Land use and land cover dynamics in a sustainable development project in the Transamazon highway region, Pará state, Brazil

Dinâmica do uso e cobertura da terra em Projeto de Desenvolvimento Sustentável na região da rodovia Transamazônica, Pará

Land use and land cover change processes are an important topic on the global environmental research agenda. They modify the energy balance and influence the climate, by altering terrestrial albedo, the chemical composition of the atmosphere, and biogeochemical cycles (FOLEY et al., 2005; VERBURG et al., 2009; LAURANCE et al., 2011). Global deforestation is one of the most impacting factors for these processes. Hansen et al. (2013) assessed that between 2000 and 2012, 32% of global losses of vegetation cover affected tropical forests, near half of which in South America.

Since the 1970s, land use and land cover dynamics in the Amazon comprises the conversion of primary forest to other land uses, particularly in the critical region known as the “Arch of Deforestation”. According to Sorrensen (2009), this process usually begins with successive episodes of selective logging with the elimination of adult individuals of commercial tree species. The use of fire is a common practice to prepare the land for agricultural systems, especially pasture, the most relevant anthropogenic land use in terms of area in the Amazon (TERRACLASS PROJECT, 2014; MAPBIOMAS PROJECT, 2019). With productivity loss due to improper land management, many of these pastures are abandoned, and natural vegetation succession occurs.

Growing needs for environmental adequacy to global climate change demand reductions in Amazon’s deforestation rates and fire use. Public policies and programs for environmental conservation have been implemented in the region, many led by the Brazilian Ministry of Environment (MMA). One of these initiatives was established by the National Colonization and Land Reform Institute (Incra), which established the land tenure modality known as Sustainable Development Project (PDS), through Incra’s Ordinance 477/1999 (INCRA, 1999), modified by Ordinance 1040/2002 (INCRA, 2002). The PDS modality was also a response to restrictions for the creation of land reform settlement projects in Amazonian forested areas, determined by Law No. 11.284 of 2006 (BRASIL, 2006).

According to Incra (1999) and Porro et al. (2015), the PDS constitutes a settlement modality whose land tenure is issued through Real Use of Rights’ Concession Contracts (CCDRU), which could be on behalf of one or more associations representing beneficiary families. Compared to other land tenure modalities, such as Extractive Reserves, PDS beneficiaries, whose main activity does not need to be extractive, may include family farmers' featuring profiles and practices with greater market integration.

In each PDS there are Legal Reserve Areas (LRA) and Alternative Use Areas (AUA). The LRA in a PDS can be used only collectively, based on sustainable forest management plans that must be approved by environmental agencies. The AUA are explored on a family basis, in plots allocated by Incra to each beneficiary family. The PDS is an environmentally differentiated modality of land reform in the Amazon, with settlers’ subsistence based on family farming, extractivism and other activities with low environmental impact (PORRO et al., 2015). The integration of these activities would contribute to reduce deforestation’s magnitude in land reform settlements.

The implementation of this tenure modality is a great challenge, particularly in critical regions with land struggles related to the occupation of and dispute over forested areas. This is the case of the area under the influence of the Transamazon Highway (BR-230), in the state of Pará (BRATMAN, 2011; DE SARTRE et al., 2012; PORRO et al., 2015). In the municipality of Anapu, the development of a proposal that articulates land reform and environmental conservation resulted mainly from the work of sister Dorothy Stang and the Pastoral Land Commission (PORRO et al., 2015).

Environmental studies can assist the proper understanding of processes of occupation in areas pressured by the expansion of pioneer frontiers in the Amazon. In such studies, the deployment of remote sensing and geoprocessing products and techniques, by providing spatial and temporal dimensions, has the potential to subsidize policies, especially those regarding the political and economic planning of natural resource use (WATRIN et al., 2011). Working on rural settlement projects in Southeast Pará, Watrin et al. (2005) found that anthropizing processes in landholdings tend to present distinct trajectories according to specific migration features and land adjudication strategies. Forest clearing is favored in areas of older colonization and smaller average plot size, thus restricting the availability of natural resources to landholders.

