#1: ~‘ 1"
4 $315“? -
2.2.» "
, _

1.,
fie“,
Q

L. .»a

 

We
2

' ,3 r" f

r').. .""

50;QI)85

This is to certify that the
thesis entitled

LAND HOLDING AND LAND COVER/USE ON AN
AMAZONIAN FRONTIER

presented by
STEPHEN PETER ALDRICH

has been accepted towards fulfillment
of the requirements for the

 

 

 

 

Master of Arts degree in Geography
M ZUW ‘
Major Professor’s Signature
MM 4 2004-
Date

MSU is an Affinnative Action/Equal Opportunity Institution

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requeSted.

 

DATE DUE DATE DUE DATE DUE

 

oljfiqfeizagg SEP 3 o 2007

 

 

 

 

NOVL1 5 20.09
—U-8-3-9 t;

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DateDue.p65-p. 15

LAND HOLDING AND LAND COVER/USE ON AN AMAZONIAN
AGRICULTURAL FRONTIER

By

Stephen Peter Aldrich

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS

Department of Geography

2004

ABSTRACT

LAND HOLDING AND LAND COVER/USE ON AN AMAZONIAN
AGRICULTURAL FRONTIER

By

Stephen Peter Aldrich

Over the course of the past fifteen years, studies of land cover and land
use change have become a major item on the global research agenda (Geist and
Lambin 2001 ). Of the numerous potential land cover and land use changes, few
have been as intensively researched as deforestation, especially in tropical areas
such as the Amazon. However, research linking human agent interactions in the
Amazon with their effects on the landscape is not yet complete.

This thesis investigates the link between smallholder cattle ranching and
deforestation on a rapidly developing agricultural frontier in Uruara, Para, Brazil
by contrasting hollow frontier and household lifecycle processes. Many of the
recent changes in household demographic and land cover and land use may
indicate a hollowing frontier in Uruaré. However, household lifecycle processes

may also be at work in shaping this agricultural landscape.

Copyright by
Stephen Peter Aldrich
2004

ACKNOWLEDGEMENTS

First and foremost I must thank my family for their support over the past
two years; your help has been invaluable. I would also like to thank my
committee members, Joseph P. Messina, Robert T. Walker, and Antoinette
M.G.A WinklerPrins for their excellent advice and patience as I struggled to
complete this work. In addition, I would like to acknowledge my fellow graduate
students, and especially Eugenio Arima, for their support through my time at
Michigan State University. I would also like to thank Cynthia Simmons for help
with information and data concerning land concentration and land conflict. My
gratitude to the funding sources which enabled me to complete this work through
graduate assistant funding and travel grants, the Wege Foundation, the Tinker
Foundation, the Graduate School at Michigan State University, the Michigan
State University Department of Geography, and the College of Social Science at
Michigan State University, cannot be overstated.

Imagery data come from Jiaguo Qi, the Tropical Rain Forest Information
Center, and Landsat.org. Some tabular data acquired from Instituto Brasileiro de
Geografia e Estatistica and the Brazilian Censo Agropecuario of 1995/96. Also,
worldwide crop yield data come from the Food and Agricultural Organization of
the United Nations. Panel data were collected as part of a project titled “Patterns
and processes of landscape change in the Brazilian Amazon: A longitudinal,

comparative analysis of smallholder land use decision-making” completed by

John O. Browder, Robert T. Walker, and Randolph H. Wynne and funded by the

United States National Science Foundation.

TABLE OF CONTENTS

LIST OF TABLES ................................................................................................ viii
LIST OF FIGURES .............................................................................................. ix
LIST OF ABBREVIATIONS ................................................................................. xi
INTRODUCTION .................................................................................................. 1
CHAPTER 1: REVIEW OF EXISTING LITERATURE ........................................... 4
A BRIEF HISTORY OF AMAZONIAN DEVELOPMENT ................................... 4
LAND HOLDING AND COLONIZATION ........................................................... 9
LAND COVER AND LAND USE ...................................................................... 13
CATTLE RANCHING ...................................................................................... 17
THE HOLLOW FRONTIER ............................................................................. 22
HOUSEHOLD LIFECYCLES ........................................................................... 26
PROBLEM STATEMENT ................................................................................ 30
CHAPTER 2: STUDY AREA, METHODS, AND METHODOLOGY .................... 34
STUDY AREA ................................................................................................. 34
LAND CONCENTRATION / GINI COEFFICIENTS ......................................... 36
PANEL DATA .................................................................................................. 37
PANEL DATA METHODS ............................................................................ 4O
LAND COVER AND LAND USE CHANGE METHODS .................................. 43
POTENTIAL ISSUES IN LAND COVER AND LAND USE CHANGE
RESEARCH ................................................................................................. 47
A BRIEF DISCUSSION OF SWIDDEN AGRICULTURE ................................. 49
CHAPTER 3: PROPERTY DYNAMICS IN URUARA .......................................... 54
LAND CONCENTRATION .............................................................................. 54
LAND HOLDING CHANGE ............................................................................. 58
LOCATION OF LAND HOLDING CHANGES .................................................. 61
CHAPTER 4: PROPERTY-LEVEL CHANGES ................................................... 63
LABOR ............................................................................................................ 65
FARMING SYSTEMS ...................................................................................... 71
SUBSISTENCE CROPS .............................................................................. 71
CORN ....................................................................................................... 72
RICE ........................................................................................................ 74
BEANS ..................................................................................................... 75

vi

SUBSISTENCE CROP SUMMARY ......................................................... 76

COMMERCIAL CROPS ............................................................................... 77
COFFEE ................................................................................................... 78
CACAO (AND COCOA) ........................................................................... 79
PEPPERCORN ........................................................................................ 80
COMMERCIAL CROP SUMMARY .......................................................... 82

CATTLE RANCHING ................................................................................... 83
CATTLE RANCHING SUMMARY ............................................................ 84

PROPERTY LAND COVER AND LAND USE ................................................. 85

CULTIVATED AREA FOR ANNUAL AND PERRENIAL CROPS ................ 85

FOREST AREA ........................................................................................... 86

PASTURE AREA ......................................................................................... 86

CAPOEIRA .................................................................................................. 37

SUMMARY .................................................................................................. 88

PROPERTY SIZE AND LAND COVER AND LAND USE STRATEGIES ........ 9o
DETECTING CHANGE WITH THE AID OF REMOTELY SENSED DATA ..... 93

PROPERTY ONE ........................................................................................ 93

PROPERTY TWO ........................................................................................ 96

PROPERTY THREE .................................................................................... 97

PROPERTY FOUR ...................................................................................... 98

CHAPTER 5: CONCLUSIONS .......................................................................... 100
LITERATURE CITED ........................................................................................ 106

vii

LIST OF TABLES

Table 1. Changes in average hectares of land uses on properties in Uruara, 1996
— 2002. Statistically significant Changes are represented in boldface. ....... 119

Table 2. Changes in property land holdings and property population in Uruara,
1996 — 2002. Statistically significant changes are represented in boldface.
................................................................................................................... 1 18

Table 3. Agricultural production for selected crops of subsistence and
commercial nature in Uruara, 1996 — 2002. Units in 60kg sacks for all
variables except cocoa and peppers (in kilograms) and cattle (in number of
animals). Statistically significant Changes are represented in boldface. 120

Table 4. Brazil's production of key subsistence crops compared to other major
agricultural regions. Source: FAOSTAT (FAO 2004) ................................. 123

Table 5. Area in hectares dedicated to selected crops of subsistence and
commercial nature in Uruara, 1996 — 2002. Statistically significant changes
are represented in boldface ....................................................................... 121

Table 6. Average area of major land covers and land uses on large and small

holdings in Uruara during 2002. Units in hectares. .................................... 122
Table 7. Brazil's production of key commercially oriented crops compared to

other major agricultural regions. Source: FAOSTAT (FAO 2004). ............. 124
Table 8. Attributes for specific properties in the Uruara area. ........................... 125

viii

LIST OF FIGURES

Figure 1. Location of Uruara and study area within the Brazilian Legal Amazon.
The bottom graphic displays the location of the Legal Amazon within all of
continental South America, the middle graphic displays the location of the
Legal Amazon within a regional context, and the top graphic shows the
location of Uruara and the extent of the study area. Source data varied,
including MODIS data obtained in 2001, and Landsat 7 ETM Imagery dated
August 9, 1999 (bands 5,4,2) ..................................................................... 127

Figure 2. The Amazon basin with hydrologic network. The Brazilian Legal
Amazon is shown as a dotted line. Though the Tocantins river basin, directly
to the east of the Amazon basin, is often considered part of the Amazon,
especially in Brazil, it is not shown here. Source data: MODIS imagery
obtained in 2001, Hydrologic network and basin boundary from LBA-ECO
project. ....................................................................................................... 128

Figure 3. An illustration of how a Hollow Frontier comes about through a complex
series of cultural, political, environmental, and ecologic interactions at
varying scales across time. Based on the Kite model, from Campbell and
Olson (1991 :18). ........................................................................................ 129

Figure 4. Relief of the Uruara area, looking Eastward toward Uruara along the
Transamazon (BR-230). 5x vertical exaggeration makes the terrain variation
more obvious. Source: Terrain information from SRTM data at 90 meter
spatial resolution resampled to 30 meters for visualization purposes.
Imagery data from Landsat ETM imagery dated August 9, 1999 (5,4,2 band
combination). ............................................................................................. 130

Figure 5. Lots in the Uruaré area (between the towns of Placas in the West and
Brazil Novo in the East), shown on satellite imagery. Source: Imagery data
from Landsat ETM sensor dated August 9, 1999 (5,4,2 band combination).
Land parcel information digitized from paper INCRA lot map. ................... 131

Figure 6. Histogram of property Change between 1996 and 2002. Panel data
Changes are shown on the left, on the right property Changes for the entire
dataset are shown. Source: Data collected via survey (Browder et al. 2001).
................................................................................................................... 133

Figure 7. Location of stable land holdings in Uruara, shown in green. Stable land
holdings occur both close to Uruara and far out on the colonization roads.
Source: Land parcel location digitized from INCRA lot map. Parcel status

collected as part of survey (Browder et al. 2001). Imagery data from Landsat
ETM imagery dated August 9, 1999 (5,4,2 band combination). ................. 132

Figure 8. Multiple lot gains and losses between 1996 and 2002 in the Uruara
area. Lot gains, displayed in blue, occur slightly more often close to the
town of Uruara. Lot losses, shown in red, occur in more remote areas.
Source: Land parcel location digitized from INCRA lot map. Parcel status
collected as part of survey (Browder et al. 2001). Imagery data from Landsat
ETM imagery dated August 9, 1999 (5,4,2 band combination). ................. 135

Figure 9. Average proportion of land covers and land uses by property size in
1996. Numbers within bars indicate average number of hectares per
property in each land cover/use category. Source: Parcel status collected as
part of survey (Browder et al. 2001) ........................................................... 136

Figure 10. Average proportion of land covers and land uses by property size in
2002. Numbers within bars indicate average number of hectares per
property in each land cover/use category. Source: Parcel status collected as
part of survey (Browder et al. 2001) ........................................................... 137

Figure 11. Property one, Shown in 1986 (on the left), 1991 (middle), and then
1999 (on the right). Colonization road shown in red. Source: Land parcel
location digitized from INCRA Iot map. Parcel status collected as part of
survey (Browder et al. 2001). Imagery data from Landsat ETM imagery
dated August 9, 1999 (5,4,2 band combination). ....................................... 138

Figure 12. Property two, shown in 1986 (left), 1991 (middle), and 1999 (right).
Colonization road Shown in red. Source: Land parcel location digitized from
INCRA lot map. Parcel status collected as part of survey (Browder et al.
2001). Imagery data from Landsat ETM imagery dated August 9, 1999 (5,4,2
band combination). .................................................................................... 139

Figure 13. Property three, shown in 1986 (left), 1991 (middle), and 1999 (right).
Colonization road shown in red Source: Land parcel location digitized from
INCRA lot map. Parcel status collected as part of survey (Browder et al.
2001). Imagery data from Landsat ETM imagery dated August 9, 1999 (5,4,2
band combination). .................................................................................... 140

Figure 14. Property four, shown in 1986 (left), 1991 (middle), and 1999 (right).
Colonization road shown in red. Source: Land parcel location digitized from
INCRA lot map. Parcel status collected as part of survey (Browder et al.
2001). Imagery data from Landsat ETM imagery dated August 9, 1999 (5,4,2
band combination). .................................................................................... 141

Figure 15. Property Changes in Uruara property panel, 1996 — 2002 Source:
Parcel status collected as part of survey (Browder et al. 2001). ................ 134

 

LIST OF ABBREVIATIONS

INCRA — Instituto Nacional de Colonizacoa e Reforrna Agraria

IPAM — Instituto de Pesquisa Ambiental Da Amazonia

EMBRAPA - Empresa Brasileira de Pesquisa Agropecuaria

SUDAM — Superintendancy for the Development of Amazonia

POLOAMAZONIA — Programa de POlos Agropecuarios e Agrominerais da
AmazOnia

AEA — Association of Amazon Businessmen

LBA—ECO - Large-Scale Biosphere-Atmosphere Experiment in Amazonia

ESRI — Environmental Science Research Institute

FAQ — United Nations Food and Agricultural Organization

MODIS - Moderate Resolution Imaging Spectroradiometer

SRTM — Shuttle Radar Topography Mission

TM - Thematic Mapper (Landsat 5)

ETM — Enhanced Thematic Mapper (Landsat 7)

GIS - Geographic Information System

SAD1969 — South American Datum of 1969

RMSE — Root Mean-Square Error

Kg - Kilogram

Cm - Centimeter

xi

 

Km2 - Square kilometer
Km — Kilometer

Ha — Hectare

xii

INTRODUCTION

With the growing realization that global changes are occurring rapidly,

land cover1 and land use2 Change has become a topic of great interest in recent
years (see 9.9. Lambin et al. 2003 for a review). Large scale land cover and land
use Changes have been strongly associated with broad environmental impacts
such as greenhouse gas increases, water quality and flow pattern Change
(Gentry and Lopez-Parodi 1980; Pontius Jr. et al. 2000; Schneider and Pontius Jr
2001; Wayland et al. 2002; Alsdorf 2003), the spread of disease and other health
problems (Meade and Earickson 2000), global and local Climate Change (Meyer
and Turner II 1992; Bounoua et al. 2002), biodiversity loss (Wilson 1999), and
the degradation of land resources (Feamside 1990; Lambin et al. 2003). Other,
non-environmental impacts of land cover and land use changes are also
observable in many locales, such as encroachment of settlers on indigenous
lands (Simmons 2002) and the decimation of local social systems due to rapid
environmental change (Endfield and O'Hara 1999).

One of the most extensively researched of these global change processes

is deforestationa, especially deforestation of tropical forest resources (Geist and

 

1 In this thesis “land cover" refers to the actual type of coverage (or lack thereof) of bare earth
and is essentially resource-oriented (see eg. Anderson et al. 1976).

In this thesis “land use” refers to the actual human use of the earth and is essentially use-
oriented (see 9.9. Anderson et al. 1976).

Though deforestation is a common term, it is often misunderstood and has widely varying
meanings. In this thesis the word “deforestation” will indicate change in primary forest cover of
any kind.

Lambin 2002). Nowhere in the world is the loss of tropical forest area as marked
as it is in the Brazilian Amazon, the largest tropical moist forest on earth (Rand
McNally 20001101; Goulding et al. 2003). While the maintenance of forest
resources is essential to regional and global environmental health (Wilson 1999),
the importance of the forests in the Amazon is widely accepted, and the study of
deforestation and its drivers comprises one of the most controversial and lively
debates in academia. Among the deliberations regarding land cover and land use
change, one of the most intense focuses on the three points of a complex
triangle of interrelated landscape changes, involving (1) land cover and land use
change (especially deforestation), (2) cattle ranching, and (3) the distribution of
land ownership.

The purpose of this thesis is to investigate the relationship between land
holding and cattle ranching and their respective influences on land cover and
land use Change in the Amazon in the recent past. The work will unfold as a
comparative assessment of household lifecycle (founded in theories of peasant
economy, see 9.9. Ellis 1993) and hollow frontier processes (which consider
multiple dimensions across time and scales, see 9.9. Rudel 2002) and will
examine the correlation between land holding change, smallholder cattle
ranching efforts, and land cover and land use Change on the relatively new
agricultural frontier surrounding Uruara, Para, Brazil (Perz and Walker 2002;
Walker 2003). Because the household lifecycle and the hollow frontier are two of
the most frequently Cited theories used to explain the causes of land cover and

land use change on terre firme agricultural frontiers in the Amazon, this thesis

seeks to show which process is more evident in the area surrounding Uruara,
Para, Brazil (Figure 1). Though the frontier in the Uruara area does display some
characteristics of a hollowing frontier (increasing cattle herds and a declining
labor force), it is likely that household lifecycle processes are primarily
responsible for land cover and land use change in this area. This thesis will test
which of these two explanations for land cover and land use change processes is
in evidence in Uruara, Para, Brazil.

This thesis is organized into five chapters: Chapter one consists of: (1) a
brief history of development in the Brazilian Amazon, (2) a review of the literature
investigating land holding and colonization, (3) cattle ranching, (4) land cover and
land use change in Brazil, and then (5) concludes with a statement of the
problem this thesis will address. Chapter two reviews the methods I use to
address my research problem, and then discusses the difficulties and criticisms
of land cover and land use research and some preconceptions of smallholder
agriculture. Chapter three discusses land holding change in Uruara, how these
Changes are distributed spatially, and their implication for land cover and land
use in the area. Chapter four discusses farm system Changes, both on their own
and in relation to property Size, and discusses some specific instances of Change
between 1996 and 2002. Chapter five contains the conclusions of this work and
its implications for the future land cover and land use of the area surrounding

Uruara.

CHAPTER 1: REVIEW OF EXISTING LITERATURE

A BRIEF HISTORY OF AMAZONIAN DEVELOPMENT

Most estimates on deforestation in the tropics place current deforestation

at very high levels, including some ominous predictions of a global tropical moist

forest loss of over 50 percent4 by the year 2022 (Wilson 1999). There are many
different estimates, however, and some of the most alarming regional
deforestation in all of the globe is presently occurring in Central and South
America (Buschbacher 1986; Rudel 1997; 2002). Recent FAO (2000) estimates
put annual deforestation loss for Latin America in its entirety at 4.4 million
hectares per year between 1990 and 2000. Though 4.4 million hectares is an
enormous amount of forest land to clear per year, an estimated 63 percent of the
total tropical forests in Latin America remain (FAO 2000).

Because of the extent of deforestation, the development impacts on
indigenous cultures, and the environmental importance of the area, the Amazon
has become one of the most researched tropical forests. The Amazon River
basin encompasses nearly 6,915,000 kmz, and is the largest river basin in the
world (Rand McNally 2000:101; Goulding et al. 2003218). Contained within is also

the world’s largest tropical moist forest (Rand McNally 2000:101; Goulding et al.

 

4 Many estimates of the extent of future deforestation (such as those in Wilson 1999) do not
establish a baseline. For instance, if 50 percent of forest is to be cut by 2022, is this 50 percent of
the forest we see today, 50 percent of forest when Europeans arrived, or 50 percent of forest
from Neolithic times?

 

2003). It is no coincidence that most land cover and land use change research
on tropical moist forests targets this region of the earth. AS mentioned above, the
colonization and land holding of this region have a significant impact on the
landscape, but the history and methods of settlement in the Amazon are

complex.

In the past, the Amazon basin5 (Figure 2) was often considered an
endless ‘virgin’ forest ripe for use and improvement (for an example see: Delson
and Dickenson 1984). However, authors such as Denevan (1992), Balée (1992),
and, in a less direct way, Zinn (2001) have called the ‘pristine myth” (the idea of a
completely forested, virgin, and sparsely populated new wor1d) into question.
Adding to this debate, recent research has begun to show there was a much
larger Native American population in the new worid than previously assumed. In
fact, Denevan (1992), Balée (1992), Smith et al. (1995), and Erickson (2000),
Claim that most of the forests (and other environments) in the Amazon Basin
were selectively manipulated by humans and eventually became what we see
today: human created landscapes (Williams 2000; Mann 2002). Many of these
supposedly ‘pristine’ landscapes are, if the traditional ‘pristine’ view is accepted,
inexplicably dominated by vegetation useful to humans. Whatever their potential
for sustainable use under traditional technologies, it is certain Amazonian
countries will continue to make their vast forested areas available to

development, often espousing policies that encourage extractive uses of the land

 

5 The “Amazon basin” is the area that naturally drains into the Amazon river or its tributaries. This
includes much of Brazil, and portions of Colombia, Ecuador, Bolivia, and Peru. Also, the
Tocantins River Basin is often included in the definition of the Amazon basin in Brazil.

5

and the degradation of environment, potentially with global ramifications (Meyer
and Turner Il 1992).

Though humans have been inhabiting and using the forests of the
Amazon for centuries, much of our present historical knowledge of the area
begins with European “discovery.” Early exploration in the Amazon had two
origins: (1) Jesuit (Christian) missions and (2) agricultural colonies granted
directly by the King of Portugal (James 1969; Santos 1984). While the inland
penetration of missionaries was significant, especially from an epidemiological
perspective (James 1969), major agricultural development in the Amazon basin
moved fonIvard more rapidly through large land concessions made by the
Portuguese Monarchy (Santos 1984). Santos (1984) argues the historical basis
for the large landholder—tendencies we see today in the Amazon started early in
Brazil, with large land concessions by the king and the simultaneous evolution of
bourgeois property rights. However, large and small land concessions were
made, and it could be argued that the foundation for smallholder agriculture was
laid simultaneously (Santos 1984).

After the rubber boom ended in the early 20th century, at the end of the
Second World War, the military government pushed areas in the Brazilian
Amazon into development in a suitably militaristic fashion with “Operation
Amazonia” (Mahar 1979). Government leaders viewed the Amazon as an
enormous area with large amounts of unused land, free for exploitation and
development. This “empty” land was considered a solution for land shortage

problems in the agriculturally developed south and northeastern parts of the

 

country (Schneider 1995). Successive govemment—sanctioned settlement
programs had essentially the same goal: economic development of Amazonia,
and the alleviation of an acute land distribution problem evident in the more
densely settled areas of Brazil. However, some research has shown the
alleviation of a land crunch has not occurred, and frontier colonization in the
Brazilian Amazon did not complete its objective (Wood and Wilson 1984;
Simmons 2004).

During the 1970s and 1980s the Instituto Nacional de Colonizacao e
Reforrna Agraria (INCRA) began new programs to colonize parts of the Amazon.
These colonization programs were placed along newly opened highways (such
as BR-230, the “Transamazon” highway) which were created by the government
to promote economic development of the Amazon. INCRA frequently built
colonization roads spaced 5 kilometers apart and perpendicularly situated to a
development highway. Land along each of these roads was typically divided into
100 hectare plots and given to new colonists under the condition that they use it
productively as a family enterprise (Perz and Walker 2002; Walker 2003). INCRA
led colonization projects exist throughout the Brazilian Amazon, with many of the
more recent settlements occurring along BR—230 in the state of Para (Walker
2003) and along BR-364 in the state of ROndonia (Browder 2002). The
settlement project surrounding Uruara, Para, Brazil is shown in Figure 1.

From the early stages of Amazonian colonization programs, each settled
area has, to a greater or lesser extent, followed a similar progression of events.

At the point of initial settlement, a frontier area is typically isolated and

inaccessible because road networks are non-existent or of poor quality (Wood
1983). Transportation is very difficult and integration with regional and national
markets is impossible. Later, as the initial settlers establish themselves,
agriculture becomes commercial in nature and the frontier area slowly integrates
with regional and national networks of social and economic interaction (Wood
1983). Eventually, as individuals with capital are attracted to the frontier due to
the establishment of defensible property rights and local government, poorer
settlers may be forced out and will either move to an urban area or to another
agricultural frontier (Wood 1983; Brown et al. 1994; Walker and Homma 1996;
Rudel 2002).

