Ecological Indicators 7 (2007) 719–730 This article is also available online at: www.elsevier.com/locate/ecolind Socioeconomic dimensions, migration, and deforestation: An integrated model of territorial organization for the Brazilian Amazon Ricardo Alexandrino Garcia a,b,*, Britaldo Silveira Soares-Filho a,c, Diana Oya Sawyer a a Center for Regional Development and Planning (Cedeplar), Universidade Federal de Minas Gerais, UFMG, Av. Antônio Carlos, 6627, 31270-900 Belo Horizonte, MG, Brazil b Department of Geography, Universidade Federal de Minas Gerais, UFMG, Av. Antônio Carlos, 6627, 31270-900 Belo Horizonte, MG, Brazil c Center for Remote Sensing (CSR) and Department of Cartography, Universidade Federal de Minas Gerais, UFMG, Av. Antônio Carlos, 6627, 31270-900 Belo Horizonte, MG, Brazil Received 13 February 2006; received in revised form 2 August 2006; accepted 12 August 2006 Abstract The territory organization of the Brazilian Amazon, understood as a socioeconomic network of municipalities, reflects the diversity of human settlements and their potentials for environmental changes, especially deforestation. This paper characterizes the urban network in the Brazilian Amazon through a model that integrates the levels of socioeconomic organization of municipalities and their interrelationships, as determined through migratory movements. The model of territorial organization combines five components: (i) the hierarchy of central places (poles) established by the concentration of urban specialized services, (ii) the geographical distance between central poles and other centers, (iii) the poles’ populations, (iv) the migratory movements among them, and (v) a socioeconomic dimension index. These components are combined into a gravitational model to produce measures and maps of the socioeconomic municipality network of the Brazilian Amazon. As a result, out of 792 municipalities in the Brazilian Amazon, 9 were classified as macro-poles, 29 were classified as meso-poles and 48 as micropoles. The areas of influence of these poles were determined according to the three hierarchy levels. The Amazon region network comprises a nested spatial pattern of municipalities not constrained by the state boundaries. Socioeconomic space and population movements influence the mobility of the deforestation frontier. This analysis provides insights to predict deforestation as well as to guide formulation of sustainable development policies suitable for each region’s specificity. # 2006 Elsevier Ltd. All rights reserved. Keywords: Socioeconomics; Migration; Territorial organization; Deforestation; Amazon * Corresponding author. Tel.: +55 31 34995449; fax: +55 31 34995410. E-mail addresses: alexandrinogarcia@gmail.com (R.A. Garcia), britaldo@csr.ufmg.br (B.S. Soares-Filho), dsawyer@cedeplar.ufmg.br (D.O. Sawyer). 1470-160X/$ – see front matter # 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.ecolind.2006.08.003 720 R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 1. Introduction Although socioeconomic dynamics drive the fast pace of deforestation in the Brazilian Amazon, the intensity of the deforestation process varies greatly across the basin due to regional differences, including physiographic attributes, access to infrastructure such as paved roads, population characteristics and dynamics and socioeconomic organization, as well as age of the frontier. Hence the understanding of the territorial distribution of both economic activity and socio-demographic dynamics is a starting point to evaluate a region’s potential for environmental changes, such as deforestation (Soares-Filho et al., 2006). This fact is implicit in the concept of anthropic pressure, which is the pressure imposed on the environment by the socioeconomic conditions of human settlement (Monteiro and Sawyer, 2001). Additionally, the establishment of a territorial hierarchy, representing nested spatial levels of socioeconomic and demographic organization, provides a basis for design of sustainable development policies according to each region’s specificities. This spatial organization can be represented in terms of the urban network and the space under its influence. According to Becker (2001), the Brazilian Amazon can be regarded as an urban forest. Indeed, IBGE (Brazilian Institute for Geography and Statistics) census data for 2000 show that more than 68% of its population lives in urban centers. Although this view stresses the need for understanding the Amazon socioeconomic space in terms of its urban network, only a few studies (e.g. Wood and Porro, 2000; Carpentier et al., 2000; Armenteras et al., 2006) have so far addressed the deforestation process in the light of the urban network and its associated socioeconomic and demographic dynamics. Other previous studies have provided methods to map the urban network layout and its associated socioeconomic space in Brazil in order to direct regional development policies (e.g. IPEA/IBGE/NESUR, 1999; Lemos et al., 1999, 2003; Garcia, 2002). It is worth mentioning that the three first studies did not explicitly involve migration, the only variable from census data able to measure flows over time and space. Only Garcia (2002), using the same framework as Lemos et al. (1999) proposed a regionalization for the Brazilian territory based on population movements among central poles and areas under their influence. Nevertheless, none of these studies provided a methodology specifically addressed to the potential of socioeconomic and demographic dimensions for environmental changes in the Amazon region. 