INTRODUCTION

The different ways in which economic development influences the environmental characteristics of a given location have been discussed both by the bias of the impact of the evolution of the economy on the environment (Grossmasn and Krueger 1991, 1995, Caviglia-Harris et al. 2009, Oliveira et al. 2011, Tritsch and Arvor 2016, Luo et al. 2017, Tsiantikoudis et al. 2019, Santiago and do Couto 2020, Yasin et al. 2020) and by the bias of the limitations imposed by the environment for economic development (Forster 1973, Gruver 1976). For the Amazon region, the evidence that there are different patterns of relationships between deforestation and the socioeconomic development profile (Margulis 2003, Prates 2008, Rodrigues et al. 2009, Andersen and Reis 2015, Caviglia-Harris et al. 2016, Tritsch and Arvor 2016, Sathler et al. 2018) arouses interest to explore and describe this relationship. Understanding the possible connections between different development pathways and deforestation patterns is particularly important when highlighting models that start from a regional perspective and prioritize the territorialities and the preservation of the resources in a forest biome.

The environmental Kuznets curve (EKC) stands out as one initiative to explore the influence of economics on environmental outcomes. Described for the first time in the early 1990s (Grossman and Krueger 1991), EKC is algebraically represented in an econometric model that aims to identify theoretically predicted relationships between economic development and environmental degradation (Grossman and Krueger 1991, 1995, Shafik and Bandyopadhyay 1992, Selden and Song 1994). In short, developing regions, in which economic activities rely on predatory exploitation of natural resources, are prone to obtaining environmental degradation as a response. However, as the region improves its overall socioeconomic conditions and rethinks its technological patterns, combined with law enforcement, economic and environmental policies, the relationship between the variables changes and the economic pressure on natural resources tends to decrease. Therefore, EKC theory corresponds to this evolutionary process and characterizes the trajectory of environmental impacts, usually taking per capita income as an economic development indicator.

In this context, EKC has been used considering several environmental indicators, different sets of complementary variables, and space-time cutouts. Grossman and Krueger (1991), precursors to EKC, explored the relationships between air quality indicators and economic development to show the influence of changes in trade barriers on environmental aspects. They observed that the levels of two main pollutants (sulfur dioxide and smoke) increased with income in countries with low levels of gross domestic product (GDP) per capita and tended to decrease in countries with a more consolidated economy and higher GDP values. In 1995, the same authors studied the possible relationships between economic development, analyzed from the GDP per capita values and a composite variable of environmental quality, covering analyses of air and water pollution in 42 countries in different regions of the globe. The authors showed that, for most environmental indicators, the result of the EKC model was established in an inverted “U” shape, presenting an initial phase of economic growth conciliated with increased pollution, followed by a phase of pollution reduction.

In the Brazilian context, Santiago and do Couto (2020) sought to analyze the relationship between deforestation and socioeconomic development for municipalities and observed the existence of a strong relationship between socioeconomic development and the use of forest resources, indicating the possibility of an inverted “U” shape relationship for this territory and analysis unit. However, the authors highlighted the existence of other factors responsible for deforestation, such as the effect of economic policies, legislation, and cultural aspects. For the Amazon region, Tritsch and Arvor (2016) observed the relationship between socioeconomic development and deforestation and confirmed the original hypothesis of an inverted “U”, emphasizing the precarious socioeconomic situation of areas still with active deforestation.

Eventually, when investigating EKC relationships, one must consider the spatial heterogeneity of variables. Using geographically weighted regressions (GWR), Tanaka and Matsuoka (2007) investigated the relationship between variables of socioeconomic development and pollution from industrial sources. The GWR method was able to respond with better adjustments in the model, concluding on the importance of considering spatial heterogeneity in the analysis of variables of a socioeconomic and environmental nature. Adopting a similar proposal, Oliveira et al. (2011) used the GWR method to investigate the relationship between deforestation and economic development in the Amazon in light of the EKC. Their results showed different relationships between deforestation and municipal GDP per capita, as well as different relationships between deforestation and other explanatory variables, confirming the hypothesis of variability for the modeled phenomena. Likewise, Kim et al. (2018) confirmed the inverted “U” shape for some regions in 22 provinces of China, testing EKC’s original relationship for economic development, mediated by income, and the emissions of three industrial pollutants (SO₂, residual water, and solid waste).

The EKC approach is also useful for exploring the process of urbanization and its environmental relationships. At a global level, Katircioğlu and Katircioğlu (2018) included dimensions of the urbanization process in the econometric model. They explored the relationship of carbon dioxide (CO₂) emissions in the function of economic development, energy consumption, and urban growth, understood as the evolution of the population living in urban areas. Environmental Kuznets curve’s relationship to urban growth and CO₂ emissions occurs in a decreasing monotonic shape, proving that the population concentration in urban areas, by itself, is not a decisive factor for the increase in greenhouse gas emissions. In the same way, Yasin et al. (2020) analyzed the impact of financial development, increase of urban population, commercial opening, presence of political institutions, and energy consumption in the ecological footprint of 110 countries. The authors confirmed the original hypothesis of EKC and observed the reduction of the ecological footprint with the increase in urbanization, reinforcing its positive environmental effect.

Despite its numerous applications, it is recognized that there are criticisms and limitations associated with the EKC conceptual-methodological proposal. As described by Stern et al. (1996), a limiting issue of this concept is to only consider the unilateral interaction of the effects of the economy on the environment. Environmental feedback, i.e., how the quality of the environment affects the economy, is not considered. In addition, the methodological assumption that income follows a normal distribution is also highlighted, which rarely applies to the units of analysis explored in research on this theme. The consequence being that these studies can estimate an average level of income, from which environmental degradation starts to decrease, which is not representative of reality (Kaika and Zervas 2013). Still, Stern et al. (1996) also discuss the limitations regarding the availability, regularity, and quality of environmental data, without which different results are generated for related studies. In addition to these technical and specific issues, there are limitations and disagreements associated with the EKC’s general hypothesis. In the current economic model, based on industrial capitalism, the perspective of continuous economic growth demands crescent exploitation of natural resources. This perspective conflicts with the proposal of environmental degradation reduction. Thereafter, ceasing the exploitation of natural resources or adopting sustainable production methods, in the first instance, would have an impact on economic growth. Thus, it is necessary to be cautious in interpreting the results, considering the limitations imposed by the conceptual model itself and the semantics and construction of the input data representing the analyzed processes.

