Animals and instrumentation

The Galápagos National Park Service approved of and granted the research permits for this work. During August, 2014, adult females (N=15) caring for chicks on the coast of Darwin Bay, Genovesa Island, Galápagos, Ecuador (0°19'5.57"N, 89°57'1.23"W) were caught by hand on their nest at night. The chick was covered to keep it warm and safe while its mother was instrumented. Using isoflurane anaesthesia and aseptic methods¹⁶, for each cerebral hemisphere, EEG sensors were placed on the dura overlying the anterior (A) and posterior (P) hyperpallium, a structure that forms a pronounced bulge in the cranium of frigatebirds (Fig. 1b; for a CT scan of a similar skull see, www.digimorph.org/specimens/Fregata_magnificens/); the sensors were 8 mm apart along the AP axis, spanning the most pronounced portion of the cranial bulge, and 4 mm from the midline (Fig. 1b). A fifth sensor was placed laterally on the left hemisphere for the electrical ground. The gold-plated, round-tipped (0.5 mm diameter) sensors were secured with a small amount of dental acrylic cured with an ultraviolet light (Clearfil SE Bond, Kuraray Noritake Dental, Japan and Tetric EvoFlow, Ivoclar Vivadent AG, Schaan, Liechtenstein) and connected to a flexible, insulated spring wire (no. 276-0146-001; DSI, St. Paul, MN). The wires were soldered to a data logger (Neurologger 2A; www.evolocus.com, see also www.vyssotski.ch/neurologger2) which included a 3.6 V lithium battery (Saft LS-14250; www.saftbatteries.com) and a three-axis accelerometer (LIS302DLH; STMicro-electronics). The logger was glued (Hystoacryl, Aesculap AG, Germany and Pattex, Repair Gel, Henkel AG & Co. KGaA, Germany) to the skin and feathers just posterior to the naso-frontal hinge (Fig. 1b). The logger was configured to record bipolar EEGs from the left and right hemispheres, and acceleration in the three cardinal directions continuously at 200 Hz for up to 10 days. A GPS data logger (i-gotU, GT-600; www.i-gotu.com) configured to record position every 5 min was attached to the back feathers with gaffer tape (tesa, no. 4671; www.tesatape.com). The total weight (55 g) of the equipment was 4.0% of the birds' weight (1366.79±24.09 g, s.e.m.). Fourteen of the 15 birds were recaptured 7.79±0.49 d (s.e.m.; range, 5.37–10.45 d) later, after returning from at least one foraging trip. In nine of the birds, we obtained recordings (16.40±3.33 h, s.e.m., in duration; Supplementary Fig. 9g) after they returned to the nest to evaluate sleep on land. At the end of the study, the equipment was removed under anaesthesia and the birds were released. On release, the birds resumed nesting behaviour indistinguishable from that observed in undisturbed birds. Finally, to validate our analysis of the flight trajectories in great frigatebirds, we used data recorded from two magnificent frigatebirds (Fregata magnificens) in a pilot study conducted in the French Guiana using a GPS data logger (GiPSy-2, www.technosmart.eu) with a 1 Hz sampling rate combined with a 3D acceleration logger (25 Hz rate; AXY-1, www.technosmart.eu; Fig. 1d,e).