ArticlePDF Available

Sleeping under the risk of predation

  • Max Planck Institute for Ornithology - Seewiesen

Abstract and Figures

Every studied animal engages in sleep, and many animals spend much of their lives in this vulnerable behavioural state. We believe that an explicit description of this vulnerability will provide many insights into both the function and architecture (or organization) of sleep. Early studies of sleep recognized this idea, but it has been largely overlooked during the last 20 years. We critically evaluate early models that suggested that the function of sleep is antipredator in nature, and outline a new model in which we argue that whole-brain or ‘blackout’ sleep may be the safest way to sleep given a functionally interconnected brain. Early comparative work also suggested that the predatory environment is an important determinant of sleep architecture. For example, species that sleep in risky environments spend less time in the relatively vulnerable states of sleep. Recent experimental work suggests that mammals and birds shift to relatively vigilant (lighter) states of sleep in response to an increase in perceived risk; these results mirror the influence of stress on sleep in humans and rats. We also outline a conceptual model of sleep architecture in which dynamic changes in sleep states reflect a trade-off between the benefits of reducing a sleep debt and the cost of predation. Overall, many aspects of plasticity in sleep related to predation risk require further study, as do the ways in which sleeping animals monitor predatory threats. More work outside of the dominant mammalian paradigm in sleep is also needed. An ecologically based view of sleeping under the risk of predation will provide an important complement to the traditional physiological and neurological approaches to studying sleep and its functions.
Content may be subject to copyright.
Sleeping under the risk of predation
*Department of Ecology & Organismal Biology, Indiana State University, Terre Haute
yDepartment of Psychiatry, University of Wisconsin, Madison, U.S.A.
(Received 2 November 2004; initial acceptance 29 November 2004;
final acceptance 26 January 2005; published online ---; MS. number: ARV-30)
Every studied animal engages in sleep, and many animals spend much of their lives in this vulnerable
behavioural state. We believe that an explicit description of this vulnerability will provide many insights
into both the function and architecture (or organization) of sleep. Early studies of sleep recognized this
idea, but it has been largely overlooked during the last 20 years. We critically evaluate early models that
suggested that the function of sleep is antipredator in nature, and outline a new model in which we argue
that whole-brain or ‘blackout’ sleep may be the safest way to sleep given a functionally interconnected
brain. Early comparative work also suggested that the predatory environment is an important determinant
of sleep architecture. For example, species that sleep in risky environments spend less time in the relatively
vulnerable states of sleep. Recent experimental work suggests that mammals and birds shift to relatively
vigilant (lighter) states of sleep in response to an increase in perceived risk; these results mirror the
influence of stress on sleep in humans and rats. We also outline a conceptual model of sleep architecture in
which dynamic changes in sleep states reflect a trade-off between the benefits of reducing a sleep debt and
the cost of predation. Overall, many aspects of plasticity in sleep related to predation risk require further
study, as do the ways in which sleeping animals monitor predatory threats. More work outside of the
dominant mammalian paradigm in sleep is also needed. An ecologically based view of sleeping under the
risk of predation will provide an important complement to the traditional physiological and neurological
approaches to studying sleep and its functions.
2005 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd.
Every animal studied to date engages in some form of
sleep or sleep-like behaviour (Hartse 1994; Tobler 2000;
Rattenborg & Amlaner 2002; Lesku et al., in press). Recent
studies on Drosophila (Shaw et al. 2000), bees, Apis
mellifera (Sauer et al. 2003, 2004) and crayfish, Procamba-
rus clarkii (Ramo
´n et al. 2004), suggest an ancient
evolutionary homology in sleep across the animal king-
dom. The actual functions of sleep are, however, still ill
defined (Rechtschaffen 1998), but new theories are arising
as advances are made in the molecular biology and
biochemistry of sleep (see Siegel 2003; Cirelli et al.
2004). Most evidence suggests that sleep has some
important restorative function (Siegel 2003, see below).
