|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
EFFECTIVENESS OF A NON-GOVERNMENTAL ORGANIZATION (NGO)
EXTENSION PROGRAM FOR FOREST CONSERVATION AND SUSTAINABLE
AGROFORESTRY IN SOUTHERN BAHIA, BRAZIL
KATHLEEN R. PAINTER
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
Kathleen R. Painter
I would like to thank my advisor, Dr. Robert Buschbacher, for inviting me to
collaborate with him on this research. Dr. Buschbacher began working with Jupara while
working for WWF and has been supporting Jupara's efforts for more than 10 years. He is
well known by both Jupara staff and community members as a respected conservationist
and an old friend. His long-standing commitment to this project contributed greatly to
the success of my research.
I would also like to thank the rest of the Buschbacher family, Margareth, Thomas
and Katharine, for spending part of their family vacation assisting me in the field.
I would like to thank the Jupara staff, Luis Souto, Mera Costa, Barbara de Carvalho
Vasconcelos, Neusa Nevis Sousa and Lucilene Mendes dos Santos, for finding the time
to introduce me to their work, their communities, and the day-to-day realities in southern
Bahia. All of them are exemplary people and work tirelessly toward the betterment of
Most importantly, I would like to thank the many Bahian families who participated
in this research and shared their time, stories, and enthusiasm with me. I would
especially like to thank Francisco Amancio dos Santos (Chico), Francisco dos Santos
(Panca), Andre Jesus da Conceicgo (Catixa), and Maurilho Jose da Silva for welcoming
me to stay at their homes while I carried out this research.
Thanks go also to UF graduate student Kye Epps for sharing her home in Ilheus
and her knowledge of doing research in Bahia.
Back at the University of Florida, I thank Dr. Karen Kainer and Dr. Peter
Hildebrand for serving as my committee members and helping to improve my research
proposal and this thesis. Special thanks go to Karen for making possible my
collaboration with Bob.
Dr. Chris Baraloto deserves a huge thank you for assisting with the statistical
I would like to thank Amy Daniels, Lucas Fortini, Dr. Mike Binford and Dr. Jane
Southworth for technical suggestions on the remote sensing component of this research.
Finally, I thank the following UF graduate students for their friendship and support:
Cara Rockwell, Meghan McGinty, Tom Henshaw, Maria DiGiano, Mandy Baily, Julie
Clingerman, Miriam Wyman, Lin Cassidy and Brian Daley.
This research was made possible by funding from the UF School of Natural
Resources and Environment, the UF Tropical Conservation and Development Program,
and St. Mary's College of Maryland.
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ................................................................................................. iii
LIST OF TABLES .................................................... ....... .. .............. viii
LIST OF FIGURES ......... ......................... ...... ........ ............ ix
A B ST R A C T ................. .......................................................................................... x
1 IN T R O D U C T IO N ............................................................................. .............. ...
2 EFFECTIVENESS OF AN NGO EXTENSION PROGRAM FOR FOREST
CONSERVATION AND SUSTAINABLE AGROFORESTRY IN TWO
COMMUNITIES IN SOUTHERN BAHIA, BRAZIL ............................................5
Introduction .................. .. ......... ....... ....... ......................... 5
A groforestry A adoption ........................ .... .................... ...................6
Integrated Conservation and Development Projects ............ .................10
R research questions ........................................ .. ...... ....... .... ... ... 12
Study Area ............. ...... ..... ..... .......... ..... .............. .........13
Jupara Project H history ............................................ ... ..... .. ........ .... 15
Selection of Study Sites............................ ................. ............... ............... 18
M e th o d s ............................................................... ................ 2 0
Description of Study Sites ............................ ........... .................. .... 20
F o rtalez a ...............................20..............................
Lagoa Santa ............................................................................. 21
F field M eth o d s ................................................................................ 2 2
D ata A n a ly sis ................................................................................................. 2 4
P participation ............................................................. 24
Use of Agroecological Practices ................................27
Multiple Regressions ...................... ......... .............28
Results ............ .. ............... ............. ...............29
Com prison of com munities ............................................................ 29
Regression Analysis of participation and family attributes on program
im plem entation ............................................................32
F ortaleza ............... .... .. ................................................. ...... 32
L ag o a S an ta ................................................ ................... ... ................3 2
U se of A agricultural Practices............................................................ ................33
F ortaleza ......................................33............................
Lagoa Santa ...................... ............ ... ............... 34
L a n d C o v e r .................................................................................................... 3 5
D isc u ssio n ........................................................................................3 7
Agroecological Practices .................................. ....................... ... ..... 37
In c o m e ........................................................................................................... 4 0
L and C over ................................... ............................41
Success of the Jupard Agroecology Project...... ........................................ 42
3 CONSERVATION AND AGRARIAN REFORM IN SOUTHERN BAHIA, THE
C A SE O F C A SC A TA ............................................................46
Introduction ................... ................ ................... 46
Jupari and Land Reform ................. ............. ........................ ...............47
Cascata, a Land Reform Settlem ent .............................................. ......50
Site Description ................. ..... ........ .......... ......... 51
M e th o d s ..............................................................................5 3
Results ............... ................................ ...............53
Participation............................ ............ 53
Adoption of Agroecological Practices.............................................. 57
L an d U se .........................................................5 7
D iscu ssion .................................................................................. 58
Agroforestry and Conservation in Cascata ....................................58
Future D directions for Jupar ...................................................................... ..... 60
4 EFFECTS OF AN NGO EXTENSION PROGRAM ON LAND USE CHANGE
ON SMALL FARM PROPERTIES IN THE ATLANTIC FOREST OF
SO U T H E R N B A H IA ........................................................................................... 62
Introduction ................................................. 62
S tu d y R e g io n ................................................................................................. 6 6
Fortaleza .............................................................................. ... ..... ...............66
The Atlantic Forest of Southern Bahia.................................. ............... 67
Remote sensing studies in the Atlantic Forest ...........................................70
M e th o d s ..............................................................................7 1
Interview s ........................................71
Participatory Mapping and Training Points ................. ................. ....72
Im age Processing ...................................................................................... ... .......72
R e su lts ................ ...........................................................................7 5
Land Use Classifications ................................................................75
Com prison w ith Interview D ata .................. ............................................... 80
D isc u ssio n ................................ ............................................................................. 8 2
Remote Sensing Analysis ................. ...................................82
Com prison w ith Interview D ata .................. ............................................... 84
Jupard Program G oals..................................................... 85
Directions for Further Study ........ .. .......... ......... .... ......86
5 C O N C L U SIO N S ..................... .... .......................... .. .... ........ .... ... .......87
L IST O F R E FE R E N C E S ............................................................................. .............. 93
B IO G R A PH IC A L SK E TCH ...................................................................... ..................99
LIST OF TABLES
2-1 Description of target agroecological practices. .....................................................17
2-2 Interview content .................. ................................ ........ .. ............ 24
2-3 Jupara sponsored activities and corresponding participation scores......................26
2-4 Descriptive statistics by groups for two communities. P=Participants, FP =
Former Participants, NP = Non-Participants.........................................................30
2-5 Multiple regression models for Fortaleza ............... .....................................32
2-6 Multiple regression models for Lagoa Santa..........................................................33
3-1 Descriptive statistics by groups for three communities .......................................55
LIST OF FIGURES
2-1 Location of study sites. Fortaleza and Lagoa Santa are shown in red. .................19
2-2 Range of participation scores in Fortaleza and Lagoa Santa.............................. 27
2-3 Component plot for principal components analysis. Component one orders
practices from ecologically destructive to ecologically beneficial. .......................28
2-4 Adoption of agroecological practices in Fortaleza............................. .............34
2-5 Adoption of agroecological practices in Lagoa Santa............................................35
2-6 Land cover distribution in Fortaleza, as reported by farmers. ................................35
3-1 Range of participation scores in Fortaleza, Lagoa Santa, and Cascata ..................54
3-2 Adoption of agroecological practices in Cascata. .................................................57
4-1 Location of study site. The blue box indicates the Southern Bahia region with
the original Atlantic Forest shown in yellow and remaining forest in green ..........68
4-2 Location of the Central and Serra do Mar Corridors..............................................69
4-3 Property sketch created by a community member. .............................................73
4-4 Land Use Classification for Fortaleza, 1986. .................... ......................... 76
4-5 Land use classification for Fortaleza, 2001 ..................................................77
4-7 Land use change in Fortaleza on properties owned by non-participants
participants in the Jupara project, and within the community forest reserve..........80
4-8 Distribution of land uses in Fortaleza as measured by classification of 2001
L a n d sat d ata ...................................... .............................................. .. 8 1
4-9 Distribution of land uses in Foratleza as reported by farmers interviewed in
2005 ................. .............. ............................. 81
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
EFFECTIVENESS OF A NON-GOVERNMENTAL ORGANIZATION (NGO)
EXTENSION PROGRAM FOR FOREST CONSERVATION AND SUSTAINABLE
AGROFORESTRY IN SOUTHERN BAHIA, BRAZIL
Kathleen R. Painter
Chair: Karen Kainer
Major Department: Interdisciplinary Ecology
This study provides an evaluation of the Jupara Agroecology Project, an integrated
conservation and development project (ICDP), as a strategy for sustainable agriculture
and forest conservation in the vicinity of a protected area. The Jupara project was
launched in 1995 in partnership with WWF, and provided extension services to 35
communities. Jupara set the following program goals: implementation of target
agroecological practices, organic certification, increased family incomes, and
maintenance of 30% of the community area in forest and 40% in agroforestry systems.
This study addresses the following research questions: Are families that
participated in the Jupara project meeting these goals? How do family dynamics and size
of landholding influence project implementation? Are project results consistent in
different types of communities?
Three communities were selected for analysis. A total of 65 household interviews
were conducted with participating and non-participating families. GPS points and basic
vegetation data for five land cover types were collected for each household. A
participation score was assessed for each family and multiple regressions were used to
analyze the effect of participation and other socioeconomic variables on program
outcomes such as use of agroecological practices, farm income, and forest and
agroforestry cover. For one community, analysis of Landsat imagery was used to verify
land cover distributions and to assess change in land cover over time.
Results of interview data show that participation in the Jupara program positively
influences the implementation of agroecological practices and farm income, but does not
influence land cover distribution or the maintenance of forest reserves by individuals.
Successfully implemented agroecological practices included the use of organic
fertilizers and composting, contour erosion barriers, diversification of agroforestry
systems, and the elimination of burning and agrochemicals.
For two of three communities, much of the forest cover was converted to
agroforestry long before the Jupara project began. One community reported an average
of 8% forest cover, and the second reported 35%, including a community reserve. Many
farmers maintain no forest reserve. The third community maintains 40% of its total area
as a community-managed reserve.
Analysis of Landsat data showed that the overall distribution of forest, agroforestry,
and open areas remained almost constant between 1986 and 2001; however, forested
areas increased on participants' lots, indicating that participants may be more likely to
allow forest regrowth to create forest reserves.
All around the world, conservationists have observed that long-term and large-scale
conservation of endangered ecosystems will require the participation of their residents
(Murniati et al. 2001; Schwartzman et al. 2000a; Schwartzman et al. 2000b).
Conservation programs have used various strategies to enlist local participation in
conservation efforts, including strategies for sustainable agriculture, extraction of non-
timber forest products, ecotourism, and payments for environmental services. Projects
which include both a conservation component and an economic development component
have come to be known as "Integrated Conservation and Development Projects" or
ICDPs have been criticized for being expensive to implement and for focusing on
development and producing too few tangible conservation results (Brandon et al. 1998;
Brechin et al. 2002). A close examination of literature on such projects, however, reveals
that few ICPDs have been rigorously evaluated. To understand the most effective ways
to involve local people in conservation efforts, it is important to understand the strengths
and weaknesses of past projects, as well as the results these projects have produced or
failed to produce.
This thesis developed out of the long-term collaboration between Dr. Robert
Buschbacher and a small conservation NGO in Ilheus, Brazil, Jupara, Assessoriapara o
Desenvolvimento Agroecologico de Comunidades Rurais (Organization for
Agroecological Development in Rural Communities), and the practical need for an
evaluation of some of Jupara's ongoing projects. Jupara works primarily with land
reform communities, and also some traditional rural communities, throughout the
Atlantic Forest of Southern Bahia. Land reform communities are communities
established relatively recently, within the last 30 years, through a government land reform
program which donates farmland to landless workers. Traditional communities also exist
in this region. In these cases, land has been passed down from father to sons for many
In 1995, Jupara began collaborating with WWF to promote diversified
agroforestry, organic agriculture, and forest conservation in communities in the buffer
zone of the Una Biological Reserve. Concrete outcome goals were set at the start of the
project, including the implementation of organic agricultural practices and the
maintenance of 30% of each community's area in natural forest reserves and 40% in
agroforestry systems. As with many conservation projects, funding constraints
prevented any rigorous evaluation or monitoring of the extent to which these outcome
goals were met (Buschbacher in prep.).
The Jupara project has been ongoing for more than 10 years and offers a suitable
example for an in-depth analysis of the outcomes of a conservation and development
project in a critical conservation area, the Atlantic Forest. This research examines the
success of the project at the level of individual families in several different communities.
Household interviews were carried out in three communities, and a remote-sensing
analysis of land cover change at the property level was conducted in one of the three
This thesis is presented in three parts. Chapter 2 was written as an independent
document, suitable for eventual publication. It focuses on the contribution of this
research toward a better understanding of the function of integrated conservation and
development projects in the buffer zones of protected areas, as well as innovation
adoption by smallholders. Within this framework, chapter two describes the Jupara
project in greater depth, including its outcome goals, and presents interview data
collected in two of the three study communities, Fortaleza and Lagoa Santa. Fortaleza is
a 30-year-old land reform settlement and Lagoa Santa is a traditional agricultural
community that has been in place for many generations. Farmers in these communities
produce a variety of products in mixed agroforestry systems, including rubber, cacao,
cloves, palm fibers, and fruit. Both participating and non-participating families were
interviewed about their participation in the project, their use of the project's target
agroecological practices, their land use patterns, including implementation of
agroforestry and conservation of native forests, and a number of socioeconomic variables
which might affect their ability to meet the project goals.
Chapter 3 is intended to complement Chapter 2 and to give readers an idea of
Jupara'a future directions. Data was collected in a third community, Cascata, using the
same methods presented in Chapter 2. Cascata's case is presented separately because this
community is very different and not easily compared to the other two. Cascata is a new
land reform community, about 8 years old, and entered the Jupara project more recently.
