Dynamics of land use and land cover in the agricultural frontier of the Brazilian Amazon: driving forces of changes and future scenarios

• Ano de defesa: 2019
• Autor(a) principal: Andrea Santos Garcia
• Orientador(a): Maria Victoria Ramos Ballester
• Banca de defesa: Mateus Batistella, Carlos Augusto Klink, Daiana Carolina Monteiro Tourne
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade de São Paulo
• Programa de Pós-Graduação: Ecologia de Agroecossistemas
• País: BR
• Link de acesso: https://doi.org/10.11606/T.91.2020.tde-22012020-112317

The Amazon`s agriculture frontier is located in the ecotone zone between the Amazon rain forest and the Cerrado (Brazilian savanic formations). The region presents high biodiversity concentrations along with a heterogeneous landscape, comprised of different types of vegetational physiognomies and land uses. In this frontier region, the Upper Xingu River Basin (UXRB), draining approximately 170,000 km2 in Mato Grosso state produces 2% of the world\'s soybeans and 0.2% of the world\'s consumed cattle meat. However, due to their heterogeneous landscapes, these frontiers are usually poorly represented by general models portraying land use and land cover detection and change, or native vegetation loss. Our goal in this research was to map land use change in the Upper Xingu River Basin and to model vegetation loss in the region. In the first chapter we present an overall overview of different concepts that were applied throughout this research. In the second chapter, we show the results of a hierarchical classification scheme built with three levels of information for improving how land cover and land use maps capture the region\'s heterogeneity. We observed that agricultural intensification occurred mainly in the Amazon while the Cerrado has undergone an expansion in agricultural area. In the last decades, the region has been experiencing a transition from a pioneer stage of development to a consolidated frontier, with commodity-oriented development. Thus, increases in agricultural areas are tied to both international markets and the American dollar/Brazilian real ratio value. In the third chapter, we compare different data sources to identify two distinct tree loss processes: deforestation and disturbance. We observed an impressive difference between datasets built to detect both disturbance and deforestation. This pattern is related to the dominant vegetation type and specificities in the different models. In the fourth chapter, we analyzed different spatial variables related to biophysical characteristics, infrastructure, economic development, and landscape composition and configuration to select a model which would adequately represent future deforestation, forest disturbance, and general native vegetation loss (including physiognomies other than forest). Our work shows that variables influence these processes in different ways, leading us to conclude that to tackle vegetation loss, both researchers and policy makers have to focus on processes other than the rain forest deforestation, which has been the traditional focus.