- Email: [email protected]
Land use changes in protected areas and their future: The legal effectiveness of landscape protection
Land Use Policy 38 (2014) 378–387
Contents lists available at ScienceDirect
Land Use Policy journal homepage: www.elsevier.com/locate/landusepol
Land use changes in protected areas and their future: The legal effectiveness of landscape protection Talita Nogueira Terra a,∗ , Rozely Ferreira dos Santos b , Diógenes Cortijo Costa c a b c
LABGEO, FEAGRI, UNICAMP, Av. Cândido Rondon, 501 – Barão Geraldo, Campinas, SP, Brazil Depto of Ecology, USP, Rua do Matão, 321 – Trav. 14, Cid. Universitária, São Paulo, SP, Brazil Department of Geotechnical and Transportation, UNICAMP, Av. Albert Einstein, 951, 13083-970 Campinas, SP, Brazil
a r t i c l e
i n f o
Article history: Received 6 September 2012 Received in revised form 28 October 2013 Accepted 2 December 2013 Keywords: Future land cover Landscape Protected area
a b s t r a c t It is expected that the application of a restrictive legal instrument would be an important barrier to human pressures on protected areas in Brazil. One aspect that remains to be determined is whether the applied restrictions will be related to the quality of scenarios at the borders of protected areas. The objective of this work was to analyze the capacity for minimizing the impacts on two protected areas and to identify the effective function of the barrier imposed by an environmental legal border. The borders of two protected areas, the Despraiado Sustainable Development Reserve and the Jureia-Itatins State Ecological Station, as well as the corresponding buffer zone were studied. The historical evolution of the land cover/land use of these regions was analyzed by dividing the regions into 900 m2 hexagonal units. The scenarios for the years 1962, 1980 and 2007 were overlaid for each hexagon. The hexagons were classified according to the possible effects of conservation, and the results were quantified in terms of the frequency of land use and ecological flows. A simulation of future land use in 2028 was performed using the Kappa index, Markov chain modeling, multi-criteria analysis and cellular automata modeling. Based on the trend for the last 45 years, a very dynamic interaction at the legal boundaries was identified; in certain cases, either conservation or degradation were stimulated, and the intended objectives of legal environmental measures were never fulfilled. The simulation showed that by 2028, the frontiers of these protected areas will retain less than 10% of the natural vegetation cover, and 43% of this area will be covered with banana plantations. © 2013 Elsevier Ltd. All rights reserved.
Introduction After 500 years of continuous fragmentation, most of the remaining Atlantic Forest has been disturbed in some way. Throughout the last century, land use/land cover changes (LUCC) have led to a high rate of deforestation and a subsequent recovery of the deforested area. These changes created a fragmented landscape dominated by young secondary forests (Teixeira et al., 2009). As one of the consequences of this process, a large number of border types were established between the forest and areas of human use. These frontiers are the result of spatial interaction patterns among neighboring patches and can be understood both as natural borders between ecosystems and as man-made borders. In legal frontiers of natural areas, adjacent communities have a characteristic set of species combined with those from another community. Thus, there is a superposition of species that strive for resources in an equilibrium condition. In contrast, borders created by human action have
∗ Corresponding author. +55 19 33057926. E-mail addresses: [email protected] (T.N. Terra), [email protected] (R.F. dos Santos). 0264-8377/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.landusepol.2013.12.003
a negative influence on the adjacent regions, affecting the density and composition of forest communities, which could lead to changes and strong imbalances (Laurance et al., 1998). This pattern is a very common situation in the areas under protection in Brazil, which suffer strong pressures along their borders. As stated by Roldán-Martín et al. (2006), ecological flows occur along these narrow borders, which define the major or minor damages occurring within the forests. Frequently, it is expected that the implementation of a highly restrictive legal environmental act will result in the formation of a more efficient barrier to human pressure within the protected area. This assumption leads to the following question: is the intensity of such restrictions related to the quality of the border scenarios? If this premise is true, one would expect that there is a damping of the human impacts compatible with the rigidity of the legal act within a certain timeframe. According to Valverde et al. (2008) and Carranza et al. (2007), it is possible to evaluate border patterns over time to understand the effects of forest exploitation by humans as well as the environmental quality and characteristics of the landscape. Nevertheless, according to Teixido et al. (2010), studies rarely relate the extent of landscape conservation over time to the attributes of the borders.
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
379
Fig. 1. Location of the study area (circled in black) on the coast of Sao Paulo – BR, covering three regions: the old Sustainable Development Reserve of Despraiado, the Jureia-Itatins State Ecological Station and the buffer zone.
Based on such considerations, this study emphasizes the possibility of recognizing the role of legal regulations in environmental protection over time in border areas with different restrictions regarding land use. The objective was to analyze and compare the buffering capacity of impacts on two protected areas as well as to verify the effective function of a legal barrier. If past scenarios are well recognized, it is possible to deduce how and where the borders will be located within few years and to support future political decisions regarding these areas. Materials and methods Study area The study area is located in southern Sao Paulo State (Brazil) and includes three adjacent regions of 100 ha each that have been subject to different legal restrictions: the Despraiado Sustainable Development Reserve (SDR), the Jureia-Itatins State Ecological Station (SES) and the Buffer Zone (BZ) (Fig. 1). Construction of the historical series of land use To determine the land use changes, scenarios of the three areas were constructed for the years 1962, 1980, 2001 and 2007 (Fig. 2). For 1962–2001, digitized aerial photographs were obtained using a Vexcel Ultrascan 5000 photogrammetric scanner with 1200 dpi (dots per inch) resolution. All required radiometric and geometric corrections were performed, resulting in panchromatic images in TIFF format. A World View satellite image with 0.5 m PAN (panchromatic channel) spatial resolution (supplied by Forest Foundation, São Paulo, Brazil) was used for mapping the land use in 2007.
