Evidence that birds sleep in mid-flight

Abstract

Many birds fly non-stop for days or longer, but do they sleep in flight and if so, how? It is commonly assumed that flying birds maintain environmental awareness and aerodynamic control by sleeping with only one eye closed and one cerebral hemisphere at a time. However, sleep has never been demonstrated in flying birds. Here, using electroencephalogram recordings of great frigatebirds (Fregata minor) flying over the ocean for up to 10 days, we show that they can sleep with either one hemisphere at a time or both hemispheres simultaneously. Also unexpectedly, frigatebirds sleep for only 0.69 h d(-1) (7.4% of the time spent sleeping on land), indicating that ecological demands for attention usually exceed the attention afforded by sleeping unihemispherically. In addition to establishing that birds can sleep in flight, our results challenge the view that they sustain prolonged flights by obtaining normal amounts of sleep on the wing.

Figure 1. Measuring the brain state and flight mode of flying frigatebirds.

(a) Great frigatebird with a head-mounted data logger for recording the electroencephalogram (EEG) from both cerebral hemispheres and head acceleration in three dimensions. A back-mounted GPS logger recorded position and altitude. Photo: B.V. (b) Overhead view of a great frigatebird skull showing (1) the position of the cranial bulge (shaded grey) overlying the hyperpallium of each hemisphere, (2) the position of the epidural electrodes (red dots, EEG; green dot, ground) and (3) the data logger (black rectangle) just posterior to the naso-frontal hinge (arrow). Scale bar is 10 mm. (c) All GPS tracks for individual birds coded with different colours. The Galapagos Islands are outlined with black lines and the study site (Genovesa) is marked by a star. Ocean depth (m) is coded with grey scale. (d) High temporal resolution (1 Hz) 10 min flight trajectory recorded with GPS from a frigatebird (see Supplementary Movie 1 for 3D visualization) showing the circling (soaring) and straight (gliding) flight modes typical of Fregatidae (Methods). (e) Altitude, ground speed and airspeed (computed from the GPS data in (d)), tangential and centripetal (radial) low-pass filtered acceleration, and the absolute value of total acceleration (measured by an accelerometer) for the flight in (d).

Figure 2. Unihemispheric and bihemispheric sleep in flight.

(a) Recording of head acceleration in three dimensions (sway, surge and heave) and electroencephalogram (EEG) activity from the left (L) and right (R) hemispheres showing the transition from wakefulness to SWS following the cessation of flapping (red bars). Brief episodes of dropping (green bars) occur after this episode of sleep. Expanded views (bottom) show wakefulness characterized by low-amplitude, high-frequency EEG activity in both hemispheres, infrequently punctuated by isolated high-amplitude, slow waves (*), and SWS characterized by continuous high-amplitude, slow waves, in this case, primarily in the left hemisphere. The red arrow (top) marks an episode of apparent REM sleep (expanded in Supplementary Fig. 13). (b) Recording from the same bird showing an episode of bihemispheric SWS (BSWS) and unihemispheric SWS (USWS), including expanded views of both states. The mean duration of episodes of sleep was shorter than these long episodes used to demonstrate USWS and BSWS in flight. These recordings are from frigatebird 13 (Supplementary Fig. 7).

Figure 3. Slow wave sleep electroencephalogram (EEG) asymmetry is related to circling flight.

(a) Distribution of awake and SWS 4 s epochs (all birds combined) occurring at different sway accelerations (0.02g₀ bins) on land and in flight at night. On land (top), the values were clustered around zero while awake and in SWS indicating that the birds held their head straight during both states. Although the birds also held their head and wings straight while awake in flight, SWS, in most cases (70.57%), occurred with circling flight to the left and right, as reflected by sway acceleration <−0.175g₀ and >0.175g₀ (dashed vertical lines). (b) Diagram showing the wing and head angle relative to the horizon during circling flight to the left calculated from the accelerometry. The corresponding brain state (see below) is also shown. (c) Recording showing the relationship between asymmetric SWS (ASWS) and acceleration along the sway axis; during ASWS with greater EEG slow wave activity (SWA; 0.75–4.5 Hz power) in the left hemisphere (ASWS-L) the sway axis showed high values corresponding to circling to the left, and when SWA was greater in the right hemisphere (ASWS-R), the sway axis showed low values corresponding to circling to the right. Same bird as in Fig. 2. (d) The relationship between sway acceleration and type of SWS in flight for all birds combined. Data from (a) are partitioned according to the type of SWS as defined in the main text; ASWS-L, ASWS-R and bihemispheric SWS (BSWS). (e) Relationship between flight mode (sway acceleration) and SWS in flight for data from (a) partitioned according to the interhemispheric asymmetry in gamma activity (30–80 Hz power); asymmetric gamma with greater gamma in the left (AGamma-L; AI⩾0.1) or right (AGamma-R; AI≤−0.1) hemisphere and bihemispheric (symmetric) gamma (BGamma; −0.1<AI<0.1). The overall relationship between circling flight, brain state and probable eye state is summarized in (b); awake hyperpallium (green) and sleeping hyperpallium (blue) and the corresponding relative difference in EEG SWA. The green arrows show the general direction of visual flow while circling to the left.

Figure 4. Frigatebirds sleep more and deeper on land than in flight.

(a) Time spent awake and in SWS and REM sleep in flight and on land. (b) Electroencephalogram (EEG) slow wave activity (SWA; 0.75–4.5 Hz power) while awake and in SWS and REM sleep in flight and on land (median and quartiles for the median). For SWS, SWA is shown for, (1) bihemispheric SWS (BSWS), (2) asymmetric SWS (ASWS) for the hemisphere with greater SWA (ASWS+) and (3) ASWS for the hemisphere with lower SWA (ASWS–). (c) Decline in SWS-related SWA during the first 12 h since landing (top) and the corresponding sleep staging (bottom) in one bird; awake (green), SWS (blue), and REM sleep (red). In the photoperiod bar (middle) grey reflects night.