Dive studies across mammals, birds, reptiles, and fish often focus on deep dives, and shallow-water diving has tended to be overlooked. For air-breathers, foraging in shallow water poses challenges since the lungs generate buoyancy, and shallow divers must trade off the extent of inhalation against the negative buoyancy needed to avoid floating to the surface. Using high-resolution depth loggers, we addressed this knowledge gap around the ecology of shallow water diving at a foraging site for hawksbill turtles (Eretmochelys imbricata) where depth was typically <3m. Contrary to predictions, dive durations were long, particularly at night (mean dive duration per turtle: 17-61 min, n=12 turtles, n=2576 nocturnal dives), despite warm water temperatures (24-37oC). Dive efficiency (% time submerged) for hawksbills was 98%, the highest recorded for any air-breathing marine vertebrate including penguins (60-78%), seals (51-91%), cetaceans (68-87%), and other sea turtle species (68-95%). Hawksbills usually dive for much longer (42-286% increase) than green and loggerhead turtles when depth and temperature are accounted for. Hawksbill turtles likely forage in very shallow water to reduce predation risk from sharks: of 423 hawksbills captured by hand, none had any evidence of shark attack, although large sharks were present in nearby deeper water. Our results challenge the prediction that shallow water dives by air breathers will usually be short and open the way for comparative studies of the ecology of shallow water diving in a range of other taxa. Our work emphasizes the likely importance of predation risk in shaping patterns of habitat utilisation.

High-resolution time-depth recorders (TDRs) were deployed on hawksbill turtles foraging in Turtle Cove at the southern tip of the horse-shoe-shaped lagoon of Diego Garcia atoll in the Chagos Archipelago, Indian Ocean (7.42oS, 72.46oE), during October 2012 (n=5) and February to March 2021 (n=7). Turtles were captured at low tide in shallow water (<1 m) by approaching slowly and carefully from behind before seizing the carapace, whereupon turtles were carried ashore. The curved carapace length (CCL) and curved carapace width (CCW) were measured using a flexible tape, and mass recorded with a suspended scale (Pesola Macro line 50 kg scale, Pesola AG, Switzerland). Maximum precision TDRs (G5 data loggers, 10 bar sensors with 0.03 m resolution, Cefas Technology Limited, Lowestoft, UK) were attached proximally to the trailing edge of flippers, using wire ties to fix them onto flipper tags placed at the thin junction between adjacent thick flipper scales (Supplementary Information Fig. S1). The TDR data sampling rate was every 1-5 seconds for pressure/depth and every 30-60 seconds for temperature. TDRs were recovered after various intervals when the turtles were next encountered in the cove (SI Table S1). The first four hours of each deployment were excluded in case of altered behaviour following capture and attachment (Thomson et al. 2012). Overall data analysis and visualisation were performed in R (R Core Team 2021). Zero-offset correction was scripted directly in base R. Individual dive characteristics and post-dive surface intervals (PDSI) were measured using MultiTrace Dive (Jensen Software Systems, Hamburg, Germany) with a minimum duration of 5 min and a threshold depth of 0.2 m, as has been described for shallow divers by Hays et al. (2007).

Any software that can open .txt files, for example Excel, R, etc.