In the Brazilian Amazon, an area of tropical forest with approximately five million km² where more than 70% of the population resides in officially urban areas (IBGE 2010a), to understand the urbanization-nature relationship we must consider that the presence and interaction with new agricultural and industrial activities have reorganized the space, historically, in its structure and socioeconomic relations. In the region, even today, the process of genesis and evolution of urban space is centered on cities and depends on urban-industrial dynamics. This urbanization replicates production systems exogenous to the Amazon, whose economic and behavioral patterns extend beyond cities. Spreading throughout the territory, these patterns overlap all the other human settlement logics immersed in the biome as a whole (Monte-Mór 1994, Lefebvre 1999). Additionally, the regional Amazon economy, which flows around the entire territory of the municipalities, is mediated by the cities and has, in this sense, an important extended urban base (Becker 2013).

The historical-geographic context of the occupation of the Amazonian territory highlights the importance of understanding the urbanization process to discuss the influence of human settlements on deforestation projections and, consequently, in the resumption of a development logic based on endogeneity, with the final objective being that of preserving the forest and the socio-cultural expressions of people who live in it. In this sense, Ribeiro et al. (2018) demonstrated that interactions between the evolution of urbanization and deforestation are not straightforward in the Brazilian Amazon. There are several agents and factors present in the territory, acting at different levels and scales, that indirectly influence deforestation and urbanization. This result points out that there are complexities and heterogeneities underlying the processes analyzed. Other studies help support these findings and show that urbanization has been seen as a process resulting from socioeconomic factors, and it is related to the evolution of deforestation in different ways (Molotch 1976, Becker 2013, Cardoso and Negrão 2013, Richards and Vanwey 2015, Tritsch and Torneau 2016).

Cities, as spaces for action and articulation between public and private agents (Moloth 1976), play an important role both in the occupation process of the territory as well as in the evolution of the regional socioeconomic development profile (Jacobs 1969). To understand the influence of different socioeconomic development profiles on environmental aspects inside the Amazon Forest biome, we no longer should think about “cities in the forest” (Castro 2009). But rather, we should go back to the idea of “cities of the forest”, which was defined, in these terms, first by Browder and Godfrey (2006) and reinterpreted by Becker and Stenner (2008). For the latter authors, the "cities of the forest" are an alternative that combines traditional knowledge and technologies with new scientific and technical advances, looking for alternatives for development, capable of promoting the environmental responsibilities related to the use of forest resources. Cities are the settlements that are the basis for action, and they can take advantage of this potential for a harmonious relationship between economic development strategies and traditional knowledge of the native population.

We seek to identify the relationship between urbanization and deforestation in the Brazilian Amazon, in light of the environmental Kuznets curve (EKC) theory. The main relationships are summarized by a general typology, considering municipalities as the unit of analysis and the evolution of a set of socioeconomic variables and deforestation data as the metrics associated with urbanization and environmental degradation. The Pará state was chosen as the pilot area for this study. The last two demographic censuses, in 2000 and 2010, defined the time interval for analyzing the interactions between changes in deforestation and the overall changes in socioeconomic variables at the municipality level for the study period. The EKC approach included spatial analysis methods to also consider the spatial heterogeneity in the variables’ relationships, adopting the proposal of Oliveira et al. (2011) as one of the main bases for methodological delimitation. It is noteworthy that the spatial approach is fundamental to understanding the possible interactions between socioeconomic, demographic, and environmental variables in the Amazon, a region with important intra-regional differences (Rindfuss and Stern 1998, Sathler et al. 2018).

However, it is recognized that the limitations of the EKC conceptual-methodological proposal discussed in the literature must be considered to ensure the transparency of the process and the integrity of the results. To get around the main problems previously highlighted, in addition to the income variable, a set of socio-demographic and economic variables derived from the census data was taken into the model. In addition, the local spatial analysis method aims to generate spatialized and individualized results. In this context, the average of the income variable is associated only with the borderline neighborhood, minimizing errors associated with the overestimation of the income turning point. Thus, this work contributes to the discussion about the influence of different strategies for regional economic development that could support “sustainability.” In the context of Amazon cities, the term sustainability is understood as the capacity for social and economic evolution dissociated from predatory uses of forest resources. The complexity between possible and probable development paths for the future of the Amazon is discussed, emphasizing the importance of thinking and acting from a regional perspective, highlighting the potential differences, in the context of this analysis, between “cities in the forest” and “cities of the forest."

METHODOLOGY

Study area

The study area comprises all 144 municipalities in the state of Pará (Fig. 1). The complexity of socioeconomic characteristics, population dynamics, and deforestation rates makes Pará an adequate object of analysis to explore the EKC relationship patterns and the different types of cities’ socioeconomic evolution. Pará is one of the nine states of the Brazilian Legal Amazon. The mosaic of land uses and land cover in Pará resulted from a multiplicity of actors and processes along with its occupation history. In the course of this history, endogenous and exogenous economic initiatives acted as the underlying driving forces of the processes of change in the forest landscape.

Until the 19th century, the process of the formation of the Amazon urban space was associated with building strategic bases to exercise control over the territory and its people, without interest in settlement and regional development. Then, in the 20th century, to favor the expansion of the agricultural frontier, the focus was on the attraction of migratory flow and organization of the labor market, culminating in tax incentives to attract companies and infrastructure networks that favored the urbanization of the population and the territory around the main cities. Later, in the 21st century, these city-centered spaces acted as a political-economic base in the industrialization process (Corrêa 1987, Becker 2013), constituting a hybrid urban reality that, although it concentrated political power in municipal cities, it still preserved layers of other ways of life and economic integration modes hidden, but alive, in the forest (Cardoso et al. 2016).

From 2000 to 2010, the population in Pará grew at an average annual rate of 2.04% (IBGE 2000, 2010a), while in Brazil, the average annual rate was 1.17% (IBGE 2000, 2010a). This fast population growth is also reflected in the urban population. Among the states of the Legal Amazon, Pará presented the highest urban population growth rate in the last decades: 16.02% between 1991 and 2010. Aside from the intense population dynamics, Pará has had the highest deforestation rates in the Brazilian Amazon since 2006. Estimates showed that more than 148 thousand km² have been deforested since 1988, which is equivalent to 34% of the deforestation recorded for the entire Legal Amazon (INPE 2020). In general, the expansion of the agricultural frontier and the intense forest loss occurred mainly where the arc of deforestation advances over the mesoregions of the southeast and southwest of Pará (IBGE 2005, INPE 2020).

Pará is a state with a complex economy. The economic activities are concentrated in the mining and timber extraction sectors, services, and agriculture; mainly with expansion in agriculture due to soy production (FAPESPA 2016). In 2015, the GDP value of the state represented a contribution of 2.2% of the national GDP and 40.8% of the GDP of the North region (FAPESPA 2017). However, even with this expressive contribution to the regional economy, Pará had the lowest human development index (HDI) in the region (0.646 in 2010), being the seventh-worst index value in the country (UNDP 2010). These characteristics and ambiguities, added to the historical-geographic context, highlight the state of Pará as an important object of study for analyzing the relationships between current development models and their possible consequences on the dynamics of forest loss.

