During the last 20 years, the Amazon region has been at the same time a place of massive ecological and social change and a laboratory of experiments aimed at promoting sustainable development. Policies and project initiatives involving diverse social groups and environmental contexts have been implemented across the region. They have resulted in mixed outcomes and trade-offs between social and environmental dimensions, making their impact at the local level difficult to assess and their successes difficult to generalize. The objective of the DURAMAZ research project was to provide a better understanding of these impacts. It produced a multi-dimensional indicator system designed to allow a holistic view of sustainable development at local and subregional levels and a comparative perspective across 12 research sites, from an isolated indigenous village to smallholders and agribusiness areas in Mato Grosso. The results of the first observation campaign (2007–2009) show that despite the claim of promoting sustainable development, no project was able to untie the ‘Gordian knot’ of development in the Amazon. Communities continue to face the old dilemma of either enjoying a preserved ecosystem but enduring adverse life conditions, or enjoying better living at the expense of forest cover. Another finding is that the subregional context is very important in shaping the impacts of regional policies. Thus, the same policy will not always have the same effect, depending on in which context it is applied. Finally, we found that cultural factors and a sense of place play a more important role than economic factors when it comes to the way people evaluate their own situation. This research provides the basis for a second phase of the project (2012–2016) in which we will continue to expand our sample and to refine our methodologies with the goal of transforming the initiative into a network of observatories of sustainable development in the Amazon.
During the last 20 years, the Brazilian Amazon has experienced massive efforts in sustainable development policies and projects aimed at reducing deforestation rates and improving social conditions . Hundreds, if not thousands, of small initiatives have been financed at the local level through mechanisms such as those related to the G7 project for protecting Brazil tropical forests (PPG-7 programme), national and international non-government organizations (NGOs), and federal and state programmes and policies . While many initiatives are hardly different from previous efforts, looking at them only in terms of success or failure would not provide a pertinent assessment of their long-term impacts. The failure to meet pre-established goals, often unrealistic and out of reach, does not mean that these initiatives did not have a deep local impact, modifying lifestyles and local perspectives, and shaping future change.
One of the most complex features in evaluating the impacts of development policies and projects in a region undergoing fast transformation such as the Amazon is the difficulty of isolating the role of social, economic and environmental factors working at different levels. From regional to local levels, the processes and outcomes of policies and projects are shaped by complex interactions between present and past conditions, and social dynamics inherent to different levels and areas . Furthermore, policies and projects tend to have ripple effects beyond the sphere that they are designed to affect. The impacts of policy changes and development projects in the Brazilian Amazon during the past two decades are good examples of these dynamics. Assessing the impacts of policies and projects requires approaches that are holistic and comparative, allowing the analysis of both specific and interrelated social transformations associated with the implementation of new instruments.
In a collaborative programme between French and Brazilian research institutions, the DURAMAZ1 project engaged in a multi-site evaluation of sustainable development initiatives aimed at providing a cross-disciplinary tool for multi-dimensional assessments of local conditions and for reviewing the impacts of such projects . A specific indicator system was developed with the goal of assessing the impact of policies and projects across four dimensions of sustainability, each capturing three levels of detail: life conditions, environmental conditions, governance and future perspectives. As described in the following, at the most disaggregated level, the indicator system includes 44 basic indicators assessed through a combination of intensive fieldwork and laboratory analysis. Additionally, a graphic/visual interface was developed to provide an effective communication tool aimed at policy-makers and stakeholders in general.
The system was applied to 12 cases in the Amazon region, covering a diverse array of situations, from an isolated indigenous village to small- and large-scale farming areas in Mato Grosso. Three principal findings arise from the analysis of the indicators and the cross-site comparisons they allow. The first is that the antinomy between the improvement of life conditions and the protection of the environment remains, despite the efforts of policies and projects to reconcile both dimensions. Local populations still face a trade-off between these two dimensions even if local decisions are not framed in such terms. Second, we see that policies produce diverse outcomes in different parts of the region. Geographical context and degree of accessibility seem to be important variables in this respect. Third, cultural factors such as sense of community and attachment to local traditions help us to explain (even ahead of economic factors) people's satisfaction regarding their situations and perspectives on what they consider sustainable development.