2. Objective and rationales The main objective of this study is to establish a model of territorial organization for the Brazilian Amazon based on the characteristics of its municipalities. The model consists of two components. The first is a measure of hierarchy of cities as a function of urban specialized services and is used to identify poles of influence. The theory of central places (Christaller, 1966) refers to a nodal urban hierarchy consisting of larger centers polarizing surrounding smaller centers. The main rationale is that the largest population centers, in which urban services are concentrated, have the largest potential for attracting socioeconomic activities and for spreading them to surrounding centers. Considering the territory as a plastic space made up of interactions among places through flows of people, resources and information (Harvey, 1989), the second component is a measure of interaction among those previously identified poles and municipalities under their influence. This measure of interaction considers: (1) for each municipality, its population size weighted by a socioeconomic dimension index (SDI); (2) for each municipality in relation to the economic poles, its distance and the overall migratory flow between them. To assess the influence of socioeconomic dimensions on the environment, we hypothesize that SDI can be used to infer the anthropic pressure exerted by a population within a specific territory unit – such as the municipalities or larger regions formed by their aggregation – on the deforestation process. We combine the following five dimensional axes to produce the SDI summary: (1) population concentration and dynamics, i.e. total growth rate; (2) economic development; (3) agrarian infrastructure; (4) agricultural and timber production; (5) social development. A positive effect on the deforestation process is ascribed for the first four dimensions and a negative effect for the fifth. The rationale for establishing these components is that population growth and migration, together with economic growth, stimulate deforestation (Skole et al., 1994; R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 Laurance, 1999; Soares-Filho et al., 2004). In addition, logging (Nepstad et al., 1999) plus agriculture and ranching expansion are listed as major current causes of deforestation in the Brazilian Amazon (Margulis, 2002; Mertens et al., 2002; Alencar et al., 2004). Conversely, social development, as illustrated in the inverted Ushape Kuznets curve (Stern et al., 1996), could represent frontier governance (Nepstad et al., 2002), counteracting environmental degradation resulting from economic development. SDI is interpreted as a proxy for the anthropic pressure on deforestation. As a measure of attraction, the population, strengthened by SDI, is employed in a gravitational model, in conjunction with the total migratory movement among the poles and the surrounding municipalities to make up their areas of influence. Therefore, this study characterizes the level of anthropic pressure and the interrelationships among municipalities, given by migratory movements, to establish a network of regions, aiming to shed light on the way these regions interconnect to influence the mobility of the deforestation frontier across the basin. 3. Methodology The measure of the hierarchy of the cities is assessed by the index of services (IS). The urban centers were identified and ranked according to their supply of services, referred to here as the ‘‘service economy’’, as follows: ISi ¼ SDPi ð1  elnð0:05ÞSDPi =SDPref Þ GDPi (1) where ISi is a ratio between the service economy domestic product (SDPi) and the gross domestic product (GDPi) of a municipality i, standardized by a reference service economy domestic product (SDPref), specifically the largest regional SDPi. As depicted in Fig. 1, a municipality tends to concentrate more urban services as its economy grows. Thus, IS measures the potential of a municipality to act as a regional center, influencing surrounding municipalities. We established three categories of centers by defining threshold values for IS: macro-poles, mesopoles, and micro-poles. Municipalities with IS higher than 20% are the macro-poles. This upper limit was defined to include all capital cities, except for Palmas, 721 Fig. 1. Scattergram of gross domestic product (GDP) vs. service economy domestic product (SDP). Ellipse embraces all macropoles, except for Palmas. Both axes are in logarithmic scale. Tocantins state capital. Two more thresholds were set by determining the natural breaks of IS distribution frequency. Those with IS between 3.2 and 20% were classified as meso-poles and those with IS between 0.5 and 3.2% were the micro-poles, while municipalities with IS below 0.5% were not considered as poles. Furthermore, the qualitative aspects related to the importance of a municipality as a regional administrative center, its geographical location with respect to neighboring poles, namely its contiguity with other municipalities classified as poles, would add or subtract municipalities from poles defined by the IS thresholds. For example, Palmas, with IS of 7.8%, was ranked as a macro-pole because of its political importance, and Jarú was