ArticlePDF Available

Regional integration and local change: Road paving, community connectivity, and social-ecological resilience in a tri-national frontier, southwestern Amazonia


Abstract and Figures

Initiatives for global economic integration increasingly prioritize new infrastructure in relatively remote regions. Such regions have relatively intact ecosystems and provide valuable ecosystem services, which has stimulated debates over the wisdom of new infrastructure. Most prior research on infrastructure impacts highlights economic benefits, ecological damage, or social conflicts. We suggest a more integrative approach to regional integration by appropriating the concepts of connectivity from transport geography and social–ecological resilience from systems ecology. Connectivity offers a means of observing the degree of integration between locations, and social–ecological resilience provides a framework to simultaneously consider multiple consequences of regional integration. Together, they offer a spatial analysis of resilience that considers multiple dimensions of infrastructure impacts. Our study case is the southwestern Amazon, a highly biodiverse region which is experiencing integration via paving of the Inter-Oceanic Highway. Specifically, we focus on the “MAP” region, a tri-national frontier where Bolivia, Brazil, and Peru meet and which differs in the extent of highway paving. We draw on a tri-national survey of more than 100 resource-dependent rural communities across the MAP frontier and employ indicators for multiple dimensions of connectivity and social–ecological resilience. We pursue a comparative analysis among regions and subregions with differing degrees of community connectivity to markets in order to evaluate their social–ecological resilience. The findings indicate that connectivity and resilience have a multifaceted relationship, such that greater community connectivity corresponds to greater resilience in some respects but not others. We conclude by noting how our findings integrate those from heretofore largely disparate literatures on infrastructure. The integration of transport geography with resilience thought thus stands to advance the study of infrastructure impacts.
Content may be subject to copyright.
Regional integration and local change: road paving, community
connectivity, and social–ecological resilience in a tri-national
frontier, southwestern Amazonia
Stephen G. Perz
Liliana Cabrera
Lucas Arau
jo Carvalho
Jorge Castillo
Rosmery Chacacanta
Rosa E. Cossio
Yeni Franco Solano
Jeffrey Hoelle
Leonor Mercedes Perales
Israel Puerta
Daniel Rojas Ce
Ioav Rojas Camacho
o Costa Silva
Received: 9 July 2010 / Accepted: 2 May 2011 / Published online: 25 May 2011
Ó Springer-Verlag 2011
Abstract Initiatives for global economic integration
increasingly prioritize new infrastructure in relatively
remote regions. Such regions have relatively intact ecosys-
tems and provide valuable ecosystem services, which has
stimulated debates over the wisdom of new infrastructure.
Most prior research on infrastructure impacts highlights
economic benefits, ecological damage, or social conflicts.
We suggest a more integrative approach to regional inte-
gration by appropriating the concepts of connectivity from
transport geography and social–ecological resilience
from systems ecology. Connectivity offers a means of
observing the degree of integration between locations, and
social–ecological resilience provides a framework to
simultaneously consider multiple consequences of regional
integration. Together, they offer a spatial analysis of resil-
ience that considers multiple dimensions of infrastructure
impacts. Our study case is the southwestern Amazon, a
highly biodiverse region which is experiencing integration
via paving of the Inter-Oceanic Highway. Specifically, we
focus on the ‘MAP’ region, a tri-national frontier where
Bolivia, Brazil, and Peru meet and which differs in the
extent of highway paving. We draw on a tri-national survey
of more than 100 resource-dependent rural communities
across the MAP frontier and employ indicators for multiple
dimensions of connectivity and social–ecological resilience.
We pursue a comparative analysis among regions and sub-
regions with differing degrees of community connectivity to
markets in order to evaluate their social–ecological
S. G. Perz (&)
Department of Sociology and Criminology and Law,
University of Florida, 3219 Turlington Hall,
Gainesville, FL 32611, USA
L. Cabrera
n de Interaccio
n Social, Universidad Amazo
de Pando, Cobija, Pando, Bolivia
L. A. Carvalho
Departamento de Economia e Mestrado em Desenvolvimento
Regional, Universidade Federal do Acre, Rio Branco, Acre,
J. Castillo R. Chacacanta Y. F. Solano
Departamento Acade
mico de Ecoturismo y Administracio
Universidad Nacional Amazo
nica de Madre de Dios,
Puerto Maldonado, Madre de Dios, Peru
R. E. Cossio
School of Natural Resources and Environment,
University of Florida, Gainesville, FL, USA
J. Hoelle
Department of Anthropology, University of Florida,
Gainesville, FL, USA
L. M. Perales
apari, Madre de Dios, Peru
I. Puerta
Universidad Amazo
nica de Pando, Cobija, Pando, Bolivia
D. Rojas Ce
Departamento de Ciencias Biologicas y Naturales,
Universidad Amazo
nica de Pando, Cobija, Pando, Bolivia
I. Rojas Camacho
Departamento de Informa
tica, Universidad Amazo
nica de
Pando, Cobija, Pando, Bolivia
A. Costa Silva
Mestrado em Desenvolvimento Regional, Universidade Federal
do Acre, Rio Branco, Acre, Brazil
Reg Environ Change (2012) 12:35–53
DOI 10.1007/s10113-011-0233-x
resilience. The findings indicate that connectivity and
resilience have a multifaceted relationship, such that greater
community connectivity corresponds to greater resilience in
some respects but not others. We conclude by noting how
our findings integrate those from heretofore largely dispa-
rate literatures on infrastructure. The integration of transport
geography with resilience thought thus stands to advance the
study of infrastructure impacts.
Keywords Infrastructure Connectivity Resilience
Amazon Globalization
In the opening years of the new millennium, initiatives for
economic integration are proliferating. This is especially
the case for investments to penetrate heretofore remote
regions in order to finally gain easier and more secure
access to their natural resources. For multilateral banks and
many national governments, it is necessary to invest in
expanded extraction of natural resources for transportation
to industrial processing and export nodes. This not only
earns foreign exchange income for the banks and countries
involved but also provides new supplies of raw materials
for the global economy.
The primary mechanism for incorporating remote
regions into commodity chains has been via infrastructure
initiatives. Interest in resource development has, however,
met with concerns about negative environmental impacts
of infrastructure projects, particularly in regions that pro-
vide important ecosystem services. A key question then
concerns the social and ecological implications of regional
We take up these issues by pursuing an integrative
approach to the issue of the social–ecological impacts of
infrastructure as a means for incorporating remote regions
into the global economy. To do so, we draw on insights
from transport geography and the resilience literature and
focus on the relationship between connectivity and social–
ecological resilience. We deploy regional transport geog-
raphy to motivate an evaluation of local and regional
connectivity by observing the extent of incorporation of
rural communities into local and regional markets. This
allows spatial differentiation of community-market con-
nectivity to permit a spatial analysis of variation in market
integration across a region receiving new infrastructure.
We invoke the concept of social–ecological resilience to
focus on the interface between rural communities and their
natural resources in the presence of infrastructure initia-
tives. Resilience thinking allows us to go beyond a narrow
economic focus to pursue a broader, interdisciplinary
evaluation of the multiple ramifications of new infrastruc-
ture. The combination of transport geography with resil-
ience thinking thus allows for a spatially differentiated
evaluation of the importance of community-market con-
nectivity for multiple dimensions of social–ecological
resilience in the context of new infrastructure. This is
innovative because the use of connectivity spatializes the
analysis of resilience, whereas the resilience approach
integrates heretofore largely disparate literatures on infra-
structure impacts.
Our study case is the Amazon, a region well known for
past infrastructure projects, frontier expansion for eco-
nomic integration, and debates about the social and eco-
logical consequences. We focus on the southwestern
Amazon, which includes substantial natural resources,
most famously its high biodiversity. This region is now
being economically integrated via paving of the Inter-
Oceanic Highway, which will link this area to Atlantic
ports in southern Brazil as well as to Pacific ports in Peru.
The heart of the southwestern Amazon is also a tri-national
frontier, called the ‘MAP’ region, named after the three
states that comprise the construction zone of the Inter-
Oceanic Highway: Madre de Dios (Peru), Acre (Brazil),
and Pando (Bolivia). The three sides of the MAP region
differ in the extent of highway paving, which permits a
comparative analysis of connectivity and resilience.
Our focal research question links regional transport
geography via the concept of connectivity to resilience
thinking and asks whether market connectivity corresponds
to greater or less social–ecological resilience. Our analysis
is multiscale and compares the three sides of the MAP
frontier as well as subregional spaces within each side. The
comparisons focus on communities with differing degrees
of connectivity, defined not only in terms of accessibility as
via road paving and market distance, but also by the extent
of community-town interactions via social ties and product
marketing. We employ these dimensions of connectivity to
evaluate community resilience in terms of collective
memory, livelihood diversity, and adaptive capacity, which
constitute different dimensions of resilience. This allows
for a multidimensional analysis of the relationship between
connectivity and resilience, which is likely to be multi-
faceted because both concepts are themselves multidi-
mensional. The findings confirm the multifaceted
relationship between connectivity and resilience, such that
greater connectivity corresponds to greater community
resilience in some dimensions but not others. These find-
ings integrate those from previous but fragmented research
on road impacts and bear implications for the study of
infrastructure via the combination of transport geography
and social–ecological resilience, to which we return in our
concluding discussion.
36 S. G. Perz et al.
We prefigure our analysis by focusing on specific types of
integration initiatives, particularly in Latin America,
especially large-scale infrastructure projects. This moti-
vates a discussion of the economic, ecological, and social
literatures on the consequences of new infrastructure,
which reveals divergent appraisals. We, therefore, propose
an integrative analytical framework organized around the
key concept of social–ecological resilience.
Regional integration in Latin America
Integration initiatives include trade agreements and infra-
structure projects. We focus on Latin America, which has
both types of initiatives. Following a period of relative
insularity under import-substitution industrialization after
WWII, the debt crisis of the 1980s prompted many coun-
tries to renegotiate multilateral bank loans (Randall 1997).
In the spirit of ‘open regionalism,’ Latin American
countries increasingly looked to each other as trading
partners and viewed integration as a means of reinvigo-
rating economic growth (Bulmer-Thomas 2001; Genna and
Hiroi 2004). In this context, several trade agreements were
formalized in Latin America. These include the North
American Free Trade Agreement (NAFTA) between the
United States, Mexico, and Canada; MERCOSUR between
Brazil and the ‘Southern Cone’ countries; and the Carib-
bean Community and Common Market (CARICOM); in
addition, there already existed the Andean Pact/Andean
Community (CAN). NAFTA is significant because it
served as a model for a Free Trade Area of the Americas
(FTAA). However, consensus on FTAA was far from clear
(Carranza 2002). By the time of FTAA negotiations in
2005, other trade blocs had emerged to counter the North
American bloc, such as the Union of South American
Nations (UNASUR) and the Bolivarian Alternative for the
Americas (ALBA). In this context, FTAA negotiators had
to set aside central issues such as agriculture and intellec-
tual property, effectively watering down the final agree-
ment (Kellogg 2007).
At the same time, other efforts at economic integration
in Latin America were advancing, notably those involving
transportation infrastructure. Cross-border infrastructure
projects have advanced open regionalism in Latin America
and East Asia (Carranza 2002; Munakata 2006). In Latin
America, no such endeavor is more significant than the
Initiative for the Integration of Regional Infrastructure in
South America (IIRSA). IIRSA was constituted by a
meeting of presidents from 12 South American countries in
2000 (IIRSA 2008). A key motivation behind IIRSA was
Latin America’s infrastructure deficit relative to its income
level as compared with competing economies in East Asia
n and Serven 2004). IIRSA was explicitly con-
ceptualized as a spatial strategy organized around ‘axes of
integration’ targeted for infrastructure investments. Such
axes constitute strategic growth corridors for regional
commerce. Subsequent IIRSA meetings included multi-
lateral development banks and led to agreements to fund
350 projects at a total projected cost of $38 billion (CEPEI
2002; IIRSA 2008; Killeen 2007; Moreira Mesquita 2007;
Van Dijk and den Haak 2007).
IIRSA has given rise to questions about environmental
impacts as well as the extent of the economic benefits. The
record of environmental impact assessments in IIRSA
projects highlights oversights and limitations (Dourojeanni
et al. 2010; Killeen 2007). There are also doubts that
regional integration under IIRSA will generate economic
growth (Moreira Mesquita 2007; Van Dijk and den Haak
2007). Such doubts are complemented by social concerns
such as the distribution of benefits among interest groups,
localities, and countries (Moreira Mesquita 2007).
Infrastructure impacts: economic, ecological and social
The concerns about IIRSA projects also take center stage in
distinct literatures on various types of road impacts. Not
only has there been extensive research on the economic
ramifications of roads and other infrastructure, a body of
work on the ecological impacts has also quickly grown. In
addition, there is a more scattered set of statements on
social consequences of roads.
Economic impacts
Recent economics literature on infrastructure was stimu-
lated by earlier studies that showed large impacts on eco-
nomic growth (Aschauer 1989; Munnell 1992). Public
sector investments in infrastructure apparently generated
large downstream increments in private investments that
expanded employment and per capita GDP. Such work
came in a period of economic recession and therefore
stimulated discussion and debate concerning infrastructure-
led growth (Boarnet 1999; Gramlich 1994; World Bank
A macroeconomic literature has consequently emerged
on infrastructure and its implications for economic growth.
The findings have generally been positive and significant,
especially in developing countries (Bourguignon and Ple-
skovic 2008; Straub 2008), including Latin America
n and Serven 2004). Positive effects of infra-
structure on growth also apply to the rural sector (Craig
et al. 1997). This literature features national-level data,
often for multiple time periods. Consequently, more
Regional integration and local change 37
emphasis has gone to dynamics over time rather than to
spatial considerations, though there is incipient recognition
of geographic variation in the economic impacts of roads
(Straub 2008).
The macroeconomic literature is complemented by a
microeconomic literature on roads and poverty in devel-
oping countries. There the emphasis is on making road
investments to maximize livelihood benefits and rural
poverty reduction through ‘geographic targeting’ of
infrastructure to facilitate ties of remote areas to local and
regional markets (Bigman and Fofack 2000; Howe and
Richards 1984; Subbarao 1997; van de Walle 2002).
Empirical studies show significant poverty reduction near
roads, whether via analysis of subnational administrative
units (Binswanger et al. 1993; Demurger 2001; Fan et al.
2000, 2004; Minot et al. 2003), local communities (Gun-
asekera et al. 2008; Pender et al. 2004), or individual
households (Gibson and Rozelle 2003; Jacoby 2000;
Windle and Cramb 1997).
Ecological impacts
If the economic literature has offered positive evaluations
of infrastructure, ecological research on roads has compiled
a litany of negative consequences. There are multiple
reviews of the ecological impacts of roads (Forman and
Alexander 1998; Trombulak and Frissell 2000). This has
fomented the emergent subspecialty of ‘road ecology’
(Coffin 2007; Forman et al. 2003).
The road ecology literature pays particular attention to
spatial aspects of road impacts on biophysical environ-
ments. Reviews often take as their point of departure the
observation that roads fragment habitats and modify stream
networks (Forman et al. 2003). From there, ecologists
catalogue a host of consequences ranging from habitat
isolation and degradation (Eigenbrod et al. 2008; Haw-
baker and Radeloff 2004), to road avoidance by wildlife
(Jaeger et al. 2005), to contagion effects such as exotic
species invasions (Christen and Matlack 2006; Hansen and
Clevenger 2005), among others.
A parallel literature on tropical forest fragmentation
has also emerged and has a similar list of fragment edge
effects on a variety of ecological outcomes (Laurance and
Bierregaard 1997; Laurance et al. 2002). Both the road
ecology and habitat fragmentation literatures focus on the
spatial configurations of habitat mosaics and the distance
up to which specific ecological effects are felt, whether
from roads (Forman et al. 2003) or habitat edges (Lau-
rance et al. 2002). One implication is that the extent of
spatial effects of a road varies considerably depending on
the type of outcome under study; another is that areas
closer to roads and habitat edges tend to be more heavily
Social impacts
A more scattered literature exists on the social impacts of
roads. One concern has been the distribution of benefits.
Social research has problematized differential benefits to
higher social status groups (Mahapa and Mashiri 2001;
Shriar 2009; Trankell 1999) and districts closer to new
infrastructure (Cross 2001; Rudel and Richards 1990),
often as the result of a lack of broad participation in
infrastructure planning (Devres Inc 1981; Robinson 2001).
Lack of participation also raises issues of exclusion of
indigenous and other minority groups, often whereby
newcomers make legalistic claims to natural resources in
order to discredit traditional claims. Such political pressure
disrupts traditional livelihoods and indigenous cultures,
something well known in frontier areas such as the Ama-
zon (Davis 1977; Leonel 1992). In addition, roads in the
Amazon are notorious for stimulating conflicts over newly
accessible natural resources (Hall 1989; Schmink and
Wood 1992).
From this review of infrastructure impacts, we draw two
insights for our present purposes. First, roads bring a wide
array of impacts, and they are decidedly different if one
highlights economic, ecological, or social outcomes. Con-
sequently, drawing on one literature yields limited and
skewed conclusions about road impacts. Moreover, it
results in oversights concerning synergies among impacts.
For example, in resource-dependent regions, disruption of
established resource management practices may result in
ecological degradation which may in turn undermine
resource-based livelihoods.
Second, there has been variable attention devoted to the
spatial ramifications of new infrastructure. In the macro-
economic literature, there is only incipient recognition of
spatial differences in economic changes due to infrastruc-
ture. In the microeconomic literature, there is recognition
of distance costs associated with ease of transportation,
though that lacks an appreciation of economic integration
in terms of the actual extent of countryside-market inter-
actions. While the arrangement of habitats in landscape
mosaics is highlighted in the road ecology literature, there
remains a focus on distance gradients that stops short of
more sophisticated treatments of habitat connectivity found
in landscape ecology (Goodwin and Fahrig 2002; Urban
and Keitt 2002). In the social impacts literature, attention
to space and location is usually subordinated to a focus on
social inequalities.
An integrative approach to integration: connectivity
and resilience
In light of these observations, we offer an integrative
approach to the analysis of infrastructure impacts on rural/
38 S. G. Perz et al.
remote regions of developing countries, which highlights
spatial variation. To broaden the analysis beyond strictly
economic, ecological, or social facets of infrastructure
impacts, we invoke the integrative concept of social–eco-
logical resilience. And to manage spatial questions con-
cerning the importance of location in the process of
economic integration, we employ the long-established
concept of connectivity from transport geography.
The concept of connectivity is the key in location theory,
evident in classic statements by von Thu
nen and Chri-
staller, with developments around equilibrium models of
the regional science school (Isard 1956) and subsequent
quantitative elaboration (Haggett et al. 1977). Transport
geography includes case studies of the extension of infra-
structure to integrate frontier regions into national econo-
mies via road and rail links, as in Siberia (North 1979), the
US Great Plains (Cronon 1991), and the Brazilian interior
(Katzman 1977), including in the Amazon (Walker et al.
2011). Transport geography highlights the importance of
infrastructure in articulating towns and rural communities
for the distribution of resources and goods via transporta-
tion networks.
We follow elements of this lineage by appropriating the
concept of connectivity as regards specific locations
(nodes) in transportation networks. Connectivity may be
evaluated at the level of a transportation system using the
ratio of the number of edges (connections) to nodes.
However, we are interested in differentiating among par-
ticular localities in terms of their market connectivity, so
we approach this concept in terms of the strength of con-
nections from a given node to others within a transportation
We view connectivity among locations in a multidi-
mensional fashion. Connectivity encompasses not only
ease of access but also the extent and type of actual
interactions between locations. One dimension of connec-
tivity between locations, therefore, involves accessibility,
defined here in terms of road paving and market distance,
with the implied transportation costs being greater on
unpaved roads and over greater distances. We refer to this
as ‘access connectivity’ and relate paving and distance to
ease of access.
A second dimension of connectivity concerns the actual
extent of interactions between nodes. We refer to this as
‘interaction connectivity’ and define it in terms of rela-
tions between rural communities and urban markets.
Interaction connectivity can be evaluated in terms of social
ties as well as of economic exchanges. The extent of such
interactions provides indexes of connectivity that indicate
the degree of integration between rural communities and
market towns which serves as contact points linked to
product distribution chains.
Analytically, we approach connectivity on two spatial
scales. On a regional scale, highways provide key con-
nections among cities and regions. Paving of highways is a
key infrastructural upgrade that reduces transportation
costs and time, especially in areas where sealed road sur-
faces afford year-round instead of only seasonal transport.
For regional access connectivity, we, therefore, differenti-
ate between areas with and without paved highways. We
also anticipate greater interaction connectivity in regions
with paved roads.
In addition, on a local scale, we view access connec-
tivity in terms of proximity to regional markets (i.e., state
capitals) and interaction connectivity as the extent of ties
between rural communities and local towns. For local
access connectivity, we can differentiate between market
spaces centered on different towns, themselves situated at
varying distances from regional cities that serve as market
hubs and distribution points in larger commodity chains.
We can thus observe distinct local market areas along a
distance gradient from a regional capital, which allows
comparisons of access connectivity between localities. In
addition, local connectivity refers to the degree of inter-
action between rural communities and local towns. This
may involve social ties such as via rural communities with
member families that reside in towns or market exchanges
observable in terms of the extent to which rural products
are commercialized in towns.
Resilience is a concept that originated in the systems
ecology literature (Gunderson 2000; Holling 1973; Walker
and Salt 2006). It has variously been defined as the
capacity of complex systems to exhibit stability or persis-
tence in the face of change, to adapt creatively to externally
induced shocks or to exhibit self-organization.
Despite its origins in ecology, geographers and other
social scientists have sought to define ‘social resilience’
(Adger 2000; Adger et al. 2002). Whereas ecological
resilience is generally defined at the level of an ecosystem,
social resilience is usually formulated in terms of the
capacity of social groups or communities to respond to
externally induced shocks. Such shocks may include dis-
ease outbreaks, storms events, swings in commodity prices,
and infrastructure projects.
Discussions of social resilience led to formulation of
‘social–ecological resilience’’ (Adger et al. 2005; Peterson
2000; Turner et al. 2003) as a conceptual foundation for
linking human action, social institutions, and market
dynamics (‘‘the social’’) to strategies for natural resource
management via livelihoods (‘‘the social–ecological’’) in
Regional integration and local change 39
order to better understand complex feedbacks from habitat
mosaics, species assemblages, and other components of
ecosystems (‘‘the ecological’’). Conceptual discussions
stimulated scholars to develop approaches for empirical
measurement and evaluation (Carpenter et al. 2001).
Among such efforts have been proposals to dimension-
alize resilience (Folke et al. 2003) by considering multiple
aspects of social–ecological resilience. This can be viewed
as a consequence of the considerable conceptual terrain
encompassed by social–ecological resilience. It is also a
result of the fact that social–ecological systems may not
only exhibit complex (non-linear) dynamics but also
encompass complicated structures involving many social
and ecological components.
Cumming et al. (2005) use this dimensional approach to
social–ecological resilience to elaborate a framework for
evaluating resilience. They propose to decompose com-
plicated social–ecological systems to feature system com-
ponents and their relationships. The focus on system
components makes the analysis of resilience more tracta-
ble, because such components are relatively easily identi-
fied as opposed to whole systems which often have ill-
defined boundaries. In this paper, we apply the Cumming
et al. (2005) framework by focusing on rural communities
as ‘social components’ of a social–ecological system.