The community is situated on a former cacao plantation, and rather than each family
settling on an individual parcel of land, families live in a central area, with production
areas located around the perimeter of the community. Farmers produce cacao almost
exclusively. Each family is assigned a small parcel of cacao to harvest, but the
processing and sale of cacao takes place collectively. This chapter focuses on Jupara's
participation in the agrarian land reform movement. Recently, Jupara has focused their
efforts on such newer land reform communities, often on cacao estates, and work in these
communities offers opportunities and challenges which are distinct from those found in
Chapter 4 is also intended as an independent article suitable for publication. This
chapter presents an analysis of land use change in Fortaleza using Landsat imagery from
1986, before the implementation of the Jupara project, and 2001, the most recent image
available. GPS points collected in the field were used to identify areas of forest,
agroforestry and forest regrowth, and cleared areas such as annual crops or pasture.
Maps provided by INCRA, the federal agency responsible for land reform, were used to
identify individual properties within the community. Land use change on participants'
properties is compared with non-participants' and former participants' properties, and the
community forest reserve area is also assessed. Land use observed on the imagery is
compared to land use data collected through interviews.
Chapter 5 presents a brief conclusion synthesizing the results and conclusions of
the three previous chapters.
EFFECTIVENESS OF AN NGO EXTENSION PROGRAM FOR FOREST
CONSERVATION AND SUSTAINABLE AGROFORESTRY IN TWO
COMMUNITIES IN SOUTHERN BAHIA, BRAZIL
In many developing countries, agriculture is the principal livelihood of the rural
poor, accounts for a large percentage of land use, and is probably the single most
powerful influence on environmental quality (Scherr 2000). The interrelated themes
of agricultural growth, environmental quality, and rural poverty alleviation have been
called the "critical triangle" of development objectives (Vosti et al. 1997). As rural
populations grow, pressure on the natural resource base increases, leading to both
environmental degradation and declining food security and human health. This is
especially true where people's access to land is limited to poor quality or
environmentally fragile lands (Scherr 2000).
The "critical triangle" becomes even more important in the regions surrounding
protected areas in the tropics. Parks and reserves can preserve only a very small
percentage of fragile or endangered ecosystems, in the late 1990s, reserves
encompassed around 5% of land worldwide, according to one estimate (Terborgh
1999). As Schwartzman et al. (2000b) point out, this small and often fragmented
percentage is unlikely to conserve much biodiversity in the long term, and, as they
state, "the rest of the forest is already inhabited." Thus efforts to protect more
biodiversity than that contained in parks will depend on local people and their ability
to prosper in and around protected areas on a sustainable basis.
Working within this political and economic reality, projects have emerged to
work with communities in the buffer zones of protected areas to improve natural
resource management practices and intensify production systems in a way that
increases incomes but directs use away from the core protected zone (Brechin et al.
2002; Murniati et al. 2001). These projects are known as "Integrated Conservation
and Development Projects," or ICDPs.
The conservation component of an ICDP might include protecting forests or
wildlife within a community, or avoiding the use of a nearby national park for
hunting or extraction of resources. The development component usually includes a
production strategy to help improve the livelihoods of the participants in an
environmentally benign way. Development projects often include ecotourism or
This study presents a comprehensive evaluation of an ICDP for forest
conservation and sustainable agroforestry, the Jupara Agroecology Project in
Southern Bahia, Brazil. The study evaluates the adoption of agroforestry practices by
families in two communities as well as the patterns of forest conservation or
deforestation both on individual family lots and community-owned property. This
research helps to better explain the effects of the intervention of an ICDP on the
adoption of agroforestry practices by individual families, as well as family and
community commitment to forest conservation.
Literature on agroforestry adoption evaluates farmers' abilities to implement
agroforestry systems and other conservation agricultural practices. Adoption studies
sometimes evaluate the effects of a specific program or intervention (Bannister &
Nair 2003; Browder & Pedlowski 2000; Browder et al. 2005; Neupane et al. 2002).
Many studies either assume that all farmers have been exposed to some specific new
practice and correlate adoption of new practices with socioeconomic variables, or
interview only adopters of a particular practice (Bannister & Nair 2003; Cooms &
A review of agroforestry adoption papers, published between 1995 and 2001,
by Pattanayak et al. (2003) reports that participation in extension programs was
considered as an explanatory variable in only 10 of the 32 studies reviewed, but
positively influenced adoption in 9 of these 10 cases. Adesina and Chianu (2002) in
Nigeria, and Boahene et al. (1999) in Ghana both demonstrated that contact with an
extension agent had a significant positive effect on farmers' adoption decisions. In
another adoption study carried out in Nepal, Neupane et al. (2002) found that
membership in an NGO positively influenced agroforestry adoption by males but
negatively influenced adoption by females.
A Brazilian example of a project-based agroforestry study analyzes the
Rond6nia Agroforestry Pilot Project (Browder & Pedlowski 2000, Browder et al.
2005). The 2000 study compared successful agroforestry plots established through the
project with unsuccessful ones and identified some of the causes of plot failure,
including biophysical variables such as inappropriate plot location, and household-
level variables such as lack of labor or family illness. The 2005 follow-up study
compared successful participants with non-participants, and found that non-
participating neighbors of successful participants were also likely to adopt
Most studies on adoption of agroforestry practices by smallholders in the
tropics have focused not on a specific project intervention but on the relationships
between other socioeconomic variables and a family's likelihood to adopt new
practices. Variables that generally have a positive influence on a farmer's ability to
adopt agroforestry practices include age of household head (Bannister & Nair 2003),
education (Boahene et al. 1999; Neupane et al. 2002), size of landholding (Adesina &
Chianu 2002; Cooms & Burt 1997; Scherr 1995), availability of labor (Adesina &
Chianu 2002; Cooms & Burt 1997; Scherr 1995), male gender of farmers (Adesina &
Chianu 2002; Cooms & Burt 1997; Scherr 1995), membership in community
organizations (Browder & Pedlowski 2000; Neupane et al. 2002), socioeconomic
level (Scherr 1995), agriculture as the primary source of income (Simmons et al.
2002), and secure land tenure (Adesina & Chianu 2002; Bannister & Nair 2003;
Simmons et al. 2002).
Many of these studies take a general survey of a large number of households
and compare groups of households, often adopters and non-adopters (Bannister &
Nair 2003), or project participants with non-participants (Browder et al. 2005), or
both (Browder & Pedlowski 2000; Neupane et al. 2002). This study offers a more in-
depth assessment of a smaller number of households. Rather than comparing groups,
we developed indices, based on detailed interviews, to measure the extent of
participation and adoption by both participating and non-participating families. In
addition, the effects of several other socioeconomic variables are considered, drawn
from this literature on the adoption of agroforestry practices.
Several authors have identified a need for further research on agroforestry
adoption in Latin America (Mercer 2004; Pattanayak et al. 2003). Despite the
prevalence of agroforestry in Southern Bahia, few agroforestry adoption studies have
been published on this region. Several studies simply describe the biophysical
characteristics of common agroforestry systems in the region (Alvim & Nair 1986;
Rice & Greenberg 2000). A few studies focus on the social aspects of agroforestry
systems and adoption decisions. Alger and Caldas (1994) surveyed Bahian cacao
farmers on their land use practices and attitudes toward conservation, and found that
the poorest farmers on the smallest lots or with the poorest soils were more likely
than farmers on better quality lands to continue to replace forests with subsistence
crops rather than agroforestry, and are more likely to use bananas or fruit trees to
provide shade for cacao than to use the traditional cabruca system, which maintains
native trees for shade cover. Johns (1999) interviewed cacao producers on medium-
sized farms on their attitudes toward the importance of shade cover, and found that
farmers resisted removing shade trees, even when encouraged to do so by a
government extension program. Trevizan (1999) interviewed cacao farmers in newly
settled and older communities with respect to "environmental risks," including
deforestation and destructive agricultural practices, and concluded that new
settlements do not present significantly more environmental risks than older ones in
terms of agricultural practices and deforestation.
In general, agroforestry adoption studies report farmers' adoption of one or
several specific practices. While the practices themselves may be environmentally
beneficial in terms of soil conservation, pest control, or reduction of agrochemical use
within the agricultural system, these studies do not relate the use of sustainable
practices within an agricultural system to a broader level of conservation across a
farmer's property or across a community-wide landscape. This study takes the idea
of agroforestry adoption one step further, by asking whether or not farmers who adopt
agroforestry practices are likely to also conserve forests.
Integrated Conservation and Development Projects
Published reviews of ICDPs worldwide have yielded mixed results, and in
many cases, criticism of the ICDP concept as a way to manage protected areas. Many
have argued that conservation programs should not "become diluted with social goals
like poverty reduction and social justice (Brechin et al. 2002)" and should simply
focus on nature protection and leave the work of social development to other
organizations (Brandon et al. 1998; Kramer et al. 1997). "What we know," writes
(Brandon et al. 1998), "is that alleviating poverty will not necessarily lead to
improvements in biodiversity conservation." Terborgh (1999) asserts that a
successful ICDP will actually increase pressure on a protected area, by attracting
newcomers to its perimeter.
Case studies support many of these claims. Kamugisha et al. (1997) published
a collection of case histories of various programs surrounding four national parks in
Kenya and Uganda and didn't find a single conservation and development program
that they felt supported local livelihoods successfully enough to slow encroachment
into parklands for grazing, water sources, and farmland. In a similar review of 20
ICDPs in Indonesia, Wells et al. (1999) stated that while monitoring and performance
data are incomplete, very few Indonesian ICDPs can realistically claim to have
enhanced biodiversity, and the most promising projects are unlikely to be financially
feasible over the long term. The authors suggest "a strategic reorientation" of
international biodiversity conservation funding. Similar assertions are made by
Peters (1998) with reference to a project in Madagascar and Blom (1998), citing
efforts in the Central African Republic.
Other authors are more optimistic. Mumiati et al. (2001) demonstrated that
families maintaining highly diversified agroforestry systems in the buffer zone of
Kerinci National Park in Indonesia were less dependent on park resources than
families with less diversified farms. Albers and Grinspoon (1997) offer a comparison
of two national parks in China, one using a policing and punishing strategy, and the
other using an ICDP involving agroforestry systems of rubber, tea, and the resin
producing tree Amomum villosum along the park's perimeter. The ICDP controlled
park degradation more effectively than the policing strategy. Browder (2002) applied
the ICDP concept to the Rondonia Agroforestry Pilot Project in the Brazilian
Amazon, and reported that in 50% of communities, the projects delivered tangible
economic benefits. Preliminary environmental data show that regions adopting the
agroforestry plots promoted by this ICDP had a lower deforestation rate than
neighboring regions from 1995-1998.
Both supporters and critics of Integrated Conservation and Development
Projects identify a need for improved evaluation and monitoring of projects (Brechin
et al. 2002; Browder 2002; Johnson et al. 2001; Kamugisha et al. 1997; Kremen et al.
1994; Wells et al. 1999). Many projects simply have not been in practice long
enough to determine their long term effects. If ICDPs are in fact an ineffective means
of managing protected areas, then the next important step is to ask why. Is the IDCP
approach a failure, or have we yet to develop adequate programs to pursue
conservation and development goals equally well (Brechin et al. 2002)?
This research is unique in that it provides an in-depth evaluation of an ICDP by
considering both the effects of the project intervention and of other causal variables
drawn from agroforestry adoption literature. The analysis is carried out at the level of
individual families in two communities and includes families who participated in the
ICDP and families who did not. The Jupard project has been ongoing for more than
10 years in communities throughout Southern Bahia, and offers a suitable example
for measuring the long-term outcomes for both conservation and development in a
variety of types of communities.
This project provides a comprehensive evaluation of the outcomes of one
ICDP, the Jupard Agroecology Project, carried out in the region of the Una Biological
Reserve in Southern Bahia, Brazil. Jupari, a Brazilian NGO, launched an extension
program in 1995, in partnership with WWF, to promote diversified agroforestry,
organic production, and forest conservation among agricultural communities in Una's
buffer zone. The Jupard-WWF project set some very specific conservation goals for
the communities in which it worked. Each community was expected to maintain 30%
of its total area in forest cover (20% forest cover is required by Brazilian law), and
40% in agroforestry systems. The remaining 30% could be used for non-agroforestry
crops, livestock, or other uses. Other goals included the appropriate use of
agroecological practices, diversity of agroforestry systems, organic certification, and
increased family income (Buschbacher in prep.).
Participatory mapping and interviews with producers in two Jupara
communities were carried out in June August 2005 to address the following
* Are families that participated in the Jupara Agroecology Project meeting project
goals for forest conservation and use of agroecologial practices?
* Are results consistent in different types of communities, including land
reform communities and traditional communities?
* How do family dynamics and size of landholding influence project
The Atlantic Forest, which once covered much of the eastern coast of Brazil as
well as parts of Argentina and Paraguay, is one of the world's most diverse
ecosystems, and also one of the most endangered. Due to intensive human
occupation, only about 7% of the forest remains intact, according to a recent mapping
project (Hirota 2003). The region is home to 108 million people, or about 60% of
Brazil's population (Hirota 2003). The Atlantic Forest Biome is on the Global 200
list of globally outstanding ecoregions, and is considered a Biodiversity Hot Spot.
The region supports over 1600 species of terrestrial vertebrates and 20,000 species of
vascular plants, and including more than 6000 endemic plant species and more than
500 endemic animal species (Mittermeier et al. 1999). Within southern Bahia,
researchers have documented extremely high tree diversity (Thomas et al. 1998), and
50% of the known endemic bird species of the Atlantic Forest (Aguiar et al. 2003).
The cacao growing region of southern Bahia contains some of the largest
remaining patches of Atlantic Forest, including the 11,000 hectare Una Biological
Reserve. Currently about 6,500 km2 of Theobroma cacao, a medium-sized
understory tree, are planted in this region (Aguiar et al. 2003). Many maintain some
of the original forest canopy or are interspersed with patches of natural forest (Alger
1998; Alves 1990; Buschbacher in prep.). This region's cacao agroforests have
generated considerable interest within the conservation community for their potential
to maintain forest-like landscapes and serve as biological corridors between blocks of
otherwise fragmented forest (Alger 1998; Alves 1990; Greenberg et al. 2000; Pardini
2004). Researchers in many countries have demonstrated that while agroforestry
systems are no substitute for intact forest, some wildlife species are able to use cacao-
based systems to some extent as a corridor between areas of more suitable habitat,
with more diverse systems providing more suitable habitat than less diverse systems
(Alves 1990; Greenberg et al. 2000; Laurance 2004; Reitsma et al. 2001). Pardini
(2004) carried out such a study on small mammals on fragmented areas of the Una
Historically, cacao in Bahia was produced on large plantations with hired labor.
Many of these plantations were planted under the cabruca system, in which some of
the forest's original overstory trees were left for shade and the understory was cleared
for cacao planting (Alves 1990). The cabruca system required smaller labor
investments than clear-cut systems, and so was amenable to largely absentee estate
owners (Ruf & Schroth 2004). On some estates, cacao is planted as a row crop,
alone or with a single overstory species, often rubber or fast growing non-native trees
that provide shade. The arrival of witches broom disease (Crinipellis perniciosa), a
fungal disease, in 1989 (Alger & Caldas 1994), combined with a decline in cacao
prices at around the same time, led to the abandonment of many plantations or to their
conversion to pastures for cattle, and to the felling of remaining large trees (Alger
1998; Ruf & Schroth 2004; Trevizan 1998). Today much of the cacao produced in
Bahia is produced by small and medium-sized landholders, because of the labor
intensity required to control the disease (Rice & Greenberg 2000).