2. Providing a tool for cross-disciplinary and cross-scale assessment of the impacts of policy change in the Amazon: the DURAMAZ proposal
In the last 20 years, the general orientation of Brazilian policies in the Amazon has changed markedly from a focus on economic and geopolitical goals to one framed by ‘socioenvironmentalism’, an approach in which sociocultural diversity is seen as an effective tool for ecological conservation . This process has roots in regional social movements and their alliances with national and international organizations . Parallel to expanding development programmes and increasing deforestation rates , regional social movements and national/international mobilization put significant pressure on the federal government to find alternative pathways to reconcile economic growth with social inclusion and environmental conservation . A multitude of policies and projects were implemented, proposing integrative responses aimed at improving the quality of life without negative environmental outcomes, and vice versa. Elusive concepts such as sustainability and governance shaped policies and projects, in discourse and in practice, throughout the region .
As a way to turn the new goal of sustainable development into reality, multiple bilateral cooperation agreements have financed projects and experiences in the Amazon. The German GTZ was the most active agency, but others such as USAID (USA), DFID (UK), NORAD (Norway) and FFEM (France) have also been important partners in this process. A showcase of multi-lateral cooperation was the PPG-7 programme, launched in 1994  and lasting for more than 15 years, which financed about 150 small-scale sustainable projects and was recognized as one of the main promoters of sustainability initiatives in the region. It was followed by other large programmes: either national, such as incentives for small-scale agriculture (e.g. PRONAF) and the pioneer Brazilian programme for environmental services (PROAMBIENTE; see Hall ), or multi-lateral, externally funded programmes, such as a recent Amazonian fund supported by Norway. International NGOs also contributed  by financing projects directly and/or acting in cooperation with regional and local NGOs or the federal government (such as the World Wildlife Fund in the ‘protected areas of the Amazon’ programme). Although less visible, the private sector has initiated various attempts at company–community partnerships, with equally diverse outcomes [13,14].
In a vast majority of projects, following international interests on community-based, integrated conservation and development projects , ‘communities’ or the ‘local level’ has been elected as the most appropriate scale for implementation. Many of these projects are based on assumptions about the efficiency of collective resource management . Frequently, projects have been designed as ‘pilots’, leading to the understanding that in cases of success, either local people would reproduce them spontaneously or the federal government would transform them into new public policies. Rarely has this happened .
A sectoral evaluative approach of these changes, in which the policies and projects would be evaluated only against their objectives, would not take into account their multiple consequences. A holistic view of local conditions and their context is needed to provide an evaluation of the multi-dimensional, long-term impacts of public policies and development projects. Local stakeholders are also constrained in their evaluation of these initiatives. It is common, and perhaps expected, that stakeholders have a limited view of the situation, constrained by their immediate needs and interests as well as limited availability of information and expertise to evaluate potential outcomes of new policies and projects. It is in this context that the DURAMAZ project focused on assessing these experiences at the local level, i.e. the scale defining the social group affected by a project and their immediate landscape. At this scale, our goal was twofold: (i) to assess on the ground and from space local social and environmental conditions, governance and future perspectives, and (ii) to examine the concrete effects of projects and understand how people define their strategies as a function of the social, cultural and economic contexts in which they live. Building a multi-dimensional and comparative indicator system represented an important part of this strategy.