Evidence from long-term sleep-deprivation studies even
suggests that death may occur with a severe lack of sleep
(Rechtschaffen & Bergmann 2002; Shaw et al. 2002). The
nature of the restorative effects of sleep is, however, still
a matter of much study and speculation.
Regardless of its functions, sleep is clearly among the
most prominent of animal behaviours. Humans spend
about one-third of their lives in this behavioural state, and
many mammal species spend even more time asleep
(Zepelin 2000). However, despite its prominence in the
lives of animals, sleep has received little attention from
animal behaviourists. If sleep is considered at all, it is
treated as something that happens when there is nothing
else to do or as some sort of unavoidable constraint.
However, just as with other, better-characterized behav-
iours, sleep is likely to be a behaviour that responds
dynamically and adaptively to a host of environmental
Here, we take such an ‘adaptationist’ perspective
(Mitchell & Valone 1990) on the functions and dynamics
of sleep. We focus on the one clear aspect of sleep that is
apparent across the animal kingdom: when compared to
alert and awake animals, sleeping animals are relatively
unresponsive and unaware of their environment. The
dangers of sleep are thus readily apparent. However,
certain ways of sleeping are probably safer than others,
Correspondence and present address: S. L. Lima, Department of Ecology
& Organismal Biology, Indiana State University, Terre Haute, IN
47809, U.S.A. (email: N. C. Rattenborg is now
at the Max Planck Institute for Ornithology-Seewiesen, 82319
Starnberg, Germany.
0003–3472/05/$30.00/0 2005 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd.
AN IM AL BE HA VI OU R, 2005, --,----
and an explicit consideration of this fact should provide
insight into many aspects of sleep in general. Consider-
ations of predation risk were, in fact, prominent in some
early comparative work on sleep (Hediger 1969; Zepelin
1970; Allison & Cicchetti 1976; Meddis 1977, 1983;
Amlaner & Ball 1983; but see Zepelin & Rechtschaffen
1974). Indeed, the antipredator implications of sleep also
figured prominently in early considerations of the func-
tion of sleep itself (e.g. Snyder 1966; Meddis 1977; see also
Horne 1988; Zepelin 1989). Since this early comparative
work, however, this focus has largely disappeared from
studies on sleep. This situation undoubtedly reflects the
clinical, human-oriented nature of the vast majority of
research on sleep, and perhaps the relative difficulty in
studying aspects of the risk of predation (e.g. Elgar et al.
1988). Nevertheless, predation-risk perspectives have
yielded several insights into behaviour in a variety of
contexts (Sih 1987, 1994; Lima & Dill 1990; Lima 1998),
and we believe they will yield similar insights in the study
of sleep.
Our goal here is thus to provide the first wide-ranging
treatment of sleep under the risk of predation. We review
past work on this topic and the historical role of anti-
predator thinking in the evolution of our view of sleep
and its function. We devote much attention to several
aspects of sleep architecture, including the distribution
and function of various states of sleep. Our focus is mainly
on the fitness or strategic consequences of various pat-
terns of sleep rather than its physiological or biochemical
functions. We highlight the things that we understand
reasonably well, the many areas in need of much more
attention, and the sorts of things that we might expect to
see in future work.
The taxonomic coverage of our review is focused almost
entirely on endothermic vertebrates (especially mam-
mals). This bias is unavoidable because the vast majority
of work on sleep has focused on humans and rats. The
majority of comparative data on sleep also comes from
mammals (Zepelin & Rechtschaffen 1974; Campbell &
Tobler 1984; Zepelin 2000; Rattenborg & Amlaner 2002;
Lesku et al., in press). A fair amount of information is
available for birds as well (Amlaner & Ball 1983, 1994).
Work on sleep in reptiles has been sparse, and work on
amphibians, fish and invertebrates has been even rarer
(Hartse 1994; Tobler 2000; Rattenborg & Amlaner 2002;
Lesku et al., in press). Nevertheless, we expect that the
strategic considerations outlined below apply to a wide
variety of animals.
