labrea/AM Assis

Brasil/AC Peruvian border), BR- 307 (Marechal Tahumaturgo/AC Benjamin

Constant/AM Venezuelan border) and BR-409 (Feij6 Santa Rosa).

Brazil's transport authority (BR) is the responsibility of the National Highway

Department (Departamento Nacional de Estras e Rodagem DNER), which delegated

some highways to the Army, some of which were sub-delegated to the State Highway

Department (Departamento de Estradas e Rodagem DERACRE). INCRA secondary

roads (to settlement projects, colonization and others) are under its responsibility, some

of which were also transferred to DERACRE.

The most recent infrastructure development in Acre, part of the Avanca Brasil

Program, is precisely the paving the BR-317/AC Brasileia Assis Brasil. This has

meant the paving of 110 km with the purpose of integrating the Southwest of Acre with









the national highway system and to promote relations with Peru, making it possible to

access the Pacific Ocean (Ministerio do Planejamento 2003).

Iiiapari and Madre de Dios in Context

The era of rubber extraction (1895 1940) marked the beginning of the non-

indigenous settlement of Madre de Dios. During this period international companies

brought the immigration of large numbers of peasants. Thus, the area of the Acre and

Manuripe rivers has the highest number of towns in the department. Between 1900 1924

peasants from Cajamarca, La Libertad, Arequipa and Loreto, as well as European and

Japanese immigrants also came to the area, and due to concern about land tenure, the

Peruvian government created the city of Puerto Maldonado in 1902 (INRENA 1998).

The decline in rubber prices first, and then the Cuzco Puerto Maldonado road

constructed in 1961, stimulated migration to Puerto Maldonado and Cuzco. The

remaining population in the Iberia Ifiapari area were based on a subsistence economy

(INRENA 1998). In the 1960s the government of Feranado Belaunde provided the first

incentives to cattle ranching in Madre de Dios. The government Office for Agricultural

Research (Servicios de Promocion e Investigaci6n Agraria SIPA) established a cattle

ranch in Madre de Dios to expand cattle raising, and encouraged the genetic improvement

of herds (Jorge Coronel cited by Varese 1999).

Gold mining also is an important activity in Madre de Dios. It is practiced in three

main forms: manual, in distant locations as a family activity or with peons; with pumps,

in the Inambari, Madre de Dios, Malinosky and Colorado rivers; and with heavy

machinery, in the Caychihue, Huaypethue, Madre de Dios and Malinosky rivers (Arbex

1997).









Logging is a major activity in Madre de Dios, though since December 6, 1999 the

Ministry of Agriculture outlawed all industrial or commercial logging in the region; after

that the President declared a ban on cedar and mahogany logging in Madre de Dios,

effective from January 1st 2000. Until 2001 the Peruvian Forestry law only authorized the

granting of logging contracts for areas less than 1000 ha, and for less than two years.

Within Madre de Dios such logging was permitted only in the Tambopata province and in

the Tahuamanu district (AIDA and SPDA 2002). A major change in logging activities

was brought with the new Forestry Law (Ley Forestal YDe Fauna Silvestre 2000) and its

regulations (Reglamento De La Ley Forestal YDe Fauna Silvestre 2001) that give forest

concessions for longer periods and over larger areas.

During the 1980s and 1990s the Government carried out two Directed Colonization

Programs in the district of Ifiapari, the Proyecto Especial Madre de Dios in Primavera and

Chilina, both of which failed: desertion of the colony in Chilina is around 56%, and in

Primavera it is around 80% (INRENA 1998).

A study on land use was conducted in the Iberia Ifiapari area in 1997 covering

204,550 ha. The area with permanent crops was 0.1%; banana was the main species, and

papaya, pineapple, coffee, and cacao were also present, usually for subsistence. Families

also had small areas with tomatoes, onions, garlic, lettuce and subsistence crops. Farming

activities were carried out along main roads and rivers. Pastures represented 3.3% of the

area, both abandoned and productive. Forests covered 92.7% of the area (INRENA

1998). According to the study made by the Brazilian Superintendency for Development

of Amazonia (Super Intend6ncia de Desenvolvimento da Amaz6nia SUDAM) and the

Peruvian National Development Institute (Instituto Nacional de Desarrollo INADE),









after the Agrarian Bank of Peru (Banco Agrario del Peru BAP) and the National Rice

Commercialization Enterprise (Empresa Comercializadora de Arroz ECASA) were

deactivated in 1991 crop production has mainly had subsistence purposes (SUDAM and

INADE 1998).

Natural protected areas and indigenous communities

At present Madre de Dios is renowned worldwide for its outstanding biological

diversity, and it has been a place of extensive research. As a result there are four areas of

strict protection: Pampas del Heath National Sanctuary; Bahuaja Sonene National Park;

Manu National Park; and the Tambopata-Candamo National Reserve. All of these are

under INRENA administration.

Madre de Dios has a low population density and is home to diverse indigenous

peoples. The valleys of the Piedras, Yaco, Chandles and Alto Manu rivers are the

ancestral territories of indigenous communities from the Pano family; these indigenous

populations live in the floodplains in the Ifiapari and Iberia provinces. The native

community Belgica (Arawak) is located in the district of Ifiapari. According to the 1991

census, they included 151 persons, but their right to the land they use was not legally

recognized (INRENA 1998).

Transportation and highways

In Madre de Dios the Tambopata and Madre de Dios rivers are important and

inexpensive transportation. Domestic flights are available in the city of Puerto

Maldonado, and sometimes flights to Rio Branco are available. The transportation

network is made up of what is now called the Pacific highway. Starting at the coast of

Peru the route is Matarani Juliaca Puerto Maldonado Iberia Ifiapari. The road is not

paved, and is difficult to transit during the rainy season in its Andean and Amazon









portions. From Puerto Maldonado there is also a road that connects with Cuzco and from

there to Lima. In Peru usually the Transportation Ministry takes care of roads and

highways; however this task is often given to Special Development Projects (Proyectos

Especiales), regional and local governments.

In the District of Ifiapari the main transportation is the road to Iberia, a 50km

unpaved road with some secondary roads, usually in bad shape. This is the road that was

built by the Proyecto EspecialMadre de Dios between 1998 and 2000. Ifiapari also has a

small airport that has hardly been used.

The Case for Comparing Iiiapari and Assis Brazil

This section looks more in depth at certain factors that are usually recognized as

land use drivers and that are also different in nature in Assis Brazil and in Ifiapari. The

discussion is centered on road infrastructure, market and credit.

The Role of Roads

Both towns were traditionally isolated from the rest of their respective countries

until recent years. From the 1960s to the late 1990s they were linked to the rest of their

countries by a dirt road suitable for a walk or to make a trip in a tractor for three days to

get to the closest town. Difficulties were mentioned, especially in cases of health

problems. Infrastructure development in Peru and Brazil towards the construction of the

Pacific highway changed the roads on both sides. There is a general sense among settlers

that the roads are the best thing the government has done for the people living in this area

so far.

In the case of Ifiapari, the unpaved road that links it to the town of Iberia and to the

city of Puerto Maldonado was built in many phases by the Proyecto EspecialMadre de

Dios. Its last portion was declared finished in October 21st, 2000. This road, although not









paved yet, is described by the engineers in the Proyecto Especial as ready to be paved.

Most farms are located along this road and there are only two secondary roads in the

district.

In the case of Assis Brazil, the road that links it to the city of Brasileia (BR-317)

was paved in its final portion in 2002, within the frame of the Avanca Brazil Program,

and for the formal inauguration presidents of both countries got together in December

2002 in Assis Brazil. There are also many secondary roads that connect to the different

farmlands in the Municipio, usually in bad shape during the rainy season.

There is a difference in the process here: both sites had very difficult dirt roads that

could only be passed by 4 x 4 trucks, referred to as "Toyota", during the rainy season;

while in 2000 after a long process an unpaved road passable all year reached Ifiapari, in

the case of Assis Brazil in 2002 the road was upgraded to a paved road in a process that

took less than two years.

The next step in the building of the Pacific highway is the construction of a bridge

over the Acre Rive; initial work is already going on and the 175m bridge, that will cost

US$7 million should be completed by the end of 2004. After that, the paving of the roads

on the Peruvian side should start, although funds are not available yet. One of the issues

that is taking time and effort from the local authorities is whether the Pacific highway

should cross through the center of the towns or if it should go around the towns. It is

apparent that most people would like the highway to cross the towns, for fear of not

getting any benefits if it does not.

The Importance of Markets

Throughout history, from the elders' references, both towns were closely related. In

terms of trading, in the beginning it was Ifiapari who was at an advantage, meaning that









settlers from Assis Brazil would buy goods on the Peruvian side, first things like soap and

candles and later manufactured goods. Although no one could pinpoint when the reverse

process started, the current scene shows a different picture. In 2000, my first visit to the

area, the presence of the "J.B" grocery store (like a small supermarket) in Assis Brazil

was unexpected; in 2003 with the new paved road I found at least two more of these

stores in town selling manufactured goods that came from different parts of Brazil. Prices

are so low compared to prices in Madre de Dios that not only people from Ifiapari but

also those from Iberia and Puerto Maldonado make the trip and cross the border to do

some shopping. But these stores not only sell goods they brought from other parts of

Brazil, they also buy a small amount of products from some local farmers like rice, beans

and fruits. So, from time to time, one can see the peculiar scene of a big ox with its

wooden cart being discharged of farm products and loaded with manufactured goods at

the front door of the grocery store. But while these markets are the places where most

people on both sides of the borders buy manufactured goods, the stores buy only a small

part from local farms in Assis Brazil.

The market for the agricultural products in both towns is much reduced, and farmers

produce mainly for subsistence. On both sides, complaints from producers are the same:

the local market demand is not enough to consume all that is produced in the area, and

transportation cost is the main restrictive factor to take the products to other places, to

Iberia in the case of Ifiapari and to Brasileia in the case of Assis Brazil. For small animals

like chickens, ducks and pigs, the market is usually Puerto Maldonado, even for the

Brazilian livestock that is not supposed to cross the border without a permit from the









Peruvian National Service for Animal and Plant Health (Servicio Nacional de Sanidad

Agropecuaria SENASA) office in Ifiapari.

But the most successful market is the market for beef; buyers from Puerto

Maldonado and from Rio Branco, respectively, go all the way to the farmlands with

trucks that may fit up to 8 -12 animals, depending on their size. Settlers do not take their

own cattle to the city; it is suggested that a "cattle mafia" arrived with the roads. They

buy cattle at your door, and with cash money, usually young bulls that are taken to fields

near the cities to be fed, processed and their meat sold. Before the road, the only way to

get the cattle to the cities was by walking them for 4 days, which was usually done by the

cattle owners.

Timber, however, has a different story in each town. To begin with, logging is not

allowed in Assis Brazil, and permits are officially necessary if one wants to use timber

from one's own land for construction. In Ifiapari, logging has been banned since 1977,

but logging activities were traditional carried out by local small loggers. Although no

farmer or settler says that logging is their only economic activity, it is certainly an

important part of the livelihood strategies for some of them. Logging became an

important source of conflict since 1999, when illegal logging permits were given to a

Peruvian logging company called Empresa Industrial Maderera Tahuamanu EIRL, who

had a joint venture with Newman Lumber Company of Mississippi. The venture installed

the first big sawmill with foreign capital, and a 100km extraction road was opened with

the purpose of extracting mahogany that was directly exported to the U.S. (Caillaux and

Chirinos 2003; AIDA and SPDA 2002). The conflict over timber led to many legal

battles between the National Institute for Natural Resources (Instituto Nacional de









Recursos Naturales INRENA) and the Newman Lumber companyover a three-year

period, as well explained by Calliux and Chirinos (2003). Almost at the same time in

2000, the new Forestry Law was approved (Ley Forestal YDe Fauna Silvestre 2000) and

in 2001 despite all the street protests that took place in Puerto Maldonado, the first forest

concession competition took place, and a total of 99 000 ha, two of the three areas that

were available in the district of Ifiapari, were given in concession to two local small

loggers associations.

Existence and Inexistence of Credit

Credit in Ifiapari has existed since the 1950s with the Banco de Fomento

Agropecuario, whose name was changed afterwards to Banco Agrario del Peru (BAP) by

the Gobierno MIitar Revolucionario of Gral. Juan Velazco Alvarado in the early 1970s.

It provided credit first for rubber, and then in materials and/or cash for agriculture and

cattle, from theearly 1980s until 1991, when the bank was closed under President

Fujimori's regime. The people who were in Ifiapari in los tiempos del banco remember

having received credit for agriculture, for cattle and for small animals. Since the Banco

Agrario there is not a single type of credit available to rural small farmers.

In Assis Brazil, however, the Constitutional Fund for the North (Fundo

Constitucional de Financiamento do Norte FNO) and the National Program for the

Strengthening of Family Agriculture (Programa Nacional de Fortalecimento da

Agriculture Familiar PRONAF) are currently available, at least for some of the farmers

and until 2001 there was also a fund available for cattle ranching.

Roads, Markets and Credits as Land Use Drivers

From Figure 1-1 to the last section, the present chapter provided the base

information to establish the first hypothesis:









HI: Access to markets, credit and road infrastructure drove more deforestation by

households in Assis Brazil than in Ifiapari.

By observing Figure 1-1 one can see that there is a bigger area deforested in the

Municipo of Assis Brazil than in the District of Ifiapari. This comparative study will

therefore focus on key differences in drivers of deforestation that may explain this

difference in land cover in the two sites.

From the information available for both sites it is apparent that there is better road

infrastructure and a better market in Assis Brazil. Credit, however, has different timings:

it was available in the are of Ifiapari for nearly 30 years until 1992, and it is currently

available in Assis Brazil since 1999. There is also a population difference: Assis Brazil

has a higher population than Ifiapari. In order to control for that difference, a household

level approach will be better than a Municipio or District approach.














CHAPTER 2
THEORETICAL APPROACH

Introduction

In an effort to improve current understanding of land use and land cover change

drivers, different analytical and empirical approaches have considered different temporal

and spatial scales. More recently, research has been conducted in Amazonia, focusing on

household level land use practices, as they are often pointed as to being responsible for

deforestation. The present chapter looks into the factors considered to be important land

use drivers, precisely into the ones that are of interest for the present research: markets,

credit and road infrastructure. Then it explores the relevant theories used in explaining

deforestation since this sort of land cover change has been the center of many studies in

the field of land use and land cover change. It finally reviews in more detail three

theoretical frameworks, and concludes by proposing an integrated theoretical framework.

Land Use Drivers and Deforestation

Markets and Credit

The socioeconomic matrix of deforestation for Amazonia elaborated by Schmink

(1994) explains deforestation as an outcome of social processes, and locates markets in

the global and national contexts. At the global level important variables are the demand

for Amazon products (e.g. timber, rubber) and foreign investment (e.g. oil, mining,

timber). At the national level the variables considered are transportation and export

orientation. Schmink (1994) describes international and national markets as an important

factor for the expansion into remote forest areas.









The debate of how economic development impacts the environment includes the

roles of markets in the use of natural resources (Godoy et al. 1997b). The topic has been

extensively researched in the case of traditional populations; however, the basic

principles may apply to any group. According to Godoy et al. (1997b) there are three

main positions in regard to market: the market works to the detriment of conservation; the

market increases conservation, if land rights are secure; and the market has ambiguous

effects on deforestation.

Households integrate into markets by selling crops, labor, or both. If integration is

achieved by selling crops, increase in market demand will increase deforestation, unless

intensification occurs. If integration is achieved by selling labor, increase in market

demand will reduce deforestation since there will be less time to work the land (Godoy et

al. 1997b). Therefore integration into both markets usually has nonlinear effects.

In their review of models of deforestation at the household level, Kaimowitz and

Angelsen (1998) include transportation costs which show an inverse relation between

market access costs and deforestation. They also find that an increase in off-farm income

sources typically decreases the pressure on forests. However, increased participation in

market oriented activities does not always have a positive impact (Schmink 2004).

Market dependency is in many cases not desirable since it leaves little room for

subsistence activities, making the producer vulnerable to changes in demand, and price

(Schmink 2004). Market integration deserves special attention in the case of traditional

groups with little market experience since some of the features of market economy, like

profit, may erode their cultural ties (Smith 1995). Increase in market demand may also

put more pressure on forest resources (Schmink 2004), resulting in increased









deforestation. Furthermore, social justice and equity are not expected outcomes from

market participation (Schmink 2004).

Government subsidies are one way of addressing the local producer disadvantages

(Schmink 2004). In the case of the Brazilian Amazon subsidies include a combination of

road building, colonization projects, and taxes and credits that have helped to foster the

frontier expansion process (Wood and Schmink 1993).

Roads Driving Land Use

Access has been recognized as the main factor in the spatial distribution and rate of

deforestation (Soares-Filho et al. 2002). Historically, rivers and roads have provided easy

access to tropical forests, but roads are especially associated with deforestation and social

conflicts (Schmink and Wood 1992). However, road improvement is an important

priority in many Amazonian countries (Maki et al. 2001) and transportation plays a

important role in development (Leinbach 2000). In most cases it is necessary, but not

sufficient for development. Local social and environmental characteristics have an

important effect in the way roads influence economic and social changes. Moreover,

roads do not always help to alleviate poverty; development will depend on the local and

regional economies' capacity to reallocate resources (Leinbach 2000).

In their review of models of tropical deforestation Kaimowitz and Angelsen (1998)

found that roads, rivers, railroads, and low gas prices provide greater access to forests.

Roads usually lead to more deforestation when there is also access to markets, especially

in areas of good soil with commercial agriculture. However, the relation is not always

direct; in some cases roads are built in previously settled and cleared areas, or settlement

and roads may be influenced by other variables.









Alves (2002) made an analysis of the geographical distribution of the deforested

areas in the Brazilian Legal Amazon. His results show that deforestation is concentrated

around major roads and pioneer settlements. Three quarters of the deforestation between

1978 and 1994 were within 50 km (on each side) of major roads.

The Brazilian Development Program Advance Brazil (Avanca Brasil ) effective

since 2000, and projected to be active until 2007, involves the paving of 6000 km of

roads (Ministerio de Planejamento e Orcamento 1999). Some of the roads to be paved

will provide access to 31 indigenous lands and 26 conservation units and remote forest

regions (Nepstad et al. 2000). This project has brought the issue of road paving and its

relation to deforestation to the front line and different scientific teams have worked in

developing deforestation scenarios for the Legal Amazon.

In modeling past deforestation, Laurance et al. (2001) found that paved roads have

more far-reaching effects than unpaved roads. On average, areas further than 25 km from

an unpaved road have less than 15% forest loss, but for paved roads average forest loss is

15% for areas between 26 and 50 km from the paved road. Nepstad et al. (2001) found

that 29-58% was deforested within 50km from paved roads and 0-9% for unpaved roads.

Paved roads produce three vicious cycles: the first is related to the cycle of cattle

ranching, annual crops and its reinforcement by the use of fire; the second is related to

conventional logging and its implications for wildfire during severe droughts; and the

third is related to rain inhibition due to the former cycles (Nepstad et al. 2000; Nepstad et

al. 2001). Investments in forest management and perennial crops would decrease the use

of fire, aswould the establishment of rules in the use of fire, and the incorporation of fire

prevention incentives in currently available credit lines (Nepstad et al. 2001).









In economic terms the opening of new frontiers increases land supply, reducing the

land value in the older frontiers. It also represents more work for the government to

monitor natural resources use, as well as to provide health, education, and technical

assistance (Nepstad et al. 2000; Carvalho et al. 2001). In general, it will encourage

colonization and forest clearing that the government does not have the capacity to control

(Laurance et al. 2001).

In economic terms the opening of new frontiers increases land supply, reducing the

land value in the older frontiers. It also represents more work for the government to

monitor natural resource use, as well as to provide health, education, and technical

assistance (Nepstad et al. 2000; Carvalho et al. 2001). In general, it will encourage

colonization and forest clearing that the government does not have the capacity to control

(Laurance et al. 2001).

The proposed alternative is to promote old frontier areas through local road

networks that allow producers to reach trading areas, with technical assistance, and with

health and education programs (Nepstad et al. 2000). Therefore, roads where settlements

already exist are desirable like Altamira-Maraba and Brasileia-Assis Brasil, and roads

that open new frontiers like Santarem-Cuiba and Humaita-Manaus are not (Nepstad et al.

2001; Nepstad et al. 2000; Carvalho et al. 2001). The most important issue is that

priorities for transport policy in rural areas must meet the necessities of the poor

population and not those of elite groups (Leinbach 2000), like has happened in the past

(Schmink and Wood 1992; Wood and Schmink 1993).The increase of governance in

frontier areas through strengthening of existing management institutions, adequate land

use planning and enforcement of the existing environmental legislation is recognized as a









very much needed measure to control land conversion following road paving (Nepstad et

al. 2002; Maki et al. 2001).

Approaches to Explain Deforestation

To analyze the effect of different land use drivers it is necessary to pay special

attention to the specific scale and place of research in order to make the appropriate

assumptions. In general, most studies have been made at the country or cross country

level in tropical areas. Four of the most important approaches for deforestation analysis

were reviewed by Barbier (2001): the Environmental Kuznets Curve (EKC), competing

land use models, forest land conversion models and institutional models. The EKC

hypothesizes that environmental problems (like deforestation) decrease as the per capital

income of a country rises. For Latin America and Africa the per capital income level at

which deforestation equals zero is two to four times higher than the current average

(Cropper and Griffiths 1994).

Competing land use models hypothesize that deforestation results from competing

land use, mainly between forests and agriculture. Therefore an opportunity cost is

calculated for agricultural conversion versus potential timber and environmental services

from forests. An important consideration is that very often agricultural conversion

follows timber extraction (Barbier 2001). Forest land conversion models assume that

households use their own labor or hire labor for land conversion, the level of cleared land

is hypothesized to be a function of output and input prices, but these data are usually

difficult to obtain. Institutional models, on the other hand, are centered on factors like

ownership and political stability.

Kaimowitz and Angelsen's (1998) more extensive review of economic models of

tropical deforestation identified three primary levels: household and firm, regional and









national and macro models. Household and firm level models, the relevant level for the

present research, were divided into three categories: analytical open economy (e.g.

Angelsen 1999), analytical subsistence and Chayanovian (e.g. Angelsen 1999), and

empirical and simulation models (e.g. Godoy et al. 1997a). Analytical models are

theoretical constructs; they allow researchers to examine the implications of their

assumptions. Empirical models quantify the relationships between variables using

statistical methods, and simulation models use parameters to assess scenarios under

different circumstances. Open economy models assume that households' and firms'

actions have no impact on prices, and that market prices (including labor) fully determine

how they value their resources. In this way household production is analyzed as profit

maximization oriented. Subsistence and Chayanovian models assume imperfect markets

(particularly labor) and that household consumption does matter in production decisions.

The household goal is to maximize utility. Empirical and simulation models require time

consuming surveys for data collection. For these models, common independent variables

are transportation costs, farmer's background, credit access, input and output prices, and

tenure security (Kaimowitz and Angelsen 1998). The different assumptions and methods

in each category show different results. The summary of findings for the analytical

models are presented in Table 2-1.

Site specific characteristics may play an important role in land use decisions, which

may result in different land cover outcomes. That is precisely why empirical research is

important and limited at the same time. While it provides site specific insights, findings

should not be extrapolated without serious considerations and reservations.









Table 2-1. Models of deforestation showing predicted effect of key variables
Variable Analytical Model
Subsistence Chayanovian Open economy
Higher agricultural prices Reduce Indeterminate Increase
Population growth Increase Increase No effect
Lower transport costs No effect or reduce Increase Increase
Higher agricultural productivity Reduce Indeterminate Increase
Higher wages NA (reduce) NA (reduce) Reduce
Higher land prices Reduce
Higher interest rates Reduce
Adapted from: Kaimowitz and Angelsen (1998)

To situate the present research within the categories at the household level set out

by Kaimowitz and Angelsen (1998) we may say that the present research is analytical and

empirical. It proposes an analytical integrated framework and uses survey data to

quantify the relationships between land use and land cover outcomes and variables

commonly perceived as land use drivers from open economy models, and household

demographic variables from chayanovian models.

Empirical models of land use that incorporate demographic variables at the

household level were reviewed by Perz (2001). His revision is specific to the Amazon

and to the Neotropical Americas. Table 2-2 is an updated version from Perz (2001).

General observations from his results show that the age of the household, length of

residence, family size, and number of family members all influence land use in terms of

area deforested, in old or secondary growth forest, as well as household agricultural

decisions, among many factors.










Table 2-2. Household demographic variables used in land use modeling for Amazonian
areas.

Demographic variables
Age of Length Adults
Variable household of Family (males,
head residence size females)b Children

(1) (2) (3) (4) (5)
Alston et al. (1993)
% area annuals +ns c +ns -ns
% area perennials -ns +ns -ns
% area pasture -ns +ns +ns
Rudel and Horowitz (1993)
% land deforested -*
Oz6rio and Campari (1995)
Ha deforested since arrival -ns +ns
Ha deforested in 1991 +ns +ns
Sydenstricker and Vosti (1993)
Ha deforested +**, +*
Jones et al. (1995)
Cleared ha per year
Total cleared ha +ns
Alston et al. (1996)
% area crops or pasture +ns c -ns
Godoy et al. (1997a)
Ha old growth forest +ns c -ns -ns
Godoy et al. (1997b)
Prob. Of cutting old growth forest
Mojeno +ns -ns
Yurucare +*
Chimane +ns -ns
Pich6n (1997)
% land in annuals -ns +**
% land in perennials +ns +**
% land in pasture +** +ns
% land in forest -** -*
Godoy et al (1998a)
Ha primary forest cut -** +** +**, +ns +ns
Godoy et al (1998b)
Ha old growth forest cut +ns +** -ns
Godoy et al (1998c)
Ha primary forest cut -ns +ns +ns
McCraken et al (1999)
Annual deforestation 1988-1991 -*
Ha forested in 1991 -*
Source: Perz 2001. Authors appear in chronological order. A "+" indicated a positive or direct effect, and a
"-" indicates a negative or inverse effect. A "ns" indicates not significant atp > 0.10, indicates
significance p < 0.10, and ** indicates significance at p < 0.05.










Table 2-2. Continued

Demographic variables
Age of Length Adults
Variable household of Family (males,
head residence size females)b Children

(1) (2) (3) (4) (5)
Wood and Walker (2000)
Ha deforested on arrival +* +*
Ha deforested at interview +* +*
Cocoa +* +*
Coffee -ns +*
Ha pasture +ns +ns
Head cattle +ns +*
Reforestation +* +ns
Gomes (2001)
Area deforested +ns +**, +ns
Pasture size +ns
Head of cattle +ns
Perz (2001)
Annuals +ns -ns +ns +ns
Perennials +ns +** +** +ns
Pasture size -ns +** +** -ns
Cattle +ns +** +** -ns
Reforestation +ns +ns +ns +ns
Perz (2002a)
Area in forest -ns +ns -**
Area in cropland -ns +** +**
Area in pasture -ns +ns -ns
Area in secondary growth +** +ns +**
Perz and Walker (2002) tobit 1,2,3
na, +ns, na, +**,
Secondary forest growth under fallow +ns, +**, +** +ns +**
Source: Perz 2001. Authors appear in chronological order. A "+" indicated a positive or direct effect, and a
"-" indicates a negative or inverse effect. A "ns" indicates not significant at p > 0.10, indicates
significance p < 0.10, and ** indicates significance at p < 0.05.



Integrated Theories

While a good understanding of the effect of individual land use drivers that are

relevant for this study, as well as an overview of the existing theories that intend to

explain deforestation reveals limitations in the tools available to adequately address land

use and land cover change issues when dealing with a complex reality. That is the









primary reason for the existence of integrated theories that intend to provide a better

explanation for changes in land use and land cover.

The framework used in this study comes from the integration of three frameworks

drawing on three different integrative theories. The first is a hierarchical framework

drawn from the bases of political ecology theory. The second looks at household

transformations from demography theory. These two frameworks explain land use at

different levels. The third framework provides the elements to link the previous ones and

to make them more flexible; it is drawn from panarchy theory

These three integrated frameworks were chosen for various reasons,. First, political

ecology and demography or Chayanovian theories have been used in explaining land use

and land cover changes in frontier areas in the Amazon region. Political ecology is

explicit in linking global and local events; it brings in the spatial dimension, and it is

preferred over open economy theories because their assumptions of perfect markets and

information are far from real in the Amazon frontier. Demography or Chayanovian

theories provide a tool to work at the household level, as required by this research, and

bring in a temporal factor by looking at household composition over time. Panarchy

presents an innovative opportunity to link political ecology and Chayanovian theories,

since it has explicit temporal and spatial dimensions and has its origins in natural

resource use.

The purpose of integration is to show that biophysical and socioeconomic drivers

of land use and land cover change do not have a linear or constant influence on

households and their land use systems (farming, ranching, logging, etc). Instead, much









depends on the stage that the household and the productive system occupy in terms of life

cycles.

All components of theory are linked together in a coherent conceptual structure

named a theoretical framework. Integration is the union of existing theory, perspectives,

approaches, models or data that are apparently disparate (Kuchka 2001). It is important to

find out how paradigms, theories and theoretical practices themselves limit integration

and how those constraints may be overcome; theoretical understanding changes through

integration (Kuchka 2001).

Integration is a difficult task. Some of the procedures and circumstances necessary

for successful theoretical integration include (Kuchka 2001):

1. Domain: the domain of the related theories must be clearly stated, to make the
development of linkages between theories more feasible.

2. Concepts: the meanings and subjects of concepts should be clear; this enables the
asking of new questions that may further integration and the development of
theory.

3. Scale and level: it should be clear if the theories are answering questions across
levels of organization or particular adjacent levels of a given scale.

For the particular research question of this study, it is my intention to integrate

frameworks from three different integrative theories: (1) Political Ecology, (2)

Demography, and (3) Panarchy. I will begin by reviewing some of their basic

components:

1. Basic conceptual devices (assumptions, definitions and concepts)
2. Framework and structure (framework and domain)

Political Ecology

Political ecology has its origins as a new research field in the 1970's. It was a

reflection of a need for 'an analytical approach integrating environmental and political









understanding' given the increase in environmental problems in the Third World (Bryant

1992; Bryant and Bailey 1997). In its first phase from the late 1970s to the mid 1980s

political ecology was mainly a critique of neo-Malthusian and cultural ecology, and had

its theoretical base in neo-Marxism. In its second phase from the late 1980s to the 1990s;

it was mainly a critique of deterministic neo-Marxism and had its theoretical base in neo-

Weberianism, social movement and household/feminist theories.

Empirical analysis in this field has been favored; this has resulted in a research

field 'grounded less in a coherent theory than in similar areas of inquiry' (Peet and Watts

1996; Bryant and Bailey 1997). These areas of inquiry are only generally similar since

different scholars have adopted different approaches to the same issues. Political

ecologists have sought to explain Third World environmental change and conflict in

terms of key environmental problems, concepts, socioeconomic characteristics, actors

and regions, or they have used various combinations of these approaches (Bryant and

Bailey 1997).

Political ecology is in part based on the assumptions and ideas of political economy

theory (Bryant and Bailey 1997). Blackie and Brookfield (1987) stated that political

ecology considers ecology concerns within a broadly defined political economy. In

general political ecologists agree on two basic points: first, the environmental forces

facing the Third World are not simply a reflection of policy or market failures, but rather

are a manifestation of broader political and economic forces associated with the

worldwide spread of capitalism. Second, there is a need for changes to local, regional and

global political-economic processes (Peet and Watts 1996).









Political ecology addresses the political, economic, and cultural factors underlying

human use of natural resources and the complex interrelations among people and groups

at different scales, from local to global (Blaikie and Brookfield 1987; Schmink and Wood

1987). Elemental political issues of structural relations of power and domination over

environmental resources have been seen by a variety of scholars as critical to

understanding the relationship of social, political, and environmental processes (Scoones

1999). The view of resources as socially and politically constructed has been central to

this discussion and has resulted in important work on how perspectives in environmental

change must be gauged from the view points of different actors (Blaikie 1995).

The perception of an unequal relationship between politics and ecology explains in

part the fact that political ecologist tend to favor consideration of the political over the

ecological (also because of the social science background of most political ecologists).

But they should not overlook advances in the understanding of ecological processes

derived from the New Ecology since, in doing so they might miss an important part of the

explanation of human-environmental interaction (Scoones 1999; Bryant and Bailey

1997).

According to Bryant and Bailey (1997) there are five main approaches or similar

areas of inquiry in Third World Political Ecology, although many times they are

combined. These approaches are the following:

In the first approach the explanation centers around a specific environmental

problem or set of problems such as soil erosion, tropical deforestation, water pollution or

land degradation. This approach constitutes in many respects a 'traditional' geographical









research theme associated with understanding the human impact on the physical

environment (Goudie 1993), but with a distinctive political-economy twist.

The second approach focuses on a concept that is perceived as having important

links to political-ecology questions. To understand the latter is partly to appreciate the

way in which ideas are developed and understood by different actors, and how attendant

discourses are developed to facilitate or block the promotion of a specific actor's interest

(Escobar 1996).

Third, inter-linked political and ecological problems are examined within the

context of a specific geographical region. 'Regional political ecology' has reflected a

concern to take into account environmental variability and the spatial variations in

resilience and sensitivity of the land, as well as 'theories of regional growth or decline'

(Blaikie and Brookfield 1987).

Fourth, political-ecological questions are explored in light of socio-economic

characteristics such as class, ethnicity or gender.

A final focus is on the interests, characteristics and actions of different types of

actors in understanding political-ecology conflicts. An actor-oriented approach seeks to

understand such conflicts (cooperation too) as an outcome of the interaction of different

actors pursuing often quite distinctive aims and interests (Long and Long 1992).

The present research will use a combination of the first and third approaches,

focusing on a particular environmental problem (deforestation) from a regional

perspective. It will center on land use and land cover outcomes driven by market,

infrastructure and credit availability.









Household Demography

The theoretical foundation of the role of household life cycle in land use was

established by Chayanov (1966). His study of peasant farming practices in Russia, in the

first half of last century, serves as a reference mainly because the October Revolution in

Russia created conditions of land abundance similar to agricultural frontiers in the

Amazon region (Walker et al. 2002). His theory explains differences in farm size and

surplus production in relation to household structure. Chayanov distinguished households

according to the ratio of consumers/workers. This relation led him to describe the

household life cycle, where young households with many children have low labor power,

and mature households with high labor power have larger holdings (Walker et al. 2002;

Perz 2001).

The basis of Chayanov's theory is that the drudgery of labor increases

exponentially as work is done, while on the other hand the marginal utility of goods

decreases as they are acquired; the household production level determined by the

intersection of these curves. Marginal utility is determined by the standard of living,

which consists of: the amount necessary to support one consumer; the number of

consumers each worker has to support (the consumer/worker ratio); the amount that has

to be reinvested in the farm to maintain its production; and any other factors that require

part of the farm's production. Drudgery is a measure of the noxiousness of labor and is

inversely related to productivity. The more productive a technique is, the greater is the

output of per unit of labor, and the lower the drudgery (Tannenbaum, 1984).

While Chayanov's theory provides the basis for demographic theory related to land

use, some assumptions about peasants do not apply directly to Amazonia. The theory

does not address the issue of migration, an important feature in frontier Amazonia where









many peasants migrated from places with a different landscape. It assumes closed

household life cycles, while in Amazonia households are not always detached from labor

or products markets, but rather have different assets. Chayanov assumed the existence of

relatively homogeneous farming practices among households, something that doesn't

necessarily happen in Amazonia where, besides farming, cattle ranching and forestry also

are land use options (Perz 2001).

At this point it becomes important to make a distinction among household assets.

Ellis (Ellis 2000) considers household assets are resources owned, controlled or claimed

by the household. These assets mediate the way in which households become involved in

production and labor markets, and participate in exchanges within their community.

Assets can be understood as capital or resource stocks that may be used for household

survival. In a general way, assets are divided in five classes. Natural capital comprises

land and water. Physical capitals are buildings, machines and roads. Human capital is the

labor available, including education, skills, and health. Financial capital is money as

savings and/or credits, while social capital is kinship and community networks (Ellis

2000).