Further, Cumming et al. (2005) highlight the ‘‘interface’
between the social and ecological by focusing on human
management of natural resources with specific ecological
ramifications. Key to the analysis is a focus on the ‘‘social–
ecological relationships’ that make up the social–ecolog-
ical interface, as manifest in resource management prac-
tices of resource-dependent rural communities. The
dimensional approach to an analysis of social–ecological
resilience in communities is especially useful since com-
munities make multiple uses of different resources
(extracting products from forests, planting crops, running
livestock, fishing rivers, hunting game, and so on) and may
therefore undergo multiple changes due to new infra-
structure. An analysis of social–ecological resilience thus
becomes a suite of manifold tests focused on resource-
dependent communities and many of their social–ecologi-
cal relationships as via resource management.
The focus on system components provides an opening to
evaluate resilience spatially. By accounting for the location
of communities, we can assess spatial variation in the
social–ecological interface. This in turn permits an analysis
of community connectivity to market towns and how that
relates to community social–ecological resilience.
We, therefore, focus on rural communities in a resource-
dependent region and evaluate their resilience using indi-
cators of distinct dimensions of social–ecological resil-
ience. We draw on prior literature (Adger 2006; Cumming
et al. 2005; Folke et al. 2003) to emphasize three
dimensions of social–ecological resilience in rural com-
munities: collective memory, livelihood diversity, and
adaptive capacity.
In the resilience literature, ‘memory’ refers to the ability
of a system to retain key properties and is viewed as crucial
for system recovery following a shock (Berke and Turner
2006; Folke et al. 2003). Ecologically speaking, this means,
e.g., persistence of seed banks for native species; socially, it
implies, e.g., collective memory to retain longstanding
resource management practices. For rural communities, we
employ demographic measures of population change via
migration to evaluate the retention of collective memory
(Adger 2000; Perz et al. 2010). Regions experiencing
infrastructure change often incur rapid population growth
and turnover via migration, as in frontier areas of the
Amazon, with the result that traditional resource manage-
ment practices are replaced with destructive alternatives
(Schmink and Wood 1992; Millikan 1992).
Another characteristic of resilience is diversity (Adger
2000; Gunderson 2000). In the ecological literature, bio-
diversity and the redundancy of species with respect to
ecological niches provide greater latitude for ecosystem
adaptation to shocks (Levin 1999). In the social science
literature, livelihood diversity receives considerable
emphasis as regards resilience and vulnerability (Ellis
2000; Goldman 1995; Perz 2005). Livelihood diversity
begets resilience by affording greater latitude for house-
holds and communities to sustain and reproduce them-
selves in the face of uncertainty or rapid change, such that
the loss of one activity can be buffered by retention or
expansion of others. Therefore, not only the level but also
change in diversity is important for an evaluation of
community resilience. For the same reasons, in resource-
dependent communities, the diversity of natural resources
is similarly important; a decline in access to resources can
signify a reduction in community resilience.
A third hallmark of resilience is the capacity for adap-
tation via self-organization (Adger 2000; Berke and Turner
2006; Folke 2006; Gunderson 2000; Holling 1973). We
formulate adaptive capacity in terms of community
responses to crises, which may affect community viability
in the future. Here, we emphasize fires and fire response. In
2005, the Amazon experienced a pronounced drought event
in which fires set to remove vegetation for agriculture
escaped control and wrought widespread economic as well
as ecological damage (Marengo et al. 2008). Fires alter
forest ecology and make future fires more likely and more
dangerous (Nepstad et al. 2008). Community fire response
is thus crucial in managing fire risk, to forests as well as to
community assets; effective fire response mitigates future
fire risk.
Our analysis features connectivity and resilience among
rural communities in a region experiencing integration via
40 S. G. Perz et al.
a large-scale infrastructure project. Per the foregoing dis-
cussion, we consider two dimensions of connectivity
(access and interaction), as well as three dimensions of
resilience (collective memory, livelihood diversity, and
adaptive capacity). This affords a multidimensional
appraisal of the potentially multifaceted relationships
between connectivity and resilience.
Methods, data, and measures
Study region
We focus on the southwestern Amazon, a study case
appropriate for purposes here because it is a relatively
remote region with a resource-dependent economy and
because it is now being integrated into national and global
economies via large-scale infrastructure. The southwestern
Amazon is well known as a tropical conservation ‘hot-
spot’ due to its extraordinarily high levels of biodiversity
and endemism (Killeen and Solo
rzano 2008; Myers et al.
In the 1970s, national governments in several countries
sharing the Amazon embarked on regional development
initiatives featuring roads, colonization projects, and fiscal
incentives for capital investment (Hemming 1985; Schm-
ink and Wood 1984). Since 2000, the southwestern Ama-
zon became the focus of a new generation of infrastructure
projects, this time under the auspices of IIRSA (2008). One
of the IIRSA’s initial priorities is the Inter-Oceanic High-
way, which links Atlantic ports in southern Brazil to Pacific
ports in Peru (CEPEI 2002). The key to the Inter-Oceanic
Highway is the Peru-Brazil axis of integration, where road
paving is still needed to provide year-round transportation
infrastructure for expanded cross-border commerce (CE-
PEI 2002). In particular, attention has focused on the heart
of the southwestern Amazon, the tri-national ‘MAP’
frontier where Madre de Dios, Peru meets Acre, Brazil, and
Pando, Bolivia (Brown et al. 2002; Iniciativa MAP 2008;
Mendoza et al. 2007; van Oosten 2004). It is in the MAP
frontier that paving of the Inter-Oceanic Highway is
Field methods
We draw on a tri-national survey of rural communities
along the Inter-Oceanic Highway and other key roads in
the MAP frontier. Using secondary data sources (censuses,
zoning plans, cadastral maps, etc.), we identified commu-
nities in a GIS as distinct land tenure units and/or popu-
lation centers along major roads in the MAP frontier. We
focused on the Inter-Oceanic Highway in Madre de Dios,
Peru, and Acre, Brazil, as well as primary roads in Pando,
Bolivia, in and near the area of influence of the Inter-
Oceanic Highway. We then sampled communities sys-
tematically to ensure variation in terms of distance from
regional capitals and (in Pando, Bolivia) distance from the
Inter-Oceanic Highway. The result is a broad geographic
distribution of communities along primary roads in the
MAP frontier, shown in Fig. 1. In Madre de Dios, Peru and
Pando, Bolivia, rural communities have nucleated popula-
tion centers on or near primary roads; in Acre, Brazil this is
less often the case. Hence, most communities sampled in
Madre de Dios and Pando have centers along the roadsides,
whereas in Acre, there is greater variation in community
distance from highways.
For our purposes, a ‘rural community’ refers to an
aggregation of families residing in a given location with
agricultural and forested land. Specific definitions of rural
communities vary across the MAP frontier because land
tenure regimes vary. In Pando, Bolivia, communities refer
to communal land tenure units (predios) consisting of
nucleated settlements where member families use the sur-
rounding lands. In Madre de Dios, Peru, communities are
constituted by local associations consisting of member
families with private individual land parcels that are gen-
erally contiguous. In Acre, Brazil, communities vary in
definition because of diverse land tenure arrangements,
ranging from settlement projects to agroextractive projects,
extractive reserves, and others.
In 2007, faculty and students from the University of
Florida (UF) collaborated with counterparts from the
National Amazonian University of Madre de Dios for
fieldwork in Madre de Dios, Peru and with counterparts
from the Amazonian University of Pando for fieldwork in
Pando, Bolivia. In 2008, UF and counterpart faculty and
students at the Federal University of Acre collaborated on
fieldwork in Acre, Brazil. In the selected communities, we
interviewed one or more ‘leaders’ (i.e., current or past
community representatives, long-time residents, etc.). In
Madre de Dios, we visited 41 associations and conducted
88 interviews; in Acre, we visited 25 distinct land tenure
areas and conducted 93 interviews; in Pando, we visited 37
predios where we conducted 111 interviews. Overall, we
obtained 292 interviews in 103 communities.
We employed a structured questionnaire with open-
ended questions to obtain comparable information on a
standardized set of topics and yet allow informants to
elaborate on their responses. Following our theoretical
In Acre, because of inconsistencies in definitions of communities
among informants, we employed state-defined land tenure units. This
reduced the number of communities ostensibly sampled in Acre, but
our weighted analysis (see note 2) offsets that by taking into account
community size, so Acre is not underrepresented.
Regional integration and local change 41
framework, we focused on items relevant to the concepts of
connectivity and social–ecological resilience but also sig-
nificant in the Amazonian context. The questionnaire
included items on the interviewee (sex, leadership position
in the community, and place of birth), general information
about the community (name, location, and land tenure
type), and sections on specific topics, including community
population (number of resident families, non-resident
member families, in-migrant families in the past 5 years,
and out-migrant families in the same period), seven com-
munity livelihood activities (timber, non-timber forest
products (NTFPs), crops, cattle, wage labor, and others),
change in the importance of those livelihood activities in
the past 5 years, change in availability (scarcity) of four
forest resources in the past 5 years (timber, NTFPs, game,
and others), marketing of key products (timber, NTFPs,
crops, cattle, other; whether sold to buyers who came to the
community, taken to towns for sale, sold through cooper-
atives, or other means), and damage from fires (as well as
community response). We also collected GPS points in
order to calculate community distances to local towns and
regional capitals.
Connectivity measures
From the questionnaire, we derived indicators of commu-
nity connectivity and social–ecological resilience. Because
we view connectivity from the standpoint of a rural com-
munity and because we recognize multiple dimensions of
connectivity, we employ four measures of this concept.
The first two are measures of access connectivity, of which
one is a regional measure and the other is a local measure;
the other two are measures of interaction connectivity, both
at the local level. At the regional scale, access connectivity
refers to road paving, and we differentiate between the
three sides of the MAP frontier, where highway paving is
complete in Acre, Brazil, underway in Madre de Dios,
Peru, and planned in Pando, Bolivia. At the local level, we
measure access connectivity by employing a distance-to-
market measure along with supplemental information in
order to group communities into ‘‘subregions’’ with distinct
distances to their respective regional capital and (where
applicable) other key distinctions (such as the availability
of a key forest resource). This resulted in the identification
of four distinct ‘subregions’ within each of the three sides
of the MAP frontier, which we discuss later.
In addition, we measure interaction connectivity with
two local-level indicators. The first is the number of
member families who are absent because they live in a
town, as a percentage of all member families. This provides
an index of social connectivity between rural communities
and market centers. The other interaction connectivity
measure concerns economic connectivity via the extent of
product marketing. We calculated this as the sum of the
number of ways key products are sold. The more products
commercialized and the more ways in which each product
is sold, the greater the economic connectivity and the
higher the value of the index.
Fig. 1 Tri-national MAP
frontier of the southwestern
Amazon, with location of
interviews with community
leaders along the Inter-Oceanic
Highway (road in Acre and
Madre de Dios) and other
primary roads (Pando)
42 S. G. Perz et al.
Resilience indicators
Our conceptual discussion highlighted three dimensions of
social–ecological resilience, namely collective memory,
livelihood diversity, and adaptive capacity. For collective
memory, we employ two migration measures. The first is
the net change in the number of member families via
migration to a community in the 5 years before the survey.
We calculated net change in terms of in-migrant minus out-
migrant families as a percentage of all member families. A
larger net change in member families, whether in gains or
especially in losses, is a proxy for loss of collective
memory. In addition, we calculated migratory turnover via
the number of in-migrant families plus out-migrant fami-
lies, as a percentage of all member families. Greater family
turnover also undermines community continuity and
implies a decline in collective memory.
For diversity, measures include livelihood diversity at
the time of interviews, change in livelihood diversity and
change in availability of forest resources. For livelihood
diversity, we drew on ranked data on the relative impor-
tance of up to seven livelihood activities (timber, NTFPs,
crops, cattle, wage labor, and others), where the most
important activity received a score of seven, the second
most important a score of six, and so on, and summed the
ranks. Communities reporting more livelihood activities
have more diverse livelihoods and thus higher livelihood
diversity scores. We interpret greater community liveli-
hood diversity as an indicator of greater resilience to
externally induced shocks. For dynamics of livelihood
diversity, we obtained information about the change in the
importance of each livelihood activity in the past 5 years.
