Be Mindful of the (Data) Gaps

June 21, 2023

In a new paper published in a special issue on detecting and attributing the causes of biodiversity change of the Philosophical Transactions of the Royal Society B, Ruth Oliver and colleagues show that wildlife images from the largest, public collection of camera trap data are closing spatial, temporal, and taxonomic biodiversity data gaps around the world. This work leveraged images from Wildlife Insights (WI), an emerging platform for processing, managing, and sharing data from camera traps. Despite having vastly less data than the largest aggregator of biodiversity records, the Global Biodiversity Information Facility (GBIF) (ca. 15 million vs. >2 billion), WI contains unique information on global biodiversity. Where camera trap data is collected, it provided much more consistent sampling (mean  = 133 vs. 57 days/year) and provided information for an additional 1% of mammal species. For 93% of mammal and 48% of bird species with images in WI, these records provided unique observations of their ranges not found in GBIF.


Since the adoption of the Global Biodiversity Framework (GBF) last December, headlines have abounded touting the ambitious goals to safeguard biodiversity. But achieving the bold goals of the GBF will rely on the (admittedly less flashy) goal of developing effective and global systems of biodiversity data collection, a.k.a. Target 21. Camera trap data has not been a prominent aspect of these systems – until now. However, as this research has revealed, camera trap data can have an outsized role in providing substantial improvements in spatial, temporal, and taxonomic coverage of species with unique potential for closing the gaps – the biodiversity data gaps, that is.


The current volume of publicly available biodiversity data out there is staggering: as of the writing of this article, more than 2.2 billion observations have been uploaded to GBIF. But this data is not without biases – certain regions and habitat types are oversampled compared to others, and species that are harder to spot in human-led surveys are underrepresented – and these biases can lead to gaps in our knowledge of where species live. Camera traps can solve some of these issues. These devices can be deployed in areas that are difficult to reach and collect data for long periods of time with minimal disturbance to species. While most camera trapping efforts today aim at documenting ground dwelling mammals and birds, new technological advancements and sampling protocols could pave the way for broader taxonomic coverage in the near future. An uptake in the use of camera traps to study arboreal species, reptiles and insects is already happening.


New results in Oliver et al. 2023 show camera trapping fills biodiversity data gaps along three dimensions – taxonomic, spatial, and temporal. Sources: (top left) Rio São Benedito, CENAP/ICMBio; ICMBio/CENAP 2017. Last updated September 2022. Rio São Benedito. Accessed via on 2023-01-21; (top right) Kon Plong 2019-20, Fauna and Flora International - Vietnam Programme, Wearn, Oliver R; Nguyen, An 2019. Last updated March 2023. Kon Plong 2019-20. Accessed via on 2023-03-10. (bottom right) Inventory of medium-to-large, terrestrial forest mammals and birds in the Tanzanian Eastern Arc Mountains, Wildlife Conservation Society; Rovero, F. 2000. Last updated August 2022. Inventory of medium-to-large, terrestrial forest mammals and birds in the Tanzanian Eastern Arc Mountains. Accessed via on 2023-01-21.


Although a powerful tool, camera traps also have some downsides. The huge amount of images collected (thousands in each survey) have to be reviewed to extract information on the species detected. The data needs to be organized and analyzed. And after they have been used for the primary purpose they have been collected for, most of the information ends up forgotten in some wildlife office’ desk despite all the effort needed to compile them.


These were the motivations for the founding of Wildlife Insights, a global platform for camera trap data analysis and sharing. Wildlife Insights not only applies advanced AI to identify species in contributors’ data, saving countless hours of manual inspection, but also provides data under fair sharing agreements and connects data contributors to decision makers. Millions of images have already been uploaded and the community is growing larger each year, aspects that led to consider camera trapping and the infrastructure provided by Wildlife Insights as promising tools to promote biodiversity information in the next decades.


Camera trap data could prove invaluable to closing the global biodiversity data gaps – in fact, as the paper by Oliver and colleagues shows, it’s already been filling these gaps along all three axes of biodiversity data coverage. As the volume of data uploaded to Wildlife Insights continues to exponentially increase and new technological advancements and sampling protocols pave the way for broader taxonomic coverage in the near future, the gaps will keep shrinking and the path to Target 21 and the rest of the GBF goals will become clearer. The collective impact of the global camera trapping community is showing all of us biodiversity and conservation scientists just how important it is to be mindful of the data gaps.