Materials and methods

Originally, the EKC model included the data per capita income, considering its quadratic term. However, some studies explore other shapes of the curve in addition to the traditional inverted “U” using the cubic and logarithmic forms of the data (Gomes and Braga 2008, Carvalho and Almeida 2010, Tsianikoudis et al. 2019). Because the study area is located in a forest biome, in this analysis, we have taken deforestation as a measure of environmental degradation because it represents loss, at different levels, of natural capital. The general specification applied also considers its cubic shape, so that other EKC formats can be explored, in addition to the original inverted “U” (Equation 1):

(1)

where Deforestation is the variable of interest (Y), INC is the per capita income, z is a vector that includes the other explanatory socioeconomic and environmental variables, ε is the term of random error, and α and βi are the regression parameters. The βi coefficients determine the shape of the curve that relates deforestation to its development profile, for each municipality.

Considering the quadratic and cubic terms of income, six relationships, or shapes of the EKC curve, could be identified from the model:

1. increasing monotonic linear relationship: when β1 is higher than 0 and β2 and β3 are not significant (Fig. 2a);
2. decreasing monotonic linear relationship: when β1 is lower than 0 and β2 and β3 are not significant (Fig. 2b);
3. “U” shaped relationship: when β1 is lower than 0, β2 is higher than 0, and β3 is not significant (Fig. 2c);
4. traditional inverted “U” format: when β1 is higher or equal than 0, β2 is lower than 0, and β3 is not significant (Fig. 2d);
5. “N” format relationship: when β1 is higher or equal to 0, β2 is lower than 0, and β3 is higher than 0 (Fig. 2-e);
6. inverted “N” format: when β1 is lower than 0, β2 is higher than 0, and β3 is lower than 0 (Fig. 2f).

Based on the theoretical foundation that supports this work, it is expected that the result of the EKC presents the inverted "U" shape, as described in the original model (Grossman and Krueger 1991, 1995). Initially, the increase in deforestation would reflect economic activities related to forest conversions, such as conversion to agriculture and livestock activities. As the municipality develops its economy, its portfolio of socioeconomic indicators should evolve with its urbanization process, and a reduction in deforestation should be expected. However, deforestation decreases can happen in different situations, not all positive, like the depletion of the forest resource, command and control policies, environmentally aware economic strategies associated with local environmentally aware populations, and environmentally aware urban policies (Selden and Song 1994, Bretschger and Smulders 2000, Becker 2009).

The EKC proposal originally reinforces the rationality of natural resources overexploitation when it suggests that the market alone can induce the necessary measures for social inclusion and environmental sustainability. However, in this work, we propose expanding the analysis of regional development beyond the income-only economic aspect. We analyze the relationship between environmental degradation and development not only by the growth of per capita income but also by a set of variables that represent different axes of territorial development. These axes represent characteristic elements of urban transformations through their infrastructure and/or services.

Therefore, we defined a set of additional explanatory variables (z; Equation 1), selected based on the literature (Moran 1993, Rodrigues et al. 2009, Guedes et al. 2012, Becker 2013, Galego-Alvarez et al. 2014, Caviglia-Harris et al. 2016, Tritsch and Arvor 2016, Sathler et al. 2018). Altogether, there were 22 initial independent variables in the exploratory analysis, comprising the following dimensions: education, health, income, labor, human development, inequality, violence, and basic services (Table 1). These variables were chosen from the compilation of work proposals whose main axis of discussion is the relationship between deforestation and socioeconomic development in the Amazon.

The values for these variables, associated with the municipality, come from census surveys and other public demographic data collection platforms. The dependent variable, deforestation, corresponds to the total accumulated deforestation by municipality up to 2000 and 2010, and it comes from the Satellite Program for Monitoring the Deforestation of the Brazilian Amazon Forest (PRODES; INPE 2015). The data availability for the socioeconomic variables used in the characterization at the municipality scale restricted the analysis to the 2000 and 2010 periods. The value difference for the period (2000-2010) was computed for each variable and its expected relationship with the difference in accumulated deforestation (2010-2010 in km²) is described in Table 2.

The analysis of the EKC relied on the results of the model expressed in a local spatial regression method, the GWR (Brunsdon et al. 1996, Fotheringham et al. 2002, Charlton et al. 2009). The GWR allows the model to estimate the parameters of localized linear regressions and it is adequate for evaluating a local model of the variable or process being investigated considering its spatial nonstationarity condition. The model estimates linear regressions for each spatial unit, i.e., the municipality, in this case, using a subsample of data based on an established neighborhood rule. In this sense, the GWR provides a set of estimates that fits local characteristics, which means, they are in accordance with the municipalities and the conditions of their surroundings, captured by the model’s neighborhood rule (Equation 2):

(2)

where (u_i,v_i) represents the coordinates of point i in space, and β_k (u_i,v_i) refers to the value that the continuous function β_k (u,v) assumes at point i.

A set of exploratory analysis of spatial data (AEDE) techniques was used to test for the presence of multicollinearity, the assumptions of linearity, and the presence of spatial dependence. From the collinearity analysis, the model was established considering the combination that reached a value of variance inflation factor (VIF; Berk 1977) less than 10 (Tamhane and Dunlop 2000, Kutner et al. 2004, Miles 2014) to ensure that there are no severe multicollinearity issues in the dataset.

For the general specifications of the GWR, a traditional Gaussian model was considered. The GWR method works with three types of regression models: continuous, binary, and count models, which are usually assigned in the statistical literature as the Gaussian, Logistics, and Poisson distribution, respectively. The Gaussian model is applied when the dependent variable can take on a wide range of values, as in the case of the variable here under analysis, deforestation. Furthermore, it was found that the values of the dependent variable are grouped symmetrically around the mean, characterizing a normal distribution, thus responding to yet another criterion for using a Gaussian model.

The geographical coordinates of the municipal seat city identified each municipality (Fig. 1). Because these are non-contiguous sample points, the adaptive kernel function with an automatic band selection criterion, i.e., golden section (Greig 1980), based on the cross-validation method (CV) was adopted. From the identification of the EKC curves type (Fig. 2) and the values of the local parameters, the municipalities were grouped by the similarity of shapes and intensities of the relationships between their urbanization stages; a complex process but represented here by the evolution of the set of sociodemographic and economic variables, and their deforestation dynamics.

RESULTS

The descriptive statistics of the data set (Table 3) show that there is great variability in the deforestation variable (DAD) in the study area. The municipalities presented on average 542 km² of deforestation difference, ranging from zero up to 9967 km², and 1111 km² of standard deviation. From the total of 22 variables, 14 presented negative minimum values, indicating a reduction in the value of the variable in the time interval analyzed for some municipalities. This result, for most variables, suggests losses of resources, infrastructure, and/or service over time. Negative mean values indicate that the variables have low values between the units of analysis in the analyzed period.