If an activity became more important, it received a positive
score; if the same, it received a score of zero; and if less
important, it received a negative score. We then summed
the change values across all reported livelihood activities to
obtain a livelihood change index. We did the same thing
for the availability of four forest resources (timber, NTFPs,
game, others). For the change indexes, positive sums
indicate a rise in livelihood diversity or forest resource
availability, and thus greater community resilience,
whereas negative sums indicating declining diversity and
thus reduced community resilience.
For adaptive capacity, we employ two measures: fire
damage and community response to fires. We know about
whether a community experienced fire damage in 2005 or
since (1 = Yes, 0 = No), and we have leader appraisals of
community response (where 1 = rapid response, fire con-
trolled, 2 = slow response, limited control of fire, and
3 = no response or inability to control fire), which we
recoded (0 = rapid or slow response, 1 = Ineffective
response). Experience of fire damage and an ineffective
response suggest a threat to future community resilience.
Indicators of different dimensions of connectivity and
resilience are likely to exhibit multifaceted relationships.
Although location theory suggests similar effects of access
and interaction connectivity, because the conceptual con-
tent of resilience is broad and given that rural communities
have many social–ecological interactions, the dimensions
of social–ecological resilience are not as tightly interre-
lated. We, therefore, anticipate that in some respects,
greater connectivity will be associated with greater resil-
ience, but in others, the relationship may be inverse, non-
linear, or zero. We detail these expectations in Table 1.
While greater connectivity may undermine social–ecolog-
ical resilience via greater migration and thus less collective
memory, greater connectivity may also encourage greater
resilience via higher livelihood diversity, though at the
same time access to forest resources may decline. And
while communities with greater connectivity may experi-
ence more uncontrolled fires, greater connectivity may also
improve fire response by fostering communication and
mobilization for more effective fire control.
Analysis and findings
Our analysis distinguishes between localities across the
MAP frontier to allow comparisons of communities inter-
pretable in terms of levels of connectivity. This provides a
geographic basis for evaluating arguments about the rela-
tionships of connectivity and social–ecological resilience.
The use of multiple measures of connectivity allows
comparisons at both the regional and local scales and
between different dimensions of connectivity. Further, the
employment of multiple indicators of social–ecological
resilience affords a broad appraisal of communities and
natural resources in the presence of new infrastructure.
Finally, a comparative analysis across degrees of connec-
tivity provides a means of evaluating relationships with
social–ecological resilience, whether strong or weak,
positive or negative, and linear or non-linear.
Our unit of analysis is the community (n = 103), but our
data were gathered from community leaders (n = 292). We,
therefore, aggregated informant responses in order to obtain
community-level information. Where informants disagreed,
for continuous variables we calculated means, for binomial
(0/1) variables we calculated proportions, and for other
categorical variables we went with the majority’s opinion.
Communities also vary considerably in size. We, therefore,
weighted communities by the number of resident member
families. Weighting had the overall effect of counting the
communities in Madre de Dios and Acre more heavily than
those in Pando (see Table 2). While this yields different
results than unweighted analysis, weighting of communities
yields findings more representative of rural families in the
Regional integration and local change 43
MAP frontier. In weighting, we did not inflate the community
sample size, so the power of inferential tests is unaffected.
Descriptive analysis
We begin with a description of the communities in the
sample and then move on to a region-level comparative
analysis, followed by multiple local-level comparative
analyses. Table 2 presents selected characteristics of the
communities for the three sides of the MAP frontier, along
with inferential tests for significance of differences. We
include indicators of highway paving, distance to regional
capital, community size, characteristics of the informants,
and livelihood activities. Table 2 shows that all commu-
nities in Acre had access to paved roads, somewhat less
than half did in Madre de Dios as of mid-2007, and rela-
tively few in Pando did as of the same year.
The highly
significant difference in highway paving constitutes the
basis for our regional comparisons between Madre de Dios,
Acre, and Pando in terms of access connectivity. Distance
to regional capital also varies among the three sides, being
greater in Acre than elsewhere. This is a result of location
geography; in Acre, the study area stretches along the
Inter-Oceanic Highway from the state capital of Rio
Branco at one end to Assis Brasil, some 340-km distant,
whereas in Madre de Dios, the capital of Puerto Maldonado
is midway along the highway corridor (and the most distant
community, In
apari, is 220 km away), and in Pando, while
the capital of Cobija is at one end of the road there, the
farthest community, Sena, is roughly 250 km away.
Community size (number of families) is greater in
Madre de Dios and Acre than in Pando, which is why the
first two sides are weighted more heavily. However, there
is the considerable variation in community size within each
side, which is why there is only a weakly significant dif-
ference between the three sides.
Informant characteristics also vary somewhat among the
three sides of the MAP frontier. We interviewed propor-
tionally more current representatives in Madre de Dios, and
additional analysis shows more interviews with former
representatives in Acre and Pando. In all three sides,
women comprised a minority of interviewees. There were
significant differences in the place of birth of informants;
proportionally fewer of our Peruvian informants were born
in Madre de Dios than Brazilians in Acre or Bolivians in
Pando. Finally, respondents varied in their durations of
residence in their communities, with longer durations in
Pando, suggestive of less population change due to less
road paving there than in Madre de Dios or Acre.
In terms of livelihood activities, Table 2 presents pro-
portions of communities engaged (to varying degrees) with
timber, NTFPs, crops, cattle, wage work, and other activities.
Most communities engaged in the first four, though some
significant differences appeared. Logging is less prevalent in
Acre where timber extraction is more heavily monitored, and
NTFPs were most important in Pando, where castan
a col-
lection is especially significant. Crop cultivation (mostly for
subsistence) is universal, and though differences in the rel-
ative prevalence of cattle were not significant, cattle ranked
highest in importance in Acre. Wage work was most wide-
spread in Acre, but other activities were more prevalent in
Madre de Dios and Pando, a reflection of greater livelihood
diversity than in Acre.
Table 1 Expected relationships of connectivity and social–ecological resilience between rural communities in the tri-national MAP frontier,
Southwestern Amazon
Relationship Explanation
Collective memory
Net migration ? Connectivity facilitates population mobility
Turnover ? Connectivity facilitates population mobility
Livelihood diversity ? Connectivity may render new activities feasible (wage labor) but render
others uncompetitive (product sales)
Change in livelihood diversity ? Connectivity may render new activities feasible (wage labor) but render
others uncompetitive (product sales)
Change in forest resource access Connectivity facilitates forest resource exploitation (timber, game, etc.)
Adaptive capacity
Fire damage ? Connectivity facilitates forest fragmentation, agricultural expansion,
use of fire
Fire response ? Connectivity facilitates emergency response
Table 1 presents unweighted analysis for highway paving and
community size. We did not weight the paving variable as the key
there is how many communities have access to paved highways, and
not weighting gives a more accurate appraisal of the geographic
extent of paving among communities. We did not weight community
size because that is the variable used for weighting, and would have
yielded distorted results.
44 S. G. Perz et al.
Regional connectivity and social–ecological resilience
We now turn to a region-level comparative analysis of
Madre de Dios, Acre, and Pando in terms of connectivity
and resilience. Table 3 presents our indicators of social–
ecological resilience among rural communities in the MAP
frontier. We base our region-level comparisons on the
assumption that Acre has greater access connectivity due to
more highway paving.
In terms of the demographic/memory variables, we find
significant regional differences. Net migration is positive in
communities in Madre de Dios and Pando and slightly
Table 2 Selected characteristics of rural communities sampled by region in the tri-national MAP frontier, Southwestern Amazon
Madre de Dios,
Acre, Brazil Pando, Bolivia MAP frontier F test
Highway paving (0 = Unpaved, 1 = Paved) 0.39 1.00 0.03 0.41 65.92**
,a, b
Km to regional capital 102.60 155.20 90.24 109.70 4.69*
Number of families 118.52 146.51 52.83 109.24 3.07
Community leader (0 = No, 1 = Yes) 0.52 0.33 0.31 0.43 4.98**
Sex of interviewee (0 = Male, 1 = Female) 0.21 0.31 0.29 0.27 1.76
Born in same state (0 = No, 1 = Yes) 0.30 0.65 0.76 0.50 22.70**
Years of residence 23.11 17.75 30.10 22.75 7.48**
Timber extraction (0 = No, 1 = Yes) 0.83 0.73 0.97 0.84 2.90
NTFP extraction (0 = No, 1 = Yes) 0.64 0.82 0.99 0.77 5.93**
Crop cultivation (0 = No, 1 = Yes) 1.00 1.00 1.00 1.00 0.00
Cattle ranching (0 = No, 1 = Yes) 0.95 0.96 0.88 0.94 0.67
Wage work (0 = No, 1 = Yes) 0.57 0.99 0.82 0.76 11.90**
Other activity 1 (0 = No, 1 = Yes) 0.92 0.63 0.78 0.81 7.12**
Other activity 2 (0 = No, 1 = Yes) 0.09 0.05 0.22 0.10 2.44
Unweighted n 41 25 37 103
Weighted n 51.4 32.2 19.4 103
a ?
P \ 0.15; * P \ 0.05; ** P \ 0.01
Unweighted analysis
Table 3 Comparative analysis of social–ecological resilience indicators by region in the tri-national MAP frontier, Southwestern Amazon
Madre de Dios,
Acre, Brazil Pando, Bolivia MAP frontier F test
Collective memory
Net migration (Pctg families in–out) ?16.08 -0.71 ?17.13 ?11.41 14.82**
Migration turnover (Pctg families in?out) 25.04 20.40 32.20 24.71 2.45
Livelihood diversity (summed ranks)
22.50 18.36 23.62 21.54 23.01**
Change in livelihood diversity (summed changes)
?1.86 ?2.74 ?0.27 ?1.84 4.23*
Change in forest resource access (summed changes)
-3.08 -3.13 -1.03 -2.70 6.76**
Adaptive capacity
Fire damage (0 = No, 1 = Yes) 0.60 0.97 0.86 0.72 9.54**
Fire response (0 = Effective, 1 = Ineffective) 0.35 0.41 0.94 0.52 19.80**
a ?
P \ 0.15; * P \ 0.05; ** P \ 0.01
We ranked seven activities, with the most important ranked highest (7), the next most important ranked next (6), and so forth. We then summed
the ranks of the activities present
For change in livelihood diversity, we obtained change scores for each of seven activities. If an activity became more important, it received a
positive value; if unchanged, a zero value; if less important, a negative value. We then summed the values
For change in forest resource access, we obtained change scores for each of four resources. If a resource became more available, it received a
positive value; if unchanged, a zero value; if less available, a negative value. We then summed the values
The test for fire response is a Chi-square (df = 4)
Regional integration and local change 45
negative in Acre. Turnover is also high, especially in
Madre de Dios and Pando. However, differences are not as
significant for turnover as for net migration, which suggests
more widespread turnover across the three sides.
In terms of diversity, Table 3 shows significant differ-
ences across the MAP frontier. Communities in Pando and
Madre de Dios had more diverse livelihoods than those in
Acre. Further analysis of the data indicates a strong focus
on cattle, crops, and wage work in Acre, with relatively
more timber, NTFPs and other activities in Madre de Dios
and Pando. Livelihood diversity increased in the 5 years
prior to surveys. What is more, the increase is greatest in
Acre, where diversity was lowest at the time of the surveys,
and lowest in Pando, where diversity was highest. That
said, the findings for forest resources show decreased
availability in the 5 years prior to the surveys. Madre de
Dios and Acre show larger decreases than Pando. Further
analysis shows large decreases in the availability of wild
game (especially in Madre de Dios and Acre) as well as
timber (especially in Madre de Dios), along with a more
moderate decline in NTFPs (in Acre, alongside a small
increase in Pando).