Some Sleep Basics
Sleep can be defined using behavioural criteria and
electroencephalographic recordings. Behaviourally, sleep
is a state characterized by (1) typical sleep posture, (2)
behavioural quiescence, (3) increased stimulus threshold
for arousal to an alert state, and (4) rapid reversibility to
wakefulness once aroused (Pie
´ron 1913; Flanigan 1972).
Tobler (1985) added a fifth characteristic of behavioural
sleep: compensatory increase following sleep deprivation.
The criterion of increased arousal thresholds during sleep
is perhaps the most important from the antipredator
perspective. According to these behavioural criteria, every
animal studied to date sleeps in some way (Rattenborg &
Amlaner 2002; Lesku et al., in press).
Sleep may also be defined according to the electrophys-
iological correlates of behaviourally defined sleep. Al-
though several electrophysiological parameters show
changes between wakefulness and sleep, the electroen-
cephalogram (EEG) has received the most attention.
Unlike simple behavioural criteria for sleep, the EEG of
birds and mammals can usually distinguish between two
basic states of sleep. The EEG of an awake mammal or bird
is characterized by low-amplitude, high-frequency waves
(Fig. 1). During a sleep state known variously as quiet
sleep, slow-wave sleep, or non-REM (rapid eye movement)
sleep, the EEG is characterized by high-amplitude, low-
frequency waves (Fig. 1). The other basic sleep state is
variously referred to as active sleep, paradoxical sleep, or
more typically as REM sleep. This sleep state is character-
ized by an EEG more typical of a waking animal (Fig. 1),
but with a general loss of voluntary muscle tone (except
in the ocular muscles). Aside from frequent small
twitches, behavioural quiescence is maintained during
REM sleep.
In some animals, the EEG can be used to distinguish
various stages of non-REM sleep. In humans, non-REM
sleep is divided into four stages that represent increasing
depth of sleep as indicated by wave amplitude in the EEG
and increasing arousal thresholds. The term slow-wave
sleep (SWS) traditionally refers only to stages 3 and 4 of
non-REM sleep in humans (Horne 1988), whereas SWS
refers to all non-REM sleep in most birds and mammals. In
rats, two stages of SWS are recognized, with one (stage 2)
deeper than other (Neckelmann & Ursin 1993). These
stages actually characterize a continuum of changes in the
intensity of SWS rather than discrete categories (Borbe
´ly &
Achermann 2000), and are therefore somewhat arbitrary.
Although they do not experience distinct stages of SWS,
birds also show variability in the intensity of SWS
(Szymczak et al. 1996; Rattenborg et al. 2004). Only one
type of REM sleep is typically identified in birds and
mammals, although two states of REM sleep, tonic and
phasic, are identified in humans (the latter is associated
with eye movement and twitching; Hartmann 1973).
In birds and mammals, the two main classes of SWS and
REM sleep are organized into ‘sleep cycles’, which are
alternating periods of SWS followed by REM sleep. Several
such cycles may occur over a given period of sleep
´ly & Achermann 2000). During an extended period
of mammalian sleep, deep SWS usually dominates the
early sleep cycles (Fig. 2), with REM sleep and less-intense
SWS becoming more prominent as sleep progresses
(Horne 1988; Borbe
´ly & Achermann 2000). A similar
pattern has been observed in some birds (Szymczak et al.
1996; Rattenborg et al. 2004).
The basic classes of SWS and REM sleep apply to most (if
not all) birds and mammals, but the existence of these two
classes of sleep in other vertebrates is far from clear (see
Rattenborg & Amlaner 2002). Obvious behaviourally de-
fined sleep, however, occurs in reptiles (Flanigan et al.
1973, 1974). Several studies have also shown distinctive
changes in EEG recordings during reptilian behavioural
sleep (Rattenborg & Amlaner 2002; Lesku et al., in press).
