Sampling method
A trail camera (Trophycam HD, and its succesor models by Bushnell Outdoor Products) is installed at each monitoring site during May 2014 to June 2021. Cameras were adjusted to video mode, and the image resolution was 1280×720 pixels (720p) and the frame rate was 30 fps in daytime mode (visible light imaging) and 18 fps in night-vision mode (infrared monochrome imaging). Although the video length was set to 30 seconds, the camera model number 119776 and the later automatically change the video length to 15 seconds in the night-vision mode due to their specifications. The trigger interval was set to 10-second, but it sometimes took more time to recover actually. Cameras were fixed to tree trunks at a height of about 1 m. Data collection and battery exchange have been conducted biannually. In 2014, the data collection and battery exchange were conducted in October and partially in July. Mammals that appeared in the videos were identified and recorded to the event list. The duration between battery exchanges was defined as the “occasion”, and the start and end times of each occasion for each camera were recorded. The occasion values provide information on the durations for which the camera traps were active, and make it possible to calculate records per unit time. An occasion end time was defined as the time when the last video file was recorded, if a camera was inactive due to battery exhaustion or mechanical failure at the visit.
The sensor cameras used in this study are essentially suitable tools for monitoring medium to large sized mammals (O'Brien et al. 2011). The capture efficiency of small mammals (e.g. rodents and shrews), can be highly sensitive to small variations in the installation of cameras, and we do not recommend treating the data on small mammals belonging to the orders Rodentia, Soricomorpha and Chiroptera as correlative indices of abundance.