The WGS84 coordinate system was used. To use the digital georeferenced data in ArcGIS 9.2 software, five ground control points (GCPs) were obtained for each year using a TOPCON HIPER® LITE+ receiver. The GCPs were distributed throughout the image, generating an RMS error of less than 5.5 m. A first-order polynomial was applied using the nearest-neighbor procedure. The georeferenced materials were orthorectified using the software ENVI 3.5 (ENVI, 2001) with the Digital Terrain Model (IGC – Geographic and Cartographic Institute of Sao Paulo State, 2004) and the altitudes obtained from the TOPCON receiver with adjustments for local geoid undulation (orthometric altitudes). The LUCC was mapped using a visual photo-interpretation of the geo-ortho-materials observed at a nominal scale of 1:3000. The mapped categories included secondary medium tropical rainforest, secondary initial tropical rainforest, tropical rainforest with banana plantation, pasture, agriculture, bare soil and human constructions. Simulating future conditions The 2028 land use was calibrated and simulated in four phases: (a) Markov Chain modeling; (b) the creation of maps of the landscape-driving forces; (c) multi-criteria analysis; and (d) the use of Markov Chains and cellular automata algorithms to generate a simulated map of land use/land cover (Fig. 2). All of the phases were constructed using the software IDRISI Andes (Kamusoko et al., 2009; Valente and Vettorazzi, 2008; Scarassatti, 2007). This method was applied to the LUCC maps in raster format with a pixel size of 3 m × 3 m. The calibration was performed from 2001 onward, adjusting the simulation for 1962 and 1980. First, a Markov Chain analysis was applied with a time interval of 18 years (from 1962 to 1980) for the maps, and an interval of 21 years was
380
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
Land use and cover change Land use maps by GIS ArcGis 9.2 from SDR, SES, ZR
Model Calibration Phase 1 - Markov Chain Phase 3 - Multi-criteria analysis
Area transition matrix Maps of 1962 from aerial photographs Maps of 1980 from aerial photographs Maps of 2001 from aerial photographs Maps of 2007 from sattelite image
Phase 4 - Markov chain cellular automata simulation
Final map from simulation (2001)
Final map with landscape factors and constraints Phase 2 - Creating the landscape factors maps and constrains Slope
Model validation
Influence of access routes
Kappa index
Influence of the agriculture areas
Formation of core residencial Simulation of future scenarios Phase 1 - Markov Chain Phase 3 - Multi-criteria analysis
Area transition matrix
Final map with landscape factors and constraints
Phase 2 - Creating the landscape factors maps and constraints
Phase 4 - Markov chain cellular automata simulation
Final map from simulation (2028)
Slope
Influence of access routes Influence of the agriculture areas Formation of core residencial Fig. 2. Description of the method for the simulation of the 2028 land cover.
projected from the second map with a proportional error of 0.15 (Pontius, 2000). The Markov chain analysis produced a matrix of transition area that was used in phase (d). The study areas were leaded for five driving forces, slopes, routes, agriculture areas and core residential areas, as factors of the landscape and the area of secondary tropical rainforest as a constraint (Table 1). The driving forces were analyzed in pair-wise comparison matrices to generate relative weights (Table 2) analyzed using an analytical hierarchy process (AHP) according to the consistency ratio. The consistency ratio is the probability that the weights from the landscape factors were randomly obtained (Saaty and Vargas, 1991). As shown in Table 2, the obtained consistency ratio ranged from 0.05 to 0.10, which was considered acceptable. The constraint factor was multiplied by the other factors to eliminate those areas without secondary tropical rainforest (Fig. 3). New maps were developed for each mapped category considering the driving forces with their respective values using a multi-criteria analysis. The seven generated maps and the matrix of transition areas were inserted into a cellular automata module coupled to the Markov chains. An iteration number of 21 were adopted as a function of the future year to be simulated. A filter was applied to the cellular automata with a 5 × 5 kernel size. Calibration of the land use change models required a time-series of reliable and consistent land use maps, which was constructed based on data from 2001. The year 2001 was simulated, and this year was compared with the real map based on the Kappa index in the software Idrisi Andes. This process was repeated until an acceptable similarity index was obtained for the calibration of the model (0.53). The procedure was repeated to simulate the landscape in 2028 using the maps from 1980 and 2007 as input. The maps of landscape factors were elaborated in accordance with the method adopted for the calibration; however, these factors were based on a land use/land cover map from 2007.
Table 1 Driving forces used to select the landscape factors and the constraints in the studied landscape. Factors of the landscape
Driving force
Slope
Related to erosion, especially in the larger areas, and the soil quality for planting. Erosion is associated with red-yellow argisols and typic cambisols (SMA, in press) with inclinations between 0 and 45◦ . Priority was given to less-erodible soils with lower slope inclination (Rosa, 2000; Bertoni and Lombardi Neto, 1990) Access roads are corridors of land-use relationships, influencing the spatial distribution of uses. The access routes represent relationships with fragmentation, housing density, agriculture and grazing (Espírito-Santo et al., 2004; Hawbaker et al., 2004; Geneletti, 2003) Bananas are the most important plantation crop in the area. Banana crops are the main income source of the resident population in the study area. We observed that since 1962, banana plantations have been growing without boundaries, and as such, these plantations are another factor of the landscape Humans are social beings; therefore, if one building is found in a certain place, the probability of finding another one next to it in time (x + 1) is high
Influence of routes
Influence of agricultural areas
Core residential areas
Constraints of the landscape
Driving force
Secondary medium tropical rainforest
The secondary medium tropical rainforest is a forest formation that originates at time (x + 1) because within the proposed time this forest type should take the place of secondary initial tropical rain forest; alternatively, the area should already be in this stage (the medium rainforest) because the category also includes the probable late successional state
Table 2 Pair-wise comparison matrix of the relative weights for each landscape factor. Categories: (A) Secondary medium tropical rainforest; (B) secondary initial tropical rainforest; (C) tropical rainforest with bananas; (D) pasture; (E) agriculture; (F) bare soils; (G) human construction. Landscape factors
Land use classes ponderation A
B
Map of slope Map of influence of access roads Map of formation of core residential areas Map of influence of the agriculture areas Consistency ratio Landscape factors
Map of influence of access roads
Map of formation Map of of core influence of residential areas the agriculture areas
1 1/3
1
1/7
1/3
1
1/5
1/3
3
1
0.05
C Map of influence of access roads
1 3
1
1/5
1/7
1
5
3
7
1
0.09
D
Map of slope
Map of influence of access roads
Map of formation Map of influence of core of the agriculture residential areas areas
1 5
1
3
1/3
1
9
5
3
1
0.10
Map of slope
Map of influence of access roads
Map of formation Map of influence of core of the agriculture residential areas areas
1 5
1
3
1/3
1
7
3
5
1
0.05
Land use classes ponderation E Map of slope
Map of slope Map of influence of access roads Map of formation of core residential areas Map of influence of the agriculture areas Consistency ratio
Map of formation Map of influence of core of the agriculture residential areas areas
Map of slope
F Map of influence of access roads
Map of formation of core residential areas
1 9
1
3
1/5
1
9
3
5
0.07
Map of influence of the agriculture areas
1
Map of slope
G Map of influence of access roads
Map of formation of core residential areas
1 9
1
3
1/7
1
7
1/3
5
0.06
Map of influence of the agriculture areas
1
Map of slope
Map of influence of access roads
Map of formation of core residential areas
1 7
1
9
3
1
7
1/3
1/3
Map of influence of the agriculture areas
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
Map of slope
1
0.09
381
382
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
Fig. 3. A map was constructed for each landscape factor and analyzed in a pair-wise comparison matrix.