During the 19903, Amazonian colonization became predominantly
unofficial and is presently driven primarily by private companies and other
individuals with an incentive to develop certain areas. While spontaneous and
unofficial colonization has occurred in the Amazon for quite some time (Moran
1984), individuals with economic interests (especially loggers) are inadvertently
promoting new colonization by extending existing colonization roads (or creating
new roads) in their quest for valuable timber species (Perz and Walker 2002;
Simmons 2002; Walker 2003). The expansion of logging operations could
accelerate present rates of deforestation, and future road building is likely to
intensify both colonization and deforestation. According to the Instituto de
Pesquisa Ambiental Da Amazonia (IPAM) 75 percent of the deforestation
occurring between 1978 and 1994 was within 50 kilometers of an official

development road (Nepstad et al. 2002). Additionally alarming is the fact that 90

 

percent of the logging that has occurred within the Brazilian Amazon is illegal
(Nepstad et al. 2002). New, govemment—sanctioned, colonization programs have
been largely put on hold due to pressures from the international community for
environmental reasons. However, when it comes to concerns involving global
capital, environmental and social trepidations take a back-seat to international
investment in ‘valuable’ projects such as hydroelectric power (Ciccantell 1999).
Frontier colonization, both old and new, has given rise to the present land holding
regimes and land cover and land use changes evident in the Amazon basin. I will
now briefly outline the problems concerning land holding commonly experienced

in settlement areas.

LAND HOLDING AND COLONIZA TION

Much of the land made available to settling small farmers in the Amazon
consists of 100 hectare lots, though many large areas, called “glebas” (large land
holdings often greater than 3,000 hectares) were set aside by the Brazilian
government and devoted to the development of a large cattle ranching industry
(James 1969; Mahar 1979; Hecht 1984; 1985; Buschbacher 1986; Hecht 1993;
Walker 1993; Schneider 1995; Walker and Homma 1996; Walker et al. 2000;
Perz and Walker 2002; Rudel 2002). Many of the smaller plots are devoted to
small scale agriculture, typically both commercial and subsistence in nature
(Buschbacher1986; Rudel 2002). However, in a trend that is occurring in all

parts of the Amazon, small land holders can find their holdings in jeopardy due to

 

the confused nature of tenure relations in frontier areas, which can lead to
conflict.

Land conflict has become a daily occurrence in some parts of the Amazon
basin. Smallholders are often in a tenuous social and financial Situation and find
themselves at odds with largeholders, commercial agriculture operations,
loggers, and some government development projects (Santos 1984; Coomes
and Burt 2001). The concentration of small holdings into commercial properties
has become more pronounced, which increases the potential for land-related
conflict in many areas (Simmons 2002). Also a factor in consolidation of land
resources is the widespread adoption of cattle ranching (Margulis 2004), which
has been shown to influence land conflict (Simmons 2004). Although it may be
difficult to conceptualize a land shortage in an area as large as the Amazon,
scarcity-driven land conflict has begun to affect the Amazonian landscape
(Simmons 2004). A notable outcome of land concentration is the encroachment
of land holders close to, and even onto, indigenous reserves, a serious problem
impacting both biological and cultural diversity (Simmons 2002).

Although smallholders dominate the Amazonian landscape, frontier areas
in most parts of Brazil experience Chronic rates of land holder turnover, even in
areas where productivity is high (Schneider 1995). In some areas, up to 60
percent of colonists sell their land to largeholders (Schneider 1995). Rapid land
turnover has been observed since the early development of the Amazon, and
was first attributed to environmental degradation of extraordinary proportions

(Myers 1981; Hecht 1984; Hecht 1985; Browder 1988b). The hypothesis of many

10

 

 

eaIIy Amazonian researchers was that because land owners sought to maximize
profits by farming in an extractive manner, productivity rapidly decreased with
increased environmental degradation (Myers 1981; Hecht 1984; Browder 1988b).
Researchers began to question not only the validity of one of the most prevalent
agricultural strategies in the Amazon, cattle ranching, but also the viability of
Amazonian agricultural development in general (Schneider 1995; Faminow
1998). Subsequent studies of colonization and agriculture in the Brazilian
Amazon suggest that the early condemnations of colonist agricultural efforts may
have confused environmental degradation with the early phases of colonization,
which can be unavoidably destructive (Hecht 1984; Hecht 1985; Browder 1988b;
Hecht and Cockbum 1990). Moran (1989a) emphasizes that all colonization
efforts move through a process of learning from various mistakes and that early
researchers too quickly criticized frontier areas which had not been given time to
complete a “learning period."

The notion that all Amazonian development efforts are completely
unprofitable has been further eroded by studies that have found frontier settler
incomes are relatively high compared to other regions of Brazil (Schneider 1995).
Schneider (1995) considers previous studies and concludes that there are more
subtle reasons for the high rate of land ownership turnover than solely declining
agricultural production. According to his argument, earIy settlers move away from
a developed frontier for “fundamental demographic and economic" reasons
(Schneider 1995: 22). As a frontier matures and becomes less “frontierish,”

property rights emerge with an increasingly visible and active government. After

11

 

some time, the enforcement of property rights opens the area to what Schneider
terms “urban capitalists,” who can afford to buy land in a frontier area as the risk
of losing their land to squatters diminishes. Urban capitalists are those that own
agricultural operations, but reside in town, away from their properties. Forced out
or voluntarily displaced by these urban capitalists, the original colonists make the
decision that benefits them most: sell their land and move to a new frontier or find
employment in an urban area (Schneider 1995). A more recent argument is that
high land turnover rates, even on highly productive plots of land, are due less to
environmental factors and more to an evolving land market and a geographically
advancing economic and social frontier (Walker and Homma 1996; Rudel 2002).
The argument that economic and social factors lead to change in land ownership
is not, in and of itself, a new idea. However, it is interesting to note that in many
frontier areas land holdings remain very dynamic, most likely due to Changes in
household demographics and regional economics that make buying, selling, or
subdividing lots more attractive to many small farmers.

The rapid change in land holding control, coupled with the continuing
maturation of frontier areas, manifests itself in the landscape through both subtle
and blatant changes to land cover and land use, among other things. Research
on land cover and land use changes in Brazil has been highly varied in location,
time, scope, and result. A brief overview of this work will be covered in the next

section.

12

 

LAND COVER AND LAND USE

Land cover and land use change literature exists in such abundance the
field is nearly its own discipline. This abundance is probably attributable to an
increased environmental awareness globally, though most concerns are
expressed by people in the more developed nations of North America and
Europe (Buschbacher1986; Hecht and Cockbum 1990; Walker 1993; Ciccantell
1999). Nevertheless, much of the early research found significant correlation
between land cover and land use change and an increasing concentration of
atmospheric greenhouse gas, water quality/flow changes (Gentry and Lopez-
Parodi 1980), and, most notably, global climate change (Meyer and Turner II
1992; Houghton et al. 1999). Major impacts on the biosphere have caused
significant interest in and support for new research surrounding land cover and
land use change simultaneously in more and less developed nations (Smith et al.
1995)

Due to rapid forest loss, especially tropical moist forest, deforestation has
become a significant focus of much of the existing literature on land cover and
land use change. Presently, high rates of deforestation are observed in the
Amazon Basin of South America, various forests in Central America, parts of
Africa, and many areas in Asia. Consequently, all of the above regions have
been, and continue to be, the subject of many land cover and land use change
studies (for examples see: Gentry and Lopez-Parodi 1980; Hecht and Cockbum
1990; Harrison 1991; Houghton et al. 1999; Messina and Walsh 2000; Coomes
and Burt 2001; FAO 2001; Geist and Lambin 2001).

13

 

While the issue of deforestation has been regularly covered by both the
popular press and academic journals, deforestation today is comparable in many
ways to lesser known deforestation processes of the past. In fact, historic
deforestation events in North America and Europe were just as dramatic in areal
scope and environmental impact as recent tropical deforestation (for European
examples see: Marsh 1864; Johnson 1980; for examples of North American
deforestation see: Whitney 1987; Williams 1989; Howarth et al. 1996; Verheyen
et al. 1999; Williams 2000; 2001). North American temperate forest loss was
significant, but today many temperate areas are experiencing a restoration of
forest cover (Johnson 1980; Walker 1993; Williams 2001). This fairly recent trend
is commonly referred to as a “forest turnaround.” In fact, in much the same way
as the “pristine myth,” there is mounting evidence that neolithic deforestation
events were much more expansive than previously thought (Williams 1989;
Williams 2000; 2001).

The occurrence of “forest turnaround” is occasionally used as a
justification for continued deforestation in tropical regions. Some researchers and
government officials argue that once tropical deforestation ends, forests will
reestablish themselves and a “forest turnaround” may occur in tropical regions
(Walker 1993; Rudel 2002). Proponents of a forest turnaround in the tropics often
point to the increasing presence of secondary forest, often referred to as
‘capoeira’ in the Brazilian Amazon, that many researchers have noted in the
Amazon (Smith et al. 1999; Sorrenson 2000; McCracken et al. 2002; Perz and

Walker 2002; Walker 2003) as proof for the existence of this phenomena.

14

 

However, there are many reasons why a forest turnaround in tropical regions is
unlikely in the Short-run, but the most significant are (1) the high cost of
protecting forests, (2) increasing demand for agricultural land, and (3) the
availability of labor saving agricultural strategies such as cattle ranching (Walker
1993; Rudel 2002).

Considering the difficulties associated with understanding forest cover
changes in the tropics, “forest turnaround” arguments may disappear as forest
resources in developed nations begin to experience increased use (Johnson and
Beale 1998; Williams 2001). Rural areas in developed countries (especially those
of North America) where forest resources have seen a sharp increase over the
past century are now experiencing Slowly declining forest cover, a process
catalyzed by nostalgia for rural areas coupled with widespread urban sprawl
(Johnson and Beale 1998).

At this juncture, few would debate the importance of land cover and land
use change research, especially research focused on deforestation, in tropical
regions around the wor1d. However, there is presently a great deal of debate
about the causes of these land cover and land use changes. In their influential
report, Geist and Lambin (2001) summarize over 150 cases of land cover and
land use change and list the causes of deforestation in each case. They separate
causes into two categories: (1) proximate causes, which “constitute (near-final or
final) human activities that directly affect the environment” (similar to: Turner ll et
al. 1990; Turner 1993), and (2) underlying causes, which are “fundamental forces

that underpin the more obvious or proximate causes of tropical deforestation”

15

 

(Geist and Lambin 2001:7-8). Geist and Lambin (2001 :23) found 96 percent of all
cases of deforestation have multiple causes, displaying the complex nature of
human-environment interactions. However, after simplifying cases of
deforestation into two-cause “tandems” Geist and Lambin (2001) find two
important tandems explain the cause of nearly half the case studies; 37 percent
of the case studies were caused by an “infrastructure-agriculture” tandem (mostly
in South America), and 10 percent were explained by the “logging-agriculture”
tandem (mostly in Asia).

The infrastructure-agriculture tandem follows a commonly described set of
events in which transportation infrastructure improvements lead to increased
deforestation. Transportation infrastructure, in turn, opens access to land for
crops and cattle ranching (Geist and Lambin 2001). Many authors have
described this process, or processes much like this (Feamside 1979; Hecht
1993; Faminow 1998; Walker et al. 2000). The logging-agriculture tandem is
similar to the infrastructure-agriculture tandem except that instead of
infrastructure improvements, logging activity opens land for agriculture. It could
be argued, however, that the logging-agriculture tandem would not be possible
without a previous infrastructure-agriculture tandem due to limited forest access
before transportation infrastructure enhancements. The logging-agriculture
tandem can be seen in the Amazon, though most often it occurs after an
infrastructure-agriculture tandem begins the land cover and land use change

process.

16

 

From studies on land use patterns of households (PichOn 1997; Coomes
and Burt 2001; Perz and Walker 2002; Walker 2003), to the social and economic
processes driving the conversion of forests to pasture (Feamside 1979; Myers
1981; Buschbacher 1986; Browder 1988b; Hecht 1993; Mattos and Uhl 1994;
Walker et al. 2000; Porro 2002), research on Amazonian deforestation continues
to be varied. A common and popular image of deforestation in the Amazon is that
of a large-scale logging operation, complete with logging skids, specialized
trucks, and huge sawmills. In fact, much deforestation in the Amazon may be
attributable not to large commercial logging ventures, but to smallholders, who
are often attempting to establish their own cattle ranching operation (Walker
1993; Faminow 1998; Walker et al. 2000; Coomes and Burt 2001; Walker et al.
2001; Browder 2002; McCracken et al. 2002; Porro 2002; Walker 2003). To date,
studies of deforestation have Shown cattle ranching, roads, markets, labor power,
and a multitude of other factors are at least partially responsible for deforestation.
Though cattle ranching is only one of many influences on deforestation, no
understanding of land cover and land use change in the Amazon would be

complete without its description.

CA TTLE RANCHING

Cattle ranching is by no means a new topic of research in Amazonia.

Mahar (1979) points out that as of 1970 there were already Close to Six million

17

 

cattle in the legal Amazon6, and that sixty million hectares of the basin were likely
to be suitable for livestock husbandry. In the late 19703, researchers and policy
makers thought cattle would be a viable and desirable use of the ‘wild’ lands in
Amazonia, and much of the research of this period was focused on improving
cattle production (Feamside 1979; Mahar1979). Actually, in dramatic contrast
with later perspectives, many researchers felt that cattle production was good for
environmental integrity in the Amazon, frequently Citing an EMBRAPA study that
made that claim (Falesi 1976). Even so, the optimism surrounding cattle ranching
was tempered with the knowledge that without government subsidies cattle
ranching was supposedly not profitable, except in the most extraordinary
circumstances (Feamside 1979; Mahar1979; Schneider 1995; Arima and Uhl
1997).

As the environmental movement awakened the world to the possibility of
irreparable environmental damage during the late 19703 and early 19803, the
idea that cattle ranches were environmentally beneficial to the Amazon rapidly
became passe (except for, possibly, research in livestock management). By
1984, the prevalent argument was that cattle ranches were counterproductive for
sustainability and development reasons. In a highly influential article, Susanna
Hecht (1985) argued that the costs associated with the development of
Amazonian cattle ranches, both in terms of environmental degradation and
economic development, were too great to encourage ranching. From a

sustainability perspective, cattle ranching came under fire because of its

 

6 The Legal Amazon contains the Brazilian States of Roraima, ROndonia, Para, Acre, Amazonas,
Mato Grosso, Tocantins, and Maranhao.

18

 

infringement on indigenous agricultural methods and excessive subsidies, which
were encouraged by the World Bank and other international development
agencies (Redclift 1987). Various highly controversial claims focused on the
negative aspects of cattle ranching were also made during the 19803.

One such claim was the much touted “hamburger connection’ that was
supposed to have created North American fast food hamburgers from hectares
upon hectares of Central American tropical moist forest (Myers 1981). In his
article outlining the “hamburger connection”, Myers (1981 :1) implies that beef
production had turned forests in all of Latin America into fast food hamburgers.
Myers goes on to paint a dramatic picture and claims that Costa Rican beef
production is environmentally harmful and economically unsound. While the
comparison Myers (1981) makes may or may not be spurious, conditions are
very different in the Amazon than in Costa Rica. Though the “hamburger
connection” was sensational at the time, and may have been an accurate
assessment of Costa Rican beef production, very little, if any, Brazilian beef was
sold during his study period in North America, and most Brazilian beef was sold
domestically (Smith et al. 1995). In fact, Browder (1988) estimates that in 1982
(shortly after Myers “hamburger connection” research) Amazonian beef
accounted for 0.007 percent of beef consumed in the United States, an
insignificant amount by any standard. Low international export levels for
Amazonian beef were due to herd health concerns such as hoof and mouth
disease and further reduced by the mad cow disease scares of the 19903.

Though Brazilian beef was not commonly exported to the United States,

19

 

domestic markets for Brazilian beef are not insignificant and absorb much of the
beef produced in Brazil. Beef is a major protein source in much of the Amazon,
though some researchers Claim that Brazil could claim more protein from the
Amazon River itself through fishing endeavors, as the Amazon has 8 times more
fish species than the Mississippi River (Pearce & Meyers 1990). Even so, cattle
ranching in the Brazilian Amazon has a bad reputation because the global
environmental impacts of cattle ranching are negative, even if the economic and
cultural benefits for ranchers are quite the opposite (Mattos and Uhl 1994).

In the late 19803 and early 19903 researchers still disparaged cattle
ranching in the Amazon, though most (including Hecht herself) thought that
Hecht’s early assessments of the inability of the Amazonian environment to
support ranching were excessive; some areas of the Amazon could profitably
support cattle ranching (Hecht 1993; Mattos and Uhl 1994; Toniolo and Uhl
1995). Indeed, for many small land holders, who primarily engaged in
subsistence agriculture, cattle ranching is the standard next step in the economic
development of their household (Hecht 1993; Walker 2003). Cattle ranching was
supposedly unprofitable, but large and smallholders alike continue to strive to
establish ranching operations, even after government subsidies have
disappeared (Walker et al. 2000; Porro 2002; Walker 2003). In fact, the cattle
herd in the state of Para has increased greatly between 1996 and 2002 with
6,182,000 head in 1990 and 10,271,000 head in 2002 (Margulis 2004). Ranching
appears profitable when more than just income per head or the amount of beef

sold is considered (Faminow 1998). A frequently cited reason settlers move

20

 

toward ranching is that cattle provide a mechanism to invest surplus funds
without using a bank (Hecht 1993; Faminow 1998). Cattle can also provide
household food services such as milk and meat products, and, given a large
enough herd, are self-reproducing. Many Amazonian households use cattle as a
risk aversion strategy and appreciate their addition to diversity of agricultural
options and financial security in times of hardship (Hecht 1993). The useful
lifespan of cleared land can be increased by cattle which can be raised on land
that might otherwise be left to fallow (Hecht 1993). Cattle also incur lower labor
costs (Hecht 1993; Faminow 1998; Walker 2003) than many other agricultural
strategies, and offer additional incidental services such as haulage ability and
manure production (Faminow 1998).

Even today many new studies on land cover and land use Change in the
Brazilian Amazon focus on ranching as the primary land use outcome of
deforestation efforts (Walker et al. 2000; Porro 2002; Rudel 2002). Though cattle
ranches are not now regarded as a totally inefficient and irrational use of land,
they are still held responsible for much of the deforestation presently occurring in
the Amazon. Because of the numerous services a herd of cattle can provide both
small land holders and large land holders, many people in frontier areas continue
to include cattle in their production strategies. In fact, as of 1990 over 85 percent
of the cleared forest area in the Brazilian Amazon was converted to cattle
pastures (Hecht and Cockbum 1990). Indeed, cattle ranches account for the
largest land use Class in the entire basin, but the commonly held belief that this

land use is universally negative seems to stem from a stereotypical reaction to

21

deforestation of tropical moist forests; deforestation, in all its forms, is often held
as detrimental to environmental health and socially irresponsible.

Whether environmentally sound or not, no one would deny that cattle
ranching and the forces that drive it are among the most important factors
impacting Amazon forests today. Many Claim that ranching is the outcome of a
social process; ranching is the result of an abandonment of cropping systems by
subsistence farmers and a conversion to pasture owned by wealthy ranchers.
This process is evident as a driving force behind much of the deforestation in the
Amazon, both old and new (Casetti and Gauthier 1977; Buschbacher1986;
Schneider 1995; Rudel 2002). This staging of land uses, with agricultural
systems changing from subsistence agriculture to cattle ranching creates what is
referred to as a ‘hollow frontier’ after research by James (1969), Taylor (1973),

and Casetti & Gauthier (1977).

THE HOLLOW FRONTIER

Historically, the most common references to the hollow frontier thesis are
in description of coffee production in Brazil, which is where the process was first
described (James 1969; Taylor 1973; Casetti and Gauthier 1977). Preston
James (1969) described the hollow frontier as a smallholder agricultural frontier
slowly being consolidated into large agricultural holdings with few inhabitants.
Recently, a hollow frontier-like process has been observed in the Ecuadorian
Amazon (Messina and Walsh 2001; Rudel 2002). Though no research has

described this process in the Brazilian Amazon, there is a distinct possibility that

22

 

the hollow frontier could exist in Uruara. A stylized version of the hollow frontier,
as applied to a frontier area such as Uruara, begins with colonization (promoted
by government initiatives or spontaneous in nature) by poor farmers practicing
subsistence agriculture consisting of staple food crops (Casetti and Gauthier
1977; Rudel 2002). The colonization process often results in successive
deforestation events which result in a smallholder dominated landscape used for --
subsistence agriculture. Eventually, as subsistence crop yields drop, land I
becomes unproductive and is abandoned (Rudel 2002). Many of the small land
holdings are sold to speculative interests (often urban capitalists) or to wealthy E
cattle ranchers who typically convert them pasture and consolidate land into
large ranches. Frontier colonists, forced out by an emerging land market, either
move on to “unused” forest resources and begin the deforestation process anew,
or depart to work in urban areas (Wood 1983; Brown et al. 1994; Rudel 2002).
Discussing the expansion of frontier areas, Wood (1983) describes the
hollow frontier process (he does not, however, name it as such), and mentions
that this series of events may lead to increased formal and informal economic
activity. Framing his discussion of the relation between smallholders and
largeholders in frontier areas of the Amazon in Marxist terms, Wood (1983)
points out that the informal activities of small subsistence farmers creates cheap
seasonal labor for the formal economy. Wood insists that formal and informal
activities, which are typically studied as separate entities, are essential and
complementary parts of the regional economic system (Wood 1983:263).

Furthermore, he suggests that these two Classes of economic activity enable

23

each other to grow, but that in its expansion the formal economy forces a hollow
frontier-like series of events (Wood 1983).

Hollow frontiers could be considered a manifestation of forces of differing
types and scale on the landscape. Because a hollow frontier occurs due to
tensions between large wealthy land holders (which in many parts of the Amazon
are frequently cattle ranchers) and smallholder farmers and ranchers, a complex
interrelationship is established. Each class of land holder influences the actions
of the other (Wood 1983), and the landscape is affected at widely differing
scales. The hollow frontier process may be interpreted as an explanation of land
use and land cover Change with a theoretical basis easily applied to political
ecology frameworks.

Using Campbell and Olson’s (1991) “Kite” model, the forces shaping a
hollow frontier can be easily recognized (Figure 3). Starting at the local level,
declining environmental quality increases the difficulty of smallholder farming on
a frontier (Figure 3). In turn, the political influence of large property holders
changes the ability of smallholders to reach their goal of ranching cattle (Figure
3). This desire to ranch cattle reinforces the frontier ethic of the area, further
increasing environmental damage, and making it easier for large holders to
purchase (degraded) land at a cheaper price (Figure 3). The influence of these
processes is not linear, and multiple interactions can take place between these
“points of the kite” and can influence interaction at multiple scales ranging from
local to global (Campbell and Olson 1991). National policies influence local

actions, and global opinion, political climate, and economic situation can

24

 

influence national and local events as well. The influence can also move the
other way; local events can shape national policy, which can then impact global
opinion as well.

For instance, global events have historically had an impact on local
landscapes in the case of cattle ranching and tropical deforestation. National
development policies in Brazil encouraged large numbers of people to settle to
the Amazon over the last fifty years (Browder 1988a; Moran 1989a; Hecht and
Cockbum 1990; Porro 2002). Of these national development policies, many
focused heavily on ranching cattle, such as the programs of the Superintendency
for the Development of Amazononia (SUDAM) and Programa de POlos
Agropecuarios e Agrominerais da AmazOnia (POLOAMAZONIA) (Simmons
2002). These development policies were, in turn, dictated or influenced by
international aid organizations touting Amazonian beef as a solution to global
hunger and national organizations such as the Association of Amazon
Businessmen (AEA) which also encouraged ranching (Simmons 2002). In this
manner, global food needs and development policies have influenced the
national policies of Brazil, especially with regard to Amazonia, creating the
conditions for a hollowing frontier in some parts of the Amazon.

