合成控制方法能够利用城市中的参与式科学数据进行更强的因果推断

1. 信息

2. 摘要

Urban environments pose unique challenges for understanding drivers of biodiversity change, as fragmented land ownership makes traditional biodiversity monitoring and randomized experiments logistically difficult. While participatory science platforms such as iNaturalist offer a promising data source by providing extensive biodiversity data from urban areas, inferring causality remains challenging because of confounding factors in observational data. To leverage these data advances, we offer a framework that combines records from iNaturalist with synthetic-control methods, a quasi-experimental approach. We demonstrate this approach in a case study assessing the impact of Hurricane Ida (2021) on the number of research-grade iNaturalist bee observations, used as a proxy for bee abundance, in Philadelphia, USA. The synthetic control estimated a 15.5-20.9% decline in bee observations in the 2 years post-event. By contrast, three conventional ecological analyses-an interrupted time-series regression, before-after comparison and a before-after control impact design-failed to detect this decline. Synthetic-control methods offer a powerful tool for estimating city-wide biodiversity responses to climate events and policy interventions, enhancing the utility of participatory science data for urban ecology.

3. 图表

4. 引用

@article{kaiser2026,
  author = {Asia Kaiser and Julian Resasco and Laura E. Dee},
  title = {Synthetic control methods enable stronger causal inference using participatory science data in cities},
  journal = {Nature Ecology \& Evolution},
  year = {2026},
  volume = {10},
  number = {6},
  pages = {1204-1215},
  doi = {10.1038/s41559-026-03084-4},
  publisher = {Springer Science and Business Media LLC},
}