Comparison of past and future land uses under different legal protections Land uses for 1962, 1980, 2007 and 2028 under different legal protections were compared by fractioning the maps in a hexagonal mesh with a unit area of 900 m2 using the Patch Analyst ArcView GIS extension (Rempel et al., 1998). This form was selected due to the necessity of building a comparable mosaic covering all of the maps with the hexagons in the same position. Furthermore, this dimension represents the smallest size that is computationally feasible to represent the reality of the area (Birch et al., 2007). The land use hexagons were overlapped in pairs. When more than one type of land use was found in one hexagon, the type of land use with the greatest impact was considered based on the following sequence of potential damage: secondary medium tropical rainforest < secondary initial tropical rainforest < tropical rainforest with banana plantation < pasture < agriculture < bare soil < human constructions. This decision was based on the fact
that, as a general rule, the use with the greatest impact enlarges its borders over time and dominates the surrounding space. These maps were overlaid, allowing for a historical analysis of land use over time from 1962 to 2028 at each hexagon. When a hexagon presented a land use type with a greater impact than at an earlier date, the change was classified as contrary to conservation; when the land use had less of an impact, the change was considered as favorable to conservation (Terra and dos Santos, 2012). The comparison of the changes in hexagons over time in areas with different levels of legal protection facilitated an evaluation of whether the land use responses were consistent with the expected goals of the respective legal acts. Results and discussion In 1980, the study area was considered an environmental protection area (EPA; IUCN category Ib); therefore, there were few environmental restrictions and only one border between the EPA
Fig. 4. Representation of the scenarios included in the legal-historical evolution of the study area.
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
383
Fig. 5. Rates (%) of deforestation and recovery for each portion of the study area and for each time period.
and the BZ (Fig. 4). In this strip, land use changes and deforestation occurred mainly in the BZ (Fig. 4 and Fig. 5), as expected, although the land use at the border, which is protected by law, was beyond that expected. In 1986, this protected area was transformed into a State Ecological Station (SES) and, as such, was subjected to
strong environmental restrictions. The expected legal result was that the forest would completely recover in the protected area and that there would be a restricted use at the BZ, thereby reducing the fragmentation at the border. However, the results of this study demonstrate that 15 years after the promulgation of this SES, the
384
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
Fig. 6. Analysis of changes in land use from 1962 to 1980, 1980 to 2007 and 2007 to 2028 in terms of conservation.
Fig. 7. (7.0) Relative frequency of the classes of land use and land cover of the series for every area of study. (7.1) Relative frequency of the classes of land use and land cover for the 2028 scenario for each unit of the study area. Categories: (A) Secondary medium tropical rainforest; (B) secondary initial tropical rainforest; (C) tropical rainforest with bananas; (D) pasture; (E) agriculture; (F) bare soils; (G) human construction.
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
385
Fig. 8. Possible land cover scenario for the study area in 2028.
situation was quite different. For example, the protected border had the effect of slightly improving conservation, increasing the forest area by only 12.7% at an intermediate state of recovery. Only 4.1% of these areas were recovered (Fig. 5), although there were no
management actions for recuperation. It is important to emphasize that until 2005, this region was subject to one Law, seven decrees (six from the Federal government and one from state government) and an environmental protection ordinance. All of these
386
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
instruments were designed to maximize the environmental protection; however, the maps indicated that the area remained under economic exploitation. As suggested by data from Fig. 4, although the legislation proclaimed that from 1980 onward, houses existing within the SES should be disassembled and the owners relocated, the number of houses and the area of housing increased, although at a lower intensity. In 2006, a portion of this area was designated as a Sustainable Development Reserve (SDR, or category VI of IUCN), which is considered much less restrictive than an SES with a large opportunity for human use, especially due to social pressure and conflicts with the traditional population that settled in this region. The result of this opportunity is that the land use at these borders increased, restricting the already reduced possibilities for environmental conservation (Fig. 6). At this time, the study area comprised three borders: the first was located between the BZ and SES; the second was located between the SES and the SDR; and the third was located between SDR and BZ. Approximately 75% of the forest areas in intermediate regeneration stages of the BZ became banana plantations. A similar conversion occurred in 34.6% in the medium-stage forest area of the SES (Fig. 5), whereas fewer changes occurred in the SDR area. In June 2009, another legal instrument re-established the area as an SES with only one border, despite losses during these four decades (Fig. 4). These results clearly demonstrate that the changes in the protected areas and borders are more strongly influenced by other forces than those of Law. Payés et al. (2013) measured the conservation success of an EPA, comparing its LUCC over time in different management categories and confirmed that the legal environmental restrictions were never fully implemented. The economic pressure in this very poor region, the lack of governmental support provided to the inhabitants within the protected area and the lack of surveillance were determinant factors for these conditions. The number of houses and the land use around them (especially banana plantations) increased independent of the legal acts (Fig. 6). These data suggest that the application of environmental regulations may direct the use of this area toward conservation or forest degradation, although such regulations are not the determinant elements of this process. The borders have slight effects, showing that higher restrictions are not mandatory to detain the advancement of its use. In addition, it became evident that the SDR status maintained the border condition better than the actual SES. This border is the most contradictory border between landscapes that was described in this study, considering that this area was under a legal instrument with high restrictions for 27 years. If these border areas continue on the same historical path as the last four decades, there is an extremely high probability that the land use in 2028 will be dominated by banana