Table 3 concludes with adaptive capacity. Fires were
widespread across the MAP region, with most communities
on all three sides reporting damage. There are significant
differences, however, with damage nearly universal in
Acre, followed by Pando and then by Madre de Dios.
Moreover, informants characterized the efficacy of com-
munity fire response differently across the MAP region.
While overall informants split roughly evenly on whether
communities with fires responded ineffectively, that pro-
portion was lower in Madre de Dios and Acre than in Pando.
Local connectivity and resilience: subregional
The remainder of the analysis involves comparisons using
local-level indicators of connectivity to evaluate social–
ecological resilience. The first local connectivity indicator
is another access measure that differentiates between sub-
regions within each side of the MAP frontier, primarily in
terms of distance to their respective regional capitals.
Within each side, we distinguish four subregions, shown in
Fig. 2. In Madre de Dios, we differentiate between (1)
communities close to (\30 km) the departmental capital of
Puerto Maldonado, which are more integrated into the
regional urban economy; (2) ‘Tambopata west,’ an older
zone of roadside settlement between Puerto Maldonado
and Cusco; (3) the ‘castan
a zone’ north of Puerto
Maldonado, where there are more castan
a trees but land is
more remote from the core of the Peruvian economy; and
(4) Tahuamanu, yet further north from Puerto Maldonado,
but where the first portion of the Inter-Oceanic Highway
was paved in Madre de Dios. Madre de Dios thus com-
prises a differentiated set of spaces defined not only by
market distance and highway paving but also by biophys-
ical differences and varying degrees of incorporation into
the Peruvian economy. We view subregion 1 as the most
connected to the regional economy, followed by 2, then 3,
and finally 4.
Fig. 2 Subregions for local-
level comparisons in the tri-
national MAP frontier,
southwestern Amazon
46 S. G. Perz et al.
In Acre, the four subregions are (1) communities
around the state capital of Rio Branco and the neigh-
boring municipality of Senador Guiomard (roughly 20 km
from the capital), where the Inter-Oceanic Highway has
been paved longest, with intensive land use and a large
urban population (250,000?); (2) communities around
Capixaba, a young town experiencing rapid growth and
located on a segment of the Inter-Oceanic Highway only
a few kilometers from the Bolivian border; (3) commu-
nities around Xapuri and the nearby sister towns of Bra-
ia and Epitaciola
ndia, which form a local market hub
along the Bolivian border roughly 230 km from Rio
Branco; and (4) Assis Brasil, located at the far end of the
Inter-Oceanic Highway from Rio Branco in Acre on the
Peruvian border, where paving in Brazil concluded. Given
the obvious distance differentials and their correspon-
dence to time since paving, we view subregion 1 as the
most integrated into the regional economy, followed in
order by 2–4.
In Pando, we defined four subregions in terms of their
integration with the departmental capital of Cobija as well
as the Inter-Oceanic Highway. Because Cobija is located
on the border with Brazil, and because the Inter-Oceanic
Highway passes close by Cobija via its sister towns of
ia and Epitaciola
ndia, our classification in Pando
reflects integration with Cobija as well as distance from the
Inter-Oceanic Highway. The four subregions in Pando are
the following: (1) communities along the Cobija-Sena road,
up to roughly 150 km from Cobija at Puerto Rico on the
near side of the Manuripi river, which lacked a bridge at
the time of surveys; (2) communities along the Cobija-
Extrema road, a smaller corridor also close to Cobija; (3)
communities in Abuna
, farther from Cobija and close to the
Brazilian border at Capixaba and perhaps more integrated
with Acre; and (4) communities along the Cobija-Sena
road, beyond the river at Puerto Rico and closer to Sena
(150–250 km from Cobija) and more oriented to trade on
the Madre de Dios river with other parts of Bolivia than to
the Inter-Oceanic Highway. Pando is thus interesting as a
space being influenced by the Highway, but more in some
subregions than others. We view subregion 1 as most
integrated with Cobija and the Inter-Oceanic Highway,
followed in order by 2–4.
Table 4 presents comparisons between the subregions
within Madre de Dios, Acre, and Pando, using the same
resilience indicators as in Table 3. The first panel in
Table 4 (Panel 4a) presents the findings for Madre de Dios.
Among the demographic/memory variables, there are
weakly significant differences. Tahuamanu, the only sub-
region that had incurred road paving by the time of the
survey, exhibits the largest relative population increases
and turnover. Turning to the diversity indicators, commu-
nities in Tambopata West exhibit less diverse livelihoods
than elsewhere; further analysis indicates that this is due to
lack of NTFPs and less wage work. Interestingly, in
Tahuamanu (paving completed) and the castan
a zone
(paving underway), community livelihoods became more
diverse in the 5 years prior to the surveys, and further
analysis indicates different changes in the two subregions.
Resource availability declined in all four subregions, but
less in the castan
a zone, which also exhibited high and
rising livelihood diversity. Fire damage was most wide-
spread in the castan
a zone and Tahuamanu, though the
results for fire response efficacy are insignificant. Madre de
Dios thus emerges as a highly spatially differentiated
region of the MAP frontier, likely due to road paving still
in progress, as well as biophysical variations (castan
a for-
ests). In subregions with lower connectivity to the regional
capital, there are especially high as well as low values on
resilience indicators, which indicate a complicated rela-
tionship between access connectivity and resilience in
Madre de Dios, likely because the paved portion of the
Inter-Oceanic Highway was not near the regional market
hub in Puerto Maldonado.
In Acre (Panel 4b), there are fewer significant differ-
ences among subregions than in Madre de Dios. Net
migration shows significant differences, with migration
gains in communities near Rio Branco and losses farther
out. In terms of diversity/change, livelihood diversity var-
ies only weakly among subregions in Acre, being some-
what higher near Rio Branco. Further, increases in
diversity appear greater closer to the capital. However,
change in access to forest resources is weakly significant
and does not exhibit a clear pattern in light of access
connectivity. While fire damage was nearly universal
among communities along the Inter-Oceanic Highway in
Acre, there was spatial differentiation in fire response,
being much more problematic closer to Rio Branco. In
sum, spatial differentiation is not as pronounced in Acre as
in Madre de Dios, and resilience does not vary as much in
terms of local access connectivity.
Panel 4c presents the resilience indicators for subregions
of Pando. In contrast to the other two sides, few indicators
exhibit significant differentiation between subregions. This
could result from relatively small weighted n’s since
communities in Pando are relatively small (see Table 2
however, an unweighted analysis of the same subregions
also shows predominantly insignificant findings. There is
the evidence of greater demographic change near Cobija
and in Abuna
, but little else varies by access connectivity.
Local connectivity and resilience: social and economic
We now consider interaction connectivity at the local level
and evaluate its implications for social–ecological
Regional integration and local change 47
resilience. We measure interaction connectivity in terms of
social ties to towns and the extent of product marketing.
Because these variables are all continuous, we categorized
them into quartiles to allow for comparative analysis.
Social interaction connectivity
Table 5 shows the findings for the social connectivity
measure and percentage of member families absent from
Table 4 Comparative analysis of social–ecological resilience indicators by local subregion in the tri-national MAP frontier, southwestern
Puerto Maldonado Tambopata West Castan
a zone Tahuamanu F test
4a. Madre de Dios, Peru
Collective memory
Net migration
?7.23 ?15.55 ?6.99 ?21.13 2.34
Migration turnover 23.51 18.86 20.52 53.94 6.27**
Livelihood diversity 24.57 19.75 24.91 23.83 23.73**
Change in livelihood diversity -0.68 ?0.13 ?3.56 ?4.19 7.42**
Change in forest resource access -3.62 -3.67 -1.68 -3.81 6.40**
Adaptive capacity
Fire damage 0.74 0.20 0.98 0.85 16.60**
Fire response 0.67 0.00 0.36 0.29 6.77
Rio Branco Capixaba Xapuri/Brasile
ia Assis Brasil F test
4b. Acre, Brazil
Collective memory
Net migration
?6.20 -0.31 -5.70 -9.02 3.67*
Migration turnover 22.38 15.36 20.34 21.80 0.26
Livelihood diversity 19.96 18.66 16.94 18.85 2.93
Change in livelihood diversity ?2.62 ?4.98 ?2.24 ?0.88 4.47*
Change in forest resource access -2.99 -1.35 -4.01 -0.75 2.47
Adaptive capacity
Fire damage 1.00 1.00 0.95 0.81 0.49
Fire response 0.64 0.75 0.25 0.00 19.03**
Cobija Extrema Sena Abuna
F test
4c. Pando, Bolivia
Collective memory
Net migration
?18.97 ?9.14 ?8.35 ?45.21 3.12
Migration turnover 32.12 27.74 29.93 49.96 1.07
Livelihood diversity 24.20 22.30 22.00 26.27 0.63
Change in livelihood diversity -0.52 ?1.68 ?0.04 -1.35 1.00
Change in forest resource access -0.38 -1.25 -2.30 -2.58 0.40
Adaptive capacity
Fire damage 0.81 0.89 1.00 0.90 0.25
Fire response 1.00 1.00 1.00 1.00 0.00
See the text and Table 3 for explanations on measurement of resilience indicators
b ?
P \ 0.15; * P \ 0.05; ** P \ 0.01
The test for fire response is a Chi-square (df = 6)
48 S. G. Perz et al.
the community. The resilience measures often vary sig-
nificantly but do not always exhibit clear patterns in dif-
ferentials across quartiles of social interaction connectivity.
Net migration is an example, where communities in the
lowest and highest social connectivity quartiles exhibit
moderate population gains, but those in the middle show
both high and low values. This suggests either strong non-
linearity or confounding by a third variable. Further anal-
ysis confirms that location may be such a factor, as many
cases in Acre appear in the second quartile, which reduces
net migration in that quartile as a reflection of Acre’s rel-
atively low net migration. Turnover, however, shows a
positive linear relationship. Percentage absent families also
exhibit an ambiguous pattern with livelihood diversity, but
again, many cases in the second quartile are in Acre, which
helps explain lower diversity there. Social connectivity
shows a positive relationship with change in availability of
forest resources. This is an intriguing finding, for it sug-
gests that communities more socially connected to towns
actually experienced smaller reductions in availability of
forest resources. Finally, community fire response varies
significantly but ambiguously across quartiles of relative
social connectivity.
Economic interaction connectivity
Table 6 evaluates the resilience indicators across quartiles
of economic connectivity measured as total products sold
by various means. The findings are significant but occa-
sionally ambiguous. Differences in net migration are
weakly significant, with a positive relationship such that
greater economic connectivity corresponds to more posi-
tive net migration. Turnover is significant and exhibits a
non-linear relationship, with the third quartile exhibiting
higher values. Livelihood diversity is significant and higher
among communities with greater marketing diversity. The
finding for change in livelihood diversity is a U-shaped
relationship, so at low and high levels of economic con-
nectivity, livelihood diversity increases the most. Further
analysis indicates that the activities increasing in impor-
tance are very different at low and high levels of economic
connectivity; at low levels, it is crops, cattle, and wages; at
high levels, it is timber, NTFPs, and crops. This again
raises suspicions that location helps explain these findings,
as several communities in Acre fall in the lowest quartile,
and in Acre, crops, cattle, and wages increased in impor-
tance. The finding for availability of forest resources is
more ambiguous, and the declines appear larger at lower
levels of economic connectivity. Fire damage is more
common among more economically connected communi-
ties. The finding for community fire response is also sig-
nificant, but no clear pattern emerges across the quartiles of
economic connectivity.
As anticipated, the findings present a multifaceted picture,
partly a reflection of the diverse impacts of infrastructure
and partly due to the multidimensional nature of both
connectivity and social–ecological resilience. We note that
the findings reflect those seen in previous though frag-
mented work on road impacts. As noted in economic
research on infrastructure, the onset of road paving accel-
erates population growth via migration (e.g., Perz et al.
2010); as seen in work on road ecology, road paving
Table 5 Comparative analysis of social–ecological resilience indicators among quartiles of social connectivity measures in the tri-national MAP
frontier, southwestern Amazon
Percentage absent member families
Lowest quartile Second quartile Third quartile Highest quartile F test
Collective memory
Net migration
?11.62 ?0.02 ?15.43 ?16.45 6.38**
Migration Turnover 16.99 18.53 28.87 32.60 4.56**
Livelihood diversity 20.60 18.74 23.50 22.35 10.77**
Change in livelihood diversity 2.51 2.23 1.78 1.15 0.85
Change in forest resource access -2.92 -3.78 -2.00 -2.26 3.30*
Adaptive capacity
Fire damage 0.60 0.76 0.77 0.87 1.54
Fire response 0.50 0.45 0.65 0.42 23.42**
See the text and Table 3 for explanations on measurement of resilience indicators
b ?
P \ 0.15; * P \ 0.05; ** P \ 0.01
The test for fire response is a Chi-square (df = 6)
Regional integration and local change 49
increases risks of fire damage due to forest fragmentation
(e.g., Forman et al. 2003); and as observed in social
research, new infrastructure affects the distribution of
benefits, as via reduced access to forest resources in areas
with road paving (e.g., Leonel 1992). In this sense, our
framework and our findings integrate prior literatures on
road impacts into a single analytical framework, which
offers a more complete and useful account of the ramifi-
cations of new infrastructure.
A conclusion from the multifaceted picture that emerges
from our framework is that the relationship of connectivity
to social–ecological resilience is not unitary. Greater con-
nectivity corresponds to greater resilience in some respects
(less population change, greater increases in livelihood
diversity), but less resilience in others (less diverse liveli-
hoods, greater declines in resource availability, more
widespread fire damage). The larger picture that emerges is
that the relationship between the structure of a regional
system (how it is organized as captured by connectivity
measures) and the social–ecological resilience of its com-
ponents (in this case, communities) cannot be reduced to a
single dimension or value.
The findings also reveal complications in relationships
between connectivity and social–ecological resilience. For
one thing, the link between access connectivity and social–
ecological resilience varies between the regional and local
levels; for another, access and interaction connectivity at
the local level do not exhibit the same relationships with
resilience indicators, and finally, relationships between
connectivity and resilience vary across sides of the MAP
frontier. To an extent, these complications reflect local
vagaries in the study region. At the regional level, Acre has
greater access connectivity via road paving but less
migration activity, a counterintuitive finding if infrastruc-
ture theoretically stimulates population mobility. However,
Acre also has a more consolidated land tenure regime,
itself a result of earlier road building and land conflicts that
have since been largely resolved (Perz et al. 2010). Further,
paving in Tahuamanu, Madre de Dios is far from Puerto
Maldonado. This yielded non-linearities in the subregional
findings on the Peruvian side of the tri-national frontier,
something that may change as paving proceeds to other
parts of Madre de Dios.
The findings nonetheless raise issues concerning the
analysis of social–ecological systems. Research on eco-
logical resilience often highlights complex dynamics in
systems with a singularly crucial component that controls
system functioning. Research on social–ecological resil-
ience, which features larger numbers of system compo-
nents, including social actors who may do several different
things with various ecological ramifications, substantially
complicates such analysis. As a consequence, in light of
external shocks such as integration initiatives, assessment
of social–ecological resilience necessitates selecting indi-
cators of different aspects of the social–ecological inter-
face, such as those highlighted in our analysis. If one views
connectivity and resilience as unidimensional concepts, the
dimensions chosen will greatly skew the conclusions. We
assert that a multidimensional approach to the analysis of
resilience offers the advantage of allowing observation of
tradeoffs evident in changes that may not only be complex
(non-linear) in their dynamics but also complicated in
terms of the multiple directions of relationships between
system components and changes therein. In complicated
systems such as social–ecological systems, it is likely that
external shocks will generate manifold consequences
Table 6 Comparative analysis of social–ecological resilience indicators among quartiles of economic connectivity measures in the tri-national
MAP frontier, southwestern Amazon
Total products sold, any means
Lowest quartile Second quartile Third quartile Highest quartile F test
Collective memory
Net migration
?5.32 ?7.77 ?14.70 ?16.03 2.43
Migration turnover 17.47 23.00 33.55 24.92 2.87*
Livelihood diversity 19.55 19.00 23.40 24.19 21.44**
Change in livelihood diversity 3.15 1.45 0.30 3.10 4.84**
Change in forest resource access -2.84 -3.60 -1.45 -2.51 4.38**
Adaptive capacity
Fire damage 0.63 0.65 0.96 0.83 3.41*
Fire response 0.58 0.39 0.70 0.38 20.18**
See the text and Table 3 for explanations on measurement of resilience indicators
b ?
P \ 0.15; * P \ 0.05; ** P \ 0.01
The test for fire response is a Chi-square (df = 6)
50 S. G. Perz et al.
including variation in system connectivity and both posi-
tive and negative implications for resilience. In our study
region, communities exhibit differing profiles of resilience;
for example, communities in Acre exhibit less migration
and less diversified livelihoods. We suggest that spatially
differentiated and multidimensional resilience analyses
will yield similar findings in other study cases.
We conclude with observations arising from our inter-
active deployment of transport geography and resilience
thinking in environmental studies of integration initiatives.
On the one hand, resilience thinking forces us to look
beyond the economic focus on transport costs and effi-
ciency, important though they are, to consider other factors
that may be crucial to understanding integration and its
implications. Employment of measures inspired by resil-
ience thinking related to memory, diversity, and adaptive
capacity reveals different dimensions of change and spatial
variation in the dynamics of regional integration. Indicators
of adaptive capacity, such as fire damage and community
response, can be crucial for understanding communities
beyond their livelihood strategies. On the other hand,
transport geography provides an essential corrective to the
often abstruse and a-spatial resilience literature. The scale
of regional integration initiatives implies that resilience
analyses need to be scaled up to consider larger geogra-
phies. Scaling up, however, leads to the unavoidable
eventuality that there will be spatial variation in system
components and relationships, presuming they are not
averaged out over conveniently large areas. Transport
geography has long-featured system-level models that
differentiate between locations and incorporate spatial
relationships in order to evaluate system-level changes in
light of system components with characteristics tied to
These considerations beg questions concerning the
endeavor to integrate transport geography and resilience
thinking as a means of bringing together the heretofore
largely disparate literatures on infrastructure impacts.
Large-scale infrastructure projects are likely to reconfigure
regional and local spaces, with the implication that social–
ecological resilience may vary spatially, with implications
for larger-scale resilience. Transport geography has shown
that improvements in infrastructure and reduced transport
costs not only expand market penetration in the countryside
but also render some local market centers less important in
distribution chains. From a spatialized resilience perspec-
tive, this system-level restructuring has very different
ramifications among localities that become distribution
points and those that are bypassed. This also bears impli-
cations for local producers whose resource management
and livelihood strategies may be modified, albeit differ-
ently in different locations. It remains an open question as
to whether and how such regional economic restructuring
may result in the loss of system-level social–ecological
resilience, if it is lost in some localities but not others.
In this narrative, changes in connectivity are crucial
determinants of local resilience and are likely also vital to
broader system changes. Evaluation of such a narrative,
however, requires a structural approach to the spatial orga-
nization of regions as systems, paired with a dynamic
analysis to fully assess the implications of changes in con-
nectivity for social–ecological resilience. Empirical models
of infrastructure impacts have yet to incorporate multiple
dimensions of social–ecological interfaces, and simulations
models have largely highlighted spatial contagions among
contiguous raster cells in landscapes rather than telecon-
nections among nodes at varying distances. Graph theoretic
approaches highlight such teleconnections for the economic,
ecological, and social (Hage and Harary 1996; Arlinghaus
2001; Urban and Keitt 2002), and if combined with spatial
insights from economic geography and a focus on the
social–ecological interface highlighted in resilience think-
ing, we gain a more integrative approach to evaluate infra-
structure impacts. Such an approach affords consideration of
local-level connections as well as system-level connectivity,
not only for community-market ties but also for landscape
connectivity. Modeling changes in connectivity also allows
for dynamic assessment of the ramifications of highway
paving for community social–ecological resilience.
Acknowledgments Financial support for this research came from
the National Science Foundation, Human and Social Dynamics Pro-
gram, Grant #0527511, and from the US Agency for International
Development, Latin America and Caribbean program in Environment,
Cooperative Agreements RLA-A-00-06-00071-00 and 512-A-00-08-
00003-00. The coauthors are coordinators and field team leaders of
the socioeconomic component of the NSF grant, and they thank the
other collaborators who contributed to the community survey field-
work and data entry in Madre de Dios, Peru (Ange
lica Almeyda,
Wendy Cueva Cueto, Eder Nicanor Chilla Pfuro, Boris Arguedas,
Erika Quispe Ruiz, Andrea Cha
vez, Rafael Rojas), Acre, Brazil
(Karla Rocha, Jesus Melo, Vera Gurgel), and Pando, Bolivia (Kelly
Biedenweg, Dave Elliott, Alexander Shenkin). For logistical support,
we thank Veronica Passos, Bertha Ikeda, and Daniel Rojas. For
helpful suggestions, we thank Julio Rojas, Frank P. de la Barra, Amy
Duchelle, Valerio Gomes, and Jacqueline Vadjunec.
Adger WN (2000) Social and ecological resilience: are they related?
Prog Hum Geog 24:347–364
Adger WN (2006) Vulnerability. Global Env Change 16:268–281
Adger WN, Kelly PM, Winkels A, Huy LG, Locke C (2002)
Migration, remittances, livelihood trajectories, and social resil-
ience. Ambio 31:358–366
Adger WN, Hughes TP, Folke C, Carpenter SR, Rockstro
m J (2005)
Social-ecological resilience to coastal disasters. Science
Arlinghaus S (2001) Graph theory and geography: an interactive
view. Wiley, New York
Regional integration and local change 51
Aschauer DA (1989) Is public infrastructure productive? J Monet
Econ 23:177–200
Berke F, Turner NJ (2006) Knowledge, learning and the evolution of
conservation practice for social-ecological system resilience.
Hum Ecol 34:479–494
Bigman D, Fofack H (2000) Geographical targeting for poverty
alleviation: an introduction to the special issue. World Bank
Econ Rev 14:129–145
Binswanger H, Khandker SR, Rosenzweig MR (1993) How infra-
structure and financial institutions affect agricultural output and
investment in India. J Dev Econ 41:337–366
Boarnet MG (1999) Road infrastructure, economic productivity, and
the need for highway finance reform. Pub Works Manag Pol
Bourguignon F, Pleskovic B (2008) Rethinking infrastructure for
development. Annual World Bank conference on development
economics—2007. World Bank, Washington, DC
Brown IF, Brilhante SCH, Mendoza E, Ribeiro de Oliveira I (2002)
Estrada de Rio Branco, Acre, Brasil aos Portos do Pacı
Como Maximizar os Benefı
cios e Minimizar os Prejuı
zos para o
Desenvolvimento Sustentable da Amazo
nia Sul-Ocidental. In:
CEPEI, La Integracio
n Regional Entre Bolivia, Brasil y Peru.
Bulmer-Thomas V (2001) Regional integration in Latin America and
the Caribbean. Bull Latin Am Res 20(3):360–369
n C, Serven L (2004) The effects of infrastructure develop-
ment on growth and income distribution. Central Bank of Chile,
working paper No. 270
Carpenter S, Walker B, Anderies JM, Abel N (2001) From metaphor
to measurement: resilience of what to what? Ecosystems
Carranza ME (2002) Neighbors or partners? NAFTA and the regional
politics of integration in North America. Latin Am Pol Soc
Centro Peruano de Estudios Internacionales (CEPEI) (2002) La
n regional entre bolivia, Brasil y Peru. CEPEI, Lima
Christen D, Matlack G (2006) The role of roadsides in plant
invasions: a demographic approach. Con Biol 20:385–391
Coffin AW (2007) From roadkill to road ecology: a review of the
ecological effects of roads. J Transp Geog 15:396–406
Craig BJ, Pardey PJ, Roseboom J (1997) International productivity
patterns: accounting for input quality, infrastructure, and
research. Am J Agr Econ 79:1064–1076
Cronon W (1991) Nature’s metropolis: Chicago and the Great West.