Household demography has its focus on human capital assets; however, according

to the specific case other assets should be considered. For the purposes of this study some

aspects of natural, physical and financial capitals will be included.

Panarchy Theory

Theories like Panarchy intend to explain not what is, but what might be. They will

not predict the details of future possibilities, but might help to identify the conditions for

future possibilities (Holling et al. 2002a). Their objective is to enable the understanding

of economic, ecological and institutional systems and their interactions. The cross-scale,









interdisciplinary, and dynamic nature of the theory gives it its name. Its essential focus is

to rationalize the interplay between change and persistence, between the predictable and

unpredictable (Holling et al. 2002a).

According to this theory, the stabilization of target variables leads to slow change

in other ecological, social, and cultural components, and those changes may lead to the

collapse of the entire system (Holling et al. 2002a). Decline in variability and diversity

creates conditions that cause a system to flip into an irreversible (typically degraded)

state controlled by unfamiliar processes. The magnitude of disturbance that can be

absorbed before the system changes its structure by changing the variables and processes

that control behavior is named ecosystem resilience (Holling and Gunderson 2002).

According to Holling and Gunderson (2002) resilience has three defining

characteristics: The first is the amount of change a system can undergo (and, therefore,

the amount of stress it can sustain) and still retain the same controls on function and

structure (still be in the same configuration-within the same domain of attraction). The

second is the degree to which the system is capable of self-organization. When managers

control certain variables in a system, they create inter-variable feedbacks that would not

be there without their intervention. The more "self-organizing" the system, the fewer

feedbacks need to be introduced by managers. And third, is the degree to which the

system expresses capacity for learning and adaptation.

Semi-autonomous levels are formed from the interactions among a set of variables

that share similar speeds. The organizations and functions we now see embracing

biological, ecological, and human systems are therefore ones that contain a nested set of

the four-phase adaptive cycles, in which opportunities for periodic reshuffling within









levels maintain adaptive opportunity, and the simple interactions across levels maintain

integrity (Holling et al. 2002a).

Integrated Frameworks

In this section I will try to integrate three frameworks from the three integrative

theories we reviewed in the preceding section. The frameworks are the following:

1. The conceptual framework that uses a three-tiered hierarchical approach to depict
the socioeconomic and biophysical drivers that led to deforestation, elaborated by
Wood (2002) which comes from a Political Ecology view.

2. A conceptual framework of household transformations, land use and environmental
change (McCracken et al. 2002), which comes from Demographic theory

3. The 'Adaptive cycle' elaborated by Holling and Gunderson (Holling and
Gunderson 2002) which is within Panarchy theory.

The Three-Tired Hierarchical Approach

This framework treats land cover outcomes as the direct effect of the land use

decisions made by rural households whose decisions are embedded in contexts that

operate at higher levels of the system. The higher level contexts consist of the proximate,

intermediate and distant drivers that comprise the socioeconomic and biophysical

subsystems. The analytical focus is on the relationships that take place within each level,

as well as the cross-level dynamics that link one level to another (Wood 2002).

This model considers socioeconomic, as well as biophysical drivers. To depict the

driver forces hierarchy, it presents the proximate, intermediate, and distant scales. The

model assumes that land use decisions made by firms and households in rural areas are

the result of interactions of a large number of variables acting at different scales within

the social and natural system. They are located at the center and within community and

kinship networks, meaning that networks at the local level may influence land use









decisions. The model also considers the feedback effect that land cover may have in the

socioeconomic and biophysical drivers (Wood 2002) (Figure 2-1).

Each land use/land cover outcome is associated with different kinds of economic

activity, and therefore with different social groups. Rubber tappers, farmers, ranchers and

loggers all engage in clearing the forest cover, but they do so in varying degrees

depending on their respective objectives, resources, and decisions (Wood 2002).

Outcomes can be arranged in five main categories: undisturbed forest, harvest of non-

timber products, selectively logged forest, cleared (annual crops, perennial crops, pasture,

mining), and regrowth (managed fallow, abandoned plots).

Household Transformations Land Use And Environmental Change

This approach is based on Chayanovian theory and on the work of the Centro

Agro-Ambiental do Tocantins (1992 cited by Walker et al. 2002), Walker and Homma

(1996) and McCraken et al.(1999). It emphasizes the role of household labor in land use

decisions in agricultural frontiers (McCracken et al. 2002; Perz 2002b; Brondizio et al.

2002) It is seen as a complement, not an alternative, to models focusing on

environmental and economic factors, like the different drivers presented by Wood (2002).

It was Walker and Homma (1996) who placed households in a context of labor and

product markets, capital availability, and land use differentiation in Amazonia (Perz

2001). According to this framework, as the household looks for its consolidation,

different stages of land use are linked to different household life cycle stages.















Distant


* International
* Exchange rat
* Commodity p


Inte


Po[
Col
Tax


* Fiscal incentives
* National and
regional markets


Proximate Socio-Economic Drivers
(Local level)


* Transportation
* Tenure security
* Access to credit


Kinship networks

Household / Firm Land
Resource allocation Use

Community networks


diate Bio-physical Drivers
Landscape/basin level)

Water table
Drainage
Meso climate


cal Drivers
el)


Y


Land-Use / Land-Cover
Outcomes

Undisturbed forest
L ucoe


j.I


* Harvested of non-
timber products
* Selectively logged
* Cleared for
annual crops
perennial crops
pasture
mining
* Regrowth
managed fallow
abandoned plots


4.
<*


Figure 2-1. The three-tiered hierarchical approach. Socioeconomic and biophysical
drivers of land use are classified in: distant, intermediate and proximate. Land
cover outcomes are direct effect of the land use decisions made by rural
households and firms. Feedback intensity is represented by the varying
thickness of return arrows.


regimes
es
rices


rmediate Socio-Economic Drivers
(National/Regional level)


Socio-Economic Drivers
(Global level)


* Trade policies
* External debt
* World markets


* Commodity prices
* Input and labor costs
* Local politics


pulation growth
onization policies
Slaws


C,


Intermec


Topography
Altitude
Precipitation

Distant Bio-physi
(Global lev

* Global climate
* Atmospheric chemistry


Proximate Bio-physical Drivers
(Local scale/farm site)

* Soil fertility Pests
* Precipitation Pathogens
* Geomorpholog, Micro-climate

I


L /-----------~ 1


4









Recent settler families in a frontier are assumed to be small young nuclear

households, with a head couple and a few young children. Due to small requirements in

land and capital, and the low level of risk, they first clear small areas of forest to cultivate

annual crops (Perz 2001). These are mainly for consumption and local markets. As the

family grows in age and size, additional site knowledge increases and more household

labor is available, more areas are cleared, and previous plots are left uncultivated, formed

into pasture, or planted in perennial crops, causing a decline in deforestation rates

(McCracken et al. 2002). This change is slow, and involves high initial capital and labor

cost.

Economic gains from cattle and perennial crops will be perceived in future years.

While perennial crops in general will not provide any returns for three to five years,

acquiring cattle may be an important capital-saving strategy because it can be quickly

purchased or sold. Perennial crops are also more labor intensive than cattle raising, one

reason why older households with less labor usually shift into pastures (Perz 2001;

McCracken et al. 2002). Access to resources like good soil, water, capital / credit,

markets, technical support and household labor affect the shift to either perennial crops or

raising cattle, or the decision to remain in annual cash crop activities. It is assumed that in

the first household stages most families exhaust their initial capital reserves. Thus, labor

of adolescent and teenage children may be a determining factor along with credit

possibilities for furthering farm investments (McCracken et al. 2002; Brondizio et al.

2002).

Finally, as Perz (2001) indicates, an important final land use is reforestation. When

children become adults, at the point of inheriting their parent's land the family may plant









trees for long-term timber production. Reforestation initially requires capital and labor,

but after the establishment of the plantation, little attention is required.

The schematic representation of McCracken et al. (2002) presented in Figure 2-2

defines five stages of a household life cycle, each one linked to a particular time of

residence in the area, demographic composition, and land use practices.

Early farm consolidation is associated with credit, capital, and large supply of

household labor. Troubles in farm consolidation are associated to reliance on annual

crops and restricted supply of household labor (McCracken et al. 2002).

The Adaptive Cycle

In case examples of regional development and ecosystem management, it has been

found that three properties seemed to shape the future responses of the ecosystem,

agencies, and people (Gunderson et al. 1995; Holling and Gunderson 2002):

1. the potential available for change, since that determines the range of possible
options;

2. the degree of connectedness between internal controlling variables and processes, a
measure that reflects the degree of flexibility or rigidity of such controls,

3. and the resilience of the systems, as a measure of their vulnerability to unexpected
or unpredictable shocks.

Potential, connectedness and resilience

The framework is partly based on the traditional view of ecosystem succession seen

as being controlled by two functions. The first is exploitation, in which rapid colonization

of recently disturbed areas is emphasized. The second is conservation, in which slow

accumulation and storage of energy and material are emphasized. In ecology, species of

the exploitive phase are named r-strategists and in the conservation phase k-strategists

(Holling and Gunderson 2002).
















Land Use & Environmental Change
Stage I Stage II Stage III Stage IV Stage V
Land-Uses/Activities
Deforestation:
Sec. Succession:.......
Annual Crops: ...................
Perennial Crops: ......
Pasture/Cattle: .........
Time 5 yrs. 10 yrs. 15 yrs. 20 yrs. 25 yrs.

Household Stages

Nuclear Young -- ------ i -
Adults w/ Small I
Children
I I
Nuclear Adults w/ ----------- --- --------- ------ -- ----- ---- ---
Older Children II
& Adolescents

Nuclear- Adults .-..-..---- a ..-.--i--- -- -- I~
w/ Teenage III
Children

Nuclear Older ...-...-...-. -.-- -... j.--. ...-
Adults w/ Teenage IV
& Young Adult
Children

Multi-Generational V ._._._._.._.. .._._._._._._._.- _.._. _.... ... ..-.... ......
& Second Generation
Households



Figure 2-2. Household transformations, land use and environmental change. This
framework highlights the role of household labor over the domestic life
course. In the upper section it suggests a pattern of land use. The thickness of
each line represents the level of activity for each land use. Land use stages (x-
axis) are linked to different household stages (y-axis). The diagonal from the
upper left to the lower right represents a general course of farm formation and
domestic life. Deviations to the right are associated to early farm
consolidation linked to credit, capital and larger supply of household labor.
Deviations downward are associated with difficulties in farm consolidation
linked to greater reliance on annual crops and a restricted supply of labor.









Later understanding in ecology indicates that two additional functions are needed.

The first is release or "creative destruction", in which the tight-bound accumulation of

biomass and nutrients becomes increasingly fragile until suddenly released. This is

named the omega (Q) phase. The second function is reorganization, in which the

remaining elements after the omega (Q) phase are rearranged; this is known as the alpha

(a) phase (Holling and Gunderson 2002).

During this cycle, as shown in Figure 4, biological time flows unevenly. The

progression in the ecosystem cycle proceeds from exploitation (r), slowly to conservation

(K), very rapidly to release (Q), rapidly to reorganization (c), and rapidly back to

exploitation (Holling and Gunderson 2002).

The cycle reflects changes in two properties: (1) Y-axis: the potential that is

inherent in the accumulated capital of biomass and nutrients. (2) X axis: the degree of

connectedness among variables (see figure 3). As the system goes from exploitation to

conservation, connectedness and potential increase (Holling and Gunderson 2002). For

resilience, as the system goes from exploitation to conservation, resilience shrinks; and it

expands as the system goes into reorganization.

Hierarchies and panarchies

There are many possible interactions among phases at one level and phases at

another level. Two are considered specially important. These are the connections named

"Revolt" and "Remember"; these connections become important at times of change in the

adaptive cycles (Holling et al. 2002b).






46





















Figure 2-3. The adaptive cycle, the four ecosystem functions (r, K, .) and the flow of












suggests the stage where the potential can leak away and where a flip is most
0 C)
--

likely into a different system.


I A eX^ p! .-, ... ^ *

r

*coiriC co ctdn,3ss -

Figure 2-3. The adaptive cycle, the four ecosystem functionis(r, K, Q,o) and the flow of
events among them. The arrows show the speed of that flow in the cycle,
where short, closely spaced arrows indicate a slowly changing situation and
long arrows indicate a rapidly changing situation. The cycle reflects changes
in two properties, (1) Y-axis: the potential that is inherent in the accumulated
capital of biomass and nutrients. (2) X axis: the degree of connectedness
among variables. The exit from the cycle indicated at the left of the figure
suggests the stage where the potential can leak away and where a flip is most
likely into a different system.

When a level in the panarchy enters its omega phase and experiences a collapse,

that collapse can cascade up to the next level by triggering a crisis, particularly if that

(higher) level is at the K phase where resilience is low. This is termed "Revolt". When a

level enters its omega phase, the opportunities and constraints for the renewal of the cycle

are strongly organized by the K phase of the next higher and slower level. This is termed


"Remember" (Holling et al. 2002b).









There are two distinctions between panarchy representation and traditional

hierarchies. The first is the importance of the adaptive cycle, and the alpha phase as the

engine of variety and generator of new experiments within each level. The second is the

connections between levels. The levels of a panarchy could therefore be drawn as a

nested set of adaptive cycles (Holling et al. 2002b).

Framework Integration

Now that the information on the theories and on the specific frameworks we are

interested in has been reviewed the next step is to bring them together in order integrate

the three frameworks. In that way it may be possible to incorporate the different levels of

land use drivers proposed by the three-tiered hierarchical approach, the household

dynamics from the household transformations approach, and finally to better understand

the spatial and temporal interactions by adapting the panarchy approach. Figure 2-4

provides an initial vision.

In Figure 2-4 there are no arrows linking the three-tiered hierarchical framework to

the panarchy framework; there is, however, a direct link. The adaptive cycle may be

applied to all hierarchical levels (household, local, proximate, intermediate and distant) to

make up a nested set of adaptive cycles, a panarchy. Looking at Figure 2-5 one can

imagine having adaptive cycles at each level and for each land use driver. However, in

this chapter the integration will be centered at the household and land use activity levels.

This is mainly for practical and methodological reasons, since information at these faster

levels may be gathered in an easier and more direct way, and examples will be easier to

understand. The logic remains the same for the higher levels and will be applied in the

next chapter when explaining the results from data analysis.










































ii-


si
0o





















sa
00





























i>s
0c,


























Go
cr0~


















0
cr-'

3 0"












Io
~'^
















Bi















-4
c~r
." 0













c -
gl








0 t













o


t 0




-t
I I











0l
ra,


U(U


0

Ho






2
.0

0 a



t 0^
o a





l- 0


S 0
a 0t


Cu,y
~4a


Z u <
?if DO l n


1|1 ~~~~ s^ ~il|l.i
|* g | | |
C)u -U- ,qO jb~C~


0 a

Scr



"0

0 0 0 .0
1 U0



c3 0
^111
F&|
m C (
















Road Paring


Market Polides


-~ -- -
.. ...........
.......'..: =....""
Figure 2-5. Levels of interaction in a nested set of adaptive cycles. In the white area the
bigger cycle represents the household and the smaller cycle represents the
land use activities. Similar cycles exist at the local, regional national and
global levels.

Looking at the household and its activities as adaptive cycles

The most obvious link between the three-tiered hierarchical approach proposed by

Wood (2002) and the household transformations through time framework proposed by

McCracken et al. (2002) is, of course, the household (as shown in Figure 2-4), our center

of analysis. While the first framework classifies the land use drivers that are external to

the household, the second framework classifies the household according to five stages in

its life cycle according to its effect on land use decisions. In that way one can imagine

that proximate, intermediate and distant socioeconomic and biophysical drivers of land

use may have a different effect depending on the household life cycle stage.


Global


I
I
I









This effect is expressed in different land use activities. At the same time, but at a

faster pace, each productive activity has its own transformations through time. In order to

make this idea clear we can look at the household's land use activities as an adaptive

cycle. Let's take the example of a cash crop in the Amazonia (see Figure 2-6). The cycle

may be seen as follows:

4. r: cash crops establishment

5. k: crops are established and producing

6. Q: due to soil depletion or a fall in market prices, crops are not in production any
more

7. a: depending on the cause for the Q phase, and on household transformations the
options may be;

a. change to different cash crops (reorganization within the same cycle),

b. change the farm place (reorganization into a different cycle) or,

c. change its productive activity (reorganization into a different cycle)

d. mix farming with other productive activities (special case, as we will see
later).

In the same way, we can look at the household itself as an adaptive cycle.

Returning to the schematic representation of McCracken et al. (2002) presented in Figure

2-2; it defines five stages of a household life cycle, each one linked to a particular time of

residence in the area, demographic composition and land use practices. We may say that

the cycle is as follows (see Figure 2-6):

1. r: this is represented by stage I, when a young couple with small children arrive to a
frontier area in Amazonia.

2. k: this will be stages II, III, and IV. The family grows from having adolescent
children to have adult children. At the end all household members are available
labor force.









3. Q: this stage in the cycle corresponds to stage V. The original family turns into
multigenerational or second generational households. Small children are present
again.

4. a: depending on the assets available, the second generation young family may:

a. remain in the original farm place (reorganization within the same cycle),

b. move to start their own farm (reorganization into a different cycle) or,

c. change from farming to another productive activity (reorganization into a
different cycle)

d. mix the farm with wage activities.


Stae~iGt~


Stage L


Stage I

land use\
delcisi on /
"'-2. -annual a soil
'Crops productive depletion


Figure 2-6. Looking at the household and its land use activities as adaptive cycles, the
bigger cycle represents the household and its four stages according to the
household transformations approach. The smaller cycle represents one of the
household land use activities (annual crops). The information in the
background represents the other levels from the three-tiered hierarchical
approach









Nonlinear effect of land use drivers

This chapter has provided an integrated framework to be used in the present

research. And it has also provided a strong basis for the formulation of the second

hypothesis:

H2: Land use drivers do not have a linear influence on households, and land use

systems; instead, much depends on the phase they occupy in their life cycle.

Now that we see the household and its activities as adaptive cycles, it should be

clear that in general the household has a slower adaptive cycle than its agricultural land

use activities. Taking this figure further, the complex interplay can be seen as the

interaction of the household adaptive cycle with other adaptive cycles within a nested set

of the four-phase adaptive cycles (see Figure 2-6). Now I will try to engage with the next

slower adaptive cycle, the household's town; in our example this is Ifiapari and Assis

Brazil. At this scale we have direct effects of the proximate and intermediate land use

drivers. Every driver has its own adaptive cycle, so we will have as many adaptive cycles

as drivers of land use determined for this particular town; of interest to us are markets,

credit, and road infrastructure.

Panarchy theory considers potentially multiple connections between phases at one

level and phases at another level, but as explained earlier, "Revolt" and "Remember"

connections become important at times of change in the adaptive cycles (Holling,

Gunderson & Peterson 2002). When the soil cycle enters its omega phase, and

experiences a collapse, that collapse can cascade up to the household cycle by triggering

a crisis, particularly if it is at the K phase (stage IV in household lifecycle) where

resilience is low: that is called revolt. The same thing happens when the household enters

its omega phase and experiences a collapse; that collapse can cascade up to the proximate









driver cycle by triggering a crisis, particularly if it is at the K phase where resilience is

low.

When the household enters its alpha phase, the opportunities and constraints for the

renewal of the cycle are strongly organized by the K phase of the proximate driver cycle,

which is called remember.

Considerations Regarding the Framework

The main objective of this section was the development of a framework to analyze

land use in Amazonia through the integration of three frameworks. The integration allows

us to see the different levels that may affect land use decisions and its interactions:

distant, intermediate and proximate drivers, household assets, and its land use activities

(productive systems). While the adaptive cycle may be applied to all levels, it is applied

in this study at the household and its land use activities level.

A main consideration should be mentioned: the time line. Ideally information taken

in at least two different points in time is desirable. In that way more accurate cycles may

be described. However it was possible to explore the issue with a single time data

collection by asking what future decisions regarding land use will be taken in the face of

anticipated changes that may include road infrastructure, and credits. Questions also

probed about key changes in land use and land cover in the past.

Work needs to be done in regard to some more specific issues. For example, some

of the resources that are considered to be household assets in one framework are

considered as proximate drivers in the three-tiered hierarchical approach (e.g. access to

credit). A second consideration is that we focused on only two of the many different

levels of interaction in the complex processes of land use decisions. The role played by









drivers/assets will vary according to different conditions, becoming more or less

important in different cases.

The nature of land use drivers is also an important fact to be considered; for

example, changes in markets at the national level may have an immediate effect on

household land use decisions, while changes in policy at the national level may not have

a direct effect, usually because environmental laws are poorly implemented. This is an

important consideration since it appears to be a contradiction to the panarchy framework

where cycles at the same level are considered to have the same speed.

Another important consideration is the nature of the land use activity. From the

example that was given in this chapter on agriculture it is obvious that annual crops have

faster cycles than households; however, other agricultural systems may have slower

cycles, like vineyards or other long lived perennial crops. That will also be the case for

rubber tapping and Brazil nuts.














CHAPTER 3
LAND USE AND LAND COVER

Introduction

In methodological terms, the research conducted in Assis Brazil and in Ifiapari was

a natural experiment. It was not an experiment conducted by a researcher; it was

evaluated through research (Bernard 2002).

The two former chapters provided a strong base for establishing the hypotheses to

be tested:

* HI: Access to markets, credit and road infrastructure drove more deforestation in
Assis Brazil than in Ifiapari.

* H2: Land use drivers do not have a linear influence on households, and land use
systems; instead, much depends on the phase they occupy in their life cycle

The present chapter deals with the methodology followed to gather the data in the

field and its analysis. Data gathering included two different steps followed in Ifiapari and

in Assis Brazil, first interviews and then questionnaires. Analysis of both qualitative and

quantitative data was performed in order to test the hypotheses set out in the first two

chapters.

Data analysis included four steps: first, the operationalization of the variables and

the presentation of descriptive statistics; second, the comparison of the means for the data

found in Assis Brazil and in Ifiapari; third, correlation analysis to observe the

relationships among variables; and fourth, multivariate modeling to observe the role of

the group of independent variables in determining land uses and land covers.









Fieldwork Methods in Iiiapari and Assis Brazil

The methodology was designed to be the same in both sites. However, during

fieldwork the different conditions made it difficult to follow the same exact steps in both

towns. There were four weeks of intense fieldwork in each place. The work included

interviews with local authorities and 45 questionnaires with small farmers. In both cases

indigenous peoples and extractive reserves were purposely excluded from in the research

since for these populations the collective ownership of resources adds a dimension to

land use decisions that is not considered in this research.

Fieldwork in Ifiapari

The primary field work in Ifiapari started on June 13th and finished on July 14th,

2004. The district is divided in 5 main sectors: Ifiapari, La Colonia, Nueva Esperanza,

Villa Primavera, and San Isidro de Chilina. The latter two were part of the government

colonization projects. There is also one native community, Belgica, which is not being

considered in this research. Ifiapari and La Colonia are urban centers, while most of the

farm lands are in Primavera, Nueva Esperanza and Chilina along the 50km road to Iberia.

Given the spatial location of Chilina, it is more linked to Iberia than to Ifiapari. Health

care, agriculture, and INRENA offices in Iberia attend Chilinas needs. From 1995 to

1997 the Special Project for Land Titulation (Proyecto Especial de Titulaci6n de Tierras

PETT) worked in the district. For this reason at least 85% of farmlands now have titles.

During the first week in Ifiapari the main activities were the interviews; these

included: the Alcalde Provincial del Tahuamanu, INRENA (National Institute for Natural

resources) manager, SENASA (National Service for Agrarian Health) manager, APEMI

(Ifiapari small loggers association), managers of the three forest concessions of the

district, presidents of the farmers and cattle ranchers association, president of the









mother's club, and the secondary school director. One key person was not possible to

interview, the manager at the Agriculture office. Statistical data and existing maps,

however, were made available for my use. Available information was not good enough to

establish the number of farmers in the district, nor the spatial location of farmlands.

The second week was spent completing the interviews and deciding on the

sampling method to apply the questionnaires. I decided to use the local health post census

data, gathered in December 2002, which provided accurate data on population and

covered most of the district except Chilina. The number of households in each sector was

established (estimated for Chilina) and a proportional number of randomly chosen

households from each sector was assigned to complete 50 households, 25% of the

households that had a farm reported for Ifiapari in 1994 (INEI 1999).

The third and fourth weeks were spent applying questionnaires (Appendix A); 45

questionnaires were carried out but only 36 were valid for this research1, representing

18.3% of the district households that had a farm (INEI 1999). I started in Ifiapari and La

Colonia, the urban areas where most families were concentrated (78%). Appointments

were made with family heads, since most work either at their farms or at the

Municipalidad, the main source of employment, during the day. The fourth week was

spent visiting Chilina (the farthest sector), Primavera, and Nueva Esperanza. See figure

3-1.








1 Questionnaires for households with more than two land holdings, with land holdings that had no
productive activities and those who could not provide accurate information were considered invalid








58








Household's farms visited in the Municipio of Assis Brazil
and the District of Ifiapari


427907 ".. 436503-- 445099 453695















N









Assis Brazil and Irapari
pattern of farms distribution


Legend
I A Bsss Bral Households
S Inaparl Households
^ ----- '~ ^^ ^^^^ h-H^


Sensor Landsat ETM+ 7
RBG Bands 3 2 1
Projection UTM
Spheroid & Datum WGS 84
Zone 19 S


0 5 10


427907 "
20


436503 445099
30


Kilometers


Figure 3-1. Household farms visited in the Municipio of Assis Brazil and in the District
of Ifiapari. While all questionnaires were made on the farm in the case of
Assis Brazil, only some of them were made on the farm in the case of Ifiapari.
The Acre River that makes the international border is highlighted in blue.


453695
40









Fieldwork in Assis Brazil

In Assis Brazil the main work was done from July 15th to August 14th. Four sectors

were identified in the area: Paraguacu, Santa Quiteria, Sao Francisco, and Assis Brazil.

The boundary between the Municipios of Assis Brazil and Brasielia is at km 8 on

the Assis Brazil-Brasieleia road. These means that there are only 8km of paved road

within Assis Brazil and, therefore, most farms in this Municipio are accessed with

secondary roads. Most farmers do not have land titles; only in Santa Quiteria where

INCRA had established Colonization Projects do farmers have land titles.

The first week, I presented my research proposal in a meeting of the Municipal

staff. I had the opportunity to present myself and the research topic, and to get feedback

and recommendations. Two interviews were carried out, one with the person in charge of

the IBAMA office and one with the person in charge of SEATER (Executive Secretary

for Technical Assistance, Rural Extension and Production Warranty).

A collaborative relation was established with SEATER, an organization which

works directly with farmers associations. Since the government only supports

associations, most farmers are part of one. Seven associations were identified in the area:

Bacia, Livramento and Estrela Brilhante in Paraguacu, Sao Felix and Fortaleza in Santa

Quiteria; Novo Progresso and Iracema in Sao Francisco. Farmers that live along the

secondary roads known as Beija Flor, Recife, Do Sete and in the main road near to Assis

Brazil were included. The number of households in each association was established

(estimated for Beija Flor, Recife, Do Sete) and a proportional number of randomly

chosen households from each sector was assigned to complete 50 households, which

represents 23.7% of the households that had a farm reported for the Municipio of Assis

Brazil in 1995/1996 (IBGE 1998).









During the second, third, and fourth week a total of 45 questionnaires (Appendix B)

were carried out and 41 were valid2, representing 19.4% of the households that had a

farm (IBGE 1998);, see Figure 3-1 for spatial distribution. A program was established to

visit each farm with a technician and transportation provided through SEATER. We used

a motorcycle for our transportation through secondary roads 8-15 km long; all of them

depart around km 4 and 7 from the recently paved Pacific highway (BR-317). Most of

them were passable only with a motorcycle.

First we visited Sdo Felix, a Colonization Project of INCRA, then we visited Bacia,

Iracema and Recife; the last one was very difficult to transit, even in August when it is

the dry season. I also took a short trip up the Acre River to visit Novo Progresso. Finally,

we visited Estrela Brilhante and Livramento. Fortaleza was not visited due to

transportation issues. Then we applied questionnaires to farmers in Beija Flor and in the

closer areas like the secondary road known as Sete and in the recently paved BR-317.

The Differences in Methodology and Their Implications

Interviews were an important component in the case of Ifapari, where two weeks

were spent in this activity and more than fifteen interviews were carried out with local

authorities and leaders. In fact, in theory, this will allow for a better contextualization of

the responses obtained in the questionnaires. It will also allow for a comparison on the

authorities' and leaders' perceptions with those of the farmers. In the case of Assis Brazil,

only two interviews were carried out; however, I expect this not to be a major problem,

since those interviewed were key persons: IBAMA and SEATER personnel.



2 Questionnaires for households with more than two land holdings, with land holdings that had no
productive activities and those who could not provide accurate information were considered invalid









In the case of questionnaires, although the number was the same in both places

(45), there were some differences. For Ifiapari 36 questionnaires were valid and were

used in this research, representing 18.3% of the district households that had a farm (INEI

1999) in that area. In Assis Brazil 41 questionnaires were valid, representing 19.4% of

the Municipio households that had a farm (IBGE 1998). This poses questions in terms of

comparisons and statistical analysis. However, although the total number of valid

questionnaires is different, they represent a very similar percentage of the total number of

farmers in each site.

Operationalization of Variables

Five groups of variables have been developed for the present research. The first

group is labeled land use outcomes; the variables in this group are indicators for

household land use activities. The second group was labeled land cover outcomes; these

variables are indicators for deforestation since arrival to the farm. The third group is

labeled background information; these variables are the control variables. The fourth

group, labeled land use drivers, is composed of two sections: markets and credit, and road

infrastructure variables. Finally, the fifth group is labeled household life cycles; these

variables account for household level variance. The variable 'place' that represents

whether a household is located in Ifiapari or in Assis Brazil is also included.

It is important to place the hypotheses and framework for the present research

within the context of Assis Brazil and Ifiapari. Market variables are at the local and

regional level, credit variables are at the national level, and road infrastructure variables

are at the national, regional and local level. Household life cycle variables are at the

household level, the same as land use, land cover, and control variables.









In order to address the central research questions, several data analysis steps will

follow. First, descriptive statistics for both Ifiapari and Assis Brazil will be presented in

order to provide a general idea of the characteristics for the whole area. Second, means

comparison of each variable for Ifiapari and Assis Brazil will be presented in order to

reveal the differences for each town. Third, in order to observe the relation between

dependent variables and dependent (outcome) and independent variables, bivariate

correlations will be presented. Fourth, to gain insights on how each group of variables

interacts and affects the outcome variables, multivariate models between each outcome

variable and each group of independent variables will be presented. Fifth, multivariate

models for each outcome variable will be developed by using the independent variables

found to be significant in the previous steps.

Tables 3-lto 3-4 present descriptive statistics for the dependent and independent

variables used in this study. Data for variables with skewness over 1 were transformed by

converting to the natural logarithm and adding 1 unit to avoid 0 values. This procedure

reduced overall skewness and improved normality for statistical analyses. Mean values

are presented for the raw data, the transformed data and for the antilog of the previously

transformed data. Skewness is presented for both the raw and transformed data. The

sample size of 77 households is the same for all variables.

Table 3-1 presents descriptive statistics for the dependent variables: land use

outcomes and land cover outcomes, for Ifiapari and Assis Brazil in the year 2003. Four

land use outcomes and four land cover outcomes are considered. Land use outcomes were

reported as hectares (ha) in annual crops, perennial crops and pasture. The number of

heads of cattle is also included since some farmers have pasture and no cattle. The area in










annuals includes rice, corn, beans, manioc, other vegetables (including tomatoes, lettuce,

herbs, spinach, and others) and the various combinations. Also included in this category

are hectares of corn mixed with pasture and with bananas. The area in hectares under

perennials includes banana, citrus, pepper, palillo, pijuayo, acai, coffee, and their various

combinations. The combinations usually include fruit trees, palms, and timber trees.

Table 3-1. Descriptive statistics for land use and land cover outcome variables. Ifiapari
and Assis Brazil, 2003.
Std.
Variable Unit Mean Mean* Deviation Skewness N
(1) (2) (3) (4) (5)
Land use outcomes
Annual crops ha 2.86 2.05 0.79 77
Perennial crops ha 0.75 0.92 2.53 77
Pasture ha 16.85 23.25 3.30 77
Heads of cattle 28.78 63.19 5.85 77
Land cover outcomes
Old growth forest ha 55.19 53.68 2.79 77
Secondary forest ha 8.84 11.84 2.82 77
Deforested area (a) ha 20.15 23.96 2.51 77
% deforested of forest (b) % 25.93 25.10 1.11 77
Transformed values In (l+var)
Land use outcomes
Annual crops 1.20 1.22 0.57 -0.40 77
Perennial crops 0.75 0.78 0.30 -1.10 77
Pasture 2.21 3.37 1.25 -0.26 77
Heads of Cattle 2.16 3.20 1.71 0.00 77
Land cover outcomes
Old growth forest 3.67 14.39 0.97 -1.42 77
Secondary forest 1.75 2.12 1.06 0.01 77
Deforested area 2.39 4.02 1.30 -0.44 77
% deforested of forest 2.68 5.36 1.33 -0.77 77
antilogg of mean logs, (a) Hectares deforested since arrival to the farm (initial old growth forest old
growth forest), (b) % forest cleared since arrival (ha deforested x 100 / ha initial old growth forest).

Land cover outcomes were reported in ha of old growth forest, secondary forest,

deforested area, and percentage of cleared forest. The deforested area was calculated by

subtracting the current area of forests from the initial (that which was existing when the









family arrived to the farm) area in hectares of forest. The percentage of cleared forest was

calculated by dividing the total area deforested by the initial area of forest.

The high standard deviation for the number of heads of cattle responds to the

inclusion of one medium size rancher household from Assis Brazil in the dataset. While

most households had less than 100 heads of cattle this household had 500. The data were

included in the analysis since they are representative of a very small group of ranchers in

Assis Brazil. For old growth forest, secondary forest, deforested area and % deforested

land, antilog transformed mean data was found to differ substantially from non-

transformed mean data.

Table 3-2 presents descriptive statistics for household background information.

This category presents seven variables. The ones measured in hectares (ha) are land size,

initial old growth forest and initial secondary forests. Household sources of off-farm

income like regular monthly income (e.g. wage, retirement), daily wage or other irregular

income sources (e.g. taxi driver) are reported through yes/no answers. The region of birth

of the household head indicates whether they were born in the Madre de Dios-Acre-

Pando region or elsewhere. The number of years the head of household has received

formal education are also provided. For the yes/no answers the mean value gives the

percentage of yes answers.

The mean area in old growth and secondary forest was 75 and 6 ha respectively.

Almost half of the families possessed a regular monthly income and almost one third had

a daily wage. More than half of the heads of households were born in the MAP region

and household heads had an average of 5 years of education.










Table 3-2. Descriptive statistics for household background information variables. Ifiapari
and Assis Brazil, 2003
Std.
Variable Unit Mean Mean* Deviation Skewness N
(1) (2) (3) (4) (5)
Background information
Farm size ha 83.51 59.47 2.3 77
Initial old growth forest ha 75.34 60.16 2.21 77
Initial secondary forest ha 5.95 11.90 3.64 77
Regular monthly income 0=no, 1=yes 0.40 0.49 0.41 77
Daily wage 0=no, 1=yes 0.27 0.45 1.04 77
Born in the MAP area 0=no, 1=yes 0.66 0.48 -0.70 77
Education 5.43 4.44 0.65 77
Transformed values In (1+var)
Farm size 4.27 26.31 0.56 0.64 77
Initial old growth forest 4.07 21.48 0.82 -1.55 77
Initial secondary forest 1.06 1.07 1.25 0.76 77
antilogg of mean logs

Table 3-3 provides descriptive statistics for place, markets, credit, and road

infrastructure. Place refers to whether the household is located in Assis Brazil or in

Ifiapari. The six market variables are: distance to the nearest market, whether the

household sells annual crops, perennial crops, small animals or cattle, and an index of

farm product commoditization. This index assigns a value to each one of the different

combinations of products sold. It ranges from 1 to 13; the lowest values are assigned to

households that sell annuals and small animals and the highest to households that sell

cattle and perennials.

