Kerk L Phillips, Mark H Showalter, Eric R Eide, James H Cardon
A dynamic optimization framework can capture important qualitative features of fragmented sleep in nesting chinstrap penguins. The findings suggest that microsleeps may reflect adaptive trade-offs between restorative sleep and wakeful activity under ecological constraint. Because the model treats sleep as a binary behavioral state, it does not distinguish unihemispheric from bihemispheric slow-wave sleep. We therefore interpret the model primarily as a model of microsleep timing and fragmentation rather than as a complete model of penguin sleep architecture or fragmented sleep across wild animals.
BACKGROUND: Sleep in wild animals reflects both physiological regulation and ecological demands. Nesting chinstrap penguins exhibit extreme sleep fragmentation, obtaining substantial sleep through thousands of brief microsleep episodes while maintaining vigilance.
OBJECTIVE: To examine whether a dynamic model integrating sleep regulation with ecological trade-offs can account for the fragmented sleep patterns observed in nesting chinstrap penguins.
METHODS: We adapted a dynamic model of sleep choice in which an animal switches between sleep and wakefulness on the basis of homeostatic sleep pressure, circadian alignment, switching costs, and stochastic waking opportunities. The model was calibrated to published sleep data from 14 nesting chinstrap penguins using five simulated moments: average hours of sleep per day, the standard deviation of hours of sleep per day, the number of sleep events, the maximum sleep duration, and the percentage of sleep durations less than 10 seconds.
RESULTS: The model reproduced several key qualitative features of penguin sleep. Simulated sleep episodes were very short, with an average duration of 2.774 seconds compared with 3.91 seconds in the observed data. The model also generated frequent transitions between sleep and wake, closely matching the observed number of daily sleep events and reproducing the maximum observed sleep duration. However, it underestimated total sleep time and produced a larger share of very short episodes than observed in the data. Estimated parameters implied minimal switching costs, rapid homeostatic adjustment over seconds-long bouts, and a strong relative weight on sleep utility within the fitted model.
CONCLUSION: A dynamic optimization framework can capture important qualitative features of fragmented sleep in nesting chinstrap penguins. The findings suggest that microsleeps may reflect adaptive trade-offs between restorative sleep and wakeful activity under ecological constraint. Because the model treats sleep as a binary behavioral state, it does not distinguish unihemispheric from bihemispheric slow-wave sleep. We therefore interpret the model primarily as a model of microsleep timing and fragmentation rather than as a complete model of penguin sleep architecture or fragmented sleep across wild animals.