Who, where, when: Observer behavior influences spatial and temporal patterns of iNaturalist participation.

Understanding the observation process is key to effective use of opportunistic biodiversity data from low-structure citizen science. We investigated how observer characteristics, including activity level (enthusiastic, moderate, or short-term) and primary location (resident or visitor), influenced spatial and taxonomic patterns of iNaturalist participation in the Hawaiian Islands from 2008 to 2021. We found that visitors represented nearly two-thirds of all observers and 96% of the enthusiastic group. Sampling bias toward developed areas, sites near roads or trails, and areas with fewer protections was relatively stronger for residents, most of whom were short-term participants. However, observations by enthusiastic residents had the greatest taxonomic diversity. Participation grew exponentially through 2019, then decreased in spring 2020. Though resident participation was comparatively steady during COVID-19 travel restrictions, it did not compensate for the decline in visitor activity. Once restrictions ended in 2021, participation recovered quickly among enthusiastic visitors but continued to be lower than expected for residents. Our results indicate that the majority and diversity of sampling relies on a small group of highly active observers, most of whom are unlikely to live in the region. Fostering sustained, local participation could improve the consistency and quality of iNaturalist observations and thus their utility in biodiversity conservation. • Majority of iNaturalist observers in Hawaii are likely to be visitors. • Visitors tend to be more active on the app than resident observers. • Resident observations contain relatively higher spatial sampling bias. • Resident activity was steadier during pandemic, but visitors show greater recovery. • More sustained participation from local community could improve data quality. [ABSTRACT FROM AUTHOR]