AgentschapPlantentuinMeise/iNatBioblitzes: Initial release

This repository and its content was created to support a review article on bioblitzes (in preparation). One of the datasets used to evidence the fact that bioblitzes trigger longer term public engagement with recording wildlife was retrieved using the iNaturalist API (https://api.inaturalist.org/v1/docs/).

iNaturalist is a smart phone application and website that forms a social network for amateur and professional naturalists. Users can collect observations with photographs, identify and comment on other people’s observations and map the results. iNaturalist users can create their own projects within the site to support events and specific surveys. Many of these user defined projects are bioblitzes. Using the iNaturalist application programming interface (API), projects were selected that contained the word ‘bioblitz’ in their title (5,509 projects on November 3, 2020).  The selection of projects based on the word ‘bioblitz’ is justified because project creators are self-identifying their projects within the bioblitz concept of a relatively intense and focused inventory of biodiversity, rather than some other short duration project with different aims to a bioblitz. From the 5,509 projects, 1860 were selected that lasted less than 10 days in duration, were restricted to a specific place, were conducted between 2013 and 2020, had more than one observer and at least one identifier. Ten days is a somewhat arbitrary cutoff length but was informed by our literature review where the average length of a bioblitz was 31 hours and all but two were four days or less. Even using this generous cutoff length 81% of the iNaturalist bioblitzes lasted four days or less. Also, we found that the word ‘bioblitz’ occurred much more frequently (27%) in iNaturalist projects lasting less than 10 days as compared to its use in only 5% of the projects lasting longer than 10 days. This filtering removed projects that were just experimental and never used, were ongoing projects, those in the future and long-term projects that do not conform to our definition of a bioblitz. From these iNaturalist projects we extracted data such as the bioblitz date, duration, number of observers, number of identifiers. From this dataset, the number of observations was summed over all bioblitzes in one year and compared to the total number of observations in iNaturalist for that given year (2013-2020). A word cloud was created from the projects’ descriptions of the same sample of projects. Both English and Spanish stopwords were used to eliminate low information-content words. Also the words “iNaturalist” and “project” were removed. Code was written in Python (3.7.3) and R (3.5.3) and can be found here: https://github.com/AgentschapPlantentuinMeise/iNatBioblitzes.

A random subsample of 100 iNaturalist bioblitz projects (of maximum 10 days) with all their users (N=3425) were selected and recording activity of each of the participants was extracted for upto a year before and a year after the bioblitz event (day -9 to day 0). For example, a bioblitz of 24 hours starts and ends on day zero, a bioblitz of 10 days starts on day minus nine and ends on day zero. For each user we calculated the difference in recording activity as the number of recording days per week before and after the bioblitz for paired weeks of the year (1-50). By taking this approach we hope to account for seasonal variation in the detectability of species. These data were used to evaluate the impact of the Bioblitz on the recording activity of the users, including only those people who used iNaturalist both before and after the event and users who used iNaturalist only during and after the event. We fitted a nonlinear mixed-effects model using the saemix R package with function y~a*exp(-bx) with ‘project’ as a random effect. For results and discussion, see our review paper currently in preparation, but will be linked to this repository as soon as it comes out.