Tropical deforestation is one of the most important land change process at the global scale and its contribution to global environmental change is significant. Yet the multiple causes of deforestation, and the geographic and inter-annual variations in their interactions are not yet fully elucidated, especially in the Brazilian Amazon. In this region, deforestation is mostly concentrated along the main road networks. A debate therefore surrounds recent infrastructure development programs launched by the Brazilian government. The representation of causal relationships has recently evolved, from simple and linear explanations based on a limited number of factors to more complex systemic explanations describing multiple causal interactions characterizing socio-ecological systems. The study of complex causal relationships poses a series of methodological challenges to scientists due to the limitations of traditional techniques of analysis in face of this complexity. In this context, the objective of the thesis is twofold. From an empirical point of view, it aims at studying the complexity of the causes of deforestation on pioneer frontiers in the Brazilian Amazon and to highlight a limited number of causal conjunctures under different natural and socio-economic conditions. It focuses more particularly on the role of roads into deforestation. From a methodological point of view, the thesis aims at demonstrating the relevance of a configurational comparative approach to face limitations of standard statistical techniques in regard of the causal complexity of socio-ecological systems, in particular of deforestation in the Brazilian Amazon. After a short review of theoretical explanations of tropical deforestation, the causes of deforestation in the Brazilian Amazon are analyzed through a syndrom approach. By highlighting causal synergies and feedbacks, this approach suggests the necessity of using methods dealing explicitly with causal complexity. The methodological chapter presents in details the configurational comparative approach used in the thesis. The Qualitative Comparative Analysis is described as an approach (via media between traditional qualitative and quantitative approaches) and as a technique (algorithmic procedure to simplify boolean expressions). The empirical part of the thesis is devoted to the application of the Qualitative Comparative Analysis to a detailed dataset collected in the context of a meta-analysis of seven local case studies (Uruará, Altamira, Paragominas, São Félix do Xingú, Guarantã do Norte, Ouro Preto do Oeste, Pedro Peixoto) at three time periods (before 1986, 1986-1992 and 1992-1999). The resulting 21 cases correspond to different situations in terms of deforestation and the role of roads in this process. Data were collected from the available literature, through field trips and through a formal expert consultation process based on a questionnaire. The explanatory model of deforestation on pioneer fronts in the Brazilian Amazon includes one outcome variable (deforestation rate) and seven potential explanatory conditions corresponding to the main dimensions of the phenomenon as identified in the literature. Results shed new light on complex interactions between causes of deforestation on pioneer frontiers in the Brazilian Amazon. In particular, they reveal a limited number of causal pathways describing in a parsimonious way the articulation of causes under different contexts, through a limited generalization. Results also suggest the catalyst role played by roads in deforestation. Biophysical conditions and the occurrence of extractive activities also appear among the most important factors of deforestation. However, the support to colonists, the accessibility to markets and the population density are apparently playing a less important role in deforestation on pioneer frontiers in the Brazilian Amazon, relatively to other factors. The thesis constitutes a significant contribution to the debate on the causes of deforestation in the Brazilian Amazon and in particular on the role of roads. It has relevant implications in the political area for the future design of policies adapted to a sustainable management of the Amazon ecosystem. The thesis also opens new methodological perspectives for land use and land cover change studies, notably in terms of modelling.
Scouvart, M. (2006). Une analyse quali-quantitative comparée des causes de la déforestation en Amazonie brésilienne. https://hdl.handle.net/2078.5/97571