The cacao crisis in Southern Bahia helped facilitate the agrarian land reform
movement in this area. High unemployment rates resulted in an increase in
unemployed workers becoming involved in the land reform movements, including the
Movimento dos Trabalhadores Rurais Sem Terra (MST) and others (Trevizan 1998).
With little prospect of future earnings in cacao, owners of the once-lucrative cacao
plantations were willing to sell their land to the Brazilian government for
redistribution among landless settlers (Buschbacher in prep.).
Depending on the biophysical conditions of the site, many producers in
Southern Bahia maintain diversified agroforestry systems rather than cabruca
systems. These may include cacao, rubber, cloves, black pepper, guarand, oranges
and other fruit trees, and various palm species. Jupard has worked with both
diversified producers and cacao producers, and with both agrarian reform settlements
and traditional communities.
Jupara Project History
Jupari is a Brazilian community development organization with origins in the
Catholic Church and the labor movement, which has been providing assistance to
both agrarian reform communities and traditional agricultural communities in
Southern Bahia since the 1980s. In 1994, Jupard entered into a partnership with
WWF. WWF was interested in the conservation of forest fragments in the buffer
zone of the Una Biological Reserve, and in particular the preservation of habitat for
the golden headed lion tamarin (Leontopithecus chrysomelas), an endemic primate
species. The Una Reserve was created specifically to protect this species, but studies
have determined that the reserve is not large enough to guarantee the long-term
survival of the golden headed lion tamarin (Deitz et al. 1994).
The Jupara-WWF project exemplifies the dual purpose of ICDPs,
encompassing the main goals of both partner organizations; to develop a model for
family agriculture that would both promote forest conservation among landowners in
Southern Bahia, and improve the socioeconomic conditions of families living in the
area (Buschbacher in prep.).
Beginning in 1995, Jupara formed several extension teams who began working
in 36 communities in the area, providing training in community leadership and
organization, and agroecological practices, such as the use of contour erosion barriers,
green manures and cover crops, organic fertilizers and composting, integrated pest
management, and diversification of agroforestry systems. Target practices are
described in greater detail in Table 2-1. Agroforestry is the most common
agricultural system for smallholders in Bahia, as most of the region's traditional cash
crops are tree crops, including cacao, rubber, cloves, palm fiber and fruit. Therefore,
the training was oriented more toward improvement of current agroforestry systems
and the use of organic methods rather than actual agroforestry adoption. As a result
of this training, 338 families in 14 communities received organic certification in
2002, including 9 settlements in which 100% of families received organic
certification (Buschbacher in prep.).
Table 2-1. Description of target agroecological practices.
1. Use of Organic Fertilizers. Four types of organic fertilizers were promoted by
Commercial Organic Fertilizer Jupara produces this fertilizer and sells it to farmers.
Homemade Liquid Organic Fertilizer A type of compost tea, using manure and
other organic ingredients, that Jupara extensionists taught farmers to make for
Compost Composting of organic materials such as cacao shells and food scraps.
Produces a small amount of fertilizer, usually used for cacao seedlings.
Green Manures Nitrogen fixing plants that are planted either between crops or
during a fallow period and allowed to decay on site. Can be used to improve soil
quality or control erosion; farmers also report using green manures to control ants.
Species commonly used were canavalia (Canavalia enisformis) and pigeon pea
2. Use of crop rotation. Rotation of the location of crops from year to year. An
annual crop should be followed by a crop from a different family. In agroforestry
systems, annual crops are gradually replaced by perennial and tree crops.
3. Multicropping. More than one species is planted in each field. Tree crops might
planted with annual crops or several tree crops planted together.
4. Mulching. Use of mulches such as leaves or cacao shells, usually around
the base of trees, to add nutrients and control erosion.
5. Contour erosion barriers. Planting along the countour, with the creation of some
kind of barrier to control erosion and gradually form terraces perpendicular to the
slope. The barrier is usually created by piling up the organic residue (weeds and
fallow vegetation hand cleared in preparation for planting) in contour rows.
Typically, annual crops are planted in the cleared area between the rows of residue,
and tree crops are planted in the rows. This way of treating the organic residue is
alternative to burning; it is labor-intensive but promotes long-term soil
6. Elimination of Agrochemicals. Agrochemicals commonly used in Bahia include
chemical fertilizers and pesticides to eliminate leafcutter ants. Jupara hoped to
these with organic products.
7. Elimination of Fire. Farmers in Bahia commonly bur to clear forest or clear
fallowed fields for planting. Jupara hoped to replace burning with intensive,
permanent agroforestry systems and the use of contour barriers.
The second stage of the Jupara-WWF program focused on commercialization of
organic cacao. In July of 1998, a local cooperative called COOPASB (a Cooperativa
dos Pequenos Produtores e Produtoras Agroecologistas do Sul da Bahia) was formed
to assist with the commercialization of agricultural products. The first large sale of
certified organic cacao for export was made in July of 2003 (Buschbacher in prep.).
In 2005, Coopasb purchased 207 tons of organic cacao produced by local
smallholders. Most of it was sold to Cargill, an international corporation that exports
cacao from the region.
Prior to this study, a comprehensive evaluation of the extent to which Jupara
communities are implementing organic agricultural practices and meeting the goals
for land use and forest conservation had not yet been performed.
Jupara has worked in varying degrees of intensity in 36 communities, with
extension efforts concentrated more intensely in those communities closer to the Una
Reserve. These 36 communities represent a wide range of economic, social, and
biophysical characteristics. Families within communities also differ in size,
socioeconomic status, farming history, size and type of landholding, levels of
participation in extension activities, and attitudes toward conservation. Given this
range of variables, we expected to find varying level of success among families in
meeting the conservation goals set forth by Jupara.
Selection of Study Sites
Two communities were selected for this evaluation, Fortaleza and Lagoa Santa
(Figure 2-1). Jupara has worked in several types of communities including land
reform settlements and traditional communities. Fortaleza is a 30-year-old land
reform settlement, and Lagoa Santa is a traditional community that has existed for
many generations. They were selected for this study because Jupara extension
support has been consistent enough to provide a reasonable test of their intervention.
Fortaleza, in particular, received a high level of extension support due to its location
close to the community reserve, and Lagoa Santa was consistently involved in Jupara
activities partly due to the efforts of one particularly dedicated community leader. In
addition, the production systems in these communities are both based on mixed
agroforestry systems and are similar enough to allow a comparison between the
These communities have strong community leaders, which facilitated both
continued participation in the Jupara project and the logistics of this research. The
political climate in the region made it unfeasible for this researcher to conduct
interviews in a community in which Jupara has not worked.
Figure 2-1. Location of study sites. Fortaleza and Lagoa Santa are shown in red.
Description of Study Sites
Fortaleza is a land reform settlement in the municipality of Una, BA. Fifty
families are settled on 1102 hectares, with an average lot size of 16 hectares (INCRA
2005). Approximately 250 hectares are in community areas, including a small forest
reserve and a community cacao plantation. Families began settling the area in the
mid 1970s. At that time, the area was completely forested. Most of the settlers in
Fortaleza grew up on their parents' farms, usually also in Southern Bahia, or worked
on nearby rubber plantations before settling there. INCRA (Instituto Nacional de
Colonizagdo e Reforma Agraria, the government agency responsible for land reform)
officially divided the land and gave titles to these landholders in 1997. At this time,
some families were moved from their original plot of land to a different plot, in some
cases, a new, completely forested plot, and other families had a portion of their
original land officially titled to a neighbor. In most cases, the male head of household
received a title to his individual parcel of land. There are also a few female heads of
household. In general, each farmer and his family work their own piece of land.
Some farmers have organized a mutirdo, or a workgroup for labor exchange in which
they take turns working together in each member's field. Other than the mutirao,
landholdings are individually managed.
Farmers in Fortaleza produce mainly rubber and cacao as cash crops, as well as
some manioc, corn, fruit and vegetable crops for consumption. Few families own
livestock beyond one mule, for hauling cacao, and a few chickens. On average, 47%
of family income comes from agricultural production and 53% comes from off-farm
sources, usually employment in the nearby town of Una, or retirement pensions.
Most of the agricultural income comes from rubber. Nine families have received
organic certification through the Jupara program.
Although this area has been settled for more than 30 years, it still lacks many
basic services. Only a small part of the community has electricity, and bus service is
sporadic at best. Water is obtained from springs along a small stream, sometimes at a
considerable distance from the family's home, and basic sanitation is completely
Many of the original farmers in Fortaleza are getting older and have children
who are married and beginning their own families. In some cases, the land has been
unofficially divided among several sons. In one case, 23 people are surviving on the
production of one 17 hectare parcel. In other cases, the children of these farmers
work in Una or on nearby plantations. Recently Jupara has helped to organize a new
group of landless rural workers, composed primarily of the sons and daughters of the
farmers in Fortaleza. The group is petitioning INCRA to grant them title to a new
settlement on a nearby piece of land. The group chose not to occupy the land and
remain in Fortaleza with their parents, awaiting a decision from INCRA.
Lagoa Santa is a traditional agricultural community located in the municipality
of Itubera. Lagoa Santa is a quilombo community, or a community that was
originally founded by a group of escaped slaves. Families of the residents have
farmed this land for several generations; as a result the landholdings have been
divided several times among farmers' descendants, and landholdings tend to be small.
Landholding size ranged from 2 to 25 hectares, with an average size of 9 hectares.
Many of the larger landholdings are shared by extended families. Little native forest
remains in this area.
INCRA does not keep maps and records of older quilombo communities as it
does for more recent settlements, so it is uncertain how many families live in Lagoa
Santa and how much area the community encompasses. About 30 families are
members of the community organization, but many other residents do not participate.
Lagoa Santa is one of several quilombos in a cluster of small communities. Some
members of these nearby communities participated in the Jupard project and several
participants from the neighboring quilombo Campo da Amancio were also
interviewed for this study to increase the sample size.
Farmers in Lagoa Santa maintain diverse agroforestry systems similar to those
in Fortaleza, producing mainly rubber, cloves, piassava palm fiber for roofing
material, and small amounts of cacao, black pepper, and manioc flour. On average,
67% of family income comes from agriculture and the remaining 33% comes from
off-farm sources, including employment in Ituberi and on nearby plantations. Only
one family maintains organic certification through the Jupard project.
Like Fortaleza, Lagoa Santa lacks electricity, potable water, basic sanitation,
and regular public transportation services.
A total of 50 household interviews were conducted in June, July and August of
2005. Jupard staff presented me to each community at the start of my fieldwork, and
I stayed in each community alone to conduct the interviews. I visited each
community twice, first spending two weeks in each community, returning to Ilheus to
conduct preliminary data analysis, and then returning to each community for a second
visit of about 10 days.
In Fortaleza, the fifty existing households were classified as participants,
former participants, or non-participants. Current members of the COOPASB
cooperative were considered participants, past members were considered former
participants, and non-members were considered non-participants. All of the
participants and former participants were interviewed, and of the remaining non-
participants, a random sample of 10 was selected, for a total of 30 interviews. In
Lagoa Santa, no map or list of present households existed, so households were
sampled opportunistically for a total of 20 interviews, 10 with participants and 10
Interview content included questions concerning family demographics,
agricultural practices, and land cover distribution, as listed in Table 2-2. To assist
participants in describing their property's land cover distribution and the size and age
of agroforestry systems, a participatory mapping activity was carried out in which
participants were asked to draw a map of their landholding. Annual family income
was calculated by asking the respondent to list the products that he or she sells, their
approximate market price, and quantity harvested per month or year.
For each household interviewed, the researcher also toured the property with
the landowner to verify information obtained in the interview, and to collect GPS
points and a brief description of each of the land cover types present on the property.
Table 2-2. Interview content
Community and regional training
Jupara extension visits
Demonstration plots and workgroups
Women's development activities
Health and environmental education
Leadership positions in community
Workers on and off farm
Number of years farming this site
Years of primary education
Size of landholding
To measure a family's level of participation in the Jupara program, participants
were asked whether or not they had participated in a number of Jupara sponsored
activities, and whether they had participated once, several times, or many times.
Participants were also asked whether or not they were a present or past member of
COOPASB or any other organization, and whether they had ever held a leadership
position in any of these organizations.
Families' involvement in the Jupara program varied considerably. In order to
consider participation as a continuous variable, rather than comparing groups of
participants and non-participants, each activity was given a numeric score. Most
activities were scored 1 for participating once, 2 for participating several times, and 3
Agricultural practices, including use
Contour erosion barriers
Land cover types present:
Forest, agroforestry, capoeira (fallow
areas and forest regrowth), annual
Diversity of products
On and off farm income
for participating many times. Activities which were more difficult for farmers to
participate in, such as traveling to attend a regional training event, or indicated a
higher level of participation, such as joining the COOPASB cooperative, were more
heavily weighted, as indicated in Table 2-3.
The sum of these was calculated to give an overall participation score for each
family. The distribution of participation scores in Fortaleza and Lagoa Santa are
shown in Figure 2-2.
Because the development of the participation score was somewhat subjective,
the validity of the score was verified by producing 10 variations of the score and
comparing them statistically to one another. Five scores were created based on the
author's judgment of five possible ways of choosing participation criteria and
weighting them, and five more were produced by randomly eliminating some
components and weighting the components randomly. A correlation matrix was
generated to compare the scores to each other, and showed that the scores were
correlated, with p<0.001 in all cases. A second correlation matrix compared the rank
of each family using the 10 participation scores; rankings were also highly correlated
in all cases, p<0.001. Because the families' rankings changed little with each
variation of the participation score, we can say that no matter which version of the
score is used, families with high participation will consistently have a high
participation score, and families with lower participation will score lower.
Table 2-3. Jupara sponsored activities and corresponding participation scores.
Regional Training in:
Women's Issues 0
Other (health, cooperatives) 0
Community Training in:
Women's Issues 0
Other (health, cooperatives, 0
theater, or environmental education)
= never, 3 = once, 4 = several times, 5 = many times
= never, 3 = once, 4 = several times, 5 = many times
= never, 3 = once, 4 = several times, 5 = many times
= never, 1 = once, 2 = several times, 3 = many times
= never, 1 = once, 2 = several times, 3 = many times
S= never, 1 = once, 2 = several times, 3 = many times
Received a visit from:
An agronomist (male) 0 = never, 1 = once, 2 = several times, 3 = many times
An environmental educator (female) 0 = never, 1 = once, 2 = several times, 3 = many times
Visited another Jupara Community
Participated in a Demonstrative Plot
or community plot.
