Seed removal plots
To assess whether the arrival of a novel predator affected the seed harvesting behaviour of granivorous melomys, we established seed removal plots at each site and sampled them each trapping session (night 6). After trapping and open field tests were conducted and melomys had been returned to their capture location, we set up 81 seed plots at each site by scraping away leaf litter with a shovel to create bare earth plots. These bare earth plots were created so that they were located in the centre between four Elliott traps within the 10x10 trapping grid. All seed plots were located randomly with respect to “distances to cover” but were all located on relatively open patches of ground. Sufficient within site replication (n = 81) significantly reduces the likelihood of distance to cover biasing population-level responses to seeds. Just before dark on the night of the seed removal experiment, we placed a single wheat seed in the centre of each bare earth plot. These seeds were either unscented, control seeds (n = 40) or predator-scented seeds that had been maintained in a sealed clip-lock bag filled with freshly collected northern quoll fur (n = 41). The placement of predator-scented and unscented seeds was alternated so that there was a chequered arrangement of scented and unscented seeds across the site. To ensure that the predator-scent was strong enough to be detected by melomys, along with the predator-scented seeds, we also placed a few strands of quoll fur around the predator-scented seeds. Before light the next morning, we returned back to each plot and counted the number of seeds of each scent-type that were removed from the plot. Melomys are the only nocturnal granivorous animal that occurs on Indian Island, and to avoid diurnal granivorous birds from removing seeds we conducted this experiment during the night only.