excluded from meso-poles due to its contiguity to Ji-Paraná, which presents higher IS. Once a hierarchy of regional poles is established, which can include a varying number of economic poles depending on a chosen cut-off threshold for IS, the measure of interaction between a pole and a municipality is calculated according to a gravitational model, as denoted: Ivi j ¼ Pi ð1 þ SDIi ÞP j ð1 þ SDI j Þ dijj (2) where Ivij represents the interaction between pole i and municipality j, given by their populations (Pi and Pj) and socioeconomic dimension indices (SDIi and SDIj), weighted by the distance (in this case, the geodetic distance) between them raised to the power of j, an attrition coefficient so that: j ¼ 1 þ eðlnð0:001Þ=vmtref Þvmti j (3) 722 R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 where vmtij is the overall migratory flow between pole i and municipality j and vmtref is the reference migratory flow, namely the largest intermunicipal migratory flow. The socioeconomic dimension index is calculated by applying the grade of membership (GOM) fuzzy classification method – see Manton et al. (1994) for theory and Gold et al. (1990); Sawyer and Beltrão (1991); Hughes et al. (1996) for application examples – to socioeconomic and agriculture census data, such as population density and growth rate, urbanization level and rate; gross domestic products, municipal income taxes and budget; number and types of agricultural implements; production from animal husbandry, agriculture, and forestry; education, habitation, and health parameters. SDI was built as a proxy of the anthropic pressure. As such, this index combines the following five dimensional axes: (1) population concentration and dynamics, (2) economic development, (3) agrarian infrastructure, (4) agricultural and timber production, and (5) social development. High levels of the first four dimensions are combined with low level of social development to produce the socioeconomic dimension index for each municipality. Only this synthetic index is presented in this paper. Variables for each of the dimensions are listed in Table 1. Their data sources are IBGE, 2000 demographic census; IBGE, 1996 population tally; IBGE, 1999 municipal database; IBGE, 1999 profiles of the Brazilian municipalities; Andrade and Serra (1999); IBGE, 1995–1996 agricultural census; IBGE, 2000 municipal cattle herd survey; IBGE, 1997 municipal cattle herd survey; IBGE, 2000 municipal agricultural survey; IBGE, 1997 municipal agricultural survey; and logging data from Verı́ssimo et al. (2001). Migratory flows were calculated based on a matrix with information for the residents in a municipality and the municipality of residence 5 years prior to August 1st, 2000 (collected in the 2000 Demographic Census). Those residing outside the municipality in the reference period are considered in-migrants; conversely those residing in the municipality in the reference period but outside it in 2000 are the outmigrants. The difference between the two is the net migration and the sum is the overall volume of population exchange (Garcia, 2002). As a measure of interaction between pole and a municipality, the overall migratory volume is employed in Eq. (3). Table 1 Variables used for the socioeconomic dimension index, according to their dimensional axis, year of reference, and source of information I. Population concentration and dynamics Total population (2000)a Population density (2000)a Level of urbanization (2000)a Population growth rate (1996/2000)a,b II. Economic development Domestic gross product (1996)c Domestic gross product: primary sector (1996)c Domestic gross product: secondary sector (1996)c domestic gross product: tertiary sector (1996)c Number of banks (1998)d Total deposits in banks—thousand Reais (1998)d Total investments—thousand Reais (1998)d Municipality revenue (1997) d Municipality total expenditure (1997) d Share in the federal funds of the municipality (1998)d Land taxes (1998)d III. Agrarian infrastructure Agricultural aggregated value (1995–1996)e Number of tractors (1995–1996) e Sowing machines (1995–1996)e Harvesters (1995–1996)e Number of trucks (1995–1996)e Total of farming machinery (1995–1996)e IV. Agricultural and timber production Total area of agricultural establishments with less than 200 ha (1995–1996)e Total area of agricultural establishments with 200 ha or more (1995–1996)e Land tenure concentration (1995–1996)e Livestock (2000)f Annual rate of increase of the livestock (1997–2000)f,g Density of cultivated area (2000)h Annual rate of increase of the cultivated area (1997–2000)h,i Number of milling companies (1997)j Timber log volume per year (1997) j Area affected by logging (1997)j V. Social development Years of schooling—population age 7–14 (1996)b Years of schooling—population age 15–24 (1996)b Years of schooling—head of household (1996)b Hospitals per 1000 population (1999)d Hospital beds per 1000 population (1999) d Ambulatories per 1000 population (1999)d Health posts per 1000 population (1999) d Health centers per 1000 population (1999) d Medical doctor offices per 1000 population (1999)d Dentist offices per 1000 population (1999)d Ambulatories in general hospitals per 1000 population (1999) d Health posts per 1000 population (1999) d Hospital bedridden patients per 1000 population (1999)d Number of households (2000)a R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 Table 