Participated in a "Mutirao"
(Community work group)
Member of Coopasb
Member of another organization
Woman is also a member.
Has held a leadership position
in one of these organizations
0 = never, 1 = once, 2 = several times, 3 = many times
0 = never, 1 = once, 2 = several times, 3 = many times
0 = never, 1 = occasionally, 2 = participated
in the past, 3 = participating regularly now.
0 = no, 3 = past member, 5 = current member
0 = no, 1 = yes
0 = no, 1 = yes
0 = no
3 = president
1 = all other leadership positions
Figure 2-2. Range of participation scores in Fortaleza and Lagoa Santa.
U) A Membership
o 10 0 A Coopasb Member
SPast Coopasb Member
U 0 O Non-Member
0.0 1.0 2.0 3.0
Community 1 = Fortaleza, 2 = Lagoa Santa
Use of Agroecological Practices
Families were asked whether or not they were currently using a variety of
traditional and innovative agroecological practices on their properties. Because the
use of some practices tended to be highly correlated with the use of other practices,
these data were reduced using a Principal Components Analysis (PCA). The first
component of the PCA produced a continuum of practices, ranging from ecologically
destructive practices on one end, such as burning and the use of chemical fertilizers,
to more innovative and ecologically desirable practices on the other end, such as the
use of organic fertilizers, as illustrated in Figure 2-3. Practices which were more
commonly practiced and possibly easier to implement, such as crop rotation and
mulching, fell somewhere in the middle. This scale is labeled the Ecological
Desirability of Practices. The PCA produced a score somewhere along this
continuum for each family. This scale of the Ecological Desirability of Practices was
used in the subsequent analysis to represent the extent to which families had
implemented agroecological practices.
Figure 2-3. Component plot for principal components analysis. Component one
orders practices from ecologically destructive to ecologically beneficial.
conto jr Legend:
Toxics -Uses agrochemicals
rotation mulchco"rI st including fertilizers.
.5 U Burning -Uses fire to clear fields
tox ics Rotation -Uses crop rotation
Contour -Uses contour erosion
multi cr barriers
S org_fert Green_m -Plants green manures
hm fert U Mulch Uses mulch around
0.0 burning plants or trees.
Compost -Composts organic
materials for fertilizer.
Multi_cr -Uses multicropping.
Hm fert -Makes and uses
-.5 organic fertilizer
SOrg_fert Purchases and uses
a commercial organic
-1.0 -.5 0.0 .5 1.0
Component 1 Ecological Desirability of Practices
For each community, six stepwise multiple regression models were used to
compare the effects of participation, using the index previously described, and nine
other family attributes, on six outcome variables that represent Jupara's goals for
conservation and adoption of agroecological practices. Family attributes included
family size, ratio of farm and off farm workers to total family members, ratio of farm
workers to total family members, age and education level of household head, size of
landholding, years the farmer has farmed this plot, income, and percent of total
income derived from farming. Outcome variables included the use of agroecological
practices (Ecological Desirability of Practices score and number of practices
implemented), percent of landholding in forest cover, percent of landholding in
agroforestry, number of products produced for sale, and farm income.
Several other family attribute variables were considered but were found to be
correlated with one or more of these variables and were eliminated. The
Kolmogorov-Smimov test was used to confirm that data were normally distributed.
Comparison of communities
Descriptive data collected in the two communities reveals some similarities
and differences between Fortaleza and Lagoa Santa (Table 2-4). Families in Lagoa
Santa are slightly larger, have more farm workers, and earn a larger percentage of
their income from farming than families in Fortaleza. Level of basic education in
Fortaleza is slightly higher, though in Fortaleza, non-participants are generally more
educated than participants, while in Lagoa Santa, participants have more years of
The most apparent difference between Fortaleza and Lagoa Santa is in the
distribution of land uses. Participants in Lagoa Santa tend to have more land than
non-participants, while in Fortaleza all landholdings are virtually the same size.
Landholders in Lagoa Santa dedicate a larger percentage of their landholdings to
agroforestry systems and have considerably less forested and fallowed areas than
farmers in Fortaleza. Landholdings in Fortaleza are much larger than those in Lagoa
Santa. In both communities, land use distribution between groups of participants and
non-participants are similar.
Table 2-4. Descriptive statistics by groups for two communities. P=Participants, FP = Former Participants, NP = Non-Participants.
Fortaleza Lagoa Santa
P FP NP All Std.dev P NP All Std.dev.
Total Family Members
Children living outside community
Average Age of all family members
Head of Household Age
Years of education
Head of Household
Years farming on current property
Age of Oldest Agroforestry System
Age of Newest Agroforestry System
Hectares Natural Forest
% Natural Forest
Years since most recent deforestation
5.44 6.07 4.19
2.03 1.85 1.30
3.42 1.23 1.00
3.29 3.65 3.19
25.55 29.54 26.80 27.28 21.9
55.00 55.11 51.67 52.17 13.3
1.58 1.22 2.78 1.83 1.91
7.70 5.92 7.46 7.00 3.37
6.75 3.75 2.36 4.07 3.11
1.44 0.63 1.31 1.05 1.25
16.92 25.00 19.44 20.10 9.09
16.42 16.06 15.24 15.96 1.30
7.75 8.00 7.72 7.82 3.27
36.4% 33.1% 37.7% 39.1% 15.6%
14.17 17.56 23.13 17.68 8.80
6.46 6.56 6.33 6.45 5.07
3.58 2.56 3.00 3.10 3.45
36.4% 33.1% 37.9% 36.0% 17.9%
9.75 12.00 12.83 11.40 7.20
Hectares Fallow and/or Forest Regrowth 1.79 2.22 3.39 2.40 2.17
% Fallow/Forest Regrowth 18.4% 10.8% 8.3% 12.3% 11.5
Hectares Annual Crops
% Annual Crops
1.71 1.42 1.25 1.48 1.38
6.4% 7.0% 8.3% 7.3% 6.5%
6.25 8.28 7.35 5.56
2.94 2.46 2.65 1.43
0.28 1.25 0.86 0.83
2.25 4.20 3.00 3.28
22.05 26.68 29.31 10.68
47.13 59.83 52.60 11.15
2.88 1.00 1.75
4.39 6.71 5.04
3.17 2.20 2.73
0.07 1.67 0.33
26.88 30.42 29.00 19.03
11.81 7.85 9.43 7.40
8.63 5.46 6.73 5.40
75.8% 73.8% 74.6% 20.3%
19.00 23.00 21.22 10.26
6.44 5.44 5.91 4.14
1.63 0.77 1.11 2.67
8.9% 7.4% 8.0% 15.2%
8.20 16.75 12.00 9.74
0.91 0.58 0.71 1.06
7.8% 8.8% 8.4% 10.4%
0.91 0..81 0.85 0.97
7.5% 13.6% 11.1% 14.4%
Table 2-4. Continued
Total Income in Brazilian Reis
% income from farming
P FP NP All Std.dev
0.88 0.75 1.66 1.08 1.13
8.6% 3.9% 4.1% 5.4% 5.8%
6202 10659 10317 8700 4617
3153 4470 3912 3775 2290
3127 6188 6405 4925 3689
50.9% 46.6% 36.8% 45.2% 26.3%
0.63 0.75 0.70 1.03
3.2% 2.9% 3.0% 3.8%
7721 10055 8581 4341
3347 8861 5375 3716
4383 1193 3206 3717
49.2% 92.7% 66.6% 30.8%
Regression Analysis of participation and family attributes on program
Participation positively influenced the implementation of sustainable agricultural
practices, as indicated by both the ecological desirability of practices score and the total
number of practices implemented, as well as farm income, but did not influence land
cover. Stepwise multiple regression analyses indicated a variety of significant
relationships between the outcome variables and other socioeconomic factors, as
illustrated in Table 2-5.
Table 2-5. Multiple regression models for Fortaleza.
Outcome Variable Significant Predictors B t Sig. (p) R2
Ecological Participation 0.491 3.426 0.002 0.511
Desirability of Ratio of Farm workers/total family 0.353 2.558 0.017
Practices Score Education 0.321 2.237 0.034
# Organic practices Participation 0.532 3.328 0.002 0.283
#Ag. Products Education 0.494 3.113 0.004 0.321
Size of lot 0.298 1.880 0.071
% Agroforestry Education 0.421 2.588 0.015 0.288
Years on Lot 0.349 2.149 0.041
% Natural Forest Years on Lot -0.466 -2.789 0.009 0.271
Farm income % income from Agriculture 0.685 6.108 0.000 0.768
Family size 0.397 3.660 0.001
Participation 0.377 3.275 0.003
Years on Lot 0.263 2.226 0.036
Size of Lot 0.211 1.769 0.090
Predictors eliminated from all models: Ratio of total workers (on and off-farm) to total family members
Age of Head of Household
Participation positively influenced the number of organic practices implemented,
the diversity of agricultural products produced for sale, and farm income, but did not
influence land cover. Other predictors were significant as indicated in Table 2-6.
Table 2-6. Multiple regression models for Lagoa Santa.
Dependant Variable Predictor B
Ecological Education 0.508
# Organic practices Ratio Farm workers/total family -0.510
# Ag. Products Participation 0.793
% Agroforestry Family Size 0.455
% Natural Forest None significant
Farm income Participation 0.480
Age of household head 0.362
Predictors eliminated from all models: Ratio of workers to total family members
Years on lot
% of income from agriculture
Size of lot
Sig. (p) R2
Use of Agricultural Practices
Multiple regression analyses examined overall implementation of agroecological
practices, but did not assess which practices were implemented by farmers and which
were not. Looking at each specific practice, we find that some practices were adopted by
Jupara participants more often than non-participants, others were in use by virtually
everyone, and others weren't adopted at all. These data help determine which practices
are most acceptable, and which are most difficult to implement.
Figure 2-4 shows the percentage of participants, former participants, and non
participants in Fortaleza who adopted each of the practices. Some practices were seldom
adopted by any group, including the use of green manures, homemade liquid organic
fertilizer, and composting. Others, contour erosion barriers, use of commercial organic
fertilizer, which is sold by Juparai, and abandonment of agrochemicals, were adopted by
participants more often than non-participants. Participants eliminated the use of burning
only slightly more often than non-participants. The remaining practices, crop rotation,
mulching, and multicropping, were in use by participants and non-participants alike and
may be considered traditional practices in Fortaleza.
Green Manures Homemade Composting Contour Commercial No Burning No Crop Rotation Mulching Multi-cropping
Organic Erosion Organic Agrochemicals
Fertilizer Barriers Fertilizer
Non-Participants 0 Former Participants U Current Participants
Figure 2-4. Adoption of agroecological practices in Fortaleza.
Results were similar in Lagoa Santa, as shown in Figure 2-5. Green manures and
homemade liquid organic fertilizer were the least adopted practices for all groups.
Composting was adopted more often by non-participants. Contour erosion barriers,
commercial organic fertilizer, and the elimination of agrochemicals were adopted more
often by participants than non-participants. Around 40% of farmers in both groups had
eliminated the use of fire, and crop rotation, mulching, and multicropping were used by
almost all farmers in both groups.
Green Homemade Composting Contour Commercial No Burning No Crop Rotation Mulching Multi-cropping
Manures Organic Erosion Organic Agrochemicals
Fertilizer Barriers Fertilizer
U Non Participants U Participants
Figure 2-5. Adoption of agroecological practices in Lagoa Santa
Average reported land use distributions in Fortaleza (Figure 2-6) and Lagoa Santa
(Figure 2-7) appear to be similar among groups of participants and non-participants, but
different between the two communities. Lagoa Santa has much less forest cover and
much more agroforestry. Graphs for Fortaleza are adjusted to include the community
Non-Participants Former Participants Current Participants
O Agroforestry U Annual Crops U Pasture O Capoeira U Forest
Figure 2-6. Land cover distribution in Fortaleza, as reported by farmers.
Participants Non Participants
I Agroforestry m Annual Crops Pasture o Capoeira m Forest
Figure 2-7. Land Cover Distribution in Lagoa Santa, as reported by farmers.
These figures report the average of the reported land use distributions. This can be
deceiving, especially in Fortaleza. On, average, the land use goal of 30% forest appears
to be met. However, these averages reflect the fact that some farmers on newer lots have
more than 30% of their area still forested, while others have no forest at all. Figure 2-8
shows the percentage of landowners meeting land use goals in each community. In all
groups, 55% or less of the farmers interviewed meet the goal for forest cover. The goal
for each property is 12.5% in Fortaleza after the community reserve is accounted for, and
30% in Lagoa Santa.
Non- Former Participants F Non- Participants -
participants F Participants F participants LS
Figure 2-8. Percentage of landowners meeting land use goals in Fortaleza (F) and Lagoa
Santa (LS) for Agroforestry and Forest.
The Jupara Project was successful in teaching agroecological practices such as the
use of organic fertilizers, composting, and contour erosion barriers, and reducing or
eliminating the use of agrochemicals and fire. Multiple regression analyses indicate that
participation in Jupara had a highly significant effect on the implementation of organic
practices on a scale defined by the Ecological Desirability of Practices Score in Fortaleza,
diversity of agricultural products in Lagoa Santa, and number of organic practices
implemented in both of these communities.
In addition to participation in the Jupara program, availability of farm workers, as
measured by ratio of farm workers to total family members, and educational level of the
household head, also positively influenced the use of agroecological practices. In
Fortaleza, education significantly influenced the diversity of agricultural products for sale
and the area of agroforestry systems established. The fact that education was often a
significant influence on more than one of these outcome variables might indicate that in
addition to extension services, rural communities benefit from access to basic education.
In both communities, some practices were implemented more often than others.
Practices tended to fall into one of three categories: practices implemented by almost
everyone, practices implemented by almost no one, and practices implemented more
often by participants than by non-participants. Almost all of the farmers interviewed
were using crop rotation, multicropping, and mulching. These three practices might be
considered traditional practices in these communities, since they are used even by those
who had no contact with the Jupara program. Participating farmers may also have been
influenced by their non-participating neighbors to adopt some practices, especially in
Fortaleza, where the program was very visible, even to non-participants.
Green manures, homemade liquid organic fertilizer, and compost were among the
least utilized practices. Green manures were used by less than 20% of both participants
and non-participants in all three communities. This practice may not have been as
vigorously promoted as some of the others, as many farmers had never even heard of the
practice, and seeds were not readily available. The few farmers who did report using
green manures reported using them as a pest control method against leaf-cutter ants,
rather than as a source of fertilizer.
Compost was used by less than 40% of farmers, and homemade liquid fertilizer
was used by only around 20% of farmers. Many farmers explained that these two
techniques required the availability of manure, and most of them do not have livestock,
so manure is not easily accessible. Some farmers compost kitchen scraps and agricultural
wastes such as the outer shells of cacao pods, but generally there is not enough
compostable organic material available to produce enough compost or compost tea to
depend on it as a main source of organic fertilizer.