Working in watersheds sites in the Lower Tapajós region, Pará state, Rozon et al. (2015) observed complex and heterogeneous changes in the spatial and temporal dynamics of land use and land cover. According to these authors, events observed in five watershed groups can be in part explained by the proximity of roads and villages.

Based on the use of geotechnology, this paper analyzes issues related to spatiotemporal changes in land use and their impacts on vegetation cover in PDS Anapu III and IV, in the state of Pará. The purpose of the paper is to understand the dynamics of land use and land cover in the PDS land reform modality, considering the selected settlement as a pilot area.

The study area corresponds to the 39,486.01 hectares of the Sustainable Development Project (PDS) Anapu III and IV, known as PDS Virola-Jatobá (PDS-VJ), located in Anapu municipality, Altamira microregion, state of Pará (Figure 1). Created by Ordinance SR01 39/2002 (INCRA, 1999), the PDS is a mosaic of 3,000-hectare land lots, previously intended for rural entrepreneurs who signed public land sale agreements with the federal government. As contract clauses were not fulfilled, the land returned to the Union. The PDS-VJ comprises two modules, PDS Anapu III, intended exclusively for forest management, while PDS Anapu IV also houses 160 alternative use plots. These plots correspond to an average of 20 hectares, intended for the development of agricultural activities by beneficiary families.

The PDS-VJ is drained by tributary streams of the Anapu and Pracurú rivers, and the existing road network is subordinated to the BR-230 (Transamazon) highway, which runs south of this area. According to Rodrigues et al. (2007), regional soils include Dystrophic Yellow Latosol (flat to soft-hilly landscapes), Dystrophic Yellow Ultisol (soft to strong-hilly landscapes) and Dystrophic Red-Yellow Latosol (hilly to strong-hilly landscapes). These soils share good physical properties (deep, well drained, good porosity and aeration), and low chemical fertility (high acidity, low base sum and low exchangeable base saturation). Such chemical limitations for agriculture are easily corrected with chemical inputs, whose cost is, however, inaccessible to the vast majority of local landholders.

Predominant land cover in PDS Virola-Jatobá is still forest, even in areas of alternative use, where 2015 forest cover was about 70% of the total. Pastures predominate in cleared areas, despite the presence of small fields of annual crops with low diversification, mostly with cassava and maize. Exceptionally, small orchards and cocoa fields are observed. In areas of abandoned agricultural use, secondary succession vegetation with variable structure and density is observed.

The georeferenced data of interest were treated and analyzed within a spatial database built in the SPRING 5.3 platform (INPE/DPI, 2015), considering IBGE’s cartographic base at the 1: 50,000 scale, and Sirgas’ 2000 projection system. Existing boundaries for the study area, obtained from a digital map provided by Incra, were also stored in the database.

Land use and land cover patterns were mapped used Landsat images, orbit/point 225/62, on the dates of 06.23.1984, 11.28.1989, 07.21.1994, 08.04.1999, 11.19.2003, 01.08.2004, 04.08.2005, 06.06.2006, 06.23.2007, 11.07.2008, 11.19.2009 and 07.04.2011 (Landsat 5 TM bands 3, 4 and 5; spatial resolution 30m); 06.17.2002, 07.25.2010, 02.21.2012 and 10.21.2013 (Landsat 7, ETM+ bands 3, 4 and 5; spatial resolution 30m); and 03.12.2014 and 07.15.2015 (Landsat 8, OLI bands 4, 5 and 6; spatial resolution 30m). The images were not submitted to the preprocessing phase given the large number of currently available images that are orthorectified (GUTMAN et al., 2013) and atmospherically corrected (MASEK et al., 2006). The latter are Landsat 8 OLI / TIRS Level 2 data products, obtained for free from USGS (2015). Images acquired through USGS are referenced to the north, which makes necessary their reprojection to the southern hemisphere.

Figure 1
PDS Virola-Jatobá location.
Source: INCRA, 2017; IBGE, 2017. Org.: by the Authors, 2019.