1 (Continued ) Improvised private household (2000)a Collective household (2000) a Water supply (2000)a Bathroom or sanitary installation (2000)a Garbage collection/destination (2000)a Electricity (2000)a Average number of television sets per household (2000) a Telephone sets in the household (2000)a Paved streets (1999)k Streets with illumination (1999)k a b c d e f g h i j k IBGE 2000 demographic census. IBGE 1996 population tally. Andrade and Serra (1999). IBGE 1999 municipal database. IBGE 1995–1996 agricultural census. IBGE 2000 municipal cattle herd survey. IBGE 1997 municipal cattle herd survey. IBGE 2000 municipal agricultural survey. IBGE 1997 municipal agricultural survey. Verı́ssimo et al. (2001). IBGE 1999 profiles of the Brazilian municipalities. Hence Ivij measures the dependence of a municipality on a regional center, defined as the attraction exerted by the center’s population. In this case, the population attraction of a central place is stressed by its socioeconomic dimension index, as we assume that larger populations with higher anthropic pressure – as determined by SDI – have more impact on the environment. In this gravitational model, the dependence of a municipality to a pole is strengthened by two-way migratory flows and weakened by geographical distance. Finally, we mapped the poles’ areas of influence by assigning to a particular pole all municipalities where its respective Ivij is greatest. 4. Results 4.1. Urban hierarchy Out of 792 municipalities in the Brazilian Amazon, 9 were classified as macro-poles, 29 as meso-poles and 48 as micro-poles. Municipalities with IS higher than 20% are the macro-poles. These include eight state capital cities with IS values above 20%: São Luis, Cuiabá, Porto Velho, Rio Branco, Manaus, Boa Vista, Belém, and Macapá. Palmas, the capital city of Tocantins state, with IS of 7.8%, was added to this list because of its administrative and political functions (Table 2). 723 The non-contiguous municipalities classified as meso-poles, with IS above 3.2%, comprise the above macro-poles, as a meso-pole is also a micro-pole – thus a macro-pole is also a meso-pole and a micro-pole – plus the cities of Rondonópolis, Ji-Paraná, Imperatriz, Santarém, Sinop, Cacoal, Marabá, Cáceres, Barra do Garças, Vilhena, Ariquenes, Tangará da Serra, Paraupebas, and Gurupi. Additional six municipalities (Caxias, Bacabal, Tucuruı́, Araguaı́na, Cruzeiro do Sul, and Tefé) were added to this level because of their known importance as regional administrative centers (Table 2). One hundred and sixteen municipalities were initially classified as micro-poles, with IS above 0.5%. After excluding contiguous poles with lower IS, they were reduced to 48: the above-mentioned macro and meso-poles plus Parintins, Guarantã do Norte, Breves, Colı́der, Juara, Alta Floresta D’Oeste, Balsas, Pontes e Lacerda, Almeirim, Barra do Corda, Paragominas, Juı́na, Itaituba, Codó, Altamira, Redenção, Jaru, Guajará-Mirim, and Alta Floresta (Table 2). 4.2. Socioeconomic dimension index Each municipality is assigned to a grade of membership within the extreme profiles that encompass the highest and lowest categories of variables constituting each of the five axes—population concentration, economic development, agrarian infrastructure, agricultural and timber production, and social development. The grade of membership varies from 0 to 1 depending on the number of equal characteristics between the municipality and the extreme profile. The higher the grade of membership of a municipality, the closer is it to the extreme profile. These indices are then combined to produce the synthetic index of Socioeconomic Dimension; high SDI values correspond to high-ranking positions in the first four dimensions and low in the fifth. For example, the municipality with grade of membership close to 1.0 for population concentration and dynamics axis is similar to its top profile, which has the largest urban population in 2000, highest population density, lowest urbanization level and rate, and highest rate of population growth. Top profile for social development index has high educational level, good medical care and garbage collecting systems, high percent of water and electricity supplies, street paving and illumination, and large number of households with 724 R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 Table 2 Populations, service domestic products (SDP) and gross domestic products (GDP) in 106 Reais, indices of service (IS), net-migration, inmigrants and out-migrants among the Amazon economic poles and other Brazilian municipalities within 1995 and 2000 State Macro-poles/ meso-poles Population Out-migrant In-migrant Migratory volume Net migration SDP GDP IS PA MT RO MA MA AC AP RR MT RO MA TO PA MT RO PA MT MT RO RO PA MT TO MT MA RO RO MA PA PA AC TO PA MA MT PA MA PA MA PA MT MA RO MT MT PA MA MT Belém a Cuiabáa Porto Velhoa São Luı́s a Manausa Rio Branco a Macapáa Boa Vistaa Rondonópolis b Ji-Paraná b Imperatrizb Palmasa Santarémb Sinopb Cacoalb Marabáb Cáceresb Barra do Garças b Vilhenab Ariquemesb Parauapebas b Tangará da Serrab Gurupib Alta Floresta Caxiasb Guajará-Mirim Jaru Bacabal b Tucuruı́b Redenção Cruzeiro do Sul b Araguaı́nab Altamira Codó Juı́na Itaituba Barra do Corda Paragominas Tefé b Almeirim Pontes e Lacerda Balsas Alta Floresta D’Oeste Colı́der Juara Breves Parintins Guarantã do Norte 1,280,614 483,346 334,661 870,028 