In Fortaleza and Lagoa Santa, the elimination of agrochemicals, the use of
commercial organic fertilizer, and the use of contour erosion barriers are practiced much
more often by participants than not participants. In this case, Jupara produces the
commercial organic fertilizer and sells it to farmers. While it appears to provide a
satisfactory replacement for chemical fertilizers for farmers who choose it, it is slightly
more expensive than chemical fertilizer, requires a larger quantity per hectare, and thus is
more difficult to transport. Many farmers stated this expense and difficulty of
transportation as primary reasons for not using the commercial organic fertilizer.
Contour erosion barriers are labor intensive and require some technical expertise to
create. Increased use by participants might indicate that Jupara was successful in
providing this technical expertise and motivating farmers to invest in a labor intensive but
ecologically beneficial process.
Contour erosion barriers should reduce the use of fire, as a permanent vegetative
barrier is planted on top of the erosion barrier, so the use of fire would destroy the
established barrier. Disappointingly, 60% of farmers in Lagoa Santa still bum their fields
before planting, and project participants are as likely to bum as non-participants. In
Fortaleza, 50% of participants, 67% of former participants, and 33% of non-participants,
burn. Many of the same farmers implementing contour erosion barriers on one part of
their property may be using fire somewhere else. However, burning was measured as a
yes or no variable, and these data do not describe the type of burning practiced. Most
respondents reported using fire only to clear recently fallowed areas for planting, and not
as a means of clearing forest.
These data also indicate that eliminating the use of fire as a management tool may
be the most difficult of Jupara's target practices to implement.
Families participating in the project tended to have higher levels of farm income
than those not participating, and in Lagoa Santa, participating families produced a greater
diversity of products for sale. While it's possible that participation in the extension
program enabled families to create more productive farms, it's also possible that those
with already productive farms, or those most dependent on farm income, were more
likely to participate in the program.
Family size, lot size, number of years farming the lot, and age of household head
were also all positively correlated with farm income in one or more communities,
according to the linear regression models. In both Fortaleza and Lagoa Santa, total
income was higher for non-participants than for participants, but farm income was higher
for participants than non-participants. Thus, participants tended to derive a higher
percentage of their total income from agriculture, especially in Lagoa Santa, where
participants earned 92.7% of their income from agriculture, while non-participants earned
only 49.2% of their income from agriculture. This may indicate that those who depend
highly on agriculture for their livelihoods have a greater incentive to invest in a program
such as Jupara, while those earning a portion of their income from an off-farm source
may be too busy to participate, or, since generally they are already earning an equal or
higher income through wage labor than they would earn farming, they may not feel that
such an investment is necessary.
Participating families did not conserve more forest or implement more agroforestry
systems than non-participating families. In Fortaleza and Lagoa Santa, extensive
conversion of forest to agroforestry had already occurred before the program began, and
the ten-year period of the program may not be a sufficient length of time for changes in
land cover patterns, such as the regrowth of secondary forests, to develop in response to
changes in practice initiated by the Jupara program.
Farmers in Fortaleza are meeting Jupara's goals of maintaining a minimum of 40%
of the area in agroforestry and 30% of natural forest more often than farmers in Lagoa
Santa. In Fortaleza, 20% of the community's total area is conserved as a forest reserve.
To maintain a total of 30% forested, individual farmers would have to conserve an
additional 12.5% of the individual landholding in natural forest. On average, 19% of
each individual landholding is forested, with 43% of farmers maintaining at least 12.5%
forested. The multiple regression model demonstrated that the factor most affecting
forest cover is the number of years the family has been farming the lot (p=0.009). Forest
cover decreases with time, and families with large amounts of forest cover are those who
were moved to new, forested lots when INCRA officially delineated the properties in
1997, as shown in Chapter 4. Most farmers reported settling on a completely forested lot
at some point in the past, and converting it to agriculture over a period of several years.
In Lagoa Santa, the area has been used for farming for many generations. After
several generations of dividing the land among family members, some lots are very small
and most are completely invested in intensive agriculture and agroforestry; on average,
75% of lots are in agroforestry systems. Due to this long history of agricultural use, it's
not surprising that Lagoa Santa has less forest cover remaining, on average 8% of each
lot, though many farmers have no forest at all. There is no community forest reserve in
Lagoa Santa, and only 2 farmers meet the goal of conserving 30% of the land in natural
forest. Some farmers stated that they are maintaining an area of forest regrowth as their
reserve, and on average 8.4% of lots were in forest regrowth. Farmers reported an
average of 12 years since the most recent deforestation on the site, so it is likely that most
of the deforestation in this area occurred before the Jupara program began working in the
Additional analysis of a time series of satellite images is provided in Chapter 4 to
more fully understand the changes in land cover patterns before and after the Jupara
Success of the Jupara Agroecology Project
Based on these results, was Jupara an effective integrated conservation and
development project? Are families meeting the project goals?
Results of this program evaluation are in agreement with several other published
ICDP evaluations (Browder 2002; Perz 2004): development goals seem to be met more
frequently than conservation ones. In both communities, participation in the Jupara
program influenced the implementation of more sustainable production practices and
enabled farmers to meet the requirements for organic certification, though the
certification itself wasn't actually maintained in all cases.
Participating farmers have higher farm incomes than non-participating farmers,
which may or may not be a result of participation in the program. The commercialization
aspect of the Jupara program hasn't been fully implemented. Currently they buy and sell
only organic cacao through the Coopasb cooperative. In communities where farmers
produce mainly cacao, an effective commercialization system is in place. For other
farmers with more diverse systems, as in Fortaleza and Lagoa Santa, commercialization
is more difficult, as it is more difficult to market smaller quantities of many things, and
for some products, little market exists for organic certified products.
Generally, these two communities are not meeting the program goals for forest
conservation. Fortaleza has a small community reserve, encompassing about 20% of the
community's total area, but less than half of the farmers there are able to maintain the
expected amount of forest on their individual properties. Lagoa Santa has no forest
reserve, and only two farmers meet the goal of 30% forest cover. Data presented here
offer no indication that farmers participating in this project conserve more forest than
These results are consistent with Alger and Caldas' 1998 finding that Bahian
farmers on smaller lots are less likely to conserve forests than those on larger lots.
Jupara'a goals for forest conservation are particularly challenging, then, since all farmers
interviewed had landholdings smaller than 20 hectares. In Lagoa Santa, lots are smaller
on average than in Fortaleza, and less forest remains.
However, this study did not consider the quantity of forest present at the start of the
Jupara project. Many farmers stated that their property was entirely deforested and
converted to agriculture years before they realized that deforestation was a problem, and
expressed regret at this loss. Obviously, people cannot be expected to conserve
something they don't have in the first place. In order to fully justify an integrated
conservation and development project like the Jupara project, monitoring is needed
before, during and after the project. It is difficult to evaluate project outcomes if we
don't have any baseline data on land cover or agricultural practices at the start of the
project. Analysis of satellite imagery might be one way to understand when most of the
forest loss occurred, as well as to observe trends in forest regrowth.
Finally, we have to make sure that both environmental and development goals of
ICDPs are realistic. It simply is not realistic to expect a family of 10 to survive on three
hectares of land, as was the case for one family in Lagoa Santa, and still maintain 30% of
their three hectares in a natural forest reserve. In an evaluation of a USAID funded
project in Madagascar, Peters (1998) recognizes this top-down approach to defining goals
and expectations as one of the major weaknesses of integrated conservation and
development projects. He suggests giving local people a participatory voice in defining
project objectives, as their perspective on what might be feasible and important might
differ greatly from the perspective of USAID, or in this case, WWF.
Browder (2002) makes the important point that ICDPs often lack functional
linkages between specific development activities and desired conservation outcomes.
Results of the Jupara project support this criticism. While the use of organic agricultural
practices may contribute to overall ecosystem health and develop a mindset of
environmental stewardship among participants, organic agriculture and resulting
increased incomes do not necessarily prevent deforestation.
In southern Bahia specifically, Alger (1998) points out that promoting agroforestry
by itself does not conserve biodiversity, in fact, it can result in further fragmentation of
existing forest fragments. In areas with many small farmers, the law requires only a
small forest reserve on each farm, and reserves on individual farms don't conserve much
biodiversity if farms and family reserves are disconnected from each other.
Conservation in communities of smallholders will require landscape-level participatory
planning to unify forest reserves and maintain connectivity between them, and this
community process will likely occur separately from the family-level process of
implementing organic agriculture.
CONSERVATION AND AGRARIAN REFORM IN SOUTHERN BAHIA, THE CASE
"You have probably passed by an encampment in your car or on the bus. Maybe
you have seen a march of the landless people. Certainly you have seen on television, or
in a newspaper or magazine, news of land conflicts or occupations," write Stedile and
Sergio (1993) in A lutapela terra no Brasil (The Fight for Land in Brazil), a small book
intended to explain to the wider Brazilian public the basic ideology of the MST
(Movimento dos Trabalhadores Rurais Sem Terra, The Landless Rural Workers
Movement). Groups of rural landless workers, living in makeshift plastic tents and
awaiting settlement by the Brazilian government, are indeed a common sight along the
highways in many parts of Brazil.
Brazil has the second most unequal distribution of land in the world, behind only
neighboring Paraguay (Domingos 2002). Eighty-nine percent of farms are less than 100
hectares in size and equal 20% of the total land area. One percent of farms are more than
1000 hectares in size, and equal 45% of the land area (Domingos 2002).
Conflicts over land tenure have been present throughout Brazilian history and
continue today. Historically, land reform in Brazil has occurred on lands considered
unproductive, often on forested or ecologically fragile lands, leading to further
environmental degradation and deforestation as settlers carve out an agricultural
livelihood from forested land. Te6filo and Garcia (2003) found that of the land
appropriated by INCRA between 1997 and 1999, only 21% of it was in agricultural use
before occupation, often because its soil, topography, or market access was poorly suited
In recent years, an outbreak of Witches' Broom Disease has changed the dynamics
of this process in Southern Bahia. As productivity of cacao plantations decreased, large
landowners were more willing to sell their lands for redistribution through land reform.
Higher quality lands are now available for settlement, and the violence once
characteristic of land invasions is decreasing (Buschbacher in prep.).
In addition to their work in agricultural extension, Jupara has been highly involved
in the rural land reform movement in Southern Bahia, and changes in the land reform
process have led to changes in Jupara's community work. The agricultural extension
project described in Chapter 2 was originally conceived to function in traditional
agricultural communities and older, well established land reform communities in which
each family managed an individual unit of land. In recent years, the project has been
adapted to also function on resettled cacao estates, where production strategies tend to be
more collective. This chapter explains Jupara's involvement in the land reform
movement and presents the results of the agroecology extension project on a newly
settled cacao estate, Cascata.
JuparA and Land Reform
INCRA (Instituto Nacional de Colonizagdo e Reforma Agraria, or National
Institute of Colonization and Agrarian Reform) is the federal agency responsible for land
reform in Brazil and has an office in each Brazilian state. INCRA is responsible for the
purchase of the land from its original owner, assessing how many families should be
settled onto the property, and deciding who those families will be.
Both INCRA and the communities involved depend greatly on a handful of land
reform NGOs to facilitate the land reform process. The MST (Movement of the Landless
Workers), and the CPT (Catholic Pastoral Commission of the Land) are some well-
known examples. Many other organizations also exist and function similarly, such as
MLT (Movement for the Struggle of the Landless) and MSLT (The Land Liberation
In addition to providing agricultural extension services in established communities,
Jupara also provides support to a number of "encampments," or communities in the
process of acquiring land on which to settle. In southern Bahia there are many of these
encampments, basically communities living in black plastic tents along the highways,
indicating their intention to settle on the adjacent land. These people are unemployed
plantation workers, sons and daughters of farmers whose landholding is too small to
support them, or even city dwellers looking for a better life. The political process
involved in the formation of a new land reform community is complicated and often takes
Most encampments receive assistance from a land reform NGO in applying to
INCRA for settlement. According to Cullen et al. (2005), only 5% of land reform
projects in Brazil were initiated by INCRA. The other 95% were initiated by land reform
NGOs, who organized a group of settlers and petitioned INCRA's cooperation.
Settlement is a community process. When land is settled, INCRA grants an
Emissdo de Posse, or right of land use, to the community association rather than to
individuals. Generally a community will work with only one land reform NGO, and
these NGOs tend to differ in their philosophies in how a community should be managed.
These communities will often live for a number of years in an encampment until they are
granted a piece of land, so community organization is very important, both before and
after settlement. The land reform NGOs assist with this organizational process and with
the development of community leaders.
The Jupara extensionists have an extensive and complex history of work with land
reform over the past 10 years. When they began working with encampments in the mid
1980s, the process usually involved the violent occupation of a piece of land, usually
forested land or land of poor soil quality, that was considered "socially unproductive" by
Brazilian law and thus eligible for reappropriation, whether the landowner wanted to give
up this land or not. Some of these lands were known as "latifundos," a term which refers
to very large rural landholdings, parts of which are often left fallow or forested. Any land
which was not currently dedicated to the active production of agricultural goods, or did
not generate employment for local workers, was considered eligible for land reform.
According to a 1993 MST report, 180 million hectares were classified as "latifundos" in
Brazil. Redistribution of these lands is the focus of the agrarian reform movement
(Stedile & Sergio 1993).
In 1989, Witches Broom Disease broke out in the cacao plantations of Bahia, and
led many plantation owners to stop investing in the labor and fertilizer inputs necessary
for cacao production. Lack of care, fertilization, and pruning allowed the disease to
spread even more rapidly through the plantations, further reducing their productivity.
This change facilitated the land reform movement in southern Bahia. Unemployment of
former plantation workers resulted in an increase in the number of workers and families
becoming involved in the land reform movements of the time, the MST and others,
including Jupard (Trevizan 1998). With little prospect of future earnings in cacao,
owners of the once-lucrative cacao plantations were willing to sell their land to the
Brazilian government for redistribution among landless settlers, with much less violent
conflict than in previous years (Buschbacher in prep.). For this reason, the Jupard
extensionists sometimes referred to Witches' Broom Disease as "a santa vassoura da
bruxa," or Saint Witches' Broom. Management of the disease is labor intensive and can
be carried out more effectively by smallholders than by large plantation owners. Most of
the recent land reform settlements are on former cacao plantations. Families receive a
small area of established cacao, which they clean and prune, replant cacao where
necessary, and eventually bring back into production.