The power structure inherent in any global to national to local (and
conversely local to national to global) interaction is quite frequently biased toward
national and global influence (Ciccantell 1999). In fact, historically the Amazon
has been viewed as a vast deposit of natural resources to be used for

exploitation (Mahar 1979), a view that has not necessarily Changed to any great

25

 

extent (Ciccantell 1999). Because this use-focused view of the Amazon has little
to do with those who inhabit the region, the structure of power can have a
significant influence on frontier inhabitants and their actions.

The presence of the hollow frontier process in some areas has contributed
to the popular notion that agriculture in the Amazon basin is universally extractive
in nature, for perennial crops, annual crops, and cattle ranching. While the hollow
frontier is mentioned in passing quite frequently in much of the land cover and
land use literature, the process is not often explained in great detail.

Though the hollow frontier tells a striking story of the development and
abandonment of frontier areas by smallholders, another process can also have a
significant impact on the landscape, resulting in a similar terminus. However, this
process, the household lifecycle, differs greatly in its explanation of the change to
frontier landscapes. In the next section, I will outline this process, its theoretical

origins, and its effects on local landscapes.

HOUSEHOLD LIFECYCLES

An alternative explanation, also mentioned frequently in much of the
research focused on smallholder influences on land cover and land use change
in the Amazon, is the household life cycle. This theory is based on the work of A.
V. Chayanov on peasant economics, and states that household demographic
change can translate into land cover and land use change, specifically increased
cultivation land area (Perz and Walker 2002). Though the effects of household

demography on agricultural frontier expansion are highly varied, many

26

investigations of the household lifecycle have observed a similar series of events.
The lifecycle of a colonist household begins with a young family settling and
beginning to use the land on a newly opened frontier. At first, their agricultural
strategy will focus solely on survival, and annual subsistence crops will be their
primary agricultural activity (Walker and Homma 1996). However, a constant
progression of deforestation will occur due to the household’s need for fertile
land, which, in turn, results in artificial land scarcity as formerly fertile lands fall
out of active use (Perz and Walker 2002). After some time, farmers gain
knowledge of how to farm successfully, and potentially have greater available
labor power as their Children grow (Moran et al. 2003). At this stage, with surplus
labor power, households Often begin to plant perennial, commercially oriented
crops due to their elevated potential for profit (Walker and Homma 1996). These
expansions in perennial crop area often lead to more labor intensive agricultural
practices and amplified land cover and land use Change (Walker and Homma
1996; Walker 2003). Alternatively, households with a smaller labor pool often
Choose to use their less fertile land to ranch cattle, a land use requiring much
less labor than the cultivation of perennials (PichOn 1997).

As a colonist household continues to mature, children or other relatives
may move off the property to start their own household. Alternatively, they may
move to another part of the same property (Walker 2003; Browder et al. 2004).
This, in turn, may lead to further deforestation, either in another area of the same
land holding, or in a completely different location (possibly close to a newly

opened frontier). This sort of property division has been noted in the state of

27

RondOnia, a part of the Amazon that has been intensively settled (Browder et al.
2004). When this property fragmentation occurs, original colonists may begin to
change their agricultural strategy due to decreased labor power. Often, if the
capital is available, households at this stage will either hire temporary labor
power or adjust by adopting a less labor-intensive land use, commonly cattle
ranching (Edelman 1985; Hecht 1993; Mattos and Uhl 1994; Faminow 1998;
Walker et al. 2000; Porro 2002; Rudel 2002). In this way, the population and age
of lot residents can significantly influence the land cover and land use of a
property in agricultural frontiers in the Amazon (Pea and Walker 2002; Walker
2003). Regardless of the final agricultural strategy employed, the cause of
property-level land cover and land use change, according to household lifecycle
theory, is a logical progression of events: (1) colonization, (2) deforestation, (3)
cultivation and/or ranching, (4) resettlement, and (5) repeat deforestation.

The explanation of land cover and land use changes through the use of
household lifecycle processes is essentially an approach founded on the
principles of peasant economy. Though a collection of individual households

each impact their local environments collectively, they operate as independent
entities. Each household7 is under a constant process of transition, closely
connected to its land, earning its subsistence through a nearly complete reliance

on household labor (Ellis 1993). This theoretical foundation does not stress the

 

7 A household, in this case, is the group of people sharing the same home after the definition set
forth by Ellis (1993).

28

multiple scales each action by smallholders can influence, and instead focuses
on local impacts.

Both household lifecycle and hollow frontier processes shape the
landscape in similar ways, and new literature suggests deforestation in the
Amazon may occur due to interactions between poor colonist holders and
wealthy ranchers (Rudel 2002). Rudel et al. (2002) show that much of the land
cover and land use change in the Ecuadorian Amazon may be caused by the
hollow frontier process. Though the economic, social, and political contexts of
deforestation the Ecuadorian Amazon are distinctly different than in Brazil, some
of the same processes that Rudel et al. (2002) observed in Ecuador are also
occurring in the Brazilian Amazon (Walker and Homma 1996; Simmons 2002).
Because the hollow frontier involves land cover and land use changes, it seems
a significant gap exists in the literature in connecting the hollow frontier process
to these changes in the wider Amazonian landscape. The connection between
colonist’s land consolidations into large holder cattle ranching operations (the
apparent result of the hollow frontier process) and land cover and land use
change has not yet been addressed. Walker (2003), among others, has pointed
to agent complexity in many critical logical tandems such as the relationship
between logging and agriculture and agriculture and infrastructure as important
to studies of deforestation. This thesis will explore the relationship between cattle
ranching, land holding change, and deforestation by focusing on the complexity
of the agents driving these processes and contrasting hollow frontier and

household life-cycle explanations.

29

 

PROBLEM S TA TEMEN T

As discussed, land cover and land use change is an increasingly
important field of research, and as such is constantly under revision. As land
cover and land use change research has matured, its explanations have
progressed from oversimplifications to the complex and intricate studies we see
presently (Lambin et al. 2003). The study of land cover and land use Change
came to the forefront of geographic research with the realization that human
actions were having an effect on regional climates and, now, is moving on to
include a much wider interdisciplinary viewpoint. Research about these important
changes has begun to synthesize environmental, social, and economic themes
and attempts to provide the scientific community and policy makers with a more
complete picture of change worldwide (Lambin et al. 2003).

Like any field of research, our knowledge of land cover and land use
change and associated drivers is far from complete; many aspects of land cover
and land use Change remain to be studied. With the increasing recognition of
complexity in human environment interactions comes a seemingly infinite number
of explanations for environmental change. In an effort to bring the number of
proximate (direct human action) and underlying (influencing social and economic
factors) causes of land cover and land use Change to a manageable level, Geist
and Lambin (2001) reduce the causes of change in 95 journal articles to a
generalized set of logical tandems. Of these, two logical proximate cause

tandems are most likely to impact deforestation in the areas surrounding Uruara,

30

Brazil; the infrastructure-agriculture tandem and the logging-agriculture tandem
explain over 40 percent of all land cover and land use changes in Latin America
according to Geist and Lambin (2002). The same study draws attention to the
complexity inherent in land cover and land use changes across Latin America;
nearly 20 percent of the tropical deforestation in the region can be explained by a
three-factor logical chain of infrastructure, logging, and agriculture (Geist and
Lambin 2002).

Much of the literature pertaining to tropical deforestation in the Amazon

focuses on the impact of swidden farmers8 and large land holders, such as
ranchers, commercial agriculturalists, and loggers, have on land cover and land
use (Hecht 1984; Hecht and Cockbum 1990; Mattos and Uhl 1994; Smith et al.
1999; Nepstad et al. 2002; Porro 2002). Researchers have long known that, in
both the low-lying floodplains along the Amazon river and on the Terra Firme in
the interior of the Amazon basin, a significant portion of land cover and land use
change in the Amazon occurs through the activity of small land holders (Dean
1995; Walker and Homma 1996; PiChOn 1997; Rudel 1997; WinklerPrins 1997;
Sorrenson 2000; Browder 2002).

Though large holder (cattle ranchers, loggers, and commercial
agriculturalists) and small holder (swidden agriculturalists, small ranchers,
subsistence settlers) land use strategies are well documented, our knowledge of
the interaction between these two distinct, though geographically interspersed,

groups remains skeletal. The important, and highly volatile, relationship between

 

8 For a discussion of swidden (shifting) agriculture see page 49.

31

small and large holders changes with time, and is an important research topic
(Wood 1983). The hollow frontier process, frequently mentioned in literature
dealing with frontier areas, may be apparent in the consolidation of land holdings
and the creation of properties with land uses requiring large areas, such as cattle
ranching. However, these processes may not be in evidence on smaller land
holdings such as the 100 hectare one-lot properties created through many
INCRA colonization projects. Instead, on these small lots, land cover and land
use changes may be governed by household lifecycle processes.

In the Amazon basin, many farmers Choose to raise cattle due to their
obvious food uses, cultural significance and their many economic benefits (Hecht
1993; Faminow 1998; Walker et al. 2000). While large ranches are common in
the Amazon, smallholders also control a significant number of the total cattle in
the basin (Faminow 1998; Walker et al. 2000). The controversy concerning the
effects of forest to pasture land cover/land use change continues, with pasture
often considered environmentally detrimental. (Feamside 1979; Hecht 1984;
Buschbacher 1986; Browder 1988b; Hecht 1993; Futemma and BrondIzio 2003).
Regardless of the effects of pasture on the quality of global and local
environments, it remains one of the most common land covers/uses in
deforested areas of the Amazon; both largeholders and smallholders raise cattle
on large pasture areas.

This thesis seeks to investigate the between link small holder cattle
ranching and deforestation in the Brazilian Amazon by contrasting hollow frontier

and household lifecycle processes. Land concentration, leading to large cattle

32

ranches, may cause what might be termed a hollow frontier in Uruara, Para,
Brazil. However, cattle ranching efforts of small farmers, frequently driven by
necessity of household demographic composition, are significant in their
contributions to land cover and land use change in the Umara area. These
competing processes leading to land cover and land use change show how two
different series of results can yield Similar landscapes. This work will attempt to
describe which process, and therefore which theory, best accounts for the land

cover and land use Change processes in Uruara, Para, Brazil.

33

CHAPTER 2: STUDY AREA, METHODS, AND
METHODOLOGY

STUDY AREA

Uruara is situated directly on BR-230, a federal road commonly referred to
as the Transamazon highway. The municipio9 and town of Uruara (located at
83°42’54”, W53°44’24”, Figure 1) are located between the cities of Altamira, 180
kilometers to the east, and ltaituba, 260 kilometers to the west. Uruara is situated
between the Amazon River, which is 175 kilometers to the north, and the lriri
River, which is 80 kilometers to the south. Due to its relatively land-locked
location, transportation in Uruara is mostly by vehicle on roads of widely varying
condition.

The area surrounding the town of Uruaré, Brazil, shown in Figure 1,
comprises a relatively young agricultural frontier. This area opened to colonists in
the late 19703 with the creation of the Transamazon highway, and a government
sanctioned settlement project administered by INCRA (Perz and Walker 2002;
Walker 2003). Before the colonization of the area through lNCRA’s settlement
efforts, populations in Uruara were sparse (Censo-Agropecuario 1995/1996). The
initial push to settle in Uruara started by the opening of the Transamazon, was
legitimized by INCRA splitting the landscape into over 3,000 lots averaging 135

hectares per lot (mode 96.64 hectares).

 

9 A municipio is roughly equivalent to a North American county.

34

Land holdingm, which is ever-Changing in much of the Amazon (Schneider
1995), is also variable in Uruara, though many small property holders have stable
land holdings (see page 58). Population, livestock production, and the cultivation
of some crops are increasing in the area, and land use regimes are gradually
shifting away from solely subsistence agriculture to more commercially focused
crops (see page 63).

Though the Amazon is often referred to as a “tropical rain forest” many of
the forests in the basin do not, in fact, possess rain forest climates. Much of the
Amazon could be more correctly classified as a “tropical savannah” climate (Aw)
according to the KOppen-Geiger climate classification system (Strahler and
Strahler 1992; Christopherson 1997). Tropical savannah climates experience
mean year-round temperatures exceeding 64.5° Fahrenheit, and have a strong
dry season with at least one month with below 6 cm of rainfall (Strahler and
Strahler 1992), and because area around Uruara has a pronounced dry season,
local climate is suggestive of tropical savanna. Even so, much of the vegetation
of the Amazon basin is tropical rain forest, which means it experiences a year-
round mean temperature greater than 645° Fahrenheit, and annual rainfall
generally exceeds evaporation in these areas (Strahler and Strahler 1992).

The terrain in Uruara is mildly undulating, and does not exhibit drastic

relief Changes (Figure 4). Though elevation differences are exaggerated in Figure

 

10 The term “land holding” is used throughout this thesis in place of “ownership” to indicate the
nature of Amazonian land possession. Because many landholders do not legally own the land
they live on, it is important to refer to the control of land as holding and not owning. A land holding
can be conceptualized in the same way as land owning, however, in the sense that land holdings
are used in much the same way.

35

 

4, it is easy to see how BR—230 follows less rugged terrain, and how the town of
Uruara itself is located in a relatively flat area. Average elevation in the area is
197.35 meters, though the range in elevations is wide, with a minimum elevation
of 16 meters and a maximum elevation of471.82 meters (0 = 66.84 meters).
Because of the relative flatness of this part of the Amazon basin, small variations
in topography can have a large impact on land cover and land use Changes due
to widely varying transportation costs (Arima 2004). However, the impact of
terrain in Uruaré is minor when compared with other areas around the globe, or
even other areas of the Amazon, especially those far to the west of Uruara in

Colombia, Peru, and Ecuador.

LAND CONCENTRATION / GINI COEFFICIENTS

Though the representation of land distribution change primarily reports
results from other research in this thesis (especially results from Censo-
Agropecuario 1995/1996) it is beneficial to briefly overview how the Gini
coefficient is calculated. The Brazilian national agricultural census (Censo-
Agropecuario 1995/1996) collects information on land holding size and divides
land holdings into classes. Using this information, it is simple to calculate Gini
coefficient values for the distribution of land resources in an area using the

following equation:

GI: (X.Y )-(X Y)

1 H! H] i

36

where “X,” is the cumulative percentage of land area divided into land holdings in
class “i”, “Y5” the cumulative percentage of the quantity (number) of land holdings
in Class “i”, and “i” is the class of land holding (after Porro 2002).

Once Gini coefficients are calculated the relative distribution of land
holdings is easily deduced. A Gini coefficient value close to one indicates more
concentrated land resources, while a Gini coefficient value close to zero indicates

evenly distributed land resources.

PANEL DA TA

A socio-economic and household attribute dataset was collected during
the summers of 1996 and 2002 as part of a collaborative project between
researchers at Virginia Polytechnic Institute and Michigan State University
(Browder et al. 2001). The study is an ongoing project (set to expire in Spring
2005) seeking to compare the processes of land cover and land use change in
Similar study areas along BR-364 in the state of RondOnia and BR-230 in the
state of Para. Because of constraints of time, resources, and knowledge I have
opted to use this dataset as an alternative to collecting my own data. However,
during the summer of 2003, l was able to work in some of the areas along BR-
364 in RondOnia, Brazil giving me some idea of the hardships experienced by
colonists and the social, economic, and physical structure of a developing frontier
very similar to Uruara.

While the two socio-economic property-level surveys are not in complete

agreement, and some properties Changed significantly between surveys,

37

properties surveyed during the two years are generally comparable. In order to
characterize the change in land use systems at the property-level, in addition to
tracking the Change in household attributes, these two survey datasets were
combined into one panel dataset in order to facilitate inquiry into the types and
extent of change in Uruara over a six year period.

Because of the arrangement of each survey and the nature of land
holdings in Uruara, this panel dataset can be divided into many different datasets
with different units of analysis, all of which may overlap with other datasets. The
ability to easily arrange the data in different ways makes the use of this data
more difficult to conceptualize, at least initially. Because both the 1996 and 2002
surveys interviewed property holders by asking about each individual lot they
controlled, the most basic level at which the panel could be analyzed is what I
term a “lot panel.” This panel compares individual INCRA delineated lots
(property parcels, cadastral map units) regardless of their owner or location. The
lot panel is useful to Characterize how varied lot land cover and land use can be,
but is not useful to illustrate land market forces or household demographic
change.

In order to more accurately Characterize changes in household
demographics, the panel dataset can be divided in order to create what I term a
“household panel.” This panel dataset compares individual family units (more
accurately termed “households” due to the fact that multiple families could
conceivably live in the same housing unit) across the area regardless of the land

resources they control or their location. As mentioned above, this data

38

arrangement makes it easy to detail the demographic and agricultural changes
households experience and the differences between individual households, but
does not necessarily make it easy to characterize changes in land holding.

A “property panel" groups data according to owner, not household or lot,
making it simple to compare changes in land holding over time. In addition, the
property panel illuminates the effects of differing property sizes on land cover and
land use in the area. Because the property panel groups lot-level data according
to owner, it makes the comparison of farms across size categories possible in a
way that the lot panel cannot. Further differentiating the property panel from the
household panel is the possibility for multiple households on each property
(though this is not in evidence in Uruara). Each of these panel datasets is distinct
but has significant overlap with the others. However, the panels listed above
stand on their own and do not rely on one another for their use in specific
analyses.

As mentioned above, the lot panel breaks the landscape into observable
units at the cadastral level (typically 100 hectare INCRA lots in this case). In this
sense, the lot panel is, for all practical purposes, the most fine-scale unit of

analysis collected in this survey. The lot panel, consisting of data collected in
1996 and 2002, covers 183 lots (91 lots in 1996, 92 lots in 2002“). As discussed

previously, this dataset does not consider ownership; instead it compares

variables on a lot to lot basis only.

 

11 One property in the lot panel dataset increased its lot holdings by 2002.

39

The household panel is an aggregation of lots into units consisting of land

controlled by a household, commonly (though not necessarily) a single family

unit. When unmatched data12 is discarded, however, the household panel
consists of only single family households of a similar demographic composition
(Table 1). The household panel consists of 76 households over two surveys for a
total of 152 observations.

The property panel, though not different from the household panel at this
stage, is conceptually distinct from both the household panel and the lot panel.
This dataset is based on property units, which are distinct from lots because a
property can consist of multiple lots. The property panel also differs from the
household panel as a property can actually house multiple households. Though
properties can consist of multiple lots, providing multiple households with a place
to live, the typical property in Uruara is actually a one-lot property with one
household. This is not always the case, however, and many properties in the
property panel consist of more than one lot. All together there are 76 properties

in the property panel in both 1996 and 2002.

PANEL DATA METHODS

Though the data collected through the 1996 and 2002 surveys is similar,
there are still some significant changes in certain variables between years. In this

thesis I asses the statistical significance of these changes through standard t-

 

12 Unmatched data includes properties that were sold between 1996 and 2002, lots that were
vacant, or missing lot identification data.

40

tests, which compare the means of different variables and quantify their
difference statistically. This statistical significance can enable a researcher to
discuss the actual significance of changes over time. Comparing the difference
between means is an efficient way to characterize the average differences
between properties in the 1996 and 2002 surveys. Because the panel dataset
contains information on the same properties during both time periods I have
opted to perform t-tests on much of the data collected during both time periods.
A standard t-test is actually a test of the hypothesis that the mean of
variable1 is equal to the mean of variable 2 (Khazanie 1997). The test of the
hypothesis that the means of two variables are equal using the Student’s t is as

follows:

t:

 

(XI TXzW
S

where X1 is the mean of variable 1, X2 is the mean of variable 2, n is the number
of observations, 3 is sample variance, and there are n — 1 degrees of freedom
(StataCorp 2001).

Because the land cover and land use data and household attribute
information is contained in respective panel datasets, I use a painIvise t-test to
quantify the difference between the mean values for a variety of variables
(Singleton Jr. et al. 1993; Khazanie 1997). This method of comparison enables
me to characterize the typical land use and land cover systems on properties in

the 1996 and 2002 surveys and to quantify the Change between survey years.

41

Though the panel data sets are versatile, there are some limitations to
their use, and a few biases may be apparent. For instance, larger properties
were not typically surveyed in 1996 or 2002 as part of the survey effort due to the
project’s focus on small farmers. Therefore, even though there are a few larger
properties (those having significantly more lots than is typical of properties in the
area) in the total dataset, they are not actually part of the panel. Agricultural
strategies in the Amazon are highly varied, including various combinations of
subsistence and commercial agriculture, a potentially complicating trend that
holds true for Uruara (though the most out-of—the-ordinary strategies are not
captured as part of this survey, e.g. land holders farming only coffee). However,
these initial biases were mitigated through survey design by implementing a
significant random component to survey site selection (Browder et al. 2001).
Also, because of the nature of the panel dataset’s identical observations and
variables over time, the results of each survey will not be completely independent
of each other; a property surveyed as part of the property panel growing 10
hectares of coffee cultivation in 1996 is more likely to have 10 hectares of coffee
cultivation in 2002 (Khazanie 1997). However, this interdependence of
observations between time periods can be compensated for by conducting a
painrvise analysis which considers only “in-pair difference (Khazanie 1997).”

While this analysis will focus on results derived primarily from the panel
datasets, at times I will differentiate between the data I use to report results.
Unless explicitly mentioned as being “outside the panel” all results are acquired

from panel analysis. Specifically, some instances of land holding change,

42

particularly examples of land consolidation, which I report on in this work,
consider properties outside the panel dataset. In other words, these properties
experienced such drastic Changes between 1996 and 2002 that they were

practically unrecognizable by the 2002 survey.

LAND COVER AND LAND USE CHANGE METHODS

Land cover and land use information, such as measurement of the amount
of primary and secondary forest cover (capoeira), pasture, and the cultivation
area of many specific crops, was collected as part of the 1996 and 2002 field
surveys conducted in Uruara. Using this information, and simple t-tests
(described above) I was able to define the average land cover and land use on
small properties in Uruara (see Table 2), and draw conclusions on the potential
for future changes (see pages 89 & 100). Though some might argue self-
reported land cover/use areas collected via survey may not be as accurate as
classified remotely sensed data, recent research has concluded that there is
often no statistically significant difference in accuracy between the two methods
of collection (Caviglia-Harris and Harris 2003). Due to constraints of knowledge
and time, I have opted to use the self-reported land cover and land use
information from the survey data to quantify land cover and land use change in
Uruara over the time period, though a method integrating both sources of data
would be most desirable. However, for display and explanation purposes, a
geographic information systems and remote sensing database was established

as part of this work. Images in this thesis are presented in color to better show

43

differences in land cover and land use and to emphasize certain aspects of
specific areas.

Using Landsat imagery acquired on a variety of dates by the Thematic
Mapper (TM) and Enhanced Thematic Mapper (ETM) sensor, survey data from
1996 and 2002, and geographic information systems (GIS) data outlining the
original official INCRA delimited lots, it is easy to identify areas of interest in
Uruara and the surrounding area. This combination of data sources allows me to
identify ’which lots have Changed hands, characterize how the agricultural system
on certain properties may be different from others, and then identify how much of
the area controlled by each property has been deforested.

