Camera trap data of small mammals at experimental dishes
Data files
Dec 25, 2024 version files 4.23 MB
-
deployments.csv
10.22 KB
-
media.csv
4.15 MB
-
observations.csv
70.64 KB
-
README.md
3.87 KB
Abstract
We conducted cafeteria experiments in 2012/13, 2018/19, 2019/20 and 2020/21. In early October, we presented 18 dishes with 10 individually-marked seeds of Fagus sylvatica, two of which were fitted with a VHF-transmitter and eight with a 25-cm wire tag with a colorful flag. To confirm the species identity of visiting rodents, we installed camera traps at each experimental dish in autumn 2019 and 2020. Rodent species composition recorded at the experimental dishes varied between 2019 and 2020, but taxonomic groups other than small mammals were extremely rare in both years. In 2019, camera traps recorded 723 independent detections of small mammals (including 524 encounters of Apodemus spp. and 160 of Myodes glareolus) and in 2020, 79 small mammal detections (including 33 encounters of Glis glis and 13 of Myodes glareolus) during 15 and 23 days of operation, respectively. Encounter rates on camera traps were in line with species composition of life trapping and highlighted the enormous variation of visitation rates at experimental dishes between subsequent years. The only camera trap observation of a larger mammal was triggered by Meles meles in 2020. In addition to the frequent encounters of small mammals at the experimental dishes, we also detected two birds in 2019 and another one in 2020.
README: Camera trap data of small mammals at experimental dishes
https://doi.org/10.5061/dryad.2ngf1vhzn
Description of the data and file structure
We conducted cafeteria experiments in 2012/13, 2018/19, 2019/20 and 2020/21. In early October, we presented 18 dishes with 10 individually-marked seeds of Fagus sylvatica, two of which were fitted with a VHF-transmitter and eight with a 25-cm wire tag with a colorful flag. To confirm the species identity of visiting rodents, we installed camera traps at each experimental dish in autumn 2019 and 2020. Rodent species composition recorded at the experimental dishes varied between 2019 and 2020, but taxonomic groups other than small mammals were extremely rare in both years.
Files and variables
File: deployments.csv
Description:
Variables
- deploymentID:Unique identifier of the deployment.
- locationID:Identifier of the deployment location.
- latitude:Latitude of the deployment location in decimal degrees, using the WGS84 datum.
- longitude:Longitude of the deployment location in decimal degrees, using the WGS84 datum.
- coordinateUncertainty:Horizontal distance from the given latitude and longitude describing the smallest circle containing the deployment location. Expressed in meters.
- deploymentStart:Date and time at which the deployment was started.
- deploymentEnd:Date and time at which the deployment was ended.
- cameraID:Identifier of the camera used for the deployment
- cameraModel:Manufacturer and model of the camera
- cameraHeight:Height at which the camera was deployed. Expressed in meters.
- cameraTilt:Angle at which the camera was deployed in the vertical plane. Expressed in degrees, with
-90
facing down,0
horizontal and90
facing up. - featureType:Type of the feature associated with the deployment.
- habitat:Short characterization of the habitat at the deployment location.
File: observations.csv
Description:
Variables
- observationID:Unique identifier of the observation.
- deploymentID:Identifier of the deployment the observation belongs to.
- eventStart:Date and time at which the event started.
- eventEnd:Date and time at which the event ended.
- observationLevel:Level at which the observation was classified. Event-based observations consider an event (comprising a collection of media files).
- observationType:Type of the observation. All categories in this vocabulary have to be understandable from an AI point of view.
- scientificName:Scientific name of the observed individual(s).
- count:Number of observed individuals.
File: media.csv
Description:
Variables
- mediaID:Unique identifier of the media file.
- deploymentID:Identifier of the deployment the media file belongs to. Foreign key to
deployments.deploymentID
. - captureMethod:Method used to capture the media file.
- timestamp:Date and time at which the media file was recorded.
- filePath:relative path to the media file.
- filePublic:FALSE if the media file is not publicly accessible
- fileName:Name of the media file.
- fileMediatype:Mediatype of the media file.
Images
Given the large number of files, we did not upload the images mentioned in "media.csv". Please contact the corresponding author if you´re interested in this data.
Code/software
Images were manually checked and labeled via the software digiKam (Version 5; https://www.digikam.org). We did not label the avi-files, but reviewed more than 9,000 video sequences to aid species identification. We used R Version 4.4.1 to format the tabular data and to summarize detections of the same taxon within 5 minutes for each location to define independent trigger events or encounters by avoiding repetitive detection of the same individual lingering in the focal view of the devices.
Methods
We installed 18 camera traps (Victure HC600) 60 cm above-ground facing the experimental dishes of our cafeteria experiment and used motion-activated photographs for the identification of mammals and birds in October 2019 and 2020. We manipulated the lenses to ensure sharp images at close distance by turning each lens about 45° counterclockwise. Our custom-built camera rack consists of 20x20 cm plexiglas for the attachment of the camera, screwed on a 1 m threaded rod (10 mm diameter) stuck into the ground (about 30 cm deep). We stick the rod through drilled holes at the center of two aluminum rulers that form a cross on top of the forest floor to enhance stability and reduce vibrations caused by wind. A threaded nut ensures the camera lens to be installed in the preferred distance to the ground (60 cm in our case).
The resulting photographs depict a small segment of the forest floor (about 35*50 cm) with animals being photographed at short distances, to enhance the identification of small mammals and birds. We chose the maximum sensitivity of the PIR-Sensor to maximize detection probability and minimum illumination of the LED-flash to avoid overexposed photographs. Cameras were set to take a series of three images and a 5-seconds video when triggered. We did not use a sleep-function to minimize the risk of missed animal encounters.
Camera traps were left for 15 days of operation in 2019 and 23 days in 2020 and made 15,593 and 13,605 images, respectively. Images were manually checked and labeled via the software digiKam (Version 5; https://www.digikam.org). We did not label the avi-files, but reviewed more than 9,000 video sequences to aid species identification. We summarized detections of the same taxon within 5 minutes for each location to define independent trigger events or encounters by avoiding repetitive detection of the same individual lingering in the focal view of the devices. We could not determine one encounter in 2019 and another 15 encounters of small mammals in 2020, due to blurred or bad quality images. All other animal encounters could be identified at the genus- or species-level.