1,405,835 253,059 283,308 200,568 150,227 106,800 230,566 137,355 262,538 74,831 73,568 168,020 85,857 52,092 53,598 74,503 71,568 58,840 65,034 46,982 139,756 38,045 53,600 91,823 73,798 63,251 67,441 113,143 77,439 111,146 38,017 94,750 78,147 76,450 64,457 33,957 43,012 60,163 26,533 28,051 30,748 80,158 90,150 28,200 105,809 34,696 19,087 48,333 44,964 11,922 12,364 10,157 12,627 15,080 32,005 8,308 31,228 6,456 11,836 18,327 7,728 5,694 5,573 9,442 6,266 6,719 7,535 8,874 8,845 3,248 11,411 9,579 7,892 10,141 4,290 14,176 9,794 6,692 4,697 21,247 7,742 10,615 4,874 6,644 5,640 3,085 2,038 6,162 3,231 7,590 6,153 2,934 57,432 24,201 20,935 55,479 85,569 15,888 28,764 28,817 8,520 8,423 17,020 38,759 12,212 12,081 5,955 18,246 6,832 4,221 8,844 6,806 12,721 6,776 5,328 3,348 6,265 2,126 3,272 4,302 9,597 6,545 2,665 10,982 5,977 3,902 3,305 5,651 3,468 11,108 3,170 2,834 4,856 5,448 2,034 2,525 1,742 3,567 3,709 3,256 163,241 58,897 40,022 103,812 130,533 27,810 41,128 38,974 21,147 23,503 49,025 47,067 43,440 18,537 17,791 36,573 14,560 9,915 14,417 16,248 18,987 13,495 12,863 12,222 15,110 5,374 14,683 13,881 17,489 16,686 6,955 25,158 15,771 10,594 8,002 26,898 11,210 21,723 8,044 9,478 10,496 8,533 4,072 8,687 4,973 11,157 9,862 6,190 48,377 10,495 1,848 7,146 40,605 3,966 16,400 18,660 4,107 6,657 14,985 30,451 19,016 5,625 5,881 81 896 1,473 3,271 2,636 6,455 57 2,207 5,526 2,580 1,122 8,139 5,277 1,705 3,596 1,625 3,194 3,817 2,790 1,392 15,596 4,274 493 1,704 3,810 784 2,363 4 3,637 1,489 4,023 2,444 322 5668.53 2956.53 2307.70 2636.66 4166.10 947.21 651.05 616.92 349.90 298.23 276.10 207.97 208.68 137.01 135.71 134.56 121.51 118.26 112.46 105.78 90.98 86.19 82.31 75.70 70.02 69.66 64.16 59.38 54.38 50.94 49.91 48.58 47.52 43.32 41.96 41.35 40.06 39.00 36.17 34.10 32.06 28.21 25.04 23.38 23.07 14.77 14.26 13.52 7676.87 3399.54 2472.55 3630.90 7159.99 1155.32 826.72 798.68 448.51 358.41 360.78 221.69 354.03 173.48 219.28 248.53 149.80 140.41 147.79 145.04 345.40 127.74 97.57 102.52 104.68 83.18 109.52 89.16 168.74 89.04 73.70 72.25 94.38 70.05 54.40 72.35 59.97 198.63 68.15 120.08 65.49 54.55 72.79 39.70 39.39 102.07 63.11 24.10 70.1 63.9 57.8 55.0 54.6 29.8 21.7 20.7 12.5 10.9 10.1 7.8 7.6 5.2 5.1 5.0 4.6 4.5 4.3 4.0 3.3 3.3 3.2 2.9 2.7 2.7 2.5 2.3 2.1 2.0 1.9 1.9 1.8 1.7 1.6 1.6 1.5 1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.9 0.6 0.5 0.5 Source: IBGE 2000 demographic census. Andrade and Serra (1999). a Macro-poles. b Meso-poles. R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 telephone and TV sets. Similar interpretation is valid for the grades of membership of the other three axes, whose variables show a direct relation to the top profiles. High values of the synthetic index SDI, therefore, combine high population concentration and growth with low urbanization, high economic development, high agrarian infrastructure, large agricultural and timber production, and low social development. Levels of socioeconomic dimensions were established using the natural breaks in the frequency distribution of this variable: low for values from 0 to 0.25; medium-low from 0.25 to 0.33; medium from 0.33 to 0.66; high medium from 0.66 to 0.75 and high from 0.75 to 1.00. Fig. 2c shows the distribution of municipalities according to the level of socioeconomic dimension. Of 729 municipalities, 7% have high socioeconomic dimension, whereas 22% have low. A comparison between the maps of density of deforested land (Fig. 2a) and the socioeconomic dimension index (Fig. 2c) demonstrates that there is a close match between municipalities with high deforested percentage and those with high to moderate SDI, notably in states of Mato Grosso, Rondônia and Acre, and more specifically in Eastern Pará and Western Maranhão state. Other municipalities with high SDI but low deforested density can be associated to regions with high deforestation within 2000 and 2001, such as the municipalities along the Cuiabá-Santarém highway in Southern Pará (Fig. 2b). In this way, this analysis also indicates that other areas with high SDI, but still low current deforestation, such as Santarém’s nearby municipalities, the municipalities surrounding Manaus towards Roraima and along the Amazon river, Aripuanã in Mato Grosso state, and Huimatá, in Amazonas state, along the Porto-Velho/Manaus highway, may potentially become hotspots of deforestation in the near future (Fig. 2c). Data recently released by DETER, INPE’s deforestation alerting system, agree with this interpretation (Valeriano et al. (2005). Moreover, the state of Mato Grosso, in which municipalities present the highest SDI, accounts for 48% of 26,130 km2 of deforestation estimated by INPE for the Brazilian Amazon in 2003–2004 (INPE, 2005). 4.3. Economic poles’ migration data Table 2 shows data for the number of in-migrants and out-migrants, the net migrants, and the overall 725 Fig. 2. (a) Density of deforested land % by 2001 (deforested land/ (municipality’s area—non-forest)), (b) 2000/2001 deforestation density % (deforestation/municipality’s area), and (c) socioeconomic dimension index for the Brazilian Amazon’s municipalities. Deforestation data come from PRODES (INPE, 2005). (1) CuiabáSantarém highway, (2) Santarém, (3) Manaus, (4) Huimatá, and (5) Aripuanã. 