Land parcel information was acquired from INCRA in the form of a paper
map without a projection (a spherical geographic coordinate system) and the
South American 1969 datum (SAD1969). This map was scanned at high spatial
resolution, and then digitized in Environmental Science Research lnstitute’s
(ESRI) ArCMap 8.1 using a heads-up process. This digital data, now in digital
GIS format, was then georeferenced and projected to match the Universal
Transverse Mercator projection (Zone 22 Zouth, World Geodetic System 1984
datum). These projection parameters match the Landsat ETM scenes I acquired
from the Tropical Rain Forest Information Center’s Landsat data archive
(Landsat.org 2004). After georeferencing and projecting GIS data, the Landsat
ETM satellite imagery and GIS data are co-registered and share the same

projection and datum information. Careful visual inspection indicates that the GIS

44

data are in spatial agreement with Landsat ETM imagery (acquired in June
1999)

However, older Landsat TM imagery, collected on the Landsat 5 platform,
are not delivered rectified. Due to gross geometric errors, Landsat ETM imagery
can be misregistered by up to 250 meters, though pre-release processing
mediates this error in many cases (EROS 2004). Because GIS data and Landsat
ETM imagery were in agreement, I rectified Landsat TM imagery, using ESRI’S
ArcMap 8.3, to the already georectified and projected ETM dataset (RMSE =
4.93 meters). Landsat TM imagery (acquired August 1986, July 1988, and July
1991), ETM imagery (acquired June 1999), and GIS data (digitized from original
cadastral maps) were in complete spatial agreement at this stage.

In order to link survey data from 1996 and 2002 to the GIS data a common
field (as mentioned eartier) must exist. The paper map of land parcels in the
Municipio of Uruara made available to the “Patterns and Processes of Landscape
Change in the Brazilian Amazon" project by INCRA contained the official lot
number of each lot (Browder et al. 2001). However, instead of recording this lot
number in the questionnaire and then matching survey and GIS data using
INCRA lot numbers, the INCRA lot number was matched to questionnaire
identification number in the field to eliminate further data processing steps. The
questionnaire number was subsequently added to the digitized INCRA lot
boundaries in the GIS and this questionnaire number field was used to link the

survey data to GIS data.

45

As mentioned above, this dataset was assembled primarily for
visualization purposes, and also to identify areas of interest in Uruaré. However,
although there are many areas displaying potentially interesting land cover and
land use changes, due to the large area considered around Uruara and the
necessarily limited number of properties surveyed, most of these areas were not
surveyed during field data collection.

Though many potential areas of interest were not covered by the 1996
and 2002 surveys, there were some properties displaying significant changes
that are characteristic of typical properties in Uruara. Some of these properties
are discussed in detail in “CHAPTER 4: PROPERTY-LEVEL CHANGES” starting
on page 93. These particular properties were selected because they display
characteristics and Changes that closely match the average land cover and land
use for farms of similar size in the study area.

At first glance, land cover and land use change research using data
collected via remote sensing seems to be a fairly non-invasive, but therefore
somewhat superfluous, form of data collection. Compared with survey based
research, it seems that many of the potential pitfalls survey data collectors have
to work carefully to avoid can be circumvented by collecting data on land cover
and land use using remote sensing. Nevertheless, as with any new technology
and concurrent shift to new methods, some researchers have raised questions
surrounding the ethical, gender, and data accuracy issues remotely sensed data

can raise.

46

POTENTIAL ISSUES IN LAND COVER AND LAND USE CHANGE
RESEARCH

Land cover and land use Change encompasses a plethora of potential
changes in the status of the land. Even “environmentally sensitive” Changes to
the environment, such as selective logging, alter the cover and/or use status of
an area. On the one hand, strictly speaking, selective logging (as an example) is
a land use change. However, on the other, actual land use Change of the area
after selective logging may not be measurable by conventional means of remote
sensing. Much of today’s land cover and land use change data is collected
through a combination of remote sensing and survey data (for selected examples
see: Edelman 1985; Harrison 1991; Geoghegan et al. 1998; Brown et al. 2000;
Walker et al. 2000; Irwin and Geoghegan 2001; Messina and Walsh 2001; Bell
and mm 2002; Walker 2003). The launch of new earth-resource observing
remote sensing technologies, such as the Landsat satellite series during the
19803, became a driving force behind the ascension of land cover and land use
change research. However, due to the frequent reliance on relatively small
spatial-scale remotely sensed data (often of Landsat TM or ETM origin), in some
studies significant land cover and land use changes may be unnoticed (Meyer
and Turner II 1992). A good example of a case where remote sensing may not
sufficiently record vital change to land cover/use regimes is the differentiation
between secondary forest cover and small holder agroforestry; this difference in
land use does not significantly alter the way many Amazonian environments
appear from space at a spatial resolution of 30 meters. Even with the potential

pitfalls involving accuracy and data sources, land cover and land use research

47

has become possible largely due to rapid advances in remote sensing and
computing technology, as well as the recognition of the need for frequently
updated data covering large portions of the globe.

The complexity of the real world versus the inherent generalization of
remotely sensed information is a potentially serious problem in land cover and
land use Change research. In addition, many social and economic theories are
not as spatially explicit as they need to be, causing the gathering of useful and
clear information from remotely sensed data to be quite difficult at times
(Geoghegan et al. 1998). Seasonality and other natural cycles can be incredibly
important in using remotely sensed data to monitor both the extent and cause of
land cover and land use Changes as phenology may hide significant human
impacts on local or even regional environments (Geoghegan et al. 1998).

Litfin (1997) raises another objection to new technology for earth
observation; she argues the source of the technology itself should be questioned.
Because science has historically been dominated by affluent males from wealthy
western nations, Litfin (1997) insists that we question the goals of remote
sensing as they exist in an unequal world. Furthermore, she criticizes global
Change science (including land cover and land use change research) because,
as she claims, human actions are frequently considered only as an afterthought
or cause of some othewvise thoroughly documented natural phenomena. Litfin’s
criticism of land use change research and the use of satellite remote sensing is
another link in a debate surrounding the ability of those who have power

(especially males) to be in control of those who do not. However, as

48

technological innovation spreads and some of the less developed countries begin
to launch earth observing satellites of their own (e.g. India’s space program) this
critique may fall out of favor. Regardless of its potential validity, this critique is not
an issue this thesis will address.

Another shortfall of land cover and land use research is the focus on the
effects of largeholders on generalized land covers and land uses. This is
especially the case in the Amazon, as many researchers have held large land
holders such as ranchers and loggers responsible for a large proportion of the
deforestation occurring there (Hecht 1984; Nepstad et al. 2002). However,
smallholders are also responsible for a large proportion of deforestation occurring
in the Amazon, as Walker et al. (2000) illustrate. Given the shared responsibility
of large and small land holders it is interesting to note the confusion surrounding
small, “shifted” land holders who practice standard fallow agriculture, and small
“shifting” land holders practicing swidden agriculture. This difference needs to be
briefly considered in any work focusing on colonists and smallholder agriculture

in the Brazilian Amazon.

A BRIEF DISCUSSION OF SWIDDEN AGRICULTURE

Because swidden, or slash-and-bum, agriculture is a passionately
debated topic in many academic circles, it requires a brief explanation. To the lay
person, agricultural methods on Amazonian frontiers are often thought of as
being solely swidden in practice. While this may be true in some cases, such as

agriculture practiced in many indigenous reserves (Simmons 2002), in most of

49

the frontier areas set up through government sanctioned development projects
and new settlement, swidden is only a transitory agricultural strategy at most

(Perz and Walker 2002; Walker et al. 2002; Walker 2003). However, because it
does occur in specific instances in Uruara, I will briefly discuss the process and

its implications for the landscape.

Swidden agriculture is the most common indigenous13 form of agricultural
practice in the Americas (Beckerman 1987). Swidden methods of clearing the
land to prepare it for agriculture have been practiced for thousands of years in
the new worId and old world alike (Conklin 1961; Beckennan 1987; Balée 1992;
Denevan 1992). In order for swidden systems to work well in the moist environs
of a tropical forest such as the Amazon, swidden agriculturalists, or “shifting
cultivators,” must clear the under story growth in an area, let the area sit during a
drying period, and then cut and burn the area (Beckerman 1987; Denevan 2001).
After this the newly burned area is planted, in Amazonia the crops are often
manioc, banana, maize (corn), beans, or plantains, cultivated continuously for a
period of 2 to 6 years (Beckerman 1987; Balée 1992). Following this cultivation
period, crop yields begin to decline, weeds typically take over, soil fertility
declines, and insect pests and fungal growth become major problems. Because
of the difficulty in maintaining the area’s cover to be suitable for agriculture, it is

left to fallow for a number of years. After some time, secondary forest eventually

 

1 . . . . .

3 In this case the term “IndIgenous” refers to persons of NatIve AmerIcan descent and not
persons who have settled in the area since the encouragement of Amazonian settlement by
respective government initiatives.

50

replaces swidden cultivated varieties and the process of slash-and-burn
agriculture can be repeated again (Beckerman 1987).

There are many estimates as to how much time it takes for swidden fields
to fallow out to ‘mature’ Amazonian forests, but most estimates range between
20 and 100 years (Beckerman 1987; Uhl 1987; Southworth et al. 2002). A typical
swidden cycle would let the recently cultivated land sit in fallow for up to 20
years, or possibly even longer, before the cultivators cycle back through the area
to reestablish temporary agriculture again (Alcom 1989). Swidden production
systems are extensive, requiring huge areas of land and large amounts of labor,
and because of their extensive nature have been disparaged in the past as
unsustainable (Alcom 1989).

Swidden agriculture has been under intense criticism for its
unsustainability and the long-term harm it supposedly causes to the local
environment, especially forest biomass and long-term species biodiversity
(Southworth et al. 2002). Though the transition to permanent agriculture can
cause fallow periods to become much too short (Conklin 1961; Beckerrnan
1987), and there is mounting evidence that many agricultural systems can create
an artificial land constraint (Simmons 2004), some studies show that practiced in
a more traditional system with enough available land, swidden agriculture can be
a sustainable agricultural practice if practiced ‘correctly’ (e.g. when fallow periods
are long enough, see: Colchester and Lohmann 1993). Some studies even point
to the biosphere services that swidden plots, sufficiently separated, provide for

an area. After forest is cut into fields, planted, and subsequently abandoned, the

51

resulting secondary growth often exhibits a higher diversity of under story plants,
trees, and animals (Kunstadter1987; Denevan 2001; Southworth et al. 2002).
Conklin (1961) warns that researchers who have severe criticisms of swidden
agriculture are usually scientists of temperate zone origin exhibiting a sort of
hemispheric-centrism; the idea that temperate agricultural practices should work
equally well in the tropics appears to be pervasive in some fields.

As Conklin (1961) briefly mentions, researchers have to be careful, and
avoid discounting any and all local knowledge, such as swidden agriculture
practices, because valuable information about an area can be permanently lost
when this knowledge is ignored (Alcom 1989; Scott and Walter 1993; Sandor
and Furbee 1996; Warren et al. 2003). In fact, in some soil quality and cultivar-
diversity based studies, researchers have found that indigenous knowledge of
soils and human influenced geophysical processes such as erosion are valuable
in the remediation of local environmental damage and the preservation of genetic
diversity in cultivars (Scott and Walter 1993; Sandor and Furbee 1996; Warren et
al. 2003). It seems that, more often than not, modern agricultural strategies
imported from the developed world do not work well in many tropical areas
(Zimmerer 1992; Scott and Walter 1993; Zimmerer 1993). In addition to being
wary of an anti-swidden bias in agricultural research, researchers should also be
aware that shifting cultivation is often blamed for more deforestation than it
actually causes. In fact, permanent cultivation and cattle ranching are
responsible for at least as much (if not more) deforestation as shifting cultivation

(Geist and Lambin 2001).

52

Regardless of the sustainability and viability of swidden agriculture, the
fact is that most frontier colonists in the Amazon do not practice “shifting”
cultivation, and are in fact “shifted” cultivators (Perz and Walker 2002). “Shifted”
cultivators differ from “shifting” cultivators in that shifted cultivators practice
standard small holder agriculture in stationary areas (including fallows) but are
displaced from their original location (Myers et al. 2000). Along the Transamazon
swidden agriculture iS typically used only as a land Clearing strategy or by
indigenous peoples (Moran 1981 ). In Uruara, small farmers only practice
swidden agriculture when they first arrive on a property to clear their land (Walker
et al. 2002; Walker 2003). Though disparaged by environmentalists as
unsustainable, farmers in Uruara have adopted swidden methods to their
advantage, often establishing their land Claims through its use (Moran 1984; Perz
and Walker 2002; Moran et al. 2003; Walker 2003). Despite the benefits of
swidden agriculture, most farmers in Uruara focus on more permanent
cultivation, typically a combination of annual subsistence crops, cattle pasture,

and a limited number of commercially oriented cultivars (Walker 2003).

53

CHAPTER 3: PROPERTY DYNAMICS IN URUARA

This chapter addresses the hollow frontier thesis by considering property
dynamics in Uruara with household panel data. It first considers land distribution
(derived from the 1995/1996 national Agricultural Census) nationally in Brazil, in
the state of Para as a whole, and the Municipio of Uruara over multiple time
periods. From there, the Chapter includes information that details land holding
change in the Uruara area using information from the property panel dataset.
Finally, this chapter concludes with a brief description of the spatial pattern of

land holding changes and why this pattern may be evident.

LAND CONCENTRATION

The Gini coefficient indicates how evenly a variable is distributed over its
range (Millimet and Slottje 2002; Porro 2002). In many geographic studies, this
measure of equal distribution is frequently employed to illuminate the land
concentration of properties in an area. If a location has more large holdings than
small holdings, the Gini coefficient value for that area will be a value close to one,
indicating an unequal distribution of land (Anh Tuan 2002; Porro 2002; Simmons
2004). If an area has many small properties the Gini coefficient value for that

area will be a value close to zero.

54

lntemationally, according to literature addressing development themes,
Brazil is known for its unequal distribution of wealth (Peet and Hartwick 1999).
This inequality also extends itself to the distribution of land resources, and
especially with the colonization of the Amazon this distribution has remained
unequal despite the best hopes of the proponents of colonization projects (Wood
and Wilson 1984; Hoffman 1998, as cited in Simmons 2004.; Simmons 2004).
Even so, many of the colonization projects throughout the Amazon have provided
land for farmers from areas where there is a dearth of land resources and little
interest in land reform programs. For some, it may be surprising to observe how
the national property distribution has not improved drastically in Brazil over time,
even though Amazonian land resources are so abundant. However, given the
history of land reform globally, and the political climate of Brazil during many of
the settlement pushes, it is not as shocking that the trend to unequal property
distribution remains (Simmons 2004).

Land distribution in the state of Para consolidated Slightly between 1985
and 1995 according to Gini coefficients derived from INCRA data and the
1995/1996 Censo Agropecuaro (agricultural census), with Gini values Changing
from 0.71 in 1985 to 0.82 in 1995. The increased property consolidation
experienced in Para is also evident nationally; in Brazil, Gini coefficients for the
entire country rose from 0.85 in 1978 to 0.89 in 1995 (Censo-Agropecuario
1995/1996; Simmons 2004), even with the continuing development of many
agricultural frontiers in the Amazon, covering vast amounts of land. However,

land distribution in the municipio of Uruaré during the mid-19903 was more equal

55

than the rest of the state of Para and the nation, with a Gini coefficient of 0.18 - a
nearly completely equal distribution of land (Censo-Agropecuario 1995/1996).

If land distribution over the entire state of Para is skewed toward larger
land holdings why does Uruara display the opposite trend? While there are many
potential reasons, the most significant for the relative equity in land distribution
for the area in 1995/1996 is the initial settlement project administered by INCRA
in the 19803. This area, as mentioned previously (page 9), is, for the most part,
divided into 100 hectare lots which were made available to new colonists one lot
at a time. Property is distributed more evenly in Uruara than the rest of Brazil
because the time since initial settlement has not been sufficient for settlers to
accumulate enough capital to acquire more lots or to clear enough land for their
lots to be attractive to “urban capitalists” who may want to establish ranching or
other commercial operations in the area. The initial, artificially imposed, property
structure was very even in its allocation of properties and this even distribution
has remained since the beginning of colonization in the area. A quick inspection
of the property structure in the area overlain on Landsat ETM imagery from
August 1999 shows that the majority of the lots in the Uruara area are small (not
much more than 100 hectares, Figure 5). In addition, there is not a great amount
of deforestation, and consequently (usually) colonization, beyond the boundaries
of the original INCRA lot structure (Figure 1, Figure 5).

However, across the remainder of Para the equity of land distribution is
not nearly as evident. In fact, the average number of hectares per property in

Para has increased every year since 1970, which may indicate a continuing land

56

consolidation already apparent in much of the state (Simmons 2004). Property
consolidation issues aside, land ownership in Para is much higher than it is in
Maranhao, where only 32 percent of lot holders own their lots (Censo-
Agropecuario 1995/1996). The low lot ownership rates in Maranhao have made
land consolidation more frequent and has lead to increased conflict between
small farmers and large ranchers. Low rates of ownership and consolidating
forces are partially responsible for the inequity in the distribution of land
resources and the violent land-related conflict evident there (Porro 2002). This
consolidation and dearth of land resources could create Similar problems in the
areas surrounding Uruara if the distribution of land becomes skewed toward
larger land holdings. Indeed, other areas of Para not far from Uruara, have seen
instances of intense land conflict involving indigenous populations, smallholders,
ranchers, and loggers (Simmons 2002; 2004).

Nonetheless, properties in Uruaré are still very equitably distributed, and
though there seems to be an active land market, census data do not indicate
strong property consolidation forces in the region. Even so, the farmers who have
acquired more lots tend to focus on commercially oriented production and over
time may require increased access to land resources. If commercially oriented
agriculture such as coffee, pepper, cacao, and cattle ranching continue and
increase, property consolidation may begin to take hold as farmers begin to
accumulate capital and acquire more land. This, in turn, could lead to land
conflict and land cover and land use Changes that will have significant impact on

local, regional, and global environmental health.

57

LAND HOLDING CHANGE

Land holdings on Amazonian agricultural frontiers tend to change hands
frequently, leading to property turnover rates many researchers have found
puzzling since the early 19803 (Schneider 1995). Uruara is no exception; like the
rest of the Amazon, land holdings are highly variable among small farmers. In
1996, the average lot holder in the surveyed area controlled 1.21 lots, with a
standard deviation of 0.62 (n=76, Table 1). In 2002, this average decreased
slightly to an average of 1.20 lots per lot holder (o=0.57, n=76, Table 1). The
average number of lots per property is greater than 1.0 in both surveys because
occasionally respondents control more than one lot. However, in order to do so,
most colonists must first establish themselves and amass enough capital to
acquire another lot. Due to the difficulties of frontier life, most colonists in the
Uruaré area control only one lot (see pages 9 & 17).

During 1995/1996, 16.5 percent of farmers leased, sharecropped, or
occupied their property without a title in the state Para, while 83.5 percent of
farmers owned their lots (held with a formal title), the highest rate of land
ownership on record in the state (Censo-Agropecuario 1995/1996). This
contrasts with agricultural land ownership rates nationally where 26 percent of
farmers lease, sharecrop, or occupy their lot without a title and 75 percent of
farmers own the land they cultivate (Censo-Agropecuario 1995/1996). However,
a comparison of Para with the rest of Brazil considers widely varying agricultural

strategies and includes states where land resources are scarcer than in Para.

58

Comparing lot ownership in Para with the rest of the Legal Amazon, one can see
that lot ownership rates are higher in Para than the rest of the Amazon basin.
This higher rate of ownership contributes to a more even land distribution, helps
solidify property rights, and aids those with sufficient capital in expanding their
holdings. In Uruara, the rate of land ownership is quite high even compared to
rates for the entire state of Para, with very few farmers leasing or sharecropping
the land and only 5 percent of farmers occupy their land with no legal
documentation such as a land title (Censo-Agropecuario 1995/1996).

While the climate of lot ownership in Para, and Uruara especially, may
make it easier for property owners to expand their holdings, between 1996 and
2002 there was no significant change in the number of lots controlled by property
owners in the study area (Table 1). Both losses and gains of lots were minor,
though some specific cases of large expansions and losses exist, between 1996
and 2002, there were impressive gains with the largest of seven lots, and the
largest loss being six lots. In the property panel, in which land holdings are nearly
identical, the only change was a loss of one lot (Figure 6). Although there are
definite changes between the surveys, and there is slightly less variation in the
number of lots owned (0 = 0.62 in 1996, o = 0.52 in 2002), the difference
between the averages is not significant at the 95 percent level for the panel
(Table 1). However, outside the panel there is significantly more variation, and it
appears that lots are changing hands more frequently. Another interesting

phenomena to note is that in the surveys the same number of lots were gained

59

by properties as were lost, though these gains and losses were not solely to or
from properties that were surveyed.

Small Changes in the average number of lots each property holder
controls may indicate the emergence of an active land market, where land
holdings Change hands frequently depending on many internal and external
social and economic factors. Though the land market seems to be active in
Uruara (see Figure 7), there is not any direct evidence of extensive land holding
consolidation or fragmentation (Figure 8). Because levels of property
fragmentation and concentration are equal, they even each other out (Figure 8).
This result may be due to the survey data used, which is collected considering
primarily small-holder lots, but the magnitude and proportion of gains and losses
in land holdings are very consistent with other parts of Amazonia (Browder et al.
2004). In addition to the development of a land market, the gradual Change of
land holdings in Uruara may be also be influenced by variations in the
demography of the surveyed area. Small changes in land holdings may signify
the beginning of a gradual shift toward greater land consolidation, which is
frequent in the rest of the state of Para (Walker and Homma 1996), though not in
evidence in Uruaré (Walker 1993).

Even though the average number of lots controlled by properties is not
significantly different between the 1996 and 2002 surveys (both for the panel and
for the full dataset), individual instances of Change are remarkable. Of the 76
properties in the property panel for 2002, 18 farmers Claimed they had not been

interviewed before, including one settler who arrived in 1997 and now owns five

60

lots. Even if land holders are not accumulating new property or fragmenting their
old property, an land resources change hands frequently, pointing to the
existence of an active land market in Uruara. Given this land market, certain

areas exhibit more rapidly changes in land holdings.

LOCATION OF LAND HOLDING CHANGES

Gains and losses of lots loosely follow a spatial pattern, as can be seen in
Figure 9. Upon inspection, Figure 9 makes it appear that property lot gains tend
to happen Closer to town and Closer to the Transamazon (the red line in Figure 9)
while properties lose lots further from town or the highway. Nevertheless,
because of the human tendency to see patterns in places where none actually
exist, this may not be the case (O’Hear 1989). Also, due to the intricacies and
problems associated with conducting field research in a remote area such as
Uruara, especially in regards to transportation, lots Closer to town were more
likely to be surveyed than lots further away. This is especially true of lots further
out on the colonization roads, which are often nearly impossible to get to
because roads are often not well maintained. The pattern of lot gains near town
is logical, however, because they are more desirable for development. This
desirability might lead one to expect properties along BR-230 to see the greatest
turnover due to land market effects. Indeed, increases in lot holdings do seem to
occur Closer to BR-230, though this is not a fixed outcome (lot losses can also be

seen close to the highway, Figure 9). In a general sense, lot losses seem to

61

occur further from BR-230 where land market forces and transportation issues
may make it harder to maintain practicable settlement.

While property land holdings Change frequently both close to and far from
BR-230, there are many more farmers who did not gain or lose lots between the
surveys (Figure 6). Because lot holders who retain the same number of lots
(often only one lot) are spread more or less evenly throughout the landscape and
far exceed the number of properties that experienced lot changes, it appears that
the majority of land holdings in Uruaré are stable over the six years between
surveys (Figure 6). This stability is not surprising, as acquiring lots is expensive
and often beyond the financial means of many colonists and the reasons for
decreasing land holdings are more pressing in areas with more advanced
agricultural frontiers, such as the colonization along BR-364 in the state of

RondOnia (Browder et al. 2004).