plantations (Fig. 7.0) as both monocultures and consorted monocultures with forests, indicative of a linear expansion (Fig. 7.1). The simulation also suggests that the anthropic fields will suffer a similar process to that experienced by the dense ombrophilous montane forest, which underwent an anthropic process and is now at an intermediate stage of secondary succession with a decreasing forest area. If this area suffers a similar evolution during the next 18 years, one can assume that the borders between the protected areas will have almost the opposite effect as that intended by the legal propositions (Fig. 8). Specifically, if the historical evolution of land use and the actions of the state government continue on the same pattern as in the last 45 years, the border areas between the SES, BZ and SDR will contain less than 10% of natural vegetation area, 42% of the area will be banana plantations and 33% of the area will be banana trees mingled with forest fragments. The simulation predicts that the aforementioned montane forest will be more abundant at the border of the SDR and not at the SES, as expected. It is quite probable that the forest will be reduced to a small scarp region at the south of
Table 3 Evaluation of the effect of changes in land use and land cover in the buffer zone (BZ), Juréia Itatins State Ecological Station (SES) and the Sustainable Development Reserve of Despraiado (SDR) based on the relative frequency (%) during the periods from 1962 to 1980, from 1980 to 2007 and from 2007 to 2028. Direction of land-use and land-cover change
Study regions BZ
SES
SDR
1962–1980 Favorable to conservation In opposition to conservation No significant changes
9 54 37
24 49 28
6 74 20
1980–2007 Favorable to conservation In opposition to conservation No significant changes
11 28 61
4 81 15
28 39 33
2007–2028 Favorable to conservation In opposition to conservation No significant changes
1 64 35
5 81 14
25 73 2
the SDR independent of the degree of legal protection. In addition, the simulation data suggest that the greatest probable outcome is that the banana plantations will form a core, making the legal borders indistinct and therefore increasingly fictitious. As a consequence, the border areas would completely lose their barrier role and, in contrast to their intended function, would become a threat to conservation (Table 3). In other words, if there is no urgent and radical change in the performance of the state management agency with respect to these borders, there will be no forest to protect nor a territory to manage by 2028. Conclusions The vast majority of the data presented in this study demonstrate that the legal borders of protection areas have been fictitious, as evidenced by intense change dynamics, which undermines the objectives and premises of the successive environmental protection units established for the area under study. The pressure on the boundary of land use overlaps the legal environmental restriction and the land use will never fulfilled the goals established for the restricted areas. If the 2028 land use occurs according to the simulation, banana plantations will dominate, making the legal borders indistinct and therefore increasingly fictitious. In this sense, little can be done in the near future to maintain the biological conservation of one of the most important Brazilian conservation areas. Acknowledgments This research was funded by the FAPESP (São Paulo Research Foundation). We would also like to thank for the Foundation Forest – Brazil, the administration of Jureia-Itatins State Ecological Station, Scientific Technical Committee (COTEC) and “BASE Aerofotogrametria e Projetos S.A.”. References Bertoni, J., Lombardi Neto, F., 1990. Conservac¸ão do solo. Ícone, São Paulo, 355 pp. Birch, C.P.D., Oom, S.P., Beecham, J.A., 2007. Rectangular and hexagonal grids used for observation, experiment and simulation in ecology. Ecological Modelling 206, 347–359. Carranza, M.L., et al., 2007. Analyzing landscape diversity in time: the use of Rènyi’s generalized entropy function. Ecological Indicators 7 (3), 505–510. ENVI, 2001. Guia do Envi 3.5, SulSoft Servic¸os de Processamento de Dados Ltda. Espírito-Santo, F. Del B., dos Santos, J.R., da Silva, P.G., 2004. Técnicas de processamento de imagens e de análise espacial para estudo de áreas florestais sob a explorac¸ão madeireira. Revista Árvore 28 (5). Geneletti, D., 2003. Biodiversity impact assessment of roads: an approach based on ecosystem rarity. Environmental Impact Assessment Review 23 (3), 343–365.
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387 Hawbaker, T.J., Radeloff, V.C., Hammer, R.B., Clayton, M.K., 2004. Road density and landscape pattern in relation to housing density, land ownership, land cover, and soils. Landscape Ecology 20, 609–625. Kamusoko, C., et al., 2009. Rural sustainability under threat in Zimbabwe – simulation of future land use/cover changes in the Bindura district based on the Markov-cellular automata model. Applied Geography 29, 435–447. Laurance, W.F., et al., 1998. Effects of forest fragmentation on recruitment patterns in Amazonian tree communities. Conservation Biology 12, 460–464. Terra, T.N., dos Santos, R.F., 2012. Measuring cumulative effects in a fragmented landscape. Ecological Modelling 228, 89–95. Payés, A.C.L.M., Pavão, T., dos Santos, R.F., 2013. The conservation success over time: evaluating the land use and cover change in a protected area under a long recategorization process. Land Use Policy 30, 177–185. Pontius Jr., R.G., 2000. Quantification error versus location error in comparison of categorical maps. Photogrammetric Engineering & Remote Sensing 66, 1011–1016. Rempel, R., Carr, A., Elkie, P., 1998. Patch Analyst 2.2. http://www.cnfer.on. ca/SEP/patchanalyst/ Roldán-Martín, M.J., de Pablo, C.L., de Agar, P., 2006. Landscape changes over time: comparison of land uses, boundaries and mosaics. Landscape Ecology 21, 1075–1088.
387
Rosa, M.C., 2000. Introduc¸ão a morfologia. Contexto, São Paulo, pp. 156. Saaty, T.L., Vargas, L.G., 1991. Prediction, Projection and Forecasting. Kluwer Academic Publishers, Boston, MA, USA. Scarassatti, D.F., Dissertac¸ão (Mestrado em concentrac¸ão de transportes) 2007. Modelagem Dinâmica na Projec¸ão de Uso do Solo em func¸ão da Rede Viária de Transportes, 119f, 2007. Faculdade de Engenharia Civil e Arquitetura, Universidade Estadual de Campinas, Campinas. Teixeira, A.M., et al., 2009. Modeling landscape dynamics in an Atlantic Rainforest region: implications for conservation. Forest Ecology and Management 257, 1219–1230. Teixido, et al., 2010. Impacts of changes in land use and fragmentation patterns on Atlantic coastal forests in northern Spain. Journal of Environmental Management 91 (4), 879–886. Valente, R.O.A., Vettorazzi, C.A., 2008. Definition of priority areas for forest conservation through the ordered weighted averaging method. Forest Ecology and Management 256 (6), 1408–1417. Valverde, V., et al., 2008. Análisis de la estructura espacial del paisaje: mosaicos del paisaje. In: Maestre, F.T., Escudero, A., Bonet, A. (Eds.), Introducción al análisis espacial de datos en ecología y ciencias ambientales: métodos y aplicaciones. Dykinson, S.L., Madrid, pp. 747–759.