Land reform in Bahia continues to evolve and many families continue to await
resettlement, while living under black plastic and depending on food aid from the
government, churches, and NGOs. While this process is happening throughout Brazil,
according to MST's website (www.mst.org.br) (MST 2006), in 2003 Bahia had more
encampments than any other state, with 20,000 MST-sponsored individuals awaiting
Cascata, a Land Reform Settlement
Most recently, Jupard's work has focused on newer land reform settlements,
usually on former cacao plantations. In addition, Jupard has worked with encampments,
or groups of people awaiting settlement through the land reform process. Extension
services are provided before the community is settled on the landholding, and landholders
have had Jupard support for the entire period of land ownership. Often potential
landowners are from urban areas or are former plantation workers, and have little
experience in agriculture and the decisions involved in land management. Extension
programs in encampments focus on community organization, environmental education,
and sustainable agriculture and land management, so that when future landowners
actually receive their own plot, they are prepared to make environmentally and
economically sound decisions. Typically, communities which have recently passed
through the encampment process are more unified than older communities, and usually
all families participate in the Jupara project. Often residents live in the workers' houses
left behind by the plantation owners, so all community members live in a central
residential area, with agricultural areas located outside of the residential area.
Cascata is one cacao plantation community that began collaborating with Jupara
while still an encampment. Many differences in the production strategies, biophysical
characteristics, age of the community, and level of participation in extension activities in
Cascata make it difficult to compare directly with older communities like Fortaleza and
Lagoa Santa. A small number of interviews were also conducted in Cascata, in order to
understand how the project may work differently in different community types, and in
order to compare Jupara's past projects with their current and future directions.
Cascata is located in the municipality of Ubaitaba, Bahia. Cascata was an
abandoned cacao plantation that was sold for land reform. Forty families were settled
there in 1998. The land titles were granted not to individuals but to the community
association, although each family was assigned an area of 4-5 hectares of the plantation
to maintain and harvest. In addition, about 180 hectares of Atlantic Forest were
designated as a community forest reserve.
Cascata farmers produce cacao almost exclusively. Most of the land was planted in
cacao already, and little open area was available for planting manioc and other food
crops. Each family harvests cacao from its individual area. Cacao seeds must be
fermented and dried before sale, and this process is done collectively in a communal area,
using infrastructure left behind by the plantation owner. The entire community
landholding is organic certified and the cacao is marketed through the Jupara cooperative.
In order to receive the higher organic price, the cacao must also be top quality, which
depends on correct fermentation and drying, so the community decided to designate a few
community members to complete this process in hopes of a more uniform and higher
In comparison to Fortaleza, Lagoa Santa, and other local smallholder communities,
the standard of living in Cascata is quite high. Unlike in the older communities studied,
very few Cascata residents work outside of the community. A few have off-farm income
from retirement pensions. The community is located along a major highway, allowing
for easy access to public transportation, local markets, secondary education and health
services. Because Cascata was established on a former plantation, much of the
infrastructure was already in place, including a dozen workers houses where families are
now living, a large building for meetings and events, a church and school, production
areas for processing cacao, and a large cacao dryer. Additional houses were constructed
to accommodate additional families. Families enjoy well-constructed brick homes,
electricity, running water, modern sanitary facilities, and even a public telephone. All of
the residences are located in the central community area and the cacao plantation areas
surround this central area. In Fortaleza and Lagoa Santa, every family lives on its own
piece of land, and residents are fairly isolated from their neighbors. Cascata's centralized
model greatly facilitates community organization and cooperation.
Twelve households were randomly selected for household interviews, as described
in Chapter 2. Data was collected on participation in the Jupara program, use of target
agroecological practices, land use patterns, and other socioeconomic conditions that
might affect a family's ability to meet project goals.
As in Fortaleza and Lagoa Santa, a participation score was calculated for each
family and data were entered into linear regression models.
Descriptive statistics were compiled for each of the three communities including a
number of factors, including family characteristics and land use patterns. A MANOVA
analysis was used to assess differences across communities. Tamhane T2 Post-Hoc Test
was used to test the statistical difference between each pair of communities. MANOVA
was selected for its ability to compare multiple dependant variables in three or more
categories while taking into account possible correlations between the dependant
Although Cascata entered the project later, participation in Cascata was comparable
to participation in the other two communities (Figure 3-1).
In Cascata, all families participated in the Jupara project (Figure 3-1) and all
maintain organic certification, and are utilizing, at minimum, the organic practices
required for certification. Cacao agroforestry was already established when residents
received their plots, and a community forest reserve sets aside 40% of the community's
total area in natural forest. Because little variation in participation, practices, or land
cover existed among families interviewed, linear regression models showed no
significant effect of participation on agricultural practices, farm income, or land cover.
A A A
1. o a Coopasb Member
c 100 0
0 0 Non-Member
0.0 1.0 2.0 3.0 4.0
Community: 1=Fortaleza, 2=Lagoa Santa, 3=Cascata
Figure 3-1. Range of participation scores in Fortaleza, Lagoa Santa, and Cascata.
Therefore, only descriptive statistics are presented here. Table 3-1 presents
household data collected in Fortaleza, Lagoa Santa, and Cascata and allows for
comparison between Cascata and older communities. On average, Cascata residents have
more years of education, fewer off-farm workers, and a larger percentage of income
coming from agriculture than Fortaleza and Lagoa Santa residents. Cascata has a smaller
family landholding size than Fortaleza and a similar size to Lagoa Santa. Including the
community forest reserves, Cascata has more forest and less area in annual crops than the
other two communities.
Table 3-1. Descriptive statistics by groups for three communities. indicates statistically significant difference (p<0.05) between
communities as measured by MANOVA analysis. Statistical relationships are shown between each community, Fortaleza
(F), Lagoa Santa (L) and Cascata (C) as measured by Tamhane T2 Post-Hoc Test.
Fortaleza Lagoa Santa Cascata Statistical Relationship
Mean Std.dev Mean Std.dev. Mean Std.dev. Between Communities
Total family members living in community
Children living outside community
Average age of all family members
Head of household age
Years of education
Head of household*
Years farming on current property*
Total area* 15.96 1.30
Age of Oldest Agroforestry System
Age of Newest Agroforestry System
% Natural Forest (includes community reserve)*
Years since most recent deforestation
% Fallow/Forest Regrowth
% Annual Crops*
5.4% 5.8% 3.0% 3.8%
Table 3-1. Continued
Total Income in Brazilian Reais
% income from farming*
Adoption of Agroecological Practices
As in the other two communities, farmers implemented some practices more often
than others. In Cascata, the preferred practices differ slightly (Figure 3-2). Green
manures, homemade liquid organic fertilizer, composting, and contour erosion barriers
were implemented least often, and crop rotation was also implemented less often than in
the other two communities. As in Fortaleza and Cascata, mulching and multicropping are
used by most farmers. All properties in Cascata obtained organic certification as a group,
so 100% of farmers have replaced chemical fertilizers with the commercial organic
fertilizer that Jupara sells, as chemical fertilizers are strictly prohibited. Only one farmer
reports burning his field, and others stated that burning is also prohibited.
Green Manures Homemade Composting Contour Erosion Commerci2
Organic Barriers Organic
Figure 3-2. Adoption of agroecological practices in Cascata.
The entire Cascata settlement meets Jupara's land cover goals of 30% natural forest
cover and 40% agroforestry. Two large community forest reserves encompass 40% of
the community's total area. Most of the other 60% of the settlement was already planted
in cacao agroforestry, although in poor condition. Each family received a parcel already
planted in cacao, and some farmers also have a small area for annual crops, usually where
cacao trees have died of Witches Broom Disease. There is also a small community
pasture area for work animals and some communally owned cattle. Figure 3-3 shows the
distribution of land use in Cascata, including the community managed forest reserve.
O Agroforestry U Annual Crops U Pasture D Capoeira U Forest
Figure 3-3. Land Use Distribution in Cascata.
Agroforestry and Conservation in Cascata
Jupara considers Cascata to be one of their greatest success stories. This
community is very different from Fortaleza and Lagoa Santa and may not be directly
comparable. In Cascata, all families participate in the Jupara program and sell products
through the Jupara cooperative. One hundred percent of families are organic certified
and as such are required to use only organic fertilizers. All interviewees reported
eliminating chemical fertilizer and using a commercial organic fertilizer, and only one
farmer reported using fire. For this reason, there may not be enough variation in levels of
participation or agricultural practices to show significant results here.
Preferred agricultural practices are slightly different in Cascata than in the other
communities. Fewer farmers reported using crop rotation, probably because this is
usually done at the beginning stages of establishing an agroforestry system, and most
Cascata farmers received already established plots. This can be a challenge in
communities on established estates, as there is little open area in which to plant annual
crops for household consumption. Homemade liquid organic fertilizer was more popular
here than in the other communities, possibly because the community association owns
several cows, so manure is readily available to use in preparing the homemade fertilizer.
A biogas project is underway to compost manure and produce methane gas to power a
cacao dryer, but as of the time of these interviews, farmers were not yet using the
compost produced by that project. Knowing that different types of communities prefer
different agroecological practices, Jupara extensionists can adapt future extension
programs to promote the practices most fitting to each type of community.
In Cascata, two large forest reserves conserve 40% of the settlement's total area in
primary Atlantic Forest. Farmers have access to only as much land as they can use for
agriculture; the forested land is strictly off-limits to agricultural activity. This is perhaps
the best scenario to ensure long-term forest conservation. Forested areas are large and
contiguous, rather than fragmented into smaller family-owned forest reserves as in
Fortaleza. Because the farming system in Cascata is intensive and permanent, and was
already established when farmers took possession of the land, Cascata farmers may feel
less need to clear forests than farmers in other communities. Soil quality is also better in
Cascata than in the other two communities, which allows for more intensive cacao
production on relatively small parcels.
Community members in Cascata have hopes of one day benefiting from their forest
reserve as an ecotourism destination. In addition to the forest reserve, Cascata has
several other unusual resources that might make it a suitable ecotourism site in the future,
including a beautiful waterfall with a large pool suitable for swimming, and a large
community house, originally the plantation owner's home, that might provide suitable
guest rooms, as well as optimal location along a major road with frequent bus service.
Although ecotourism is only an idea now, the presence of these resources may provide an
additional incentive to conserve forests and other natural resources.
Future Directions for Jupara
Jupara's current work focuses mainly on communities like Cascata: new land
reform settlements on former cacao estates. While it is important to recognize the need
for extension services to improve both livelihoods and conservation practices in all rural
communities, regardless of age, location, or production strategy, communities like
Cascata are in many ways the ideal places for Jupara's work, especially from a
conservation perspective. There are several reasons for this.
First, Cascata and communities like it tend to be highly organized. Community
members live close together in a small residential area, rather than in isolated homes on
individual landholdings. This facilitates the participation of most, if not all, community
members, and makes it easy to organize meetings and activities. Some important
management decisions, such as the designation of community reserves and organic
certification of the entire estate, are made collectively rather than individually.
Secondly, the biophysical conditions on abandoned cacao estates tend to be much
more suited for agriculture than those of older land reform settlements. Historically, the
best lands were occupied by wealthy estate owners, and settlers invaded poorer lands.
Many settlements still exist on these lands, and in several cases, agriculture has proven
inviable given the poor soil quality. In Cascata, landowners can produce cacao, a crop
that is demanding in soil nutrients but produces a lucrative product, with a minimum of
fertilizer inputs. Cacao is also advantageous because there is a market for organic cacao.
Finally, by actively participating in the land reform process, Jupara is able to gain
access to these communities before they are settled on the land and begin educating
future farmers before destructive practices are in place. In many cases, new settlers have
little farming experience, and it may be easier to teach agroecological practices to a new
farmer than to change the well-established practices of a veteran.
Also, as facilitators of the land reform process, Jupara extensionists are able to
choose landholdings to target for land reform, and are currently working to choose areas
that have conservation value. In this way, larger remaining patches of Atlantic Forest can
be preserved as community reserves within new settlements.
EFFECTS OF AN NGO EXTENSION PROGRAM ON LAND USE CHANGE ON
SMALL FARM PROPERTIES IN THE ATLANTIC FOREST OF SOUTHERN BAHIA
The use of remote sensing technology has become a common method for assessing
land use and land cover change (LULCC) in many areas of the world. Researchers have
used satellite imagery to observe vegetation patterns at a landscape-wide scale, and also
to observe changes over time through analysis of a time series of images. Remote
sensing plays an important role in monitoring tropical deforestation in regions such as the
Brazilian Amazon (McCracken et al. 2002), Mexico's Yucatan Peninsula (Vance &
Geoghegan 2002), and Southeast Asia (Rindfuss et al. 2002).
While remote sensing is a powerful tool, it allows us to examine only the physical,
spatial, and temporal aspects of land cover change. Many researchers have paired remote
sensing studies with on-the-ground data collection, including interviews of local
landowners, in order to better understand the social, economic, and political drivers of
deforestation and other land cover changes. Although interviews may take place at the
level of households or individual landholdings, remote sensing analyses are commonly
broader in scale, encompassing entire regions, watersheds, protected areas and their
buffer zones, or villages. In one example, Sunderlin et al. (2000) used satellite imagery
to observe a nationwide trend in deforestation following an economic crisis, and a survey
of 5000 households in Cameroon to help explain this change, but households were not
linked to a specific location on the landscape.
Although landscape-wide studies are useful for determining broad-scale trends in
land cover change, understanding region-wide habitat availability for a particular species,
assessing the connectivity of landscape elements, or planning future protected areas,
many land use decisions are made at a much finer scale. Especially in developing
countries where small-scale agriculture is common, individual landowners make
decisions about which crops to plant and where, whether or not to convert forest to
agriculture, which trees to cut, how much fertilizer and pesticide to use, and how to
manage water resources. A comprehensive understanding of the forces driving land use
and land cover change requires an understanding of the household-level decision making
process (Rindfuss et al. 2002).
Very few LULCC studies have focused specifically on change at the household
plot level, due in part to methodological difficulties inherent in identifying the precise
location of a small farm plot on a satellite image. Several studies in the Brazilian
Amazon have linked households with specific plots of land (McCracken et al. 2002;
Walsh et al. 2002). These studies have been facilitated by the fact that plots in most
newer Amazonian settlements are of a uniform size and shape, and characterized by a
"fishbone" pattern of deforestation, allowing for easy identification of each landholding
on the satellite image. McCracken et al. (2002) initially used digital property maps
obtained from INCRA, the government agency responsible for delineating new
settlements. They found that these maps were often inaccurate or incomplete, since their
purpose was simply to identify which plots belonged to whom, not to conduct detailed
GIS analyses. They were able to develop a suitable farm property grid for 402
households using a combination of the INCRA maps, GPS points collected in the field,
and mathematical interpolation of GIS layers.
In some study regions, no property maps are available, and properties may be
irregularly shaped or not legally owned by the user. In these cases, researchers have had
to create property maps using GPS points collected in the field, combined with sketch
maps of the property and identification of key landscape features visible on the image.
This method is time and labor intensive. Walsh et al. (2002) employed this method,
along with a base maps of roads, to identify 418 landholdings in the Ecuadorean Amazon.
Vance and Geoghegan (2002) used GPS points and participatory mapping to
outline 188 properties within the ejido systems in Quintana Roo and Campeche, Mexico.