W.W. Norton, New York
Cross C (2001) Why does South Africa need a spatial policy?
Population migration, infrastructure and development. J Contemp
Afr St 19:111–127
Cumming GS, Barnes G, Perz SG, Schmink M, Sieving KE,
Southworth J, Binford M, Holt RD, Stickler C, Van Holt T
(2005) An exploratory framework for the empirical measurement
of resilience. Ecosystems 8:975–987
Davis SH (1977) Victims of the miracle: development and the Indians
of Brazil. Cambridge University Press, Cambridge
Demurger S (2001) Infrastructure and economic growth: an expla-
nation for regional disparities in China? J Comp Econ 29:95–117
Devres Inc (1981) Socio-economic and environmental impacts of low
volume rural roads: a review of the literature. US Agency for
International Development, Washington, DC
Dourojeanni MJ, Barandiara
n A, Dourojeanni D (2010) Amazonı
Peruana en 2021. SPDA, Lima
Eigenbrod F, Hecnar SJ, Fahrig L (2008) Accessible habitat: an
improved measure of the effects of habitat loss and roads on
wildlife populations. Landsc Ecol 23:159–168
Ellis F (2000) Rural livelihoods and diversity in developing countries.
Oxford University Press, Oxford
Fan S, Hazell P, Haque T (2000) Targeting public investments by
agro-ecological zone to achieve growth and poverty alleviation
goals in Rural India. Food Pol 25:411–428
Fan S, Zhang L, Zhang X (2004) Reform, investment, and poverty in
China. Econ Dev Cultural Change 52:395–421
Folke C (2006) Resilience: the emergence of a perspective for social-
ecological systems analyses. Global Env Change 16:253–267
Folke C, Colding J, Berkes F (2003) Building resilience and adaptive
capacity in social-ecological systems. In: Berkes F, Colding J,
Folke C (eds) Navigating social-ecological systems: building
resilience for complexity and change. Cambridge University
Press, Cambridge, pp 352–387
Forman RTT, Alexander LE (1998) Roads and their major ecological
effects. Ann Rev Ecol Syst 29:207–231
Forman RTT, Sperling D, Bisonette JA, Clevenger AP, Cutshall CP,
Dale VH, Fahrig L, France R, Goldman CR, Heanue K, Jones
JA, Swanson FJ, Turrentine T, Winter TC (2003) Road ecology:
science and solutions. Island Press, Washington, DC
Genna GM, Hiroi T (2004) Power preponderance and domestic
politics: explaining regional economic integration in Latin
America and the Caribbean, 1960–1997. Int Interact 30:143–164
Gibson J, Rozelle S (2003) Poverty and access to roads in Papua New
Guinea. Econ Dev Cultural Change 52:159–185
Goldman A (1995) Threats to sustainability in African agriculture:
searching for appropriate paradigms. Human Ecol 23(3):291–334
Goodwin BJ, Fahrig L (2002) How does landscape structure influence
landscape connectivity? Oikos 99:552–570
Gramlich E (1994) Infrastructure investment: a review essay. J Econ
Lit 32:1176–1196
Gunasekera K, Anderson W, Lakshmanan TR (2008) Highway-
induced development: evidence from Sri Lanka. World Dev
Gunderson LH (2000) Ecological resilience—in theory and applica-
tion. Ann Rev Ecol Syst 31:425–439
Hage P, Harary F (1996) Island networks: communication, kinship,
and classification structures in Oceania. Cambridge University
Press, Cambridge
Haggett P, Cliff AD, Frey A (1977) Locational analysis in human
geography. Wiley, New York
Hall AL (1989) Developing Amazonia: deforestation and social
conflict in Brazil’s Caraja
s programme. Manchester University
Press, Manchester
Hansen MJ, Clevenger AP (2005) The influence of disturbance and
habitat on the presence of non-native plant species along
transport corridors. Bio Con 125:249–259
Hawbaker TJ, Radeloff VC (2004) Roads and landscape pattern in
Northern Wisconsin based on a comparison of four road data
sources. Con Biol 18:1233–1244
Hemming J (1985) Change in the Amazon Basin, 2 vols. Manchester
University Press, Manchester
Holling CS (1973) Resilience and stability of ecological systems. Ann
Rev Ecol Syst 4:1–23
Howe J, Richards P (1984) Rural roads and poverty alleviation. ILO,
Iniciativa MAP (2008) URL available at
Accessed 20 Jan 2008
Iniciativa para la Integracion del a Infraestructura Regional de
America del Sur (IIRSA) (2008) URL available at
. Accessed 20 Jan 2008
Isard W (1956) Location and space-economy: a general theory
relating to industrial location, market areas, land use, trade, and
Urban Structure. MIT/Wiley, Cambridge
Jacoby HG (2000) Access to markets and the benefits of rural roads.
Econ J 110:713–737
Jaeger JAG, Bowman J, Brennan J, Fahrig L, Bert D, Bouchard J,
Charbonneau K, Frank K, Gruber B, von Toschanowitz KT
52 S. G. Perz et al.
(2005) Predicting when animal populations are at risk from
roads: an interactive model of road avoidance behavior. Ecol
Mod 185:329–348
Katzman MT (1977) Cities and frontiers in Brazil: regional dimen-
sions of economic development. Harvard University Press,
Cambridge, MA
Kellogg P (2007) Regional integration in Latin America: dawn of an
alternative to neoliberalism? New Political Sci 29:187–209
Killeen TJ (2007) A perfect storm in the Amazon Wilderness:
development and conservation in the context of the initiative for
the integration of regional infrastructure of South America
(IIRSA). Conservation International, Arlington, VA
Killeen TJ, Solo
rzano LA (2008) Conservation strategies to mitigate
impacts from climate change in Amazonia. Phil Trans R Soc B
Laurance WF, Bierregaard RO (1997) Tropical forest remnants:
ecology, management and conservation of fragmented commu-
nities. University of Chicago Press, Chicago
Laurance WF, Lovejoy TE, Vasconcelos HL, Bruna EM, Didham RE,
Stouffer PC, Gascon C, Bierregaard RO, Laurance SG, Sampaio
E (2002) Ecosystem decay of amazonian forest fragments: A
22-year investigation. Con Biol 16:605–618
Leonel M (1992) Roads, Indians and the environment in the Amazon:
from Central Brazil to the Pacific Ocean. International Working
Group for Indigenous Affairs, Copenhagen
Levin S (1999) Fragile dominion: complexity and the commons.
Perseus, Reading, MA
Mahapa SM, Mashiri S (2001) Social exclusion and rural transport:
gender aspects of a road improvement project in Tshitwe,
Northern Province. Dev South Afr 18:365–376
Marengo JA, Nobre CA, Tomasella J, Oyama MD, Sampaio de
Oliveira G, de Oliveira R, Camargo H, Alves LM, Brown IF
(2008) The drought of Amazonia in 2005. J Climate 21:495–516
Mendoza E, Perz SG, Schmink M, Nepstad D (2007) Participatory
stakeholder workshops to mitigate impacts of road paving in the
Southwestern Amazon. Con Soc 5:382–407
Millikan BH (1992) Tropical deforestation, land degradation, and
society: lessons from Rondonia, Brazil. Latin Am Perspectives
Minot N, Baulch B, Epprecht M (2003) Poverty and inequality in
Vietnam: spatial patterns and geographic determinants. Interna-
tional Food Policy Research Institute
Moreira Mesquita M (2007) Trade Costs and Economic Fundamen-
tals of the Initiative for Integration of Regional Infrastructure of
South America (IIRSA). Inter-American Development Bank:
INTAL-ITD Working Paper No. 30
Munakata N (2006) Transforming East Asia: the evolution of regional
economic integration. Research Institute Economy, Trade and
Industry, Tokyo
Munnell A (1992) Public infrastructure investment and economic
growth. J Econ Persp 6:189–198
Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J
(2000) Biodiversity hotpots for conservation priorities. Nature
Nepstad DC, Stickler CM, Soares-Filho B, Merry F (2008) Interac-
tions among amazon land use, forests, and climate: prospects for
a near-term forest tipping point. Phil Trans R Soc B
North RN (1979) Transport in Western Siberia: Tsarist and Soviet
development. University of British Columbia Press, Vancouver
Pender J, Jagger P, Nkonya E, Sserunkuuma D (2004) Development
pathways and land management in Uganda. World Dev
Perz SG (2005) The effects of household asset endowments on
agricultual diversity among frontier colonists in the Amazon.
Agrofor Syst 63:263–279
Perz SG, Cabrera L, Carvalho LA, Castillo J, Barnes G (2010) Global
economic integration and local community resilience: new
infrastructure projects and demographic change in the south-
western Amazon. Rural Sociol 75:300–325
Peterson G (2000) Political ecology and ecological resilience: an
integration of human and ecological dynamics. Ecol Econ
Randall L (1997) The political economy of Latin America in the
postwar period. University of Texas Press, Austin
Robinson R (2001) Decentralization of road administration: case
studies in Africa and Asia. Pub Admin Dev 21:53–64
Rudel TK, Richards S (1990) Urbanization, roads, and rural
population change in the ecuadorian andes. St Comp Int Dev
Schmink M, Wood CH (1984) Frontier expansion in Amazonia.
University Press of Florida, Gainesville, FL
Schmink M, Wood CH (1992) Contested frontiers in Amazonia.
Columbia University Press, New York
Shriar AJ (2009) Roads to poverty: exploring the impacts of
economic integration on socioeconomic conditions and land
use in northern Guatemala. J Plan Ed Res 28:456–469
Straub S (2008) Infrastructure and growth in developing countries:
recent advances and research challenges. World Bank policy
research working paper No. 4460, Washington, DC
Subbarao K (1997) Public works as an anti-poverty program: an
overview of cross-country experience. Am J Agr Econ
Trankell IB (1999) On the road in Laos: an anthropological study of
road construction and rural communities. White Lotus Press,
Trombulak SC, Frissell CA (2000) Review of ecological effects of
roads on terrestrial and aquatic communities. Con Biol
Turner NJ, Davidson-Hunt IJ, O’Flaherty M (2003) Living on the
edge: ecological and cultural edges as sources of diversity for
social-ecological resilience. Hum Ecol 31:439–461
Urban D, Keitt T (2002) Landscape connectivity: a graph-theoretic
perspective. Ecology 82:1205–1218
Van de Walle D (2002) Choosing rural road investments to help
reduce poverty. World Dev 30:575–589
Van Dijk P, den Haak S (2007) Troublesome construction: IIRSA and
public-private partnerships in road infrastructure.
van Oosten C (2004) Fading frontiers? Local development and cross-
border partnerships in Southwest Amazonia. University of
Utrecht, Utrecht
Walker BJ, Salt D (2006) Resilience thinking: sustaining ecosystems
and people in a changing world. Island Press, Washington, DC
Walker RT, Perz SG, Arima E, Simmons C (2011) The transamazon
highway: past, present, future. In: Brunn SD (ed) Chap 33 in
engineering earth. Springer, New York
Windle J, Cramb RA (1997) Remoteness and rural development:
economic impacts of rural roads on upland farmers in Sarawak,
Malaysia. Asia Pacific Viewpoint 38:37–53
World Bank (1994) World development report 1994: infrastructure
for development. World Bank, Washington, DC
Regional integration and local change 53
... En este capítulo, se revisan los cambios y los procesos subyacentes en la Amazonía como SSE. En concreto, nos centramos en la Amazonía suroccidental, específicamente la frontera trinacional, donde Bolivia, Brasil y Perú se encuentran (Perz et al., 2012). Este foco permite una mayor flexibilidad en nuestra discusión, pues la Amazonía es sin duda un espacio enorme. ...