Credit refers to the number of times the household had used credit since arriving to

their farm. Road infrastructure groups six variables: whether the household is located in a

main road, a secondary road, a tertiary road, or a walking path; the distance in kilometers

from the main road, and a transportation time index. This index was created to combine

road infrastructure variables, It was calculated by dividing the known distances (obtained

from the interviews and tracked roads) to the markets of either Assis Brazil or Ifiapari by










approximate average travel velocities for primary, secondary, and tertiary roads, and

walking paths (45, 20, 15, and 4 kilometers per hour, respectively) and obtaining a total

travel time by adding each section. For the yes/no answers the mean value gives the

percentage of yes answers.

Table 3-3. Descriptive statistics for place, markets, credit and road infrastructure
variables. Ifiapari and Assis Brazil, 2003
Std.
Variable Unit Mean Mean* Deviation Skewness N
(1) (2) (3) (4) (5)
Place (li apari/Assis Brazil) 0=1, 1=A 1.53 0.502 -0.133 77
Market and credit
Distance from nearest market km 12.41 6.28 0.33 77
Sells annual crops 0=no, l=yes 0.70 0.46 -0.90 77
Sells perennial crops 0=no, 1=yes 0.13 0.34 2.25 77
Sells small animals 0=no, 1=yes 0.42 0.50 0.35 77
Sells Cattle 0=no, l=yes 0.56 0.50 -0.24 77
Farm product
commoditization index 6.61 4.79 -0.12 77
Times credit was received 1.06 1.49 2.68 77
Road Infrastructure
Lives in main road 0=no, l=yes 0.12 0.32 2.43 77
Lives in secondary road 0=no, 1=yes 0.34 0.48 0.70 77
Lives in tertiary road 0=no, l=yes 0.29 0.46 0.97 77
Lives in walking path 0=no, 1=yes 0.26 0.44 1.12 77
Distance from main road km 5.59 4.62 0.33 77
Transportation time hours 0.60 0.38 1.10 77
Transformed values In (l+var)
Times credit was received 0.55 0.64 0.56 0.77 77
Transportation time 0.45 0.57 0.22 0.46 77
* antilog of mean logs

The mean distance to nearest market was short (12.41 km) compared to other areas.

More than two thirds of households sell annual crops and only a small portion of

households sell perennial crops. As for animals, almost half of households sell small

animals while more than half of households sell cattle.

On average all households had received credit once in the past. Spatially a small

percentage of households lives along a primary road while more that one third lives along










a secondary road and almost one third lives along a tertiary path and along a walking

respectively. The mean distance to the main road was 5.59 km while the mean

transportation time, as obtained from the index, was 36 minutes.

Table 3-4 provides descriptive statistics for the household life cycle; this category

groups eight variables. The first is the time the household has lived on the farm (years

on the farm), the age of household head, the number of family members currently living

on the farm, the number of family members that participate in land use activities, the

number of children in the family and the number of adults in the family. Additional

indices are presented for labor hired and labor exchanged. Values range from one to four,

one meaning no labor was exchanged or hired during the last 12 months; two, labor

exchanged for reasons other than forest clearing; three, labor exchanged for forest

clearing; and four, labor exchanged for 2 and 3. For the labor hired index, five represents

labor hired all year round.

Table 3-4. Descriptive statistics for household life cycle variables. Ifiapari and Assis
Brazil, 2003.
Std.
Variable Unit Mean Mean* Deviation Skewness N
(1) (2) (3) (4) (5)
Years on farm 13.90 10.63 1.26 77
Age of household head 44.78 14.29 0.67 77
Family members on lot 4.55 1.88 0.58 77
Family members working
farm 2.48 1.54 1.13 77
Number of children 1.70 1.57 0.87 77
Number of adults 4.18 2.89 1.34 77
Labor hired index 2.58 1.44 0.36 77
Labor exchanged index 2.06 1.23 0.57 77
Transformed values In (1+var)
Years on farm 2.43 4.16 0.80 -0.49 77
Family members working
farm 1.15 1.16 0.47 -0.59 77
Number of adults 1.52 1.68 0.48 0.73 77
antilogg of mean logs









The families had spent an average of 14 years on their plot by the summer, 2003,

when this study was conducted. The mean head of the household was 44 years old.

Though there were an average of 5 family members, only 2 were actively working on the

farm. The number of children was half the number of adults when the definition of adult

was older than 14 years of age. Families, on average, hired labor and exchanged labor for

purposes other than land clearing, although there seems to be a trend in which labor is

hired for land clearing purposes.

Comparing Variable Means for Assis Brazil and Iiiapari

Table 3-5 provides the results of independent t-tests for the means of land use and

land cover outcomes, background information, markets, credit, infrastructure, and

household life cycle variables between Ifiapari and Assis Brazil. Both raw and

transformed data means are compared, but means are considered significantly different

according to the transformed means.

All means for land use outcomes were significantly different. Assis Brazil had a

significantly larger area of annual and perennial crops and pasture, and nearly four times

the heads of cattle than did Ifiapari, in the year 2003. Land cover outcomes means were

not found to be different except the percentage of forest cleared since arrival to the farm,

which is significantly less in Assis Brazil than in Ifiapari, mainly due to the larger average

farms in Assis Brazil. Although not significant, there was an apparent trend of larger

areas of old growth forest in Assis Brazil than Ifiapari.










Table 3-5. T-test of means for land use outcomes, land cover outcomes, background
information, markets and credit, road infrastructure and household life cycle
variables according to location in Ifiapari or Assis Brazil, 2003.
Means
Variables unit Ifiapari Assis Brazil T
(1) (2) (3)
Land use outcomes
Annual crops ha 2.38 3.27 -1.95+
Perennial crops ha 0.65 0.85 -0.95
Pasture ha 11.79 21.29 -1.82+
Heads of cattle count 13.31 42.37 -2.17 (a)*
Land cover outcomes
Old growth forest ha 45.69 63.54 -1.55 (a)
Secondary forest ha 8.57 9.09 -0.19
Deforested area (b) ha 20.46 19.88 0.11
% deforested of forest (c) % 28.07 24.05 0.70
Background information
Farm size ha 68.4 96.78 -2.25(a)*
Initial old growth forest ha 66.15 83.42 -1.33 (a)
Initial secondary forest ha 2.18 9.26 -2.86 (a)**
Regular monthly income 0=no, l=yes 0.31 0.49 -1.64 (a)
Daily wage 0=no, l=yes 0.42 0.15 2.69 (a)**
Born in the MAP area 0=no, l=yes 0.50 0.80 -2.90 (a)**
Education years 7.75 3.39 4.91**
Transformed values
Land use outcomes
Annual crops 1.06 1.33 -2.12*
Perennial crops 0.67 0.82 -2.12*
Pasture 1.89 2.50 -2.2*
Heads of cattle 1.44 2.80 -3.79**
Land cover outcomes
Old growth forest 3.68 3.65 0.11 (a)
Secondary forest 1.68 1.81 -0.57
Deforested area (b) 2.58 2.23 1.19
% deforested of forest (c) 2.96 2.43 1.82 (a)+
Background information
Farm size 4.16 4.36 -1.60(a)
Initial old growth forest 4.11 4.03 0.48 (a)
Initial secondary forest 0.55 1.52 -3.77 (a)**
Regular monthly income 0=no, l=yes 0.31 0.49 -1.64 (a)
Daily wage 0=no, l=yes 0.42 0.15 2.69 (a)**
Born in the MAP area 0=no, l=yes 0.50 0.80 -2.90 (a)**
Education years 7.75 3.39 4.91**
+ p < 0.1, p < 0.05, ** p < 0.01, (a) F test found variance significantly different (p<0.05) for T test equal
variance is not assumed, (b) Hectares deforested since arrival to the farm (initial old growht forest old
growth forest), (c) % forest cleared since arrival (ha deforested x 100 / ha initial old growth forest).










Table 3-5. Continued
Means
Variables unit Ifiapari Assis Brazil T
(1) (2) (3)
Market and Credit
Distance from nearest market km 14.59 10.50 2.89 (a)**
Sells annual crops 0=no, l=yes 0.64 0.76 -1.11 (a)
Sells perennial crops 0=no, l=yes 0.14 0.12 0.22
Sells small animals 0=no, l=yes 0.28 0.54 -2.37 (a)*
Sells Cattle 0=no, l=yes 0.33 0.76 -4.064**
Farm product commoditization index 5.08 7.95 -2.732**
Times credit was received (e) 1.42 0.76 1.90 (a)+
Road Infrastructure
Lives in main road 0=no, l=yes 0.22 0.02 2.66 (a)*
Lives in secondary road 0=no, l=yes 0.22 0.44 -2.06 (a)*
Lives in tertiary road 0=no, l=yes 0.00 0.54 -6.81 (a)**
Lives in walking path 0=no, 1=yes 0.56 0.00 6.61 (a)**
Distance from main road km 3.24 7.65 -4.74**
Transportation time hours 0.76 0.46 3.56 (a)**
Household life cycle
Years on farm years 14.08 13.73 0.14
Age of household head years 44.72 44.83 -0.03
Family members on lot count 4.53 4.56 -0.08
Family members working farm 2.17 2.76 -1.70+
Number of children count 1.56 1.83 -0.76
Number of adults count 3.86 4.46 -0.91
Labor hired index 2.00 3.10 -3.60**
Labor exchanged index 2.14 2.00 0.49
Transformed values In (1+var)
Times credit was received (e) 0.66 0.45 1.688+
Transportation time 0.53 0.37 3.24 (a)**
Years on farm 2.38 2.47 -0.48
Family members working farm 1.03 1.25 -2.16*
Number of adults count 1.46 1.57 -0.96
+ p < 0.1, p < 0.05, ** p < 0.01, (a) F test found variance significantly different (p<0.05) for T test equal
variance is not assumed, (e) Since arrival to the property.

Variables that represent the farmer's background had means that were not

significantly different except for initial area in secondary forest and percentage of family

chiefs that were born in the MAP area. Both variables presented higher values for Assis

Brazil. The percentage of households receiving a daily wage, and years of education,


presented significantly higher values for Ifiapari.









Market variables showed that surveyed households in Ifiapari were located

significantly further from the nearest market than those in Assis Brazil. A significantly

larger number of households in Assis Brazil sold small animals and cattle than

households in Ifiapari. This was reflected in the farm product commoditization index

which was significantly greater for Assis Brazil than for Ifiapari

As for credit, interestingly, households in Ifiapari had received significantly more

credit more than households in Assis Brazil since their arrival to the property. Credits

were provided by the Agrarian Bank in Peru from the late 1950s until 1991. In Assis

Brazil credit access is more recent; most households did not have credit available until

the late 1990s.

The road infrastructure variables show that farmers in Ifiapari are significantly

more likely to live along a main road than those in Assis Brazil, but are significantly less

likely to live along a secondary road. The majority of households in Assis Brazil (54%)

lived along tertiary roads while no households in Ifiapari were found on tertiary roads,

and instead the majority of households in Ifiapari (56%) were found along walking paths,

where none were found for Assis Brazil (see Figure 3-1). Walking paths and tertiary

roads are in different categories because walking paths were found in Ifiapari only, they

are not passable by motor vehicles and they start in the border of the main road.

Households in Assis Brazil were found more than twice as far (7.65 km), on average,

than those in Ifiapari (3.24 km). Transportation time to the nearest market (either in Assis

Brazil, Ifiapari or Iberia) is greater (31.8 minutes) for Ifiapari households than for those in

Assis Brazil (22.2 minutes). This is explained because the index assumes motor vehicle









transportation for roads, and walking speed for walking paths, which makes

transportation time greater for Ifiapari households

Of the eight household life cycle variables only two were found to be significant.

The number of family members working on the farms and the days of labor hired were

significantly higher in Assis Brazil than in Ifiapari.

Relating the results of the mean analysis to the first hypothesis for this research we

may say that land use outcomes are larger in area (annual crops, perennial crops, pasture)

and in number (heads of cattle) in Assis Brazil than in Ifiapari as was expected. This may

be explained by significantly different market and road infrastructure variables. However

the second part of the hypothesis is not as expected: differences in land cover outcomes,

in particular in area deforested at the household level, are not significant. This can be

explained by significantly different background variables such as initial area of

secondary forest, that is four times higher in Assis Brazil. The higher percentage of old

growth forest cleared in Ifiapari is in part explained by the smaller farm sizes in Ifiapari.

As for the second hypothesis we may say that household life cycle variables are not

significantly different from Ifiapari to Assis Brazil except for the family members

working the farm and labor hired; therefore differences in land use outcomes may be due

to the effect of the variables external to the household .

However it is necessary to explore the way in which the dependent and independent

variables interact to be able to draw more meaningful statements with respect to both

hypotheses. The following analysis steps will treat all the questionnaires as a single

sample and not as separate ones since the sample number is very low. Instead, the










variable "place" will be included; this will allow us to use the whole sample and to

observe the differences between the two sites.

Correlations Between Independent and Dependent Variables

Bivariate correlations are the third step in this analysis. Significant positive or

negative correlations provide insight into relationships between individual variables

and/or outcomes. Tables 3-6 to 3-14 present bivariate correlation coefficients (Pearson

correlation) between each land use and land cover outcome, as well as against the entire

suite of measured variables.

Table 3-6 presents the correlations among the land use outcomes and land cover

outcomes. These outcomes were natural log transformed prior to analysis. All significant

correlations were found to be positive. Households having larger areas in perennials also

had larger areas in annuals. For obvious reasons, farms with more heads of cattle also had

a greater area in pasture.

Table 3-6. Correlations between land use outcomes and land cover outcomes variables.
Ifiapari and Assis Brazil, 2003.
Variable Annuals Perennials Pasture Cattle
Variable
(1) (2) (3) (4)
Land use outcomes
Ln ha annual crops 1
Ln ha perennial crops 0.983** 1
Ln ha pasture 0.145 0.061 1
Ln number heads of cattle 0.128 0.052 0.750** 1
Land cover outcomes
Ln ha old growth forest 0.200+ 0.222+ 0.030 -0.049
Ln ha secondary forests 0.296** 0.317** 0.004 -0.095
Ln ha deforested (a) 0.264* 0.184 0.490** 0.304**
Ln % deforested of forest (c) 0.196 0.127 0.343** 0.216+
+ p < 0.1, p < 0.05, ** p < 0.01, (a) ha deforested since arrival to the farm (initial old growth forest old
growth forest), (b) % forest cleared since arrival (ha deforested x 100 / ha initial old growth forest), N=77.

The correlations with the land cover outcomes provided less obvious results. The

area in old growth, secondary growth, and deforested were all positively correlated with









the area in annuals. These same land cover outcomes, with the exception of area

deforested, were also significantly correlated with area in perennials. Increasing area of

pasture was found with increasing deforested area and % forest cleared since arrival.

Greater numbers of cattle also were related to increasing levels of deforestation and %

forest cleared.

Table 3-7 shows correlations between land use outcomes and background

information variables. All significant correlations were found to be positive. Larger initial

farm sizes (ha), and head of household born within the MAP region, were both related to

larger areas of annuals and perennials. Increasing area of pasture was related with

increasing areas of: initial farm size and initial area of old growth forest, as well as the

household having a regular monthly income. Initial farm size, initial area in old growth,

initial area in secondary growth, having a regular monthly income and having a daily

wage were all higher as the head of cattle owned by the household increased.

Table 3-7. Correlations between land use outcomes and background information. Ifiapari
and Assis Brazil, 2003
Variable Annuals Perennials Pasture Cattle
(1) (2) (3) (4)
Background information
Ln ha of farm size 0.228* 0.194+ 0.503** 0.335**
Ln initial ha old growth forest 0.165 0.127 0.338** 0.192+
Ln initial ha secondary forest 0.027 0.057 0.097 0.196+
Regular monthly income (0=no, l=yes) -0.141 -0.157 0.229* 0.229*
Daily wage (0=no, l=yes) -0.130 -0.122 -0.150 -0.192+
Born in the MAP (area 0=no, l=yes) 0.243* 0.254* -0.064 -0.018
Years of education 0.066 0.033 -0.027 -0.160
+ p < 0.1, p < 0.05, **p< 0.01, N=77

Table 3-8 shows correlations between land use outcomes and place, market and

credit, and road infrastructure variables. Living in Assis Brazil was correlated with

increases in all the land use outcomes. In the market category selling perennials was not

correlated to any land use outcome. Selling annuals and small animals was positively










correlated with area under annuals and under perennials. As expected, selling cattle and

the commoditization index were positively correlated with pasture and heads of cattle.

Table 3-8. Correlations between land use outcomes market and credit and road
infrastructure. Ifiapari and Assis Brazil, 2003
Variable Annuals Perennials Pasture Cattle
(1) (2) (3) (4)
Place liapari/Assis Brazil (0=I, 1 A) 0.238* 0.238* 0.246* 0.402**
Market and credit
Distance in km from nearest market 0.129 0.185 -0.209+ -0.315**
Sells annual crops (0=no, l=yes) 0.387** 0.423** -0.026 -0.117
Sells perennial crops (0=no, l=yes) 0.109 0.067 0.009 0.067
Sells small animals (0=no, l=yes) 0.361** 0.352** 0.234* 0.190+
Sells cattle (0=no, 1=yes) 0.232* 0.178 0.592** 0.732**
Farm product commoditization (index) 0.024 -0.035 0.569** 0.715**
Times credit was received 0.103 0.083 0.083 -0.126
Road infrastructure
Lives in main road (0=no, l=yes) -0.089 -0.137 -0.179 -0.276*
Lives in secondary road (0=no, l=yes) 0.384** 0.368** 0.359** 0.404**
Lives in tertiary road (0=no, l=yes) 0.017 0.055 -0.029 -0.022
Lives in walking path (0=no, l=yes) -0.367** -0.353** -0.227* -0.211+
Distance in km from main road 0.161 0.217+ 0.050 0.213+
Ln transportation time (index) -0.016 0.049 -0.251* -0.213+
+ p < 0.1, p < 0.05, ** p < 0.01, N=77

The credit variable was not correlated with any land use outcome. Households

living on secondary roads had larger areas of all land use outcomes and owned more head

of cattle; however, those living on walking paths had smaller areas of annuals and pasture

and owned less head of cattle. Cattle herd size decreased for households living on main

roads, tertiary roads, or walking paths, but increased for those families living on

secondary roads.

Table 3-9 presents the correlations between land use outcomes and household

cycles. The area in annuals or perennials increased for households ranking higher on the

"labor exchanged" index. Households that had more family members working on the

farm had greater areas planted with perennial crops. The area of pasture and the number

of heads of cattle increased with the time on the farm, the number of children, the number










of adults, but, declined for older farms, with more adults in the family, and for the ones

that ranked higher in the labor exchange index.

Table 3-9. Correlations between land use outcomes and household cycles. Ifiapari and
Assis Brazil, 2003
Variable Annuals Perennials Pasture Cattle
Variable
(1) (2) (3) (4)
Household life cycle
Ln years on farm 0.135 0.062 0.564** 0.391**
Age of household head -0.133 -0.172 0.218+ 0.263*
Family members on lot -0.091 -0.048 -0.262* -0.184
Ln family members working the farm 0.208 0.264* -0.041 -0.036
Number of children -0.054 -0.004 -0.309** -0.297**
Ln number of adults -0.015 -0.050 0.352** 0.325**
Labor hired index 0.146 0.106 0.296** 0.365**
Labor exchanged index 0.360** 0.349** -0.168 -0.182
+ p < 0.1, p< 0.05, ** p< 0.01, N=77

Table 3-10 shows the correlations between land use outcomes and land cover

outcomes. The only significant, and fairly obvious, relationship in this table is between

increasing hectares of forest deforested since arrival accompanying higher percentages of

initial forest area being deforested.

Table 3-10. Correlations between land use outcomes and land cover outcomes. Ifiapari
and Assis Brazil, 2003
Old growth Secondary Ha %
Variable forest forest deforested deforested

(1) (2) (3) (5)
Land cover outcomes
Ln ha old growth forest 1
Ln ha secondary forests -0.049 1
Ln ha deforested (a) 0.186 -0.080 1
Ln % deforested of forest (b) -0.025 -0.080 0.929** 1
+ p < 0.1, p < 0.05, ** p < 0.01, (a) ha deforested since arrival to the farm (initial old growth forest old
growth forest), (b) % forest cleared since arrival (ha deforested x 100 / ha initial old growth forest), N=77

In table 3-11 the correlations between land cover outcomes and background

information are shown. Larger farms and farms with more initial area under old growth

forest were more likely to have larger areas of old growth forest and at the same time

more deforested area. The area and percentage of forest area cleared since the










household's arrival to the parcel was smaller on farms with larger initial areas of

secondary forest.

Table 3-11. Correlations between land cover outcomes and background information.
Ifiapari and Assis Brazil, 2003
Old growth Secondary Ha % of forest
Variable forest forest deforested cleared
(1) (2) (3) (5)
Background information
Ln ha farm size 0.746** 0.107 0.459** 0.147
Ln initial ha old growth forest 0.799** -0.132 0.636** 0.437**
Ln initial ha secondary forest -0.105 0.299** -0.526** -0.564**
Regular monthly income (0=no, l=yes) 0.046 0.046 -0.010 -0.034
Daily wage (0=no, l=yes) -0.074 -0.191+ -0.089 -0.049
Born in the MAP area (0=no, l=yes) 0.025 0.043 -0.035 -0.049
Years of education 0.060 0.012 0.152 0.131
+ p < 0.1, p< 0.05, ** p< 0.01, N=77

Table 3-12 shows correlations between land cover outcomes, markets and credit,

and road infrastructure. Farms located further from, and having longer transportation

times, to the nearest markets, or selling annual crops, had a larger area in old growth

forest. Those households that were selling annual crops and that did not live on walking

paths had increasingly larger areas in secondary forest. More hectares had been

deforested since arrival by households that sold cattle but fewer had been deforested by

those who live on tertiary roads. The percentage of initial forest deforested was higher for

households that sold cattle, had received more credit since arrival, or that ranked highly

on the "farm product commoditization" index. A smaller percentage was deforested,

however, by families living on tertiary roads.

Table 3-13 shows the correlations between land cover outcomes and household

cycles. A greater area in secondary forest was found for older farms, that had more

family members working on them, or that ranked highly on the "labor exchanged" index.

Hectares deforested since arrival also was higher for older farms, with older household










heads, that had a larger numbers of adults, or that ranked higher on the "labor

exchanged" index. A greater percentage of the initial forest had been deforested on older

farms that had more adults in their households.

Table 3-12. Correlations between land cover outcomes, markets and credit and road
infrastructure. Ifiapari and Assis Brazil, 2003
%
Old growth Secondary Ha
deforested
Variable forest forest deforested d
(a)
(1) (2) (3) (5)
Place liiapari/Assis Brazil (0=1, 1 A) -0.013 0.065 -0.136 -0.201
Market and Credit
Distance in km from nearest market 0.221+ -0.026 0.077 0.108
Sells annual crops (0=no, l=yes) 0.390** 0.262* 0.108 -0.002
Sells perennial crops (0=no, l=yes) 0.034 -0.13 0.05 0.076
Sells small animals (0=no, l=yes) 0.076 0.070 0.133 0.144
Sells Cattle (0=no, l=yes) -0.157 -0.082 0.238* 0.257*
Farm product commoditization (index) -0.158 -0.100 0.214+ 0.214+
Times credit was received 0.135 0.146 0.329** 0.244*
Road infrastructure
Lives in main road (0=no, l=yes) 0.035 0.119 0.188 0.189
Lives in secondary road (0=no, l=yes) -0.011 0.141 0.181 0.133
Lives in tertiary road (0=no, l=yes) 0.026 -0.003 -0.192+ -0.233*
Lives in walking path (0=no, l=yes) -0.041 -0.236* -0.136 -0.042
Distance in km from main road 0.176 -0.073 -0.109 -0.145
Ln transportation time (index) 0.260* -0.074 -0.027 0.005
+ p < 0.1, p < 0.05, ** p < 0.01, (a) % forest cleared since arrival (ha deforested x 100 / ha initial old
growth forest), N=77

Table 3-13. Correlations between land cover outcomes and household cycles. Ifiapari and
Assis Brazil, 2003

Old growth Secondary Ha
deforested
Variable forest forest deforested d
(a)
(1) (2) (3) (5)
Household life cycle
Ln years on farm 0.013 0.280* 0.562** 0.460**
Age of household head -0.148 0.048 0.205+ 0.165
Family members on lot -0.076 -0.008 -0.05 0.041
Ln family members working the farm -0.099 0.269* 0.011 0.036
Number of children 0.004 -0.107 -0.143 -0.079
Ln number of adults -0.129 0.086 0.256* 0.215+
Labor hired (index) 0.116 0.101 -0.026 -0.107
Labor exchanged (index) 0.180 0.190+ 0.203+ 0.165
+ p < 0.1, p < 0.05, ** p < 0.01, (a) % forest cleared since arrival (ha deforested x 100 / ha initial old
growth forest), N=77










Results from this section mainly provide a general idea of the existing relationships

between dependent and independent variables. Land use outcomes were more likely to be

significantly correlated to the market, credit and road infrastructure while valuables land

cover outcomes were more likely to be significantly correlated with background

information variables. For a better understanding of these relationships, the next section

presents multivariate models for land use and land cover outcomes.

Multivariate Models

This section constitutes the fourth step in the data analysis. Here Ordinary Least

Square (OLS) regression models were run to gain insights into the way variables interact,

since correlations provide limited information in this aspect. A total of five models were

run for each land use and land cover outcome. Model 1 presents the results (beta

coefficient, significance, r square and F) after regressing a given outcome (e.g. area in

annual crops) against the background information (BI) variables. Model 2 presents

market and credit (MC) variables, Model 3 presents road infrastructure (RI) variables,

and model 4 presents household life cycle (HLC) variables.

The fifth model (termed the "comprehensive" model) integrates the variables that

were found to be significant in Models 1 to 4, as well as variables that were significantly

correlated to the outcome (from Tables 3-6 to 3-13).

Land Use Models

For area in annual crops (Table 3-14) the BI model shows that those who had a

regular monthly income had less area in annual crops. Conversely, those born in the MAP

area, having more years of education, and living in Assis Brazil rather than in Ifiapari had

a larger area of annual crops.










Table 3-14. Models of farm area in annual crops outcome regressed on background
information, market & credit road infrastructure, and household life cycle in
Assis Brazil and Ifiapari, 2003.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
BI MC RI HLC Comp.
Background information
Constant -0.288
Ln ha of farm size 0.154
Ln initial ha old growth forest 0.02
Ln initial ha secondary forest -0.01
Regular monthly income (0=no, l=yes) -0.327*
Daily wage (0=no, l=yes) -0.170
Born in the MAP (area 0=no, l=yes) 0.246+
Years of education 0.04* 0.042**
Place 0.363*
Market & credit
Constant 0.626*
Distance in km from nearest market 0.01
Sells annual crops (0=no, l=yes) 0.328* 0.335**
Sells perennial crops (0=no, l=yes) 0.02
Sells small animals (0=no, l=yes) 0.189
Sells cattle (0=no, 1=yes) 0.969** 1.109**
Farm product commoditization (index) -0.09** -0.088**
Times credit was received 0.05
Place 0.08
Road infrastructure
Constant 0.897+
Lives in main road (0=no, l=yes) -0.183
Lives in secondary road (0=no, l=yes) 0.312+
Lives in tertiary road (0=no, l=yes)
Lives in walking path (0=no, l=yes) -0.595* -0.463**
Distance in km from main road -0.02
Ln transportation time (index) 0.867* 0.648*
Place 0.06
Household life cycle
Constant 0.463
Ln years on farm 0.103
Age of household head -0.01+
Family members on lot -0.05
Ln family members working the farm 0.286+
Number of children -0.01
Ln number of adults 0.125
Labor hired index 0.08
Labor exchanged index 0.184** 0.101*
Place 0.134
R2 0.241 0.380 0.260 0.322 0.538
F 2.705* 5.220** 4.091** 3.573** 11.483**
+ p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized
slopes, N=77










The MC model shows that those who sold annuals or cattle and who ranked lower

in the "farm product commoditization" index had more land in annual crops. In Model 3

(RI variables) farms on secondary roads or with higher transportation times (as calculated

in the "transportation time" index) had a larger area, and farms located on walking path

had a smaller area, of annual crops.

The HLC model shows that households with a younger "head of household", with

more family members working in the farm, and/or those who exchanged more labor kept

more area in annual crops. By observing the models R2 values we may say that market

and credit variables (R2 = 0.380) are the ones that explain the most variation, followed by

household life cycle variables (R2 = 0.322), road infrastructure variables (R2 = 0.260),

and finally the background information variables (R2 = 0.241).

The area in perennial crops (Table 3-15) for the BI model (incorporating

"background information" variables, as previously explained) shows similarity to the

case of area in annual crops. The MC model (incorporating market and credit variables)

shows that those who were farther from the nearest market, who sold annual crops, or

cattle, and those who ranked lower in the farm product commoditization index had more

area in perennial crops.

The RI model (incorporating road infrastructure variables) shows that households

located on walking paths and households who had lower transportation times to the

nearest market maintained a smaller area in perennial crops.










Table 3-15. Models of farm area in perennial crops outcome regressed on background
information, market & credit road infrastructure, and household life cycle in
Assis Brazil and Ifiapari, 2003
Variables Model 1 Model 2 Model 3 Model 4 Model 5
BI MC RI HLC Comp.
Background information
Constant 0.055
Ln ha of farm size 0.066
Ln initial ha old growth forest 0.013
Ln initial ha secondary forest 0.001
Regular monthly income (0=no, l=yes) -0.176*
Daily wage (0=no, l=yes) -0.082
Born in the MAP (area 0=no, l=yes) 0.137+
Years of education 0.018** 0.019**
Place 0.175**
Market & credit
Constant 0.413+
Distance in km from nearest market 0.009+
Sells annual crops (0=no, l=yes) 0.184** 0.182**
Sells perennial crops (0=no, l=yes) -0.023
Sells small animals (0=no, l=yes) 0.085
Sells cattle (0=no, 1=yes) 0.518** 0.597**
Farm product commoditization (index) -0.051** -0.052**
Times credit was received 0.012
Place 0.067
Road infrastructure
Constant 0.621*
Lives in main road (0=no, l=yes) -0.162
Lives in secondary road (0=no, l=yes) 0.138
Lives in tertiary road (0=no, l=yes)
Lives in walking path (0=no, l=yes) -0.362* -0.261**
Distance in km from main road -0.089
Ln transportation time (index) 0.532* 0.448**
Place 0.005
Household life cycle
Constant 0.404*
Ln years on farm 0.028
Age of household head -0.007+
Family members on lot -0.027
Ln family members working the farm 0.183*
Number of children -0.003
Ln number of adults 0.069
Labor hired index 0.034
Labor exchanged index 0.091** 0.044*
Place 0.069
R2 0.222 0.414 0.284 0.312 0.549
F 2.432* 6.007** 4.619** 3.370** 11.983**
+ p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized
slopes, N=77










The variables influencing the HLC model (incorporating household life cycle

variables) are similar to those that influenced the area in annual crops. In the case of

perennials, as in the case of annuals, market and credit variables (R2 = 0.414) were the

ones that had the most explanatory power, followed by household life cycle variables (R2

= 0.312), road infrastructure variables (R2 = 0.284) and background information variables

(R2 = 0.222).

The variables explaining area in pasture (Table 3-16) from the BI model showed

that larger sized farms had a larger area in pasture. The MC model shows that farms

closer to the nearest market, that sold small animals, who ranked higher in the farm

product commoditization index and those who received credit a greater number of times

had a greater area of pasture.

The RI model shows that farms on secondary roads had more area in pasture.

Finally, the HLC model shows that households who had greater residence times on their

farms and those whose families were of younger age had more area in pasture.

Conversely, those households with more adults and those who exchanged less labor had

more area in pasture. According to the R2 coefficients household life cycle variables (R2

= 0.508) are the ones that perform better, followed by market and credit variables (R2

0.446), background variables (R2 = 0.327), and road infrastructure variables (R2 = 0.179).

For heads of cattle owned (Table 3-17) the BI model shows that farms in Assis

Brazil had more cattle than those in Ifiapari. The MC related variables of the second

model shows that farms located closer to the nearest market, and farms who ranked

higher in the farm product commoditization index, owned more cattle. The RI model

shows that farms on secondary roads and farms in Assis Brazil had more cattle.










Table 3-16. Models of farm area in pasture outcome regressed on background
information, market & credit, road infrastructure, and household life cycle in
Assis Brazil and Ifiapari, 2003.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
BI MC RI HLC Comp.
Background information
Constant -3.063**
Ln ha of farm size 1.382** 0.632**
Ln initial ha old growth forest -0.294
Ln initial ha secondary forest -0.038
Regular monthly income (0=no, l=yes) 0.333
Daily wage (0=no, 1=yes) -0.004
Born in the MAP (area 0=no, l=yes) -0.433
Years of education 0.010
Place 0.469
Market & credit
Constant 1.571*
Distance in km from nearest market -0.045*
Sells annual crops (0=no, 1=yes) 0.020
Sells perennial crops (0=no, 1=yes) -0.179
Sells small animals (0=no, l=yes) 0.464+
Sells cattle (0=no, 1=yes) 0.634
Farm product commoditization (index) 0.093+ 0.115**
Times credit was received 0.454*
Place -0.134
Road infrastructure
Constant 2.771+
Lives in main road (0=no, 1=yes) -0.484
Lives in secondary road (O=no, l=yes) 0.683+
Lives in tertiary road (0=no, 1=yes)
Lives in walking path (0=no, 1=yes) -0.020
Distance in km from main road -0.008
Ln transportation time (index) -0.844
Place 0.259
Household life cycle
Constant 0.377
Ln years on farm 0.838** 0.538**
Age of household head -0.025*
Family members on lot -0.100
Ln family members working the farm 0.054
Number of children -0.094
Ln number of adults 0.702+
Labor hired index 0.093 0.136*
Labor exchanged index -0.181+
Place 0.356
R2 0.327 0.446 0.179 0.508 0.632
F 4.139** 6.843** 2.551* 7.681** 30.948**
+ p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized
slopes, N=77










Table 3-17. Models of head of cattle outcome regressed on background information,
market & credit, road infrastructure, and household life cycle in Assis Brazil
and Ifiapari, 2003
Variables Model 1 Model 2 Model 3 Model 4 Model 5
BI MC RI HLC Comp.
Background information
Constant -2.944+
Ln ha of farm size 0.794
Ln initial ha old growth forest 0.001
Ln initial ha secondary forest 0.075
Regular monthly income (0=no, l=yes) 0.444
Daily wage (0=no, 1=yes) -0.021
Born in the MAP (area 0=no, l=yes) -0.593
Years of education -0.002
Place 1.219*
Market & credit
Constant 1.088
Distance in km from nearest market -0.071** -0.048*
Sells annual crops (0=no, 1=yes) -0.054
Sells perennial crops (0=no, 1=yes) 0.332
Sells small animals (0=no, l=yes) 0.219
Sells cattle (0=no, 1=yes) 0.924
Farm product commoditization (index) 0.157* 0.200**
Times credit was received 0.014
Place 0.194
Road infrastructure
Constant -1.010
Lives in main road (0=no, 1=yes) 0.554
Lives in secondary road (O=no, l=yes) 1.582** 0.693**
Lives in tertiary road (0=no, 1=yes)
Lives in walking path (0=no, 1=yes) 1.589
Distance in km from main road 0.057
Ln transportation time (index) -1.237
Place 1.562*
Household life cycle
Constant -1.084
Ln years on farm 0.642* 0.305*
Age of household head 0.000
Family members on lot 0.033
Ln family members working the farm -0.083
Number of children -0.266+ -0.236**
Ln number of adults 0.234
Labor hired index 0.173
Labor exchanged index -0.228+
Place 1.151* 0.482+
R2 0.277 0.635 0.351 0.425 0.712
F 3.251** 14.776** 5.358** 5.492** 28.800**
+ p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized
slopes, N=77










Model 4, incorporating HLC variables, shows that households who had greater

tenure on the farm, lower numbers of children, who exchanged less labor, and farms in

Assis Brazil had more cattle. Observing the R2 values we may say that market and credit

variables are the ones that performed best (R2 = 0.635), followed by the household life

cycle variables (R2 = 0.425), the road infrastructure variables (R2 = 0.351), and

explaining the least of the variation, the background information variables (R2 = 0.277).