The multicollinearity analysis reduced the set of variables from 22 to 15 variables. The final set presented a VIF value equal to 5.86. The excluded variables were: DTHEILI, DPPHE, DPPPE, DPHVUL, DINFM, DPSUR, DPPE (see Table 2 for an expansion of abbreviations). After excluding the collinear variables, linearity and spatial dependence were also tested. The normality of the residuals rejected the null hypothesis of normality. The homoscedasticity of the residuals did not reject the null hypothesis of homoscedasticity for one test and rejected it for the other. Moran’s test I, with the analysis of pseudo-significance (999 permutations), observed the presence of statistically significant spatial effects in the regression residuals, confirming the need for spatial methods (Table 4).

Because the model with original variables did not respond favorably to the assumptions of linearity, a log transformation was applied to the dependent variable (DAD). Logarithmic transformations enable adjusting the probabilities to a Gaussian curve, obtaining consistent results when dealing with primarily positive data (Pino 2014). From testing the linearized model, Santa Cruz do Arari was removed from the data set because it had a zero value in the deforestation variable (DAD). The model (Equation 3), adjusted after the analysis of the VIF and dependent variable transformation, is:

(3)

After the logarithmic transformation, the assumptions of linearity and the presence of spatial effects in the model were again evaluated. The normality test of the residuals failed to reject the null hypothesis and similarly, the homoscedasticity tests of the residuals failed to reject the null hypothesis. With the model aligned with the assumptions of linearity, Moran’s test I, with the analysis of pseudo-significance (999 permutations), observed the presence of statistically significant spatial effects. The global results of general adjustment of the variables in the GWR model showed that, compared to the non-spatial regression, the determination coefficient had a better adjustment using the spatial method. In addition to the determination coefficient, the Akaike value also indicated an improvement in the fit of the spatial model compared to the values found in the non-spatial regression (Table 4).

Figures 3 and 4 represent, respectively, EKC patterns and the local parameters estimated by the GWR model on municipalities’ maps (Tabular data can be made available upon reader’s request.). The dynamics of deforestation (Log (DAD)) as a function of per capita income (DINC, DINC², and DINC³) and other socioeconomic variables were not statistically significant (95% confidence were considered +/- 1.96 t-statistic) for the northeastern region of Pará, comprising the Metropolitan Region of Belém and its surroundings. This result is associated with the historical and geographical characteristics of the region. Because this area is contiguous with Belém, it receives the influence of an established and consolidated urban infrastructure, consequently of higher values of socioeconomic indexes. Additionally, forest resources have long been intensively exploited, as indicated by the low values of deforestation in recent dates. In 2019, Belém presented only 21.46% of forest area concerning the municipality area, whereas for the entire state of Pará, for example, this value reached 69% (INPE 2020).

When considering the spatial relationships present in the model of 15 complementary socioeconomic variables, only 4 were statistically significant. This result was expected because the theoretical-conceptual basis of this work assumes that, although interconnected, there are no direct and/or causal relationships between urbanization and deforestation in the Brazilian Amazon. These processes are subject to forces and agents that respond to different interests and act on different scales. It is also noteworthy that three of the four significant variables correspond to types of work, which shows the relevance of this aspect in the urban development process, corroborating the ideas proposed by Jacobs (1969) and discussed by Becker (2013) when highlighting the importance of restructuring of forms of work as a means to guarantee a more sustainable future for the Amazon, based on production and work systems compatible with a forest region.

For all municipalities in which the relationships were statistically significant, the variables difference in the percentage of employees with a formal contract (DPEFC; Appendix 4, Table A4.1), difference in the percentage of employees in the mineral extraction sector (DPEMES; Appendix 4, Table A4.2), and difference in the percentage of employees in the service sector (DPESS) were positively related to the variation in deforestation (Log (DAD)). On the other hand, the difference in the percentage of the population with garbage collection (DPPGC; Appendix 4, Table A4.4) had a negative relationship with the variation in deforestation during that period. It is noteworthy that the spatial variation of the local coefficients for these complementary variables (DPEFC, DPEMES, DPESS, and DPPGC), as a function of the deforestation dynamics (Log (DAD)), should not be interpreted as a direct contribution of these variables to the forest conversion process. Instead, they are a synthesis of the socioeconomic profile, variating according to the pattern of deforestation evolution.

DISCUSSION

The different shapes of EKC, observed in the context of the proposed analysis, respond to different occupation processes and socioeconomic contexts that represent different types of urbanization and their dynamics of deforestation. The "U" shaped priority pattern may be related to a type of urbanization and socioeconomic development that historically establishes a predatory relationship with natural resources. This result reflects the historical development of industrialization in the Amazon territory, described by Becker (1982) as an unequal process based on the valorization of raw materials and submission to the economic interests of the international market. Defined as a “resource frontier,” i.e., a settlement and productivity area in a virgin territory (Becker 1982), the presence of large-scale and economically profitable natural resources made the Amazon a place for development orbiting the economy of large centers, submitted to political force as a potential for national integration (Santos 2013).

The irregular distribution of the Amazonian population promotes physical and cultural isolation that suppresses the sense of community, facilitating the penetration of modern society in these spaces (Becker 1982, 2013, Pandolfo 1994). The construction of federal roads, between the 1960s and 1980s, linked the Amazon to the country’s major economic centers, facilitating the trade of raw materials that reinforced the colonialist and unbalanced economy (Becker 1982, Santos 2013). As well, the emergence of the capitalist industry changed the power of urban mediation in the Amazon territory (Silva 2021) understood since the colonial period as a space of power, celebration, and exchange and which with the advent of the urban-industrial model became interpreted as a space for production and consumption.

The reduction in deforestation at the beginning of the “U”-shaped EKC curve describes a profile of the evolution of socioeconomic development that reflects the trend to contain deforestation in this region up to a given moment, represented by the inflection of the curve. However, the change in the trajectory of the curve demonstrates that the command-and-control policies are not capable, by themselves, of refraining forest conversion practices in the long term. This dynamic shows the strength of industrialization in the process of transforming the territory, and its relationships that seeking to homogenize and align production, consumption, and lifestyle patterns with commercial logic, ignores the potential of diversity and reinforces the domain of exogenous agents and export logic (Monte-Mór 1994, Santos 2009, 2013).