(a) The DURAMAZ indicator system
(i) Reviewing existing systems
Many attempts have been made to quantify or measure sustainable development through indicator systems. We define an indicator as a significant element that permits an evaluation of a broader ensemble. Nowadays, indicators are very frequently referred to as a reduced set of figures that sum up complex environmental, social or economic situations or dynamics. In general, indicator systems can be categorized into two main groups: some that reduce the information into one or a few figures and others that try to monitor several parameters at the same time, presenting them as a dashboard. The latter allows sharing of information in several themes, and presents a more complex and detailed view of the studied situation. Since the 1980s, several multi-dimensional systems, such as the human development index , have been used at the international level.
During the 1990s, new indicator systems were designed to take into account the links among social, economic and ecological factors. Some remained linked to economic variables, such as the ‘index of sustainable economic welfare’  or the ‘genuine progress indicator’ , whereas others aimed at standardizing the impact of behaviour in terms of land units, such as the ecological footprint indicator  and the ‘living planet index’ . In parallel, international efforts were made to assemble systems that could provide parameters to evaluate new objectives set by the United Nations, such as Agenda 21. One of the results was the ‘dashboard of sustainability’ of the United Nations Commission on Sustainable Development (UNCSD), which was promoted in two stages. Introduced in 2001 and revised in 2007, it was particularly innovative in proposing a series of multi-dimensional themes (such as natural hazards, governance and global partnership) instead of focusing on the famous three pillars of sustainability (economy, environment, society).
Such holistic systems are not easily deployed. Most of them rely on figures computed at a national scale, which are not always reliable, as the heavy criticism encountered by the ‘environmental sustainability index’ has shown . They also incorporate a great number of variables (e.g. 130 subvariables aggregated into 46, as in the UNCSD ‘dashboard of sustainability’), which severely limits them as readable communication tools. Many of them also fail to effectively link different dimensions of sustainable development by using a mere juxtaposition of the indicators relative to social, economic and environmental aspects. One attempt to overcome these limitations was the French national indicator system for sustainable development , which proposed a dashboard approach, including thematic modules that took into account the interactions among the four dimensions.
After the 2002 Johannesburg world summit on sustainable development, cultural diversity became an important part of this agenda. New systems integrating this dimension have readily appeared, such as the bio-cultural index  and the FAO system for indigenous peoples . Other systems tried to focus on assessing linkages between objective and subjective dimensions of well-being of a given population, such as the happy planet index , which quantifies the number of ‘healthy and happy’ life years obtained at the cost of the use of natural resources.
Besides these global systems, there were many attempts at regional indices, adapted to the environmental and cultural particularities of a given region. Several innovative efforts were proposed for the Amazon region such as the integrated index of sustainability built by Lobato Ribeiro (2002), the Amazonian Cooperation Treaty Organization's system2 and the PPG-7 projects evaluation system .
(ii) The DURAMAZ design
The systems described above did not match the specific needs of the DURAMAZ project—a system to be used at local scale adapted to assess and compare various social and environmental dimensions in multiple contexts and populations. For this reason, we engaged in building our own, adopting a ‘fitness-for-purpose’ contextualized approach . Table 1 describes the different dimensions and two levels of aggregation of the DURAMAZ indicator system (detailed information about data and calculations can be found in the electronic supplementary material).
The question of data availability is, in general, crucial in defining the composition of indicators. Because specific fieldwork was scheduled for each of our study sites,3 we were in a position to gather primary data for almost every indicator and to generate others from laboratory analysis of remotely sensed data. The principles of the Bellagio conference on sustainable development indicators  were taken into account, as the design was to be multi-dimensional and as we chose to incorporate a demographic component in order to assess current population structure and intergenerational changes.
