This manuscript presents a synthesis of my work on spatial inequalities, and my research perspectives for the coming years. With my co-authors, I studied different forms of spatial inequalities: socio-economic and quality of life disparities, employment growth disparities, residential segregation and environmental inequalities. My perspectives are partly a continuation of this work, and partly a new research direction toward a spatialized analysis of biodiversity conservation policies. Synthesis The first part of the synthesis deals with regional inequalities in metropolitan France. It presents my contributions to the analysis of the socio-economic diversity of rural and mountain regions, quality of life disparities, and disparities in manufacturing employment growth. We first built typologies of rural and mountain regions, using a wide range of socioeconomic and landscape indicators computed at fine spatial scales, and standard methods of exploratory data analysis and hierarchical clustering. In addition, we built a typology of rural remoteness by combining two key indicators: proximity to the main employment centres and to the main service centres. The resemblance between the typology of rural regions and the one of rural remoteness highlights the structuring influence of the urban network on regional disparities. This influence is also visible in the typology of mountain regions, where some types have urban or suburban locations and others are away from urban centres. These typologies also highlight regional specificities at wider scales: for example, some types of rural regions are almost specific to the north or to the south of mainland France, or to its so-called ‘empty diagonal’ (i.e., low-density north-east-south-west diagonal), and the mountain regions that prevail in the different mountain chains (the Alps, the Vosges…) are not the same. Alongside my contribution to these empirical studies, I offered some reflections on the guiding principles of a rural policy in the light of distributive justice conceptions centred on efficiency, responsibility, or the satisfaction of needs. Regarding quality of life disparities, we used a variable clustering approach – the ClustOfVar method –and a set of regional quality of life indicators to reveal the components (i.e., the clusters of indicators and their synthetic variables) that structure the living conditions at this regional scale in mainland France. These components appear to be broadly organised around two main branches: the resources and employment conditions of inhabitants, on the one hand, and the local environment (access to services, housing, environmental and social conditions), on the other. The computation of spatial autocorrelation indices confirm the existence of spatial interactions between regions, with a specific intensity and spatial scope for each of the quality of life components. Furthermore, the mapping of synthetic quality of life variables once again highlights the structuring influence of the urban network, as well as regional specificities at wider geographical scales. Another contribution concerns regional disparities in employment growth. The question under consideration is whether these disparities reflect more or less advantageous regional sectoral specialisations, or differences in competitiveness between locations that are more or less conducive to economic activity as a whole. The Shift-Share Analysis (SSA) framework is generally used to answer such a question. However, standard SSA techniques can only be applied to data that are highly aggregated – in space, in time, and in the number of industrial sectors – to avoid certain sectors being absent from certain regions. Our new SSA technique – called ‘Scalable SSA’ – overcomes this limitation. It allows us to conduct a year-by-year study of regional disparities in manufacturing employment growth over the period 1994-2015, considering 24 manufacturing industries and 3 regional levels, including a fine spatial division into about 2000 small regions. The results obtained suggest that these disparities stem primarily from differences in regional competitiveness. We also show that competitiveness effects are positively auto-correlated and that clusters of regions with low or high competitiveness can be identified. These results lead to some reflections on public action, inviting in particular to take into account such clusters when designing regional development policies. The second part of the synthesis deals with socio-spatial inequalities within urban regions. Natural amenities occupy a special place in this context, as a factor of residential attractiveness and source of segregation processes on the one hand and as sources of unevenly distributed benefits on the other hand. A preliminary contribution consists of a review of the social science literature using the notions of natural or environmental amenities, which allows for a better understanding of the definitions of these notions, the indicators used, the research questions and the academic dynamics. We analysed 328 peer-reviewed articles published over a 40-year period. We observed an exponential growth in publications, with a small set of economics journals publishing most of them. Research objectives have changed from an initial focus on recovering the implicit prices of natural/environmental amenities to a more recent and growing interest in evaluating their impacts on regional development and quality of life. Unidimensional measures related to water and land cover are the most common indicators. Despite this body of work, there are still no widely acknowledged definitions of these notions. We propose revised definitions that emphasise the heterogeneity of impacts on people and firms (and hence inequality) and bridge the gap between the concepts of natural amenity and ecosystem services. A related empirical contribution focuses on the role played by natural amenities in the emergence of residential socio-spatial structures in the urban regions of Grenoble and Aix-Marseille-Toulon. In a first step, econometric estimations confirm that natural attributes (forests, open spaces, rivers, lakes, coastline and altitude) influence residential location choices, and that their effects differ according to the size and socio-professional status of households. The second step is the segregation analysis: household choice probabilities for each of the urban locations are computed using the full models estimated in the first step (realistic scenario), or partial models in which the parameters corresponding to the natural attribute variables are set to zero (counterfactual scenario). We can then run Monte Carlo experiments to simulate the socio-spatial distribution of households for each scenario, calculate the associated segregation indices, and test the difference