Abstract

Biodiversity is crucial for human health, but previous methods of measuring biodiversity require intensive resources and have other limitations. Crowdsourced datasets from citizen scientists offer a cost-effective solution for characterizing biodiversity on a large spatial scale. This study has two aims: 1) to generate fine-resolution plant species diversity maps in California urban areas using crowdsourced data and extrapolation methods; and 2) to examine their associations with sociodemographic factors and identify subpopulations with low biodiversity exposure. We used iNaturalist observations from 2019 to 2022 to calculate species diversity metrics by exploring the sampling completeness in a 5 × 5-km2 grid and then computing species diversity metrics for grid cells with at least 80 % sample completeness (841 out of 4755 grid cells). A generalized additive model with ordinary kriging (GAM OK) provided moderately reliable estimates, with correlations of 0.64–0.66 between observed and extrapolated metrics, relative mean absolute errors of 21 %–23 %, and relative root mean squared errors of 27 %–30 % for grid cells with ≥80 % sample completeness from 10-fold cross-validation. GAM OK was further applied to extrapolate species diversity metrics from saturated grid cells (N = 841) to the remaining grid cells with <80 % sample completeness (N = 3914) and generate diversity maps that cover the grid. Further, generalized linear mixed models were used to examine the associations between species diversity and sociodemographic indicators at census tract level. The wild vascular plant species diversity metrics were inversely associated with neighborhood socioeconomic status (i.e., unemployment, linguistic isolation, educational attainment, and poverty rate). Minority populations (i.e., African American, Asian American, and Hispanic) and children had significantly lower diversity exposure in their neighborhoods. Crowdsourcing data offers a cost-effective solution for characterizing large-scale biodiversity in urban areas.