The resulting DURAMAZ indicator system [4,31] is composed of 44 subindicators associated with several field data collection instruments ranging from structured and unstructured questionnaires,4 life history and migration assessments, and archival research to field assessments and GPS point collection, GIS analysis, and remote sensing image processing (see the electronic supplementary material, figure S1). The system is structured in four dimensions inspired by the UNCSD approach, mixing elements from the three classical pillars (economy, environment, society) and from the fourth dimension of ‘governance’, which is increasingly pointed out as crucial.5 However, our holistic approach did not mesh well with the idea of four independent dimensions, because many elements are intrinsically linked. We thus chose to elect four dimensions that thematically overlap. Our first dimension, ‘life conditions’, integrates economic (e.g. income) and social subindicators (health conditions, gender issues, etc.). Our second dimension, ‘environmental conditions’, relies on ecological (e.g. biodiversity, forest cover) and social (e.g. use of forest resources, perception of environmental change) data. The third dimension of our system, ‘present needs and future perspectives’, is based on demographics and satisfaction report data. The fourth dimension, ‘governance and local social interactions’, combines data on social (e.g. participation in social networks) and institutional aspects (e.g. partnerships). It includes the measuring of local satisfaction as an important element of the quality of life.
The DURAMAZ indicator system proposes two different levels of aggregation of the 44 base indicators (table 1). One, aimed at local stakeholders and policy-makers, provides concrete information into what needs to be done in order to transform the current situation of a site. It consists of an aggregation of the 44 indicators into 14 variables, each covering a specific thematic area linked with sustainability. The other level of aggregation shows performance in each of our four dimensions of sustainability. It was designed for researchers interested in comparing performances and profiles at the site level and between sites. Both products are associated with graphic presentations, similar to the representation proposed by Foley et al. .
Obviously, as Munda  points out, the methods used to calculate the indicators can have an important influence on the outcome and the global interpretation of the resulting set. Following a standardized procedure of data collection, each indicator was given a score from 0 to 10, the value of which was computed as a function of values taken from the pertinent literature and known reference parameters valid for the Amazon region. Transparency and simplicity were the reasons why we chose not to weight the values when aggregating the subindicators, in an attempt to avoid difficulties in the dialogue between stakeholders and researchers . Also, we did not produce a unique synthetic index in order to prevent any substitutability between the four modules of the DURAMAZ indicator system.
3. Sites studied by the DURAMAZ project
The DURAMAZ project selected a sample of 12 sites where sustainable development projects had been implemented. This sample was constituted by taking into account several factors, with the inclusion of various situations representative of the Amazon region being an important one. In addition, we considered different financial schemes (i.e. from projects funded individually by international NGOs to projects funded by federal or state programmes), population types (i.e. from forest people to large-scale farmers) and geographical location (i.e. from isolated regions to well-connected areas). Table 2 describes the sites and their main characteristics. Figure 1 locates these sites in the Amazon region, while the electronic supplementary material gives more detail about each of them. The sample of sites was limited to 12 owing to funding limitations. With such a small number, it is evident that our conclusions are not statistically representative at the scale of the whole region. At the same time, given the importance of the fieldwork done in each site, the DURAMAZ project sample of sites is bigger and more diverse than most of the projects working in the Amazon and its conclusions can be considered as relevant.
In each site, interviews were conducted under a sampling scheme. When the site was with less than 75 families, every household was surveyed. Beyond 75 families, a sampling was realized in order to conduct 75 interviews. This sampling was random, but the survey team was instructed to verify that all parts of the studied area were represented in the sample. In each household, the chief of the family was interviewed. The project ended with 900 interviews and about 1250 cumulated days of fieldwork across all sites.
(a) The four-dimensions model: cross-site comparisons and trend analysis
The four-dimension aggregation level (figure 2) shows several trends. First, we see that life conditions in the extractive and indigenous areas are generally worse than at the agricultural frontier, but that the environmental conditions are better oriented. Small-scale agriculture sites tend to have better life conditions than extractive and indigenous areas. Also, the governance dimension appears to be the least-developed dimension in comparative terms, configuring a problem in nearly every site. However, the situation is very contrasted in this respect, demonstrating that the influence of social organization goes beyond social and geographical differences. For instance, although very different from each other, the site in Mato Grosso representing large-scale farmers (Sorriso Vivo) and the extractivist community of Iratapuru show very similar performances in terms of governance. ‘Life conditions’ and ‘present needs and future perspectives’, on the other hand, tend to have opposite trends across sites. Small-scale agriculture and agribusiness sites have better indicators of ‘life conditions’, but at the expense of ‘present needs and future perspectives’, whereas in the indigenous or extractive context the situation is the opposite.
