Answers are grounded in this paper and cite their sources.
Environmental impact assessment (EIA) is an essential instrument for anticipating and mitigating the ecological effects of development projects before they are authorized. Yet when national jurisdictions make these assessments contingent on technical parameters such as surface-based thresholds, some projects are deemed insignificant regardless of their ecological context. In this paper, we ask what proportion of projects consuming natural land in mainland France undergo an effective assessment. Combining spatial analysis of 14,915 planning permits issued in 2022 and 2023 with national land cover data, and comparing eligible projects with the opinions and decisions issued by environmental authorities, we estimate that approximately 89% of permits are granted without any project-level environmental assessment, and that permits not subject to systematic EIA collectively account for over 33% of the natural and semi-natural surfaces consumed over the study period. The open and semi-open land cover class, which includes the grassland and heath formations that are among the most rapidly declining ecosystems in Europe, is the class most exposed to these authorizations. A directed content analysis of ten complete screening dossiers identifies five recurring argumentative patterns in prefectural reasoning. A regionally stratified random sample of 100 screening decisions establishes their prevalence: all five recur, 77% of exemption decisions mobilize at least three of them, and 94% contain no substantive treatment of cumulative effects or connectivity. These results show that neither systematic thresholds nor the case-by-case screening procedure functions as an effective ecological filter, and argue for an assessment grounded in ecological context and governed by transparent and documented criteria.
No takes yet. Share an insight, caveat, or question.
Padilla et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: