Measuring predation pressures and identification of predators
The modes that predators use to attack gastropods are classified as shell entry, shell crushing, shell boring, and whole-shell swallowing (Vermeij, 1979; DeWitt & Robinson, 2000). Some recaptured snails that were dead showed evidence of shell-crushing, so we considered that they died from attacks by predators rather than old age. We compared the number of predated snails between the mainland and the island using Fisher’s exact test. To examine whether the number of predations was associated with shell colour and growth stage, we also analysed them with a generalised linear mixed model (GLMM). Predation (1) and non-predation (0) were used as binary response variables, while the shell colour, growth stage (adults 1; juvenile 0), and their interaction were used as fixed effects. The month when the dead snails were recaptured was used as a random effect. “lme4” (Bates et al. , 2007) was used with logit as a link function in the statistical package R (R Core Team, 2020).
To identify the snails’ predators, camera-trap experiments were conducted at the mainland study site from December 2019 to July 2020 and at the island study site from April 2017 to February 2018. At these study sites, snails were secured to trees with 5-cm-long strings (Daiso Ltd., Hiroshima, Japan; diameter 0.2 mm) and monitored using trail cameras (TROPHYCAM HD3 ESSENTIAL, Bushnell Ltd., Kansas, USA). Each camera was attached to a tree or on a tripod and operated on a motion trigger system. The camera automatically recorded videos when an animal entered the camera’s sensor range. 4 and 5 camera traps were set up at the mainland and island study sites, respectively. The predators were identified from these videos in our laboratory.