The four-dimensional system allows us to explore possible groupings between the studied sites6 by comparing their profiles. Two groups and two atypical profiles appear. In the first group (upper row in figure 2), dimensions 2 (environmental conditions) and 3 (present and future needs) perform better than the other two dimensions. It is composed of sites with preserved forest cover and includes mainly extractive or indigenous sites. The other group (second row in figure 2) is composed of frontier areas, in which dimensions 1 (life conditions) and 4 (governance) are better oriented than the other two dimensions. Some sites exhibit a typical profile, such as the agribusiness site in Mato Grosso, in which the life conditions indicators outperform the other three dimensions—a clear reflection of the prosperity brought by agribusiness at the expense of other considerations.
(b) The 14-dimensions model: from an analytical system to local ownership of indicators
The 14-indicators aggregation provides a more detailed and differentiated view of each site (figure 3). It is better suited for performance assessment and policy-making recommendations. Our intention here is not to compare the sites with one another, but to evaluate the impacts of the sustainable development projects implemented in each site. This level of aggregation exposes the situations of different communities in relation to thematic areas. It allows us to easily assess what are the weakest ones and where effort should be concentrated. A majority of the sustainable development projects had, for instance, the objective of improving life conditions. As one can see in figure 3, the first four sectors of each site drawing (starting at the top and going clockwise) are, in general, little developed, showing that this objective was not achieved in most cases.
Each site has its own difficulties, but we see common problems across sites in terms of access to energy, tap water, healthcare and education. Even the agribusiness site in Mato Grosso, which was expected to have good results in the function of its economic prosperity, has many weaknesses in terms of environmental variables (7, 8, 9), intergenerational prospects (12) and governance (14).
It is expected that these results will foster discussions between local stakeholders and institutions, leading to a better orientation of future projects. For instance, RDS Mamirauá seems to present a specific gender issue, given that the ‘gender equality dimension’ is very poorly developed, even compared with other extractive and traditional areas, such as PAE Chico Mendes or RDS Iratapuru. This problem, also underpinned in specific social science works,7 probably needs to be addressed specifically.
This 14-dimensions system was presented to local stakeholders (recognized or designated leaders representing each site) during a 2-day meeting in Brasília. This quasi-experimental interaction was very informative for the participants, because printed copies of figure 3 were distributed. We could see that they immediately understood the purpose of the model and started to compare results, either between dimensions inside their own sites or between sites. Moreover, they immediately started to establish links between local programmes or initiatives (or the lack of such) with observed results. For instance, a representative from the small-scale agriculture site PDS Esperança pointed out that the absence of a school inside the settlement was the obvious reason for the low mark in dimension 7 (education). The representative immediately stated that this figure should be presented to the municipality's mayor, as it was an obvious plea for the immediate installation of a local school. The representative of the agribusiness site pointed out that despite its reputation as a highly deforested area, its results in dimension 8 (environmental dynamics) were not so bad and even better than the RESEX site of Ciriaco, interpreting this as a demonstration of the new commitment of large-scale farmers to ethical behaviour towards the protection of legally protected areas and to the development of services such as better waste management.
5. Social–environmental trade-offs and local expectations of development
Indicators have been analysed using Pearson's correlation method and principal component analysis, using both the four-dimensions and 14-dimensions aggregations.8 From the results of both analytical approaches, a number of recurrent features appeared, which represent both important findings and key factors related to sustainable development in the Amazon.
(a) The ‘Gordian knot’ of the Amazon remains: facing a trade-off between environmental preservation and improvement in life conditions
The analysis of the indicators shows that there is a recurrent opposition between improvement in life conditions and environmental preservation, and that accessibility stands out as the most relevant variable explaining this trade-off. Regardless of the region of the Amazon in which sites are located, the system tends to show that the typical opposition is between well-connected sites having relatively good life conditions but poorer environmental conditions and isolated sites having a preserved environment but adverse life conditions.9 This opposition does not seem to be a result of environmental policies, as it appears at a general level and when comparing sites subject to the same environmental and agricultural incentive policies, respectively, ‘extractive’ and ‘small-scale agriculture’ sites (each representing five sites).
What the analysis seems to point out is that despite the reorientation of public policies in the Amazon towards socioenvironmentalism, communities continue to face the old dilemma of either enjoying a preserved ecosystem but enduring adverse life conditions, or enjoying better living at the expense of forest cover. This conclusion tends to indicate a failure of both global policies and the small-scale initiatives represented by the projects evaluated by DURAMAZ. No site suggested a solution to untie this Amazonian ‘Gordian knot’.
At the same time, the terms of this trade-off are not necessarily clear for local stakeholders, as qualitative interviews on the topic of sustainable development clearly showed. This is where the 14-dimensions model proved very useful. The indicator system is an efficient way to pinpoint this mechanism and to present it to the communities so that they can identify that, for the moment, they face such a trade-off between environmental and life conditions. They could then adopt a position on the issue, determining up to what point they accept the trade-off and consider ways of monitoring the effects through this indicator system.
(b) The same policies did not produce the same effects across the region
Aside from the general trend observed in the previous paragraph, contrasted tendencies may also be observed. As each of our sites reflects the impact of broader regional policies (from the creation of protected areas to the expansion of agricultural settlements), we can, for instance, observe their different outcomes. Using k-means and hierarchical clustering on the four- and 14-indicators' outputs, we have thus explored possible groupings of the sites. What we found was that there was some consistency between the a priori groups based on sociocultural and economic characteristics and those resulting from cluster analysis (e.g. small-scale agriculture sites). At the same time, some sites were always grouped outside their a priori group. For instance, the small-scale agriculture site PDS Esperança was always grouped with extractive and indigenous sites, whereas the extractive economy site RESEX Ciriaco was repeatedly associated with small-scale agriculture ones. The clustering processes seem to emphasize the importance of the regional context in the profile of each site, regardless of the policies and projects to which they were subjected. This finding is important because it begs a reorientation of sustainable development projects and policies. Instead of reasoning about general regional categories, they should favour an intraregional approach to avoid applying panacea tools  to a reality that does not correspond to a perceived average condition at a regional level (figure 4).
(c) Cultural factors remain very important
If the main explicative factors for the overall sites' conditions seem to reside in the antinomy environment/economy as it relates to degree of accessibility, then interesting correlations between other indicators of our set appear. One of them is that the satisfaction of people regarding their site's situation (indicator 11b in table 1) and the will of young people to stay (vis-à-vis their desire to seek a better life in another place; indicator 12a) are not influenced by economic indicators such as life conditions (indicator 3a, for instance) or their parents' income (indicator 1a). We interpret this fact as a demonstration that there is a strong cultural component in terms of satisfaction and sense of place. This is confirmed by the fact that indigenous and extractive areas have far better marks in the first two indicators cited (meaning more satisfaction and less disposition for youngsters to leave) than other sites.
Also, there is no apparent correlation between the position of young people about staying or leaving (indicator 12a) with the availability of schooling programmes (indicator 7a). This is a provocative finding because many observers have concluded that an improved level of education would turn the interest of young people towards urban areas in an effort to find jobs corresponding to their qualifications. Again, the cultural dimension of belonging to an area and a community seems here more explicative than other factors. In general, young people will be prone to migrating more frequently when they come from areas where people have already experienced long-term migration, as we observed in most small-scale agricultural areas. Such behaviour is less common for people coming from areas with a deeper cultural background, such as extractive or indigenous sites.