On the other hand, the EKC’s decreasing monotonic shape illustrates a type of socioeconomic development that converges with the reduction of deforestation. As the municipality reaches higher socioeconomic conditions, deforestation reduces. All municipalities with decreasing monotonic EKC curves (Fig. 3) registered an increase in their rural population, suggesting that EKC considers the importance of the rural economy for the region. In the Marajó Archipelago, the municipalities presented less than 10% of deforestation over the total area of the municipality (INPE 2020). The increased value of the floodplain products in the market, together with the policies of legalization of land tenure, helped to maintain the viability of the rural economy (Costa and Brondízio 2011). In the region, the increase of livestock occupied mainly native pasture areas of poorly drained savannas (Arima and Uhl 1996), conserving the natural forest landscape. In addition to the livestock, the island’s economy is based on fishing and extraction of açaí and rubber.

Other municipalities in the group, such as those in the Mocajuba region, are part of an important açaí producing region, comprising the mesoregions of Northeast of the state of Pará, some municipalities of the Marajó archipelago, and of the Metropolitan Region of Belém (MRB; Costa et al. 2018), which have the fruit extraction as the main base of the local economy. The açaí economy is known to be important for sustainable and inclusive development. It is based on peasant forms of management and the industrial process is associated with small and medium-sized companies based on local socioeconomic potential (Costa et al. 2018). The production and consumption of açaí in this region illustrate how the deepening of relations between the urban base and the agro-extractive economies can create alternative urban development models (Silva 2021). The idea of “producing to conserve” (Becker 2013) overcomes the industrial logic, seeking development strategies based on economic dynamism and maintenance of natural and social diversity.

In contrast, the capitalist-industrial expansion, by transforming resources and means of production, changes the relationship between society and nature. Underlying industrialization and corporate urbanization (Santos 2013) direct its interests toward economic relations and neglects the existence of people who inhabit and make a living from the territory, manifesting what Castro (2009) calls “cities in the forest.” Development models, which are environmentally responsible and socially fair, require an understanding of the possibilities of economic growth, mediated by urbanization and by the inherent dynamics of nature itself (Siva 2017). It may be possible to promote cities as strategic places for the endogenous organization of the regional productive structure, resuming the society-nature relationship as represented by the “cities of the forest” (Browder and Godfrey 2006).

Additionally, the results of the complementary variables express the spatial variability of the municipal socioeconomic profiles, or the precariousness of the urban territories, according to the pattern of deforestation evolution. As previously described, the variable DPEFC (difference in the percentage of employees with a formal contract between 2000 and 2010) is positively related to the variation in deforestation (Fig. 4a). This relationship suggests that the evolution of deforestation could be associated with the creation of new formal jobs, which could lead to the heating of the local economy. Higher DPEFC values are concentrated in the northeast of the state, close to the Metropolitan Region of Belém, which was expected because public agencies, service companies, and commerce are concentrated in this region of the state. A non-significant area in the southeastern portion of the state is where the deforestation frontier advances. This spatial distribution follows the economic dynamics of Pará. Industrialized regions and centralities, such as the Metropolitan region, generate more formal jobs and offer services. Consequently, in these regions, economic activities associated with formal jobs contribute more to explain the growth of deforestation.

Furthermore, there are municipalities whose productive structures are mainly supported by exploiting the potential of Amazonian biodiversity, and the production and transformation processes create demands in the labor market. For example, the production of manioc, açaí, and coconut in the lower Tocantins region; of palm oil in the northeast of Pará, mainly in the municipalities of Acará, Bonito, Tailândia, Tomé-Açu, and Concórdia do Pará; and of black pepper, concentrated in Tomé-Açu, Igarapé-Açu, Baião, and Cametá, showcase some of the examples of the potential biodiversity in the productive systems that have at their center ways of producing associated with adequate management of the culture in coexistence with the surrounding forest landscape.

The DPEMES variable (difference in the percentage of employed persons in the mineral extraction sector) is also positively related to deforestation in a small group of municipalities in the northeast of Pará (Fig. 4b). Among the statistically significant municipalities for this relationship, Barcarena, Paragominas, Ipixuna do Pará, and Cachoeira do Piriá are some of the main mining municipalities in the state. Because mineral extraction and transformation are their main economic base activities, these municipalities stand out in the results for this variable. The positive relationship between mining and deforestation is highlighted by Sonter et al. (2017) when explaining that due to combined factors, i.e., land use displacement, development of commodity supply chains, urban expansion, and waste disposal, the effects of deforestation can reach large extensions and areas far beyond their operational limits.

The DPESS variable (difference in the percentage of employed persons in the service sector) presented a positive relationship with deforestation, observed in two groups of municipalities in the southeast and southwest of Pará (Fig. 4c). These are municipalities whose economic base is associated with both mineral exploration (the regions of Carajás and the lower Amazon (with emphasis on the exploration of bauxite in Oriximiná, and gold in Itaituba) and with livestock. Therefore, the operational aspects of their activities and their production chains, which provide infrastructure and services for the development of the activity and labor, are two associated factors that influence the dynamics of deforestation in these municipalities.

Finally, the variable DPGC (difference in the percentage of the population with garbage collection) is negatively related to deforestation and presents higher values in two deforestation frontiers, southeast and southwest of Pará (Fig. 4d). Municipalities with household waste collection services indicate urban structures that are more consolidated, opposing the dynamics of deforestation, associated with agricultural regions, and less structured urban development.

Considering the spatial variability of the evolution of the processes under analysis, the results confirm Sathler et al.’s (2018) hypothesis on the need for spatial methods to investigate socioeconomic and environmental processes from spatialized responses. Relationships between deforestation and socioeconomic development in the Amazon were previously discussed by Santiago and do Couto (2010) and Tritsch and Arvor (2016), however, these authors used non-spatial analysis approaches. Santiago and do Couto (2010) studied the entire Brazilian territory, and therefore, data were considered to provide generalized answers for the country, without considering regional particularities. On the other hand, studying the Brazilian Legal Amazon, Tritsch and Arvor (2016) considered only three socioeconomic variables to represent the general relationship between socioeconomic development and different classes of deforestation frontiers. Although in this study, the authors tried to observe a relationship between deforestation and a small set of socioeconomic variables, however, the objective is to attempt to characterize factors associated with deforestation frontiers and not with the urbanization process.

Therefore, our results complement previous analyses by Tritsch and Arvor (2016) by considering a more robust dataset, with a greater capacity to represent the socioeconomic reality and evolution of deforestation in the Amazon region. Our results also contribute to the discussions raised by Santiago and do Couto (2010) by providing a more specific, and consequently, more in-depth view, which contextualizes the interpretation of socio-demographic and economic variable analysis as a dimension of urbanization. That said, the greatest contribution of this work lies in the proposition of a methodological approach that considering the spatial heterogeneity of the Amazon territory and the spatial dependence of the variables, provides individualized EKC curves for each municipality through local spatial analysis. Furthermore, the set of explanatory variables in this work aims to represent the social, demographic, and economic transformations related to the population and the territory urbanization at the municipal scale.