62

CHAPTER 4: PROPERTY-LEVEL CHANGES

This chapter considers the changes occurring on properties by
considering household and property level changes on properties in Uruara. The
analysis in this chapter was simplified by the fact that at this stage the household
panel and property panel are in fact identical datasets, therefore from this point
on I will make no distinction as to which dataset was used. This chapter begins
with an outline of the Changes in labor power over the households in the study
area and the implications these Changes may have for agricultural production
and land cover and land use. Next, the chapter details the average change in
property-level production of annual subsistence and perennial crops (which are
typically commercially oriented), and outlines the changes in the production of
cattle that many properties exhibit. This chapter then outlines the typical land
cover and land use strategy of properties in Uruara and how this strategy
Changes as property sizes increase. The chapter then concludes with a brief
description of a few properties which display Changes in agricultural production
and land cover and land use strategies which closely match what is typical on
properties of similar sizes.

In order to envision the conversion of a frontier area from primary forests
to an explicitly human landscape it is often helpful to assume a series of events
must occur. Though the process behind deforestation and agricultural

development in the Amazon is not normally as linear as this progression, and

63

does not follow a prescribed recipe, most observations of property-level changes
note some sort of pattern. This pattern often starts with a change from forest to
subsistence agriculture, followed by another Change from subsistence-focused
agriculture to commercially oriented crops, at both regional (Wood 1983) and
local (Perz and Walker 2002; Walker et al. 2002) levels.

This process, with a few exceptions, can be seen in Uruara as well; early
efforts by colonists were primarily focused on subsistence agriculture, and as
time has passed have become increasingly commercially oriented. Uruara,
Brazil, is therefore an example of a rapidly changing agricultural frontier
displaying a trend toward more commercially oriented crops over a surprisingly
short time span. Between 1996 and 2002, Uruara and the surrounding area have
experienced a significant decrease in the planting and production of some key
subsistence crops, while, at the same time, experiencing a slight increase in the
production of some commercial cultivars (Table 3, Table 4).

Average property population has decreased, though these decreases are
of mixed significance (95 percent level, Table 1). These population decreases
may indicate a gradual shift in average property demographic is occurring in the
area, which is typical for a developing agricultural frontier over time. A change in
farming system orientation toward commercially valuable crops and low-labor
agricultural strategies such as cattle ranching is noticeable in the typical land use
properties in the area exhibit (Table 2, Table 5). These gradual changes point
toward continuing development in the Uruara area and may indicate an eventual

change to a more urbanized than agricultural frontier area.

64

Land cover and land use, agricultural strategy, and population change,
many properties in the Uruara area are changing rapidly. With planned and
impromptu improvements to the transportation network in Uruara along the
Transamazon and into settlement roads (such as seasonal repairs of necessity),
it is likely the commercialization of agriculture will continue in the future. As more
remote areas become increasingly accessible, land values may rise near roads
and more densely settled areas and agricultural goods will become easier to
bring to marketplaces. These Changes may lead to further deforestation in areas

that remain forested due to their isolation.

LABOR

Of late, one of the most persuasive explanations of Amazonian
deforestation at local and sub-regional (county-level) scales is the household life-
cycle. According to this theory, deforestation and other land cover/use changes
can be explained by the progression of a household’s demographic structure
through time, often framed within a common series of household events
(discussed on page 17). Regardless of household demographic situation, labor
power availability is often a driving force behind the selection of frontier
agricultural strategies (Zimmerer 1993; Walker et al. 2000; Coomes and Burt
2001).

In Uruara, the location of farming households along the household life-
cycle continuum can have a noticeable impact on the land cover and land use of

both individual properties and the greater landscape (Perz and Walker 2002;

65

Walker et al. 2002). Changes in the demographic structure of individual
households in the Uruara area are important to consider when observing
concomitant change in agricultural strategies and land cover and land use
regimes. The panel survey, which collected data in 1996 and 2002, gathered
information on property populations and, through simple change analysis and t-
tests, can illuminate the significance of demographic Change in the area.

Recognizing that property population and other household attributes can
have a significant impact on land cover and land use, in addition to agricultural
strategy, it is important to place properties in Uruara along the continuum of the
household life-cycle. In 1996, there were a total of 509 people living on properties
surveyed as part of the panel dataset. Average total property population at the
time was 6.70 people with a standard deviation of 4.3 from a total of 76
properties (Table 1). In 2002, there were 407 people on the surveyed properties,
with an average total property population of 5.36 and a standard deviation of 3.93
(Table 1).

In Uruara, average total property population change between 1996 and
2002 is significant at the 95 percent level (Table 1), with a total decrease of 102
persons on properties between the two surveys. The significant decrease in
overall population between the panel surveys would be expected of a “hollowing”
frontier; as older Children move to their own properties, the elderly may pass
away, or parents age and are unable to have children, lots will gradually

consolidate under fewer owners. This process may be due to household

66

demographic Changes, or an evolving land market. In either case, population
decreases can be an indicator of frontier consolidation and maturation.

Traditionally, researchers across the globe studying phenomena in many
fields have considered male populations as the most important for continuing
agricultural viability (James 1969; Wood 1983 refers to labor in almost completely
male terms). Recently, however, studies have shown that female and child labor
are often at least as important for household survival as male populations
(Carney 1996; Rocheleau and Edmunds 1997). Regardless of the proportion of
labor females contribute, the elderty and young Children are often a drain on
household resources. Taken together, Changes in the populations of the very
young and very old can have an impact on the production of properties as
increases in their populations change the ability of the property and available
labor to meet subsistence needs. These changes, coupled with population
change in other categories (female and male populations) can have serious
implications for both property livelihood and land cover and land use in a frontier
area such as Uruara.

In 1996, the average male population on each property was 2.42 males
with a standard deviation of 1.79 from a total of 76 properties (Table 1). Of the 76
surveyed properties surveyed in 2002, there was an average of 1.79 males per
property with a standard deviation of 1.36 (Table 1).

Over the properties covered by the panel there was a decrease in total
male populations between 1996 and 2002, comprising the largest proportion of

the total population loss during this period. The lower number of males between

67

panel surveys is interesting, as the loss may indicate a significant decrease in
available labor on the surveyed lots. However, a supposed loss in labor
resources is debatable, as women and children often perform key agricultural
tasks (Carney 1996; Rocheleau and Edmunds 1997). Regardless of the decline
in male populations and their implication for labor in the area, the decline of
males is an interesting phenomenon and quite significant at the 95 percent level
(Table 1). This decrease in male labor may precede a noticeable decline in the
Clearance of forest cover on many farms, though this has yet to be evaluated.

Average female population was 1.63 females per property in 1996, with a
standard deviation of 0.96 (n = 76). In 2002, the average number of females per
property was slightly lower at 1.37 females and a standard deviation of 1.09 (n =
76).

The female populations (like male populations) also experienced declines
on many properties, and the overall trend was a slight decrease between 1996
and 2002 from 124 females to 104 females. Though a change in the female
population by 20 persons seems like a significant decline, especially given the
low total number of females on surveyed properties, the average female
population on properties did not Change significantly between 1996 and 2002 at
the 95 percent level (Table 1).

The average number of children (people 0 — 15 years of age) on the
surveyed properties in 1996 was 2.47 (n = 76) with a standard deviation of 2.79

children. In 2002, average child populations on the surveyed properties (n = 76)

68

decreased slightly to an average of 1.97 children per lot and a standard deviation
of 2.45 Children (Table 1).

Similar to female populations in the area, there was also a slight total
decrease in the population of children; Uruara experienced a decrease from 188
children in 1996 to 150 children in 2002. This decrease indicates that households
in the Uruara area may be maturing beyond the point of optimal labor force,
though this remains to be seen. The majority of properties, however, retained the
same number of children. The average number of children on each property was
not significantly different between 1996 and 2002 at the 95 percent level of
confidence (Table 1). If children are moving out of the area, it is interesting to
note where they may be moving to. While the 1996 and 2002 surveys did not
focus on where children have moved, some researchers have directly addressed
this question in other parts of the Amazon. The link between urban and rural
sectors is very strong in the Amazon, and in floodplain locations near Santarém,
Para, Brazil, many households send their children to school in urban areas, while
the rest of the family remains in rural areas (WinklerPrins 2002; 2004). While this
does occur in the areas surrounding Santarém, this is not necessarily the case in
Uruara. In other parts of the Brazilian Amazon, as children mature they seek jobs
in urban areas, or move to other lots along the agricultural frontier (Browder
1997; Browder et al. 2004).

On all the properties surveyed for the panel, the number of elderly persons
(greater than age 65) per property is generally quite low. In 1996, the average

number of elderly persons per property was 0.19 with a standard deviation of

69

0.61 (n = 76), and in 2002 the average elderly population per property remained
the nearly same at 0.21 (c = 0.57, n = 76, Table 1).

Elderly populations, which would, in addition to child populations, increase
the dependency ratio between years (and drain labor power), remained almost
exactly the same over the time period. By 2002, the total elderly population
increased to 24 people, a slight gain of 4 people in 1996. This slight increase
also conforms to a household lifecycle explanation of changes in the area, as
elderly family members may move to the frontier as a colonist household
establishes itself and life on the frontier becomes more stable. However this is
not usually the case in Uruara, as most properties had no elderly population in
1996 and maintained no elderly population in 2002. The average number of
elderly persons on each property did not change in any significant way between
surveys (or = 0.05, Table 1). Again, research has shown that many elderly people
in the Amazon move away from the frontier as they age, living in urban areas yet
remaining closely tied to their families in rural areas (WinklerPrins 2002; 2004).

Dependent populations, those who are old and very young, did not change
significantly at the 95 percent level during the six years between surveys (Table
1). However, the stability of dependants together with a decrease in male and
female working populations could mean a failing labor force with an increasing
dependency ratio for the Uruara area. In fact, the dependency ratio did increase
between 1996 and 2002 for the properties surrounding Uruara (Table 1), though

not significantly at the 95 percent level.

70

FARMING SYSTEMS

In the Uruara area farming systems are highly diverse, and include
subsistence-oriented annual crops, commercially-oriented perennial crops, and
multi-purpose cattle ranching (Walker 2003). Though commercially-oriented
crops are grown primarily for profit, a certain proportion of these crops may be
consumed on the farm for subsistence purposes. Likewise, subsistence-oriented
crops are grown primarily for consumption on the property, though some may be
sold at a profit, used to barter with neighbors, or as gifts.

Because of the high level of diversification and the large number of crops

grown in Uruara“, I will only talk about the three most ubiquitous of these crops
which fall into two categories: (1) crops of which most of the production is
consumed on the property, is subsistence crops and (2) crops of which most of
the production is sold or exchanged, i.e. commercial crops. Cattle are addressed
separately because of their multi-faceted presence on the frontier of Uruaré and

fall into both categories at different stages of frontier life.

SUBSISTENCE CROPS

The surveys conducted in 1996 and 2002 collected information concerning
the agricultural strategies of a large number of colonist farmers. Among the

information collected, subsistence agriculture (crops consumed primarily by

 

14 The 2002 panel data survey asked respondents specifically about 19 different crops and also
had an open-ended component for respondents to discuss other crops. For each perennial and
annual crop respondents were asked to estimate how much of each crop was consumed on the
property, how much was sold at market, and the price received per unit for each crop.

71

persons on each property) is among the most varied between the two surveys.
Because an evolving agricultural frontier is not an easy place to maintain a viable
farming establishment, certain crops are essential for a household’s survival.
Much of the food consumed by people on properties in the Uruara area is grown
on the property itself, which means that subsistence crops are very important. In
the study area there are many different subsistence crops of varying importance
to the livelihood of small farmers, the three most important being corn, rice, and
beans. AS mentioned above, these crops are not necessarily used solely for

subsistence (see page 71).

CORN

Throughout the world, corn is among the most important subsistence
crops. In Brazil this is no exception; the national yield of corn exceeded 3,000
kilograms per hectare in 2002 which is an increase over yields in 1996, and more
than double the yield of the entire continent of Africa (Table 6). In Uruara, the
situation is similar to the rest of Brazil and corn is considered one of the more
important food crops. However, the use of corn is different in many areas of the
Amazon than in much of North America; corn is grown for feed purposes, and
most com production is fed to chickens and pigs in the Uruara area (Arima
2004).

In 1996, the average property area to dedicated to corn cultivation was
2.34 hectares per property with a standard deviation of 2.97 hectares (n = 76,
Table 4). The average property area planted in corn decreased to 1.11 hectares

per property in 2002, with a standard deviation of 1.70 hectares (n = 76, Table 4).

72

As would be expected considering average area decreases, the total area
dedicated to corn cultivation also declined between the 1996 survey and the
2002 survey. The overall decline of land planted in corn between surveys is
93.18 hectares, nearly the size of an entire INCRA lot. While more properties
decreased their corn cultivation area than increased it by 2002, there were still a
fair number of corn cultivation increases as well (48 instances).

Corn production in both surveys was recorded in 60 kg sacks. Average
property corn production in 1996 was 28.88 sacks (or 1,732.8 kg) with a standard
deviation of 39.72 sacks (2,383.2 kg), while in 2002 average property corn
production declined to 14.99 sacks (899.4 kg) with a standard deviation of 27.56
sacks (1,653.6 kg, Table 3, Table 4). In general, most properties decreased their
corn production, though the decreases are not much more than the number of
properties maintaining no corn production between surveys. A total of 48
properties increased their corn production, with most production increases being
3,000kg or less between 1996 and 2002. Corn production displays a significant
overall decrease, totaling 63,330 kg (Table 3), and may indicate the beginning of
a transition from subsistence agriculture to commercially oriented agriculture.
This trend in agricultural strategies supports previous research by many authors
showing a ramping-up of commercial integration as a common step in the
development of a frontier area such as Uruara (Walker 1999; Coomes and Burt
2001; Camagni et al. 2002; Rudel 2002). However, though an increasing number
of properties grow commercial cultivars in Uruara, decreases in corn planting and

production could also be explained by local and regional environmental

73

 

conditions, including events such as drought, excess rain, heat, pest invasion,

etc.

RICE

Rice, in much the same way as corn, is an important subsistence crop in
much Brazil. In fact, and quite surprisingly, rice yields in Brazil are almost as high
as rice yields in Asia, at 3,326 kilograms per hectare in 2002 (Table 6). In Uruara
rice is a staple part of the colonist’s diet, which makes it a vital subsistence crop.

Average property area planted in rice was 2.30 hectares (a = 2.44) per
property in 1996 and 0.87 hectares per property in 2002 (c = 1.06, Table 4). The
decline in total rice cultivation area over all the properties surveyed between
1996 and 2002 is even more impressive than the loss of area planted in corn,
covering 108.8 hectares. The majority of properties decreased their cultivation of
rice (Table 4), though some properties did increase their rice areas (most
expansions less than two hectares).

Average rice production (also recorded in the number of 60kg sacks) over
surveyqed properties was 44.54 sacks (2,672.4 kg) per property with a standard
deviation of 43.09 sacks (2,585.4 kg) in 1996 and 16.70 sacks (1,002 kg) per
property with a standard deviation of 27.71 sacks (1,662.6 kg) in 2002 (Table 3).
Total overall rice production decreased by 126,960 kg over the surveyed
properties, as would be expected from the decline in areas planted in rice
between surveys. Between surveys, most properties decreased their rice
production with most decreases being small (2,400 kg or less), though increases

in rice production were not unimportant, and some properties increased their

74

 

production (though most increases were small at 2,400 kg or less). Even with
greater production on some properties, average rice yields experienced a

significant decrease by 2002 (or = 0.05, Table 3).

BEANS

Beans are another highly popular and ubiquitous subsistence crop in
much of Brazil. Though there are many ways to cook beans and a great many

specific bean varieties, this is an aggregated category with one exception:

soybeans are not included”. National bean yields increased between surveys

 

from 571 kilograms per hectare in 1996 to 735 kilograms per hectare in 2002
(Table 6). To illustrate the substantial nature of bean production, Brazil’s bean
yields were higher than either the continent of Africa or the continent of Asia in
2002. In Uruara, bean production covers less area than corn or rice, but still
remains a very important subsistence crop.

In 1996, the average property area planted in beans was relatively minor
when compared to the typical areas dedicated to corn and rice. Bean cultivation,
on average, covered only 0.64 hectares per property (a = 0.84, Table 4). In 2001,
average bean cultivation decreased to 0.25 hectares per property in 1996 (c =
0.46, Table 4). As would be expected given the decline in average area, total
bean cultivation decreased between 1996 and 2002 by 29.88 hectares. Only 29
properties increased their bean cultivation area, which was not enough to counter

balance the 47 properties that decreased their bean cultivation area. With the

 

1 . . . -.
5 Soybeans are a growrng land cover/use In BraZII, though on this part of the Transamazon they
are not yet in great evidence.

75

number of properties decreasing bean cultivation area and the relatively small
decrease in total area dedicated to bean cultivation it is apparent that most
properties are moving away from bean production. Average property bean
cultivation area is significantly lower between 1996 and 2002 surveys at the 95
percent level (Table 4).

Individual property bean production (measured in 60 kg sacks) averaged
4.06 sacks (243.6 kg) with a standard deviation of 5.59 sacks (335.4 kg) in 1996
and 2.18 sacks (130.8 kg) with a standard deviation of 5.59 sacks (335.4 kg) in
2002 (Table 3). Of all the properties surveyed, most decreased their bean
production by 300 kg or less. On the other hand, increases in bean production
were not uncommon, with typical increases being of the same magnitude (300 kg
or less). The overall decrease in bean production between 1996 and 2002 was
1,792.8 kg, which is not a large amount when compared to the decreases in corn
and rice production. However, the difference in average property bean production

is still significant at the 95 percent level between 1996 and 2002 (Table 3).

SUBSISTENCE CROP SUMMARY

As subsistence crops rice, beans, and corn show a Clear overall decrease
in both average lot area and average lot production (Table 3, Table 4). These
decreases are puzzling at first, as one would expect the population’s need for
sustenance would remain fairly constant over time. This stability in needs would
be even more likely as decreases in the area under cultivation and production of
subsistence oriented crops seem to outweigh the decreases in population. While

there are many explanations for this trend of subsistence crop declines, a main

76

 

reason could be the potential transition from subsistence oriented agriculture to a
more commercially oriented agricultural outlook. An agricultural strategy
displaying a decrease in subsistence crops coupled with a stable population
would also be consistent with a household lifecycle explanation; properties have
more labor power, due to more children, and can focus on more intensive
agricultural practices and commercially oriented production. In turn, households
may Choose to purchase some of their food with funds they may have earned
through commercially oriented crops or cattle ranching. Household lifecycles are
not the only explanation for a shift in farming strategies, however, and other
factors such as environmental conditions, culture, and an increase in beef
consumption may have lead to the decrease in crops that are mostly subsistence

oriented (corn, rice, beans).

COMMERCIAL CROPS

Though many properties in the region surrounding Uruara are primarily
oriented toward subsistence crops, property owners also dedicate a portion of
their cultivation area to crops that can be sold in markets for profit. These crops
are a key way for farmers to diversify their agricultural strategies to include more
flexible crops, increase earnings and, often, begin to raise livestock. In fact, as
mentioned above, cattle ranching is a common end goal for farmers in much of
the Amazon and commercial cultivars are often the small holder's best chance at
gaining enough capital to purchase livestock or hire additional labor in order to

realize this goal. As mentioned above, some of these crops may be consumed

77

on the property they are grown on, but most of the production is sold (see page

71).

COFFEE

Coffee is a major revenue generating crop in rural Brazil, with an annual
production of 399,000,000kg for the North region of the country (USDA 2002).
Coffee can be quite profitable for large and small producers alike, but is also
quite susceptible to global price fluctuations. One such price change occurred
shortly after World War II when coffee prices were so low railroad lines in Brazil
used raw coffee beans as combustible fuel for steam engines (Peet and Hartwick
1999). In 1996 and 2002, Brazil’s coffee yields outranked much of the rest of the
world, contributing to Brazil’s fame as a leading coffee producer (Table 7).
However, much of the Amazon’s coffee production is not used for export
purposes, instead being consumed at local and regional levels.

In the Uruara area, average coffee cultivation was 1.62 hectares per
property, with a standard deviation of 4.48 hectares (Table 4) in 1996. In 2002,
this average area changed to 1.31 hectares per property (a = 3.30, Table 4).
Between 1996 and 2002 there was an overall decrease of 23.91 hectares
dedicated to coffee in Uruara, though average decreases are insignificant at the
95 percent level (Table 4).

ln Uruara, it appears that coffee cultivation has become more profitable,
potentially due to increased national, regional, and global market integration,
coupled with a slight rise in coffee prices over the past two years (Lee 2002).

Coffee production in 1996 averaged 6.46 60 kg sacks (o = 21.43 sacks or

78

 

 

1,285.8 kg) per property for an average production of 387.6 kg (Table 3). In 2002
this average production for coffee increased to 9.82 sacks (589.2 kg) per
property, with a standard deviation of 18.82 sacks (1,129.2 kg, Table 3). Coffee
production has obviously increased on properties in Uruara and these increases
in production are quite high, though quite variable (Table 3). Of the relatively few
properties that produced coffee in 1996, 12 decreased their production by 480 kg
or less between 1996 and 2002. However, 36 properties increased their
production, accounting for the overall increase in coffee production of 15,2952
kg between 1996 and 2002. Most increases were fairly small at the individual
property level (60 kg or less), though a handful of properties increased their
production by more than 6,000 kg between the two years. Regardless, the
differences in average property coffee production between the surveys are not

significant at the 95 percent level (Table 3).

CACAO (AND COCOA)

Cacao, the tree from which cocoa beans are derived, can yield uncertain
benefits for smallholders with limited monetary resources. Similar to coffee,
cacao prices are highly variable and the crop takes more time than most
subsistence crops to grow to maturity. This does not stop small farmers from
growing cacao, however, because when prices are high, profits from cacao are a
powerful incentive to grow this perennial. Considering other parts of the globe, it
is easy to see that Brazil is among the leading cacao producers. However, cacao
yields are not as impressive as some other crops (such as coffee) and the

Amazon does not produce the majority of Brazil’s cacao (Table 7).

79

 

During the 1996 survey (n = 76), the average cacao area was 2.35
hectares per property, with a standard deviation of 4.94 hectares (Table 4). In
2002 this changed to an average of 1.74 hectares of cacao planted on each
property (n = 76), with a standard deviation of 4.11 (Table 4). Cacao is another
example of a slight decrease in total cultivation area over the panel dataset
between 1996 and 2002, with a decrease of 45.89 hectares during (insignificant
at 95 percent level, Table 4). Most increases in cacao cultivation area on
properties were small in size, at 11 hectares or less, though a few properties
decreased their cacao area by more than 11 hectares. Most properties did not
increase their cacao planted areas, however, and many properties did not have
cacao planted at all during both surveys.

Cacao production averaged 565.66 kg per property in 1996, with a
standard deviation of 1,199.59 kg (Table 3). During the 2002 survey, average
property cocoa production increased slightly to 567.37 kg with a standard
deviation of 1,288.81 kg. The increase in cocoa production in the area was minor
at only 109 kg between 1996 and 2002, and was not a significant increase at the
95 percent confidence level (Table 3). Whatever influences the production of
cocoa in the area, it appears that a major shift to commercial cacao production is

not in evidence in the area surrounding Uruara.

PEPPERCORN

Peppercorn production is not often considered to be a major agricultural
crop, but in Brazil that perception is certainly spurious. In 2002, Brazil’s yields of

pepper were double that of North & Central America’s, and far exceeded

80

production in Africa and Asia (Table 7). Pepper is a high-yield commercially
oriented crop, which might be quite attractive to small holder agriculturalists
attempting to begin small-scale commercial agriculture. However, pepper
production in the Amazon has had a different historical basis than many of the
other frequently grown cultivars. Introduced as a plantation crop by Japanese
immigrants in the years leading up to the Second World War, pepper plantations
were highly profitable especially as Indonesian exports of the crop declined
(James 1969:554).

During the 1996 survey, the average area planted in peppers on
properties was small at 0.65 hectares and a standard deviation of 0.85 hectares
(Table 4).This average area declined to 0.44 hectares per property with a
standard deviation of 0.83 hectares in 2002 (Table 4). Peppers do not cover a
significant area on each property, and this area does not change greatly between
surveys (Table 4). However, many properties (36) in the 1996 survey did not
have peppers planted at all.

Total pepper production over the surveyed properties decreases
noticeably from an average of 917.99 kg per property (a = 1,337.06 kg, Table 3)
in 1996 to 340.04 kg per property (0 = 740.30 kg) in 2002 (Table 3), for a total
overall decrease of 35,924 kg pepper production. This means that most
smallholders in the panel are not focusing their agricultural efforts on the
production of pepper for commercial purposes.

The decrease in pepper production may be due to opportunities lucrative

crops such as coffee and cacao offer many of the small scale commercial

81

agriculturalists in the area. In addition, coffee and cacao have on-property uses
as most household members drink coffee and juice made from cacao fruit is
frequently prepared. Another explanation for decreased pepper production could
be environmental conditions were such that yields were lower than normal in
2002. Yet another possibility is that many farmers have decided to decrease
pepper cultivation, due to its high labor costs, in favor of pasture creation and
cattle ranching as these strategies are often a more fiscally attractive use of

available land resources.

COMMERCIAL CROP S_UMMARY

Commercial crops have not changed significantly in production between
1996 and 2002 with the exception of pepper, which experienced a noticeable and
Significant decrease in production (Table 3). On the other hand, between 1996
and 2002, property production and planted area decreases are significant with
regards to subsistence crops (Table 3, Table 4).

Commercially-oriented crops yields are likely to continue to increase as
the frontier in Uruara matures. Another positive influence on the continued
production of commercially-oriented crops is the fact that the Uruara area is
becoming increasingly integrated with the rest of the region as transportation
networks improve. This could be creating a larger market area in Uruara, further
influencing the area and production of these commercial crops. However, the
average total cultivated area decrease between the surveys is interesting and
may be due to changes in growing conditions or the prices growers can get for

their products. Another potential explanation for area differences between the

82

surveys concerning commercially oriented crops could be differences in data
collection methods between the two surveys. Because the 1996 survey collected
an estimated number of plants of each of these crops and all subsequent 1996
area estimates are interpolated from these, estimates for the area of these
cultivars may be slightly biased. Furthermore, many smallholders in the Uruara
area may finally be acquiring enough capital to start raising more cattle, a

common goal of many smallholders.

CATTLE RANCHING

As an agricultural frontier matures and labor power on properties
decreases, smallholders frequently convert from subsistence agriculture to more
commercially oriented agriculture, including cattle ranching (Walker et al. 2000;
Walker et al. 2002). However, many frontier colonists acquire livestock very early

in their agricultural decision making process as the benefits of ranching cattle are

many and often hard for smallholders to ignore (Faminow 1998)”.

In 1996 the average cattle heard was 27.45 head per property with a
standard deviation of 27.24 head, while in 2002 the average was 51.95 head per
property with a standard deviation of 45.13 (Table 3). Total survey head counts
were 2,089 head in 1996 and 3,948 head in 2002. This near doubling of cattle
populations in the surveys speaks strongly of the benefits (see footnote 14) cattle
may offer the frontier colonist, and is a statistically significant increase in cattle

numbers at the 95 percent level (Table 3). This increase may have substantial

 

16 For further discussion on the potential non-monetary benefits of cattle ranching see page 17.

83

implications for property land use patterns and the orientation of agricultural
production in the area (whether production is commercial or subsistence in
nature) and mirrors the expansions in cattle herds throughout the Brazilian
Amazon (Walker et al. 2000; McCracken et al. 2002; Porro 2002).

Cattle ranching is widely variable in Uruara, and comprises a major part of
the rural agricultural economy in the area. Many farmers manage cattle that are
not their own in order to gain more capital, which they often use to establish their
own cattle herd. If the trend toward increasing cattle herds continues in Uruara as
it did in RondOnia during the 19903 (see: Margulis 2004), significant land cover
and land use impacts, especially with respect to forest cover, will occur in the

near future on this rapidly developing frontier.

CATTLE RANCHING SUMMARY

The farming system of the average colonist in Uruara changed slightly
between the two surveys. As would be typical of a maturing agricultural frontier,
lot holders in the area are growing fewer subsistence-oriented crops, and slowly
switching their agricultural production system to commercially oriented cultivars
and cattle ranching. This changing strategy is not surprising clue to the decline in
population on most properties, which in turn means that property labor power has
decreased. Cattle ranching is a good, low labor, option for maturing households.

A farming system change from subsistence to commercial orientation
would be expected of a maturing frontier and a maturing household (Walker et al.
2002). As accessibility and transportation networks improve in many remote

areas, households increase their labor power, and farmers realize an improved

84

financial situation. As this occurs in Uruara, the area will probably experience
Increased deforestation, increased pasture (discussed on page 86), and an

increase in areas dedicated commercial crops.

PROPERTY LAND COVER AND LAND USE

CULTIVATED AREA FOR ANNUAL AND PERRENIAL CROPS

Average area planted in annual and perennial cultivars obviously varies
among properties, but in 1996 the average cultivated was 9.90 hectares (o =
8.66, n = 76, Table 2) per property. Over all the surveyed properties in 1996, the
total cultivated area was 752.45 hectares. By 2002, the average area in
cultivation was 5.72 hectares (c = 6.70, n = 76) per property (Table 2). Over all
surveyed properties, 434.86 hectares of land in the Uruara area was under active
annual or perennial cultivation in 2002.

The change in areas dedicated to subsistence (primarily annuals such as
beans, rice, and corn) and commercially oriented (mainly perennials like cacao,
coffee, and peppers) crops between 1996 and 2002 is significant at the 95
percent level. It is Clearly evident that less land is being converted into active
cultivation over the surveyed properties, and that properties are not increasing
areas under cultivation. This is not surprising, and is a commonly noticed
phenomenon; as the frontier ages, more forest is cut down, but this area is

typically converted to pasture (Hecht 1993; Faminow 1998; Pfaff 1999; BrondIzio

85

et al. 2002). However, the decline in cultivated area and the concomitant decline
in crop production seems to indicate that agricultural efforts are not intensifying.

Instead, it appears that other agricultural strategies are being pursued.

FOREST AREA

Average primary forest cover changed from 74.29 hectares per property I
(c = 58.31) in 1996 to 56.52 (c = 35.31) hectares per property in 2002 (Table 2).

Forest cover over all surveyed properties (n = 76) was 5,646.3 hectares in 1996,

 

while in 2002 the total forest cover decreased (4,295.39 hectares). Similar to the
decreases in average cultivated area on properties between 1996 and 2002,
there is a significant decrease in the amount of average property primary forest
cover in 2002 (o = 0.05, Table 2).

No Amazonian researcher debates the fact that deforestation is occurring
in the Amazon, and the trend over the six years between surveys is in no way
surprising. Primary forest must be cleared for the increase in pasture land and
the slight increases in perennial agriculture which are evident in Uruara.
However, on average as of 2002, smallholders in the Uruara area had not yet
exceeded the legal limit to land clearance at 50 percent of land holdings
(assuming a 100 hectare lot average, see Table 2), though the average amount

of Cleared land is quite Close (Walker 2003).

PASTURE AREA

86

As mentioned above, cattle herds in the surveyed area experienced a
marked increase between 1996 and 2002 (Table 3), which, in turn, is likely to
have an influence on pasture in the area. Average pasture area was 27.94
hectares per property in 1996, with a standard deviation of 18.59 hectares (n =
76, Table 2). In 2002 (n = 76), the average pasture area on each property was
28.76 hectares with a standard deviation of 21.27 (Table 2). Total pasture on the
surveyed properties in 1996 was 2,123.4 hectares, while in 2002 the total pasture
was 2,186.0 hectares.

The property average for pasture area did not increase significantly (or =
0.05) between 1996 and 2002, and appears to be highly variable (Table 2). A
slight average increase in property pasture area is consistent with other
research, though in other areas it has been greater, often occurring due to
increased demand for beef (Walker et al. 2000). These small expansions are an
indication that the small holder agriculturalist in Uruara often focuses on
establishing a cattle ranching operation, and most farmers are succeeding at this.
However, because forests are difficult to clear, it is likely that many properties are
not expanding their cultivation of perennial or annual crops or their pasture area

to any great extent.

CAPOEIRA

Capoeira, or secondary forest growth, is common throughout the Amazon
(Smith et al. 1999; Sorrenson 2000; McCracken et al. 2002; Walker 2003), and

has been observed in Uruara (Perz and Walker 2002; Walker 2003). In fact,

87

capoeira can be a significant land cover on many agricultural frontiers as areas
planted in crops lose their fertility and pastures become overrun with weeds or
pests. In Uruara, capoeira is a significant land cover, and in 1996 it covered an
average of 6.82 hectares on each property (Table 2). In 2002, capoeira covered,
on average, more of each property than crop cultivation at 12.05 hectares per
property (Table 2). Capoeira areas increased significantly between 1996 and
2002 at the 95 percent level (Table 2), which is consistent with observations of

the area conducted by Perz & Walker (2002).

SUMMARY

The average 100 hectare, one-lot property surveyed in 1996 consisted of
nearly 75 hectares primary forest cover, almost 28 hectares of pasture, nearly 10
hectares cultivated area, and almost 7 hectares of capoeira (Table 2). This land
use system is typical of a frontier area; most of the property is still forested, lot
food supplies are grown on a small area, a portion of the property is covered by
land that has fallen out of use, and the less labor-intensive cattle ranching is
practiced on a small scale (Faminow 1998; Walker et al. 2000). However, six
years later, in 2002, the typical small holder land use system is arguably more
mature. The typical 100 hectare property in the 2002 survey has 56 hectares of
primary forest, 29 hectares of pasture, 12 hectares of capoeira, and 6 hectares of
cultivated area (Table 2). Again, this land use pattern is typical of a maturing
frontier moving toward a greater integration with commercial activities, more

disused land, and an increased focus on cattle ranching (Faminow 1998; Walker

88

 

et al. 2000). These trends may indicate many small properties in the area are
moving toward less labor-intensive agricultural solutions, possibly because of the
decrease in male populations and increase of the dependency ratio in the area
(Table 1).

Even though the land use system is changing as the agricultural frontier in
Uruara matures, and average forest area on lots is significantly different between
the 1996 and 2002 surveys, pasture has not experienced a significant increase
(a = 0.05) since 1996. This is interesting because average cultivation area and
primary forest cover have both experienced significant decreases, while only
capoeira has experienced a significant increase (Table 2). These small
inconsistencies are likely due to the way inter-cropping (raising two or more
crops on the same area of a property simultaneously) was handled as part of the
1996 survey. Regardless, the property land use strategies will continue to
change as Uruara continues to mature as an agricultural frontier, and if studies in
other areas of the Amazon are any indication, pasture land will continue to
increase rapidly (Hecht 1984; Schneider 1995; Walker et al. 2000; Rudel 2002).
One potential issue for further research in this area is the impact of government
subsidies, use of technology, and other support. Though very few of the
properties included in the property panel received aid of this sort (19 properties in
1996 and 11 properties in 2002), the impact of technology and financial

assistance could be substantial.

89

PROPERTY SIZE AND LAND COVER AND LAND USE
STRATEGIES

Larger properties in the 2002 survey (those controlling more than a single
lot) often have a distinctly different land use strategy than smaller, single-lot
properties. As would be expected, average property size increases with the
number of lots held (Table 1), potentially impacting the landscape in different
ways due to the special strategies largeholders have when using the land (e.g.
large cattle pasture areas). However, taking land holding size and land cover/use
into account, two general trends can be observed in Uruara: (1) as properties
increase in size, the percentage of the property covered in primary forest
decreases slightly, and (2) as properties increase in size up to three or four lots
the proportion of total property area covered by pasture and cultivars remains
fairly stable (Table 2, Table 4, Table 5, Figure 10).

Figure 10 displays the average proportional land cover and land use by
the number of lots controlled by individual farmers in the Uruara area in 1996 for
the entire survey. For this analysis, lot holdings are used in place of actual
property area due to inaccuracies in the self-reporting of respondent lot holdings.
I believe this is a reasonable approach because, in general, lot sizes are highly
uniform within the study area (see Figure 5, Figure 6, Figure 9). In 1996, land
holdings spanning up to three lots have very similar land cover/use regimes.
These smaller properties tend to be about 60 percent primary forest, 10 percent
cultivated area, and 30 percent pasture land (Figure 10). This land cover/use
regime begins to change subtly on four-lot properties, which often have

proportionally more primary forest cover and less pasture (Figure 10). Things

90

become radically different on five and six lot properties, which appear to possess

different land cover/use strategies, in particular more cultivated area, greater

proportions of capoeira and less pasture (Figure 10)". These differences on
larger properties are interesting because they run counter to the hollow frontier
thesis and, when taken at face value, seem to indicate that larger holdings do not
typically move toward cattle ranching. This may not actually be the case,
however, because there are very few five and six lot properties in the survey (in
fact, there is only one example of each in 1996) and these few instances
misreported land cover/use, tallying only land cover/use for only one lot out of
many.

In 2002, these land cover and land use regimes are noticeably different.
Proportional property land use is not nearly as similar across property sizes, and
there seems to be more varied land cover/use strategies. Single-lot properties,
which comprise the most significant portion of the survey, have much the same
land cover/use strategy as single-lot properties in 1996; they are about 75
percent primary forest, 5 percent cultivated area, and 20 percent pasture (Figure
11). However, the similarities between the 1996 and 2002 surveys end with
single-lot properties. Two, three, and four-lot properties are similar to each other,
but have far more pasture and much less primary forest cover than two, three,
and four-lot properties in 1996 (Figure 11). The larger properties, spanning from

six to eight lots, have drastically different land use regimes than large properties

 

17 For the 4 and 6 lot properties shown in Figure 10 total hectare counts are not accurate
because missing hectare count data. However, this information was included anyway because
the respondent's proportional estimation of each land cover/use class was nearly identical to this
proportional display. Also, there are only on example of a 4-lot and 6-lot farm in 1996.

91

surveyed in 1996. Pasture is the dominant land cover/use on larger properties in
2002, when in 1996 the dominant land cover was primary forest (Figure 11). In
addition, proportionally, cultivated area declined drastically by the time of the
2002 survey, further highlighting the focus on pasture land, especially on the one
eight lot property, which is completely dedicated to pasture (Figure 11). These
trends on larger properties are not surprising, especially when labor
considerations are taken into account. Because the few larger properties in this
study have the land resources for extensive agricultural operations, strategies
requiring less labor, such as cattle ranching, are attractive.

While there are a few general trends in land cover and land use on
properties as land holdings increase, there are some substantial differences
between surveys as well. In general, considering the differences between the
1996 and 2002 surveys, there is a shift in land use regimes from a highly
diversified farming system toward a greater focus on pasture land, especially on
larger properties in Uruara (Figure 11). There are a lot of potential driving forces
behind this land use strategy change, but among the most significant is the fact
that Uruara and the surrounding area are beginning to become more regionally
integrated. This change supports both the hollow frontier and household lifecycle
theories, though it is not certain which of these processes (or if other processes)
drive land cover and land use change in the area. Farmers in Uruara may be
increasing their focus on cattle ranching in order to store any capital they may
have accumulated, or as a solution to waning farm labor power if their children

have moved off the property. On average, properties seem to be focusing more

92

on cattle ranching than the cultivation of crops, a phenomena that can be found

on many properties in the area.

DETECTING CHANGE WITH THE AID OF REMOTELYSENSED
DA TA

To illustrate the changes that are commonly evident in a property's land
cover and land use over time I will now briefly describe four small properties of
different sizes in the Uruaré area. These properties are, in general, very similar to
properties of similar size in the study area in their demographic composition, land
cover and land use, crop production, and cattle ranching activities yet show

surprisingly varied agricultural strategies.

PROPERTY ONE

Individual properties differ greatly in the Uruara area, even among the few
properties that cover larger land area. This 325 hectare property included in the
panel survey, covering 4 lots, has some particularly interesting characteristics
(Figure 12). One of the most striking changes between 1996 and 2002 on this
particular property is the decline in population. Total property population
decreases from 13 in 1996 to 6 in 2002, with the biggest loss in population being
males. Coupled with this change in the demographic composition of the property
is a rapid change in land cover/use strategy. Total primary forest decreased from
300 hectares to 240 hectares coupled with a decrease in area of the property

under cultivation, from 13.95 hectares in 1996 to 6.5 hectares in 2002 (Table 8).

93

According to the survey data, pasture area supposedly declined between the
surveys as well, from 75 hectares in 1996 to 55 hectares in 2002, but given the
decrease in property size without any property disaggregation occurring, this is
probably due to errors in the self-reporting of the property owner (Table 8). On
this property, capoeira covered 5.5 hectares in 1996, with secondary growth
present on only two of the four lots. In 2002 capoeira doubled to cover 11
hectares, with secondary forest growth in evidence to some extent on all four lots
(Table 8). Regardless of capoeira and pasture area changes, the number of
cattle on this property has increased from 82 animals in 1996 to 122 animals in
2002 (Table 8). It is interesting to note that cultivated area has decreased for
most cultivars excepting cacao, and the number of cattle on the property has
increased between the surveys, which may be related to the significant declines
in property population and the concomitant increase in capoeira growth. As
people have left the property, owners are often forced to change to less labor
intensive agricultural strategies (Pichén 1997; Walker et al. 2000), in this case
cattle ranching. In fact, this particular farmer plans to maintain the same area of
the property in subsistence crops and pasture, but increase the area of the
property dedicated to perennial (commercial) crops and the number of cattle.
On this same property there were significant changes in agricultural
production between the panel surveys. Corn production increased, while other
annuals such as rice and bean production decreased (Table 8). However,
commercial cultivars were a slightly different case; the production of perennials

such as coffee and cacao increased greatly between the surveys, while pepper

94

production greatly decreased. This trend is puzzling, especially considering the
decline in property population. However, the increase in corn, cacao, coffee, and
the cattle herd is not so puzzling when hired labor is taken into account. In 1996,
this farmer did not hire any off-property labor, but by 2002 this farmer had begun
to hire labor for a daily wage. This increase in labor daily wage hire provides
made it easier for this particular farmer to increase the output of perennial crops
even while the on-farrn labor power decreased. In addition, the cattle herd
increase can provide income and capital storage for this smallholder using only
available labor.

These substantial changes in production, land cover and land use,
population, and cattle herd aside, it is impressive to view the change occurring on
this property on satellite imagery. Figure 12 shows the change that has occurred
through the use of unclassified Landsat TM and ETM imagery in a 5,4,2 band
combination. With this particular band combination heavily disturbed areas are
shown in magenta and white. This property was mostly covered by primary forest
in 1986, with most clearing occurring in the southeast corner of this property,
which is, not coincidentally, closest to the road and town. By 1991, some of this
cleared land had been revegetated, but land cover changes penetrate further into
the lot. These changes continue in 1999, and it becomes clear that the extent of
clearance and how farmers in more remote locations (northern part of figure) are
rapidly converting forest cover to other intensive uses. Seeing this simple

imagery time-series of a fairly small property it is easy to understand why land

95

cover and land use change is such an important topic of research in Amazonian

environments.

PROPERTY TWO

This property, larger than most in Uruara (over 300 ha), is shown in Figure
13. This particular property covers three lots, and is a much more stable property
as far as land cover and land use and population are concerned. Property
population decreased by one person, a child, and property area has remained
the same between the surveys, covering two lots. Land use is also more even
over time, at least with respect to cultivation area, when compared to the
property discussed above. In 1996, 4.8 hectares of the property were dedicated
to cultivation, while there were 178 hectares of primary forest, about 100
hectares of pasture, and only 5.7 hectares of capoeira. The property’s land
cover/use changed by 2002 to 4.3 hectares of cultivated area, 75 hectares of
primary forest, nearly 50 hectares of new pasture, and an incredible 89 hectares
of capoeira (Table 8). The changes in the agricultural production of this property
varied: corn, coffee, and pepper production declined between 1996 and 2002.
Cacao production increased slightly, as did rice production, but cattle declined
over the six years between the surveys (Table 8). Similar to the four lot property
mentioned above (Figure 12), this farmer planned to increase the property’s

cattle herd and the property’s area dedicated to commercial (perennial) cultivars.

96

mmvflmm

PROPERTY THREE

Small properties differ from large properties in more than just area. One
small property, shown in Figure 14 covers one 100 hectare lot. This property has
nearly as many people as the larger properties in the area with eight people in
2002, though one male left the property and an elderly person moved to the
property between 1996 and 2002. Land use strategy is more variable on this
small property than some of the larger properties, and changed greatly between
surveys. In 1996, the property’ had 6.75 hectares of cultivars, 67 hectares of
primary forest, 25 hectares of pasture, and 2 hectares of capoeira (Table 8). By
2002, this lot land use had changed with 3.5 hectares of cultivated area, 54
hectares of primary forest, 40 hectares of pasture, and only 1 hectare of capoeira
(Table 8). What is interesting is that even though the pasture on the property
nearly doubled between 1996 and 2002, cattle actually declined. This may be
due to speculation on the part of the farmer or a desire to start ranching other
people’s cattle on the property. This sort of pasture increase without a
concomitant increase in the property cattle heard has been noticed elsewhere in
the Amazon, and may be associated with land clearance in anticipation of cattle
heard increases (Faminow 1998; Walker et al. 2000).

As would be expected from this property’s decline in cultivated area, most
crops experienced a decline in production between the two surveys. Corn, rice,
coffee, and pepper production decreased in 2002 (pepper saw the largest
decrease of over 1,280 kg, Table 8). Of all cultivars, only bean production

increased between the two surveys. The Landsat TM and ETM imagery in Figure

97

14 show the land cover and land use change activity on this lot over the years,
and is quite impressive, especially considering that this property has been active
since 1974. In 1986 only a small portion in the lower left corner of the property
had been deforested, but by 1991 this disturbed area increased along the lot
boundary (Figure 14). By 1999, just before the 2002 survey, nearly half of the
property had been deforested, most of which was turned into pasture land,

according to the survey (Table 8, Figure 14).

PROPERTY FOUR

Another small property in the area, this one covering two lots and
spanning across one of the colonization roads in the area (Figure 15), was also
highly variable between 1996 and 2002. Total property population remained
stable, though there were more children and fewer men in 1996 than in 2002.
The real changes by 2002 on this 203 hectare property were realized in its
agricultural strategy. Cultivation area increased from 18 hectares in 1996 to 20
hectares in 2002, while primary forest area supposedly increased by four
hectares (from 50 hectares in 1996 to 54 hectares in 2002, Table 8). Pasture
area increased by 10 hectares to cover 110 hectares of the property by 2002 and
capoeira experienced a decrease of 5 hectares to cover 20 hectares of the
property by 2002 (Table 8). It is interesting to note that the land cover and land
use on this property are quite close to the averages for two lot properties in 2002,

shown in Figure 11.

98

This property’s production did not change significantly between the two
surveys (with the exception that cacao and coffee production and areas seemed
to switch between surveys, Table 8), which puts this particular property in a
category with many others in Uruara. For many farmers the main objective is
property maintenance; the farmer will make no major changes in agricultural
strategy but instead will focus on maintaining production and diversifying
agricultural production enough to cope with problems as they arise. This stability
can easily be seen on the satellite imagery of this property, as the land cover
does not change to any great extent between 1986 and 1999, though there may
be more secondary forest and disturbance further into the lots (Figure 15).

Because the 1996 and 2002 panel surveys focused on smallholders and
ignored larger land holders such as loggers, large ranchers, and commercial
agriculturalists, there are few lot holders in the dataset who have more than one
or two lots (Table 1). However, even over the six years between surveys there
does appear to be an active land market and properties are slowly changing.
Some land fragmentation, observed in other parts of the Amazon (Browder et al.
2004), does occur in Uruara, but it is not widespread (Arima 2004). While
changes are equally distributed between gains and losses of lots, it seems that
property fragmentation is not prevalent in the panel (Figure 8), and there may be
a slow consolidation of properties. Land concentration can cause conflict, and
runs counter to many of the Brazilian govemment’s reasons for the development
of many Amazonian agricultural frontiers (Wood and Wilson 1984; Smith et al.