between the segregation levels obtained 'with' and 'without' the partial effects of natural attributes. The results show that natural attributes have a significant impact on the socio-spatial sorting of households. For example, in Aix-Marseille-Toulon, they are associated with a slight increase in the segregation of higher professions and a strong increase in the segregation of retired people. In Grenoble, they are associated with a strong increase of household segregation by size, but with a decrease of the segregation of retired people and higher professions. The place-specific interplay between the ‘natural-amenity driven segregation channel’ and the socio-spatial sorting induced by other location factors (access to jobs, services, social amenities, etc.) appears to be crucial for the dynamics of segregation. These results suggest that the natural environment should be taken into account in urban social mix policies. The last noticeable contribution is the development of a robust statistical approach for measuring urban environmental inequalities. Conceptually, we call 'environmental segregation' the spatial separation between a social group and an environmental (dis)amenity, and 'segregation-based environmental inequality' a difference between two social groups in their degrees of environmental segregation. We then propose two original segregation-based environmental inequality indices: (i) the environmental dissimilarity gap index (ΔEDk) for inequalities related to zonal environmental data (such as vegetation cover), and (ii) the environmental centralization index (ECd) for inequalities related to point environmental data (such as hazardous sites). The environmental dissimilarity for a group is the proportion of individuals to be moved from one neighbourhood to another to reach the situation of no environmental segregation, and ΔEDk provides the dissimilarity gap between two groups (with k the degree of contiguity for the consideration of interactions between neighbourhoods). ECd compares the distribution of two social groups in terms of proximity to the environmental sites (with d indicating the maximum distance beyond which the site is supposed to have no more effect). We also suggest a randomisation strategy to test the statistical significance of our indices and a procedure to identify and map hotspots that have a strong influence on environmental inequality. These methods are made available in the R package SegEnvIneq. This approach could help urban decision-makers to better target their policies. Empirical results highlight the unequal access to green spaces and exposure to industrial risks between poor and non-poor households in the urban regions of Grenoble and Aix-Marseille. Perspectives The last part of the manuscript presents my research perspectives. Complementary work on the analysis of residential location choices, regional disparities in quality of life and environmental inequalities is already underway. The first is a longitudinal and dynamic analysis of residential mobility in metropolitan France, focusing on household preferences for residential density. The second deepens our analysis of the quality of life and examines its stability over time. The third is an in depth analysis of the empirical relationship between urban segregation and environmental inequalities in French urban regions. In the longer term, however, my ambition is to enter the field of biodiversity conservation economics. More specifically, I aim to explore the relationship between economic geography, economic inequality and biological diversity. The ultimate goal is to be able to consider simultaneously ecological, social, and spatial effects associated with public policies. This ecological-economic framework should have solid foundations in both economics and ecology. This is why I have started working with ecologists and plan to invest heavily in such interdisciplinary collaborations in the future. Several cross-national studies have shown correlations between economic activity, economic inequality and biodiversity indicators. Yet these links are poorly understood, and even less so at sub-national scales. One challenge is to make progress in empirically identifying the different channels through which socio-economic factors are likely to affect biodiversity. Another challenge is to develop a theoretical modelling of these links, which would improve the ability of public actors to consider the impacts (including unforeseen ones) of their decisions on biodiversity, and to design effective conservation policies. Although I plan to contribute to both challenges, the second is at the heart of my perspectives. New Economic Geography (NEG) will be a privileged reference to start this work. NEG models emphasize the importance of considering the endogeneity of the economic spatial structure when thinking about the impact of public policies. A first approach will be to model a system of regions - featuring monocentric or polycentric cities and their hinterlands – taking into account that different spatial patterns of activity will be associated with different economic (e.g. land rents and wages) and ecological (e.g. habitat degradation, fragmentation and disturbance) impacts. Interdisciplinary collaborations will be beneficial to help define ecologically sounded ‘biodiversity production functions’. The ecological literature shows that the responses to anthropogenic pressures of different biodiversity components (e.g., taxonomic) and metrics (e.g., stock vs. flow) can vary significantly. This diversity of responses should be taken seriously, as it may create conflicts between competing conservation objectives. In a further step, I will consider turning to dynamic modelling frameworks to study the consequences of potential non-linearities and threshold effects, latencies and irreversibilities. The coupling of economic geography and meta-community models is also an avenue that could be considered in the long term. As for the link between economic inequality and biodiversity, taking it into account would make it possible to better understand the social impacts of conservation policies, as well as the ecological impacts of redistribution policies. Economic growth also needs to be considered: the conflict between growth, spatial equity and social cohesion has already been analysed in NEG models, but not the concomitant conflict with biodiversity. A final perspective concerns land policy. Land use management is deemed crucial for biodiversity conservation. The French ‘Climate and Resilience’ law of 2021 sets a target of ‘no net land take by 2050’. This raises the question of its effects on inequality and growth and its effectiveness for biodiversity conservation. Modelling this policy and its dynamics in a geographical economy would be a useful line of research.