CONCLUSIONS

Our work proposed to analyze the impact of different urbanization developments based on the analysis of some variables linked to socio-demographic and economic development features, with the dynamics of the deforestation in the Pará Amazon region, establishing municipalities as the unit of analysis. In the light of the environmental Kuznets curve (EKC) theory, we identified two base-type curves, i.e., decreasing monotonic and “U” shapes, which describe the relationship between the subset of socio-demographic and economic variables, representing the evolution of urbanization and the dynamics of deforestation for the municipalities of the state of Pará.

The use of GWR to map the spatial variability intrinsic to the modeled phenomena enabled the observation of different forms of relationships between the variables. In general, in the context of the Pará Amazon and consolidated economy scenarios, deforestation and income tend to converge. The prevalence of the “U”-shaped EKC illustrates the contradiction generated by the incorporation of the Amazon into urban-industrial economic dynamics. The industrial economic model, based on the logic of production and accumulation, acts as a mediator for the transformation of natural resources, space, and consequently, the relationship between people and their territory. Although the Amazon is considered a frontier region of natural resources, the forest assets will be exploited in an accelerated and predatory way unless other development paths are considered. A development agenda without a sustainability strategy is unsustainable in the long term. Searching for alternative models is essential for the conservation and maintenance of the Amazon economic, cultural wealth, and diversity, i.e., its socio-biodiversity. Planning and building a future, not only for the forest but also for the people who live there, is essential.

The results of the EKC curves show that pre-industrial production systems remain alive in contemporary Amazonia and represent alternatives for the recovery of the links between development and sustainable use of forest resources (Costa 2019). The decreasing monotonic EKC curve shows a disconnect between the evolution of socio-demographic and economic variables and deforestation over a period of time. The municipalities in this group are characterized by productive systems that operate in greater harmony with a forested landscape, that is, based on regional socio-biodiversity resources, according to the concept of the “cities of the forest.” On the other hand, the “U”-shaped EKC curve shows an initial trend toward a reduction in deforestation which, according to socio-demographic and economic variable evolution, undergoes an inversion point and starts to have a concurrent trajectory in the other direction. This curve was predominant in the state of Pará, representing the strong influence of capitalist industrialization on the current urban patterns of development in the Amazon, as characterized by the concept of “cities in the forest.” This development project is based on the association between exogenous and endogenous capital, which, from the actions of a local elite, imposes an urban-industrial pattern of territory transformation that disregards the dynamics of nature and the interests of the population.

Regionally based planning, designed to contribute to the autonomy of cities, develop the economy, and add value to the diversity of regional resources, can promote socio-demographic advances with a concomitant reduction in deforestation activity. In this context, it is possible to highlight the importance of the geopolitical strategies described by the concept of “cities of the forest” as an alternative to detach themselves from industrial standards and to prioritize new economies and social and environmental relations. The resumption of traditional production bases, associating local technologies and knowledge with new ones, aiming at valuing regional potential, would result in a spatial awareness more capable of promoting the use and management of the forest’s natural resources without destroying them, updating the traditional knowledge, i.e., about the ways of producing and living in Amazonia, and promoting economic autonomy. Thus, we can think of building a pattern of urban development that is socially and environmentally compatible with a forest biome.

LITERATURE CITED

Andersen, L. E., and E. J. Reis. 2015. Deforestation, development, and government policy in the Brazilian Amazon: an econometric analysis. Discussion Paper No. 69. Institute for Applied Economic Research, Brasília, Brazil. [online] URL: https://www.econstor.eu/bitstream/10419/220158/1/dp_069.pdf

Arima, E., and C. Uhl. 1996. Pecuária na Amazônia Oriental: desempenho atual e perspectivas futuras. No. 1. Imazon, Belém, Brazil.

Becker, B. 2013. A urbe amazônida: a floresta e a cidade. First edition. Garamond, Rio de Janeiro, Brazil.

Becker, B. K. 1982. Geopolítica da Amazônia: a nova fronteira de recursos. First edition. Zahar, Rio de Janeiro, Brazil.

Becker, B. K., and C. Stenner. 2008. Um futuro para a Amazônia. First edition. Oficina de Textos, Rio de Janeiro, Brazil.

Berk, K. N. 1977. Tolerance and condition in regression computations. Journal of the American Statistical Association 72(360a):863-866. https://doi.org/10.1080/01621459.1977.10479972

Bretschger, L., and S. Smulders. 2000. Explaining environmental Kuznets curves: how pollution induces policy and new technologies. Tilburg University CentER Working Paper No. 2000-95. Tilburg University, Tilburg, The Netherlands. http://dx.doi.org/10.2139/ssrn.249023

Browder, J., and B. Godfrey. 2006. Cidades da Floresta: urbanização, desenvolvimento e globalização na Amazônia Brasileira. First edition. EDUA-Editora da Universidade Federal do Amazonas, Manaus, Brazil.

Brunsdon, C., A. S. Fotheringham, and M. E. Charlton. 1996. Geographically weighted regression: a method for exploring spatial nonstationarity. Geographical Analysis 28(4):281-298. https://doi.org/10.1111/j.1538-4632.1996.tb00936.x

Cardoso, A. C. D., A. C. Melo, and T. V. Gomes. 2016. O urbano contemporêneo na fronteira de expansêo do capital: padrões de transformações espaciais em seis cidades do Pará, Brasil. Revista de Morfologia Urbana 4(1):5-28.

Cardoso, A. C. D., and M. R. G. Negrão. 2013. Das máquinas autoritárias de planejamento do século XX, às máquinas de crescimento urbano do século XXI: o caso do sudoeste do Pará. Pages 758-767 Proceedings of the 15th Encontro nacional da Associação Nacional de Pós-Graduação e Pesquisa em planejamento urbano e regional. Belém do Pará, Pará, Brazil.

Carvalho, T. S., and E. Almeida. 2010. A hipótese da curva de Kuznets ambiental global: uma perspectiva econométrico-espacial. Estudos Econômicos (São Paulo) 40(3):587-615. https://doi.org/10.1590/s0101-41612010000300004

Castro, E. M. R. 2009. Cidades na floresta. Annablume, São Paulo, Brazil.

Caviglia-Harris, J. L., D. Chambers, and J. R. Kahn. 2009. Taking the “U” out of Kuznets: a comprehensive analysis of the EKC and environmental degradation. Ecological Economics 68(4):1149-1159. https://doi.org/10.1016/j.ecolecon.2008.08.006

Caviglia-Harris, J., E. Sills, A. Bell, D. Harris, K. Mullan, and D. Roberts. 2016. Busting the boom-bust pattern of development in the Brazilian Amazon. World Development 79:82-96. https://doi.org/10.1016/j.worlddev.2015.10.040

Charlton, M., A. S. Fotheringham, and C. Brunsdon. 2009. Geographically weighted regression. White Paper No. 2. National University of Ireland, Maynooth, Maynooth, Ireland.