Whether these distinctive changes are homologous to
avian and mammalian SWS or REM sleep remains to be
shown. Furthermore, some behavioural states (postures)
in sleeping reptiles are sometimes associated with differing
degrees of responsiveness to external stimuli and thus
probably represent different intensities of sleep (Flanigan
et al. 1973). Few studies have examined sleep in amphib-
ians and fish, but clear behavioural sleep occurs in these
taxa as well (Tobler 2000; Rattenborg & Amlaner 2002;
Lesku et al., in press). Invertebrates also engage in
behavioural sleep according to the definition outlined
above, and distinct changes in neural activity have been
seen in sleeping bees (Apis sp.: Kaiser & Steiner-Kaiser
1983; Schuppe 1995) and Drosophila (Nitz et al. 2002);
however, classes of sleep analogous to SWS or REM sleep
have yet to be clearly identified in these taxa (but see
´n et al. 2004).
Animals may be classified as monophasic or polyphasic
sleepers. Monophasic sleepers tend to concentrate sleep
into a distinct portion of the day (Tobler 1989; Ball 1992).
Humans are classic monophasic sleepers, as are most other
primates. Polyphasic sleepers sleep during several periods
that may occur at any time of day. Rats are typically
polyphasic sleepers, as are many rodents and other small
mammals such as insectivores. Some types of mammals
such as ungulates show no clear tendency towards one
category or another (Tobler 1989). Overall, birds are
largely monophasic sleepers, but waterfowl and shorebirds
Sleep stages
Figure 2. Distribution of sleep states across a typical night’s sleep in humans. In humans, non-REM sleep is divided into four distinct stages;
stage 1 is the lightest stage and stage 4 is the deepest stage as measured by EEG amplitude and arousal thresholds. W represents the waking
state and R represents REM sleep. Reproduced from Borbe
´ly & Achermann (2000) with permission from Elsevier.
(a) SWS
(b) REM
(c) Awake
Frequency (Hz)
1 s
100 uV
Figure 1. Electroencephalogram (EEG) of sleep in the laboratory mouse. Shown are representative EEGs of the mouse while (a) in slow-wave
sleep (SWS) or non-REM sleep, (b) in REM sleep and (c) awake. Plots to the right indicate the relative power (in arbitrary units) of various wave
frequencies in the adjacent EEG.
(which can feed both day and night) may be largely
polyphasic in their sleep (Amlaner & Ball 1983).
Finally, many species of birds sometimes engage in
unihemispheric slow-wave sleep (USWS, as opposed to
the bihemispheric sleep typical of most mammals) in
which one of the cerebral hemispheres is awake (to some
extent) while the other sleeps (Amlaner & Ball 1994;
Rattenborg et al. 2000). During USWS, the eye contralat-
eral (opposite and neurologically connected to) the awake
hemisphere is open and monitoring the environment.
Behavioural evidence also suggests that some form of
unihemispheric sleep probably occurs in reptiles (Ratten-
borg et al. 2000). Unihemispheric sleep may be a trait
ancestral to both birds and mammals that was subse-
quently lost in mammals (see Rattenborg et al. 2000).
As mentioned earlier, there are many theories about the
function of sleep. Below are some hypotheses that take
a predation-risk perspective. See Moorcroft (1995), Re-
chtschaffen (1998) and Siegel (2003) for a more complete
discussion of the many proposed functions of sleep.
The Immobilization Hypothesis
Prominent among theories of the function of sleep is
the ‘immobilization hypothesis’ developed by Meddis
(1975, 1977; see also Webb 1975). The basic idea behind
this early hypothesis is that sleep serves a protective role
during times when an animal cannot be usefully engaged
in any other activity. Meddis reasoned that animals not
immediately threatened by predators would be safer if
they passed the time as quietly as possible, in a state of
sleep. Sleep would prevent an animal from moving or
responding to nonthreatening stimuli in ways that might
attract the attention of predators.