6. Concluding remarks
The indicator system elaborated by the DURAMAZ project has been effective insofar as it allowed our group to study and compare 12 local situations, identifying a number of repetitive patterns among them, which led us to some key findings. The main one is that besides their hopes and claims, small-scale development projects seem to have failed to untie the ‘Gordian knot’ of the Amazon, which is to improve life conditions while at the same time conserving the initial condition of the environment. But the project has also brought one additional conclusion. In each and every site, the situation encountered was much more complex than expected, with mixed influences from the application of several public policies (health, education, etc.), the influence of the overall economic context, and the role played by the development initiatives funded by diverse types of institutions, from local NGOs to international cooperation.
The complexity of social–environmental interactions continues to expand in the Amazon. Public policies related to infrastructure (e.g. electric energy) and cash transfer programmes (e.g. different family aid programmes and extending retirement benefits) are reaching deep into the Amazon and shaping the region in new ways. The institutional and analytical success of the first phase of the project allowed us to launch a new phase—DURAMAZ 2—focusing on developing an observatory system based on longitudinal analysis (2003–2015). Recollecting and recalculating indicators in these sites since 2003 will allow us to evaluate the impacts of the public policies and the long-term effects of local development projects. For this phase, we have added a new dimension of institutional analysis as well as modules related to the analysis of local dimensions of climate change and scenario building. We continue to expand our sample, to refine our methodologies, and to reach out to new partners with the goal of transforming the initiative into a network of observatories of sustainable development in the Amazon.
One contribution of 18 to a Theme Issue ‘Ecology, economy and management of an agroindustrial frontier landscape in the southeast Amazon’.
↵1 DURAMAZ is an acronym in French for sustainable development in the Amazon.
↵2 The Amazon Cooperation Treaty was signed in 1978 by Bolivia, Brazil, Colombia, Ecuador, Guyana, Peru, Suriname and Venezuela. Its objectives were to promote joint actions towards the harmonious development of the Amazon Basin. In 1995, the members of the treaty decided to create a permanent organization, which was installed in 2002 in Brasilia (Brazil). As part of its tasks, this organization did elaborate an indicator system for the Amazon region.
↵3 A special fieldwork guide was elaborated to ensure the consistency of all the project teams working in the field.
↵4 The fieldwork protocol included three questionnaires: one for the households (socioeconomic questionnaire), one for the individuals (life history and events questionnaire) and one for the local institutions (governmental bodies, NGOs, associations … ). All questionnaires had multiple-choice questions and open-end questions such as ‘How do you understand sustainable development?’ All households of a given site were surveyed when the total number of families was less than 80; the sampling of larger sites was defined on the basis of parameters that would ensure representativeness. More than 900 households were surveyed in the 12 sites during more than 1200 days in the field.
↵5 See Tables S1–S4 in electronic supplementary material.
↵6 The studied groupings were also confirmed by statistical analysis of the indicators, like hierarchical or k-means clustering.
↵7 See Nelissa Peralta, ‘Ecoturismo e Mudança Social na Amazônia Rural: efeitossobre o papel da mulher e as relações de gênero’, http://www.academia.edu/739678/Ecoturismo_e_Mudanca_Social_na_Amazonia_Rural_efeitos_sobre_o_papel_da_mulher_e_as_relacoes_de_genero.
↵8 Some of the 44 indicators have also been individually tested against the 14 aggregated ones.
↵9 Of course, the repartition of those sites is somewhat dependent on the geography, because isolated places are more frequent in the centre and northwestern part of the Amazon than along the southern border around Mato Grosso or Southern Pará. But even at the heart of the frontier, there are still numerous areas with difficult access.
- © 2013 The Author(s) Published by the Royal Society. All rights reserved.