Corrêa, R. L. 1987. A periodizaçêo da rede urbana da Amazônia. Revista Brasileira de Geografia 4(3):39-68

Costa, F. A. 2019. A brief economic history of Amazon (1720-1970). Cambridge Scholars, Cambridge, UK.

Costa, S. M., and E. S. Brondízio. 2011. Cities along the floodplain of the Brazilian Amazon: characteristics and trends. Pages 83-97 in M. Pinedo-Vasquez, M. Ruffino, C. Padoch, and E. Brondízio, editors. The Amazon Várzea. Springer,Dordrecht, The Netherlands. https://doi.org/10.1007/978-94-007-0146-5_6

Costa, F. A., D. A. Fernandes, and C. N. S. Crispim. 2018. Constituição, Situação e Dinâmica de Arranjos Produtivos Locais: o caso do APL de açaí na região do Grão-Pará (2002 a 2010). Análise Econômica 36(69). https://doi.org/10.22456/2176-5456.55792

Forster, B. A. 1973. Optimal consumption planning in a polluted environment. Economic Record 49(4):534-545. https://doi.org/10.1111/j.1475-4932.1973.tb01954.x

Fotheringham, A. S., C. Brunsdon, and M. Charlton. 2002. Geographically weighted regression: the analysis of spatially varying relationships. John Wiley and Sons, West Sussex, UK.

Fundação Amazônia de Amparo a Estudos e Pesquisas do Pará (FAPESPA). 2016. Estimativas e projeções do PIB paraense 2014-2020. FAPESPA, Belém do Pará, Brazil.

Fundação Amazônia de Amparo a Estudos e Pesquisas do Pará (FAPESPA). 2017. Estimativas e projeções do PIB Paraense 2015-2020. FAPESPA, Belém do Pará, Brazil.

Gallego-Alvarez, I., M. Vicente-Galindo, M. Galindo-Villardón, and M. Rodríguez-Rosa. 2014. Environmental performance in countries worldwide: determinant factors and multivariate analysis. Sustainability 6(11):7807-7832. https://doi.org/10.3390/su6117807

Gomes, S. C., and M. J. Braga. 2008. Desenvolvimento econômico e desmatamento na Amazônia Legal: uma análise econométrica. In 46th Congresso da Sociedade Brasileira De Economia, Administração e Sociologia Rural. Rio Branco, Acre, Brazil. https://doi.org/10.22004/ag.econ.103095

Greig, D. M. 1980. Optimisation. Longman, Harlow, UK.

Grossman, G. M., and A. B. Krueger. 1991. Environmental impacts of a North American free trade agreement. Working Paper No. 3914. National Bureau of Economic Research, Cambridge, Massachusetts, USA. https://doi.org/10.3386/w3914

Grossman, G. M., and A. B. Krueger. 1995. Economic growth and the environment. Working Paper No. 4634. National Bureau of Economic Research, Cambridge, Massachusetts, USA. https://doi.org/10.3386/w4634

Gruver, G. W. 1976. Optimal investment in pollution control capital in a neoclassical growth context. Journal of Environmental Economics and Management 3(3):165-177. https://doi.org/10.1016/0095-0696(76)90016-4

Guedes, G. R., E. S. Brondízio, A. F. Barbieri, R. Anne, R. Penna-Firme, and A. O. D’Antona. 2012. Poverty and inequality in the rural Brazilian Amazon: a multidimensional approach. Human Ecology 40(1):41-57. https://doi.org/10.1007/s10745-011-9444-5

Hoffmann, R. 1991. O índice de desigualdade de Theil-Atkinson. Brazilian Review of Econometrics 11(2):143-160. https://doi.org/10.12660/bre.v11n21991.3001

Instituto Brasileiro de Geografia e Estatística (IBGE). 2000. Censo demográfico 2000. Instituto Brasileiro de Geografia e Estatística, Rio de Janeiro, Brazil. [online] URL: https://www.ibge.gov.br/estatisticas/downloads-estatisticas.html

Instituto Brasileiro de Geografia e Estatística (IBGE). 2005. Mapa da Amazônia Legal: fronteira í;cola. Instituto Brasileiro de Geografia e Estatística, Rio de Janeiro, Brazil. [online] URL: https://www.ibge.gov.br/geociencias/cartas-e-mapas/mapas-regionais/15976-fronteira-agricola-amazonia-legal.html?=&t=downloads

Instituto Brasileiro de Geografia e Estatística (IBGE). 2010a. Censo demográfico 2010. Instituto Brasileiro de Geografia e Estatística, Rio de Janeiro, Brazil. [online] URL: http://censo2010.ibge.gov.br/

Instituto Brasileiro de Geografia e Estatística (IBGE). 2010b. Sinopse do censo demográfico 2010. Instituto Brasileiro de Geografia e Estatística, Rio de Janeiro, Brazil. [online] URL: https://censo2010.ibge.gov.br/sinopse/index.php?dados=8

Instituto Nacional de Pesquisas Espaciais (INPE). Coordenação-Geral de Observação da Terra. 2020. Programa de monitoramento da Amazônia e demais biomas. Desmatamento-Amazonia Legal, São Paulo, Brazil. [online] URL: http://terrabrasilis.dpi.inpe.br/downloads/

Jacobs, J. 1969. The economy of cities. Random House, New York, New York, USA.

Kaika, D., and E. Zervas. 2013. The environmental Kuznets curve (EKC) theory. Part B: critical issues. Energy Policy 62:1403-1411. https://doi.org/10.1016/j.enpol.2013.07.130

Katircioğlu, S., and S. Katircioğlu. 2018. Testing the role of urban development in the conventional environmental Kuznets curve: evidence from Turkey. Applied Economics Letters 25(11):741-746. https://doi.org/10.1080/13504851.2017.1361004

Kim, Y., K. Tanaka, and C. Ge. 2018. Estimating the provincial environmental Kuznets curve in China: a geographically weighted regression approach. Stochastic Environmental Research and Risk Assessment 32(7):2147-2163. https://doi.org/10.1007/s00477-017-1503-z

Kutner, M. H., J. Neter, C. J. Nachtsheim, and W. Li. 2004. Applied linear statistical models. Fifth edition. McGraw- Hill Irwin, Boston, Massachusetts, USA.

Lefebvre, H. 1999. A revolução urbana. UFMG, Belo Horizonte, Brazil.