Land Cover Models

The BI model variables that explained best why larger areas in old growth forest

(Table 3-18) existed were larger area of secondary and old growth forests in farms on

arrival. These farms had more area in old growth forest in the year 2003. The MC model

shows that those who sold annual crops had more area in old growth forest. The third

model (with RI variables) shows that farms with greater transportation times to the

nearest market or farms who exchanged more labor (as shown in the HLC model) had

more area in old growth forest. The model R2 values show that, background variables (R2

= 0.695) explain most of the variation, followed by market & credit variables (R2

0.197), road infrastructure variables (R2 = 0.116) and, household life cycle variables (R2

=0.101).

For the area of the parcel in secondary forest (Table 3-19) the BI model shows that

larger farms, those with smaller initial areas in old growth forest, and households without

daily wages had a greater area of secondary forests. The second model (with MC

variables) shows that households who sold annual crops had more area in secondary

forest. The one significant RI variable in the third model shows that those who lived on

walking paths had less area in secondary forest.










Table 3-18. Models of area of old growth forest outcome regressed on background
information, market and credit, road infrastructure, and household life cycle in
Assis Brazil and Ifiapari, 2003.
Variables Model 1 Model 2 Model 3 Model 4 Model 5
BI MC RI HLC Comp.
Background information
Constant -1.136+
Ln ha of farm size 0.241
Ln initial ha old growth forest 0.918** 0.977**
Ln initial ha secondary forest 0.168* 0.167**
Regular monthly income (0=no, l=yes) -0.083
Daily wage (0=no, 1=yes) 0.152
Born in the MAP (area 0=no, l=yes) 0.104
Years of education 0.000
Place -0.136
Market & credit
Constant 2.644**
Distance in km from nearest market 0.027
Sells annual crops (0=no, 1=yes) 0.069** 0.240+
Sells perennial crops (0=no, 1=yes) 0.122
Sells small animals (0=no, l=yes) -0.055
Sells cattle (0=no, 1=yes) -0.360
Farm product commoditization (index) 0.011
Times credit was received 0.135
Place 0.170
Road infrastructure
Constant 2.891**
Lives in main road (0=no, 1=yes) 0.339
Lives in secondary road (0=no, l=yes) 0.090
Lives in tertiary road (0=no, 1=yes)
Lives in walking path (0=no, 1=yes) -0.363
Distance in km from main road 0.013
Ln transportation time (index) 1.572* 0.707*
Place 0.017
Household life cycle
Constant 3.692**
Ln years on farm 0.110
Age of household head -0.011
Family members on lot -0.023
Ln family members working the farm -0.179
Number of children 0.030
Ln number of adults -0.058
Labor hired index 0.104
Labor exchanged index 0.187+
Place -0.083
R2 0.695 0.197 0.116 0.101 0.724
F 19.333** 2.086* 1.525 0.836 47.110**
+ p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized
slopes.




Full Text

PAGE 1

LAND USE AND LAND COVER IN IAP ARI, PERU, AND ASSIS BRAZIL, BRAZIL, SOUTHWEST AMAZONIA By ANGELICA MARIA ALMEYDA ZAMBRANO 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 2004

PAGE 2

Copyright 2004 by Anglica Mara Almeyda Zambrano

PAGE 3

To the people of the MAP region.

PAGE 4

ACKNOWLEDGMENTS I would like to thank the people of Assis Brazil and Iapari, especially the 90 families that allowed me to interview them in order to gather the data for this research. I also would like to thank Manuel Batista de Araujo, Prefeito of the Municpio of Assis Brazil, and Mario Enrique Montes Len, Alcalde of the Province of Iapari. I would like to thank the different funding sources that allowed me to attend the Center for Latin American Studies at the University of Florida and to those who funded the field research: the Organization of American States, the Tropical Conservation and Development program at University of Florida, the Tinker Foundation and the Setor de Estudos do Uso da Terra e Mudanas Globais at the Federal University of Acre. I would like to thank the members of my committee, for their help and support throughout the research process and especially for their comments on my various final draftsMichael Binford, Stephen Perz, and especially to Marianne Schmink, my chair. I would like to thank I. Foster Brown for the opportunity to visit Acre for the first time in 2001, and for his support and comments on my research, as well as for all his logistical support during my field work. I would like to thank my field assistants, Mercedes in Iapari and Marco in Assis Brazil, for all their hard work that made my field research possible. I also would like to thank a number of institutional offices that helped me in the field: INRENA, INADE, MADERIJA and MADERACRE, the health post in Iapari, SEATER, IBAMA, and SETEM. iv

PAGE 5

I would like to thank my fianc, Eben Broadbent for all his support during the writing process. v

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................iv LIST OF TABLES ...........................................................................................................viii LIST OF FIGURES ...........................................................................................................xi ABSTRACT ......................................................................................................................xii 1 INTRODUCTION........................................................................................................1 Land Use and Land Cover Change ...............................................................................1 Proximate Sources and Driving Forces of Change .......................................................3 Assis Brazil and Iapari ................................................................................................4 Site Description .....................................................................................................5 Assis Brazil and Acre in Context ..........................................................................6 Population composition ..................................................................................8 Natural protected areas .................................................................................10 Transportation and highways .......................................................................11 Iapari and Madre de Dios in Context ................................................................12 Natural protected areas and indigenous communities ..................................14 Transportations and highways ......................................................................14 The Case for Comparing Iapari and Assis Brazil .....................................................15 The Role of Roads ...............................................................................................15 The Importance of Markets .................................................................................16 Existence and Inexistence of Credit ....................................................................19 Roads, Markets and Credits as Land Use Drivers ...............................................19 2 THEORETICAL APPROACH..................................................................................21 Introduction .................................................................................................................21 Land Use Drivers and Deforestation ..........................................................................21 Markets and Credit ..............................................................................................21 Roads Driving Land Use .....................................................................................23 Approaches to Explain Deforestation ..................................................................26 Integrated Theories .....................................................................................................30 Political Ecology .................................................................................................32 Household Demography ......................................................................................36 Panarchy Theory ..................................................................................................37 vi

PAGE 7

Integrated Frameworks ...............................................................................................39 The Three-Tired Hierarchical Approach .............................................................39 Household Transformations Land Use And Environmental Change ..................40 The Adaptive Cycle .............................................................................................43 Potential, connectedness and resilience .......................................................43 Hierarchies and panarchies ...........................................................................45 Framework Integration ........................................................................................47 Looking at the household and its activities as adaptive cycles ....................49 Nonlinear effect of land use drivers .............................................................52 Considerations Regarding the Framework ..........................................................53 3 LAND USE AND LAND COVER............................................................................55 Introduction .................................................................................................................55 Fieldwork Methods in Iapari and Assis Brazil .........................................................56 Fieldwork in Iapari ............................................................................................56 Fieldwork in Assis Brazil ....................................................................................59 The Differences in Methodology and Their Implications ...................................60 Operationalization of Variables ..................................................................................61 Comparing Variable Means for Assis Brazil and Iapari ...........................................68 Correlations Between Independent and Dependent Variables ...................................73 Multivariate Models ....................................................................................................79 Land Use Models .................................................................................................79 Land Cover Models .............................................................................................86 Land Use and Land Cover Final Models .............................................................92 4 CONCLUSIONS......................................................................................................101 APPENDIX A QUESTIONNAIRE APPLIED IN IAPARI, PERU..............................................104 B QUESTIONNAIRE APPLIED IN ASSIS BRAZIL, BRAZIL................................109 LIST OF REFERENCES.................................................................................................114 BIOGRAPHICAL SKETCH...........................................................................................121 vii

PAGE 8

LIST OF TABLES Table page 1-1 Comparisons of land area and population from the country level to the Municipio of Assis Brazil and the District of Iapari................................................5 1-2 Comparing historical processess for the Municpio of Assis Brazil and the District of Iapari.......................................................................................................8 2-1 Models of deforestation showing predicted effect of key variables.........................28 2-2 Household demographic variables used in land use modeling for Amazonian areas..........................................................................................................................29 3-1 Descriptive statistics for land use and land cover outcome variables. Iapari and Assis Brazil, 2003.................................................................................63 3-2 Descriptive statistics for household background information variables. Iapari and Assis Brazil, 2003.................................................................................65 3-3 Descriptive statistics for place, markets, credit and road infrastructure variables. Iapari and Assis Brazil, 2003.................................................................................66 3-4 Descriptive statistics for household life cycle variables. Iapari and Assis Brazil, 2003.................................................................................67 3-5 T-test of means for land use outcomes, land cover outcomes, background information, markets and credit, road infrastructure and household life cycle variables according to location in Iapari or Assis Brazil, 2003.............................69 3-6 Correlations between land use outcomes and land cover outcomes variables. Iapari and Assis Brazil, 2003.................................................................................73 3-7 Correlations between land use outcomes and background information. Iapari and Assis Brazil, 2003.................................................................................74 3-8 Correlations between land use outcomes market and credit and road infrastructure. Iapari and Assis Brazil, 2003..........................................................75 3-9 Correlations between land use outcomes and household cycles. Iapari and Assis Brazil, 2003.................................................................................76 viii

PAGE 9

3-10 Correlations between land use outcomes and land cover outcomes. Iapari and Assis Brazil, 2003.................................................................................76 3-11 Correlations between land cover outcomes and background information. Iapari and Assis Brazil, 2003.................................................................................77 3-12 Correlations between land cover outcomes, markets and credit and road infrastructure. Iapari and Assis Brazil, 2003..........................................................78 3-13 Correlations between land cover outcomes and household cycles. Iapari and Assis Brazil, 2003.................................................................................78 3-14 Models of farm area in annual crops outcome regressed on background information, market & credit road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003.................................................................................80 3-15 Models of farm area in perennial crops outcome regressed on background information, market & credit road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003.................................................................................82 3-16 Models of farm area in pasture outcome regressed on background information, market & credit, road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003......................................................................................................84 3-17 Models of head of cattle outcome regressed on background information, market and credit, road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003.............................................................................................................85 3-18 Models of area of old growth forest outcome regressed on background information, market and credit, road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003.................................................................................87 3-19 Models of area of secondary forest outcome regressed on background information, market and credit, road infrastructure, and household life cycle in Assis Brazil and Inapari, 2003.................................................................................88 3-20 Models of area (ha) deforested outcome regressed on background information, market and credit, road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003......................................................................................................90 3-21 Models of percentage of initial forest deforested since arrival outcome regressed on background information, market and credit, road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003............................................91 3-22 Final models for land use outcomes showing all the independent variables that were significant in final multivariate land use and land cover models. Assis Brazil and Iapari, 2003.................................................................................93 ix

PAGE 10

3-23 Final models for land cover outcomes showing all the independent variables that were significant in final multivariate land use and land cover models. Assis Brazil and Iapari, 2003.................................................................................94 x

PAGE 11

LIST OF FIGURES Figure page 1-1 Study area: Tri-national borde r Iapari (Peru), Assis Braz il (Brazil) and Bolpebra (Bolivia).....................................................................................................7 2-1 The three-tiered hierarchical approach.....................................................................41 2-2 Household transformations, land use and environmental change............................44 2-3 The adaptive cycle, the four ecosystem functions (r, K, ,) and the flow of events among them...................................................................................................46 2-4 Framework integration, the three-tiered hierarchical approach, household transformations approach and the panarchy approach.............................................48 2-5 Levels of interaction in a nested set of adaptive cycles...........................................49 2-6 Looking at the household and its land use activities as adaptive cycles..................51 3-1 Household farms visited in the Municipio of Assis Brazil and in the District of Iapari.......................................................................................................................58 xi

PAGE 12

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 LAND USE AND LAND COVER IN IAPARI, PERU, AND ASSIS BRAZIL, BRAZIL, SOUTHWEST AMAZONIA By Anglica Mara Almeyda Zambrano December 2004 Chair: Marianne Schmink Major Department: Center for Latin American Studies The present thesis research has its roots in the growing field of land use and land cover change. It compares land use and land cover across space in two bordering areas: the district of Iapari in Madre de Dios, Peru, and the municpio of Assis Brazil in Acre, Brazil. This study case has a micro-level approach that focuses on small farm households in this area know as the tri-national border Peru-Brazil-Bolivia. The research analyzes the differences in land use and land cover in Assis Brazil and Iapari and proposes an innovative integrative framework with roots on Political Ecology, Household Demography and Panarchy theories. The research also has an empirical component, the assessment of the proposed framework is done by modeling land use and land cover in both tows. The findings reveal the relevance of road infrastructure variables, market variables and background variables in explaining differences in land use and similarities in land cover outcomes. It also suggests that xii

PAGE 13

further development of the proposed integrative framework may contribute to a better understanding of different level process that drive land use and land cover change. xiii

PAGE 14

CHAPTER 1 INTRODUCTION The subject matter of this research has its roots in the growing field of land use and land cover change; however, differently from most studies it does not deal with changes across time. It compares land use and land cover across space in two bordering areas: the district of Iapari in Madre de Dios, Peru and the municpio of Assis Brazil in Acre, Brazil. The present document has been organized in four chapters. The chapter 1 briefly reviews the bases of the field of land use and land cover change, and presents the study area from a historical and current perspective observing its similarities and differences. The chapter 2 goes more in depth on relevant theories used to explain land use and land cover changes, and it ends by proposing an integrated theoretical framework for the present research. Chapter 3 describes the methods used to gather and analyze the data. It also presents and interprets the results obtained form the analysis. Chapter 4 presents a general discussion of the findings from Chapter 3 and elaborates general conclusions for the present research. Land Use and Land Cover Change It is only relatively recently that humans have taken a large role in modifying landscapes across the globe; although the process of massive change began in temperate areas, it is currently centered in the tropics (Ojima et al. 1994). Changes are so severe that when globally added they notably impinge on important Earth System functions (Lambin et al. 2001). Traditionally, research on human dimensions of global change has focused 1

PAGE 15

2 on two broad fields: industrial metabolism and land use and land cover change. The present research is framed within the latter. Land use and land cover change is recognized as an interdisciplinary research area. The land use component refers to the utilization of land, and it has been traditionally studied by social scientists, while the land cover component refers to the biotic and physical components of land surface, and has been traditionally studied by natural scientists (Meyer and Turner 1992). Land cover changes occur in two ways: conversion and modification. Conversion implies change in land category and modification implies change within a category (Meyer and Turner 1992). Land cover conversion is usually related to land use changes in area, and land cover modification is usually related to land use changes in intensity. The majority of existing literature deals with land cover conversion, although land cover modifications are also widespread and probably as important as the former but more difficult to asses (Meyer and Turner 1992; Lambin et al. 2001). Houghton (1994) explains that changes in land use have both intended and unintended consequences. The intended consequence is to increase area or productivity of a certain type of product, although some land uses have the opposite effect. The unintended consequence is to have a negative effect on global climate. Land use change has contributed to the enhancement of the greenhouse effect, 25% of human-caused emissions. Land use change rapidly changes ecosystems properties, regular inputs and exchanges of energy; water and nutrients in ecosystems are being severely changed and it may also create greater opportunities for exotic species invasion (Ojima et al. 1994). Well-recognized changes are trace-gas emissions, detriment of water quality, changes in water flows, and soil alteration and erosion.

PAGE 16

3 Proximate Sources and Driving Forces of Change Meyer and Turner (1992) differentiate between the proximate sources of change and the driving forces of change. Proximate sources are the human actions that alter the land cover, while driving forces are the underlying causes of proximate sources. At the global level, much research on driving forces has focused on human population pressure, but different changes may involve different driving forces, and the same changes may involve different driving forces in different areas of the world (Meyer and Turner 1992). According to Lambin et al. (2001) poverty and population growth are not the major causes of land cover change at the global scale. Significant correlations between land cover conversion and population were found only when the research area possessed similar social and environmental characteristics (Meyer and Turner 1992). The role of political, cultural and other demographic factors in land use decisions are gradually taking more relevance in the effort to understand global change (Ojima et al. 1994). Relationships are complex; the current higher rates of change in the so called developing areas may be explained by the demand from developed countries, since international trade is often an important land use change driver (Houghton 1994). Tropical deforestation has been a central concern among other land cover conversions. In general agricultural expansion is considered to be the main proximate source (Barbier 2001) and has been found to be driven by changing economic factors that are associated to institutional factors like social and political changes (Hecht 1985; Lambin et al. 2001). On average, 50% of the forest area lost in the tropics per year is used to replace agricultural areas that are no longer productive and were abandoned. Therefore only 50% goes to actually increase the area in agriculture (Houghton 1994). Data on past land use and land cover are not enough to improve current models of land use and land

PAGE 17

4 cover change. Data should be accompanied by a better understanding of the causes of land use and land cover change (Lambin et al. 2001). Barbier (2001) reported a list of factors that play an important role in tropical deforestation. Among those listed at the cross-country and country level are factors like income, population growth and density, agricultural prices and returns, agricultural yields, logging prices and returns, roads and road building and institutional factors. In general changes in markets, credit and roads are associated with changes in land cover, and therefore we will review them in more detail in the following chapter. Assis Brazil and Iapari Comparison studies in land use and land cover change are rare. One of the main constraints is to adequately address differences at the land use driver level. Socioeconomic and biophysical drivers may interact in ways that are difficult to understand when trying to make comparisons. To assess the role of biophysical land use drivers one would like to compare populations with very similar socioeconomic characteristics that are located in different landscapes. In a similar way, to assess the role of socioeconomic land use drivers one would like to compare populations that are in the same biophysical landscape that have different socioeconomic characteristics. The latter is precisely the case I expect to make for Assis Brazil and Iapari in this section. Both towns, due to their proximity, are located in very similar biophysical landscapes. This fact will allow us to compare the effect of socioeconomic land use drivers. Socioeconomic land use drivers are especially interesting in this bi-national context since they largely result from different development policies applied by Peruvian and Brazilian governments since this area was first reached by the white man in the early 1900s.

PAGE 18

5 Site Description General differences between Brazil and Peru are very evident at national levels. Brazil is larger in size and population (Table 1-1). However when we look at the lower administration levels, obvious differences diminish. The municpio of Assis Brazil has 2884.2 km 2 while the district of Iapari has 3,793.9 km 2 In the year 2000, Assis had a population of 3,490 persons. In 1993 Iapari had a population of 841; the population of Iapari had changed little by the year 2000, due to internal migration to bigger cities in Peru. Table 1-1. Comparisons of land area and population from the country level to the Municipio of Assis Brazil and the District of Iapari Variable Brazil Peru Federative Republic Constitutional Republic 26 states 24 departments Government 5 regions 12 regions Land area km 2 8.5 million 1.3 million Population 172.6 million 26.1 million State: Acre Department: Madre de Dios Land area km 2 153,149.9 1 85,182.6 2 Administration 5 Development regions 3 Provinces Municpio: Assis Brasil District: Iapari Land area km 2 2,884.2 3 3,793.9 4 Population 3490 6 (2000) 841 5 (1993) Municipal Seat: Assis Brasil District capital: Iapari 1 Governo do Estado do Acre GEA (2000a), 2 Insituto Nacional de Estadstica e Informtica INEI (1997), 3 GEA (2000a), 4 Instituto Nacional de Recursos Naturales INRENA (1998), 5 INEI (2002), 6 Instituto Brasileiro de Geografia e Estatstica IBGE (2002). The study sites are located in what is called Southwest Amazonia. Assis Brazil is a municipal seat in the state of Acre in Brazil and Iapari is a district capital in the department of Madre de Dios in Peru. The area of study consists of the Municpio of Assis Brazil and the District of Iapari (Figure 1-1). In geomorphologic terms, the area is quite similar on both sides of the border, and results from the interaction of tectonic, climatic, and erosive factors that have shaped its

PAGE 19

6 surface. The climate is the same: hot and tropical, seasonally humid, and with abundant rains and a short dry season that usually lasts from June to August. There are some initiatives that cross the international border between Peru and Brazil such as the Development Program for the Peruvian-Brazilian Border Communities, a bi-national program led by the ministers of foreign affairs of Peru and Brazil with the assistance of the General Secretariat of the Organization of American States (GS/OAS). The Brazilian area covers the Municpio of Assis Brazil and the Peruvian side covers the Province of Tahuamanu; the district of Iapari is located within this province (SUDAM and INADE 1998). Since 2000 there have been annual meetings known as MAP (Madre de Dios-Acre-Pando) that bring together academic institutions, international cooperation agencies, non-governmental organizations, and local, state and national governments committed to sustainable development and conservation in the MAP area. According to the document elaborated by the SUDAM and INADE (1998), cities in the area are mainly rural, work on farms being the principal economic activities. Assis Brazil and Iapari are the main centers with urban characteristics. Both are small towns cut by a road. Table 1-2 presents a general time line for both towns making a comparison of the different events that should explain land cover differences. Assis Brazil and Acre in Context Acre was the traditional territory of different indigenous groups that fled, were killed or were forced to move by the correrias when the white man arrived looking for rubber (Hevea brasiliensis). No missions or indigenous slavery existed in the area during the colonial period (GEA 2000b).

PAGE 20

7 Fi g ure 1-1. Stud y Area: Tri-national border Ia p ari ( Peru ), Assis Brazil ( Brazil ) and Bol p ebra ( Bolivia )

PAGE 21

8 Table 1-2. Comparing historical processess for the Municpio of Assis Brazil and the District of Iapari Year Iapari Assis Brazil 1900s Indigenous territory 1900-13 Rubber boom until plantations in Asia took over rubber production 1914-50 Migration to Puerto Maldonado and Cuzco Tire industry and World War II maintain rubber tapping in Acre Seringa 1950s Banco de Fomento Agropecuario 1960s Unpaved main roads are built 1970s Agrarian Bank is established Operation Amazonia 1980s Directed Settlement Conflicts over land 1990s Agrarian Bank is closed Secondary roads are built Main road is paved 2000s Main road is improved Credit programs are established Acre was settled when rubber extraction expanded beyond Belm and Manaus; between 1850 and 1870, suppliers spread their network westward to the Madeira and Purus rivers (Schmink and Wood 1992). This was Bolivian territory, incorporated to Brazil in 1903 after a war, with the signing of the Petropolis Treaty (GEA 2000a). Population composition Household productive structures in Acre are classified in the following categories (GEA 2000b): Ribeirinhos. The first settlements in Acre were reached by river; along their banks are found most of the municipal seats. The riparian populations established communities based on family productive activities. They had a diversified subsistence production, adapted to the Amazonian environment, without large-scale shifting cultivation practices (GEA 2000b). Seringueiros. Migrants from the northeast, nordestinos, were the labor force for rubber extraction. However, English plantations in Asia took over world production of

PAGE 22

9 rubber after 1913, giving origin to the rubber crisis in Brazil. In the 1920s in areas endowed with dense stands of the castanheira tree (Bertholletia excelsa), Brazil nuts became an especially important export item (Schmink and Wood 1992). During World War II the North American and Brazilian governments coordinated to stimulate rubber extraction in Brazil (Batalha da Borracha) with a second nordestino migration to Acre. After the armed forces took control of the Brazilian government in 1964, the Operation Amazonia (Wood and Schmink 1993) began in 1966, and many landlords from the south and Southeast moved into Acre stimulated by federal incentives on cattle, logging and mining. This generated major land conflicts in the mid 1970s that resulted in the institutionalization of Extractivist Settlement Projects in 1987 (Projetos de Assentamento Extrativista PAEs under Instituto Nacional de Colonizao e Reforma Agrria INCRA administration) and Extractive Reserves in 1990 (Reservas Extrativistas RESEX, under Instituto Brasileiro de Meio Ambiente e dos Recursos Naturais Renovveis IBAMA administration). Due to the decline in rubber price, many families now migrate to agricultural settlements, ranches and urban peripheries. Families that remain in PAE and RESEX areas move from extractive to farming, cattle and logging activities (GEA 2000b). Colonos. Farming families are located in the Directed Settlement Projects (Projeto de Assentamento Dirigido PAD) and Colonization Projects of INCRA. Traditional state models of rural settlement in Acre present many problems; for example, family plots may be designed without considering soil quality, topography or water courses. Their main farming activities are corn, rice and beans, coffee, papaya, passion fruit, banana and pineapple. Extractive activities may be directed to wood, brazil nut, aa and game (GEA

PAGE 23

10 2000b). There is also a strong tendency to establish cattle for milk and meat, with consequences in the increase of deforested areas. Pecuaristas. The first expansion phase of cattle raising occurred in the 1960s-70s, when the military regime began the Amazonian Operation, offering incentives for cattle raising. Prior to this they had cut financial help to rubber traders who were forced to sell their lands at low prices to the landlords from the South. In the 1970s both the federal and state governments wanted to convert Acre into a major meat producer. The second phase (1979-1989) was marked by an increase in degraded areas, mainly because of a pest that thrived in established pastures. At the same time new pasture species adapted to tropical weather were introduced, as well as better cattle management practices. The third phase was initiated in 1989 by the federal government prohibition on the use of official credit for development activities that result in deforestation in the Amazonia. The development of environmental laws, the increase in environmental control and the availability of new technologies and pressures on tropical forest conservation made pecuaristas adopt strategies to recover deforested areas. But it is clear that there is a strong tendency to increase cattle raising areas in households of all kinds and size (GEA 2000b). Natural protected areas Acre has two Indirect Use Conservation Units (Unidade de Conservao de Uso Indireto UCUI), the Serra do Divisor National Park and the Acre River Ecological Station, both are under the administration of IBAMA. The latter is located in the southeast, distributed in the municpios of Assis Brasil and Sena Madureira. It was created in 1981 by the federal government. The Ecological Station also borders two Indigenous Lands (Terras Indigenas TIs under the administration of Fundao Nacional do ndio FUNAI). The TI Cabeceira Do Rio Acre covers 78.513 ha, and has 134

PAGE 24

11 persons, from the Jaminawa ethnic group. Environment conflicts in the Ecological Station and TIs are caused by the proximity to Assis Brazil, which increases logging and game activities by poachers (GEA 2000b). Transportation and highways The oldest transportation system in Acre is the rivers, the Madeira River being the most important as it is the cheapest way to get to Manaus. The Acre River is also important during the rainy season. There is aerial domestic transportation from Rio Branco (flights from Rio Branco to Puerto Maldonado are offered in a less consistent manner). Finally there is a network of federal and state highways, municipal roads and INCRAs secondary roads. The federal government during the military dictatorship, seekign to promote Brazils industrialization, gave priority to the construction of highways. In the case of Acre they were: BR-364 (Rio Branco Peruvian border), BR-317 (Lbrea/AM Assis Brasil/AC Peruvian border), BR307 (Marechal Tahumaturgo/AC Benjamin Constant/AM Venezuelan border) and BR-409 (Feij Santa Rosa). Brazils transport authority (BR) is the responsibility of the National Highway Department (Departamento Nacional de Estras e Rodagem DNER), which delegated some highways to the Army, some of which were sub-delegated to the State Highway Department (Departamento de Estradas e Rodagem DERACRE). INCRA secondary roads (to settlement projects, colonization and others) are under its responsibility, some of which were also transferred to DERACRE. The most recent infrastructure development in Acre, part of the Avana Brasil Program, is precisely the paving the BR-317/AC Brasilia Assis Brasil. This has meant the paving of 110 km with the purpose of integrating the Southwest of Acre with

PAGE 25

12 the national highway system and to promote relations with Peru, making it possible to access the Pacific Ocean (Ministerio do Planejamento 2003). Iapari and Madre de Dios in Context The era of rubber extraction (1895 1940) marked the beginning of the non-indigenous settlement of Madre de Dios. During this period international companies brought the immigration of large numbers of peasants. Thus, the area of the Acre and Manuripe rivers has the highest number of towns in the department. Between 1900 1924 peasants from Cajamarca, La Libertad, Arequipa and Loreto, as well as European and Japanese immigrants also came to the area, and due to concern about land tenure, the Peruvian government created the city of Puerto Maldonado in 1902 (INRENA 1998). The decline in rubber prices first, and then the Cuzco Puerto Maldonado road constructed in 1961, stimulated migration to Puerto Maldonado and Cuzco. The remaining population in the Iberia Iapari area were based on a subsistence economy (INRENA 1998). In the 1960s the government of Feranado Belaunde provided the first incentives to cattle ranching in Madre de Dios. The government Office for Agricultural Research (Servicios de Promocin e Investigacin Agraria SIPA) established a cattle ranch in Madre de Dios to expand cattle raising, and encouraged the genetic improvement of herds (Jorge Coronel cited by Varese 1999). Gold mining also is an important activity in Madre de Dios. It is practiced in three main forms: manual, in distant locations as a family activity or with peons; with pumps, in the Inambari, Madre de Dios, Malinosky and Colorado rivers; and with heavy machinery, in the Caychihue, Huaypethue, Madre de Dios and Malinosky rivers (Arbex 1997).

PAGE 26

13 Logging is a major activity in Madre de Dios, though since December 6, 1999 the Ministry of Agriculture outlawed all industrial or commercial logging in the region; after that the President declared a ban on cedar and mahogany logging in Madre de Dios, effective from January 1 st 2000. Until 2001 the Peruvian Forestry law only authorized the granting of logging contracts for areas less than 1000 ha, and for less than two years. Within Madre de Dios such logging was permitted only in the Tambopata province and in the Tahuamanu district (AIDA and SPDA 2002). A major change in logging activities was brought with the new Forestry Law (Ley Forestal Y De Fauna Silvestre 2000) and its regulations (Reglamento De La Ley Forestal Y De Fauna Silvestre 2001) that give forest concessions for longer periods and over larger areas. During the 1980s and 1990s the Government carried out two Directed Colonization Programs in the district of Iapari, the Proyecto Especial Madre de Dios in Primavera and Chilina, both of which failed: desertion of the colony in Chilina is around 56%, and in Primavera it is around 80% (INRENA 1998). A study on land use was conducted in the Iberia Iapari area in 1997 covering 204,550 ha. The area with permanent crops was 0.1%; banana was the main species, and papaya, pineapple, coffee, and cacao were also present, usually for subsistence. Families also had small areas with tomatoes, onions, garlic, lettuce and subsistence crops. Farming activities were carried out along main roads and rivers. Pastures represented 3.3% of the area, both abandoned and productive. Forests covered 92.7% of the area (INRENA 1998). According to the study made by the Brazilian Superintendency for Development of Amazonia (Super Intendncia de Desenvolvimento da Amaznia SUDAM) and the Peruvian National Development Institute (Instituto Nacional de Desarrollo INADE),

PAGE 27

14 after the Agrarian Bank of Peru (Banco Agrario del Peru BAP) and the National Rice Commercialization Enterprise (Empresa Comercializadora de Arroz ECASA) were deactivated in 1991 crop production has mainly had subsistence purposes (SUDAM and INADE 1998). Natural protected areas and indigenous communities At present Madre de Dios is renowned worldwide for its outstanding biological diversity, and it has been a place of extensive research. As a result there are four areas of strict protection: Pampas del Heath National Sanctuary; Bahuaja Sonene National Park; Manu National Park; and the Tambopata-Candamo National Reserve. All of these are under INRENA administration. Madre de Dios has a low population density and is home to diverse indigenous peoples. The valleys of the Piedras, Yaco, Chandles and Alto Manu rivers are the ancestral territories of indigenous communities from the Pano family; these indigenous populations live in the floodplains in the Iapari and Iberia provinces. The native community Belgica (Arawak) is located in the district of Iapari. According to the 1991 census, they included 151 persons, but their right to the land they use was not legally recognized (INRENA 1998). Transportations and highways In Madre de Dios the Tambopata and Madre de Dios rivers are important and inexpensive transportation. Domestic flights are available in the city of Puerto Maldonado, and sometimes flights to Rio Branco are available. The transportation network is made up of what is now called the Pacific highway. Starting at the coast of Peru the route is Matarani Juliaca Puerto Maldonado Iberia Iapari. The road is not paved, and is difficult to transit during the rainy season in its Andean and Amazon

PAGE 28

15 portions. From Puerto Maldonado there is also a road that connects with Cuzco and from there to Lima. In Peru usually the Transportation Ministry takes care of roads and highways; however this task is often given to Special Development Projects (Proyectos Especiales), regional and local governments. In the District of Iapari the main transportation is the road to Iberia, a 50km unpaved road with some secondary roads, usually in bad shape. This is the road that was built by the Proyecto Especial Madre de Dios between 1998 and 2000. Iapari also has a small airport that has hardly been used. The Case for Comparing Iapari and Assis Brazil This section looks more in depth at certain factors that are usually recognized as land use drivers and that are also different in nature in Assis Brazil and in Iapari. The discussion is centered on road infrastructure, market and credit. The Role of Roads Both towns were traditionally isolated from the rest of their respective countries until recent years. From the 1960s to the late 1990s they were linked to the rest of their countries by a dirt road suitable for a walk or to make a trip in a tractor for three days to get to the closest town. Difficulties were mentioned, especially in cases of health problems. Infrastructure development in Peru and Brazil towards the construction of the Pacific highway changed the roads on both sides. There is a general sense among settlers that the roads are the best thing the government has done for the people living in this area so far. In the case of Iapari, the unpaved road that links it to the town of Iberia and to the city of Puerto Maldonado was built in many phases by the Proyecto Especial Madre de Dios. Its last portion was declared finished in October 21 st 2000. This road, although not

PAGE 29

16 paved yet, is described by the engineers in the Proyecto Especial as ready to be paved. Most farms are located along this road and there are only two secondary roads in the district. In the case of Assis Brazil, the road that links it to the city of Brasilia (BR-317) was paved in its final portion in 2002, within the frame of the Avana Brazil Program, and for the formal inauguration presidents of both countries got together in December 2002 in Assis Brazil. There are also many secondary roads that connect to the different farmlands in the Municpio, usually in bad shape during the rainy season. There is a difference in the process here: both sites had very difficult dirt roads that could only be passed by 4 x 4 trucks, referred to as Toyota, during the rainy season; while in 2000 after a long process an unpaved road passable all year reached Iapari, in the case of Assis Brazil in 2002 the road was upgraded to a paved road in a process that took less than two years. The next step in the building of the Pacific highway is the construction of a bridge over the Acre Rive; initial work is already going on and the 175m bridge, that will cost US$7 million should be completed by the end of 2004. After that, the paving of the roads on the Peruvian side should start, although funds are not available yet. One of the issues that is taking time and effort from the local authorities is whether the Pacific highway should cross through the center of the towns or if it should go around the towns. It is apparent that most people would like the highway to cross the towns, for fear of not getting any benefits if it does not. The Importance of Markets Throughout history, from the elders references, both towns were closely related. In terms of trading, in the beginning it was Iapari who was at an advantage, meaning that

PAGE 30

17 settlers from Assis Brazil would buy goods on the Peruvian side, first things like soap and candles and later manufactured goods. Although no one could pinpoint when the reverse process started, the current scene shows a different picture. In 2000, my first visit to the area, the presence of the J.B grocery store (like a small supermarket) in Assis Brazil was unexpected; in 2003 with the new paved road I found at least two more of these stores in town selling manufactured goods that came from different parts of Brazil. Prices are so low compared to prices in Madre de Dios that not only people from Iapari but also those from Iberia and Puerto Maldonado make the trip and cross the border to do some shopping. But these stores not only sell goods they brought from other parts of Brazil, they also buy a small amount of products from some local farmers like rice, beans and fruits. So, from time to time, one can see the peculiar scene of a big ox with its wooden cart being discharged of farm products and loaded with manufactured goods at the front door of the grocery store. But while these markets are the places where most people on both sides of the borders buy manufactured goods, the stores buy only a small part from local farms in Assis Brazil. The market for the agricultural products in both towns is much reduced, and farmers produce mainly for subsistence. On both sides, complaints from producers are the same: the local market demand is not enough to consume all that is produced in the area, and transportation cost is the main restrictive factor to take the products to other places, to Iberia in the case of Iapari and to Brasilia in the case of Assis Brazil. For small animals like chickens, ducks and pigs, the market is usually Puerto Maldonado, even for the Brazilian livestock that is not supposed to cross the border without a permit from the

PAGE 31

18 Peruvian National Service for Animal and Plant Health (Servicio Nacional de Sanidad Agropecuaria SENASA) office in Iapari. But the most successful market is the market for beef; buyers from Puerto Maldonado and from Rio Branco, respectively, go all the way to the farmlands with trucks that may fit up to 8 -12 animals, depending on their size. Settlers do not take their own cattle to the city; it is suggested that a cattle mafia arrived with the roads. They buy cattle at your door, and with cash money, usually young bulls that are taken to fields near the cities to be fed, processed and their meat sold. Before the road, the only way to get the cattle to the cities was by walking them for 4 days, which was usually done by the cattle owners. Timber, however, has a different story in each town. To begin with, logging is not allowed in Assis Brazil, and permits are officially necessary if one wants to use timber from ones own land for construction. In Iapari, logging has been banned since 1977, but logging activities were traditional carried out by local small loggers. Although no farmer or settler says that logging is their only economic activity, it is certainly an important part of the livelihood strategies for some of them. Logging became an important source of conflict since 1999, when illegal logging permits were given to a Peruvian logging company called Empresa Industrial Maderera Tahuamanu EIRL, who had a joint venture with Newman Lumber Company of Mississippi. The venture installed the first big sawmill with foreign capital, and a 100km extraction road was opened with the purpose of extracting mahogany that was directly exported to the U.S. (Caillaux and Chirinos 2003; AIDA and SPDA 2002). The conflict over timber led to many legal battles between the National Institute for Natural Resources (Instituto Nacional de

PAGE 32

19 Recursos Naturales INRENA) and the Newman Lumber companyover a three-year period, as well explained by Calliux and Chirinos (2003). Almost at the same time in 2000, the new Forestry Law was approved (Ley Forestal Y De Fauna Silvestre 2000) and in 2001 despite all the street protests that took place in Puerto Maldonado, the first forest concession competition took place, and a total of 99 000 ha, two of the three areas that were available in the district of Iapari, were given in concession to two local small loggers associations. Existence and Inexistence of Credit Credit in Iapari has existed since the 1950s with the Banco de Fomento Agropecuario, whose name was changed afterwards to Banco Agrario del Peru (BAP) by the Gobierno MIitar Revolucionario of Gral. Juan Velazco Alvarado in the early 1970s. It provided credit first for rubber, and then in materials and/or cash for agriculture and cattle, from theearly 1980s until 1991, when the bank was closed under President Fujimoris regime. The people who were in Iapari in los tiempos del banco remember having received credit for agriculture, for cattle and for small animals. Since the Banco Agrario there is not a single type of credit available to rural small farmers. In Assis Brazil, however, the Constitutional Fund for the North (Fundo Constitucional de Financiamento do Norte FNO) and the National Program for the Strengthening of Family Agriculture (Programa Nacional de Fortalecimento da Agricultura Familiar PRONAF) are currently available, at least for some of the farmers and until 2001 there was also a fund available for cattle ranching. Roads, Markets and Credits as Land Use Drivers From Figure 1-1 to the last section, the present chapter provided the base information to establish the first hypothesis:

PAGE 33

20 H1: Access to markets, credit and road infrastructure drove more deforestation by households in Assis Brazil than in Iapari. By observing Figure 1-1 one can see that there is a bigger area deforested in the Municipo of Assis Brazil than in the District of Iapari. This comparative study will therefore focus on key differences in drivers of deforestation that may explain this difference in land cover in the two sites. From the information available for both sites it is apparent that there is better road infrastructure and a better market in Assis Brazil. Credit, however, has different timings: it was available in the are of Iapari for nearly 30 years until 1992, and it is currently available in Assis Brazil since 1999. There is also a population difference: Assis Brazil has a higher population than Iapari. In order to control for that difference, a household level approach will be better than a Municpio or District approach.

