Search for: All records

Abstract The availability of citizen science data has resulted in growing applications in biodiversity science. One widely used platform, iNaturalist, provides millions of digitally vouchered observations submitted by a global user base. These observation records include a date and a location but otherwise do not contain any information about the sampling process. As a result, sampling biases must be inferred from the data themselves. In the present article, we examine spatial and temporal biases in iNaturalist observations from the platform's launch in 2008 through the end of 2019. We also characterize user behavior on the platform in terms of individual activity level and taxonomic specialization. We found that, at the level of taxonomic class, the users typically specialized on a particular group, especially plants or insects, and rarely made observations of the same species twice. Biodiversity scientists should consider whether user behavior results in systematic biases in their analyses before using iNaturalist data. 

Understanding spatiotemporal variation in environmental conditions is important to determine how climate change will impact ecological communities. The spatial and temporal autocorrelation of temperature can have strong impacts on community structure and persistence by increasing the duration and the magnitude of unfavorable conditions in sink populations and disrupting spatial rescue effects by synchronizing spatially segregated populations. Although increases in spatial and temporal autocorrelation of temperature have been documented in historical data, little is known about how climate change will impact these trends. We examined daily air temperature data from 21 General Circulation Models under the business-as-usual carbon emission scenario to quantify patterns of spatial and temporal autocorrelation between 1871 and 2099. Although both spatial and temporal autocorrelation increased over time, there was significant regional variation in the temporal autocorrelation trends. Additionally, we found a consistent breakpoint in the relationship between spatial autocorrelation and time around the year 2030, indicating an acceleration in the rate of increase of the spatial autocorrelation over the second half of the 21^stcentury. Overall, our results suggest that ecological populations might experience elevated extinction risk under climate change because increased spatial and temporal autocorrelation of temperature is expected to erode both spatial and temporal refugia.