Contents lists available at ScienceDirect
Land Use Policy journal homepage: www.elsevier.com/locate/landusepol
Land use changes in protected areas and their future: The legal effectiveness of landscape protection Talita Nogueira Terra a,∗ , Rozely Ferreira dos Santos b , Diógenes Cortijo Costa c a b c
LABGEO, FEAGRI, UNICAMP, Av. Cândido Rondon, 501 – Barão Geraldo, Campinas, SP, Brazil Depto of Ecology, USP, Rua do Matão, 321 – Trav. 14, Cid. Universitária, São Paulo, SP, Brazil Department of Geotechnical and Transportation, UNICAMP, Av. Albert Einstein, 951, 13083-970 Campinas, SP, Brazil
a r t i c l e
i n f o
Article history: Received 6 September 2012 Received in revised form 28 October 2013 Accepted 2 December 2013 Keywords: Future land cover Landscape Protected area
a b s t r a c t It is expected that the application of a restrictive legal instrument would be an important barrier to human pressures on protected areas in Brazil. One aspect that remains to be determined is whether the applied restrictions will be related to the quality of scenarios at the borders of protected areas. The objective of this work was to analyze the capacity for minimizing the impacts on two protected areas and to identify the effective function of the barrier imposed by an environmental legal border. The borders of two protected areas, the Despraiado Sustainable Development Reserve and the Jureia-Itatins State Ecological Station, as well as the corresponding buffer zone were studied. The historical evolution of the land cover/land use of these regions was analyzed by dividing the regions into 900 m2 hexagonal units. The scenarios for the years 1962, 1980 and 2007 were overlaid for each hexagon. The hexagons were classified according to the possible effects of conservation, and the results were quantified in terms of the frequency of land use and ecological flows. A simulation of future land use in 2028 was performed using the Kappa index, Markov chain modeling, multi-criteria analysis and cellular automata modeling. Based on the trend for the last 45 years, a very dynamic interaction at the legal boundaries was identified; in certain cases, either conservation or degradation were stimulated, and the intended objectives of legal environmental measures were never fulfilled. The simulation showed that by 2028, the frontiers of these protected areas will retain less than 10% of the natural vegetation cover, and 43% of this area will be covered with banana plantations. © 2013 Elsevier Ltd. All rights reserved.
Introduction After 500 years of continuous fragmentation, most of the remaining Atlantic Forest has been disturbed in some way. Throughout the last century, land use/land cover changes (LUCC) have led to a high rate of deforestation and a subsequent recovery of the deforested area. These changes created a fragmented landscape dominated by young secondary forests (Teixeira et al., 2009). As one of the consequences of this process, a large number of border types were established between the forest and areas of human use. These frontiers are the result of spatial interaction patterns among neighboring patches and can be understood both as natural borders between ecosystems and as man-made borders. In legal frontiers of natural areas, adjacent communities have a characteristic set of species combined with those from another community. Thus, there is a superposition of species that strive for resources in an equilibrium condition. In contrast, borders created by human action have
∗ Corresponding author. +55 19 33057926. E-mail addresses: [email protected] (T.N. Terra), [email protected] (R.F. dos Santos). 0264-8377/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.landusepol.2013.12.003
a negative influence on the adjacent regions, affecting the density and composition of forest communities, which could lead to changes and strong imbalances (Laurance et al., 1998). This pattern is a very common situation in the areas under protection in Brazil, which suffer strong pressures along their borders. As stated by Roldán-Martín et al. (2006), ecological flows occur along these narrow borders, which define the major or minor damages occurring within the forests. Frequently, it is expected that the implementation of a highly restrictive legal environmental act will result in the formation of a more efficient barrier to human pressure within the protected area. This assumption leads to the following question: is the intensity of such restrictions related to the quality of the border scenarios? If this premise is true, one would expect that there is a damping of the human impacts compatible with the rigidity of the legal act within a certain timeframe. According to Valverde et al. (2008) and Carranza et al. (2007), it is possible to evaluate border patterns over time to understand the effects of forest exploitation by humans as well as the environmental quality and characteristics of the landscape. Nevertheless, according to Teixido et al. (2010), studies rarely relate the extent of landscape conservation over time to the attributes of the borders.
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
379
Fig. 1. Location of the study area (circled in black) on the coast of Sao Paulo – BR, covering three regions: the old Sustainable Development Reserve of Despraiado, the Jureia-Itatins State Ecological Station and the buffer zone.
Based on such considerations, this study emphasizes the possibility of recognizing the role of legal regulations in environmental protection over time in border areas with different restrictions regarding land use. The objective was to analyze and compare the buffering capacity of impacts on two protected areas as well as to verify the effective function of a legal barrier. If past scenarios are well recognized, it is possible to deduce how and where the borders will be located within few years and to support future political decisions regarding these areas. Materials and methods Study area The study area is located in southern Sao Paulo State (Brazil) and includes three adjacent regions of 100 ha each that have been subject to different legal restrictions: the Despraiado Sustainable Development Reserve (SDR), the Jureia-Itatins State Ecological Station (SES) and the Buffer Zone (BZ) (Fig. 1). Construction of the historical series of land use To determine the land use changes, scenarios of the three areas were constructed for the years 1962, 1980, 2001 and 2007 (Fig. 2). For 1962–2001, digitized aerial photographs were obtained using a Vexcel Ultrascan 5000 photogrammetric scanner with 1200 dpi (dots per inch) resolution. All required radiometric and geometric corrections were performed, resulting in panchromatic images in TIFF format. A World View satellite image with 0.5 m PAN (panchromatic channel) spatial resolution (supplied by Forest Foundation, São Paulo, Brazil) was used for mapping the land use in 2007.