PAGE 34

CHAPTER 2 THEORETICAL APPROACH Introduction In an effort to improve current understanding of land use and land cover change drivers, different analytical and empirical approaches have considered different temporal and spatial scales. More recently, research has been conducted in Amazonia, focusing on household level land use practices, as they are often pointed as to being responsible for deforestation. The present chapter looks into the factors considered to be important land use drivers, precisely into the ones that are of interest for the present research: markets, credit and road infrastructure. Then it explores the relevant theories used in explaining deforestation since this sort of land cover change has been the center of many studies in the field of land use and land cover change. It finally reviews in more detail three theoretical frameworks, and concludes by proposing an integrated theoretical framework. Land Use Drivers and Deforestation Markets and Credit The socioeconomic matrix of deforestation for Amazonia elaborated by Schmink (1994) explains deforestation as an outcome of social processes, and locates markets in the global and national contexts. At the global level important variables are the demand for Amazon products (e.g. timber, rubber) and foreign investment (e.g. oil, mining, timber). At the national level the variables considered are transportation and export orientation. Schmink (1994) describes international and national markets as an important factor for the expansion into remote forest areas. 21

PAGE 35

22 The debate of how economic development impacts the environment includes the roles of markets in the use of natural resources (Godoy et al. 1997b). The topic has been extensively researched in the case of traditional populations; however, the basic principles may apply to any group. According to Godoy et al. (1997b) there are three main positions in regard to market: the market works to the detriment of conservation; the market increases conservation, if land rights are secure; and the market has ambiguous effects on deforestation. Households integrate into markets by selling crops, labor, or both. If integration is achieved by selling crops, increase in market demand will increase deforestation, unless intensification occurs. If integration is achieved by selling labor, increase in market demand will reduce deforestation since there will be less time to work the land (Godoy et al. 1997b). Therefore integration into both markets usually has nonlinear effects. In their review of models of deforestation at the household level, Kaimowitz and Angelsen (1998) include transportation costs which show an inverse relation between market access costs and deforestation. They also find that an increase in off-farm income sources typically decreases the pressure on forests. However, increased participation in market oriented activities does not always have a positive impact (Schmink 2004). Market dependency is in many cases not desirable since it leaves little room for subsistence activities, making the producer vulnerable to changes in demand, and price (Schmink 2004). Market integration deserves special attention in the case of traditional groups with little market experience since some of the features of market economy, like profit, may erode their cultural ties (Smith 1995). Increase in market demand may also put more pressure on forest resources (Schmink 2004), resulting in increased

PAGE 36

23 deforestation. Furthermore, social justice and equity are not expected outcomes from market participation (Schmink 2004). Government subsidies are one way of addressing the local producer disadvantages (Schmink 2004). In the case of the Brazilian Amazon subsidies include a combination of road building, colonization projects, and taxes and credits that have helped to foster the frontier expansion process (Wood and Schmink 1993). Roads Driving Land Use Access has been recognized as the main factor in the spatial distribution and rate of deforestation (Soares-Filho et al. 2002). Historically, rivers and roads have provided easy access to tropical forests, but roads are especially associated with deforestation and social conflicts (Schmink and Wood 1992). However, road improvement is an important priority in many Amazonian countries (Mki et al. 2001) and transportation plays a important role in development (Leinbach 2000). In most cases it is necessary, but not sufficient for development. Local social and environmental characteristics have an important effect in the way roads influence economic and social changes. Moreover, roads do not always help to alleviate poverty; development will depend on the local and regional economies capacity to reallocate resources (Leinbach 2000). In their review of models of tropical deforestation Kaimowitz and Angelsen (1998) found that roads, rivers, railroads, and low gas prices provide greater access to forests. Roads usually lead to more deforestation when there is also access to markets, especially in areas of good soil with commercial agriculture. However, the relation is not always direct; in some cases roads are built in previously settled and cleared areas, or settlement and roads may be influenced by other variables.

PAGE 37

24 Alves (2002) made an analysis of the geographical distribution of the deforested areas in the Brazilian Legal Amazon. His results show that deforestation is concentrated around major roads and pioneer settlements. Three quarters of the deforestation between 1978 and 1994 were within 50 km (on each side) of major roads. The Brazilian Development Program Advance Brazil (Avana Brasil ) effective since 2000, and projected to be active until 2007, involves the paving of 6000 km of roads (Minsterio de Planejamento e Oramento 1999). Some of the roads to be paved will provide access to 31 indigenous lands and 26 conservation units and remote forest regions (Nepstad et al. 2000). This project has brought the issue of road paving and its relation to deforestation to the front line and different scientific teams have worked in developing deforestation scenarios for the Legal Amazon. In modeling past deforestation, Laurance et al. (2001) found that paved roads have more far-reaching effects than unpaved roads. On average, areas further than 25 km from an unpaved road have less than 15% forest loss, but for paved roads average forest loss is 15% for areas between 26 and 50 km from the paved road. Nepstad et al. (2001) found that 29-58% was deforested within 50km from paved roads and 0-9% for unpaved roads. Paved roads produce three vicious cycles: the first is related to the cycle of cattle ranching, annual crops and its reinforcement by the use of fire; the second is related to conventional logging and its implications for wildfire during severe droughts; and the third is related to rain inhibition due to the former cycles (Nepstad et al. 2000; Nepstad et al. 2001). Investments in forest management and perennial crops would decrease the use of fire, aswould the establishment of rules in the use of fire, and the incorporation of fire prevention incentives in currently available credit lines (Nepstad et al. 2001).

PAGE 38

25 In economic terms the opening of new frontiers increases land supply, reducing the land value in the older frontiers. It also represents more work for the government to monitor natural resources use, as well as to provide health, education, and technical assistance (Nepstad et al. 2000; Carvalho et al. 2001). In general, it will encourage colonization and forest clearing that the government does not have the capacity to control (Laurance et al. 2001). In economic terms the opening of new frontiers increases land supply, reducing the land value in the older frontiers. It also represents more work for the government to monitor natural resource use, as well as to provide health, education, and technical assistance (Nepstad et al. 2000; Carvalho et al. 2001). In general, it will encourage colonization and forest clearing that the government does not have the capacity to control (Laurance et al. 2001). The proposed alternative is to promote old frontier areas through local road networks that allow producers to reach trading areas, with technical assistance, and with health and education programs (Nepstad et al. 2000). Therefore, roads where settlements already exist are desirable like Altamira-Marab and Brasilia-Assis Brasil, and roads that open new frontiers like Santarm-Cuib and Humait-Manaus are not (Nepstad et al. 2001; Nepstad et al. 2000; Carvalho et al. 2001). The most important issue is that priorities for transport policy in rural areas must meet the necessities of the poor population and not those of elite groups (Leinbach 2000), like has happened in the past (Schmink and Wood 1992; Wood and Schmink 1993).The increase of governance in frontier areas through strengthening of existing management institutions, adequate land use planning and enforcement of the existing environmental legislation is recognized as a

PAGE 39

26 very much needed measure to control land conversion following road paving (Nepstad et al. 2002; Mki et al. 2001). Approaches to Explain Deforestation To analyze the effect of different land use drivers it is necessary to pay special attention to the specific scale and place of research in order to make the appropriate assumptions. In general, most studies have been made at the country or cross country level in tropical areas. Four of the most important approaches for deforestation analysis were reviewed by Barbier (2001): the Environmental Kuznets Curve (EKC), competing land use models, forest land conversion models and institutional models. The EKC hypothesizes that environmental problems (like deforestation) decrease as the per capita income of a country rises. For Latin America and Africa the per capita income level at which deforestation equals zero is two to four times higher than the current average (Cropper and Griffiths 1994). Competing land use models hypothesize that deforestation results from competing land use, mainly between forests and agriculture. Therefore an opportunity cost is calculated for agricultural conversion versus potential timber and environmental services from forests. An important consideration is that very often agricultural conversion follows timber extraction (Barbier 2001). Forest land conversion models assume that households use their own labor or hire labor for land conversion, the level of cleared land is hypothesized to be a function of output and input prices, but these data are usually difficult to obtain. Institutional models, on the other hand, are centered on factors like ownership and political stability. Kaimowitz and Angelsens (1998) more extensive review of economic models of tropical deforestation identified three primary levels: household and firm, regional and

PAGE 40

27 national and macro models. Household and firm level models, the relevant level for the present research, were divided into three categories: analytical open economy (e.g. Angelsen 1999), analytical subsistence and Chayanovian (e.g. Angelsen 1999), and empirical and simulation models (e.g. Godoy et al. 1997a). Analytical models are theoretical constructs; they allow researchers to examine the implications of their assumptions. Empirical models quantify the relationships between variables using statistical methods, and simulation models use parameters to assess scenarios under different circumstances. Open economy models assume that households and firms actions have no impact on prices, and that market prices (including labor) fully determine how they value their resources. In this way household production is analyzed as profit maximization oriented. Subsistence and Chayanovian models assume imperfect markets (particularly labor) and that household consumption does matter in production decisions. The household goal is to maximize utility. Empirical and simulation models require time consuming surveys for data collection. For these models, common independent variables are transportation costs, farmers background, credit access, input and output prices, and tenure security (Kaimowitz and Angelsen 1998). The different assumptions and methods in each category show different results. The summary of findings for the analytical models are presented in Table 2-1. Site specific characteristics may play an important role in land use decisions, which may result in different land cover outcomes. That is precisely why empirical research is important and limited at the same time. While it provides site specific insights, findings should not be extrapolated without serious considerations and reservations.

PAGE 41

28 Table 2-1. Models of deforestation showing predicted effect of key variables Analytical Model Variable Subsistence Chayanovian Open economy Higher agricultural prices Reduce Indeterminate Increase Population growth Increase Increase No effect Lower transport costs No effect or reduce Increase Increase Higher agricultural productivity Reduce Indeterminate Increase Higher wages NA (reduce) NA (reduce) Reduce Higher land prices Reduce Higher interest rates Reduce Adapted from: Kaimowitz and Angelsen (1998) To situate the present research within the categories at the household level set out by Kaimowitz and Angelsen (1998) we may say that the present research is analytical and empirical. It proposes an analytical integrated framework and uses survey data to quantify the relationships between land use and land cover outcomes and variables commonly perceived as land use drivers from open economy models, and household demographic variables from chayanovian models. Empirical models of land use that incorporate demographic variables at the household level were reviewed by Perz (2001). His revision is specific to the Amazon and to the Neotropical Americas. Table 2-2 is an updated version from Perz (2001). General observations from his results show that the age of the household, length of residence, family size, and number of family members all influence land use in terms of area deforested, in old or secondary growth forest, as well as household agricultural decisions, among many factors.

PAGE 42

29 Table 2-2. Household demographic variables used in land use modeling for Amazonian areas. Demographic variables Age of household head Length of residence Family size Adults (males, females)b Children Variable (1) (2) (3) (4) (5) Alston et al. (1993) % area annuals +ns c +ns -ns % area perennials -ns +ns -ns % area pasture -ns +ns +ns Rudel and Horowitz (1993) % land deforested -** Ozrio and Campari (1995) Ha deforested since arrival -ns +ns Ha deforested in 1991 +ns +ns Sydenstricker and Vosti (1993) Ha deforested +**, +* Jones et al. (1995) Cleared ha per year Total cleared ha +ns Alston et al. (1996) % area crops or pasture +ns c -ns Godoy et al. (1997a) Ha old growth forest +ns c -ns -ns Godoy et al. (1997b) Prob. Of cutting old growth forest Mojeno +ns -ns Yurucare +* Chimane +ns -ns Pichn (1997) % land in annuals -ns +** % land in perennials +ns +** % land in pasture +** +ns % land in forest -** -** Godoy et al (1998a) Ha primary forest cut -** +** +**, +ns +ns Godoy et al (1998b) Ha old growth forest cut +ns +** -ns Godoy et al (1998c) Ha primary forest cut -ns +ns +ns McCraken et al (1999) Annual deforestation 1988-1991 -* Ha forested in 1991 -* Source: Perz 2001. Authors appear in chronological order. A "+" indicated a positive or direct effect, and a "-" indicates a negative or inverse effect. A "ns" indicates not significant at p > 0.10, indicates significance p < 0.10, and ** indicates significance at p < 0.05.

PAGE 43

30 Table 2-2. Continued Demographic variables Age of household head Length of residence Family size Adults (males, females)b Children Variable (1) (2) (3) (4) (5) Wood and Walker (2000) Ha deforested on arrival +* +* Ha deforested at interview +* +* Cocoa +* +* Coffee -ns +* Ha pasture +ns +ns Head cattle +ns +* Reforestation +* +ns Gomes (2001) Area deforested +ns +**, +ns Pasture size +ns Head of cattle +ns Perz (2001) Annuals +ns -ns +ns +ns Perennials +ns +** +** +ns Pasture size -ns +** +** -ns Cattle +ns +** +** -ns Reforestation +ns +ns +ns +ns Perz (2002a) Area in forest -ns +ns -** Area in cropland -ns +** +** Area in pasture -ns +ns -ns Area in secondary growth +** +ns +** Perz and Walker (2002) tobit 1,2,3 Secondary forest growth under fallow +ns, +**, +** na, +ns, +ns na, +**, +** Source: Perz 2001. Authors appear in chronological order. A "+" indicated a positive or direct effect, and a "-" indicates a negative or inverse effect. A "ns" indicates not significant at p > 0.10, indicates significance p < 0.10, and ** indicates significance at p < 0.05. Integrated Theories While a good understanding of the effect of individual land use drivers that are relevant for this study, as well as an overview of the existing theories that intend to explain deforestation reveals limitations in the tools available to adequately address land use and land cover change issues when dealing with a complex reality. That is the

PAGE 44

31 primary reason for the existence of integrated theories that intend to provide a better explanation for changes in land use and land cover. The framework used in this study comes from the integration of three frameworks drawing on three different integrative theories. The first is a hierarchical framework drawn from the bases of political ecology theory. The second looks at household transformations from demography theory. These two frameworks explain land use at different levels. The third framework provides the elements to link the previous ones and to make them more flexible; it is drawn from panarchy theory These three integrated frameworks were chosen for various reasons,. First, political ecology and demography or Chayanovian theories have been used in explaining land use and land cover changes in frontier areas in the Amazon region. Political ecology is explicit in linking global and local events; it brings in the spatial dimension, and it is preferred over open economy theories because their assumptions of perfect markets and information are far from real in the Amazon frontier. Demography or Chayanovian theories provide a tool to work at the household level, as required by this research, and bring in a temporal factor by looking at household composition over time. Panarchy presents an innovative opportunity to link political ecology and Chayanovian theories, since it has explicit temporal and spatial dimensions and has its origins in natural resource use. The purpose of integration is to show that biophysical and socioeconomic drivers of land use and land cover change do not have a linear or constant influence on households and their land use systems (farming, ranching, logging, etc). Instead, much

PAGE 45

32 depends on the stage that the household and the productive system occupy in terms of life cycles. All components of theory are linked together in a coherent conceptual structure named a theoretical framework. Integration is the union of existing theory, perspectives, approaches, models or data that are apparently disparate (Kuchka 2001). It is important to find out how paradigms, theories and theoretical practices themselves limit integration and how those constraints may be overcome; theoretical understanding changes through integration (Kuchka 2001). Integration is a difficult task. Some of the procedures and circumstances necessary for successful theoretical integration include (Kuchka 2001): 1. Domain: the domain of the related theories must be clearly stated, to make the development of linkages between theories more feasible. 2. Concepts: the meanings and subjects of concepts should be clear; this enables the asking of new questions that may further integration and the development of theory. 3. Scale and level: it should be clear if the theories are answering questions across levels of organization or particular adjacent levels of a given scale. For the particular research question of this study, it is my intention to integrate frameworks from three different integrative theories: (1) Political Ecology, (2) Demography, and (3) Panarchy. I will begin by reviewing some of their basic components: 1. Basic conceptual devices (assumptions, definitions and concepts) 2. Framework and structure (framework and domain) Political Ecology Political ecology has its origins as a new research field in the 1970s. It was a reflection of a need for an analytical approach integrating environmental and political

PAGE 46

33 understanding given the increase in environmental problems in the Third World (Bryant 1992; Bryant and Bailey 1997). In its first phase from the late 1970s to the mid 1980s political ecology was mainly a critique of neo-Malthusian and cultural ecology, and had its theoretical base in neo-Marxism. In its second phase from the late 1980s to the 1990s; it was mainly a critique of deterministic neo-Marxism and had its theoretical base in neo-Weberianism, social movement and household/feminist theories. Empirical analysis in this field has been favored; this has resulted in a research field grounded less in a coherent theory than in similar areas of inquiry (Peet and Watts 1996; Bryant and Bailey 1997). These areas of inquiry are only generally similar since different scholars have adopted different approaches to the same issues. Political ecologists have sought to explain Third World environmental change and conflict in terms of key environmental problems, concepts, socioeconomic characteristics, actors and regions, or they have used various combinations of these approaches (Bryant and Bailey 1997). Political ecology is in part based on the assumptions and ideas of political economy theory (Bryant and Bailey 1997). Blackie and Brookfield (1987) stated that political ecology considers ecology concerns within a broadly defined political economy. In general political ecologists agree on two basic points: first, the environmental forces facing the Third World are not simply a reflection of policy or market failures, but rather are a manifestation of broader political and economic forces associated with the worldwide spread of capitalism. Second, there is a need for changes to local, regional and global political-economic processes (Peet and Watts 1996).

PAGE 47

34 Political ecology addresses the political, economic, and cultural factors underlying human use of natural resources and the complex interrelations among people and groups at different scales, from local to global (Blaikie and Brookfield 1987; Schmink and Wood 1987). Elemental political issues of structural relations of power and domination over environmental resources have been seen by a variety of scholars as critical to understanding the relationship of social, political, and environmental processes (Scoones 1999). The view of resources as socially and politically constructed has been central to this discussion and has resulted in important work on how perspectives in environmental change must be gauged from the view points of different actors (Blaikie 1995). The perception of an unequal relationship between politics and ecology explains in part the fact that political ecologist tend to favor consideration of the political over the ecological (also because of the social science background of most political ecologists). But they should not overlook advances in the understanding of ecological processes derived from the New Ecology since, in doing so they might miss an important part of the explanation of human-environmental interaction (Scoones 1999; Bryant and Bailey 1997). According to Bryant and Bailey (1997) there are five main approaches or similar areas of inquiry in Third World Political Ecology, although many times they are combined. These approaches are the following: In the first approach the explanation centers around a specific environmental problem or set of problems such as soil erosion, tropical deforestation, water pollution or land degradation. This approach constitutes in many respects a traditional geographical

PAGE 48

35 research theme associated with understanding the human impact on the physical environment (Goudie 1993), but with a distinctive political-economy twist. The second approach focuses on a concept that is perceived as having important links to political-ecology questions. To understand the latter is partly to appreciate the way in which ideas are developed and understood by different actors, and how attendant discourses are developed to facilitate or block the promotion of a specific actors interest (Escobar 1996). Third, inter-linked political and ecological problems are examined within the context of a specific geographical region. Regional political ecology has reflected a concern to take into account environmental variability and the spatial variations in resilience and sensitivity of the land, as well as theories of regional growth or decline (Blaikie and Brookfield 1987). Fourth, political-ecological questions are explored in light of socio-economic characteristics such as class, ethnicity or gender. A final focus is on the interests, characteristics and actions of different types of actors in understanding political-ecology conflicts. An actor-oriented approach seeks to understand such conflicts (cooperation too) as an outcome of the interaction of different actors pursuing often quite distinctive aims and interests (Long and Long 1992). The present research will use a combination of the first and third approaches, focusing on a particular environmental problem (deforestation) from a regional perspective. It will center on land use and land cover outcomes driven by market, infrastructure and credit availability.

PAGE 49

36 Household Demography The theoretical foundation of the role of household life cycle in land use was established by Chayanov (1966). His study of peasant farming practices in Russia, in the first half of last century, serves as a reference mainly because the October Revolution in Russia created conditions of land abundance similar to agricultural frontiers in the Amazon region (Walker et al. 2002). His theory explains differences in farm size and surplus production in relation to household structure. Chayanov distinguished households according to the ratio of consumers/workers. This relation led him to describe the household life cycle, where young households with many children have low labor power, and mature households with high labor power have larger holdings (Walker et al. 2002; Perz 2001). The basis of Chayanovs theory is that the drudgery of labor increases exponentially as work is done, while on the other hand the marginal utility of goods decreases as they are acquired; the household production level determined by the intersection of these curves. Marginal utility is determined by the standard of living, which consists of: the amount necessary to support one consumer; the number of consumers each worker has to support (the consumer/worker ratio); the amount that has to be reinvested in the farm to maintain its production; and any other factors that require part of the farms production. Drudgery is a measure of the noxiousness of labor and is inversely related to productivity. The more productive a technique is, the greater is the output of per unit of labor, and the lower the drudgery (Tannenbaum, 1984). While Chayanovs theory provides the basis for demographic theory related to land use, some assumptions about peasants do not apply directly to Amazonia. The theory does not address the issue of migration, an important feature in frontier Amazonia where

PAGE 50

37 many peasants migrated from places with a different landscape. It assumes closed household life cycles, while in Amazonia households are not always detached from labor or products markets, but rather have different assets. Chayanov assumed the existence of relatively homogeneous farming practices among households, something that doesnt necessarily happen in Amazonia where, besides farming, cattle ranching and forestry also are land use options (Perz 2001). At this point it becomes important to make a distinction among household assets. Ellis (Ellis 2000) considers household assets are resources owned, controlled or claimed by the household. These assets mediate the way in which households become involved in production and labor markets, and participate in exchanges within their community. Assets can be understood as capital or resource stocks that may be used for household survival. In a general way, assets are divided in five classes. Natural capital comprises land and water. Physical capitals are buildings, machines and roads. Human capital is the labor available, including education, skills, and health. Financial capital is money as savings and/or credits, while social capital is kinship and community networks (Ellis 2000). Household demography has its focus on human capital assets; however, according to the specific case other assets should be considered. For the purposes of this study some aspects of natural, physical and financial capitals will be included. Panarchy Theory Theories like Panarchy intend to explain not what is, but what might be. They will not predict the details of future possibilities, but might help to identify the conditions for future possibilities (Holling et al. 2002a). Their objective is to enable the understanding of economic, ecological and institutional systems and their interactions. The cross-scale,

PAGE 51

38 interdisciplinary, and dynamic nature of the theory gives it its name. Its essential focus is to rationalize the interplay between change and persistence, between the predictable and unpredictable (Holling et al. 2002a). According to this theory, the stabilization of target variables leads to slow change in other ecological, social, and cultural components, and those changes may lead to the collapse of the entire system (Holling et al. 2002a). Decline in variability and diversity creates conditions that cause a system to flip into an irreversible (typically degraded) state controlled by unfamiliar processes. The magnitude of disturbance that can be absorbed before the system changes its structure by changing the variables and processes that control behavior is named ecosystem resilience (Holling and Gunderson 2002). According to Holling and Gunderson (2002) resilience has three defining characteristics: The first is the amount of change a system can undergo (and, therefore, the amount of stress it can sustain) and still retain the same controls on function and structure (still be in the same configurationwithin the same domain of attraction). The second is the degree to which the system is capable of self-organization. When managers control certain variables in a system, they create inter-variable feedbacks that would not be there without their intervention. The more "self-organizing" the system, the fewer feedbacks need to be introduced by managers. And third, is the degree to which the system expresses capacity for learning and adaptation. Semi-autonomous levels are formed from the interactions among a set of variables that share similar speeds. The organizations and functions we now see embracing biological, ecological, and human systems are therefore ones that contain a nested set of the four-phase adaptive cycles, in which opportunities for periodic reshuffling within

PAGE 52

39 levels maintain adaptive opportunity, and the simple interactions across levels maintain integrity (Holling et al. 2002a). Integrated Frameworks In this section I will try to integrate three frameworks from the three integrative theories we reviewed in the preceding section. The frameworks are the following: 1. The conceptual framework that uses a three-tiered hierarchical approach to depict the socioeconomic and biophysical drivers that led to deforestation, elaborated by Wood (2002) which comes from a Political Ecology view. 2. A conceptual framework of household transformations, land use and environmental change (McCracken et al. 2002), which comes from Demographic theory 3. The Adaptive cycle elaborated by Holling and Gunderson (Holling and Gunderson 2002) which is within Panarchy theory. The Three-Tired Hierarchical Approach This framework treats land cover outcomes as the direct effect of the land use decisions made by rural households whose decisions are embedded in contexts that operate at higher levels of the system. The higher level contexts consist of the proximate, intermediate and distant drivers that comprise the socioeconomic and biophysical subsystems. The analytical focus is on the relationships that take place within each level, as well as the cross-level dynamics that link one level to another (Wood 2002). This model considers socioeconomic, as well as biophysical drivers. To depict the driver forces hierarchy, it presents the proximate, intermediate, and distant scales. The model assumes that land use decisions made by firms and households in rural areas are the result of interactions of a large number of variables acting at different scales within the social and natural system. They are located at the center and within community and kinship networks, meaning that networks at the local level may influence land use

PAGE 53

40 decisions. The model also considers the feedback effect that land cover may have in the socioeconomic and biophysical drivers (Wood 2002) (Figure 2-1). Each land use/land cover outcome is associated with different kinds of economic activity, and therefore with different social groups. Rubber tappers, farmers, ranchers and loggers all engage in clearing the forest cover, but they do so in varying degrees depending on their respective objectives, resources, and decisions (Wood 2002). Outcomes can be arranged in five main categories: undisturbed forest, harvest of non-timber products, selectively logged forest, cleared (annual crops, perennial crops, pasture, mining), and regrowth (managed fallow, abandoned plots). Household Transformations Land Use And Environmental Change This approach is based on Chayanovian theory and on the work of the Centro Agro-Ambiental do Tocantins (1992 cited by Walker et al. 2002), Walker and Homma (1996) and McCraken et al.(1999). It emphasizes the role of household labor in land use decisions in agricultural frontiers (McCracken et al. 2002; Perz 2002b; Brondizio et al. 2002) It is seen as a complement, not an alternative, to models focusing on environmental and economic factors, like the different drivers presented by Wood (2002). It was Walker and Homma (1996) who placed households in a context of labor and product markets, capital availability, and land use differentiation in Amazonia (Perz 2001). According to this framework, as the household looks for its consolidation, different stages of land use are linked to different household life cycle stages.

PAGE 54

41 Figure 2-1. The three-tiered hierarchical approach. Socioeconomic and biophysical drivers of land use are classified in: distant, intermediate and proximate. Land cover outcomes are direct effect of the land use decisions made by rural households and firms. Feedback intensity is represented by the varying thickness of return arrows.

PAGE 55

42 Recent settler families in a frontier are assumed to be small young nuclear households, with a head couple and a few young children. Due to small requirements in land and capital, and the low level of risk, they first clear small areas of forest to cultivate annual crops (Perz 2001). These are mainly for consumption and local markets. As the family grows in age and size, additional site knowledge increases and more household labor is available, more areas are cleared, and previous plots are left uncultivated, formed into pasture, or planted in perennial crops, causing a decline in deforestation rates (McCracken et al. 2002). This change is slow, and involves high initial capital and labor cost. Economic gains from cattle and perennial crops will be perceived in future years. While perennial crops in general will not provide any returns for three to five years, acquiring cattle may be an important capital-saving strategy because it can be quickly purchased or sold. Perennial crops are also more labor intensive than cattle raising, one reason why older households with less labor usually shift into pastures (Perz 2001; McCracken et al. 2002). Access to resources like good soil, water, capital / credit, markets, technical support and household labor affect the shift to either perennial crops or raising cattle, or the decision to remain in annual cash crop activities. It is assumed that in the first household stages most families exhaust their initial capital reserves. Thus, labor of adolescent and teenage children may be a determining factor along with credit possibilities for furthering farm investments (McCracken et al. 2002; Brondizio et al. 2002). Finally, as Perz (2001) indicates, an important final land use is reforestation. When children become adults, at the point of inheriting their parents land the family may plant

PAGE 56

43 trees for long-term timber production. Reforestation initially requires capital and labor, but after the establishment of the plantation, little attention is required. The schematic representation of McCracken et al. (2002) presented in Figure 2-2 defines five stages of a household life cycle, each one linked to a particular time of residence in the area, demographic composition, and land use practices. Early farm consolidation is associated with credit, capital, and large supply of household labor. Troubles in farm consolidation are associated to reliance on annual crops and restricted supply of household labor (McCracken et al. 2002). The Adaptive Cycle In case examples of regional development and ecosystem management, it has been found that three properties seemed to shape the future responses of the ecosystem, agencies, and people (Gunderson et al. 1995; Holling and Gunderson 2002): 1. the potential available for change, since that determines the range of possible options; 2. the degree of connectedness between internal controlling variables and processes, a measure that reflects the degree of flexibility or rigidity of such controls, 3. and the resilience of the systems, as a measure of their vulnerability to unexpected or unpredictable shocks. Potential, connectedness and resilience The framework is partly based on the traditional view of ecosystem succession seen as being controlled by two functions. The first is exploitation, in which rapid colonization of recently disturbed areas is emphasized. The second is conservation, in which slow accumulation and storage of energy and material are emphasized. In ecology, species of the exploitive phase are named r-strategists and in the conservation phase k-strategists (Holling and Gunderson 2002).

PAGE 57

44 Figure 2-2. Household transformations, land use and environmental change. This framework highlights the role of household labor over the domestic life course. In the upper section it suggests a pattern of land use. The thickness of each line represents the level of activity for each land use. Land use stages (x-axis) are linked to different household stages (y-axis). The diagonal from the upper left to the lower right represents a general course of farm formation and domestic life. Deviations to the right are associated to early farm consolidation linked to credit, capital and larger supply of household labor. Deviations downward are associated with difficulties in farm consolidation linked to greater reliance on annual crops and a restricted supply of labor.

PAGE 58

45 Later understanding in ecology indicates that two additional functions are needed. The first is release or creative destruction, in which the tight-bound accumulation of biomass and nutrients becomes increasingly fragile until suddenly released. This is named the omega () phase. The second function is reorganization, in which the remaining elements after the omega () phase are rearranged; this is known as the alpha () phase (Holling and Gunderson 2002). During this cycle, as shown in Figure 4, biological time flows unevenly. The progression in the ecosystem cycle proceeds from exploitation (r), slowly to conservation (K), very rapidly to release (), rapidly to reorganization (), and rapidly back to exploitation (Holling and Gunderson 2002). The cycle reflects changes in two properties: (1) Y-axis: the potential that is inherent in the accumulated capital of biomass and nutrients. (2) X axis: the degree of connectedness among variables (see figure 3). As the system goes from exploitation to conservation, connectedness and potential increase (Holling and Gunderson 2002). For resilience, as the system goes from exploitation to conservation, resilience shrinks; and it expands as the system goes into reorganization. Hierarchies and panarchies There are many possible interactions among phases at one level and phases at another level. Two are considered specially important. These are the connections named Revolt and Remember; these connections become important at times of change in the adaptive cycles (Holling et al. 2002b).

PAGE 59

46 Figure 2-3. The adaptive cycle, the four ecosystem functions (r, K, ,) and the flow of events among them. The arrows show the speed of that flow in the cycle, where short, closely spaced arrows indicate a slowly changing situation and long arrows indicate a rapidly changing situation. The cycle reflects changes in two properties, (1) Y-axis: the potential that is inherent in the accumulated capital of biomass and nutrients. (2) X axis: the degree of connectedness among variables. The exit from the cycle indicated at the left of the figure suggests the stage where the potential can leak away and where a flip is most likely into a different system. When a level in the panarchy enters its omega phase and experiences a collapse, that collapse can cascade up to the next level by triggering a crisis, particularly if that (higher) level is at the K phase where resilience is low. This is termed Revolt. When a level enters its omega phase, the opportunities and constraints for the renewal of the cycle are strongly organized by the K phase of the next higher and slower level. This is termed Remember (Holling et al. 2002b).