726 R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 migratory volume among each of the 48 regional poles and the remaining Brazilian municipalities between 1995 and 2000. There is an association between the polarizing capacity of a pole and its overall migration volume. In other words, the higher the economic importance of a pole, measured in terms of its IS, the larger is its overall migratory volume, even if its net migration is relatively low (Garcia et al., 2004). For example, although the macro-pole of São Luis and the micro-pole of Parauapebas possess similar net migration figures (7100 and 6400, respectively), São Luis’s overall migration is five-fold that of Paraupebas, which makes evident the greater importance and dynamism of São Luis. 4.4. Areas of influence of the economic poles Fig. 3. The micro- (a), meso- (b) and macro-poles (c), and their areas of influence for the Brazilian Amazon in 2000. Poles: Macapá, 1; Rio Branco, 2; Porto Velho, 3; Manaus, 4; São Luı́s, 5; Cuiabá, 6; Palmas, 7; Belém, 8; Boa Vista, 9; Ji-Paraná, 10; Cacoal, 11; Gurupi, 12; Sinop, 13; Vilhena, 14; Tangará da Serra, 15; Barra do Garças, 16; Cáceres, 17; Rondonópolis, 18; Santarém, 19; Tefé, 20; Tucuruı́, 21; Bacabal, 22; Caxias, 23; Marabá, 24; Imperatriz, 25; Parauapebas, 26; Araguaı́na, 27; Cruzeiro do Sul, 28; Almeirim, 29; Breves, 30; Guarantã do Norte, 31; Jaru, 32; Colı́der, 33; Guajará-Mirim, 34; Juı́na, 35; Pontes e Lacerda, 36; Parintins, 37; The index of interaction (Iv) measures a two-way influence: of the pole over remaining municipalities and of these over the pole. High values of this index mean high socioeconomic dimensions – translated as a high anthropic pressure – and large populations, both for the pole and municipality, as well as short distance and large migration exchange. Thus, high Iv can be interpreted as a strong connection between these two regions facilitated by proximity and population movement. As a result of this connection, an anthropic pressure gradient, influencing the deforestation process, is established from the pole to the satellite regions. The highest Iv between a given municipality and an economic pole defines which center is the pole of influence for this municipality. After the calculation of the Iv, a network of regions under influence of the economic poles was identified (Fig. 3). Of the 48 micro-poles, the municipality of São Luis polarized the highest number of municipalities; a total of 109. This effect may be ascribed not only to its economic importance but also to the large number of small municipalities in Maranhão state. The second pole was Belém, with 72 municipalities under influence, followed by Palmas, with 67. The centers with smallest number of municipalities were Almeirim Paragominas, 38; Altamira, 39; Ariquemes, 40; Itaituba, 41; Codó, 42; Barra do Corda, 43; Balsas, 44; Redenção, 45; Alta Floresta, 46; Alta Floresta D’Oeste, 47; Juara, 48. Acronyms for the Brazilian States: TO, Tocantins; PA, Pará; GO, Goiás; MT, Mato Grosso; RO, Rondônia; RR, Roraima; AP, Amapá; AC, Acre. R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 (2), Guajará Mirim (3) and Tefé (2). But, in terms of area, the poles influencing the largest regions were Manaus, Cuiabá, Belém, Altamira and Rio Branco. Manaus, a burgeoning economic metropolis situated in the heart of the forest, is a particular case, due to voluminous fiscal incentives conveyed to its industrial park. Altamira, the largest Brazilian municipality and thus an anomaly in itself, hinders a detailed view of the migratory flows within this region, and Rio Branco, in spite of its low economic importance, is the most remote Brazilian state capital. The set of municipalities polarized by the same micro-pole defines the microregions of influence, as depicted in Fig. 3a. Like the micro-pole regions, the meso-pole of São Luis had the highest number of influenced municipalities (143), followed by Belém (87) and Palmas (69). The 29 meso-poles polarize directly the nonpole municipalities, the other micro-poles and consequently the municipalities polarized by the micro-poles. The set of micro-poles and the area of influence polarized by the same meso-pole configure the meso-regions. Fig. 3b illustrates this configuration, in which the meso-region of São Luis comprises the micro-region of São Luis (109 municipalities) and the 26 municipalities of the micro-regions of Barra da Corda and 8 municipalities of the micro-region of Codó. The regional macro-poles directly polarize the nonpole municipalities plus other meso-poles and indirectly their micro-poles of influence. The set of municipalities polarized by a macro-pole is the macroregion of influence. Fig. 3c exhibits the state capital macro-poles and their areas of influence. One can observe that their areas of influence are not constrained by state boundaries. 