Methods

Tracking data

The validity of our modelling approach hinges on constructing ENMs from multiple colonies. We thus compiled the most comprehensive database of tropical seabird tracking data to date, featuring 16 species representing 60 colonies around the globe and totalling ~12000 individual trips (Table 1; Supporting information). Species that had more than 10 tracking studies were modelled individually. Tropicbird Phaethon and frigatebirdFregata species were grouped respectively (Table 1) to boost sample size and make family level models, acknowledging that foraging differences between species within these families have been observed (e.g. Mott et al. 2016). Noddy Anous species were modelled as a family group (following evidence of foraging niche overlap; Surman & Wooller 2003, Shephard et al. 2018), and large tern Thalasseusand Hydroprogne species were aggregated into a meta-family group for modelling. The exceptions were sooty terns, which were modelled as an individual species (despite few tracking studies) because of their unique, highly pelagic foraging niche among terns (Surman & Wooller 2003). Wedge-tailed shearwaters were also modelled separately for short trips and long trips because we had good a-priori knowledge of dual-foraging behaviours during breeding on the GBR (Congdon et al. 2005, McDuie et al. 2015) and in New Caledonia (Weimerskirch et al. 2020), and sufficient tracking studies to do so. Both individual species (e.g. brown booby BRBO) and family groups (e.g. large terns TERN) are hereafter referred to as “modelled species”.
Table 1 Summary of quantity and coverage of tracking data analysed for each species, and the grouping of species with less data.