The WGS84 coordinate system was used. To use the digital georeferenced data in ArcGIS 9.2 software, five ground control points (GCPs) were obtained for each year using a TOPCON HIPER® LITE+ receiver. The GCPs were distributed throughout the image, generating an RMS error of less than 5.5 m. A first-order polynomial was applied using the nearest-neighbor procedure. The georeferenced materials were orthorectified using the software ENVI 3.5 (ENVI, 2001) with the Digital Terrain Model (IGC – Geographic and Cartographic Institute of Sao Paulo State, 2004) and the altitudes obtained from the TOPCON receiver with adjustments for local geoid undulation (orthometric altitudes). The LUCC was mapped using a visual photo-interpretation of the geo-ortho-materials observed at a nominal scale of 1:3000. The mapped categories included secondary medium tropical rainforest, secondary initial tropical rainforest, tropical rainforest with banana plantation, pasture, agriculture, bare soil and human constructions. Simulating future conditions The 2028 land use was calibrated and simulated in four phases: (a) Markov Chain modeling; (b) the creation of maps of the landscape-driving forces; (c) multi-criteria analysis; and (d) the use of Markov Chains and cellular automata algorithms to generate a simulated map of land use/land cover (Fig. 2). All of the phases were constructed using the software IDRISI Andes (Kamusoko et al., 2009; Valente and Vettorazzi, 2008; Scarassatti, 2007). This method was applied to the LUCC maps in raster format with a pixel size of 3 m × 3 m. The calibration was performed from 2001 onward, adjusting the simulation for 1962 and 1980. First, a Markov Chain analysis was applied with a time interval of 18 years (from 1962 to 1980) for the maps, and an interval of 21 years was
380
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
Land use and cover change Land use maps by GIS ArcGis 9.2 from SDR, SES, ZR
Model Calibration Phase 1 - Markov Chain Phase 3 - Multi-criteria analysis
Area transition matrix Maps of 1962 from aerial photographs Maps of 1980 from aerial photographs Maps of 2001 from aerial photographs Maps of 2007 from sattelite image
Phase 4 - Markov chain cellular automata simulation
Final map from simulation (2001)
Final map with landscape factors and constraints Phase 2 - Creating the landscape factors maps and constrains Slope
Model validation
Influence of access routes
Kappa index
Influence of the agriculture areas
Formation of core residencial Simulation of future scenarios Phase 1 - Markov Chain Phase 3 - Multi-criteria analysis
Area transition matrix
Final map with landscape factors and constraints
Phase 2 - Creating the landscape factors maps and constraints
Phase 4 - Markov chain cellular automata simulation
Final map from simulation (2028)
Slope
Influence of access routes Influence of the agriculture areas Formation of core residencial Fig. 2. Description of the method for the simulation of the 2028 land cover.
projected from the second map with a proportional error of 0.15 (Pontius, 2000). The Markov chain analysis produced a matrix of transition area that was used in phase (d). The study areas were leaded for five driving forces, slopes, routes, agriculture areas and core residential areas, as factors of the landscape and the area of secondary tropical rainforest as a constraint (Table 1). The driving forces were analyzed in pair-wise comparison matrices to generate relative weights (Table 2) analyzed using an analytical hierarchy process (AHP) according to the consistency ratio. The consistency ratio is the probability that the weights from the landscape factors were randomly obtained (Saaty and Vargas, 1991). As shown in Table 2, the obtained consistency ratio ranged from 0.05 to 0.10, which was considered acceptable. The constraint factor was multiplied by the other factors to eliminate those areas without secondary tropical rainforest (Fig. 3). New maps were developed for each mapped category considering the driving forces with their respective values using a multi-criteria analysis. The seven generated maps and the matrix of transition areas were inserted into a cellular automata module coupled to the Markov chains. An iteration number of 21 were adopted as a function of the future year to be simulated. A filter was applied to the cellular automata with a 5 × 5 kernel size. Calibration of the land use change models required a time-series of reliable and consistent land use maps, which was constructed based on data from 2001. The year 2001 was simulated, and this year was compared with the real map based on the Kappa index in the software Idrisi Andes. This process was repeated until an acceptable similarity index was obtained for the calibration of the model (0.53). The procedure was repeated to simulate the landscape in 2028 using the maps from 1980 and 2007 as input. The maps of landscape factors were elaborated in accordance with the method adopted for the calibration; however, these factors were based on a land use/land cover map from 2007.
Table 1 Driving forces used to select the landscape factors and the constraints in the studied landscape. Factors of the landscape
Driving force
Slope
Related to erosion, especially in the larger areas, and the soil quality for planting. Erosion is associated with red-yellow argisols and typic cambisols (SMA, in press) with inclinations between 0 and 45◦ . Priority was given to less-erodible soils with lower slope inclination (Rosa, 2000; Bertoni and Lombardi Neto, 1990) Access roads are corridors of land-use relationships, influencing the spatial distribution of uses. The access routes represent relationships with fragmentation, housing density, agriculture and grazing (Espírito-Santo et al., 2004; Hawbaker et al., 2004; Geneletti, 2003) Bananas are the most important plantation crop in the area. Banana crops are the main income source of the resident population in the study area. We observed that since 1962, banana plantations have been growing without boundaries, and as such, these plantations are another factor of the landscape Humans are social beings; therefore, if one building is found in a certain place, the probability of finding another one next to it in time (x + 1) is high
Influence of routes
Influence of agricultural areas
Core residential areas
Constraints of the landscape
Driving force
Secondary medium tropical rainforest
The secondary medium tropical rainforest is a forest formation that originates at time (x + 1) because within the proposed time this forest type should take the place of secondary initial tropical rain forest; alternatively, the area should already be in this stage (the medium rainforest) because the category also includes the probable late successional state
Table 2 Pair-wise comparison matrix of the relative weights for each landscape factor. Categories: (A) Secondary medium tropical rainforest; (B) secondary initial tropical rainforest; (C) tropical rainforest with bananas; (D) pasture; (E) agriculture; (F) bare soils; (G) human construction. Landscape factors
Land use classes ponderation A
B
Map of slope Map of influence of access roads Map of formation of core residential areas Map of influence of the agriculture areas Consistency ratio Landscape factors
Map of influence of access roads
Map of formation Map of of core influence of residential areas the agriculture areas
1 1/3
1
1/7
1/3
1
1/5
1/3
3
1
0.05
C Map of influence of access roads
1 3
1
1/5
1/7
1
5
3
7
1
0.09
D
Map of slope
Map of influence of access roads
Map of formation Map of influence of core of the agriculture residential areas areas
1 5
1
3
1/3
1
9
5
3
1
0.10
Map of slope
Map of influence of access roads
Map of formation Map of influence of core of the agriculture residential areas areas
1 5
1
3
1/3
1
7
3
5
1
0.05
Land use classes ponderation E Map of slope
Map of slope Map of influence of access roads Map of formation of core residential areas Map of influence of the agriculture areas Consistency ratio
Map of formation Map of influence of core of the agriculture residential areas areas
Map of slope
F Map of influence of access roads
Map of formation of core residential areas
1 9
1
3
1/5
1
9
3
5
0.07
Map of influence of the agriculture areas
1
Map of slope
G Map of influence of access roads
Map of formation of core residential areas
1 9
1
3
1/7
1
7
1/3
5
0.06
Map of influence of the agriculture areas
1
Map of slope
Map of influence of access roads
Map of formation of core residential areas
1 7
1
9
3
1
7
1/3
1/3
Map of influence of the agriculture areas
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
Map of slope
1
0.09
381
382
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
Fig. 3. A map was constructed for each landscape factor and analyzed in a pair-wise comparison matrix.