A provocative aspect of the immobilization hypothesis
is that sleep serves no restorative function (Meddis 1975,
1977). Sleep is viewed as an optional activity whose
duration depends on the opportunities to do something
else. For instance, under this hypothesis, animals would
sleep longer during longer winter nights to pass the time
safely until daylight brings opportunities for useful activ-
ity. During the shorter nights of summer, sleep times
would be accordingly reduced. There is evidence in
humans and other animals that sleep times can
vary seasonally in this fashion (Amlaner & Ball 1983;
Rattenborg et al. 2004; but see Kohsaka et al. 1992), but
a minimum daily core of sleep may be needed for
restoration (Horne 1988).
Rechtschaffen (1998) outlined several problems with the
immobilization hypothesis. For instance, this hypothesis
cannot easily explain why one often observes a rebound in
sleep time or intensity following a period of sleep depriva-
tion. The immobilization hypothesis also does not readily
explain the existence of various states of sleep, which
themselves may be associated with differing functions,
such as memory consolidation (Walker & Stickgold 2004;
but see Vertes 2004) or energy conservation in endotherms
(Berger & Phillips 1995; see also Pravosudov & Lucas 2000).
The same holds for several metabolic activities largely
unique to sleep (Cirelli et al. 2004). Furthermore, remain-
ing very quietly awake would seem to be safer than sleep,
especially if sleep is an optional activity.
Like virtually all hypotheses of sleep under the risk of
predation, the immobilization hypothesis has never been
modelled formally to determine whether its conclusions
actually follow from its assumptions. Here we address the
question of whether the profound behavioural ‘blackout’
of sleep provides more safety than quiet wakefulness.
From a strategic perspective, two aspects of the immobi-
lization hypothesis are most relevant: (1) the degree to
which immobilization prevents detections and attacks by
predators and (2) the degree to which an animal can
detect or respond effectively to attack as a function of the
degree or depth of sleep. We assume (as per Meddis 1977)
that a sleeping animal is less prone to detection than is
a quietly awake animal (which may be more prone to react
to nonthreatening stimuli, as hypothesized). We also
assume that a quietly awake animal would be better able
to respond to an attack, should one occur.
These two key assumptions are clearly antagonistic and
thus, the option favoured evolutionarily is not obvious.
The simplest way to approach the problem is to assume
that an animal must survive a period Tduring which it
can usefully engage in no activity other than rest or sleep.
Assuming that a predator may appear at any time with
equal probability, the probability of surviving the interval
Tis simply exp(aDT ), where ais the encounter rate with
predators and Dis the probability of death given an
encounter. The variables aand Tare fixed; hence,
maximizing survival is equivalent to minimizing D. The
probability Ditself is the product of the probability of
being detected and attacked by a predator (d) and the
probability of capture given attack (c); each of these
probabilities is a function of the intensity of sleep (s),
which varies from 0 to 1. Quiet wakefulness and deep
sleep are represented by sZ0 and sZ1, respectively.
Intermediate values for sleep intensity are possible, which
might represent lighter stages of sleep.
The animal has behavioural control over the intensity
of sleep; hence, we seek the value of sthat minimizes
DZd(s)c(s). The optimal sleep intensity (s*) depends
strongly on the form of the functions chosen to describe
the probabilities d(s) and c(s). For instance, if both d(s) and
c(s) are linear in s, then Dwill be minimized at one of the
extremes (Fig. 3a). That is, the animal would be either
quietly awake (s*Z0) or deeply asleep (s*Z1). These two
extreme solutions also hold when dand care both
concave-downward functions of s(Fig. 3b). The extreme
favoured is determined by the details of the functions.
Furthermore, an intermediate value of s* is possible when
both dand care concave-upward functions of s(Fig. 3c).
