Although environmental justice scholars have been addressing spatial scale for at least thirty years, one of its components remains largely overlooked, namely extent. In this paper, we investigate the effects of extent variation in environmental (in)justice patterns by analyzing the statistical associations between socio-economic marginality and environmental vulnerability at three different spatial scopes: metropolitan areas, municipalities, and boroughs. Using census and geospatial data, our case study focuses on the relations between income and color/race, on the one hand, and susceptibility to flooding and landslides, on the other, in Brazil’s two largest cities (São Paulo and Rio de Janeiro). We constructe an integrated, census tract-level database using areal weighting interpolation combining census with susceptibility data to perform k-means multivariate clustering analysis. Although our results show that the most vulnerable to landslides and flooding are non-white, low-income people – confirming common environmental justice claims – they also suggest that spatial extent impinges on statistical patterns. While the clusters are very similar in the metropolitan and municipal scales, pointing to a fractal-like structure, they differ significantly at the borough scale. This smaller scale reveals a different picture, one in which well-to-do white people are just as exposed to hazards (especially flooding) as non-white, middle- to lower-middle-class people. We argue this is a result of elite self-exposure to risk, which is supported by socio-spatially unjust distribution of risk-mitigation infrastructures. Policymakers should pay attention to such scale-dependent complexities in devising ways to cope with the increasing inequities brought about by climate change.

尽管环境正义学者至少三十年来一直在研究空间尺度，但它的一个组成部分仍然在很大程度上被忽视，即范围。在本文中，我们通过分析三个不同空间范围内的社会经济边缘性和环境脆弱性之间的统计关联来研究环境（不）公正模式的程度变化的影响：大都市地区、直辖市和自治市镇。利用人口普查和地理空间数据，我们的案例研究一方面侧重于巴西两个最大城市（圣保罗和里约热内卢）的收入与肤色/种族之间的关系，另一方面侧重于洪水和山体滑坡的易感性。我们构建一个集成的，人口普查区域级数据库使用面积加权插值将人口普查与敏感性数据相结合，以执行 k 均值多变量聚类分析。尽管我们的结果表明，最容易受到山体滑坡和洪水影响的是非白人、低收入人群——证实了常见的环境正义主张——但它们也表明空间范围会影响统计模式。虽然这些集群在大都市和市政尺度上非常相似，指向类似分形的结构，但它们在自治市级尺度上却存在显着差异。这个较小的规模揭示了另一幅图景，其中富裕的白人与非白人、中下层中产阶级的人一样面临危险（尤其是洪水）。我们认为这是精英自我暴露于风险的结果，这得到了风险缓解基础设施的社会空间不公平分布的支持。政策制定者在设计应对气候变化带来的日益加剧的不平等时，应注意这种依赖于规模的复杂性。