Comparison of past and future land uses under different legal protections Land uses for 1962, 1980, 2007 and 2028 under different legal protections were compared by fractioning the maps in a hexagonal mesh with a unit area of 900 m2 using the Patch Analyst ArcView GIS extension (Rempel et al., 1998). This form was selected due to the necessity of building a comparable mosaic covering all of the maps with the hexagons in the same position. Furthermore, this dimension represents the smallest size that is computationally feasible to represent the reality of the area (Birch et al., 2007). The land use hexagons were overlapped in pairs. When more than one type of land use was found in one hexagon, the type of land use with the greatest impact was considered based on the following sequence of potential damage: secondary medium tropical rainforest < secondary initial tropical rainforest < tropical rainforest with banana plantation < pasture < agriculture < bare soil < human constructions. This decision was based on the fact
that, as a general rule, the use with the greatest impact enlarges its borders over time and dominates the surrounding space. These maps were overlaid, allowing for a historical analysis of land use over time from 1962 to 2028 at each hexagon. When a hexagon presented a land use type with a greater impact than at an earlier date, the change was classified as contrary to conservation; when the land use had less of an impact, the change was considered as favorable to conservation (Terra and dos Santos, 2012). The comparison of the changes in hexagons over time in areas with different levels of legal protection facilitated an evaluation of whether the land use responses were consistent with the expected goals of the respective legal acts. Results and discussion In 1980, the study area was considered an environmental protection area (EPA; IUCN category Ib); therefore, there were few environmental restrictions and only one border between the EPA
Fig. 4. Representation of the scenarios included in the legal-historical evolution of the study area.
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
383
Fig. 5. Rates (%) of deforestation and recovery for each portion of the study area and for each time period.
and the BZ (Fig. 4). In this strip, land use changes and deforestation occurred mainly in the BZ (Fig. 4 and Fig. 5), as expected, although the land use at the border, which is protected by law, was beyond that expected. In 1986, this protected area was transformed into a State Ecological Station (SES) and, as such, was subjected to
strong environmental restrictions. The expected legal result was that the forest would completely recover in the protected area and that there would be a restricted use at the BZ, thereby reducing the fragmentation at the border. However, the results of this study demonstrate that 15 years after the promulgation of this SES, the
384
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
Fig. 6. Analysis of changes in land use from 1962 to 1980, 1980 to 2007 and 2007 to 2028 in terms of conservation.
Fig. 7. (7.0) Relative frequency of the classes of land use and land cover of the series for every area of study. (7.1) Relative frequency of the classes of land use and land cover for the 2028 scenario for each unit of the study area. Categories: (A) Secondary medium tropical rainforest; (B) secondary initial tropical rainforest; (C) tropical rainforest with bananas; (D) pasture; (E) agriculture; (F) bare soils; (G) human construction.
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
385
Fig. 8. Possible land cover scenario for the study area in 2028.
situation was quite different. For example, the protected border had the effect of slightly improving conservation, increasing the forest area by only 12.7% at an intermediate state of recovery. Only 4.1% of these areas were recovered (Fig. 5), although there were no
management actions for recuperation. It is important to emphasize that until 2005, this region was subject to one Law, seven decrees (six from the Federal government and one from state government) and an environmental protection ordinance. All of these
386
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387
instruments were designed to maximize the environmental protection; however, the maps indicated that the area remained under economic exploitation. As suggested by data from Fig. 4, although the legislation proclaimed that from 1980 onward, houses existing within the SES should be disassembled and the owners relocated, the number of houses and the area of housing increased, although at a lower intensity. In 2006, a portion of this area was designated as a Sustainable Development Reserve (SDR, or category VI of IUCN), which is considered much less restrictive than an SES with a large opportunity for human use, especially due to social pressure and conflicts with the traditional population that settled in this region. The result of this opportunity is that the land use at these borders increased, restricting the already reduced possibilities for environmental conservation (Fig. 6). At this time, the study area comprised three borders: the first was located between the BZ and SES; the second was located between the SES and the SDR; and the third was located between SDR and BZ. Approximately 75% of the forest areas in intermediate regeneration stages of the BZ became banana plantations. A similar conversion occurred in 34.6% in the medium-stage forest area of the SES (Fig. 5), whereas fewer changes occurred in the SDR area. In June 2009, another legal instrument re-established the area as an SES with only one border, despite losses during these four decades (Fig. 4). These results clearly demonstrate that the changes in the protected areas and borders are more strongly influenced by other forces than those of Law. Payés et al. (2013) measured the conservation success of an EPA, comparing its LUCC over time in different management categories and confirmed that the legal environmental restrictions were never fully implemented. The economic pressure in this very poor region, the lack of governmental support provided to the inhabitants within the protected area and the lack of surveillance were determinant factors for these conditions. The number of houses and the land use around them (especially banana plantations) increased independent of the legal acts (Fig. 6). These data suggest that the application of environmental regulations may direct the use of this area toward conservation or forest degradation, although such regulations are not the determinant elements of this process. The borders have slight effects, showing that higher restrictions are not mandatory to detain the advancement of its use. In addition, it became evident that the SDR status maintained the border condition better than the actual SES. This border is the most contradictory border between landscapes that was described in this study, considering that this area was under a legal instrument with high restrictions for 27 years. If these border areas continue on the same historical path as the last four decades, there is an extremely high probability that the land use in 2028 will be dominated by banana