How fishing cats Prionailurus viverrinus Bennett, 1833 fish: describing a felid’s strategy to hunt aquatic prey

Study area, data collection, analysis: Our study area consists of three sites, including the Howrah district (22°27′40.66″N, 88°0′32.71″E) in the lower Gangetic floodplains of West Bengal and two coastal wetlands in the Mahanadi floodplains along the eastern coast – Paradip (20°18′50.63″N, 86°36′45.49″E) and Chilika (Figure 1). All are human-dominated and/or human-used landscapes. We initiated a participatory-science initiative, ‘Know Thy Neighbour’, in which we used 20 camera traps (14 Covert Illuminator 9 and 6 Browning IR) at each site opportunistically over 2.5 years, starting from November 2016. We consulted local people and as suggested by them we searched for probable sites to place camera traps along shallow ponds or water channels by locating scats and/or tracks of felids. The cameras were set on video mode and were active from 5:30 p.m.–6:00 a.m. Each trained volunteer picked up traps in the early morning and repositioned them at the same location at night to reduce chances of misplacement/theft. We assumed that we would not fail to miss fishing cat activity when camera traps were nonactive, since fishing cats are primarily nocturnal (Hunter 2019) and villagers mostly reported seeing them after dark. We collected over 200 camera-trap video recordings from the participatory-citizen-science initiative of which 197 videos were selected to analyse hunting behaviours of fishing cats. All videos were analysed by the first author to avoid introducing observer bias. Consecutive videos from the same camera trap on the same night, which were taken less than a minute apart, were considered as single data, as we assumed lack of independence between closely occurring activities so that we could minimise bias towards individual behavioural tendencies in our analysis. ‘Behaviour States’ were defined as behavioural patterns of relatively long duration (Martin and Bateson 1993). These were further divided into ‘behaviour sub-states’. Behavioural patterns of relatively short duration were classified as ‘Behaviour Events’ (Martin and Bateson 1993). We calculated the proportion of time spent performing each State and the frequency of each Event using an ethogram framework. If two separate individuals were detected in the videos, these were considered independent behavioural events and analysed as two different samples. To examine activity patterns, videos (n = 98) were selected, which were recorded at least 30 min apart to ensure independence. These videos were then used to generate line graphs to understand peak activity patterns of fishing cats. In total, we analysed 134.43 min from 197 camera trap video footages constituting 198 data points from all three study sites combined; 77.48 min of video footage from Howrah, 15.95 min of video footage from Paradip and 41 min of video footage from Chilika.