Segregation, socioeconomics, and urban design produce spatial bias in community science biodiversity data: A case study in Saint Louis, Missouri, USA

Abstract

Community science is a powerful tool to understand biodiversity and species richness globally and in urban settings. Community science initiatives deployed as smartphone apps such as iNaturalist are increasingly used by both members of the public and scientists. Previous research shows that socioeconomic factors affect biodiversity patterns and that people of varying ages, racial, and ethnic backgrounds use smartphone applications in different ways; combined, these patterns may produce systematic biases in community science databases. Recently, spatial biases in community science data have been linked to user identity, income, race, or 20th century housing segregation practices. Here we investigate social and environmental factors driving the spatial distribution of iNaturalist data across census tracts in Saint Louis, Missouri, USA. We gathered 11,066 iNaturalist observations and 18 potential predictor variables from 106 U.S. census tracts and built generalized linear models based on a priori hypotheses about the effects of predictor variables. We compared models using Akaike Information Criterion and determined that the best predictors of iNaturalist observations are socioeconomic status, racial and ethnic diversity, relative location to historically residentially segregated areas, and open accessible green space. Metrics that quantify local segregation and inequality are critical to describe the uneven distribution of iNaturalist observations within Saint Louis. Results have significant implications for how researchers incorporate community science data into biodiversity studies, how they interpret spatial trends of community science databases, and how they might strategize future campaigns to advocate for increasing involvement in community science social networks.

Bibliographic Information