plantations (Fig. 7.0) as both monocultures and consorted monocultures with forests, indicative of a linear expansion (Fig. 7.1). The simulation also suggests that the anthropic fields will suffer a similar process to that experienced by the dense ombrophilous montane forest, which underwent an anthropic process and is now at an intermediate stage of secondary succession with a decreasing forest area. If this area suffers a similar evolution during the next 18 years, one can assume that the borders between the protected areas will have almost the opposite effect as that intended by the legal propositions (Fig. 8). Specifically, if the historical evolution of land use and the actions of the state government continue on the same pattern as in the last 45 years, the border areas between the SES, BZ and SDR will contain less than 10% of natural vegetation area, 42% of the area will be banana plantations and 33% of the area will be banana trees mingled with forest fragments. The simulation predicts that the aforementioned montane forest will be more abundant at the border of the SDR and not at the SES, as expected. It is quite probable that the forest will be reduced to a small scarp region at the south of
Table 3 Evaluation of the effect of changes in land use and land cover in the buffer zone (BZ), Juréia Itatins State Ecological Station (SES) and the Sustainable Development Reserve of Despraiado (SDR) based on the relative frequency (%) during the periods from 1962 to 1980, from 1980 to 2007 and from 2007 to 2028. Direction of land-use and land-cover change
Study regions BZ
SES
SDR
1962–1980 Favorable to conservation In opposition to conservation No significant changes
9 54 37
24 49 28
6 74 20
1980–2007 Favorable to conservation In opposition to conservation No significant changes
11 28 61
4 81 15
28 39 33
2007–2028 Favorable to conservation In opposition to conservation No significant changes
1 64 35
5 81 14
25 73 2
the SDR independent of the degree of legal protection. In addition, the simulation data suggest that the greatest probable outcome is that the banana plantations will form a core, making the legal borders indistinct and therefore increasingly fictitious. As a consequence, the border areas would completely lose their barrier role and, in contrast to their intended function, would become a threat to conservation (Table 3). In other words, if there is no urgent and radical change in the performance of the state management agency with respect to these borders, there will be no forest to protect nor a territory to manage by 2028. Conclusions The vast majority of the data presented in this study demonstrate that the legal borders of protection areas have been fictitious, as evidenced by intense change dynamics, which undermines the objectives and premises of the successive environmental protection units established for the area under study. The pressure on the boundary of land use overlaps the legal environmental restriction and the land use will never fulfilled the goals established for the restricted areas. If the 2028 land use occurs according to the simulation, banana plantations will dominate, making the legal borders indistinct and therefore increasingly fictitious. In this sense, little can be done in the near future to maintain the biological conservation of one of the most important Brazilian conservation areas. Acknowledgments This research was funded by the FAPESP (São Paulo Research Foundation). We would also like to thank for the Foundation Forest – Brazil, the administration of Jureia-Itatins State Ecological Station, Scientific Technical Committee (COTEC) and “BASE Aerofotogrametria e Projetos S.A.”. References Bertoni, J., Lombardi Neto, F., 1990. Conservac¸ão do solo. Ícone, São Paulo, 355 pp. Birch, C.P.D., Oom, S.P., Beecham, J.A., 2007. Rectangular and hexagonal grids used for observation, experiment and simulation in ecology. Ecological Modelling 206, 347–359. Carranza, M.L., et al., 2007. Analyzing landscape diversity in time: the use of Rènyi’s generalized entropy function. Ecological Indicators 7 (3), 505–510. ENVI, 2001. Guia do Envi 3.5, SulSoft Servic¸os de Processamento de Dados Ltda. Espírito-Santo, F. Del B., dos Santos, J.R., da Silva, P.G., 2004. Técnicas de processamento de imagens e de análise espacial para estudo de áreas florestais sob a explorac¸ão madeireira. Revista Árvore 28 (5). Geneletti, D., 2003. Biodiversity impact assessment of roads: an approach based on ecosystem rarity. Environmental Impact Assessment Review 23 (3), 343–365.
T.N. Terra et al. / Land Use Policy 38 (2014) 378–387 Hawbaker, T.J., Radeloff, V.C., Hammer, R.B., Clayton, M.K., 2004. Road density and landscape pattern in relation to housing density, land ownership, land cover, and soils. Landscape Ecology 20, 609–625. Kamusoko, C., et al., 2009. Rural sustainability under threat in Zimbabwe – simulation of future land use/cover changes in the Bindura district based on the Markov-cellular automata model. Applied Geography 29, 435–447. Laurance, W.F., et al., 1998. Effects of forest fragmentation on recruitment patterns in Amazonian tree communities. Conservation Biology 12, 460–464. Terra, T.N., dos Santos, R.F., 2012. Measuring cumulative effects in a fragmented landscape. Ecological Modelling 228, 89–95. Payés, A.C.L.M., Pavão, T., dos Santos, R.F., 2013. The conservation success over time: evaluating the land use and cover change in a protected area under a long recategorization process. Land Use Policy 30, 177–185. Pontius Jr., R.G., 2000. Quantification error versus location error in comparison of categorical maps. Photogrammetric Engineering & Remote Sensing 66, 1011–1016. Rempel, R., Carr, A., Elkie, P., 1998. Patch Analyst 2.2. http://www.cnfer.on. ca/SEP/patchanalyst/ Roldán-Martín, M.J., de Pablo, C.L., de Agar, P., 2006. Landscape changes over time: comparison of land uses, boundaries and mosaics. Landscape Ecology 21, 1075–1088.
387
Rosa, M.C., 2000. Introduc¸ão a morfologia. Contexto, São Paulo, pp. 156. Saaty, T.L., Vargas, L.G., 1991. Prediction, Projection and Forecasting. Kluwer Academic Publishers, Boston, MA, USA. Scarassatti, D.F., Dissertac¸ão (Mestrado em concentrac¸ão de transportes) 2007. Modelagem Dinâmica na Projec¸ão de Uso do Solo em func¸ão da Rede Viária de Transportes, 119f, 2007. Faculdade de Engenharia Civil e Arquitetura, Universidade Estadual de Campinas, Campinas. Teixeira, A.M., et al., 2009. Modeling landscape dynamics in an Atlantic Rainforest region: implications for conservation. Forest Ecology and Management 257, 1219–1230. Teixido, et al., 2010. Impacts of changes in land use and fragmentation patterns on Atlantic coastal forests in northern Spain. Journal of Environmental Management 91 (4), 879–886. Valente, R.O.A., Vettorazzi, C.A., 2008. Definition of priority areas for forest conservation through the ordered weighted averaging method. Forest Ecology and Management 256 (6), 1408–1417. Valverde, V., et al., 2008. Análisis de la estructura espacial del paisaje: mosaicos del paisaje. In: Maestre, F.T., Escudero, A., Bonet, A. (Eds.), Introducción al análisis espacial de datos en ecología y ciencias ambientales: métodos y aplicaciones. Dykinson, S.L., Madrid, pp. 747–759.