In other regions, landowners reside in one place, while agricultural activities occur
somewhere else. This was the case in northeastern Thailand, where Rindfuss et al.
(2002) selected 310 villages and experimented with various methodologies to link
households to their field plots. This process is even more difficult and labor intensive
when agricultural plots do not contain the dwelling unit.
In each of these studies, researchers interviewed parcel owners and compiled data
on socioeconomic conditions and land use decisions of each family. These data were
then related to land cover data in order to better understand why different families might
make different land use decisions.
This study will employ a combination of these methods to carry out a property-
level study of land use change in Fortaleza, a rural settlement near Una, Brazil, and to
further develop the methodology for such research. As noted by previous researchers,
difficulties in accurately identifying individual properties limit the accuracy of results.
Thirty household interviews were conducted in June and July of 2005, as part of an
evaluation of the Jupara Agroecology Extension Program. Through a partnership with
WWF, Jupara has been working in the area since 1995 to promote organic agroforestry
and forest conservation. Specific program goals included the implementation of
agroecological practices and maintenance of 40% of each landholding in agroforestry and
30% in natural forest cover. Taking into account a large community forest reserve,
individual landholders were expected to maintain 12.5% of their own landholding in a
forest reserve, to maintain 30% of the overall community landscape in forest cover.
Results of the program evaluation interviews indicated that while many participants
have implemented organic agricultural practices, only 42% of participants, and 43% of all
farmers interviewed, have maintained a forest reserve equal to at least 12.5% of their
Interview data offer only a snapshot of one point in time. Many landowners report
that although the area was almost completely forested when the first settlers arrived in the
1970s, extensive deforestation had occurred before the Jupara program began in 1995.
More recent settlers reported significantly more remaining forest cover than those who
had been living in Fortaleza for longer periods of time. Knowing the distribution of land
cover types that landowners report at present, this remote sensing analysis was designed
to assess the extent of forest cover in the community before and after the Jupara project.
Images from 1986, well before the project began, and 2001, the most recent image
available, were analyzed to address the following research questions:
* How has forest cover changed in Fortaleza between 1986 and the present?
* How much forest cover was present before the Jupara program began?
* Is the program asking landowners to conserve something that wasn't present to
* Are program participants more likely to allow forest regrowth than non-
* Is land cover data reported in interviews consistent with land cover data observed
on satellite images?
Fortaleza is a land reform settlement near Una, BA. Fifty families are settled on
1102 hectares, with an average lot size of 16 hectares (INCRA 2006). Approximately
250 hectares are in community areas including a forest reserve and a small community
cacao plantation. Families began settling the area in the mid 1970s. At that time, the
area was completely forested. Most of the settlers in Fortaleza grew up on their parents'
farms or worked on nearby rubber plantations before settling there.
INCRA (Instituto Nacional de Colonizagdo e Reforma Agraria, the government
agency responsible for land reform) officially divided the land and gave titles to these
landholders in 1997. At this time, eight new lots were created along the westernmost
perimeter of the community, in an area that had been a part of the community forest
reserve. By locating families along the perimeter of the reserve, INCRA hoped to avoid
invasion of this land by outsiders. Eight families were moved from their original plot of
land to one of these new, completely forested plots, and other families had a portion of
their original land officially titled to a neighbor.
In most cases, the male head of household received a title to his individual parcel
of land. Farmers in Fortaleza produce mainly rubber and cacao as cash crops, as well as
some manioc, corn, fruit and vegetable crops for consumption.
Of the 50 families settled in Fortaleza, nine are currently members of Jupara and
COOPASB, Jupara's agricultural cooperative. Nine families were formerly members and
dropped out, and the remainder of residents may have participated in some Jupara
training events but never formally joined the cooperative. For the purposes of this
analysis, we compare members, former members, non-members, and families on new
lots. The families on new lots are also non-members but are analyzed separately due to
the more recent process of land cover change on these lots.
The Atlantic Forest of Southern Bahia
Fortaleza is located in the Atlantic Forest Region of Southern Bahia, Brazil, and
has been targeted for conservation efforts due to it's proximity to the Una Biological
Reserve. Figure 4-1 shows the location of the study site and the reserve. The Una
Reserve encompasses about 11,000 hectares of what remains of the Atlantic Forest
ecosystem (Buschbacher in prep.). The Atlantic Forest, which once covered much of the
eastern coast of Brazil as well as parts of Argentina and Paraguay, is one of the world's
most diverse ecosystems, and also one of the most endangered. The first mapping of the
Brazilian Atlantic Forest was undertaken in 1990 with the participation of IBAMA
(Instituto Brasiliero do Meio Ambiente e dos Recursos Naturais Renovaveis, or the
Brazilian Institute of the Environment and Renewable Natural Resources) and the
Fundacgo SOS Mata Atlantica (SOS Atlantic Forest Foundation), a prominent Brazilian
NGO. According to this project, the Atlantic Forest originally covered 15% of the
Brazilian national territory, and has been reduced to a mere 7% of its original area, or
about 1% of the Brazilian National Territory. The Atlantic Forest is home to 108 million
people, or about 60% of the population of Brazil (Hirota 2003).
SIESB 2004 Image, 5121101' -
rea around ~J Reserve-w =
Figure 4-1. Location of study site. The blue box indicates the Southern Bahia region with
the original extent of the Atlantic Forest shown in yellow and remaining forest
in green. The smaller black box indicates the area around Fortaleza and the
Una Reserve. May 2001 Landsat image of the reserve and Foratleza is shown
on the right.
The Atlantic Forest Biome is on the Global 200 list of globally outstanding
ecoregions, and is considered a Biodiversity Hot Spot (Mittermeier et al. 1999). The
region supports over 1600 species of terrestrial vertebrates and 20,000 species of vascular
plants, including more than 6000 endemic plant species and more than 500 endemic
animal species (Mittermeier et al. 1999).
In order to set priorities for conserving what remains of the fragmented Atlantic
Forest, two biodiversity corridors have been designated; the Serra do Mar Corridor,
which extends southwest of Rio de Janeiro through Minas Gerais, Sao Paulo, and Parana
states, and the Corridor Central, in Southern Bahia and Espiritu Santo (Figure 4-2)
(Aguiar et al. 2003). The Una Reserve and Fortaleza fall within the Central Corridor.
Within the Central Corridor, researchers have documented extremely high tree diversity
(Thomas et al. 1998), twelve endemic primate species (Pinto 1994) and 50% of the
known endemic bird species of the Atlantic Forest (Aguiar et al. 2003). The Central
Corridor encompasses almost 12 million hectares, and about 12% of its total area is
covered in native forest (Conservation International 2006). The corridor contains at least
40 protected areas, 70% of which are state-owned. The average protected area size in the
region is 93.13 km2 (Aguiar et al. 2003). Financial resources needed to enforce protected
areas and establish new ones are limited, so an emphasis on privately owned protected
areas and community managed ones, like the community reserve in Fortaleza, is essential
for the long-term conservation of the Atlantic Forest.
Figure 4-2. Location of the Central and Serra do Mar Corridors.
The Central Corridor includes the cacao growing region of Southern Bahia.
Currently about 600,000 hectares of Theobroma cacao, a medium-sized understory tree,
are planted in this region (Buschbacher in prep.), many maintain some of the original
forest canopy or are interspersed with patches of natural forest (Alger 1998; Alves 1990;
Buschbacher in prep.). This region's cacao agroforests have generated considerable
interest within the conservation community for their potential to maintain forest-like
landscapes and serve as biological corridors between blocks of fragmented forest (Alves
1990). Researchers in many countries have demonstrated that while agroforestry systems
are no substitute for intact forest, some wildlife species are able to use cacao-based
systems to some extent as a corridor between areas of more suitable habitat, with more
diverse systems providing more suitable habitat than less diverse systems (Alves 1990;
Greenberg 1998; Greenberg et al. 2000; Laurance 2004; Reitsma et al. 2001). Pardini
(2004) carried out such a study on small mammals on fragmented areas of the Una
Biological Reserve. For this reason, conservation programs like the one implemented by
Jupara, targeting cacao producers and producers of other agroforestry crop such as
rubber, cloves, palms, and fruit, are important to maintain connectivity of the region's
remaining forest fragments.
Remote sensing studies in the Atlantic Forest
The presence of extensive agroforestry further complicates remote sensing studies
in this region. In regions such as the Amazon and Mexico's dry tropical forests, land
cover can usually be divided into forested and non-forested, as annual crops or pastures
for livestock are common land uses. Agroforests, on the other hand, can be extremely
difficult to separate from natural forests and forest regrowth, especially in the case of
cabruca systems, in which cacao is planted in the shade of natural forest overstory trees
(Lemos Costa 2000; Saatchi et al. 2001), or very old agroforestry systems. Fortaleza
residents harvest rubber and cacao from agroforestry systems that may be more than
thirty years old, although these are usually not cabruca systems but mixed plantations of
rubber, cacao, and other fruit species.
Only a few remote sensing studies have been carried out in the Atlantic Forest
region. Lemos Costa (2000); Saatchi et al. (2001), and Araujo (1997) all carried out
broad scale classifications of the Southern Bahia region using Landsat Imagery and
NDVI (Normalized Difference Vegetation Index). NDVI is a vegetation index
commonly used as a measure of plant biomass. Lemos and Araujo both defined forested
areas, agroforests and open areas by assigning a characteristic range of NDVI values to
each. Use of NDVI alone, however, may be unreliable, because NDVI has been shown
to be ineffective in classifying uneven aged, humid tropical forests where the biomass can
be very high (Sader et al. 1989).
Several other studies have relied on more expensive, higher resolution imagery to
distinguish between similar land cover classes like forests and agroforestry, such as
IKONOS imagery (Guanes Rego & Koch 2003), SIR-C Radar imagery (Saatchi et al.
2001), and CASI imagery, which is similar to an aerial photograph (Olson 1998). Due to
the financial constraints of this study, this approach was not considered.
A total of 30 household interviews were conducted in Fortaleza in June and July of
2005. The fifty existing households were classified as participants, former participants,
or non-participants in the Jupara project. Current members of the Jupara's agricultural
cooperative, Coopasb, were considered participants, past members were former
participants, and non-members were considered non-participants. All of the participants
and former participants were interviewed, and of the remaining non-participants, a
random sample of 10 was selected, for a total of 30 interviews. Interview content
included questions concerning family demographics, agricultural practices, and land
cover distribution, and participation in Jupara training activities.
Participatory Mapping and Training Points
For each household interviewed, the researcher also toured the property with the
landowner to verify information obtained in the interview, and to collect GPS training
points and a brief description of each of the land cover types present on the property for
use in image classification. Farmers reported five land cover types: Forests,
Agroforestry, "Capoeira," which includes forest regrowth and fallowed areas, Annual
Crops, and Pasture. To assist participants in describing their property's land cover
distribution, a participatory mapping activity was carried out in which participants were
asked to draw a simple sketch map of their landholding, including agroforestry systems
and forested areas, areas of annual crops and pastures, roads, waterways, and houses. A
representative sketch is shown in Figure 4-3. GPS points were collected along the
perimeter and covers of these sketches, and at other important landscape features, to
assist in placing the property boundary on the image.
Landsat images from September 11, 1986 and May 23, 2001 were acquired from
the University of Maryland's Global Land Cover Facility (University of Maryland 2006).
Image dates were chosen based on availability of cloud-free imagery, and seasonal
differences in vegetation may influence results somewhat. The most recent Landsat
images (2003-present) are affected by an equipment malfunction and were not used for
Figure 4-3. Property sketch created by a community member.
Bahia). Fortaleza is located in a very rural area, roads are not paved and are not visible'
I / /
", T/ / |
Figure 4-3. Property sketch created by a community member.
Images were subset to a manageable size. The 2001 image was georectified to a
map ofwaterways provided by IESB (Instituto de Estudos Socioambientais do Sul da
Bahia). Fortaleza is located in a very rural area, roads are not paved and are not visible
on the image. The 1986 image was then georectified to the 2001 image. Images were
calibrated to remove differences in haze and light angle between the two image dates. An
NDVI layer and a texture layer were added to the original image, to provide additional
information for image classification.
The most difficult part of a property-level remote sensing study is locating the
properties on the image. INCRA in Salvador provided a digital map of Fortaleza. The
map included only the community outline, roads, and streams; individual property
outlines were not included, and UTM coordinates appeared to be off by a factor of 1000.
Jupara provided a photocopy of another map, also produced by INCRA, which did
include property boundaries. This photocopied map was scanned and georectified in
Erdas Imagine, using points extracted by hand and multiplied by 1000 from the digital
map. The scanned, georectified map was then traced with a mouse to create an
independent vector layer that could be projected on top of the image. This multistep
process created some error in the location of property boundaries, and some GPS points
collected in the field did not fall into the correct property polygons. INCRA maps are not
particularly accurate to begin with, as noted by other researchers (McCracken et al.
2002), since their main purpose is simply to keep records of who owns which lot, not to
conduct scientific research. Accuracy of this analysis also depends on the accuracy of the
IESB map used for image georectification. Additional GPS data from the field could
improve the accuracy of this map. However, the location of several large landscape
features, such as the community forest reserve and a large cleared ranch just to the south
of the community boundary, confirm that the vector layer is in approximately the correct
The vector layer was projected onto the images and images were subset again to
include only Fortaleza. Images were classified using methods of unsupervised
classification (Jensen 2005). Computer-generated classes were identified as Forest,
Agroforestry and Forest Regrowth, Cleared Areas, and Clouds and Shadows, based on
training points collected in the field and general knowledge of the landscape. Farmers
reported five land cover categories, but since some of these are difficult to separate on the
image, images were classified into only three classes, plus clouds and shadows. "Forest"
on the image could include forest and some mature forest regrowth or very old
agroforestry systems. "Agroforestry and Forest Regrowth" were one category for the
purposes of image classification, and include agroforestry and regrowth that farmers
reported as "capoeira." "Cleared Areas" include annual crops, pastures, and some
recently fallowed capoeiras.
Classifications were imported into ArcGIS and separate layers were created for
each land use class for each image date. Clouds and shadows were eliminated from the
analysis. The vector layer was modified so that each landholding was a separate polygon
with its own identification number. Using the zonal statistics tool, the area of Forest,
Agroforestry/Regrowth, and Cleared Areas within each polygon was calculated for the
1986 and 2001 classifications. Percent change in each land cover type between 1986 and
2001 was calculated for each polygon.
Land Use Classifications
Classifications of 1986 and 2001 images show areas of forest, agroforestry or
regrowth, and open areas within property polygons and communal areas (Figures 4-4 and
4-5). Areas in which the land use changed from the 1986 image to the 2001 image are
highlighted in Figure 4-6. Properties were divided into several categories for the
purposes of this analysis. Table 4-1 shows distribution of Forest, Agroforestry (AF), and
Cleared Areas for property owners who participated in the Jupara program, former Jupara
participants, and non-participants. Lots which were not yet in use in 1986 were analyzed
separately and are labeled New Lots. (All of the families on new lots were also non-
participants, but results for non-participants on new lots are not included in results for
non-participants on older lots.) The community reserve was also analyzed separately.
