Can citizen science monitor whale-shark aggregations? Investigating bias in mark–recapture modelling using identification photographs sourced from the public

Context The conservation status of the whale shark, Rhincodon typus, remains uncertain. Throughout their range, whale sharks are a focus for wildlife tourism and many identification photographs taken by tourists have become available online, presenting a potentially valuable source of data for monitoring populations using mark–recapture techniques. However, the suitability of these photographs for mark–recapture models has yet to be investigated. Aims We explore the suitability of identification-photographs available from online databases to produce assessments of life-history parameters and conservation status of whale sharks in the Maldives. Methods To test the validity of using publically sourced images, we used photo-identification images collected from both experienced researchers and tourists between 2003 and 2008 to construct two databases. Images taken by tourists were compiled from online databases. Researcher and public databases were analysed separately and the results of mark–recapture models then compared. Key results The dataset constructed from online public databases did not violate the assumptions of mark–recapture modelling. Estimates of parameters and abundance obtained from models using these data were similar to those produced using data provided by experienced researchers. Conclusions Publically sourced data allowed for the accurate estimation of abundance of whale sharks. These estimates were not confounded by the suitability of photographs, probably because of the high encounter rate in the aggregation, the high residency rate of sharks and the retrospective nature of photo-identification, which limited heterogeneity in capture probability between marked and unmarked sharks. Implications Our findings support the use of publically sourced data for use in mark–recapture studies of whale sharks, at least in situations where sharks are resident to the location. This approach will be useful in regions where data collected by tourists are available online, and research funding is limited.