PAGE 60

47 There are two distinctions between panarchy representation and traditional hierarchies. The first is the importance of the adaptive cycle, and the alpha phase as the engine of variety and generator of new experiments within each level. The second is the connections between levels. The levels of a panarchy could therefore be drawn as a nested set of adaptive cycles (Holling et al. 2002b). Framework Integration Now that the information on the theories and on the specific frameworks we are interested in has been reviewed the next step is to bring them together in order integrate the three frameworks. In that way it may be possible to incorporate the different levels of land use drivers proposed by the three-tiered hierarchical approach, the household dynamics from the household transformations approach, and finally to better understand the spatial and temporal interactions by adapting the panarchy approach. Figure 2-4 provides an initial vision. In Figure 2-4 there are no arrows linking the three-tiered hierarchical framework to the panarchy framework; there is, however, a direct link. The adaptive cycle may be applied to all hierarchical levels (household, local, proximate, intermediate and distant) to make up a nested set of adaptive cycles, a panarchy. Looking at Figure 2-5 one can imagine having adaptive cycles at each level and for each land use driver. However, in this chapter the integration will be centered at the household and land use activity levels. This is mainly for practical and methodological reasons, since information at these faster levels may be gathered in an easier and more direct way, and examples will be easier to understand. The logic remains the same for the higher levels and will be applied in the next chapter when explaining the results from data analysis.

PAGE 61

48 2) Household transformations, land use and environmental change. The thickness of each line represents the level of activity for each land use. Land use stages (x-axis) are linked to different household stages (y-axis). The diagonal from the upper left to the lower right represents a general course ofarm formation and domestic life. Figure 2-4. f 1) The three-tiered hierarchical approach, socioeconomic and biophysical drivers of land use are classified in: distant, intermediate and proximate. Lan cover outcomes are direct effect of the land use decisions made by rural households and firms. Feedback intensity is represented by the varying thickness of return arrows. Fi g ure 2-4. Framework inte g ration the three-tiered hierarchical a pp roach household transformations a pp roach and the p anarch y a pp roach. 3) The adaptive cycle, a schematic representation of two levels in a Panarchy. Each level or structural element experiences its own adaptivecycle dynamics, within the constraintsof the broader, slower hierarchical levels. d

PAGE 62

49 Figure 2-5. Levels of interaction in a nested set of adaptive cycles. In the white area the bigger cycle represents the household and the smaller cycle represents the land use activities. Similar cycles exist at the local, regional national and global levels. Looking at the household and its activities as adaptive cycles The most obvious link between the three-tiered hierarchical approach proposed by Wood (2002) and the household transformations through time framework proposed by McCracken et al. (2002) is, of course, the household (as shown in Figure 2-4), our center of analysis. While the first framework classifies the land use drivers that are external to the household, the second framework classifies the household according to five stages in its life cycle according to its effect on land use decisions. In that way one can imagine that proximate, intermediate and distant socioeconomic and biophysical drivers of land use may have a different effect depending on the household life cycle stage.

PAGE 63

50 This effect is expressed in different land use activities. At the same time, but at a faster pace, each productive activity has its own transformations through time. In order to make this idea clear we can look at the households land use activities as an adaptive cycle. Lets take the example of a cash crop in the Amazonia (see Figure 2-6). The cycle may be seen as follows: 4. r: cash crops establishment 5. k: crops are established and producing 6. : due to soil depletion or a fall in market prices, crops are not in production any more 7. : depending on the cause for the phase, and on household transformations the options may be; a. change to different cash crops (reorganization within the same cycle), b. change the farm place (reorganization into a different cycle) or, c. change its productive activity (reorganization into a different cycle) d. mix farming with other productive activities (special case, as we will see later). In the same way, we can look at the household itself as an adaptive cycle. Returning to the schematic representation of McCracken et al. (2002) presented in Figure 2-2; it defines five stages of a household life cycle, each one linked to a particular time of residence in the area, demographic composition and land use practices. We may say that the cycle is as follows (see Figure 2-6): 1. r: this is represented by stage I, when a young couple with small children arrive to a frontier area in Amazonia. 2. k: this will be stages II, III, and IV. The family grows from having adolescent children to have adult children. At the end all household members are available labor force.

PAGE 64

51 3. : this stage in the cycle corresponds to stage V. The original family turns into multigenerational or second generational households. Small children are present again. 4. : depending on the assets available, the second generation young family may: a. remain in the original farm place (reorganization within the same cycle), b. move to start their own farm (reorganization into a different cycle) or, c. change from farming to another productive activity (reorganization into a different cycle) d. mix the farm with wage activities. Figure 2-6. Looking at the household and its land use activities as adaptive cycles. the bigger cycle represents the household and its four stages according to the household transformations approach. The smaller cycle represents one of the household land use activities (annual crops). The information in the background represents the other levels from the three-tiered hierarchical approach

PAGE 65

52 Nonlinear effect of land use drivers This chapter has provided an integrated framework to be used in the present research. And it has also provided a strong basis for the formulation of the second hypothesis: H2: Land use drivers do not have a linear influence on households, and land use systems; instead, much depends on the phase they occupy in their life cycle. Now that we see the household and its activities as adaptive cycles, it should be clear that in general the household has a slower adaptive cycle than its agricultural land use activities. Taking this figure further, the complex interplay can be seen as the interaction of the household adaptive cycle with other adaptive cycles within a nested set of the four-phase adaptive cycles (see Figure 2-6). Now I will try to engage with the next slower adaptive cycle, the households town; in our example this is Iapari and Assis Brazil. At this scale we have direct effects of the proximate and intermediate land use drivers. Every driver has its own adaptive cycle, so we will have as many adaptive cycles as drivers of land use determined for this particular town; of interest to us are markets, credit, and road infrastructure. Panarchy theory considers potentially multiple connections between phases at one level and phases at another level, but as explained earlier, Revolt and Remember connections become important at times of change in the adaptive cycles (Holling, Gunderson & Peterson 2002). When the soil cycle enters its omega phase, and experiences a collapse, that collapse can cascade up to the household cycle by triggering a crisis, particularly if it is at the K phase (stage IV in household lifecycle) where resilience is low: that is called revolt. The same thing happens when the household enters its omega phase and experiences a collapse; that collapse can cascade up to the proximate

PAGE 66

53 driver cycle by triggering a crisis, particularly if it is at the K phase where resilience is low. When the household enters its alpha phase, the opportunities and constraints for the renewal of the cycle are strongly organized by the K phase of the proximate driver cycle, which is called remember. Considerations Regarding the Framework The main objective of this section was the development of a framework to analyze land use in Amazonia through the integration of three frameworks. The integration allows us to see the different levels that may affect land use decisions and its interactions: distant, intermediate and proximate drivers, household assets, and its land use activities (productive systems). While the adaptive cycle may be applied to all levels, it is applied in this study at the household and its land use activities level. A main consideration should be mentioned: the time line. Ideally information taken in at least two different points in time is desirable. In that way more accurate cycles may be described. However it was possible to explore the issue with a single time data collection by asking what future decisions regarding land use will be taken in the face of anticipated changes that may include road infrastructure, and credits. Questions also probed about key changes in land use and land cover in the past. Work needs to be done in regard to some more specific issues. For example, some of the resources that are considered to be household assets in one framework are considered as proximate drivers in the three-tiered hierarchical approach (e.g. access to credit). A second consideration is that we focused on only two of the many different levels of interaction in the complex processes of land use decisions. The role played by

PAGE 67

54 drivers/assets will vary according to different conditions, becoming more or less important in different cases. The nature of land use drivers is also an important fact to be considered; for example, changes in markets at the national level may have an immediate effect on household land use decisions, while changes in policy at the national level may not have a direct effect, usually because environmental laws are poorly implemented. This is an important consideration since it appears to be a contradiction to the panarchy framework where cycles at the same level are considered to have the same speed. Another important consideration is the nature of the land use activity. From the example that was given in this chapter on agriculture it is obvious that annual crops have faster cycles than households; however, other agricultural systems may have slower cycles, like vineyards or other long lived perennial crops. That will also be the case for rubber tapping and Brazil nuts.

PAGE 68

CHAPTER 3 LAND USE AND LAND COVER Introduction In methodological terms, the research conducted in Assis Brazil and in Iapari was a natural experiment. It was not an experiment conducted by a researcher; it was evaluated through research (Bernard 2002). The two former chapters provided a strong base for establishing the hypotheses to be tested: H1: Access to markets, credit and road infrastructure drove more deforestation in Assis Brazil than in Iapari. H2: Land use drivers do not have a linear influence on households, and land use systems; instead, much depends on the phase they occupy in their life cycle The present chapter deals with the methodology followed to gather the data in the field and its analysis. Data gathering included two different steps followed in Iapari and in Assis Brazil, first interviews and then questionnaires. Analysis of both qualitative and quantitative data was performed in order to test the hypotheses set out in the first two chapters. Data analysis included four steps: first, the operationalization of the variables and the presentation of descriptive statistics; second, the comparison of the means for the data found in Assis Brazil and in Iapari; third, correlation analysis to observe the relationships among variables; and fourth, multivariate modeling to observe the role of the group of independent variables in determining land uses and land covers. 55

PAGE 69

56 Fieldwork Methods in Iapari and Assis Brazil The methodology was designed to be the same in both sites. However, during fieldwork the different conditions made it difficult to follow the same exact steps in both towns. There were four weeks of intense fieldwork in each place. The work included interviews with local authorities and 45 questionnaires with small farmers. In both cases indigenous peoples and extractive reserves were purposely excluded from in the research since for these populations the collective ownership of resources adds a dimension to land use decisions that is not considered in this research. Fieldwork in Iapari The primary field work in Iapari started on June 13 th and finished on July 14 th 2004. The district is divided in 5 main sectors: Iapari, La Colonia, Nueva Esperanza, Villa Primavera, and San Isidro de Chilina. The latter two were part of the government colonization projects. There is also one native community, Belgica, which is not being considered in this research. Iapari and La Colonia are urban centers, while most of the farm lands are in Primavera, Nueva Esperanza and Chilina along the 50km road to Iberia. Given the spatial location of Chilina, it is more linked to Iberia than to Iapari. Health care, agriculture, and INRENA offices in Iberia attend Chilinas needs. From 1995 to 1997 the Special Project for Land Titulation (Proyecto Especial de Titulacin de Tierras PETT) worked in the district. For this reason at least 85% of farmlands now have titles. During the first week in Iapari the main activities were the interviews; these included: the Alcalde Provincial del Tahuamanu, INRENA (National Institute for Natural resources) manager, SENASA (National Service for Agrarian Health) manager, APEMI (Iapari small loggers association), managers of the three forest concessions of the district, presidents of the farmers and cattle ranchers association, president of the

PAGE 70

57 mothers club, and the secondary school director. One key person was not possible to interview, the manager at the Agriculture office. Statistical data and existing maps, however, were made available for my use. Available information was not good enough to establish the number of farmers in the district, nor the spatial location of farmlands. The second week was spent completing the interviews and deciding on the sampling method to apply the questionnaires. I decided to use the local health post census data, gathered in December 2002, which provided accurate data on population and covered most of the district except Chilina. The number of households in each sector was established (estimated for Chilina) and a proportional number of randomly chosen households from each sector was assigned to complete 50 households, 25% of the households that had a farm reported for Iapari in 1994 (INEI 1999). The third and fourth weeks were spent applying questionnaires (Appendix A); 45 questionnaires were carried out but only 36 were valid for this research 1 representing 18.3% of the district households that had a farm (INEI 1999). I started in Iapari and La Colonia, the urban areas where most families were concentrated (78%). Appointments were made with family heads, since most work either at their farms or at the Municipalidad, the main source of employment, during the day. The fourth week was spent visiting Chilina (the farthest sector), Primavera, and Nueva Esperanza. See figure 3-1. 1 Questionnaires for households with more than two land holdings, with land holdings that had no productive activities and those who could not provide accurate information were considered invalid

PAGE 71

58 Figure 3-1. Household farms visited in the Municipio of Assis Brazil and in the District of Iapari. While all questionnaires were made on the farm in the case of Assis Brazil, only some of them were made on the farm in the case of Iapari. The Acre River that makes the international border is highlighted in blue.

PAGE 72

59 Fieldwork in Assis Brazil In Assis Brazil the main work was done from July 15 th to August 14 th Four sectors were identified in the area: Paragua, Santa Quitria, So Francisco, and Assis Brazil. The boundary between the Municpios of Assis Brazil and Brasielia is at km 8 on the Assis Brazil-Brasieleia road. These means that there are only 8km of paved road within Assis Brazil and, therefore, most farms in this Municpio are accessed with secondary roads. Most farmers do not have land titles; only in Santa Quitria where INCRA had established Colonization Projects do farmers have land titles. The first week, I presented my research proposal in a meeting of the Municipal staff. I had the opportunity to present myself and the research topic, and to get feedback and recommendations. Two interviews were carried out, one with the person in charge of the IBAMA office and one with the person in charge of SEATER (Executive Secretary for Technical Assistance, Rural Extension and Production Warranty). A collaborative relation was established with SEATER, an organization which works directly with farmers associations. Since the government only supports associations, most farmers are part of one. Seven associations were identified in the area: Bacia, Livramento and Estrela Brilhante in Paragua, So Felix and Fortaleza in Santa Quitria; Novo Progresso and Iracema in So Francisco. Farmers that live along the secondary roads known as Beija Flor, Recife, Do Sete and in the main road near to Assis Brazil were included. The number of households in each association was established (estimated for Beija Flor, Recife, Do Sete) and a proportional number of randomly chosen households from each sector was assigned to complete 50 households, which represents 23.7% of the households that had a farm reported for the Municipio of Assis Brazil in 1995/1996 (IBGE 1998).

PAGE 73

60 During the second, third, and fourth week a total of 45 questionnaires (Appendix B) were carried out and 41 were valid 2 representing 19.4% of the households that had a farm (IBGE 1998);, see Figure 3-1 for spatial distribution. A program was established to visit each farm with a technician and transportation provided through SEATER. We used a motorcycle for our transportation through secondary roads 8-15 km long; all of them depart around km 4 and 7 from the recently paved Pacific highway (BR-317). Most of them were passable only with a motorcycle. First we visited So Felix, a Colonization Project of INCRA, then we visited Bacia, Iracema and Recife; the last one was very difficult to transit, even in August when it is the dry season. I also took a short trip up the Acre River to visit Novo Progresso. Finally, we visited Estrela Brilhante and Livramento. Fortaleza was not visited due to transportation issues. Then we applied questionnaires to farmers in Beija Flor and in the closer areas like the secondary road known as Sete and in the recently paved BR-317. The Differences in Methodology and Their Implications Interviews were an important component in the case of Iapari, where two weeks were spent in this activity and more than fifteen interviews were carried out with local authorities and leaders. In fact, in theory, this will allow for a better contextualization of the responses obtained in the questionnaires. It will also allow for a comparison on the authorities and leaders perceptions with those of the farmers. In the case of Assis Brazil, only two interviews were carried out; however, I expect this not to be a major problem, since those interviewed were key persons: IBAMA and SEATER personnel. 2 Questionnaires for households with more than two land holdings, with land holdings that had no productive activities and those who could not provide accurate information were considered invalid

PAGE 74

61 In the case of questionnaires, although the number was the same in both places (45), there were some differences. For Iapari 36 questionnaires were valid and were used in this research, representing 18.3% of the district households that had a farm (INEI 1999) in that area. In Assis Brazil 41 questionnaires were valid, representing 19.4% of the Municpio households that had a farm (IBGE 1998). This poses questions in terms of comparisons and statistical analysis. However, although the total number of valid questionnaires is different, they represent a very similar percentage of the total number of farmers in each site. Operationalization of Variables Five groups of variables have been developed for the present research. The first group is labeled land use outcomes; the variables in this group are indicators for household land use activities. The second group was labeled land cover outcomes; these variables are indicators for deforestation since arrival to the farm. The third group is labeled background information; these variables are the control variables. The fourth group, labeled land use drivers, is composed of two sections: markets and credit, and road infrastructure variables. Finally, the fifth group is labeled household life cycles; these variables account for household level variance. The variable place that represents whether a household is located in Iapari or in Assis Brazil is also included. It is important to place the hypotheses and framework for the present research within the context of Assis Brazil and Iapari. Market variables are at the local and regional level, credit variables are at the national level, and road infrastructure variables are at the national, regional and local level. Household life cycle variables are at the household level, the same as land use, land cover, and control variables.

PAGE 75

62 In order to address the central research questions, several data analysis steps will follow. First, descriptive statistics for both Iapari and Assis Brazil will be presented in order to provide a general idea of the characteristics for the whole area. Second, means comparison of each variable for Iapari and Assis Brazil will be presented in order to reveal the differences for each town. Third, in order to observe the relation between dependent variables and dependent (outcome) and independent variables, bivariate correlations will be presented. Fourth, to gain insights on how each group of variables interacts and affects the outcome variables, multivariate models between each outcome variable and each group of independent variables will be presented. Fifth, multivariate models for each outcome variable will be developed by using the independent variables found to be significant in the previous steps. Tables 3-1to 3-4 present descriptive statistics for the dependent and independent variables used in this study. Data for variables with skewness over 1 were transformed by converting to the natural logarithm and adding 1 unit to avoid 0 values. This procedure reduced overall skewness and improved normality for statistical analyses. Mean values are presented for the raw data, the transformed data and for the antilog of the previously transformed data. Skewness is presented for both the raw and transformed data. The sample size of 77 households is the same for all variables. Table 3-1 presents descriptive statistics for the dependent variables: land use outcomes and land cover outcomes, for Iapari and Assis Brazil in the year 2003. Four land use outcomes and four land cover outcomes are considered. Land use outcomes were reported as hectares (ha) in annual crops, perennial crops and pasture. The number of heads of cattle is also included since some farmers have pasture and no cattle. The area in

PAGE 76

63 annuals includes rice, corn, beans, manioc, other vegetables (including tomatoes, lettuce, herbs, spinach, and others) and the various combinations. Also included in this category are hectares of corn mixed with pasture and with bananas. The area in hectares under perennials includes banana, citrus, pepper, palillo, pijuayo, aai, coffee, and their various combinations. The combinations usually include fruit trees, palms, and timber trees. Table 3-1. Descriptive statistics for land use and land cover outcome variables. Iapari and Assis Brazil, 2003. Unit Mean Mean* Std. Deviation Skewness N Variable (1) (2) (3) (4) (5) Land use outcomes Annual crops ha 2.86 2.05 0.79 77 Perennial crops ha 0.75 0.92 2.53 77 Pasture ha 16.85 23.25 3.30 77 Heads of cattle 28.78 63.19 5.85 77 Land cover outcomes Old growth forest ha 55.19 53.68 2.79 77 Secondary forest ha 8.84 11.84 2.82 77 Deforested area (a) ha 20.15 23.96 2.51 77 % deforested of forest (b) % 25.93 25.10 1.11 77 Transformed values ln (1+var) Land use outcomes Annual crops 1.20 1.22 0.57 -0.40 77 Perennial crops 0.75 0.78 0.30 -1.10 77 Pasture 2.21 3.37 1.25 -0.26 77 Heads of Cattle 2.16 3.20 1.71 0.00 77 Land cover outcomes Old growth forest 3.67 14.39 0.97 -1.42 77 Secondary forest 1.75 2.12 1.06 0.01 77 Deforested area 2.39 4.02 1.30 -0.44 77 % deforested of forest 2.68 5.36 1.33 -0.77 77 *antilog of mean logs, (a) Hectares deforested since arrival to the farm (initial old growth forest old growth forest), (b) % forest cleared since arrival (ha deforested x 100 / ha initial old growth forest). Land cover outcomes were reported in ha of old growth forest, secondary forest, deforested area, and percentage of cleared forest. The deforested area was calculated by subtracting the current area of forests from the initial (that which was existing when the

PAGE 77

64 family arrived to the farm) area in hectares of forest. The percentage of cleared forest was calculated by dividing the total area deforested by the initial area of forest. The high standard deviation for the number of heads of cattle responds to the inclusion of one medium size rancher household from Assis Brazil in the dataset. While most households had less than 100 heads of cattle this household had 500. The data were included in the analysis since they are representative of a very small group of ranchers in Assis Brazil. For old growth forest, secondary forest, deforested area and % deforested land, antilog transformed mean data was found to differ substantially from non-transformed mean data. Table 3-2 presents descriptive statistics for household background information. This category presents seven variables. The ones measured in hectares (ha) are land size, initial old growth forest and initial secondary forests. Household sources of off-farm income like regular monthly income (e.g. wage, retirement), daily wage or other irregular income sources (e.g. taxi driver) are reported through yes/no answers. The region of birth of the household head indicates whether they were born in the Madre de Dios-Acre-Pando region or elsewhere. The number of years the head of household has received formal education are also provided. For the yes/no answers the mean value gives the percentage of yes answers. The mean area in old growth and secondary forest was 75 and 6 ha respectively. Almost half of the families possessed a regular monthly income and almost one third had a daily wage. More than half of the heads of households were born in the MAP region and household heads had an average of 5 years of education.

PAGE 78

65 Table 3-2. Descriptive statistics for household background information variables. Iapari and Assis Brazil, 2003 Unit Mean Mean* Std. Deviation Skewness N Variable (1) (2) (3) (4) (5) Background information Farm size ha 83.51 59.47 2.3 77 Initial old growth forest ha 75.34 60.16 2.21 77 Initial secondary forest ha 5.95 11.90 3.64 77 Regular monthly income 0=no, 1=yes 0.40 0.49 0.41 77 Daily wage 0=no, 1=yes 0.27 0.45 1.04 77 Born in the MAP area 0=no, 1=yes 0.66 0.48 -0.70 77 Education 5.43 4.44 0.65 77 Transformed values ln (1+var) Farm size 4.27 26.31 0.56 0.64 77 Initial old growth forest 4.07 21.48 0.82 -1.55 77 Initial secondary forest 1.06 1.07 1.25 0.76 77 *antilog of mean logs Table 3-3 provides descriptive statistics for place, markets, credit, and road infrastructure. Place refers to whether the household is located in Assis Brazil or in Iapari. The six market variables are: distance to the nearest market, whether the household sells annual crops, perennial crops, small animals or cattle, and an index of farm product commoditization. This index assigns a value to each one of the different combinations of products sold. It ranges from 1 to 13; the lowest values are assigned to households that sell annuals and small animals and the highest to households that sell cattle and perennials. Credit refers to the number of times the household had used credit since arriving to their farm. Road infrastructure groups six variables: whether the household is located in a main road, a secondary road, a tertiary road, or a walking path; the distance in kilometers from the main road, and a transportation time index. This index was created to combine road infrastructure variables, It was calculated by dividing the known distances (obtained from the interviews and tracked roads) to the markets of either Assis Brazil or Iapari by

PAGE 79

66 approximate average travel velocities for primary, secondary, and tertiary roads, and walking paths (45, 20, 15, and 4 kilometers per hour, respectively) and obtaining a total travel time by adding each section. For the yes/no answers the mean value gives the percentage of yes answers. Table 3-3. Descriptive statistics for place, markets, credit and road infrastructure variables. Iapari and Assis Brazil, 2003 Unit Mean Mean* Std. Deviation Skewness N Variable (1) (2) (3) (4) (5) Place (Iapari/Assis Brazil) 0=I, 1=A 1.53 0.502 -0.133 77 Market and credit Distance from nearest market km 12.41 6.28 0.33 77 Sells annual crops 0=no, 1=yes 0.70 0.46 -0.90 77 Sells perennial crops 0=no, 1=yes 0.13 0.34 2.25 77 Sells small animals 0=no, 1=yes 0.42 0.50 0.35 77 Sells Cattle 0=no, 1=yes 0.56 0.50 -0.24 77 Farm product commoditization index 6.61 4.79 -0.12 77 Times credit was received 1.06 1.49 2.68 77 Road Infrastructure Lives in main road 0=no, 1=yes 0.12 0.32 2.43 77 Lives in secondary road 0=no, 1=yes 0.34 0.48 0.70 77 Lives in tertiary road 0=no, 1=yes 0.29 0.46 0.97 77 Lives in walking path 0=no, 1=yes 0.26 0.44 1.12 77 Distance from main road km 5.59 4.62 0.33 77 Transportation time hours 0.60 0.38 1.10 77 Transformed values ln (1+var) Times credit was received 0.55 0.64 0.56 0.77 77 Transportation time 0.45 0.57 0.22 0.46 77 antilog of mean logs The mean distance to nearest market was short (12.41 km) compared to other areas. More than two thirds of households sell annual crops and only a small portion of households sell perennial crops. As for animals, almost half of households sell small animals while more than half of households sell cattle. On average all households had received credit once in the past. Spatially a small percentage of households lives along a primary road while more that one third lives along

PAGE 80

67 a secondary road and almost one third lives along a tertiary path and along a walking respectively. The mean distance to the main road was 5.59 km while the mean transportation time, as obtained from the index, was 36 minutes. Table 3-4 provides descriptive statistics for the household life cycle; this category groups eight variables. The first is the time the household has lived on the farm (years on the farm), the age of household head, the number of family members currently living on the farm, the number of family members that participate in land use activities, the number of children in the family and the number of adults in the family. Additional indices are presented for labor hired and labor exchanged. Values range from one to four, one meaning no labor was exchanged or hired during the last 12 months; two, labor exchanged for reasons other than forest clearing; three, labor exchanged for forest clearing; and four, labor exchanged for 2 and 3. For the labor hired index, five represents labor hired all year round. Table 3-4. Descriptive statistics for household life cycle variables. Iapari and Assis Brazil, 2003. Unit Mean Mean* Std. Deviation Skewness N Variable (1) (2) (3) (4) (5) Years on farm 13.90 10.63 1.26 77 Age of household head 44.78 14.29 0.67 77 Family members on lot 4.55 1.88 0.58 77 Family members working farm 2.48 1.54 1.13 77 Number of children 1.70 1.57 0.87 77 Number of adults 4.18 2.89 1.34 77 Labor hired index 2.58 1.44 0.36 77 Labor exchanged index 2.06 1.23 0.57 77 Transformed values ln (1+var) Years on farm 2.43 4.16 0.80 -0.49 77 Family members working farm 1.15 1.16 0.47 -0.59 77 Number of adults 1.52 1.68 0.48 0.73 77 *antilog of mean logs

PAGE 81

68 The families had spent an average of 14 years on their plot by the summer, 2003, when this study was conducted. The mean head of the household was 44 years old. Though there were an average of 5 family members, only 2 were actively working on the farm. The number of children was half the number of adults when the definition of adult was older than 14 years of age. Families, on average, hired labor and exchanged labor for purposes other than land clearing, although there seems to be a trend in which labor is hired for land clearing purposes. Comparing Variable Means for Assis Brazil and Iapari Table 3-5 provides the results of independent t-tests for the means of land use and land cover outcomes, background information, markets, credit, infrastructure, and household life cycle variables between Iapari and Assis Brazil. Both raw and transformed data means are compared, but means are considered significantly different according to the transformed means. All means for land use outcomes were significantly different. Assis Brazil had a significantly larger area of annual and perennial crops and pasture, and nearly four times the heads of cattle than did Iapari, in the year 2003. Land cover outcomes means were not found to be different except the percentage of forest cleared since arrival to the farm, which is significantly less in Assis Brazil than in Iapari, mainly due to the larger average farms in Assis Brazil. Although not significant, there was an apparent trend of larger areas of old growth forest in Assis Brazil than Iapari.

PAGE 82

69 Table 3-5. T-test of means for land use outcomes, land cover outcomes, background information, markets and credit, road infrastructure and household life cycle variables according to location in Iapari or Assis Brazil, 2003. Means Variables unit Iapari Assis Brazil T (1) (2) (3) Land use outcomes Annual crops ha 2.38 3.27 -1.95+ Perennial crops ha 0.65 0.85 -0.95 Pasture ha 11.79 21.29 -1.82+ Heads of cattle count 13.31 42.37 -2.17 (a)* Land cover outcomes Old growth forest ha 45.69 63.54 -1.55 (a) Secondary forest ha 8.57 9.09 -0.19 Deforested area (b) ha 20.46 19.88 0.11 % deforested of forest (c) % 28.07 24.05 0.70 Background information Farm size ha 68.4 96.78 -2.25(a)* Initial old growth forest ha 66.15 83.42 -1.33 (a) Initial secondary forest ha 2.18 9.26 -2.86 (a)** Regular monthly income 0=no, 1=yes 0.31 0.49 -1.64 (a) Daily wage 0=no, 1=yes 0.42 0.15 2.69 (a)** Born in the MAP area 0=no, 1=yes 0.50 0.80 -2.90 (a)** Education years 7.75 3.39 4.91** Transformed values Land use outcomes Annual crops 1.06 1.33 -2.12* Perennial crops 0.67 0.82 -2.12* Pasture 1.89 2.50 -2.2* Heads of cattle 1.44 2.80 -3.79** Land cover outcomes Old growth forest 3.68 3.65 0.11 (a) Secondary forest 1.68 1.81 -0.57 Deforested area (b) 2.58 2.23 1.19 % deforested of forest (c) 2.96 2.43 1.82 (a)+ Background information Farm size 4.16 4.36 -1.60(a) Initial old growth forest 4.11 4.03 0.48 (a) Initial secondary forest 0.55 1.52 -3.77 (a)** Regular monthly income 0=no, 1=yes 0.31 0.49 -1.64 (a) Daily wage 0=no, 1=yes 0.42 0.15 2.69 (a)** Born in the MAP area 0=no, 1=yes 0.50 0.80 -2.90 (a)** Education years 7.75 3.39 4.91** + p < 0.1, p < 0.05, ** p < 0.01, (a) F test found variance significantly different (p<0.05) for T test equal variance is not assumed, (b) Hectares deforested since arrival to the farm (initial old growht forest old growth forest), (c) % forest cleared since arrival (ha deforested x 100 / ha initial old growth forest).

PAGE 83

70 Table 3-5. Continued Means Variables unit Iapari Assis Brazil T (1) (2) (3) Market and Credit Distance from nearest market km 14.59 10.50 2.89 (a)** Sells annual crops 0=no, 1=yes 0.64 0.76 -1.11 (a) Sells perennial crops 0=no, 1=yes 0.14 0.12 0.22 Sells small animals 0=no, 1=yes 0.28 0.54 -2.37 (a)* Sells Cattle 0=no, 1=yes 0.33 0.76 -4.064** Farm product commoditization index 5.08 7.95 -2.732** Times credit was received (e) 1.42 0.76 1.90 (a)+ Road Infrastructure Lives in main road 0=no, 1=yes 0.22 0.02 2.66 (a)* Lives in secondary road 0=no, 1=yes 0.22 0.44 -2.06 (a)* Lives in tertiary road 0=no, 1=yes 0.00 0.54 -6.81 (a)** Lives in walking path 0=no, 1=yes 0.56 0.00 6.61 (a)** Distance from main road km 3.24 7.65 -4.74** Transportation time hours 0.76 0.46 3.56 (a)** Household life cycle Years on farm years 14.08 13.73 0.14 Age of household head years 44.72 44.83 -0.03 Family members on lot count 4.53 4.56 -0.08 Family members working farm 2.17 2.76 -1.70+ Number of children count 1.56 1.83 -0.76 Number of adults count 3.86 4.46 -0.91 Labor hired index 2.00 3.10 -3.60** Labor exchanged index 2.14 2.00 0.49 Transformed values ln (1+var) Times credit was received (e) 0.66 0.45 1.688+ Transportation time 0.53 0.37 3.24 (a)** Years on farm 2.38 2.47 -0.48 Family members working farm 1.03 1.25 -2.16* Number of adults count 1.46 1.57 -0.96 + p < 0.1, p < 0.05, ** p < 0.01, (a) F test found variance significantly different (p<0.05) for T test equal variance is not assumed, (e) Since arrival to the property. Variables that represent the farmers background had means that were not significantly different except for initial area in secondary forest and percentage of family chiefs that were born in the MAP area. Both variables presented higher values for Assis Brazil. The percentage of households receiving a daily wage, and years of education, presented significantly higher values for Iapari.

