The purple line traces the (smoothed) percentage of 30-s epochs at night during which a bird (no. 38) broke an infrared beam by crossing the cage. The blue line shows the same information for the hours of light. Note the broad peak in activity between March and June and the broader and lower peak in activity between August and December, corresponding to the spring and fall migrations, respectively. The low levels of activity during July and December through March correspond to the summer and winter nonmigratory periods. The sharp high peaks (orange ovals) in early August of 2003 and early February of 2004 mark brief periods of experimenter-induced sleep restriction.

Behavioral state was scored across 24-h (noon to noon) periods using a combination of video and electrophysiological recordings for birds in a nonmigratory (n = 5) and migratory (n = 8) state. The plots and table reflect the average for all birds in each group. All recordings were performed under a 12:12 LD photoperiod with lights turned off at 18:00 and on at 06:00.
(A) Proportion of time in each behavioral state for nonmigrating (top) and migrating (bottom) birds. The proportion of every 10-min period spent in each sleep/wakefulness state was calculated for each bird and then averaged across all birds: wakefulness (black), drowsiness (gray), SWS (blue), and REM sleep (red). Note that overall sleep propensity in migrating birds is greatly diminished between approximately 22:30 and 06:00. Note also the increased propensity for REM sleep from 18:00 to 20:00 as compared to the same time period when not migrating.
(B) Sleep and REM sleep latencies. Sleep latency was calculated as the length of time from lights out until the first occurrence of sleep (in all cases SWS) for birds in a nonmigratory and migratory state; average sleep latency did not differ significantly between nonmigrating and migrating birds. REM sleep latency was calculated as the length of time from sleep onset to the first occurrence of REM sleep. Note that REM sleep occurred earlier in sleep during migration for all five birds that were recorded in both a nonmigratory and migratory state (t = 3.3, paired, two-tailed, p < 0.05). Note also that the REM sleep latencies for the three birds recorded only in a migratory state were shorter than the shortest REM sleep latency in nonmigrating birds.
(C) Sleep percentages. Average daily percentages of sleep and wakefulness states for birds in a nonmigrating (n = 5) and migrating (n = 8) state. Total sleep is the sum of SWS and REM sleep. For all states of vigilance, values for the migrating condition differed significantly from the nonmigrating condition (p < 0.01, after Bonferroni correction). The proportion of total sleep occupied by REM sleep was not significantly different between migratory states.

This figure shows the time course of EEG power density in the 1.5- to 2.5-Hz band in SWS during the dark phase for the left (blue) and right (red) hemisphere in nonmigrating sparrows. Curves represent mean 2-h values with standard error of the mean (n = 5). The EEG power density in the 1.5- to 2.5-Hz band of each 2-h interval is expressed as a percentage of the average EEG power in the 1.5- to 2.5-Hz band over all SWS epochs (dashed line = 100% of average 1.5- to 2.5-Hz power). The last 2-h interval is excluded since not all birds exhibited SWS during this time. A two-way, repeated measures ANOVA, with “hemisphere” and “2-h intervals” as factors, revealed a significant effect of the 2-h interval (F = 5.60, p < 0.05 with the Greenhouse–Geisser correction); neither an effect of hemisphere, nor an interaction between hemisphere and interval reached statistical significance (p > 0.1).

The average value for all birds of nighttime (purple) and daytime (blue) activity (percentage of 30-s epochs containing at least one infrared beam break) and the accuracy (green) and total number of responses (red) on the acquisition task (n = 8). As in Figure 1, the sharp high peaks (orange ovals) in early August of 2003 and early February of 2004 mark brief periods of experimenter-induced sleep restriction.

This figure shows four 1-min samples each containing three EEG recordings (left hemisphere vs. right hemisphere [L v. R], left hemisphere vs. neutral reference [L], and right hemisphere vs. neutral reference [R]) and one EMG from one sparrow (no. 65) depicting the typical electrophysiological correlates of each behavioral state.
(A) Transition from SWS (blue) to wakefulness (black).
(B) Drowsiness (gray).
(C) Transition from SWS (blue) to a bout of REM sleep (red), and then a brief awakening (black), followed by a return to SWS (blue).
(D) Wakefulness during a period of migratory restlessness.
High-amplitude artifacts associated with gross movements are shaded with a gray background. (A–C) were recorded on 11 August 2003, during the summer while the bird was in a nonmigratory state, and (D) was recorded on 11 October 2003, during the fall migratory period.

REM sleep as a proportion of total sleep time is plotted for every 10-min period during the dark phase for birds in nonmigrating (n = 5) and migrating (n = 8) states. The individual dots represent the average for each 10-min period; the solid line is a spline fit to these data. The dashed line represents the absolute amount of REM sleep, as a percentage of recording time. Note that the fit for the migrating birds is truncated, not only because very few periods after midnight had any REM sleep, but also because no point after midnight was based on more than one bird.

Average values for accuracy and the total number of responses on the acquisition and performance tasks are shown for the 3 d preceding sleep restriction, the 3 d of sleep restriction, and the 3 d following sleep restriction; boxes represent the 25th to 75th percentile of data with the median indicated by the line across the box. The “whiskers” extend from the quartiles to the most extreme value less than 1.5 times the interquartile range. Points outside the whiskers are plotted with small circles. Accuracy and total number of responses decreased significantly in both the acquisition and performance tasks following sleep restriction (n = 7).

Accuracy and the total number of responses during the acquisition task are shown for the 3 wk of spring and fall during which each bird was most active (orange) at night and for the 3 wk during summer and winter during which each bird was the least active (purple) at night. Note that the data for winter are plotted twice to facilitate comparison. In contrast to sleep restriction imposed by the experimenters (see Figure 6), sparrows maintained high levels of accuracy on the acquisition and performance tasks during periods of sleep restriction associated with migration. Furthermore, the total number of responses reached the highest values during the fall migration, in contrast to the decline in responding following experimenter-imposed sleep restriction.