SNAPSHOT USA 2019-2023: The first five years of data from a coordinated camera trap survey of the United States

SNAPSHOT USA is an annual, multi-contributor camera trap survey of mammals across the United States. The growing SNAPSHOT USA dataset is intended for tracking the spatial and temporal responses of mammal populations to changes in land use, land cover, and climate. These data will be useful for exploring the drivers of spatial and temporal changes in relative abundance and distribution, as well as the impacts of species interactions on daily activity patterns. SNAPSHOT USA 2019–2023 contains 987,979 records of camera trap image sequence data and 9,694 records of camera trap deployment metadata. Data were collected across the United States of America in all 50 states, 12 ecoregions, and many ecosystems. Data were collected between August 1st and December 29th each year from 2019 to 2023. The dataset includes a wide range of taxa but is primarily focused on medium to large mammals. SNAPSHOT USA 2019–2023 comprises two .csv files. The original data can be found within the SNAPSHOT USA Initiative in the Wildlife Insights platform.

Methods
The first three annual SNAPSHOT USA surveys were coordinated by Roland Kays, Michael Cove, and William McShea. The 2019, 2020, and 2021 datasets are accessible for public use through the Supporting Information of their respective publications. Although the 2019 and 2020 surveys were originally processed and stored in eMammal (https://www.emammal.si.edu), all data are now housed in Wildlife Insights (WI) within the SNAPSHOT USA Initiative. The two most recent surveys, 2022 and 2023, were coordinated by the SNAPSHOT USA Survey Coordinator Brigit Rooney. This dataset represents the first publication of 2022 and 2023 SNAPSHOT USA data.

The SNAPSHOT USA project developed a standard protocol in 2019 to survey mammals >100 g and large identifiable birds. Cameras are unbaited and set at approximately 50 cm height across an array of at least 7 cameras with a minimum distance of 200 m and a maximum of 5 km between them. The collection period for SNAPSHOT USA data is between September and October and the target minimum of camera trap-nights per array is 400. Some contributors to SNAPSHOT USA 2019–2023 started collecting data earlier or deployed cameras later based on locations or logistics, and we chose to include data from August 1st through December 29th each year in this dataset.

The first two years of SNAPSHOT USA data incorporated an Expert Review Tool to verify the accuracy of every identification, as that was built in to the eMammal repository. This tool required SNAPSHOT USA project managers (Cove and Kays in 2019, with more taxon-specific reviewers in 2020) to review and confirm all species identifications, in an effort to minimize identification errors. As eMammal automatically grouped all uploaded images into “sequences” of images taken within 60 seconds of each other, by using the image timestamps, species identifications were made for individual sequences rather than images. These data have since been transferred to WI, where they underwent opportunistic review and correction by the SNAPSHOT USA Survey Coordinator.

In contrast, SNAPSHOT USA 2021, 2022, and 2023 were managed and identified entirely in WI. All SNAPSHOT USA projects in this repository were created as “Sequence” projects, to enable the identification of sequences in the same manner as eMammal. Each 60-second sequence of images was classified to the narrowest taxonomic level possible by three iterations of validation. First, WI’s Artificial Intelligence algorithm suggested a taxonomic identification. This algorithm consists of a multiclass classification deep convolutional neural network model that uses pre-trained image embedding from Inception, a model used to identify objects. Second, each array’s Principal Investigator was responsible for validating the data, fixing Artificial Intelligence identification mistakes, and approving the data they contributed to the survey. Lastly, the SNAPSHOT USA Survey Coordinator quality-checked the deployment data and as many identified sequences as possible. This was a multistep process that began with checking the sequence metadata for obvious timestamp errors by organizing them chronologically in Microsoft Excel, and the deployment metadata for location errors by mapping their coordinates and looking for outliers. Next, the coordinator checked the sequence metadata for unlikely identifications, including species detections in places outside their known range, and verified their accuracy by viewing the images in WI. Finally, identifications for the most common species were verified by using the “Species” filter on WI to look for mistakes, one species at a time. 

When combining the five years of SNAPSHOT USA data to create SNAPSHOT USA 2019–2023, several aspects of the data were standardized to ensure consistency across all years. These were camera array names, camera location names, and taxonomy classifications. To match protocol requirements, all camera locations less than 5 km apart were classified as one array. This resulted in combining several arrays that were originally recorded under different names and ensuring that arrays in the same place maintained the same name each year. The camera location names were standardized by ensuring that all locations with geographic coordinates that were the same to four decimal places, in Decimal Degrees notation, had the same name. However, the original coordinates were retained in the dataset. Finally, all species taxonomy classifications for the 2019 and 2020 datasets (identified in eMammal) were standardized to match those used by WI. As part of this process, all subspecies of mammals in the dataset were changed to species level (e.g., Florida black bear (Ursus americanus floridanus) became American black bear (Ursus americanus)).

For mammal taxonomy classifications, WI uses a combination of the International Union for Conservation of Nature (IUCN) Red List of Threatened Species (2023; https://iucnredlist.org) and the American Society of Mammalogists Mammal Diversity Database (2024; https://www.mammaldiversity.org). For bird species, WI uses Birdlife International’s taxonomy classifications (2024; https://datazone.birdlife.org/species/search). The WI taxonomy is continually updated in response to public user suggestions and the taxonomy used in the SNAPSHOT USA 2019–2023 dataset reflects the WI taxonomy used in June 2024.