PAGE 84

71 Market variables showed that surveyed households in Iapari were located significantly further from the nearest market than those in Assis Brazil. A significantly larger number of households in Assis Brazil sold small animals and cattle than households in Iapari. This was reflected in the farm product commoditization index which was significantly greater for Assis Brazil than for Iapari As for credit, interestingly, households in Iapari had received significantly more credit more than households in Assis Brazil since their arrival to the property. Credits were provided by the Agrarian Bank in Peru from the late 1950s until 1991. In Assis Brazil credit access is more recent; most households did not have credit available until the late 1990s. The road infrastructure variables show that farmers in Iapari are significantly more likely to live along a main road than those in Assis Brazil, but are significantly less likely to live along a secondary road. The majority of households in Assis Brazil (54%) lived along tertiary roads while no households in Iapari were found on tertiary roads, and instead the majority of households in Iapari (56%) were found along walking paths, where none were found for Assis Brazil (see Figure 3-1). Walking paths and tertiary roads are in different categories because walking paths were found in Iapari only, they are not passable by motor vehicles and they start in the border of the main road. Households in Assis Brazil were found more than twice as far (7.65 km), on average, than those in Iapari (3.24 km). Transportation time to the nearest market (either in Assis Brazil, Iapari or Iberia) is greater (31.8 minutes) for Iapari households than for those in Assis Brazil (22.2 minutes). This is explained because the index assumes motor vehicle

PAGE 85

72 transportation for roads, and walking speed for walking paths, which makes transportation time greater for Iapari households Of the eight household life cycle variables only two were found to be significant. The number of family members working on the farms and the days of labor hired were significantly higher in Assis Brazil than in Iapari. Relating the results of the mean analysis to the first hypothesis for this research we may say that land use outcomes are larger in area (annual crops, perennial crops, pasture) and in number (heads of cattle) in Assis Brazil than in Iapari as was expected. This may be explained by significantly different market and road infrastructure variables. However the second part of the hypothesis is not as expected: differences in land cover outcomes, in particular in area deforested at the household level, are not significant. This can be explained by significantly different background variables such as initial area of secondary forest, that is four times higher in Assis Brazil. The higher percentage of old growth forest cleared in Iapari is in part explained by the smaller farm sizes in Iapari. As for the second hypothesis we may say that household life cycle variables are not significantly different from Iapari to Assis Brazil except for the family members working the farm and labor hired; therefore differences in land use outcomes may be due to the effect of the variables external to the household However it is necessary to explore the way in which the dependent and independent variables interact to be able to draw more meaningful statements with respect to both hypotheses. The following analysis steps will treat all the questionnaires as a single sample and not as separate ones since the sample number is very low. Instead, the

PAGE 86

73 variable place will be included; this will allow us to use the whole sample and to observe the differences between the two sites. Correlations Between Independent and Dependent Variables Bivariate correlations are the third step in this analysis. Significant positive or negative correlations provide insight into relationships between individual variables and/or outcomes. Tables 3-6 to 3-14 present bivariate correlation coefficients (Pearson correlation) between each land use and land cover outcome, as well as against the entire suite of measured variables. Table 3-6 presents the correlations among the land use outcomes and land cover outcomes. These outcomes were natural log transformed prior to analysis. All significant correlations were found to be positive. Households having larger areas in perennials also had larger areas in annuals. For obvious reasons, farms with more heads of cattle also had a greater area in pasture. Table 3-6. Correlations between land use outcomes and land cover outcomes variables. Iapari and Assis Brazil, 2003. Annuals Perennials Pasture Cattle Variable (1) (2) (3) (4) Land use outcomes Ln ha annual crops 1 Ln ha perennial crops 0.983** 1 Ln ha pasture 0.145 0.061 1 Ln number heads of cattle 0.128 0.052 0.750** 1 Land cover outcomes Ln ha old growth forest 0.200+ 0.222+ 0.030 -0.049 Ln ha secondary forests 0.296** 0.317** 0.004 -0.095 Ln ha deforested (a) 0.264* 0.184 0.490** 0.304** Ln % deforested of forest (c) 0.196 0.127 0.343** 0.216+ + p < 0.1, p < 0.05, ** p < 0.01, (a) ha deforested since arrival to the farm (initial old growth forest old growth forest), (b) % forest cleared since arrival (ha deforested x 100 / ha initial old growth forest), N=77. The correlations with the land cover outcomes provided less obvious results. The area in old growth, secondary growth, and deforested were all positively correlated with

PAGE 87

74 the area in annuals. These same land cover outcomes, with the exception of area deforested, were also significantly correlated with area in perennials. Increasing area of pasture was found with increasing deforested area and % forest cleared since arrival. Greater numbers of cattle also were related to increasing levels of deforestation and % forest cleared. Table 3-7 shows correlations between land use outcomes and background information variables. All significant correlations were found to be positive. Larger initial farm sizes (ha), and head of household born within the MAP region, were both related to larger areas of annuals and perennials. Increasing area of pasture was related with increasing areas of: initial farm size and initial area of old growth forest, as well as the household having a regular monthly income. Initial farm size, initial area in old growth, initial area in secondary growth, having a regular monthly income and having a daily wage were all higher as the head of cattle owned by the household increased. Table 3-7. Correlations between land use outcomes and background information. Iapari and Assis Brazil, 2003 Annuals Perennials Pasture Cattle Variable (1) (2) (3) (4) Background information Ln ha of farm size 0.228* 0.194+ 0.503** 0.335** Ln initial ha old growth forest 0.165 0.127 0.338** 0.192+ Ln initial ha secondary forest 0.027 0.057 0.097 0.196+ Regular monthly income (0=no, 1=yes) -0.141 -0.157 0.229* 0.229* Daily wage (0=no, 1=yes) -0.130 -0.122 -0.150 -0.192+ Born in the MAP (area 0=no, 1=yes) 0.243* 0.254* -0.064 -0.018 Years of education 0.066 0.033 -0.027 -0.160 + p < 0.1, p < 0.05, ** p < 0.01, N=77 Table 3-8 shows correlations between land use outcomes and place, market and credit, and road infrastructure variables. Living in Assis Brazil was correlated with increases in all the land use outcomes. In the market category selling perennials was not correlated to any land use outcome. Selling annuals and small animals was positively

PAGE 88

75 correlated with area under annuals and under perennials. As expected, selling cattle and the commoditization index were positively correlated with pasture and heads of cattle. Table 3-8. Correlations between land use outcomes market and credit and road infrastructure. Iapari and Assis Brazil, 2003 Annuals Perennials Pasture Cattle Variable (1) (2) (3) (4) Place Iapari/Assis Brazil (0=I, 1=A) 0.238* 0.238* 0.246* 0.402** Market and credit Distance in km from nearest market 0.129 0.185 -0.209+ -0.315** Sells annual crops (0=no, 1=yes) 0.387** 0.423** -0.026 -0.117 Sells perennial crops (0=no, 1=yes) 0.109 0.067 0.009 0.067 Sells small animals (0=no, 1=yes) 0.361** 0.352** 0.234* 0.190+ Sells cattle (0=no, 1=yes) 0.232* 0.178 0.592** 0.732** Farm product commoditization (index) 0.024 -0.035 0.569** 0.715** Times credit was received 0.103 0.083 0.083 -0.126 Road infrastructure Lives in main road (0=no, 1=yes) -0.089 -0.137 -0.179 -0.276* Lives in secondary road (0=no, 1=yes) 0.384** 0.368** 0.359** 0.404** Lives in tertiary road (0=no, 1=yes) 0.017 0.055 -0.029 -0.022 Lives in walking path (0=no, 1=yes) -0.367** -0.353** -0.227* -0.211+ Distance in km from main road 0.161 0.217+ 0.050 0.213+ Ln transportation time (index) -0.016 0.049 -0.251* -0.213+ + p < 0.1, p < 0.05, ** p < 0.01, N=77 The credit variable was not correlated with any land use outcome. Households living on secondary roads had larger areas of all land use outcomes and owned more head of cattle; however, those living on walking paths had smaller areas of annuals and pasture and owned less head of cattle. Cattle herd size decreased for households living on main roads, tertiary roads, or walking paths, but increased for those families living on secondary roads. Table 3-9 presents the correlations between land use outcomes and household cycles. The area in annuals or perennials increased for households ranking higher on the labor exchanged index. Households that had more family members working on the farm had greater areas planted with perennial crops. The area of pasture and the numbrr of heads of cattle increased with the time on the farm, the number of children, the number

PAGE 89

76 of adults, but, declined for older farms, with more adults in the family, and for the ones that ranked higher in the labor exchange index. Table 3-9. Correlations between land use outcomes and household cycles. Iapari and Assis Brazil, 2003 Annuals Perennials Pasture Cattle Variable (1) (2) (3) (4) Household life cycle Ln years on farm 0.135 0.062 0.564** 0.391** Age of household head -0.133 -0.172 0.218+ 0.263* Family members on lot -0.091 -0.048 -0.262* -0.184 Ln family members working the farm 0.208 0.264* -0.041 -0.036 Number of children -0.054 -0.004 -0.309** -0.297** Ln number of adults -0.015 -0.050 0.352** 0.325** Labor hired index 0.146 0.106 0.296** 0.365** Labor exchanged index 0.360** 0.349** -0.168 -0.182 + p < 0.1, p < 0.05, ** p < 0.01, N=77 Table 3-10 shows the correlations between land use outcomes and land cover outcomes. The only significant, and fairly obvious, relationship in this table is between increasing hectares of forest deforested since arrival accompanying higher percentages of initial forest area being deforested. Table 3-10. Correlations between land use outcomes and land cover outcomes. Iapari and Assis Brazil, 2003 Old growth forest Secondary forest Ha deforested % deforested Variable (1) (2) (3) (5) Land cover outcomes Ln ha old growth forest 1 Ln ha secondary forests -0.049 1 Ln ha deforested (a) 0.186 -0.080 1 Ln % deforested of forest (b) -0.025 -0.080 0.929** 1 + p < 0.1, p < 0.05, ** p < 0.01, (a) ha deforested since arrival to the farm (initial old growth forest old growth forest), (b) % forest cleared since arrival (ha deforested x 100 / ha initial old growth forest), N=77 In table 3-11 the correlations between land cover outcomes and background information are shown. Larger farms and farms with more initial area under old growth forest were more likely to have larger areas of old growth forest and at the same time more deforested area. The area and percentage of forest area cleared since the

PAGE 90

77 households arrival to the parcel was smaller on farms with larger initial areas of secondary forest. Table 3-11. Correlations between land cover outcomes and background information. Iapari and Assis Brazil, 2003 Old growth forest Secondary forest Ha deforested % of forest cleared Variable (1) (2) (3) (5) Background information Ln ha farm size 0.746** 0.107 0.459** 0.147 Ln initial ha old growth forest 0.799** -0.132 0.636** 0.437** Ln initial ha secondary forest -0.105 0.299** -0.526** -0.564** Regular monthly income (0=no, 1=yes) 0.046 0.046 -0.010 -0.034 Daily wage (0=no, 1=yes) -0.074 -0.191+ -0.089 -0.049 Born in the MAP area (0=no, 1=yes) 0.025 0.043 -0.035 -0.049 Years of education 0.060 0.012 0.152 0.131 + p < 0.1, p < 0.05, ** p < 0.01, N=77 Table 3-12 shows correlations between land cover outcomes, markets and credit, and road infrastructure. Farms located further from, and having longer transportation times, to the nearest markets, or selling annual crops, had a larger area in old growth forest. Those households that were selling annual crops and that did not live on walking paths had increasingly larger areas in secondary forest. More hectares had been deforested since arrival by households that sold cattle but fewer had been deforested by those who live on tertiary roads. The percentage of initial forest deforested was higher for households that sold cattle, had received more credit since arrival, or that ranked highly on the farm product commoditization index. A smaller percentage was deforested, however, by families living on tertiary roads. Table 3-13 shows the correlations between land cover outcomes and household cycles. A greater area in secondary forest was found for older farms, that had more family members working on them, or that ranked highly on the labor exchanged index. Hectares deforested since arrival also was higher for older farms, with older household

PAGE 91

78 heads, that had a larger numbers of adults, or that ranked higher on the labor exchanged index. A greater percentage of the initial forest had been deforested on older farms that had more adults in their households. Table 3-12. Correlations between land cover outcomes, markets and credit and road infrastructure. Iapari and Assis Brazil, 2003 Old growth forest Secondary forest Ha deforested % deforested (a) Variable (1) (2) (3) (5) Place Iapari/Assis Brazil (0=I, 1=A) -0.013 0.065 -0.136 -0.201 Market and Credit Distance in km from nearest market 0.221+ -0.026 0.077 0.108 Sells annual crops (0=no, 1=yes) 0.390** 0.262* 0.108 -0.002 Sells perennial crops (0=no, 1=yes) 0.034 -0.13 0.05 0.076 Sells small animals (0=no, 1=yes) 0.076 0.070 0.133 0.144 Sells Cattle (0=no, 1=yes) -0.157 -0.082 0.238* 0.257* Farm product commoditization (index) -0.158 -0.100 0.214+ 0.214+ Times credit was received 0.135 0.146 0.329** 0.244* Road infrastructure Lives in main road (0=no, 1=yes) 0.035 0.119 0.188 0.189 Lives in secondary road (0=no, 1=yes) -0.011 0.141 0.181 0.133 Lives in tertiary road (0=no, 1=yes) 0.026 -0.003 -0.192+ -0.233* Lives in walking path (0=no, 1=yes) -0.041 -0.236* -0.136 -0.042 Distance in km from main road 0.176 -0.073 -0.109 -0.145 Ln transportation time (index) 0.260* -0.074 -0.027 0.005 + p < 0.1, p < 0.05, ** p < 0.01, (a) % forest cleared since arrival (ha deforested x 100 / ha initial old growth forest), N=77 Table 3-13. Correlations between land cover outcomes and household cycles. Iapari and Assis Brazil, 2003 Old growth forest Secondary forest Ha deforested % deforested (a) Variable (1) (2) (3) (5) Household life cycle Ln years on farm 0.013 0.280* 0.562** 0.460** Age of household head -0.148 0.048 0.205+ 0.165 Family members on lot -0.076 -0.008 -0.05 0.041 Ln family members working the farm -0.099 0.269* 0.011 0.036 Number of children 0.004 -0.107 -0.143 -0.079 Ln number of adults -0.129 0.086 0.256* 0.215+ Labor hired (index) 0.116 0.101 -0.026 -0.107 Labor exchanged (index) 0.180 0.190+ 0.203+ 0.165 + p < 0.1, p < 0.05, ** p < 0.01, (a) % forest cleared since arrival (ha deforested x 100 / ha initial old growth forest), N=77

PAGE 92

79 Results from this section mainly provide a general idea of the existing relationships between dependent and independent variables. Land use outcomes were more likely to be significantly correlated to the market, credit and road infrastructure while valuables land cover outcomes were more likely to be significantly correlated with background information variables. For a better understanding of these relationships, the next section presents multivariate models for land use and land cover outcomes. Multivariate Models This section constitutes the fourth step in the data analysis. Here Ordinary Least Square (OLS) regression models were run to gain insights into the way variables interact, since correlations provide limited information in this aspect. A total of five models were run for each land use and land cover outcome. Model 1 presents the results (beta coefficient, significance, r square and F) after regressing a given outcome (e.g. area in annual crops) against the background information (BI) variables. Model 2 presents market and credit (MC) variables, Model 3 presents road infrastructure (RI) variables, and model 4 presents household life cycle (HLC) variables. The fifth model (termed the comprehensive model) integrates the variables that were found to be significant in Models 1 to 4, as well as variables that were significantly correlated to the outcome (from Tables 3-6 to 3-13). Land Use Models For area in annual crops (Table 3-14) the BI model shows that those who had a regular monthly income had less area in annual crops. Conversely, those born in the MAP area, having more years of education, and living in Assis Brazil rather than in Iapari had a larger area of annual crops.

PAGE 93

80 Table 3-14. Models of farm area in annual crops outcome regressed on background information, market & credit road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003. Variables Model 1 Model 2 Model 3 Model 4 Model 5 BI MC RI HLC Comp. Background information Constant -0.288 Ln ha of farm size 0.154 Ln initial ha old growth forest 0.02 Ln initial ha secondary forest -0.01 Regular monthly income (0=no, 1=yes) -0.327* Daily wage (0=no, 1=yes) -0.170 Born in the MAP (area 0=no, 1=yes) 0.246+ Years of education 0.04* 0.042** Place 0.363* Market & credit Constant 0.626* Distance in km from nearest market 0.01 Sells annual crops (0=no, 1=yes) 0.328* 0.335** Sells perennial crops (0=no, 1=yes) 0.02 Sells small animals (0=no, 1=yes) 0.189 Sells cattle (0=no, 1=yes) 0.969** 1.109** Farm product commoditization (index) -0.09** -0.088** Times credit was received 0.05 Place 0.08 Road infrastructure Constant 0.897+ Lives in main road (0=no, 1=yes) -0.183 Lives in secondary road (0=no, 1=yes) 0.312+ Lives in tertiary road (0=no, 1=yes) Lives in walking path (0=no, 1=yes) -0.595* -0.463** Distance in km from main road -0.02 Ln transportation time (index) 0.867* 0.648* Place 0.06 Household life cycle Constant 0.463 Ln years on farm 0.103 Age of household head -0.01+ Family members on lot -0.05 Ln family members working the farm 0.286+ Number of children -0.01 Ln number of adults 0.125 Labor hired index 0.08 Labor exchanged index 0.184** 0.101* Place 0.134 R2 0.241 0.380 0.260 0.322 0.538 F 2.705* 5.220** 4.091** 3.573** 11.483** + p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized slopes, N=77

PAGE 94

81 The MC model shows that those who sold annuals or cattle and who ranked lower in the farm product commoditization index had more land in annual crops. In Model 3 (RI variables) farms on secondary roads or with higher transportation times (as calculated in the transportation time index) had a larger area, and farms located on walking path had a smaller area, of annual crops. The HLC model shows that households with a younger head of household, with more family members working in the farm, and/or those who exchanged more labor kept more area in annual crops. By observing the modells R 2 values we may say that market and credit variables (R 2 = 0.380) are the ones that explain the most variation, followed by household life cycle variables (R 2 = 0.322), road infrastructure variables (R 2 = 0.260), and finally the background information variables (R 2 = 0.241). The area in perennial crops (Table 3-15) for the BI model (incorporating background information variables, as previously explained) shows similarity to the case of area in annual crops. The MC model (incorporating market and credit variables) shows that those who were farther from the nearest market, who sold annual crops, or cattle, and those who ranked lower in the farm product commoditization index had more area in perennial crops. The RI model (incorporating road infrastructure variables) shows that households located on walking paths and households who had lower transportation times to the nearest market maintained a smaller area in perennial crops.

PAGE 95

82 Table 3-15. Models of farm area in perennial crops outcome regressed on background information, market & credit road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003 Variables Model 1 Model 2 Model 3 Model 4 Model 5 BI MC RI HLC Comp. Background information Constant 0.055 Ln ha of farm size 0.066 Ln initial ha old growth forest 0.013 Ln initial ha secondary forest 0.001 Regular monthly income (0=no, 1=yes) -0.176* Daily wage (0=no, 1=yes) -0.082 Born in the MAP (area 0=no, 1=yes) 0.137+ Years of education 0.018** 0.019** Place 0.175** Market & credit Constant 0.413+ Distance in km from nearest market 0.009+ Sells annual crops (0=no, 1=yes) 0.184** 0.182** Sells perennial crops (0=no, 1=yes) -0.023 Sells small animals (0=no, 1=yes) 0.085 Sells cattle (0=no, 1=yes) 0.518** 0.597** Farm product commoditization (index) -0.051** -0.052** Times credit was received 0.012 Place 0.067 Road infrastructure Constant 0.621* Lives in main road (0=no, 1=yes) -0.162 Lives in secondary road (0=no, 1=yes) 0.138 Lives in tertiary road (0=no, 1=yes) Lives in walking path (0=no, 1=yes) -0.362* -0.261** Distance in km from main road -0.089 Ln transportation time (index) 0.532* 0.448** Place 0.005 Household life cycle Constant 0.404* Ln years on farm 0.028 Age of household head -0.007+ Family members on lot -0.027 Ln family members working the farm 0.183* Number of children -0.003 Ln number of adults 0.069 Labor hired index 0.034 Labor exchanged index 0.091** 0.044* Place 0.069 R2 0.222 0.414 0.284 0.312 0.549 F 2.432* 6.007** 4.619** 3.370** 11.983** + p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized slopes, N=77

PAGE 96

83 The variables influencing the HLC model (incorporating household life cycle variables) are similar to those that influenced the area in annual crops. In the case of perennials, as in the case of annuals, market and credit variables (R 2 = 0.414) were the ones that had the most explanatory power, followed by household life cycle variables (R 2 = 0.312), road infrastructure variables (R 2 = 0.284) and background information variables (R 2 = 0.222). The variables explaining area in pasture (Table 3-16) from the BI model showed that larger sized farms had a larger area in pasture. The MC model shows that farms closer to the nearest market, that sold small animals, who ranked higher in the farm product commoditization index and those who received credit a greater number of times had a greater area of pasture. The RI model shows that farms on secondary roads had more area in pasture. Finally, the HLC model shows that households who had greater residence times on their farms and those whose families were of younger age had more area in pasture. Conversely, those households with more adults and those who exchanged less labor had more area in pasture. According to the R 2 coefficients household life cycle variables (R 2 = 0.508) are the ones that perform better, followed by market and credit variables (R 2 = 0.446), background variables (R 2 = 0.327), and road infrastructure variables (R 2 = 0.179). For heads of cattle owned (Table 3-17) the BI model shows that farms in Assis Brazil had more cattle than those in Iapari. The MC related variables of the second model shows that farms located closer to the nearest market, and farms who ranked higher in the farm product commoditization index, owned more cattle. The RI model shows that farms on secondary roads and farms in Assis Brazil had more cattle.

PAGE 97

84 Table 3-16. Models of farm area in pasture outcome regressed on background information, market & credit, road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003. Variables Model 1 Model 2 Model 3 Model 4 Model 5 BI MC RI HLC Comp. Background information Constant -3.063** Ln ha of farm size 1.382** 0.632** Ln initial ha old growth forest -0.294 Ln initial ha secondary forest -0.038 Regular monthly income (0=no, 1=yes) 0.333 Daily wage (0=no, 1=yes) -0.004 Born in the MAP (area 0=no, 1=yes) -0.433 Years of education 0.010 Place 0.469 Market & credit Constant 1.571* Distance in km from nearest market -0.045* Sells annual crops (0=no, 1=yes) 0.020 Sells perennial crops (0=no, 1=yes) -0.179 Sells small animals (0=no, 1=yes) 0.464+ Sells cattle (0=no, 1=yes) 0.634 Farm product commoditization (index) 0.093+ 0.115** Times credit was received 0.454* Place -0.134 Road infrastructure Constant 2.771+ Lives in main road (0=no, 1=yes) -0.484 Lives in secondary road (0=no, 1=yes) 0.683+ Lives in tertiary road (0=no, 1=yes) Lives in walking path (0=no, 1=yes) -0.020 Distance in km from main road -0.008 Ln transportation time (index) -0.844 Place 0.259 Household life cycle Constant 0.377 Ln years on farm 0.838** 0.538** Age of household head -0.025* Family members on lot -0.100 Ln family members working the farm 0.054 Number of children -0.094 Ln number of adults 0.702+ Labor hired index 0.093 0.136* Labor exchanged index -0.181+ Place 0.356 R2 0.327 0.446 0.179 0.508 0.632 F 4.139** 6.843** 2.551* 7.681** 30.948** + p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized slopes, N=77

PAGE 98

85 Table 3-17. Models of head of cattle outcome regressed on background information, market & credit, road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003 Variables Model 1 Model 2 Model 3 Model 4 Model 5 BI MC RI HLC Comp. Background information Constant -2.944+ Ln ha of farm size 0.794 Ln initial ha old growth forest 0.001 Ln initial ha secondary forest 0.075 Regular monthly income (0=no, 1=yes) 0.444 Daily wage (0=no, 1=yes) -0.021 Born in the MAP (area 0=no, 1=yes) -0.593 Years of education -0.002 Place 1.219* Market & credit Constant 1.088 Distance in km from nearest market -0.071** -0.048* Sells annual crops (0=no, 1=yes) -0.054 Sells perennial crops (0=no, 1=yes) 0.332 Sells small animals (0=no, 1=yes) 0.219 Sells cattle (0=no, 1=yes) 0.924 Farm product commoditization (index) 0.157* 0.200** Times credit was received 0.014 Place 0.194 Road infrastructure Constant -1.010 Lives in main road (0=no, 1=yes) 0.554 Lives in secondary road (0=no, 1=yes) 1.582** 0.693** Lives in tertiary road (0=no, 1=yes) Lives in walking path (0=no, 1=yes) 1.589 Distance in km from main road 0.057 Ln transportation time (index) -1.237 Place 1.562* Household life cycle Constant -1.084 Ln years on farm 0.642* 0.305* Age of household head 0.000 Family members on lot 0.033 Ln family members working the farm -0.083 Number of children -0.266+ -0.236** Ln number of adults 0.234 Labor hired index 0.173 Labor exchanged index -0.228+ Place 1.151* 0.482+ R2 0.277 0.635 0.351 0.425 0.712 F 3.251** 14.776** 5.358** 5.492** 28.800** + p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized slopes, N=77

PAGE 99

86 Model 4, incorporating HLC variables, shows that households who had greater tenure on the farm, lower numbers of children, who exchanged less labor, and farms in Assis Brazil had more cattle. Observing the R 2 values we may say that market and credit variables are the ones that performed best (R 2 = 0.635), followed by the household life cycle variables (R 2 = 0.425), the road infrastructure variables (R 2 = 0.351), and explaining the least of the variation, the background information variables (R 2 = 0.277). Land Cover Models The BI model variables that explained best why larger areas in old growth forest (Table 3-18) existed were larger area of secondary and old growth forests in farms on arrival. These farms had more area in old growth forest in the year 2003. The MC model shows that those who sold annual crops had more area in old growth forest. The third model (with RI variables) shows that farms with greater transportation times to the nearest market or farms who exchanged more labor (as shown in the HLC model) had more area in old growth forest. The model R 2 values show that, background variables (R 2 = 0.695) explain most of the variation, followed by market & credit variables (R 2 = 0.197), road infrastructure variables (R 2 = 0.116) and, household life cycle variables (R 2 = 0.101). For the area of the parcel in secondary forest (Table 3-19) the BI model shows that larger farms, those with smaller initial areas in old growth forest, and households without daily wages had a greater area of secondary forests. The second model (with MC variables) shows that households who sold annual crops had more area in secondary forest. The one significant RI variable in the third model shows that those who lived on walking paths had less area in secondary forest.

PAGE 100

87 Table 3-18. Models of area of old growth forest outcome regressed on background information, market and credit, road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003. Variables Model 1 Model 2 Model 3 Model 4 Model 5 BI MC RI HLC Comp. Background information Constant -1.136+ Ln ha of farm size 0.241 Ln initial ha old growth forest 0.918** 0.977** Ln initial ha secondary forest 0.168* 0.167** Regular monthly income (0=no, 1=yes) -0.083 Daily wage (0=no, 1=yes) 0.152 Born in the MAP (area 0=no, 1=yes) 0.104 Years of education 0.000 Place -0.136 Market & credit Constant 2.644** Distance in km from nearest market 0.027 Sells annual crops (0=no, 1=yes) 0.069** 0.240+ Sells perennial crops (0=no, 1=yes) 0.122 Sells small animals (0=no, 1=yes) -0.055 Sells cattle (0=no, 1=yes) -0.360 Farm product commoditization (index) 0.011 Times credit was received 0.135 Place 0.170 Road infrastructure Constant 2.891** Lives in main road (0=no, 1=yes) 0.339 Lives in secondary road (0=no, 1=yes) 0.090 Lives in tertiary road (0=no, 1=yes) Lives in walking path (0=no, 1=yes) -0.363 Distance in km from main road 0.013 Ln transportation time (index) 1.572* 0.707* Place 0.017 Household life cycle Constant 3.692** Ln years on farm 0.110 Age of household head -0.011 Family members on lot -0.023 Ln family members working the farm -0.179 Number of children 0.030 Ln number of adults -0.058 Labor hired index 0.104 Labor exchanged index 0.187+ Place -0.083 R2 0.695 0.197 0.116 0.101 0.724 F 19.333** 2.086* 1.525 0.836 47.110** + p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized slopes.

PAGE 101

88 Table 3-19. Models of area of secondary forest outcome regressed on background information, market and credit, road infrastructure, and household life cycle in Assis Brazil and Inapari, 2003. Variables Model 1 Model 2 Model 3 Model 4 Model 5 BI MC RI HLC Comp. Background information Constant 0.453 Ln ha of farm size 1.230** 1.012** Ln initial ha old growth forest -0.867** -0.962** Ln initial ha secondary forest 0.080 Regular monthly income (0=no, 1=yes) -0.072 Daily wage (0=no, 1=yes) -0.506+ Born in the MAP (area 0=no, 1=yes) 0.058 Years of education 0.019 Place -0.315 Market & credit Constant 1.381* Distance in km from nearest market -0.015 Sells annual crops (0=no, 1=yes) 0.511+ 0.669** Sells perennial crops (0=no, 1=yes) -0.518 Sells small animals (0=no, 1=yes) 0.178 Sells cattle (0=no, 1=yes) -0.187 Farm product commoditization (index) 0.002 Times credit was received 0.265 Place 0.104 Road infrastructure Constant 1.850+ Lives in main road (0=no, 1=yes) -0.075 Lives in secondary road (0=no, 1=yes) 0.111 Lives in tertiary road (0=no, 1=yes) Lives in walking path (0=no, 1=yes) -0.956+ Distance in km from main road -0.058 Ln transportation time (index) 1.033 Place -0.012 Household life cycle Constant 0.249 Ln years on farm 0.388* 0.367* Age of household head -0.007 Family members on lot 0.017 Ln family members working the farm 0.820** Number of children -0.127 Ln number of adults -0.224 Labor hired index 0.155 Labor exchanged index 0.135 Place -0.171 R2 0.218 0.119 0.100 0.233 0.309 F 2.365* 1.152 1.294 2.260* 8.042** + p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized slopes.

PAGE 102

89 The HLC model shows that those who had more years on the farm and more family members working in the farm had a greater area in secondary forest. According to the R 2 value, household life cycle variables (R 2 = 0.233) performed better, followed by background information variables (R 2 = 0.218), market and credit variables (R 2 = 0.119) and road infrastructure (R 2 = 0.100) variables. The background information variables (from model 1) show more hectares of land deforested since arrival to the farm (Table 3-20) for those farms who had larger initial areas in old growth forest, while farms who had smaller initial areas of secondary forests had deforested a larger area. The MC model shows that those who received credit more times since arrival, and households who lived in Iapari, had deforested a larger area. There were no significant RI variables. In the HLC model it is shown that older farms (years since establishment) had cleared more hectares of forest. The R 2 values for the models: background information, market & credit, and household life cycle (excluding road infrastructure which had no significant variables), were 0.523, 0.260, and 0.375, respectively. For percentage of old growth forest cleared since arrival (Table 3-21) the BI model shows that those who had smaller farms, larger areas of initial area in old growth forest, and farms which had smaller initial areas of secondary forests, had higher percentages of cleared forest. Model 2 (MC variables) shows that those who sold cattle, who ranked lower in the farm product commoditization index, who received more credit, and farms in Iapari had cleared a higher percentage of the initial old growth forest since arriving.

PAGE 103

90 Table 3-20. Models of area (ha) deforested outcome regressed on background information, market and credit, road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003. Variables Model 1 Model 2 Model 3 Model 4 Model 5 BI MC RI HLC Comp. Background information Constant -1.048 Ln ha of farm size 0.388 Ln initial ha old growth forest 0.528+ 0.577** Ln initial ha secondary forest -0.432** -0.390** Regular monthly income (0=no, 1=yes) -0.110 Daily wage (0=no, 1=yes) -0.321 Born in the MAP (area 0=no, 1=yes) 0.047 Years of education 0.022 Place 0.054 Market & credit Constant 2.185** Distance in km from nearest market 0.004 Sells annual crops (0=no, 1=yes) 0.242 Sells perennial crops (0=no, 1=yes) -0.132 Sells small animals (0=no, 1=yes) 0.354 Sells cattle (0=no, 1=yes) 0.694 Farm product commoditization (index) 0.030 Times credit was received 0.796** Place -0.700* Road infrastructure Constant 3.123* Lives in main road (0=no, 1=yes) 0.413 Lives in secondary road (0=no, 1=yes) 0.520 Lives in tertiary road (0=no, 1=yes) Lives in walking path (0=no, 1=yes) -0.765 Distance in km from main road -0.018 Ln transportation time (index) 0.662 Place -0.621 Household life cycle Constant 0.868 Ln years on farm 0.971** 0.693** Age of household head -0.016 Family members on lot -0.018 Ln family members working the farm 0.029 Number of children -0.018 Ln number of adults 0.198 Labor hired index 0.026 Labor exchanged index 0.148 Place -0.465 R2 0.523 0.260 0.125 0.376 0.667 F 9.329** 2.990** 1.672 4.492** 48.789** + p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized slopes, N=77

PAGE 104

91 Table 3-21. Models of percentage of initial forest deforested since arrival outcome regressed on background information, market and credit, road infrastructure, and household life cycle in Assis Brazil and Iapari, 2003. Variables Model 1 Model 2 Model 3 Model 4 Model 5 BI MC RI HLC Comp. Background information Constant 2.767* Ln ha of farm size -0.813+ -1.144** Ln initial ha old growth forest 0.915* 0.905** Ln initial ha secondary forest -0.405** -0.413** Regular monthly income (0=no, 1=yes) -0.027 Daily wage (0=no, 1=yes) -0.340 Born in the MAP (area 0=no, 1=yes) 0.076 Years of education 0.013 Place 0.047 Market & credit Constant 3.137** Distance in km from nearest market 0.002 Sells annual crops (0=no, 1=yes) -0.051 Sells perennial crops (0=no, 1=yes) -0.064 Sells small animals (0=no, 1=yes) 0.432 Sells cattle (0=no, 1=yes) 1.247+ 0.687** Farm product commoditization (index) -0.021* Times credit was received 0.579* Place -0.977** Road infrastructure Constant 3.386** Lives in main road (0=no, 1=yes) 0.521 Lives in secondary road (0=no, 1=yes) 0.527 Lives in tertiary road (0=no, 1=yes) Lives in walking path (0=no, 1=yes) -0.441 Distance in km from main road -0.015 Ln transportation time (index) 0.521 Place -0.639 Household life cycle Constant 1.630* Ln years on farm 0.852** 0.640** Age of household head -0.016 Family members on lot 0.044 Ln family members working the farm 0.049 Number of children -0.030 Ln number of adults 0.168 Labor hired index -0.020 Labor exchanged index 0.090 Place -0.596+ R2 0.417 0.261 0.112 0.289 0.634 F 6.074** 2.995** 1.470 3.023** 24.619** + p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized slopes, N=77

PAGE 105

92 The second model (incorporating MC variables) shows that households who sold cattle, that ranked lower in the farm product commoditization index, who received credit a greater number of times, and those who lived in Iapari had a higher percentage of old growth forest cleared since arrival. The RI model shows that the road infrastructure variables are not significant. The HLC model shows that households who had lived more years on their farms and who lived in Iapari had a cleared a higher percentage of old growth forest since their arrival Land Use and Land Cover Final Models Tables 3-22 and 3-23 present the final models obtained for each land use and land cover outcome. The tables show all the independent variables that were found to be significant for land use and land cover outcomes for comparison purposes. It is evident that the are planted in annual or perennial crops is determined by market and road infrastructure variables. Household and background variables are not important for determining crops. Area under pasture and the number of heads of cattle on the other hand, were higher on farms that had been occupied for a longer time. For land cover outcomes, background variables are more helpful in explaining deforestation and current areas under old growth and secondary forests. However, time of residence in the farm way also an important variable to explain area deforested since arrival, and percentage deforested. Market and credit variables do not seem to explain land cover outcomes. However there seem to be larger ares of old growth forest in farms with higher transportation times. Selling cattle is linked to higher percentages of deforestation

PAGE 106

93 Table 3-22. Final models for land use outcomes showing all the independent variables that were significant in final multivariate land use and land cover models. Assis Brazil and Iapari, 2003. Land Use Variables annuals perennials pasture cattle Background information Ln ha of farm size 0.632** Ln initial ha old growth forest Ln initial ha secondary forest Years of education 0.042** 0.019** Market & credit Distance in km from nearest market -0.048* Sells annual crops (0=no, 1=yes) 0.335** 0.182** Sells cattle (0=no, 1=yes) 1.109** 0.597** Farm product commoditization (index) -0.088** -0.052** 0.115** 0.200** Road infrastructure Lives in secondary road (0=no, 1=yes) 0.693** Lives in walking path (0=no, 1=yes) -0.463** -0.261** Ln transportation time (index) 0.648* 0.448** Household life cycle Ln years on farm 0.538** 0.305* Number of children -0.236** Labor hired index 0.136* Labor exchanged index 0.101* 0.044* Place 0.482+ R2 0.538 0.549 0.632 0.712 F 11.483** 11.983** 30.948** 28.800** Constant 0.324+ 0.306** -2.896** 0.130 + p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized slopes, N=77

PAGE 107

94 Table 3-23. Final models for land cover outcomes showing all the independent variables that were significant in final multivariate land use and land cover models. Assis Brazil and Iapari, 2003. Land Cover old growth secondary area percentage Variables forest forest deforest. deforest. Background information Ln ha of farm size 1.012** -1.144** Ln initial ha old growth forest 0.977** -0.962** 0.577** 0.905** Ln initial ha secondary forest 0.167** -0.390** -0.413** Years of education Market & credit Distance in km from nearest market 0.240+ 0.669** Sells cattle (0=no, 1=yes) Farm product commoditization (index) Road infrastructure Sells annual crops (0=no, 1=yes) 0.687** Lives in secondary road (0=no, 1=yes) Lives in walking path (0=no, 1=yes) Ln transportation time (index) 0.707* Household life cycle Ln years on farm 0.367* 0.693** Number of children Place 0.640** Labor hired index Labor exchanged index R2 0.724 0.667 0.309 0.634 F 47.110** 8.042** 48.789** 24.619** Constant -0.968* -0.011 -1.220* 2.382** + p < 0.1, p < 0.05, ** p < 0.01, (a) All models use OLS estimation, and coefficients are unstandardized slopes, N=77

PAGE 108

95 With respect to hypotheses H1: Access to markets, credit, and road infrastructure, drove more deforestation in Assis Brazil than in Iapari. In comparing Iapari and Assis Brazil, we did find differences in land use outcomes, but not in land cover outcomes. These differences, however, are not a result of location per se but rather of the differences found in key variables that influence land use decisions. Important variables were found to be significantly different from one place to the other, especially road infrastructure variables, market variables and background information variables. This is evident when comparing correlations between the variable place and land use outcomes (Table 3-8 and 12), with the final multivariate models for land use and land cover outcomes (Tables 2-22 and 23). When all variables are considered, the differences between places are well explained by road infrastructure, market and background information variables, although place as a variable by itself does not show significant relations with land use nor land cover outcomes Regarding market access, from the six market variables considered in this research, four were significantly different: households in Assis Brazil were on average closer to markets, more households in Assis Brazil sold small animals and cattle, and households in Assis Brazil were ranked higher in the farm product commoditization index. Multivariate models showed that market variables were important in determining land use outcomes, especially the area planted in annual and perennial crops. Market variables also were important in determining the heads of cattle owned, and the area in pasture, although these outcomes were also correlated. According to the market variables observed, the effects in land cover outcomes may vary. For example, if a household is

PAGE 109

96 selling annual crops it is likely to have a larger area in old growth and secondary forest, while selling cattle is related to having higher levels of deforestation. Access to credit differed between the two study sites, with households in Iapari having received significantly more credit than households in Assis Brazil. Since the late 1950s and through the early 1990s, households in Iapari had access to credit. Households in Assis Brazil have had access to credit only more recently, since the late 1990s. However, other sources of off-farm income, such as retirement funds, were available for households in Assis Brazil. In the multivariate model that included market and credit variables, the number of times a household had received credit had a significant influence only on the ha in pasture and the percentage of forest cleared. However, in the final multivariate model it did not have a significant influence on any land use or land cover outcomes. This may be may be due to temporal scale issues, such as the different times credit resources become available in each town. All six road infrastructure variables at the local level were significantly different. While in Assis Brazil there was a larger network of roads, and most households were located adjacent to one, in Iapari more than half of the households were located on only walking paths. Like market variables, road variables were important in determining the area in annual and perennial crops. However, they were not relevant for determining the area in pasture or heads of cattle owned. Road infrastructure variables are significant in modeling the area in old growth forest, with longer transportation to market times being related with larger areas of old growth forest, but they are not relevant in any other land cover model. These variables were less influential than had been expected. This may be due to the small spatial scale over which the research was conducted. Spatially more

PAGE 110

97 extensive studies are likely necessary to capture the effect of increasing distance to the road, since in this study all households were located within 15km from the main road. With respect to H2: Land use drivers do not have a linear influence on households; instead, much depends on the phase they occupy in their life cycle. This part of the analysis is more related to the integrated framework that was proposed in Chapter 2. Household life cycles in Iapari and Assis Brazil did match the adaptive cycles proposed from panarchy theory. Households that had older heads of households, had a longer residence time on the farm, had less children and more adults, and, therefore, seemed to follow the household transformation frame. The matter is to characterize the adaptive cycle phases according to household demographic characteristics. As for deforestation, the variables used as indicators for land cover were not significantly different between Assis Brazil and Iapari at the household level. This poses interesting questions because the variables used as indicators for land use outcomes were significantly different. Land cover outcomes were more determined by background variables related to the size of the property, the initial area in old growth and secondary forest than by the variables considered as land use drivers in the present study: market, credit, road infrastructure. The variables that were chosen as land use drivers were not good to model land cover outcomes as, while they may help in understanding who would be expected to have more area in annual crops, perennial crops, and pasture, they do not help to understand if an area in old growth forest or an area in secondary forest would be allocated to a certain land use activity.