Luo, J., P. Du, A. Samat, J. Xia, M. Che, and Z. Xue. 2017. Spatiotemporal pattern of PM_2.5 concentrations in mainland China and analysis of its influencing factors using geographically weighted regression. Scientific reports 7(1):40607. https://doi.org/10.1038/srep40607

Margulis, S. 2003. Causas do desmatamento da Amazônia brasileira. Banco Mundial, Brasília, Brazil. [online] URL: https://www.terrabrasilis.org.br/ecotecadigital/pdf/causas-do-desmatamento-da-amazonia-brasileira.pdf

Miles, J. 2014. Tolerance and variance inflation factor. Wiley StatsRef: statistics reference online. Wiley, New York, New York, USA. https://doi.org//10.1002/9781118445112.stat06593

Molotch, H. 1976. The city as a growth machine: toward a political economy of place. American Journal of Sociology 82(2):309-332. https://doi.org/10.1086/226311

Moran, E. F. 1993. Deforestation and land use in the Brazilian Amazon. Human Ecology 21(1):1-21. https://doi.org/10.1007/BF00890069

Monte-Mór, R. L. M. 1994. Urbanização extensiva e lógicas de povoamento: um olhar ambiental. In M. Santos, M. A. Souza, and M. L. Silveira, editors. Território, globalização e fragmentação. Hucitec/Anpur, São Paulo, Brazil.

Oliveira, R. C. D., E. Almeida, R. D. S. Freguglia, and R. C. S. Barreto. 2011. Deforestation and economic growth in Brazil: an analysis of the environmental Kuznets curve for the legal Amazon. Journal of Rural Economics and Sociology 49(3):709-739.

Pandolfo, C. M. 1994. Amazônia brasileira: ocupação, desenvolvimento e perspectivas atuais e futuras. First edition. Editora Cejup, Belém do Pará, Brazil.

Pino, F. A. 2014. A questão da não normalidade: uma revisão. Revista de economia agrícola 61(2):17-33.

Prates, R. C. 2008. O desmatamento desigual na Amazônia brasileira: sua evolução, suas causas e conseqüências sobre o bem-estar. Dissertation. Universidade de São Paulo, São Paulo, Brazil. https://doi.org/10.11606/T.11.2008.tde-06082008-162358

Ribeiro, R. M., S. Amaral, A. M. V. Monteiro, and A. P. Dal’Asta. 2018. The urbanization and forest conversion processes in the Paraense Amazon -a multiescalar study. Revista Brasileira de Estudos de População 35(3). https://doi.org/10.20947/s102-3098a0068

Richards, P., and L. VanWey. 2015. Where deforestation leads to urbanization: how resource extraction is leading to urban growth in the Brazilian Amazon. Annals of the Association of American Geographers 105(4):806-823. https://doi.org/10.1080/00045608.2015.1052337

Rindfuss, R. R., and P. C. Stern. 1998. Linking remote sensing and social science: the need and the challenges. Pages 1-27 in D. Liverman, E. Moran, R. Rindfuss, and P. Stern, editors. People and pixels: linking remote sensing and social science. National Academy, Washington, D.C., USA.

Rodrigues, A. S. L., R. M. Ewers, L. Parry, C. Souza, Jr., A. Veríssimo, and A. Balmford. 2009. Boom-and-bust development patterns across the Amazon deforestation frontier. Science 324(5933):1435-1437. https://doi.org/10.1126/science.1174002

Santiago, A. R., and H. T. Z. do Couto. 2020. Socioeconomic development versus deforestation: considerations on the sustainability of economic and social growth in most Brazilian municipalities. Environmental Development 35:100520. https://doi.org/10.1016/j.envdev.2020.100520

Santos, M. 2009. Pobreza urbana. Third edition. Edusp, São Paulo, Brazil.

Santos, M. 2013. A urbanização brasileira. Fifth edition. Edusp, São Paulo, Brazil.

Sathler, D., S. B. Adamo, and E. E. C. Lima. 2018. Deforestation and local sustainable development in Brazilian Legal Amazonia. Ecology and Society 23(2):30. https://doi.org/10.5751/es-10062-230230

Selden, T. M., and D. Song. 1994. Environmental quality and development: is there a Kuznets curve for air pollution emissions? Journal of Environmental Economics and Management 27(2):147-162. https://doi.org/10.1006/jeem.1994.1031

Shafik, N., and S. Bandyopadhyay. 1992. Economic growth and environmental quality: time-series and cross-country evidence. World Bank Publications, Vol. 904. World Bank, Washington, D.C., USA.

Silva, H. 2021. A economia do açaí em Belém-PA: vida urbana e biodiversidade em uma experiência singular de desenvolvimento econômico. Novos Cadernos NAEA 24(3):259-286.

Sonter, L. J., D. Herrera, D. J. Barrett, G. L. Galford, C. J. Moran, and B. S. Soares-Filho. 2017. Mining drives extensive deforestation in the Brazilian Amazon. Nature Communications 8(1):1013. https://doi.org/10.1038/s41467-017-00557-w

Stern, D. I., M. S. Common, and E. B. Barbier. 1996. Economic growth and environmental degradation: the environmental Kuznets curve and sustainable development. World Development 24(7):1151-1160. https://doi.org/10.1016/0305-750X(96)00032-0

Tamhane, A. C., and D. D. Dunlop. 2000. Statistics and data analysis - from elementary to intermediate. Prentice-Hall, Upper Saddle River, New Jersey, USA.

Tanaka, K., and S. Matsuoka. 2007. Reconsidering the environmental Kuznets curve: geographically weighted regression approach. Hiroshima University. The 21st Century COE Program ‘Social Capacity Development for Environmental Management and International Cooperation’. Discussion Paper Series Vol. 8. Hiroshima International Center for Environmental Cooperation, Hiroshima, Japan.

Tritsch, I., and D. Arvor. 2016. Transition in environmental governance in the Brazilian Amazon: emergence of a new pattern of socio-economic development and deforestation. Land Use Policy 59:446-455. https://doi.org/10.1016/j.landusepol.2016.09.018

Tritsch, I., and F.-M. Le Tourneau. 2016. Population densities and deforestation in the Brazilian Amazon: new insights on the current human settlement patterns. Applied Geography 76:163-172. https://doi.org/10.1016/j.apgeog.2016.09.022

Tsiantikoudis, S., E. Zafeiriou, G. Kyriakopoulos, and G. Arabatzis. 2019. Revising the environmental Kuznets curve for deforestation: an empirical study for Bulgaria. Sustainability 11(16):4364. https://doi.org/10.3390/su11164364

United Nations Development Programme (UNDP). 2010. Human development index 2010. Ranking by states. United Nations Development Programme, New York, New York, USA. [online] URL: https://www.br.undp.org/content/brazil/pt/home/idh0/rankings/idhm-uf-2010.html

Yasin, I., N. Ahmad, and M. A. Chaudhary. 2020. Catechizing the environmental-impression of urbanization, financial development, and political institutions: a circumstance of ecological footprints in 110 developed and less-developed countries. Social Indicators Research 147(2):621-649. https://doi.org/10.1007/s11205-019-02163-3