Land Use in Fortaleza, 1986
- Former Participants
- New Lots (Non-Participants)
- Community Areas
Cleared Areas/Annual Crops/Pasture
= | Clouds and Shadows
460 920 1,840 2,760 3,680
m I Mete
Figure 4-4. Land Use Classification for Fortaleza, 1986.
Land Use in Fortaleza, 2001
0 480 920
1,840 2,760 3,680
- Former Participants
- New Lots (Non-Participants)
- Community Areas
Cleared Areas/Annual Crops/Pasture
Figure 4-5. Land use classification for Fortaleza, 2001
Land Use Change in Fortaleza, 1986 2001
- Former Participants
- New Lots (Non-Participants)
- Community Areas
M Cleared-> Forest
AF/Reg. -> Forest
I I Cleared -> AF/Reg
SForest -> AF/Reg.
SAF/Reg. -> Cleared
M Forest -> Cleared
= | No Change
0 460 920
1,840 2,760 3,680
Figure 4-6. Area of Land Cover Change in Fortaleza,1986 2001
Table 4-1. Land use change in Fortaleza, 1986-2001.
Forest Forest % Change AF/R AF/R % Change Cleared Cleared % Change
1986 2001 in Forest 1986 2001 in AF/R 1986 2001 in Cleared
Non-Participants 16% 24% 8% 40% 46% 6% 38% 34% -4%
Former Participants 25% 27% 2% 50% 48% -2% 21% 25% 5%
Participants 14% 34% 20% 44% 35% -8% 41% 35% -5%
New Lots 84% 16% -68% 5% 16% 11% 7% 28% 22%
Community Reserve 81% 78% -3% 14% 16% 2% 5% 6% 1%
Total 32% 31% -2% 35% 39% 3% 28% 31% 2%
Figure 4-7 shows the percent change in each cover class for lots owned by
participants, former participants, and non-participants, as well as on new lots and within
the community forest reserve, and for the entire community area. The overall quantity of
these three land use classes remained essentially the same in Fortaleza between 1986 and
2001, but the spatial distribution of forest patches, agroforestry, and cleared areas have
changed. Several new lots have been settled in what was once a forested area, forest
cover has decreased in these areas as well as in the community forest reserve.
Meanwhile, forest cover has increased on older agricultural lots. Forest cover increased
by 20% on lots owned by Jupari participants, compared with a 9% increase on non-
participants' lots, and 2% increase on former participants' lots. In addition, the land
cover maps show that much of the increase in forest cover occurs along two riparian
Land Use Change in Fortaleza, 1986-2001
0 N FP P EW RESERVE TOTAL
Figure 4-7. Land use change in Fortaleza on properties owned by non-participants (NP),
former participants (FP), and current participants (P) in the Jupara project, as
well as on newer lots and within the community forest reserve.
Comparison with Interview Data
The participants in this study reported a land use distribution similar to the one
observed on the Landsat image. Figure 4-8 shows the average land use distribution
reported by interviewees, and Figure 4-9 shows the community-wide distribution
observed on the 2001 classification. The interview data included additional classes
which are difficult to discern on satellite imagery; forest regrowth and fallows are a
separate category, and open areas are divided into annual crops and pastures.
Land Use in Fortaleza as observed on 2001 Landsat Image
E Agforestry/Regrowth 0 Cleared Areas 0 Forest
Figure 4-8. Distribution of land uses in Fortaleza as measured by classification of 2001
Land use in Fortaleza as reported by farmers
[ Agroforestry m Annual Crops Pasture m Capoeira Forest
Figure 4-9. Distribution of land uses in Foratleza as reported by farmers interviewed in
Remote Sensing Analysis
This analysis shows that the overall proportions of forest, agroforestry or forest
regrowth, and open areas in Fortaleza were nearly the same in 2001 as in 1986, with
slight increases in agroforestry (3%) and open spaces (2%), and a slight decrease in forest
(-2%). However, analysis of the satellite imagery reveals that the spatial distribution of
forest patches has changed considerably since 1986. In the eastern part of the
community, where settlers have been present longest, and where little forest existed in
1986, we see regeneration of forested areas, especially in two riparian corridors which
run north-south through the community. Areas classified as "forest," then, include not
only the remaining original forest but some areas of forest regrowth. Once the secondary
forest reaches a certain stature, it becomes indistinguishable from mature forest on the
In some areas of the community, riparian zones are marked by a steep decline
toward a small stream, and farmers might find these sloped areas unsuitable for
agriculture. Many interviewees cited the protection of water resources as a main reason
for conserving forest. Most families depend on natural springs along this stream as their
source of drinking water, and several farmers told of incidences in which streams had
dried up after the removal of a forest patch, and reappeared as the forest was allowed to
Alternately, GPS points collected in the field suggest that a few of these areas that
appear to be "forested" are actually mature agroforestry systems, which appeared as
young trees in 1986 but are now large enough to be mistaken for forest in the Landsat
classification. Some farmers reported agroforestry systems more than 30 years old that
included some very large trees, primarily rubber (Hevea brasilensis), shade species
planted with cacao, and a few very large fruit trees including j ackfruit (Artocarpus
heterophyllus) and mango (Mangifera spp. ).
The eight westernmost lots in the community were not yet settled in 1986 and were
still completely or mostly forested. INCRA officially demarcated and titled the entire
community in 1997, and at that time several families were relocated to the perimeter of
this forested area. According to local residents, INCRA reasoned that by locating
families around the perimeter of the community forest reserve, invasion of this land by
non-community members could be prevented. Unfortunately, most of the relocated
families were granted completely forested land, and so had no choice but to deforest it.
Sixty-eight percent of the forest on these lots had disappeared by 2001.
Within the community forest reserve, we also see a 3% decrease in forest cover.
This could be accounted for by the fact that a lot on the edge of the reserve was used to
build a second school in the mid 1990s. A visit to the area does indicate some level of
human disturbance and removal of logs, and some families along the perimeter of the
reserve may be using land actually located within the reserve. Although some
degradation is evident, the community forest reserve is still by far the largest patch of
forest in the community, and still the easiest way to conserve forest in Fortaleza, as no
one person has autonomous decision-making power for this land, and the community has
established an expectation for conservation of the area.
A landowner-by-landowner analysis of the lots in the eastern part of Fortaleza
offers good news for the effectiveness of the Jupara program. Current program
participants appear to be far more likely than non-participants to allow forest regrowth on
their properties (or the development of very old agroforestry systems). Forest cover
increased by 20% on participants' lots, compared to an 8% increase on non-participants'
lots, and a 2% increase on former participants' lots. Participants had the least forest in
1986 (14%) and have the most forest of any group now (34%). Participation in a
conservation program may be only one factor influencing this change. Location of the
lots along steep or riparian areas may also be a factor. Some of the participants were
among the first settlers to arrive in Fortaleza in the 1970s, so the development of very old
agroforestry systems or regeneration of more mature forest patches may be a pattern that
develops only after many years of settlement.
Comparison with Interview Data
By interviewing farmers, we can get a more specific idea of the possible land uses
within each of the three categories observable on the image. Interview data represents
the mean of a sample of 30 farmers, while the image analysis includes all 50 properties in
Fortaleza, the community reserve, and two smaller community areas.
Farmers reported a land use distribution fairly consistent with the distribution
observed through image analysis. Farmers reported slightly more forest and agroforestry
than appeared on the image, and slightly less cleared area. This could be because some
young agroforestry systems or recently fallowed areas appeared as cleared areas on the
image classification, or because cleared areas actually decreased between 2001, when the
image was collected and 2005, when the interviews were conducted. Interview data are
based on a representative sample of 30 households, weighted so that participating and
non-participating households are proportionately represented, while the Landsat
classification generated data for all landholdings; this could also explain slight
discrepancies. Overall, the results were similar, and this serves to confirm both the
accuracy of data reported by farmers and of our Landsat classification.
Jupara Program Goals
At the start of the extension program, Jupara program leaders, along with WWF
partners, set goals for forest conservation in participating communities. In Fortaleza, the
goal was to maintain 30% of the community's total area in natural forest. The
community reserve protects about 20% of the total area. On the remaining 80% of the
land, each farmer would have to conserve 12.5% of his or her own plot, to reach the
overall goal of 30% forest cover. Of the 30 farmers interviewed, 14 of them, or 46%,
report that their property meets this goal, when only original forest cover is considered.
According to the image classification, which includes forest regrowth as well as original
forest cover, 46 of the 50 properties, or 92%, meet this goal. The image analysis used a
broader definition of "forest:" regrowth areas that farmers usually define as "( i,,,,it/"
were included here, and very old agroforestry systems may also be included. Including
both the reserve and forested areas on individual lots, 31% of the community area was
forested in 2001, indicating that Fortaleza does meet Jupara's conservation goal, when
areas of regrowth are taken into consideration.
Areas of regrowth may not provide the same conservation benefits as the original
forest, as regrowth areas are likely to be more fragmented and include a different species
composition, and may or may not provide suitable wildlife habitat. At the same time, it is
important that conservation programs like Jupara set goals that are attainable, in order to
maintain the enthusiasm and participation of local residents. Adopting a broader
definition of forest conservation may make goals more attainable, especially in an already
deforested area like Fortaleza. Lots that were settled in the 1970s included little forest
cover in 1986, so regeneration of forested areas is really the best that conservationists can
hope for in this case. Regrowth areas and very old agroforestry systems can provide
important environmental services, such as the protection of water sources, and can act as
corridors between areas of remaining Atlantic Forest. The tendency of participants to
allow forest regrowth to a greater extent than non-participants indicates that the Jupara
program has succeeded in instilling this conservation ethic in Fortaleza.
Directions for Further Study
The accuracy of this assessment could be improved by returning to the field and
collecting additional GPS points in Fortaleza, especially in areas of forest regrowth. A
random sample of these areas could be visited to determine whether they are in fact areas
of forest regrowth or very old agroforestry, and the age of the regrowth or agroforestry.
Additional fieldwork could also help to more accurately locate properties on the satellite
image. Walking the perimeter of a few representative properties with a GPS might be
one way to accomplish this.
The community forest reserve also merits further research. While forest cover is
increasing outside of the reserve, the remote sensing analysis indicates some degradation
of the reserve itself. Who is using this area? Do users have the permission of the larger
community to use this community resource? Community reserves are one effective way
to set aside some of the larger fragments of remaining Atlantic Forest. The agrarian
reform movement is constantly establishing new communities throughout Southern
Bahia, and many include a community forest reserve. An in-depth study of the
management of these reserves in well-established communities such as Fortaleza might
allow planners to more appropriately select areas for community reserves and create
management plans for the long term-conservation of these reserves.
As with any large project, the Jupara Agroecology Project had its strengths and
weaknesses. The project did have a notable impact on the use of organic agricultural
practices in the three communities studied, as shown in Chapters 2 and 3. The project
also had many qualitative benefits that were not necessarily captured here, including
improving community organizations, developing leaders, strengthening women's
participation in agriculture and community organizations, and improving farmers'
attitudes toward conservation. Almost all of the farmers interviewed, participants and
non-participants alike, spoke very highly of the program and its benefits to the
community. Specific comments from participants included the following:
"I learned to preserve nature, to work in an agroecological system, and to stop
killing insects that could be beneficial." Aloisio, Lagoa Santa
"I learned to speak in an assembly and not be timid. I gained knowledge of both
agriculture and commercialization." -Andre, Lagoa Santa
"I learned to dialogue better with people. If you have a problem, you can discuss it
with others and know that it's not so serious." -Maria, Fortaleza
Interview data indicated that the project may not have consistently met its goals for
forest conservation in Fortaleza and Lagoa Santa. In Cascata, the forest conservation
goal was met through the establishment of a large community reserve. The remote
sensing analysis in Chapter 4 helps present a clearer picture of the dynamics of land use
change in Fortaleza. Although the overall amount of forest cover remained almost
constant between 1986 and 2001, forest cover increased on older family lots and
decreased within the forest reserve area as some parts of the reserve were converted to
family lots after properties were reassigned by INCRA. All of the project participants
were situated on older lots and had little natural forest left to conserve when the project
began in 1995, but we do see a trend in forest regrowth in Fortaleza, and it appears that
project participants are allowing more forest regrowth on their properties than non-
This research could be strengthened by designing similar remote sensing analyses
for Lagoa Santa and Cascata, and for the regional landscape. Each community has a
distinct land use history, and each could offer an independent and interesting land use and
land cover change study.
The land use data presented here also highlight the importance of collecting
baseline data when a project like this begins, in order to set feasible conservation goals
and more accurately monitor progress toward these goals. Communities are very large,
and some kinds of agroforestry systems and forests in this region may look alike at first
glance. It would be easy for an extensionist attending a community meeting or visiting a
limited number of families to misestimate the distribution of land uses if no rigorous data
collection were carried out. Jupara's original conservation goals, which seem to have
been set somewhat arbitrarily, may not have been very feasible if little forest was left to
conserve in the target communities. Goals might have included the establishment of
forest reserves by allowing forest regrowth, rather than, or in addition to, conserving
existing forests. The participatory mapping activity indicated that many farmers had
never given much thought to the overall landscape of their farms and the spatial
arrangement of land uses. A similar activity at the outset of a conservation program
would allow farmers to conceptualize what is present and plan changes that might be
possible for future years. The use of remote sensing is also an excellent tool for
conservation planning, when this technology is available. Collecting baseline data could
allow Jupara and other conservation organizations to focus their efforts on communities
where conservation goals are most likely to be met. Communities involved in this project
were selected mainly based on their proximity to the Una Reserve, or in some cases, their
longstanding relationship with Jupara and involvement in past Jupara projects.
Jupara also hoped to increase family incomes. Participating families do tend to
have higher agricultural incomes than non-participating ones, but this could be because
wealthier families, or families depending more heavily on agriculture for their income,
tended to participate in the project more often than other families. When fieldwork was
completed in 2005, farmers in Cascata were marketing their cacao collectively through
the Coopasb cooperative and receiving a higher price because the cacao was organic.
Cargill, an international exporter of various agricultural products, was the main buyer of
Coopasb's organic cacao. In 2006, Cargill decided to stop offering a premium price for
organic certified cacao, and this year Cascata farmers sold their organic cacao to local
buyers for the same price as non-certified cacao.
Farmers in Fortaleza and Lagoa Santa did not report an increase in income as a
result of participation in the cooperative. Most did not regularly sell products through the
cooperative. Coopasb has been most active in the marketing of organic cacao, which is
advantageous for cacao communities like Cascata, but less advantageous for communities
with mixed production, like Fortaleza and Lagoa Santa. Perhaps in the future they will