PAGE 111

98 Considering the role of household demography as a mediator for the effect of land use drivers, from the correlation analysis it is apparent that older households had more heads of cattle and more pasture. Demographic variables were less important in modeling the area in annual and perennial crops. It is suggested that the general subsistence character of the farming activities would encourage the continuity in annual crops. On the other hand, perennial crops are discouraged by the number of cases of burned plantations and the frequent attacks of fungi; besides, the households that established perennial plantations were often supported by a national program, as high initial investments were usually required. The most important demographic variable seems to be the time of residence in the farm. In the multivariable models this variable was determinant of the area in pasture and the number of heads of cattle. Furthermore, this was one of the few variables that was important for both land use and land cover outcomes. Households that had lived on their farms for longer periods appeared to have more area in secondary forests, more are deforested since arrival, and a higher percentage of forest cleared since arrival. These results suggest that households slowly move from (exploitation) r to k (conservation) and major changes are produced once children reach adulthood (over 14 years old) allowing for the release and reorganization of household resources, after which a cattle ranching cycle is strengthened or started. However, other land use activities are also carried out. Land use drivers considered in the present study were conceptualized as local level factors; however they were not always only local. Market was, in general, at the local level, but cattle had a more regional character since buyers often came from state capital

PAGE 112

99 cities. Credit was a factor at the national level since in both places they were part of national policies whose objective was to give support to rural farmers. Road infrastructure had a mixed character; main road improvement and paving are part of national policies with broad support at the regional level. Secondary roads have only infrequently been created and maintained by regional or local governments. Cross-scale interactions at the broader and slower levels seem to be more obvious; road infrastructure upgrading (national level) does improve market access (local and regional level), by diminishing transportation time. In the case of cattle it has an interesting effect since it provides buyers access to the farms. In terms of faster levels, household demography does influence the way in which the households in Assis Brazil and Iapari allocate their resources. The pattern, as was mentioned before, is similar to the one presented by McCracken (2002), although older families in Assis Brazil and Iapari do not tend to rely more on perennials but rather on pasture and cattle. When heads of households were asked what their future plans were, and what they would do if they could get credit for the next year, one-third responded, in both sites, that they would invest in cattle-related activities. Policy may also play a role in future decisions; the legal reserve law was brought up by many of the farmers that expressed having future plans of planting perennial crops and increasing the number of small animals. It is clear that land use drivers do not have a linear influence on a household land use decisions; instead much depends on the phase they occupy in their life cycle. It is uncertain how future changes will alter the influence of the household life cycle on land use decision making. Assis Brazil and Iapari may soon be in the center of an important international trading route, and although most farmers have expressed positive

PAGE 113

100 expectations, the Pacific Highway is already bringing new settlers to the urban areas of both towns. Further research is required in order to be able to better delineate the household adaptive cycle. A second time survey and a bigger sample would allow for better results. Other indicators for household demography should also be tested

PAGE 114

CHAPTER 4 CONCLUSIONS This study compared current land use and land cover in Assis Brazil and Iapari. However it actually compared the outcome of a historical process that has led to the current landscape. The area of study is also currently undergoing important changes. The first steps toward the building of the Interoceanic highway are underway. Important differences were found between Assis Brazil and Iapari that influence the land use and land cover outcomes in each place. For the present study they did so via the operation of widely used variables in explaining land use and land cover change. The most important were found to be road infrastructure and market variables. Variables that are not so widely used were also found to be important. like background information variables. The findings show that place does make a difference, but exactly how that is so remains an empirical question, that may have roots in local, national or global processes that comparative studies may help to identify. In regard to the role of roads in defining land use and land cover outcomes, the study indicates that road infrastructure for Assis Brazil and Iapari is relevant in affecting decisions about areas to be used for annual or perennial crops. However, the activity showing the most rapid expansion in the area is cattle ranching. Under current conditions road infrastructure by itself is not a relevant factor when taking decisions on cattle ranching expansion in the area, since market factors are stronger for this specific activity. If under current conditions cattle ranching is expanding, with the Interoceanic highway the process may accelerate. 101

PAGE 115

102 Agriculture may also expand with the Interoceanic highway. The presence of one case of mechanized agriculture for rice production in Iapari, and the use of a singular greenhouse system for vegetables production in Assis Brazil, suggest that technological improvements for agriculture may also expand. Current land use practices, however, present an uncertain future for the area. While household life cycle stages do not seem to influence decisions on agriculture, older households tend to have more pasture and cattle. The hypothesized trend towards perennial crops for older households is not present in the study area. Limiting factors are two-fold. First, the widespread use of fire as part of farming practices, without adequate rules, has led to accidental burning of many of the perennial crops established in the area in the past. The second limiting factor is the lack of adequate technical assistance for perennial crop establishment and management, especially in regard to pests. More land under annual crops, perennial crops and pasture does not necessarily translate into more deforestation at the household level. Deforestation in this case is better explained by the initial extent of old growth and secondary forests, since we do not know if an area in old growth forest or an area in secondary forest will be allocated to a certain land use activity. While road infrastructure, market and background information variables explain differences in land use and land cover, these variables themselves are explained in great part by policies defined at different levels. This is more evident for the case of road infrastructure; in Assis Brazil roads were developed from a combination of, national, state and local policies, while in Iapari road infrastructure was almost entirely a national level policy matter, and therefore the site did not have secondary roads. In providing

PAGE 116

103 systematic information about existing differences and similarities among the authorities and farmers in Assis Brazil and Iapari, this research contributes to a cross-border development process that recognizes the distinct aspects of each side of the border. The integrated framework proposed in this research remains to be further developed and critiqued. Its assessment in the present research is limited to cross-sectional data, and it really requires a multitemporal approach, ideally complemented by panel data and satellite image analysis. Despite the high quality and large amount of data required to test integrated theoretical approaches for modeling land use and land cover change, such efforts can contribute to research in the land use and land cover change field by differentiating local processes from broader ones.

PAGE 117

APPENDIX A QUESTIONNAIRE APPLIED IN IAPARI, PERU Investigacin de tesis de maestra Cambio en el uso y en la cobertura de la tierra Angelica Almeyda Universidad de Florida Centro de Estudios Latinoamericanos Programa Desarrollo y Conservacin en los Trpicos Cuestionario confidencial para pobladores de Iapari (Madre de Dios, Per) Julio del 2003 Nmero de cuestionario ______________________ Fecha de entrevista ______________________ Fecha de revisin ______________________ Coordenadas geogrficas ____________________________________________ A. Parmetros demogrficos de la familia A.i Jefe de familia: 1). Cuantos aos tiene? _________________________ (edad) 2). Desde cuando reside en Iapari? _________________________ (aos) 3). Cual es su lugar de nacimiento? _____________________________________ (distrito, departamento) 4). Cual es su lugar de procedencia? _____________________________________ (distrito, departamento) 5). A que se dedica actualmente? ________________________________________________________ 6). A que se ha dedicado antes? ________________________________________________________ 7). Cual es su grado de educacin? ________________________________________(aos, nivel) 104

PAGE 118

105 A.ii Familia: 8). Cuantas personas hay en su familia/casa? _____________________________ 9). Edad, genero, actividad principal N Edad Genero Actividad principal Actividad en el uso de la tierra 1 2 3 4 5 6 7 B. Actividades econmicas 10). Cuales son sus fuentes de ingreso? (agricultura, ganado, recoleccin, caza, comercio, sueldos) ______________________________________________________________ 11). Posee o usted o usa propiedad en rea urbana? Donde? Ha? ______________________________________________________________ 12). Posee usted o propiedad en alguna rea rural? Donde? Ha? ______________________________________________________________ 13). Cuando fue la ultima vez que intercambio labor con otras familias? Como? Frecuencia? ______________________________________________________________ 14). Cuando fue la ultima vez que contrato mano de obra? para realizar que labores? Frecuencia? ______________________________________________________________ C. Cambio en el uso y en la cobertura de la tierra C.i Rgimen de tenencia de la tierra 15). Cual es la condicin de tenencia de la tierra q usted ocupa actualmente? (1) Propietario (2) Arrendador (3) Encargado (4) Ocupante (5) Otro______ 16). Desde cuando est en estas tierras? ____________________ (ao)

PAGE 119

106 17). Cual fue la situacin de la propiedad cuando lleg? (1) Propietario (2) Arrendador (3) Encargado (4) Ocupante (5) Otro______ 18). Cual es el rea de la tierra? ____________________________(ha) 19). Hubieron dueos anteriores o es usted el primero en vivir aqu? ______________________________________________________________ D. Actividades relacionadas al uso de la tierra 20). Cual es el rea actual bajo produccin de: (1) Anuales__________________________________________________(ha) (2) Perennes_________________________________________________(ha) (3) Pasto______________________________cuantas cabezas ?_______(ha) (4) Otro_____________________________________________________(ha) 21). Cual fue el rea inicial bajo produccin de: (1) Anuales__________________________________________________(ha) (2) Perennes_________________________________________________(ha) (3) Pasto______________________________cuantas cabezas? _______(ha) (4) Otro_____________________________________________________(ha) 22). Cual es el rea actual de bosques (1) Primarios_________________________________________________(ha) (2) Secundarios_______________________________________________(ha) 23). Cual fue el rea inicial de bosques (1) Primarios_________________________________________________(ha) (2) Secundarios_______________________________________________(ha) 24). Que emplea usted en sus actividades agropecuarias? Cuando? (1) Fertilizantes _________________________________________________ (2) Herbicidas___________________________________________________ (3) Vacunas_____________________________________________________ (4) Maquinaria__________________________________________________ (5) Otros_______________________________________________________ 25). Como clasificara a sus suelos en relacin a otros suelos de Iapari? (1) Frtiles (2) Regulares (3) Malos

PAGE 120

107 26). Realiza usted alguna practica de reforestacin? Cual? Por que? Desde cuando? __________________________________________________________________ ___________________________________________________________________ E. Crdito, carreteras y mercado 27). Actualmente tiene usted acceso a lneas de crdito? ________________________ 28). Cuando llego a estas tierras tuvo usted acceso a lneas de crdito? ________________________ 29). Si hizo uso antes de lneas de crdito, como? cuando? en que lo empleo? ______________________________________________________________ 30). Cual es la distancia actual a la carretera principal? __________________________(km) 31). Cual era la distancia inicial a la carretera principal? __________________________(km) 32). Cual es la distancia actual a las carreteras secundarias?________________(km) 33). Cual era la distancia inicial a las carreteras secundarias? __________________________ (km) 34). Como calificara las condiciones actuales de acceso al predio? (1) Buenas (2) Regulares (3) Malas 35). Como calificara las condiciones iniciales de acceso al predio? (1) Buenas (2) Regulares (3) Malas 36). A quien vende actualmente sus productos? (1) Anuales ____________________________________________________ (2) Perennes ____________________________________________________ (3) Ganado _____________________________________________________ (4) Forestales maderables _________________________________________ (5) Forestales no maderables _______________________________________ (6) Otros _______________________________________________________

PAGE 121

108 37). Cual era el mercado inicial para sus productos? (1) Anuales ____________________________________________________ (2) Perennes ____________________________________________________ (3) Ganado _____________________________________________________ (4) Forestales maderables _________________________________________ (5) Forestales no maderables _______________________________________ (6) Otros _______________________________________________________ F. Planes a futuro 38). Cuales son sus planes para el futuro? por que? ______________________________________________________________ ______________________________________________________________ 39). Si en este momento accediera a una lnea de crdito, en que la empleara? ______________________________________________________________ ______________________________________________________________ 40). Si estuviera mas cerca de la carretera, cambiaria sus actividades? ______________________________________________________________ ______________________________________________________________ 41). Cual ser el efecto de la carretera interoceanica en sus actividades? _____________________________________________________________ _____________________________________________________________ 42). Que medidas se deberan tomar para mejorar la vida del poblador de Iapari? ______________________________________________________________ ______________________________________________________________

PAGE 122

APPENDIX B QUESTIONNAIRE APPLIED IN ASSIS BRAZIL, BRAZIL Pesquisa de tese de mestrado Mudana no uso e na cobertura da terra Angelica Almeyda Universidade da Florida Centro de Estudos Latino-americanos Programa de Desenvolvimento e Conservao nos Trpicos Questionrio confidencial para moradores do Municpio de Assis Brasil (Acre, Brasil) Julio e Agosto de 2003 Nmero de questionrio ______________________ Data da entrevista ______________________ Data da reviso ______________________ Coordenadas geogrficas _____________________________________________________ B. Parmetros demogrficos da famlia B.i Chefe da famlia: 1). Quantos anos tem? _______________________ (idade) 2). Desde quando reside em Assis Brasil? ________________________ (anos) 3). Onde voc nasceu? _________________________________________ (municpio, estado) 4). Em que lugar voc morou anteriormente? _________________________________________ (municpio, estado) 5). A que se dedica atualmente? ___________________________________________________________ 6). A que se dedicava antes? ___________________________________________________________ 7). Qual o seu grau de estudo? _________________________________________________(anos, nvel) 109

PAGE 123

110 B.ii Famlia: 8). Quantas pessoas tem na sua famlia/casa? _____________________________ 9). Idade, gnero, atividade principal N Idade Gnero Atividade principal Atividade no uso da terra 1 2 3 4 5 6 7 C. Atividades econmicas 10). Quais so suas fontes de renda? (agricultura, gado, coleta, caa, comrcio, salrios) ______________________________________________________________ 11). Possui ou usa propriedade em rea urbana? Onde? Ha? ______________________________________________________________ 12). Possui propriedade em alguma rea rural? Onde? Ha? ______________________________________________________________ 13). Quando foi a ltima vez que fez adjunto com outras famlias? Como? Freqncia? ______________________________________________________________ 14). Quando foi a ltima vez que contratou mo-de-obra? para que? Freqncia? ______________________________________________________________ D. Mudana no uso e na cobertura da terra D.i Tipe da terra 15). Qual o tipo de condio de posse da terra que voc ocupa atualmente? (1) Proprietrio (2) Arrendado (3) Encarregado (4) Ocupante (5) Outro____ 16). Desde quando trabalha nestas terras? ____________________ (ano)

PAGE 124

111 17). Qual era a situao de posse da propriedade quando chegou? (1) Proprietrio (2) Arrendado (3) Encarregado (4) Ocupante (5) Outro____ 18). Qual a rea da terra? ____________________________(ha) 19). Haviam donos antes de voc? ______________________________________________________________ E. Atividades de uso da terra 20). Qual a rea atual de produo: (1) Anuais___________________________________________________(ha) (2) Perenes__________________________________________________(ha) (3) Pasto_______________________________quantas cabeas?_______(ha) (4) Outro____________________________________________________(ha) 21). Qual foi a rea inicial de produo: (1) Anuais___________________________________________________(ha) (2) Perenes__________________________________________________(ha) (3) Pasto_______________________________quantas cabeas? _______(ha) (4) Outro____________________________________________________(ha) 22). Qual a rea atual de bosques (1) Mata bruta, virgem ________________________________________(ha) (2) Capueira_________________________________________________(ha) 23). Qual foi a rea inicial de bosques (1) Mata bruta, virgem_________________________________________(ha) (2) Capueira_________________________________________________(ha) 24). Que usa em suas atividades agropecurias? Quando? (1) Fertilizantes__________________________________________________ (2) Herbicidas___________________________________________________ (3) Vacinas_____________________________________________________ (4) Maquinaria__________________________________________________ (5) Outros______________________________________________________ 25). Como classificaria os seus solos? (1) Frteis (2) Regulares (3) Maus

PAGE 125

112 26). Voc realiza alguma prtica de reflorestamento? Qual? Por que? Desde quando? ______________________________________________________________ ______________________________________________________________ F. Crdito, estradas e mercado 38). Atualmente voc tem acesso a linhas de crdito? ____________________________________ 39). Quando chegou a estas terras teve acesso a linhas de crdito? ____________________________________ 40). Voc usou linhas de crdito antes, como? quando? Em que o empregou? ______________________________________________________________ 41). Qual a distncia atual at a estrada principal? __________________________(km) 42). Qual era a distncia inicial at a estrada principal? __________________________(km) 43). Qual a distncia atual at o ramal mais prximo? __________________________(km) 44). Qual era a distncia inicial at o ramal mais prximo? __________________________(km) 45). Como qualificaria as condies atuais de acesso moradia? (1) Boas (todo o ano) (2) Regulares (no vero) (3) Ruins (difcil acesso) 46). Como qualificaria as condies iniciais de acesso moradia? (1) Boas (todo o ano) (2) Regulares (no vero) (3) Ruins (difcil acesso) 47). Para quem vende seus produtos atualmente? (1) Anuais ____________________________________________________ (2) Perenes ____________________________________________________ (3) Gado _______________________________________________________ (4) Florestais madeirabeis__________________________________________ (5) Florestais no madeirabeis ______________________________________ (6) Outros ______________________________________________________

PAGE 126

113 48). Para quem vendeu seus produtos no inicio? (1) Anuais _____________________________________________________ (2) Perenes _____________________________________________________ (3) Gado _______________________________________________________ (4) Florestais maderabeis _________________________________________ (5) Florestais no madeirabeis _____________________________________ (6) Outros ______________________________________________________ G. Planos para o futuro 41). Quais so seus planos para o futuro? por que? ______________________________________________________________ ______________________________________________________________ 42). Si neste momento tivesse uma linha de crdito, em que usaria? ______________________________________________________________ ______________________________________________________________ 43). Si estivesse mais perto da estrada, mudaria suas atividades? ______________________________________________________________ ______________________________________________________________ 41). Que efeito esta causando a estrada interocenica em suas atividades? Que efeito poder causar a continuidade da estrada e a ponte que nos ligar ao Pacfico? _____________________________________________________________ _____________________________________________________________ 42). Que medidas seriam necessrias para melhorar a vida do morador de Assis Brasil? _____________________________________________________________ _____________________________________________________________

PAGE 127

LIST OF REFERENCES AIDA and SPDA. 2002. Illegal Mahogany Logging in the Districts of Iapari and Iberia, Madre de Dios, Peru: The Role of the Newman Lumber Company: Asociacin Interamericana para la Defensa del Ambiente AIDA, Sociedad Peruana de Derecho Ambiental SPDA. Alston, L. J., G. D. Libecap, and R. Schneider. 1993. The Settlement Process, Property Rights, and Land Use on the Brazilian Amazon Frontier: Preliminary Results and Lessons from US Economic History. Paper read at All-University of California Economic History Conference, at Pasadena, April 2-3. . 1996. The Determinants and Impact of Property Rights: Land Titles on the Brazilian Frontier. Journal of Law Economics & Organization 12 (1):25-61. Alves, D. S. 2002. An Analysis of the Geographical Patterns of Deforestation in the Brazilian Amazon in the Period 1991-1996. In Deforestation and Land Use in the Amazon, edited by C. H. Wood and R. Porro. Gainesville: University Press of Florida. Angelsen, A. 1999. Agricultural Expansion and Deforestation: Modelling the Impact of Population, Market Forces and Property Rights. Journal of Development Economics 58:185-218. Arbex, X. 1997. El Futuro del Departamento de Madre de Dios (Per): Temas Sobre el Medio Ambiente y Actividad Humana. Cuzco: Centro de Estudios Regionales Andinos "Bartolom de Las Casas". Barbier, E. B. 2001. The Economics of Tropical Deforestation and Land Use: An Introduction to the Special Issue. Land Economics 77 (2):155-171. Bernard, H. R. 2002. Research Methods in Anthropology: Qualitative and Quantitative Approaches. 3rd ed. Walnut Creek: Altamira Press. Blaikie, P. 1995. Changing Environments or Changing ViewsA Political Ecology for Developing-Countries. Geography 80 (348):203-214. Blaikie, P. M., and H. C. Brookfield. 1987. Land Degradation and Society. London; New York: Methuen. Brondizio, E. S., S. D. McCracken, E. F. Moran, A. D. Siqueira, D. R. Nelson, and C. Rodriguez-Pedraza. 2002. The Colonist Footprint. In Deforestation and Land Use 114

PAGE 128

115 in Amazonia, edited by C. H. Wood and R. Porro. Gainesville: University Press of Florida. Bryant, R. L. 1992. Political EcologyAn Emerging Research Agenda in Third-World Studies. Political Geography 11 (1):12-36. Bryant, R. L., and S. Bailey. 1997. Third World Political Ecology. London ; New York: Routledge. Caillaux, J., and C. Chirinos. 2003. El Caso Tahuamanu: Cuando el Bosque Toca Madera. 1st ed. Lima: SPDA Sociedad Peruana de Derecho Ambiental. Carvalho, G., A. C. Barros, P. Moutinho, and D. Nepstad. 2001. Sensitive Development Could Protect Amazonia Instead of Destroying It. Nature 409:409. Chayanov, A. V. 1966. The Theory of Peasant Economy. Homewood, IL: Richard D. Irwin for American Economic Association. Cropper, M., and C. Griffiths. 1994. The Interaction of Population Growth and Environmental Quality. American Economic Review 84 (2):250-254. Ellis, F. 2000. Rural Livelihoods and Diversity in Developing Countries. New York: Oxford University Press. Escobar, A. 1996. Constructing Nature: Elements for a Postestructural Political Ecology. In Liberation Ecologies: Environment, Development, Social Movements, edited by M. Watts. London: Routledge. Godoy, R., S. Groff, and K. O'Neill. 1998a. The Role of Education in Neotropical Deforestation: Household Evidence from Amerindias in Honduras. Human Ecology 26 (4):649-675. Godoy, R., M. Jacobson, J. De Castro, J. Aliaga, J. Romero, and A. Davis. 1998b. The Role of Tenure Security and Private Time Preference in Neotropical Deforestation. Land Economics 74 (2):162-170. Godoy, R., M. Jacobson, and D. Wilkie. 1998c. Strategies of Rain-Forest Dwellers against Misfortunes: The Tsimane' Indians of Bolivia. Ethnology 37 (1):55-69. Godoy, R., K. O'Neill, S. Groff, P. Kostishack, A. Cubas, J. Demmer, K. McSweeney, J. Overman, D. Wilkie, N. Brokaw, and M. Martinez. 1997a. Household Determinants of Deforestation by Amerindians in Honduras. World Development 25 (6):977-987. Godoy, R., D. Wilkie, and J. Franks. 1997b. The Effects of Markets on Neotropical Deforestation: A Comparative Study of Four Amerindian Societies. Current Anthropology 38 (5):875-878.

PAGE 129

116 Gomes, C. V. A. 2001. Dynamics of Land Use in an Amazonian Extractive Reserve: The Case of the Chico Mendes Extractive Reserve in Acre, Brazil, Masters Thesis University of Florida, Gainesville, FL. Goudie, A. 1993. The Human Impact on the Natural Environment. Oxford: Blackwell. Governo do Estado do Acre, P. E. d. Z. E.-E. 2000a. Zoneamento Ecolgico-Econmico do Estado do Acre. 3 vols. Vol. I Recursos Naturais e Meio Ambiente. Rio Branco: Secretaria de Estado de Cincia Tecnologia e Meio Ambiente-SECTMA. . 2000b. Zoneamento Ecolgico-Econmico do Estado do Acre. 3 vols. Vol. II Aspectos Econmicos e Ocupao Territorial. Rio Branco: Secretaria de Estado de Cincia Tecnologia e Meio Ambiente-SECTMA. Gunderson, L. H., C. S. Holling, and S. S. Light. 1995. Barriers Broken and Bridges Built: A Synthesis. In Barriers and Bridges to the Renewal of Ecosystems and Institutions, edited by S. S. Light. New York: Columbia University Press. Hecht, S. B. 1985. Environment, Development and Politics: Capital Accumulation and the Livestock Sector in Amazonia. World Development 13 (6):663-684. Holling, C. S., and L. H. Gunderson. 2002. Resilience and Adaptive Cycles. In Panarchy: Understanding Transformations in Human and Natural Systems, edited by C. S. Holling. Washington: Island Press. Holling, C. S., L. H. Gunderson, and D. Ludwig. 2002a. In Quest of a Theory of Adaptive Change. In Panarchy: Understanding Transformations in Human and Natural Systems, edited by C. S. Holling. Washington: Island Press. Holling, C. S., L. H. Gunderson, and G. D. Peterson. 2002b. Sustainability and Panarchies. In Panarchy: Understanding Transformations in Human and Natural Systems, edited by C. S. Holling. Gainesville: University Press of Florida. Houghton, R. A. 1994. The Worldwide Extent of Land-Use Change. Bioscience 44 (5):305-313. IBGE. 2004. Censo Agropecurio 1995/1996. Instituto Brasileiro de Geografia e Estatstica-IBGE 1998 [accessed April 09, 2004]. Available from http://www.ibge.gov.br/. IBGE. 2002. Cidades: Assis Brasil. Instituto Brasileiro de Geografia e Estatstica-IBGE 2002 [accessed September 23, 2002]. Available from www.ibge.gov.br/cidadesat/xtras/temas.php. INEI. 2004. III Censo Nacional Agropecuario 1994. Instituto Nacional de Estadistica e Informtica-INEI 1999 [accessed April 09, 2004]. Available from http://www.inei.gob.pe/.

PAGE 130

117 INEI. 1997. Conociendo Madre de Dios. Lima: Oficina Tcnica de Difusin Estadstica y Tecnologa Informatica-OTDETI, Instituto Nacional de Estadstica e Informtica-INEI, Lima, Peru. . 2002. Banco de Informacion Distrital: Distrito Iapari. Instituto Nacional de Estadistica e Informtica-INEI [accessed September 23, 2002]. Available from www.inei.gob.pe/cpi/bancogeo/170301.htm. INRENA. 1998. Zonificacin Ecolgica-Econmica Yaco-Iapari e Iberia-Iapari, Madre de Dios. Instituto Nacional de Recursos Naturales-INRENA, Lima, Peru. Jones, D. W., V. H. Dale, J. J. Beauchamp, M. A. Pedlowski, and R. V. O'Neill. 1995. Farming in Rondnia. Resource and Energy Economics 17 (2):155-188. Kaimowitz, D., and A. Angelsen. 1998. Economic Models of Tropical Deforestation: A Review. Bogor: Centre for International Forestry Research. Kuchka, H. E. 2001. Method for Theory: A Prelude to Human Ecosystems. Ecological Anthropology 5 (special issue). Lambin, E. F., B. L. Turner, H. J. Geist, S. B. Agbola, A. Angelsen, J. W. Bruce, O. T. Coomes, R. Dirzo, G. Fischer, and C. Folke. 2001. The Causes of Land-Use and Land-Cover Change: Moving Beyond the Myths. Global Environmental Change 11 (4):261-269. Laurance, W. F., M. A. Cochrane, S. Bergen, P. M. Fearnside, P. Delamnica, C. Barber, S. D'Angelo, and T. Fernandes. 2001. The Future of the Brazilian Amazon. Science 291 (5503):438-439. Leinbach, T. R. 2000. Mobility in Development Context: Changing Perspectives, New Interpretations, and the Real Issues*1. Journal of Transport Geography 8 (1):1-9. Ley Forestal y de Fauna Silvestre. LEY No. 27308. July 16th 2000. Long, N., and A. Long. 1992. Batllefields of Knowledge: The Interlooking of Theory and Practice in Social Research and Development. London: Routledge. Mki, S., R. Kalliola, and K. Vuorinen. 2001. Road Construction in the Peruvian Amazon: Process, Causes and Consequences. Environmental Conservation 28 (3):199-214. McCracken, S. D., E. S. Brondizio, D. R. Nelson, A. D. Siqueira, and C. Rodriguez-Pedraza. 1999. Remote Sensing and GIS at the Farm Property Level: Demography and Deforestation in the Brazilian Amazon. Photogrammetric Engineering and Remote Sensing (65):1311-20.

PAGE 131

118 McCracken, S. D., A. D. Siqueira, E. F. Moran, and E. Brondizio. 2002. Land Use Patterns on an Agricultural Frontier on Brazil. In Deforestation and Land Use in the Amazon, edited by R. Porro. Gainesville: University Press of Florida. Meyer, W. B., and B. L. Turner. 1992. Human Population Growth and Global Land-Use/Cover Change. Annual Review of Ecology and Systematics 23:39-61. Ministrio do Planejamento e Oramento. 1999. Plano Plurianual 2000-2003 Avana Brasil. [accessed April 20, 2004]. Available from http://www.abrasil.gov.br/index.htm. Ministerio do Planejamento e Oramento. 2003. Avanca Brasil, Construindo o Brasil, Acre. [accessed December 23, 2003]. Available from http://www.abrasil.gov.br/infra/transportes/index.asp?uf=AC. Nepstad, D., J. P. Capobianco, A. C. Barros, G. Carvalho, P. Moutinho, U. Lopes, and P. Lefebvre. 2000. Avana Brasil: Os Custos Ambientais para a Amaznia. Belm: Instituto de Pesquisa Ambiental da Amaznia-IPAM, Instituto Scio Ambiental-ISA. Nepstad, D., G. Carvalho, A. Cristina Barros, A. Alencar, J. Paulo Capobianco, J. Bishop, P. Moutinho, P. Lefebvre, J. Lopes Silva, Urbano, and E. Prins. 2001. Road Paving, Fire Regime Feedbacks, and the Future of Amazon Forests. Forest Ecology and Management 154 (3):395-407. Nepstad, D., D. McGrath, A. Alencar, A. C. Barros, G. Carvalho, M. Santilli, and M. d. C. Vera Diaz. 2002. Frontier Governance in Amazonia. Science 295:629-631. Ojima, D. S., K. A. Galvin, and B. L. Turner. 1994. The Global Impact of Land-Use Change. Bioscience 44 (5):300-304. Ozrio de Almeida, A. L., and J. S. Campari. 1995. Sustainable Settlement in the Brazilian Amazon. Oxford: Oxford University Press. Peet, R., and M. Watts. 1996. Liberation Ecologies : Environment, Development, Social Movements. London ; New York: Routledge. Perz, S. G. 2001. Household Demographic Factors as Life Cycle Determinants of Land Use in the Amazon. Population Research and Policy Review 20 (3):159-186. . 2002a. Household Demography and Land Use Allocation among Small Farms in the Brazilian Amazon. Research in Human Ecology 9 (2):1-16. . 2002b. Household Demography and Land Use Allocation among Small Farms in the Brazilian Amazon. Human Ecology Review 9 (2):1-16.

PAGE 132

119 Perz, S. G., and R. T. Walker. 2002. Household Life Cycles and Secondary Forest Cover among Small Farm Colonists in the Amazon. World Development 30 (6):1009-1027. Pichn, F. J. 1997. Colonist Land-Allocation Decisions, Land Use and Deforestation in the Ecuadorian Amazon Frontier. Economic Development and Culture Change 45:707-744. Reglamento de la Ley Forestal y de Fauna Silvestre. DS No. 014-2001-AG(09.04.01). April 9th 2001. Rudel, T. R., and B. Horowitz. 1993. Tropical Deforestation: Small Farmers and Land Clearing in the Ecuadorian Amazon. New York: Columbia. Schmink, M. 1994. The Socioeconomic Matrix of Deforestation. In Population and Environment: Rethinking the Debate, edited by L. Arizpe, P. Stone and D. Major. Boulder: Westview Press. . 2004. Communities, Forests, Markets and Conservation. In Working Forests in the Tropics: Conservation Trough Sustainable Management?, edited by D. Zarin, M. Schmink and J. Alavalapati. New York: Columbia University Press. Schmink, M., and C. H. Wood. 1987. The 'Political Ecology' of Amazonia. In Lands at Risk in the Third World: Local Level Perspectives, edited by M. M. Horowitz. Boulder: Westview. . 1992. Contested Frontiers in Amazonia. New York: Columbia University Press. Scoones, I. 1999. New Ecology and the Social Sciences: What Prospects for a Fruitful Engagement? Annual Review of Anthropology 28:479-507. Smith, R. C. 1995. The Gift That Wounds: Charity, The Gift Economy and Social Solidarity in Indigenous Amazonia. Paper read at Symposium on Community Forest Management and Sustainability in the Americas, at University of Wisconsin, Madison, February 3-4. Soares-Filho, B., A. Alencar, D. Nepstad, G. Cerqueira, M. d. C. Vera Diaz, S. Rivero, L. Solrzano, and E. Voll. 2002. Simulating the Response of Land-Cover Changes to Road Paving and Governance Along a Major Amazon Highway: The Santarm-Cuiab Corridor. Paper read at Healthy Ecosystems, Healthy People ISEH/CI, at Washington, D.C, June 6-11. SUDAM and INADE. 1998. Plan de Desarrollo de las Comunidades Fronterizas Peruano-Brasileas Iapari Assis Brasil. Superinendencia de Desenvolvimento da Amazonia-SUDAM, Instituto Nacional de Desarrollo-INADE. Lima, Peru.

PAGE 133

120 Sydenstricker, N., and S. A. Vosti. 1993. Household Size, Sex Composition, and Land Use in Tropics Moist Forests: Evidence from Machadinho Colonization Project, Rondonia, Brazil. Varese, M. 1999. Drivers of Investment in Cattle among Landholders in the Southern Peruvian Amazon. Masters Thesis, University of Florida, Gainesville. Walker, R., S. Perz, M. Caldas, and L. G. Teixeira Silva. 2002. Land Use and Land Cover Change in Forest Frontiers: The Role of Household Life Cycles. International Regional Science Review 25: 169-199. Walker, R. T., and A. Homma. 1996. Land Use and Land Cover Dynamics in the Brazilian Amazon: An Overview. Ecological Economics 18: 67-80. Wood, C. H. 2002. Introduction: Land Use and Deforestation in the Amazon. In Deforestation and Land Use in the Amazon, edited by R. Porro. Gainesville: University Press of Florida. Wood, C. H., and M. Schmink. 1993. The Military and the Environment in the Brazilian Amazon. Journal of Political and Military Sociology 21 (Summer):81-105. Wood, C. H., and R. T. Walker. 2000. Tenure Security, Investment Decisions and Resource Use among Small Farmers in the Amazon, unpublished paper.

PAGE 134

BIOGRAPHICAL SKETCH Angelica Almeyda was born in Lima, Peru. After spending most of her childhood in the Peruvian Amazon, she graduated as a Bachelor in Forestry Sciences at Universidad Nacional Agraria La Molina in 1999. Her undergraduate thesis was conducted in the Peruvian central Amazon on secondary forest biodiversity. After working with the Instituto del Bien Comun from 2000 to 2002 in the Amazon Community-based natural resources managements research Initiative she started masters studies at the Center for Latin American Studies at University of Florida in 2002 where she conducted the present thesis research. 121