Photographing Carnivores with Snapshot Serengeti

Exploring the dynamics of natural systems requires large-scale, multispecies monitoring approaches that are often infeasible for long-lived, wide-ranging species. Remote camera surveys provide a much-needed non-invasive, cost-effective approach for large-scale multispecies monitoring. Camera surveys are growing in popularity and can produce overwhelming amounts of data; there is clear need for efficient data processing techniques and accessible analytical approaches. Snapshot Serengeti, a project on the citizen science platform The Zooniverse (www.zooniverse.org), employs members of the general public to help process >1 million photographs produced by a large-scale camera survey operating in Serengeti National Park, Tanzania. I will present Snapshot Serengeti as a case study in (1) the use of citizen science to rapidly and accurately process large volumes of imagery produced by large-scale camera surveys, and (2) preliminary analyses of novel applications of camera trap data. Through the Snapshot Serengeti website (www.snapshotserengeti.org), ~30,000 volunteers contributed classifications, processing three years of data with >96% accuracy. Preliminary analyses of Snapshot Serengeti data further suggest that large-scale camera surveys accurately reflect community dynamics through space and time. The image processing capacity provided by citizen science projects such as Snapshot Serengeti can expand the types of questions that camera traps can address in ecological research and demonstrates that citizen science can effectively process large quantities of ecological data.