1999; Porro 2002; Simmons 2002; 2004).

99

CHAPTER 5: CONCLUSIONS

With the growing recognition of the significant impacts local land cover
and land use changes have on regional and global environmental integrity,
researchers have begun to assemble a more complete picture of land cover and
land use processes (Lambin et al. 2003). This work is not complete, and many
competing explanations for land cover and land use change are presently being
tested.

Among the plethora of competing theories, two of the most frequently cited
causes of deforestation (and other land cover and land use changes) in
Amazonia are the hollow frontier thesis, and household lifecycle theory.
Interestingly, both of these explanations for land cover and land use changes
often result in a similar agricultural landscape; a frontier with larger land holdings,
very little forest cover, and a property-level emphasis on perennial crops of a
commercial nature or cattle ranching. Many of the agricultural areas of the
Amazon display this pattern of land cover and land use, as the result of a high
rate of deforestation spanning many years.

A variation on the hollow frontier has been documented in the Ecuadorian
Amazon, and outside of the Amazon in the region surrounding Séo Paulo (James
1969; Taylor 1973; Delson and Dickenson 1984; Dean 1995:233; Messina and
Walsh 2000; Rudel 2002). Also apparent in much of the Amazon is the effect of a

household’s demographic change on the land cover and land use of a property

100

(Walker and Homma 1996; McCracken et al. 2002; Perz and Walker 2002;
Walker et al. 2002; Walker 2003). Both of these processes, or a variation on
them, may be in operation in Uruara, but which of the two best captures the
actual process driving the widespread land cover and land use change in this
area? This thesis has discussed the political ecologic forces that may be driving
the development of a hollow frontier in Uruara, and the forces behind the impact
of household lifecycles on the landscape.

Household lifecycles influence agricultural frontiers in most developing
areas to some degree, but in the Amazon they are a highly significant force
shaping agricultural landscapes (McCracken et al. 2002; Perz and Walker 2002;
Walker 2003). In Uruara, household lifecycle processes have driven substantial
land cover and land use changes, especially with regard to secondary forest
(Perz and Walker 2002). The findings presented by Perz and Walker (2002) do
not necessarily translate into a direct impact of household demographics on
agricultural strategy. Changes in the frontier may indicate the development of a
hollow frontier and not a maturing frontier changing due to maturing colonist
households.

As mentioned above, a hollow frontier is an agricultural frontier where
small farmers have moved away from the frontier they originally opened, either of
their own will or through force. On a hollow frontier large land holders dominate
and agricultural tasks are frequently performed by daily laborers (James 1969;
Taylor 1973; Casetti and Gauthier 1977; Rudel 2002). A hollow frontier typically

has very few on-property residents, and instead property holders live off-property

101

in nearby towns (Rudel 2002). In most cases, property consolidation must occur
for a hollow frontier to develop, a process which has yet to occur on any
significant level in Uruara (see page 54).

Properties are not consolidating very rapidly in Uruara, and land resources
are more evenly distributed in this area than most other parts of the Amazon (see
page 54). This fact does not support the hollow frontier theory, though on its own
the lack of land consolidation does not completely eliminate the possibility of the
development of a hollow frontier in Uruara. Specifically, if land owners started
moving to urban areas (such as the town of Uruara itself), running their
properties remotely or by hiring permanent help, a hollow frontier would be likely
to develop in Uruara. Indeed, significant population decline did occur in the panel
between 1996 and 2002 (Table 1), and average daily workers hired per year on
properties has increased significantly from 30 per year in 1996 to 61 per year in
2002 (t= 3.12, n=76). Specific land cover and land use changes seem to indicate
a gradual hollowing of the frontier as well. Properties tend to have more area
dedicated to pasture, and decreasing (though insignificantly) areas dedicated to
cultivation of commercially valuable crops, primary forest cover, and subsistence
crops (Table 3, Table 4). The decline in property populations and increased daily
labor hire, coupled with trends in property land cover and land use, could be
construed to indicate the presence a hollowing process on the frontier in Uruara.

However, I would argue these changes are not drastic enough to hollow
the agricultural frontier in Uruara in the near future. In fact, the land cover and

land use changes that could indicate a hollowing of the frontier in this area are

102

not significantly different among panel properties; average property pasture land
only increased mildly, primary forest and cultivated area declined greatly (Table
2, Table 5). If pasture did not increase significantly, and primary forest cover and
cultivated area decreased significantly, in what way are farmers using the
remainder of their deforested land in Uruara? One explanation would be that
farmers may be abandoning production of any sort on parts of their property,
leaving land fallow for a period before restarting production. Indeed, capoeira
(secondary forest growth) increased significantly between 1996 and 2002 (Table
2, Table 5) indicating many farmers are letting areas become reforested.

If active use of the land on properties moved through a progression of use
and abandonment, which was suggested by some researchers in the 19803 and
early 19905 (Hecht 1984; Feamside 1990; Hecht and Cockbum 1990), the land
cover and land use strategies of farmers would be of great concern to anyone
interested in the health of local and global environments (Morello et al. 2000;
Gerbens-Leenes and Nonhebel 2002). However, there is not, at this time,
evidence suggesting this consumption of land resources is occurring any more in
Uruara than in other locales in Amazonia beyond what is normally attributed to
what many researchers term “the frontier ethic” (Walker et al. 20002687). In all
actuality, secondary forest re-growth does occur in Uruara, but its existence is
not necessarily attributable to use and discard practices and is instead related to
household lifecycles (Perz and Walker 2002; Walker et al. 2002).

In fact, much of the land cover and land use change in Uruara could be

attributable to property household demographic progression. For instance, as a

103

household matures children are likely to move off lots to work in urban areas or
to establish their own property (Browder et al. 2004). As the younger property
inhabitants move away, populations decline and labor power becomes more
limited (Pichon 1997; Walker et al. 2000; McCracken et al. 2002). At this point a
farmer has few options to maintain the property; the farmer can either hire labor
power if he or she has the capital, or can leave cleared land to fallow and
eventually return to forest.

In Uruara we see a slight, yet insignificant increase in pasture land on
properties, significant decreases in the production and planted area of all major
crops (except coffee), and a significant decrease in forest cover (Table 2, Table
3, Table 4, Table 5). In addition, cattle production increased significantly between
1996 and 2002 (Table 1). Keeping in mind the significant decrease in property
population, the land cover/use and agricultural activities on these properties are
adequately explained by household lifecycle theory. As labor has moved from
lots, in the form of children or adults (Table 1), many farmers in UnJara may have
been unable to maintain previous levels of agricultural production (Table 3).
Because of the lower labor requirements of cattle ranching (Hecht 1985; Mattos
and Uhl 1994; Faminow 1998; Walker et al. 2000), many farmers have left their
crop production land to fallow, choosing instead to ranch cattle (Table 3).

Between 1996 and 2002 there have been many changes to properties and
their demographic composition, socio-economic status, land holdings, and land
cover and land use strategy in Uruara. Though some of these changes seem to

indicate the development of a hollow frontier, this is not the case with small

104

farmers in the area. In fact, the household lifecycle is the more likely explanation
of these changes. The knowledge that land consolidation and large scale
agriculture are not very common in the Uruaré area makes it difficult to say with
any certainty that a hollow frontier is developing. Instead, household
demographic change is beginning to yield a terminal landscape that appears
quite similar to a hollow frontier, but without the widespread land consolidation.

Though hollow frontiers have been observed in many areas, there does
not appear to be a hollowing of the agricultural frontier surrounding Uruara.
Indeed, in areas of Rondonia, a part of the Amazon quite similar to Uruara in its
history but more advanced in its development, a hollow frontier has not yet
developed. Instead a combination of increasing regional integration and property
fragmentation has been observed, with cattle ranching as a common land use
(Browder et al. 2004). The development of this sort of landscape is highly likely in
Uruara, and as time passes by there will continue to be a focus on cattle

ranching and a continuing of the deforestation already so frequently observed.

105

LITERATURE CITED

Alcom, J.B. (1989). "Process as resource: The traditional agricultural ideology of
Bora and Haustec resource management and its implication for research."
Advances in Economic Botany 7: 63-77.

Alsdorf, Douglas E. (2003). "Water storage of the central amazon floodplain
measured with GIS and remote sensed imagery." Annals of the
Associatiion of American Geographers 93(1): 55-66.

Anderson, J.R., E. Hardy, et al. (1976). A land-use and land-cover classification
system for use with remote sensor data. Washington, D.C., USA, US.
Geological Survey: 28.

Anh Tuan, Nguyen Do (2002). Land use and agricultural commercialization. The
case of Na_mdinh Province (Vietnam). Ninth Conference of the
International Association for the Study of Common Property, Victoria Falls,
Zimbabwe.

Arima, E. Y. and C. Uhl (1997). "Ranching in the Brazilian Amazon in a national

context: economics, policy, and practice." Sociey and Natural Resources
10(5): 433-451 .

Arima, Eugenio (2004). Personal Communication. East Lansing, MI, USA.

Balée, W. (1992). People of the fallow: A historical ecology of foraging in lowland
South America. Conservation of Neot_ropical Forests: Working from
Traditional Resource is; K. H. Redford and C. Padoch. New York, NY,
Columbia University Press.: 35-57.

Beckerman, Stephen (1987). Swidden in Amazonia and the Amazon rim.
Comparative Farming§vstems B. L. Turner II and S. B. Brush. New York,
NY, The Guilford Press: 55-94.

Bell, Kathleen P. and Elena G. lnrvin (2002). "Spatially explicit micro-level
modeling of land use change at the rural-urban interface." Agricultural
Economics 27: 217-232.

Bounoua, L., R. Defries, et al. (2002). "Effects of land cover conversion on
surface climate.” Climatic Change 52: 29-64.

BrondIzio, Eduardo, Stephen D. McCracken, et al. (2002). The colonist footprint:
Toward a conceptual framework of land use and deforestation trajectories
among small farmers in the Amazonian frontier. Deforestation and Land
ULse in t_he Amazon. C. H. Wood and R. Porro. Gainesville, FL, USA,
University Press of Florida: 133-161.

106

Browder, J. O. (1997). gainforest cities: the urban transfonnation of Amazonia.
New York, NY, USA, Columbia University Press.

Browder, John O. (1988a). Public policy and deforestation in the Brazilian
Amazon. fllblic policies and_the misuse of forest resources. R. Repetto,
Cambridge University Press, for Wor1d Resources Institute: 247-297.

Browder, John O. (1988b). "The social costs of rain forest destruction: A critique
and economic analysis of the 'Hamburger Debate.” lnterciencia 13(3):
115-120.

Browder, John O. (2002). Reading colonist landscapes: Social factors influencing
land use decisions by small farmers in the Brazilian Amazon.
Deforestation and Land Lfie in the Amazon. C. H. Wood and R. Porro.
Gainsville, FL, University Press of Florida: 218-240.

Browder, John 0., Robert T. Walker, et al. (2001). Patterns and processes of
landscape change in the Brazilian Amazon: A longitudinal, comparative
analysis of smallholder land use decision-making. East Lansing, MI,
National Science Foundation Proposal.

Browder, John O., Randolph H. Wynne, et al. (2004). Landscape Fragmentation
and Land Cover Change in the Brazilian Amazon: Results of a Household
Level Survey of Small Farmers in Rond6nia, 1992-2002. Philadelphia, PA,
USA, Association of American Geographers Annual Meeting.

Brown, D.G., B.C. Pijanowski, et al. (2000). "Modeling the relationships between
land use and land cover on private lands in the Upper Midwest, USA."
Journal oflEnvironmergal Management 59: 247-263.

Brown, Lawrence A., Rodrigo Sierra, et al. (1994). ”Urban-System Evolution in
Frontier Settings.” Gepgraphical Review 84(3): 249-265.

Buschbacher, Robert J. (1986). "Tropical deforestation and pasture
development." BioScience 36(1): 22-28.

Camagni, Roberto, Maria Cristina Gibelli, et al. (2002). "Urban mobility and urban
form: the social and environmental costs of different patterns of urban
expansion.” Ecological Economics 40: 199-216.

Campbell, David J. and Judy Olson (1991). Framework for environment and
development: The Kite. Michigan State University: East Lansing, MI,
CASID: 30.

Carney, Judith A. (1996). Converting the wetlands, engineering the environment:
The intersection of gender with agrarian change in Gambia. Liberation
Ecologies: Environment, Development Socia_l Movements. R. Feet and M.
Watts. New York, NY, Routledge: 165-187.

Casetti, Emilio and Howard L. Gauthier (1977). "A formalization and test of the
"Hollow Frontier" hypothesis." Economic Geography. 53(1): 70-78.

107

 

 

 

Caviglia-Harris, Jill and Daniel W. Harris (2003). Using remote sensing data to
verify survey data, estimate response bias, and improve on deforestation
modeling: An application fore households in Rond6nia, Brazil. Portsmouth,
NH, USA.

Censo-Agropecuario (1995/1996). Censo Agropecuario. Brazil.

Christopherson, Robert W. (1997). Geosystems: An introduction to physical
geography. Saddle River, NJ, USA, Prentice-Hall, Inc.

Ciccantell, Paul S. (1999). "It's all about power: The political economy and
ecology of redefining the Brazilian Amazon." ThLSociological Quarterly
40(2): 293-315.

Colchester, Marcus and Larry Lohmann, Eds. (1993). The struggle for land and
the fate of the forests. Atlantic Highlands, NJ, USA, Zed Books Ltd.

Conklin, Harold C. (1961 ). "The Study of shifting cultivation." Current
Anthropology 2(1): 27-61.

Coomes, Oliver T. and Graeme J. Burt (2001). "Peasant charcoal production in
the Peruvian Amazon: Rainforest use and economic reliance." Forest
Ecology and Management 140(1): 39-50.

Dean, Warren (1995). With broadax and firebrand: The destruction of the
Brazilian Atlantic forest. Berkely, CA, USA, University of California Press.

Delson, Roberta M. and John P. Dickenson (1984). "Perspectives on landscape
change in Brazil." Journal of Latin American Studies 16(1): 101-125.

Denevan, WM. (1992). "The pristine myth: The landscape of the Americas in
1492." Annals of the Association of American Geographers 82(3): 369-
385.

Denevan, William (2001). Cultivated Landscapes of Native Amazonia and the
Andes. New York, NY, USA, Oxford University Press.

Edelman, M. (1985). "Extensive land use and the logic of the Iatifundio: a case
study in Guanacaste Province, Costa Rica." Human Ecology 13(2): 153-
185.

Ellis, Frank (1993). Peasant economics: liarm hopsehplgs and agrarian
development. Cambridge, UK, Cambridge University Press.

Endfield, G. and S.L. O'Hara (1999). "Degradation, drought, and dissent: An
environmental history of colonial Michoacan, West-Central Mexico."
Annals of the Associatpn of American Geographers 89(3): 420-442.

Erickson, CL. (2000). "An artificial landscape-scale fishery in the Bolivian
Amazon." Nature 408: 190-193.

EROS (2004). Earth Resource Observation Systems Data Center. Enhanced
Thematic Mapper Plus (ETM+) Product Description, United States
Geological Survey. 2004.

 

 

 

108

Falesi, l. (1976). Ecosystema de Pastagem Cultivada na Amazonia Brasileira,
Boletim Tenecio 1. Belém, Para, Brazil, EMBRAPA.

Faminow, Merle D. (1998). Cattle; Deforestatim and Development in the
Ama_zon: An economic, agronomic and environmental perspective. New
York, NY, USA, CAB International.

FAO (2000). Global Forest Resources Assessment 2000: Chapter I, Forest Area
and Area Change. Rome, Italy, Food and Agricultural Organization: 1-15.

FAO (2001). State of The World's Forests, Food and Agricultural Organization of
The United Nations: Forestry Department: 13.

FAQ (2004). FAOSTAT Data, United Nations Food and Agriculture Organization.
2004.

Feamside, Philip M. (1979). "Cattle yield prediction for the transamazon highway
of Brazil." lnterciencia 4(4): 220-225.

Feamside, Philip M. (1990). Environmental destruction in the Brazilian Amazon.
The Future of Amazonia: Destrpction or smnable development? New
York, NY, USA, St. Martin's Press: 179-225.

Futemma, Célia and Eduardo S. BrondIzio (2003). ”Land reform and land-use
changes in the lower Amazon: Implications for agricultural intensification.”
Human Ecology 31(3): 369-402.

Geist, Helmut J. and Eric F. Lambin (2001). What Drives Tropical Deforestation?:
A meta-analysis of proximate and underlying causes of deforestation
based on subnational case study evidence. Louvain-la-Neuve, Belgium,
LUCC lntemational Project: 116 + Introductory.

Geist, Helmut J. and Eric F. Lambin (2002). "Proximate causes and underlying
driving forces of tropical deforestation." BioScience 52(2): 143-150.

Gentry, AH. and J. Lopez-Parodi (1980). "Deforestation and increased flooding
of the upper Amazon." Science 210(4476): 1354-1356.

Geoghegan, Jaqueline, Lowell Pritchard Jr., et al. (1998). 'Socializing the pixel'
and 'Pixelizing the social' in land-use and land cover change. People and
Pixels: aning Remote Sensingand Social Science. D. Liverrnan, E. F.
Moran, R. R. Rindfuss and P. C. Stem. Washington, DC, National
Academy Press: 51-69.

Gerbens-Leenes, P.W. and S. Nonhebel (2002). "Consumption Patterns and
their effects on land required for food." Ecological Economics 42: 185-199.

Goulding, Michael, Ronaldo Barthem, et al. (2003). The Smithsonian atlas of the
Amazon. Washington, DC, USA, Smithsonian Institution Books.

Harrison, Susan (1991). ”Population growth, land use and deforestation in Costa
Rica, 1950-1984.” lnterciencia 16(2): 83-93.

109

Hecht, SB. (1985). "Environment, development and politics: Capital
accumulation and the livestock sector in Eastern Amazonia.” World
Development 13(6): 663-684.

Hecht, Susanna B. (1984). Cattle ranching in Amazonia: Political and ecological
considerations. Frontier Expansion in Amazonia. M. Schmink and C. H.
Wood. Gainesville, FL, University of Florida Press: 366-398.

Hecht, Susanna B. (1993). “The logic of livestock and deforestation in Amazonia:
Considering land markets, value of ancillaries, the larger macroeconomic
context, and individual economic strategies." Bioscience 43(10): 687-695.

Hecht, Susanna and Alexander Cockbum (1990). The Fate of The Forest:
Developers, Destroyers, anflefenders of the Amazon. New York, NY,
HarperPerennial.

Hoffman, Rodolfo (1998, as cited in Simmons 2004.). A estrutura fundiaria no
Brasil de acordo com o cadastro do INCRA: 1967 a 1998. Working Paper,
lNCRA/UNICAMP.

Houghton, R.A., J.L. Hackler, et al. (1999). "The US carbon budget: Contributions
from land-use change." Science 285(23 July 1999): 574-578.

Howarth, R.W., G. Billen, et al. (1996). "Regional nitrogen budgets and riverine N
& P fluxes for the drainages to the North Atlantic Ocean: Natural and
human influences." Bipgeochemistm 35: 75-139.

lnlvin, Elena G. and Jacqueline Geoghegan (2001). "Theory, data, methods:
Developing spatially explicit economic models of land use change."
Agriculture. Ecosystems and Environment 85: 7-23.

James, Preston E. (1969). Latin America. New York, New York, Odyssey Press.

Johnson, Charles W. (1980). The Nature of Vermont: Introduction and guide to a
New England Environment. Hanover, New Hampshire, USA, The
University Press of New England.

Johnson, Kenneth M. and Calvin Lunsford Beale (1998). "The rural rebound."
The Wilson Quarterly 22(2): 16-27.

Khazanie, Ramakant (1997). Statistics in a world of applications, Addison-
Wesley Publishing Company, Inc.

Kunstadter, Peter (1987). Swiddeners in Transition: Lua Farmers in Northern
Thailand. Comparative Farming Systems. B. L. Turner II and S. B. Brush.
New York, NY, The Guilford Press: 130-155.

Lambin, Eric F., Helmut J. Geist, et al. (2003). "Dynamics of land-use and land-
cover change in tropical regions." Annual Review of Environmental
Resources 28: 205-241.

Landsat.org (2004). Center for Global Change and Earth Observations.
(http:/llandiattlg), Michigan State University.

110

Lee, Frank (2002). Statement on Coffee. _$u_bcommittee on the Western
Hemisphere: Hopse lntemational Rela_tionsgCfiommittee. Washington, DC,
USA, United States Department of Agriculture.

Litfin, K.T. (1997). "The gendered eye in the sky: A feminist perspective on earth
observation satellites." Frontiers: A Joml of Women's Studies. 18(2):
26-47.

Mahar, Dennis J. (1979). Frontier Development Policy in Brazil: A Study of
Amazonia. New York, New York, Praeger Publishers.

Mann, Charles C. (2002). ”1491." The Atlantic Monthly March: 41-53.

Margulis, Sergio (2004). Causes of deforestation of the Brazilian Amazon.
Washington, D.C., USA, The World Bank: 77.

Marsh, George Perkins (1864). Introduction. Man and Natpre or. Physical
Geography as Modifietl by Hpman Action. Cambridge, MA, Belknap Press
of Harvard University.

Mattos, Marli Maria and Christopher Uhl (1994). "Economic and ecological
perspectives on ranching in the Eastern Amazon." World Development
22(2): 145-158.

McCracken, Stephen 0., Andrea D. Siqueira, et al. (2002). Land use patterns on
an agricultural frontier in Brazil: Insights and examples from a
demographic perspective. Deforestation and Land Use in the Amazon. C.
H. Wood and R. Porro. Gainsville, FL, University Press of Florida: 162-
192.

Meade, Melinda S. and Robert J. Earickson (2000). Medical Geography. New
York, NY, USA, The Guilford Press.

Messina, Joseph P. and Stephen J. Walsh (2000). The application of a cellular
a‘Wtomaton model for predicting_deforestation: Patterns and processes of
LULCC in the Ecuadorian Ama_zon. 4th lntemational Conference on
Integrating GIS and Environmental Modeling (GlS/EM4): Problems,
Prospects and Research Needs, Banff, Alberta, Canada.

Messina, Joseph P. and Stephen J. Walsh (2001). "2.5D morphogenesis:
Modeling landuse and landcover dynamics in the Ecuadorian Amazon."
Plant Ecology 158(1): 75-88.

Meyer, William B. and BL Turner Il (1992). "Human population growth and
global Iand-use/cover change.” Annpgl Review of Ecology and
Systematics 23: 39-61.

Millimet, Daniel L. and Daniel Slottje (2002). "An environmental Paglin-Gini."
Applied Economics Letters 9: 271-274.

Moran, Emilio F. (1981). Developing the Amazon. Bloomington, IN, USA, Indiana
University Press.

111

Moran, Emilio F. (1984). Colonization in the transamazon and Rond6nia. Frontier
mansion in Amazonia. M. Schmink and C. H. Wood. Gainesville, FL,
University of Florida Press: 285-303.

Moran, Emilio F. (1989a). Govemment—directed settlement in the 1970s: An
Assessment of transamazon highway colonization. The Human Ecology of
Tropical Land Settlement in Latin America. D. A. Scumann and W. L.
Partridge. Boulder, CO., Westview Press: 172-198.

Moran, Emilio F., Andrea Siqueira, et al. (2003). Household demographic
structure and its relationship to deforestation in the Amazon basin. People
and the environment: Approaches to linki_ng hodsehflag commpnity
s_urveys to remote sensing and GIS. J. Fox, V. Mishra, R. Rindfuss and S.
Walsh. Norwell, MA, USA, Kluwer Academic Publishers: 61-89.