... Nos centramos en la Amazonía suroccidental, en concreto la frontera trinacional donde Bolivia, Brasil y Perú se encuentran (Mendoza et al., 2008;Perz et al., 2012). Específicamente, nos centramos en la región «MAP», integrada por las divisiones políticas que forman la frontera trinacional: Madre de Dios (Perú), Acre (Brasil) y Pando (Bolivia). ...
... El primer período de proyectos de IIRSA contó con la pavimentación de la Carretera Interoceánica. Esta obra fue un proyecto principal de IIRSA, y actualmente es la carretera principal que cruza la frontera MAP (Perz et al., 2012). Si se considera la anterior generación de proyectos de infraestructura en la Amazonía, que produjeron deforestación generalizada, conflictos sociales y nuevos tipos de uso de los recursos, incluidos en la frontera MAP (Leonel, 1992), IIRSA y la Carretera Interoceánica han estimulado el debate polémico sobre las posibles consecuencias para los bosques, la biodiversidad, las poblaciones locales y las reivindicaciones territoriales tradicionales (Killeen, 2007;Dourojeanni et al., 2010). ...
Full-text available
El capítulo se enfoca específicamente en la Amazonía suroccidental, particularmente en la frontera trinacional de Bolivia, Brasil y Perú. Esta área se selecciona por su diversidad y complejidad, permitiendo una evaluación detallada de las interacciones socio-ecológicas en un espacio heterogéneo donde los cambios y sus procesos subyacentes varían en el tiempo y entre localidades. Se examina el cambio histórico en la región hasta la década de 1990, distinguiendo entre aspectos sociales y ecológicos antes de abordar de manera explícita las interacciones socio-ecológicas. El análisis de estas interacciones facilita la comprensión de los impulsores específicos de los cambios y las reacciones, lo que a su vez ayuda a explicar los cambios no lineales y a entender la estructura de los mecanismos subyacentes .
... This characteristic makes the MAP region strategic for studies that focus on forest conservation and environmental change mitigation programs with transboundary implications, contemplating fire risk and assessment of its impacts. The MAP region economy is based on the extraction of wood, nuts, and livestock, mainly in the states/departments of Pando and Madre de Dios; therefore, the lack of alignment of countryspecific public policies, combined with distinct political interests in this transnational region, may directly impact these activities, magnifying the advancement of deforestation and forest fire degradation (Souza et al., 2006;Marsik et al., 2011;Perz et al., 2012;Michaelsen et al., 2013). For instance, the annual total deforested area in the states/departments of Acre and Madre de Dios increased by 137 and 103%, respectively, in 2020 compared to 2012 (INPE-National Institute for Space Research, 2022a; RAISG-Amazon Network of Georeferenced Socio-Environmental Information, 2022). ...
... The results showed that, between 148 to 163 grid cells, representing ca. 31 ± 1.17% of the MAP region, had 25 or more fire counts, defined based on the third quartile used for calibrating the model. The patterns found followed the historical geography of fire occurrence in this region (Perz et al., 2012;Baraloto et al., 2015). Fires were concentrated in the most anthropized regions, in the northwest, extending along the northern borders, reaching the northeast flank and central east portions of the study area. ...
Full-text available
Fires are among the main drivers of forest degradation in Amazonia, causing multiple socioeconomic and environmental damages. Although human-ignited sources account for most of the fire events in Amazonia, extended droughts may magnify their occurrence and propagation. The southwestern Amazonia, a transnational region shared by Brazil, Peru, and Bolivia and known as the MAP region, has been articulating coordinated actions to prevent disasters, including fire, to reduce their negative impacts. Therefore, to understand the fire patterns in the MAP region, we investigated their main drivers and the changes in the suitability of fire occurrence for the years 2005, 2010, 2016, and 2020. We used a maximum entropy (MaxEnt) model approach based on active fire data from satellites, climatic data, and land use and land cover mapping to spatially quantify the suitability of fire occurrence and its drivers. We used the year 2015 to calibrate the models. For climatic data and active fire count, we only considered grid cells with active fire count over the third quartile. All our models had a satisfactory performance, with values of the area under the curve (AUC) above 0.75 and p < 0.05. Additionally, all models showed sensitivity rates higher than 0.8 and false positive rates below 0.25. We estimated that, on average, 38.5% of the study region had suitable conditions for fire occurrence during the study period. Most of the fire-prone areas belong to Acre, representing approximately 74% of the entire MAP region. The percentage of deforested areas, productive lands, forest edges, and high temperatures were the main drivers of fire occurrence in southwestern Amazonia, indicating the high vulnerability of fragmented landscapes extreme climatic conditions to fire occurrence. We observed that the modeling approach based on Maxint is useful for useful for evaluating the implications of climatic and anthropogenic variables on fire distribution. Furthermore, because the model can be easily employed to predict suitable and non-suitable locations for fire occurrence, it can to prevent potential impacts associated with large-scale wildfire in the future at regional levels.
... Resilience is understood as an emerging property of socio-ecosystems that determines ecosystem stability and its ability to increase its capacity to learn and adapt in response to natural or human-induced perturbations (Holling 1973;Lenton et al. 2008). The concept of 'social-ecological resilience' integrates the study of landscape change intensity as a consequence of socioeconomic alterations (Perz et al. 2012;Salvati et al. 2013), since landscape is the result of interactions among society, economy and ecosystems (Lepart and Debussche 1992;Sirami et al. 2010). Quantifying and understanding SES resilience allows to predict and adapt to landscape changes (Hu et al. 2018) and, together with vulnerability to land degradation, is an important tool for sustainable land use planning and decision-making (Salvati et al. 2013). ...
Full-text available
Context The difficulty of analysing resilience and threshold responses to changing environmental drivers becomes evident in the social-ecological systems framework due to their inherent complexity. Research is needed to develop new tools able to deal with such challenges and determine potential thresholds for SES variables that primarily influence tipping point behaviour. Objectives In this paper, a methodology based on the application of Bayesian Networks (BNs) has been developed to quantify the social-ecological resilience along an urban–rural gradient in Madrid Region, detecting the tipping point values of the main socioeconomic indicators implying critical transitions at landscape stability thresholds. Method To do this, the spatial–temporal trends of the landscape in an urban–rural gradient from Region de Madrid (Spain) were identified, to then quantify the intensity of the changes and explain them using BNs based on regression models. Finally, through inference propagation the thresholds of landscape change were detected. Results The results obtained for the study area indicate that the most resilient landscapes analysed are those where the traditional silvo-pastoral activity was maintained by elderly people and where there is cohesion between neighbouring rural municipalities. Conclusion The method developed has allowed us to detect the tipping points from which small changes in socioeconomic indicators generate large changes at the landscape level. We demonstrate that the use of BNs is a useful tool to achieve an integrated social-ecological spatial planning.
... In the Amazon basin, there have been many cases of infrastructure projects without transparent planning and implementation, which has made the governance of project impacts the focus of long-term debate. Highways such as the Cuiabá-Santarém, the Inter-Oceanic, the Marginal de la Selva in Colombia, and the proposed road across Isiboro Sécure National Park and Indigenous Territory, more commonly referred to as TIPNIS, have all involved polemics over their impacts (e.g., Alencar et al. 2004, Perz et al. 2012, Achtenburg 2013, Dominguez Ossa 2019. Similarly, hydroelectric dams including the Tucuruí, Jirau, Santo António, and Belo Monte have been the focus of competing claims that yielded histories of complications and stoppages, but which nevertheless advanced to implementation (e.g., Moretto et al. 2012, Athayde 2014, Fearnside 2014, Chen et al. 2015, Jiang et al. 2018, Santos et al. 2018. ...
Full-text available
2022. The wisdom of hindsight: a comparative analysis of timelines of environmental governance of infrastructure across the Pan-Amazon. Ecology and Society 27(1):28. https://doi. ABSTRACT. The planning and implementation of infrastructure projects is a long-term enterprise that involves debates over the many positive and negative impacts. Previous work has examined questions about conditions for effective environmental governance of infrastructure, but has typically focused on individual projects and short time frames. We therefore pursued an historical approach to environmental governance of infrastructure projects across multiple cases, taking up examples of highways and dams in the Amazon. Through multi-stakeholder workshops, conservation partners developed historical timelines of events concerning governance of infrastructure in four regions within the basin. Timelines permit analysis to identify periods of particular dynamism, improvements and declines in governance effectiveness, identification of influential stakeholders and events, and conditions that define the effectiveness of governance. We conclude with lessons within and across cases about conditions and strategies for effective environmental governance of infrastructure.
... Drawing on the large body of literature dealing with rural livelihoods in the region (e.g., Cousins, 1999;Cocks and Wiersum, 2003;Shackleton and Shackleton, 2004), we focus on livelihood diversity, household health, and perceptions of environmental change to measure adaptive capacity. Previously, adaptive capacity was considered a function of livelihood diversity (Turner, Davidson-Hunt, and O'Flaherty, 2003;Adger et al., 2005;Uy, Shaw, and Takeuchi, 2008;Perz et al., 2012), as well as health and wellbeing. These were most closely linked together in case studies on water and health (Ebi & Sebenza, 2008;Bunch, 2011). ...
Full-text available
This article explores the problem of studying the temporal dimensions of human-environmental interactions, especially in light of not having available longitudinal data. We utilize a methodology that highlights the past, present, and future, in order to approximate these kinds of results in rural South Africa. The present is measured in the form of livelihood surveys. For the past, oral histories were conducted with elderly people in four villages to acquire information about past adaptive strategies. For the future, focus groups and fuzzy cognitive maps (FCM) of household participants in a workshop setting were conducted so as to understand what adaptations they envisage. We found that present conditions for adaptive capacity do not always align with those described in the past or envisaged for the future, but linkages emerge in a number of instances. Studies like this provide a means for temporal analysis without necessitating the use of longitudinal data.
Full-text available
Both farmers and traders benefit from trade networking, which is crucial for the local economy. Therefore, it is crucial to understand how these networks operate, and how they can be managed more effectively. Throughout this study, we examine the economic networks formed between farmers and traders through the trade of food products. These networks are analyzed from the perspective of their structure and the factors that influence their development. Using data from 18 farmers and 15 traders, we applied exponential random graph models. The results of our study showed that connectivity, Popularity Spread, activity spread, good transportation systems, and high yields all affected the development of networks. Therefore, farmers’ productivity and high market demand can contribute to local food-crop trade. The network was not affected by reciprocity, open markets, proximity to locations, or trade experience of actors. Policy makers should consider these five factors when formulating policies for local food-crop trade. Additionally, local actors should be encouraged to use these factors to improve their network development. However, it is important to note that these factors alone cannot guarantee success. Policy makers and actors must also consider other factors such as legal frameworks, economic policies, and resource availability. Our approach can be used in future research to determine how traders and farmers can enhance productivity and profit in West Africa. This study addresses a research gap by examining factors influencing local food trade in a developing country.
There is a large research literature on the impacts of roads and other infrastructure, which highlights the economic benefits, environmental harms and social problems. Most previous research on infrastructure impacts adopts a top-down approach, such as via the use of governmental or remotely-sensed data. This paper argues that a bottom-up approach that features stakeholder perspectives offers complementary advantages to understanding infrastructure impacts that can support improved planning and governance. We conducted stakeholder workshops about impacts of the Interoceanic Highway in the tri-national “MAP” frontier of the southwestern Amazon. The findings confirm previous research in several respects, but also indicate several contrasts. The range of impacts is much broader than topics featured in previous research, and some of the most commonly reported problems, such as diverse forms of crime, have been rarely studied as infrastructure impacts. We conclude by discussing the implications, in terms of criminological research on infrastructure impacts, synergies among diverse impacts of infrastructure, and improved planning of infrastructure for better governance of impacts.