Camera set up
For the camera setup, we used a Raspberry Pi 4 microcomputer circuit board with a 5-megapixel 1080P camera (25 mm x 24 mm x 17 mm) with a 130⁰ adjustable night vision fisheye lens (focus 2.5 cm) and a single 5 mm infrared LED light (Fig. 1). We removed the round part of the infrared LED light using a belt sander and added a small semi-transparent piece of sticky tape to diffuse the light throughout the filming chamber (Fig. 1). A Rpi-RTC DS1307 clock module keeps accurate time while not connected to the internet. The microcomputer was placed in an IP65 waterproof electrical junction box (158 x 90 x 60 mm) with four 1 g silica gel desiccant packets. Three holes were drilled in the junction box: two on the long side for waterproof cable glands (fits cables between 3–6.5 mm) and one on the bottom for the PVC male adaptor (SKU 436-010; 3 cm). The power source cable (USB type C) and the USB extension cable were inserted individually through one of the two cable glands and together into a 1.5 m long stainless-steel shower hose to prevent damage by rodents (which chew exposed rubber or plastics). On the side of the stainless-steel hose, the cables were inserted individually through cable glands in an IP67-rated waterproof PVC box (38.8 x 28.9 x 24.3 cm). In this box, the USB power cable is connected to a power converter (12 V to 5 V) connected to a 130 Ah LiFePO4 lithium battery (10 kg). A 1 TB USB drive is connected to the USB extension cable (Fig. 1).
Trials by (Mouy et al. , 2020) have found that USB storage used more energy than SD card storage, therefore lowering battery life, and USB storage was less reliable because of its more fragile connection. However, O’Brien et al. (2023a, b) and Kallmyer et al.(2017) successfully used 64 GB and 32 GB USB storage in their study, respectively. In this study, we chose a 1 TB USB storage because the USB drive and battery are separated from the microcomputer; which has the benefit that the USB drive and battery are replaced in the field after 21 days (often during rain and 90%> humidity), we do not have to open the box with the microcomputer and camera to potential damage by moisture. The total cost for the 21-day camera setup is AUS$790.75 (Table 1).
The camera length can be easily adjusted for different habitats. For example, when filming animals under leaves or artificial shelters (Fig. 4), we used the camera setup with the camera directly placed in the PVC male adaptor (4 cm width) and the lens focusing at 2.5 cm. To ensure that water could not enter the PVC male adaptor, a PVC coupling was placed with a glass lens glued inside and a rubber O-ring was placed between the connections. This way, if necessary, we could still change the focus of the camera by removing the coupling. For filming under natural or artificial wood shelters (‘logs’; see paragraph “artificial shelters” below), we use a 20 cm long and 4.2 cm wide PVC pipe in the converter, which goes over the male adaptor. The camera and the infrared LED light were held in place at the end of the PVC pipe with a 3D-printed holder against the glass lens. Both the camera and battery box were covered with square mesh (mesh size 1 x 1 cm) to protect the plastics from chewing by rodents. To record the temperature and humidity a datalogger (Hygrochron iButton) was placed in the chamber when the camera was recording (Fig. 5).
The Raspberry Pi was coded to start recording when it powered on, and for continuous video recordings of 15 minutes each. These recordings run back-to-back but a 15-minute duration was chosen to keep the file size of each recording manageable (O’Brien et al. , 2023a). The ‘continuous’ recording stretched for 21 days. Videos were saved to a 1 TB USB drive, creating folders of the month followed by folders of the day. The videos themselves were labelled with time (hour.minute.second) (Appendix 1; for detailed instructions on how to program the Raspberry Pi see e.g. Hereward et al. , 2021; Youngblood, 2019).