Morello, Jorge, Gustavo Buzai, et al. (2000). "Urbanization and the consumption
of fertile land and other ecological changes: the Case of Buenos Aires."
Environment & Urbanization 12(2): 119-131.

Myers, Norman (1981). "The hamburger connection: How Central America's
forests become North America's hamburgers." Ambio 10(1): 3-8.

Myers, Norman, Jefferson Fox, et al. (2000). "Shifting versus shifted cultivators.”
BioScience 50(10): 846-846.

Nepstad, Daniel, Joao Paulo Coapobianco, et al. (2002). Roads in the rainforest:
Environmental costs for the Amazon. Belém , Brazil, Amazon Institute of
Environmental Research (IPAM, Brazil): 24.

O'Hear, Anthony (1989). An Introduction Mhe Philosophy of Science. Oxford,
Clarendon Press.

Peet, Richard and Elaine Hartwick (1999). Theories of Development. New York,
New York, Guilford Press.

Perz, Stephen G. and Robert T. Walker (2002). "Household life cycles and
secondary forest cover among small farm colonists in the Amazon." World
Development. 30(6): 1009-1027.

Pfaff, Alexander SP. (1999). 'What drives deforestation in the Brazilian
Amazon?" Journal of Environmental Economics and Management 37: 26-
43.

Pichdn, Francisco J. (1997). "Settler households and land-use patterns in the
Amazon frontier: Farm-level evidence from Ecuador." WorId_Development
25(1): 67-91 .

Pontius Jr., R.G., L. Claessens, et al. (2000). Scenarios of land-use change and
nitrogen release in the Ipswich watershed, Massacltpsefis. USA. 4th
international conference on integrated geographic information systems
and environmental modeling: problems, prospects, and needs for
research, Boulder, CO, University of Colorado, cooperative institute for
research in environmental science.

112

Porro, Roberto (2002). Land use, cattle ranching and the concentration of
landownership in Maranhao, Brazil. Deforestation and Land Use in the
Amazon. C. H. Wood and R. Porro. Gainsville, FL, University Press of
Florida: 315-337.

Rand McNally, & Company (2000). Atlas of World Geography. USA, Rand
McNally & Company.

Redclift, M. (1987). Sustainable development: The problem. Sustainable
Development: ExplorinLthe Contradictions. London, Routledge: 52-78.

Rocheleau, D. and D. Edmunds (1997). "Women, men, and trees: Gender,
power, and property in forest and agrarian landscapes." World
Development 25(8): 1351 -1 371 .

Rudel, Thomas K. (1997). "The paths to rain forest destmction: Crossnational
patterns of tropical deforestation, 1975-90." World_Development 25(1): 53-
65.

Rudel, Thomas K. (2002). "A tropical forest transition? Agricultural change, out-
migration, and secondary forests in the Ecuadorian Amazon." Annals of
the Associatign of American Gepgraphers 92(1): 87-102.

Sandor, J. and L. Furbee (1996). ”Indigenous knowledge and classification of
soils in the Andes of Southern Peru." Soil Science Society of America
Journal 60(5): 1502-1512.

Santos, Roberto (1984). Law and social change: The problem of land in the
Brazilian Amazon. Frontier Expansion in Amazonia. M. Schmink and C. H.
Wood. Gainesville, FL, University of Florida Press: 439-462.

Schneider, Laura C. and R. Gil Pontius Jr (2001). "Modeling land-use change in
the Ipswich Watershed, Massachusetts, USA.” Agriculture, Ecosystems
and Environment 85: 83-94.

Schneider, Robert R. (1995). Government and the economy on the Amazon
frontier. Washington, DC, World Bank: 65.

Scott, CA. and M.F. Walter (1993). ”Local knowledge and conventional soil
science approaches to erosional processes in the Shivalik Himalaya.”
Mopntain Research and Development 13(1): 61-72.

Simmons, Cynthia S. (2002). "The local articulation of policy conflict: Land use,
environment, and Amerindian rights in Eastern Amazonia." flip
Professional Geographer 54(2): 241-258.

Simmons, Cynthia S. (2004). "The political economy of land conflict in the
Eastern Brazilian Amazon." Annals of the Associa_tion of A_mericin

Geographers: 24.

Singleton Jr., Royce A., Bruce C. Straits, et al. (1993). Approaches to Sorfl
Research. New York, NY, USA, Oxford University Press.

113

Smith, Joyotee, Petra van de Kop, et al. (1999). "Dynamics of secondary forests
in slash-and-burn farming: Interaction among land use types in the
Peruvian Amazon." AgricultureLEcosystems and Environment 76: 85-98.

Smith, Nigel J.H., Emanuel Adilson S. Serrao, et al. (1995). Amazonia:
Resiliency and dynamism of the land and its people. Toyo, Japan, United
Nations University Press.

Sorrenson, CL. (2000). "Linking smallholder land use and fire activity: Examining
biomass burning in the Brazilian Lower Amazon." Forest Ecology and
Management 128(1-2): 11-25.

Southworth, Jane, Harini Nagendra, et al. (2002). ”Fragmentation of a landscape:
incorporating landscape metrics into satellite analyses of land-cover
change.” Landscape Research 27(3): 253-269.

StataCorp (2001). Stata Statistical Software: Release 7.0. College Station, TX,
USA, Stata Corporation.

Strahler, Alan H. and Arthur N. Strahler (1992). Modern Physical W.
New York, NY, John Wiley & Sons, Inc.

Taylor, Harry W. (1973). "$50 Paulo's Hollow Frontier." Revista Geographica 79:
149-166.

Toniolo, Angelica and Christopher Uhl (1995). "Economic and ecological
perspectives on agriculture in the Eastern Amazon." World Development
23(6): 959-973.

Turner ll, B.L., W.C. Clark, et al., Eds. (1990). The earth as transformed by
Man acjtion. Global and regonal changes in th_et)iosphere over the past
300 years. Cambridge, New York, Port Chester, Melbourne & Sydney.,
Cambridge University Press (with Clark University).

Turner, Matthew (1993). 'Overstocking the range: A critical analysis of the
environmental science of Sahelian pastoralism.” Economic Geography
69(4): 402-421.

Uhl, Christopher (1987). "Factors controlling succession following slash-and-burn
agriculture in Amazonia.” Jo_umal ofEcology 75: 377-407.

USDA (2002). Tropical Products: World Markets and Trade, Coffee Update,
United States Department of Agriculture. 2003.

Verheyen, K., M. Bossuyt, et al. (1999). "The land use history (1278-1990) of a
mixed hardwood forest in western Belgium and its relationship with
chemical soil characteristics." Journal of Biogeography 26: 1115-1128.

Walker, Robert, Scott. A. Drzyzga, et al. (2001). "A behavioral model of
landscape change in the Amazon Basin." Manuscript.

Walker, Robert, Emilio Moran, et al. (2000). "Deforestation and cattle ranching in
the Brazilian Amazon: External capital and household processes." World
Develooment 28(4): 683-699.

114

 

Walker, Robert T. (1993). ”Deforestation and Economic Development." Canadian
_Jpgmal of Regional Science 16(3): 481-497.

Walker, Robert T. (1999). "The Structure of Uncultivated Wilderness: Land Use

Beyond the Extensive Margin." Journal of Regional Science 39(2): 387-
410.

Walker, Robert T. (2003). "Mapping process to pattern in the landscape change
of the Amazonian frontier." Annals ofthe Associa_tion of American
Geographers 93(2): 376-398.

Walker, Robert T. and Alfredo Kingo Oyama Homma (1996). "Land use and land
cover dynamics in the Brazilian Amazon: An overview." Ecological
Economics 18: 67-80. [E

Walker, Robert T., Stephen Perz, et al. (2002). "Land use and land cover change
in forest frontiers: The role of household life cycles.” ntemational Regional
Science Review 25(2): 169-199.

Warren, A., H. Osbahr, et al. (2003). ”Indigenous view of soil erosion at Fandou
Beri, southwestern Niger." Geoderma 111(3-4): 439-456.

Wayland, Karen G., David W. Hyndman, et al. (2002). "Modelling the impact of £-
historical land uses on surface-water quality using groundwater flow and

solute-transport models." Lakes & Reservoirs: Research at Management
7: 10.

Whitney, G.G. (1987). "An ecological history of the Great Lakes Forest of
Michigan." flimal of Ecology 75: 667-684.

Williams, M. (1989). "Deforestation: past and present." Progss in Human
Geography 13(2): 176-208.

Williams, Michael (2000). "Dark ages and dark areas: Global deforestation in the
deep past." Mmal of Historical Geography 26(1): 28-46.

Williams, Michael (2001). "The history of deforestation." History Today 51 (7): 30-
37.

Wilson, Edward O. (1999). The Diversity of lip. New York, NY, USA, WW.
Norton 8 Company, Inc.

WinklerPrins, Antoinette M.G.A. (1997). "Geographical Field Notes: Land-use
decision making using local soil knowledge on the lower Amazon
floodplain." Geographical Review 81(1): 105-110.

WinklerPrins, Antoinette M.G.A. (2002). "House-lot Gradens in Santarem, Para,
Brazil: Linking Rural with Urban." Urban Ecosystems 6(1-2): 43-65.

WinklerPrins, Antoinette M.G.A. (2004). Df Social Networks and Plagt
Exchanges: Homegardeng in the Amazon. Michigan State University
Department of Geography Colloquium Series, Michigan State University.

Wood, Charles H. (1983). Peasant and capitalist production in the Brazilian
Amazon: A conceptual framework for the study of frontier expansion. mg

 

115

Dilemma of AmazoniaLDevelopment. E. F. Moran. Boulder, CO,
Westview Press: 259-277.

Wood, Charles H. and John Wilson (1984). The magnitude of migration to the
Brazilian frontier. Frontier Expansion in Ama_zonia. M. Schmink and C. H.
Wood. Gainesville, FL, University of Florida Press: 142-152.

Zimmerer, KS. (1992). "The loss and maintenance of native crops in mountain
agriculture." GeoJoumal 27(1): 61-72.

Zimmerer, KS. (1993). "Soil erosion and labor shortages in the Andes with
special reference to Bolivia, 1951-91: Implication for 'conservation-with-
development.” World Development 21 (10): 1659-1675.

Zinn, Howard (2001). A People's History of the United States: 1492 to Present=
New York, New York, Perennial.

116

APPENDIX A

117

Difference of

 

 

1996 2002
Variable Means
Y 0 Y o t a
Number of Lots Owned 1.21 0.62 1.20 0.57 -0.14 0.05
Total Population 6.70 4.30 5.36 3.93 -2.01 0.05
Male Population 2.42 1.48 1.79 1.36 -2.73 0.05
Female Population 1.63 0.96 1.37 1.09 -1.57 0.05
Child Population 2.47 2.79 1.97 2.45 -1.17 0.05
Elderly Population 0.19 0.61 0.21 0.57 0.14 0.05
Dependency Ratio 0.70 0.71 0.74 0.80 0.37 0.05

Table 1. Changes in property land holdings and property population in Uruara, 1996 -

 

n = 152, 150 dof, criticalt=1.97

 

 

2002. Statistically significant changes are represented in boldface.

118

 

 

Difference of
1996 2002
Property Area Means
Y 6 Y o t a
Cultivars 9.90 8.66 5.72 6.70 -3.33 0.05
Forest 74.29 58.31 56.52 35.31 -2.27 0.05
Pasture 27.94 18.59 28.76 21.27 0.25 0.05
Capoeira 6.82 8.07 12.05 15.37 2.63 0.05

 

 

 

n =152, 150 dof, cn'ticalt=1.97

Table 2. Changes In average hectares of land uses on propertles In Uruara, 1996 - 2002.
Statistically significant changes are represented In boldface.

119

 

 

 

Difference of
1996 2002
Crop Means
7(- o 7(- o t 0:

Corn 28.88 39.72 14.99 27.56 -2.50 0.05
Rice 44.54 43.09 16.70 27.71 -4.73 0.05
Beans 4.06 5.59 2.18 5.59 -2.07 0.05
Coffee 6.46 21.43 9.82 18.82 1.03 0.05
Cocoa 565.66 1,199.59 567.37 1,288.81 0.01 0.05
Pepper 917.99 1 ,337.06 340.04 740.30 -3.30 0.05
Head of Cattle1 27.45 27.24 51.95 45.13 4.05 0.05

 

 

 

n = 152, 150 dof, criticalt= 1.97

1a property with 700 cattle removed. n = 151, 149 dof, critical t = 1.97

Table 3. Agricultural production for selected crops of subsistence and commercial nature
in Uruara, 1996 - 2002. Units in 60kg sacks for all variables except cocoa and peppers (in
kilograms) and cattle (in number of animals). Statistically significant changes are
represented in boldface.

120

Difference of

 

 

1996
Crop Means
Y 0 7 o t or

Corn 2.34 2.97 1.11 1.70 -3. 13 0.05
Rice 2.30 2.44 0.87 1.06 4.70 0.05
Beans 0.64 0.84 0.25 0.46 -3.59 0.05
Coffee1 1.62 4.48 1.31 3.30 -0.49 0.05
(330301 2.35 4.94 1.74 4.11 -0.82 0.05
Pepper1 0.65 0.85 0.44 0.83 -1.52 0.05

 

 

 

n = 152, 150 dof, cn'ticalt=1.97

1 interpolated from number of plants according to measurements

from Walker et al. 2002 (p. 194).

Table 4. Area in hectares dedicated to selected crops of subsistence and commercial
nature in Uruara, 1996 - 2002. Statistically significant changes are represented. In boldface.

121

 

 

 

Properties Average
Forest Agriculture Pasture
Lots in Property Other
Area Area Area
Category Area
48.61 4.73 24.90 16.68
1 63 94.92
51.21% 4.98% 26.23% 17.57%
90.50 11.07 48.50 33.60
2 7 183.67
49.27% 6.03% 26.41% 18.29%
3 or 170.67 11.50 60.17 58.49
6 300.83
More 56.73% 3.82% 20.00% 19.44%

 

 

 

 

 

 

Table 5. Average area of major land covers and land uses on large and small holdings In
Uruara during 2002. Units in hectares.

122

 

 

 

 

 

 

 

 

 

 

 

 

 

Yield in Kilograms / Hectare 1996 2002
Corn
Africa 1,703 1 ,632
Asia 3,853 3,876
Europe 5,087 5,781
North & Central America 6,485 6,777
Brazil 2,697 3,011
Rice
Africa 2,310 1 ,989
Asia 3,854 3,977
Europe 5,090 5,693
North 8 Central America 5,421 6,074
Brazil 2,657 3,326
Beans
Africa 588 641
Asia 600 579
Europe 1 .222 1 ,380
North & Central America 903 1,046
Brazil 571 735

 

 

 

." Ania-unsung.”

Table 6. Brazil's production of key subsistence crops compared to other major agricultural
regions. Source: FAOSTAT (FAO 2004)

123

 

 

 

 

 

 

 

 

 

 

 

Coffee
Africa 469 365
Asia 743 961
North & Central America 630 581
Brazil 713 1 ,055
Cacao {IT
Africa 476 374
Asia 830 929
North & Central America 421 423 é_
Brazil 388 300
Pepper
Africa 629 596
Asia 541 673
North & Central America 1,009 1,214
Brazil 1,939 2,465

 

 

 

Table 7. Brazil's production of key commercially oriented crops compared to other major
agricultural regions. Source: FAOSTAT (FAO 2004)

124

 

2 8 2 mm
2 o: 2 2:
an em 2 8
m: 8 m: m:

 

  

2 F 2 N
a: e. 2 mm
2 B 2 km

2 9m 2 E a

 

  

     

0: : 0: m6
0: mm 0: wk
0: QVN 0: 00m
0: 090 m: @064

    

0000000
95me
0me8
00004530

 

cozosuoi

 

 

 

 

 

 

 

 

 

 

 

 

 

E 8.0 2 8.0 E E .0 2 8.0 2 Re .880
m: 00 .0 8 00 .0“ m: 00 .0 m: 00 .0 m: 00 .0 m: 00 .0 0: m0.“V omomo
0: 00 .0m 8 00 .0 m: 00.“ 0: m0.m 0: 00 .0 0: mm .0 m: 00 .0 omtoo
0: 00 .0 m: 00 .0 m: 00 .0 m: 00 .0 m: 00 .0 m: 00 .0 m: 00.“ mcmom
0: 00.0 m: 00 .0 a: 00.“ 0: 0mg 0: 00.“ m: 00 .0 m: 00 .0 moi
0: 00 .0 m: 00 .0... a: 00.“ m: 009 m: 00 N 8 wk .0 m: 00 .0 :80
new: EL 000: E“ 000: 2 000: mm 000: 00 000: 03 000: NE 000: mm 0:30
$0000 0“ 03000 0 F 03000 0 20000 0 90er 0 03000 n 03000 0 03000 2 0.009".
«com 009 meow 000V Noom 009 meow 009
So". 3300...". 09.5. 3.002.". 03... 5.2.9.0 0:0 3.6005

Table 8. Attributes for specific properties in the Uruara area.

125

APPENDIX B

126

 

  

‘8 .1 “7‘ El:

l'. - d
'9
\
‘j‘0510203040Kilomehrs
.3. r—i—i—i—i .

 

‘

 

 

 

 

 

 

0 750 1.500 3.000
Figure 1. Location of Uruara and study area within the Brazilian Legal Amazon. The bottom
graphic displays the location of the Legal Amazon within all of continental South America,
the middle graphic displays the location of the Legal Amazon within a regional context.
and the top graphic shows the location of Uruara and the extent of the study area. Source
data varied, including MODIS data obtained in 2001, and Landsat 7 ETM Imagery dated
August 9, 1999 (bands 5,4,2).

127

 

 

 

 

 

 

 

Figure 2. The Amazon basin with hydrologic network. The Brazilian Legal Amazon is
shown as a dotted line. Though the Tocantins river basin, directly to the east of the
Amazon basin, is often considered part of the Amazon, especially in Brazil, it is not shown
here. Source data: MODIS imagery obtained in 2001, Hydrologic network and basin
boundary from LBA-ECO project.

128

Political
World Development Organizations
Encourage Ranching

Social/Cultural
Global Development

Over Local Environment \ GIobal

 

 

 

 

\ Economlc
Unequal Development 8
Rapid Globalization

V...
Environmental
Encouraged Exploitation of Resources
by More Developed Countries
Political
Support for Large

Commercial Agricultural Properties '

    

SocIaIICuItural
Percieved Wealth of
Amazon Land Resources

 

    

 

National

 

Economlc
\ Encoura ement of

Cattle anching
Industry

 

Political
Reinforcement of Environmental

Large Property Tendencies Conversion of Forest
to Human Use

Social/Cultural
“Frontier Ethic” 8
Cattle Ranching Elite Local

0
O
O
O
O
I
O
O
I
.0
......
eI'
.......

Economlc
Difficulty of Frontier
Life

Environmental
Declining Local
Environmental Health
Figure 3. An illustration of how a Hollow Frontier comes about through a complex series of
cultural, political, environmental, and ecologic interactions at varying scales across time.
Based on the Kite model, from Campbell and Olson (1991:18).

129

 

 

€531”? .* ., _ -.
gejfiifztfzrg‘ =-~ ~
*1

.,.> _‘_. A. '
«I: 345-” '; «w- ”I“ *‘fi—a . .
._ -.

<‘vm- ,

 

Figure 4. Relief of the Uruaré area, looking Eastward toward Uruaré along the
Transamazon (BR-230). 5x vertical exaggeration makes the terrain variation more obvious.
Source: Terrain information from SRTM data at 90 meter spatial resolution resampled to 30
meters for visualization purposes. Imagery data from Landsat ETM imagery dated August
9, 1999 (5,4,2 band combination).

130

 

 

‘5

 

Figure 5. Lots in the Uruara area (between the towns of Placas in the West and Brazil Move
in the East), shown on satellite imagery. Source: Imagery data from Landsat ETM sensor
dated August 9, 1999 (5,4,2 band combination). Land parcel information digitized from
paper INCRA lot map.

131

 

A °-£5—5=“Iam [:1 No Lot Holding Change

Figure 6. Location of stable land holdings in Uruara, shown in green. Stable land holdings
occur both close to Uruara and far out on the colonization roads. Source: Land parcel
location digitized from INCRA lot map. Parcel status collected as part of survey (Browder
et al. 2001). Imagery data from Landsat ETM imagery dated August 9, 1999 (5,4,2 band
combination).

132

 

 

 

Full Dataset

 

 

 

-10+1-5 -4 -3 -2 -1 0 +1 +2 +3 +4 +5

5
50:-
5

Figure 7. Histogram of property change between 1996 and 2002. Panel data changes are
shown on the left, on the right property changes for the entire dataset are shown. Source:
Data collected via survey (Browder et al. 2001).

133

 

Frequency

 

 

 

 

20 -
3
0 0 r “’L
Stable Properties Aggregating Disaggregating
Properties Properties

Figure 8. Property changes in Uruara property panel, 1996 — 2002 Source: Parcel status
collected as part of survey (Browder et al. 2001).

 

N
A Lag-EQGIometers E L01 L033 1:] L01 Gain

Figure 9. Multiple lot gains and losses between 1996 and 2002 in the Uruara area. Lot
gains, displayed in blue, occur slightly more often close to the town of Uruara. Lot losses,
shown in red, occur in more remote areas. Source: Land parcel location digitized from
INCRA lot map. Parcel status collected as part of survey (Browder et al. 2001). Imagery
data from Landsat ETM imagery dated August 9, 1999 (5,4,2 band combination).

135

100% - _ 6

4 . 4
14 fl @ DCapoeira
(53 I] Pasture
80% _ {£6} {37 flPrimary Forest
I] Cultivation
59
60% -
80
288
40% - 137
58 92
20% - 38
24
1° 12 1,7 20
0% I I II, I I I

 

 

 

1 2 3 4 5 6
Farm Size (Lots)

Figure 10. Average proportion of land covers and land uses by property size in 1996.
Numbers within bars indicate average number of hectares per property in each land
cover/use category. Source: Parcel status collected as part of survey (Browder et al. 2001).

136

100%

 

I] Capoeira

D Pasture

a Primary Forest
D Cultivation

80%

 

 

60%

 

 

40%

 

 

 

 

20%

 

 

 

 

0% —rJ
1 2 3 4 5 6 7 8
Farm Size (Lots)

Figure 11. Average proportion of land covers and land uses by property size in 2002.
Numbers within bars indicate average number of hectares per property in each land
cover/use category. Source: Parcel status collected as part of survey (Browder et al. 2001).

137

'1 .

 

Figure 12. Property one, shown in 1986 (on the left), 1991 (middle), and then 1999 (on the
right). Colonization road shown in red. Source: Land parcel location digitized from INCRA
lot map. Parcel status collected as part of survey (Browder et al. 2001). Imagery data from
Landsat ETM imagery dated August 9, 1999 (5,4,2 band combination).

138

 

Figure 13. Property two, shown in 1986 (left), 1991 (middle), and 1999 (right). Colonization
road shown in red. Source: Land parcel location digitized from INCRA lot map. Parcel
status collected as part of survey (Browder et al. 2001). Imagery data from Landsat ETM
imagery dated August 9, 1999 (5,4,2 band combination).

139

 

Figure 14. Property three, shown in 1986 (left), 1991 (middle), and 1999 (right). Colonization
road shown in red Source: Land parcel location digitized from INCRA lot map. Parcel
status collected as part of survey (Browder et al. 2001). Imagery data from Landsat ETM
imagery dated August 9, 1999 (5,4,2 band combination).

140

 

Figure 15. Property four, shown in 1986 (left), 1991 (middle), and 1999 (right). Colonization
road shown in red. Source: Land parcel location digitized from INCRA lot map. Parcel
status collected as part of survey (Browder et al. 2001). Imagery data from Landsat ETM
imagery dated August 9, 1999 (5,4,2 band combination).

141

   

ililliijilijjijllillillii