Examining spatial biases in the community science platform, iNaturalist, using British Columbia, Canada, as a case study




Geurts, Ellyne

Journal Title

Journal ISSN

Volume Title



The ever-growing interest in community science platforms like iNaturalist and eBird is ushering in a new era of biodiversity and ecology research where researchers are overflowing with data across large geographical and temporal scales. However, these big and often unstructured data come with a cost, biases. These biases include temporal, spatial, and taxonomic biases in opportunistically collected community science datasets like the popular biodiversity platform, iNaturalist. There is a need to improve our knowledge of the biases on these platforms, so that the data can be used effectively. My thesis tackles this gap by examining spatial biases on the iNaturalist platform. My first study uses Maxent to model broad-scale spatial bias in iNaturalist observations in British Columbia, Canada. I ask: Where are iNaturalist users primarily observing? and What landscape features best explain the spatial bias? I find that distance to roads is the most important landscape variable explaining spatial bias. In my second chapter, I experimentally tested whether fine-scale spatial biases of trails affected taxonomic richness estimates on iNaturalist using paired timed transects with a team of iNaturalist observers. I found greater taxonomic richness on trails compared to away from trails and no difference in rare species observations between on and off trails, suggesting there is no loss of information by primarily surveying along trails. Overall, this research shows important variables to include to control for spatial bias when using iNaturalist data and provides reassuring evidence that fine-scale bias does not impede biodiversity surveying from community scientists.



biodiversity, spatial biases, community science, iNaturalist, citizen science