5. Conclusion The urban network, in conjunction with the regional patterns of migration and the municipal socioeconomic dimensions, helps identify the Amazon frontier mobility, pointing out the current hotspots of deforestation – the deforestation arc – along with the emerging new frontiers. Fig. 4b portrays this integrated view, in which arrows indicate the deforestation trend towards the Amazon innermost frontiers. The bigger the arrow, the stronger is the front. Patterns of migration also 727 indicate that most of the in-migrants to these inner fronts are coming from nearby consolidated frontiers, showing the way the frontier perpetuates itself (Fig. 4a). Two particular regions stand out from these results. First, the high SDI values assigned to municipalities along the Cuiabá-Santarém highway in Southern Pará may account for the expectation of paving this road (Fig. 2c). Second, Manaus, although with high SDI, concentrates much of its potential for deforestation nearby, namely on its urban–rural fringe. This can be attributed to its large flow of urban in-migrants, who are attracted to the numerous jobs offered by its burgeoning industry park (Table 2). This effect also demonstrates that the economic model of Manaus is exogenous to the region under its influence. The integrated model presented here, considering the flow of people over economic poles and other centers, has added a new dimension to the grand regional compartments of the Brazilian Amazon, as first drawn up by Becker (1990) and later developed by Kampel et al. (2001) and Becker (2001). This type of model also has great potential for predicting environmental changes due to anthropic pressure, considering that the socioeconomic space network plays a decisive role in governing human settlement patterns. For example, its framework provides a basis for forecasting deforestation at different spatial levels, equivalent to the regions defined at each hierarchy level (SoaresFilho et al., 2006). In comparison to the previous regionalization methods of Lemos et al. (2003) and Garcia (2002), the approach presented here is more sensitive to territorial diversity, providing a wider range of spatial arrangements, as it employed, instead of microregions, municipal census data. Also, as exhibited in Fig. 3, all municipalities assigned to a particular region in each of the three hierarchy levels form continuous spatial clusters, demonstrating that the adopted method is highly consistent in terms of spatial continuity, a prime requirement for any regionalization method. The set of equations used in the present methodology was conceived to provide a general framework to map territorial organization and thus can be modified to incorporate different views of the urban network and its associated socioeconomic space. For example, geodetic distance could be replaced by other measures of distance that embody the concept of accessibility. In 728 R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 Fig. 4. (a) 1995–2000 migration net rates for the Brazilian Amazon municipalities and (b) major deforestation fronts, derived from the integrated analysis of the Amazon urban network, population movements and socioeconomic dimension index, laid over 2000–2001 deforestation hotspots from Alencar et al. (2004). addition, the gravitational mass, in the numerator of Eq. (2), can be modified to address different geographic approaches. Still, further population studies can incorporate patterns of migration Brazil, especially in Peru, Bolivia, and Guiana. The maps of the poles’ areas of influence, presented in Fig. 3, are not meant to define regions in a strict sense, but rather to depict the way the Amazon socioeconomic space is structured in relation to its urban network and, consequently, how this spatial organization influences the deforestation process. In reality, there are no clear-cut boundaries between those regions, as all municipalities hold multiple interactions among themselves and with the identified R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 poles. Moreover, the territorial organization presented here reflects the Amazon socioeconomic space at the turn of the millennium; this picture is expected to change as new economic centers emerge and other urban connections are established. It is particularly interesting to note that the identified regions depict a nested spatial pattern not constrained by state boundaries. This socioeconomic layout not only highlights the diversity of Amazon territory, but may also be useful for redirecting interstate public policies and other proactive measures aiming to conciliate socioeconomic development with the conservation of the Amazon’s natural heritage. Acknowledgments This study is part of the ‘‘Amazon Scenarios’’ project. We thank the Gordon and Betty Moore Foundation and LBA-ECO – the Large-Scale Atmosphere Biosphere Experiment in Amazonia – for funding. Second author also receives support from CNPQ (Conselho Nacional de Desenvolvimento Cientı́fico e Tecnológico). References Alencar, A., Nepstad, D., McGrath, D., Moutinho, P., Pacheco, P., Diaz, M.C.V., Soares-Filho, B.S., 2004. Desmatamento na Amazônia: Indo Além da Emergência Crônica. IPAM, Belém, 85 pp. Andrade. T.A., Serra. R.V., 1999. Estimativas para o Produto Interno Bruto dos Municı́pios Brasileiros: 1975, 1980, 1985 e 1996. www.nemesis.org.br/docs/thomp5.pdf (last accessed 12 December 2005). Armenteras, D., Rudas, G., Rodriguez, N., Sua, S., Romero, M., 2006. Patterns and causes of deforestation in the Colombian Amazon. Ecol. Indicators 6, 353–368. Becker, B.K., 2001. Revisão das polı́ticas de ocupação da Amazônia: é possı́vel identificar modelos para projetar cenários? Revista Parcerias Estratégicas 12, 135–159. Becker, B.K., 1990. Amazônia. Ática, São Paulo, 122 pp. Carpentier, C.L., Vosti, S.A., Witcover, J., 2000. Intensified production systems on western Brazilian Amazon settlement. Ecosyst. Environ. 82, 73–88. Christaller, W., 1966. Central Places in Southern Germany. PrenticeHall, New Jersey, 230 pp. Garcia, R.A., 2002. A migração como variável endógena: uma proposta de regionalização baseada em pólos econômicos e suas áreas de influência. Ph.D. Thesis. University of Minas Gerais, 181 pp. 729 Garcia, R.A., Lemos, M.B., Carvalho, J.A.M., 2004. As transformações das áreas de influência migratória dos pólos econômicos brasileiros nos perı́odos 1980–1991 e 1991–2000. Revista Brasileira de Estudos de População 22, 259–282. Gold, D.T., Woodbury, M.A., George, L.K., 1990. Relationship classification using grade of membership analysis—a typology of sibling relationships in later life. J. Gerontol. 45 (2), 43–51. Harvey, D., 1989. The Condition of Postmodernity. Basil Blackwell, London, 378 pp. Hughes, D.C., Fillenbaum, G.G., Woodbury, M.A., Siegler, I.C., 1996. Psychometric characteristics of the mini-mental state examination in a community population—a grade of membership analysis. Int. J. Geriatr. Psychiatr. 11, 439–455. INPE, Instituto Nacional de Pesquisas Espaciais, 2005. Monitoramento da Floresta Amazônica. www.obt.inpe.br/prodes (last accessed 10 December 2005). IPEA/IBGE/NESUR, 1999. Caracterização e Tendências da Rede Urbana do Brasil: Desenvolvimento Regional e Estruturação da Rede Urbana. Universidade Estadual de Campinas, Campinas, 127 pp. Kampel S.A., Câmara, G., Monteiro, A.M.V., 2001. Configurações Espaciais do Processo de Desflorestamento da Amazônia. www.dpi.inpe.br/geopro/modelagem/relatorio_configuracoes_desmatamento.pdf (last accessed 09 December 2005). Laurance, W.F., 1999. Reflections on tropical deforestation crisis. Biol. Conserv. 91, 109–117. Lemos, M.B., Diniz, C.C., Guerra, L.P., 1999. Pólos econômicos do nordeste e suas áreas de influência: uma aplicação do modelo gravitacional utilizando Sistema de informações geográficas (SIG). Revista Econômica do Nordeste 30, 568–584. Lemos, M.B., Diniz, C.C., Guerra, L.P., Moro, S., 2003. A nova configuração regional brasileira e sua geografia econômica. Revista Estudos Econômicos 33, 665–700. Manton, K., Woodbury, M., Tolley, D., 1994. Statistical Applications Using Fuzzy Sets. John Wiley & Sons, New York, 312 pp. Margulis, S., 2002. Quem são os agentes dos desmatamentos na Amazônia e por que eles desmatam? www.obancomundial.org/ index.php/content/view_folder/87.html (last accessed 13 December 2005). Mertens, B., Poccard-Chapuis, R., Piketty, M.G., Lacques, A.E., Venturieri, A., 2002. Crossing spatial analyses and livestock economics to understand deforestation processes in the Brazilian Amazon: the case of São Félix do Xingú in South Pará. Agric. Econ. 27, 269–294. Monteiro, M.P., Sawyer, D., 2001. Diagnóstico demográfico, socioeconômico e de pressão antrópica na região da AmazôniaLegal. In: Capobianco, J.P.R., Verı́ssimo, A., Moreira, A., Sawyer, D., Santos, I., Pinto, L.P. (Eds.), Biodiversidade na Amazônia Brasileira: Avaliação e Ações Prioritárias Para a Conservação, Uso Sustentável e Repartição de Benefı́cios. Instituto Socioambiental & Estação Liberdade, São Paulo, pp. 308–320. Nepstad, D., McGrath, D., Alencar, A., Barros, A.C., Carvalho, G., Santilli, M., Diaz, M.C.V., 2002. Frontier governance in Amazonia. Science 295, 629–631. Nepstad, D., Verı́ssimo, A., Alencar, A., Nobre, C., Lima, E., Lefebrve, P., Schlesinger, P., Potter, C., Moutinho, P., Mendoza, 730 R.A. Garcia et al. / Ecological Indicators 7 (2007) 719–730 E., Cochrane, M., Brooks, V., 1999. Large-scale impoverishment of Amazonian forests by logging and fire. Nature 398, 505–508. Sawyer, D.O., Beltrão, K.I., 1991. ‘‘Healthy households’’ and child survival in Brazil. In: Demographic and Health Surveys World Conference, Washington, DC, August 5–7. IRD/Macro International, Inc., Columbia, pp. 205–221. Skole, D.L., Chomentowski, W.H., Salas, W.A., Nobre, A.D., 1994. Physical and human dimensions of deforestation in Amazonia. BioScience 44, 314–322. Soares-Filho, B.S., Alencar, A., Nepstad, D., Cerqueira, G.C., Diaz, M.C.V., Rivero, S., Solórzano, L., Voll, E., 2004. Simulating the response of land-cover changes to road paving and governance along a major Amazon highway: the santarém-cuiabá corridor. Global Change Biol. 10, 745–764. Soares-Filho, B.S., Nepstad, D.C., Curran, L.M., Cerqueira, G.C., Garcia, R.A., Ramos, C.A., Voll, E., McDonald, A., Lefebvre, P., Schlesinger, P., 2006. Modelling conservation in the Amazon basin. Nature 440, 520–523. Stern, D., Common, M., Barbier, E., 1996. Economic growth and environmental degradation: the environmental kuznets curve and sustainable development. World Dev. 24, 1151– 1160. Valeriano, D.M., Shimabukuro, Y.E., Duarte, V., Anderson, L.O., Espı́rito-Santo, F., Arai, E., Maurano, L.E., Souza, R.C., Freitas, R.M., Aulicino, L., 2005. Detecção do desflorestamento da Amazônia Legal em tempo real—projeto DETER. In: Simpósio Brasileiro de Sensoriamento Remoto, vol. XII, Goiânia, Brasil, April 16–21. INPE, São José dos Campos, pp. 3403–3409. Verı́ssimo, A., Arima, E., Lima, E., 2001. O Diagnóstico do Uso da Terra na Amazônia: exploração madeireira, agricultura e agropecuária. In: Capobianco, J.P.R., Verı́ssimo, A., Moreira, A., Sawyer, D., Santos, I., Pinto, L.P. (Eds.), Biodiversidade na Amazônia Brasileira: Avaliação e Ações Prioritárias Para a Conservação, Uso Sustentável e Repartição de Benefı́cios. Instituto Socioambiental & Estação Liberdade, São Paulo, pp. 327– 337. Wood, C., Porro, R., 2000. In: Wood, C., Porro, R. (Eds.), Deforestation and Land Use in the Amazon. University Press of Florida, Gainesville, 400 pp.