Can data mining from various internet platforms systematically accelerate detection of alien species invasions across the EU?

Abstract

Invasive alien species (IAS) expansions are increasingly impacting the biodiversity and economy of Europe. To more effectively allocate the limited resources available for their management, it is pertinent to accelerate detection of IAS spread and distribution. One largely untapped secondary data source showing much potential lies in the automated tracking of internet activity such as IAS search intensity or mentions across different internet platforms. In this study, we tested if internet activity increases systematically when IAS expand into new EU countries utilizing the combined data of 88 invasive species from various internet platforms. In total, 14 internet platforms were screened and evaluated based on their database accessibility, mined data quality and utility for systematic IAS expansion tracking. We found that the procedure to obtain researcher access to minimal data required for IAS tracking (i.e., information about location, time and place) varies widely across platforms, and is particularly difficult without incurring significant costs for many of the larger ones (X, Google and Tiktok). From the explored species, more charismatic species (i.e., mammals) overall gained more online traction than more cryptic ones (i.e., plants), though online activity of the first proved a worse representation of real-world occurrence patterns. Moreover, while the final five selected internet platforms showed increased activity surrounding the year of invasion in many of the explored invasion scenarios (particularly Wikipedia and Facebook), inconsistencies between species groups, trends per platform and the large variability in data quality currently still hampers systematic integration of such data into existing databases. We conclude that combining IAS activity data from various internet platforms shows potential to accelerate IAS expansion detection across the EU (especially for fish, crustaceans, reptiles, birds and plants). However, incorporation in automated early warning systems is currently hampered by variable data quality, limited researcher access to online data and the few open, accurate and generalizable species classification algorithms with API access.