Selected images were classified with the region-supervised method, after being submitted to the segmentation process, in this case considering the similarity and area thresholds of respectively 400 and 11. After the generation of context files, statistical attribute regions were extracted in order to allow their classification. Image classification was performed by the Bhattacharya algorithm, based on the previously generated segmentation grid. Preliminary thematic products were analyzed in a field survey and subsequently subjected to thematic editions to reduce omission and commission errors.

A thematic caption was generated in the image classification process, comprising, besides Water and Cloud/Shade classes, nine mapping units, including Dense Forest and two stages of secondary vegetation (Old and Young Fallow). The separation of these classes is due to internal shading promoted by their structural differences, such as stratification and canopy height (WATRIN et al., 2005). The definition of other land use classes considered stages of development and various cultural and management practices adopted in the region. The Clean Pasture class refers to newly installed areas or areas with low weed infestation, which are features opposite to the ones in areas defined as Weedy Pasture. Agricultural areas were named as Annual and Perennial Crops, according to their cycle, while areas under agricultural preparation were mapped as Burnt and Exposed Soil classes.

Area quantification of mapped classes and analysis of land use and land cover dynamics in the studied years were performed upon obtaining thematic images of interest. The analysis of land use and land cover dynamics was possible by comparing images of all consecutive dates (cross-tabulation), totaling 16 temporal contrasts. Transition matrices were then generated based on the spatial distribution of thematic information.

Four phases were established considering equidistant intervals for the purpose of analyzing the selected series of images, taking the date of PDS establishment as reference: pre-installation (1984, 1989, 1994 and 1999; PRE), post-installation I [2002 to 2006; POST (1)], post-installation II [2007 to 2011; POST (2)] and post-installation III [2012 to 2015; POST (3)].

The analyzes considered both the temporal contrasts and the aggregation of these contrasts in the four phases. The general linear model (GLM) was adopted for the tests, considering the effect of the phases, with a significance level of 5%. Both tests and graphical representations were obtained with the aid of the Statistica software.

During the early stages of the analysis, deforestation remained practically stable, occasionally registering very modest annual increments. A more prominent rate was only recorded seven years after the installation of the PDS. This performance is linked to an unusual strategy of land occupation in this land tenure modality. The expected pattern was the highest deforestation rates already occurring in the first years of settlement. In the last phase analyzed, deforestation rates tended to decrease, as some lots reached the 20-ha limit previously defined for agricultural use.

Figure 5
Land use and land cover classification for 1984 and 2015, PDS Virola-Jatobá, Anapu, PA.
Source: by the Authors, 2019.

The Dense Forest class predominates in the study area landscape, denoting a state of incipient transformation, typical of agricultural frontiers. Dominant land use classes in cleared areas are pastures in different stages, with the Clean Pasture class being more frequent. Agricultural areas were absent in the pre-installation phase, and almost irrelevant during all post-installation phases.

Dense Forest areas were found to have high rates of stability over the period being assessed. Such level of stability is more often seen at consolidated landscapes of former colonization sites, or in areas of lower anthropogenic pressure, such as extractive reserves or indigenous lands. Comparatively, the areas of secondary succession presented lower stability rates, as they are an active component of agricultural fallows in family farming, especially the Young Fallow stage.

For cleared areas, pastures presented the higher stability rates, with Clean Pastures presenting the highest values, as opposed to Weedy pastures, which showed some variation. As the adopted agricultural systems predominantly consist of shifting cultivation, stability rates of the Annual Crop class were generally low.

This work was carried out under the project “Local governance and sustainability of community-based forest management in Sustainable Development Projects in Anapu, Transamazon - AutoManejo”. It was supported by Embrapa's Macroprogram 6 and Native Forest Resources Portfolio, as well as by the Fundação Amazônia Paraense de Amparo à Pesquisa - Fapespa (Icaaf 104/2014). The authors would like to thank Funcate’s geographer João Felipe Cerqueira Pinto for his support in the satellite image classification audit process employed in this work.