This simple exercise in modelling suggests that immo-
bilization can indeed be superior to quiet wakefulness as
a predator avoidance strategy. However, quiet wakefulness
or even an intermediate degree of sleep can also be
favoured under some circumstances. Thus, it is clear that
immobilization (defined broadly as sO0) is not a unique
outcome of the assumptions put forth by Meddis (1975,
1977). Unfortunately, we do not have nearly enough
information on the behavioural events surrounding at-
tacks on sleeping animals to assess which of the above
outcomes would be more likely (see also Lima 2002). It
does seem reasonable, however, to conclude that the
immobilization hypothesis cannot account generally for
the existence or maintenance of sleep. We nevertheless
suspect that immobilization is an important aspect of the
nature of sleep.
A New Hypothesis: ‘Blackout’ Sleep Makes
Sleeping Safer
Here, we outline a simple strategic model that views the
whole-brain ‘blackout’ nature of sleep as an antipredator
response. More specifically, the function of blackout sleep
may be to achieve the specific functions of sleep as safely
as possible. Our approach differs from (and is comple-
mentary to) the traditional approach to the subject, in
which insights into the function of sleep are sought by
comparing various neural/metabolic processes between
the sleeping and waking brain. We believe that theories
of the function of sleep must also address why these sleep-
associated processes are best accomplished via a behav-
ioural blackout. Without the blackout and resulting
vulnerability of sleep, there would not be anything
enigmatic to explain. Space limitations require that this
model be presented in full elsewhere, so here we will
simply summarize its main points.
We take the position held by a plurality of sleep
researchers and assume that sleep has some sort of critical
maintenance or restorative effect on neural tissue (see
Siegel 2003). We assume that the brain is composed of
distinct neural units that are linked to other such units to
perform various functions for the organism in question.
We further assume that these units could ‘sleep’ separately
from other units (Mahowald & Schenck 1992; Huber et al.
2004). Sleep in a given unit would require that the unit be
taken off-line much as envisioned by Krueger & Oba
(1993, 2002; see also Tononi & Cirelli 2003). With this
configuration, it would be possible for one part of the
brain to sleep while another part is awake and in
a relatively good position to react to threats.
The form of sleep favoured evolutionarily depends on
the nature of the interactions between neural units. With
additive interactions among units, if 50% of the units are
online (awake), then the animal has a 50% chance of
surviving attack. If interactions are multiplicative, then
having 50% of units online leads to much less than 50%
survival. Given additive interactions, it matters little
whether sleep is concentrated into one blackout period
or spread out evenly over a much longer time spent
partially awake. However, multiplicative interactions
among units would lead quickly to blackout sleep; that
is, survival is maximized if all units sleep at the same time.
Simply put, it is safer to sleep by consolidating sleep into
one single (short) block, which increases the time spent
awake and fully able to deal with predators. The option of
spending time partially awake will not lead to greater
0 0.2 0.4 0.6 0.8 1
0 0.2 0.4 0.6 0.8 1
0 0.2 0.4 0.6 0.8 1
D = d(s)c(s)
D = d(s)c(s)
D = d(s)c(s)
Figure 3. Representative views of the immobilization hypothesis for
the function of sleep. Shown is the probability of death (D) given an
encounter with a predator as a function of the intensity of sleep (s). D
is the product of the probability of being detected and attacked by
a predator (d(s)) and the probability of capture given attack (c(s)).
Quiet wakefulness is represented by sZ0, and deep sleep by sZ1.
The large dots indicate the value of sminimizing D(or s*). (a) Case in
which both d(s) and c(s) are linear. Here, s* occurs at one of the
extremes of 0 or 1; s*Z1 for the functions shown. (b) Case in which
both d(s) and c(s) are concave-downward functions of s. Here, s*
also occurs at one of the extremes of 0 or 1; s*Z0 for the functions
shown. (c) Case in which both d(s) and c(s) are concave-downward
functions of s. Here, s* may occur at an intermediate value of s.
safety when interactions between units are multiplicative.
A model addressing the actual time spent asleep should
also consider the various ‘states of vigilance’ that exist in
sleeping animals (see below) and the fact that the level of
risk experienced while sleeping will determine (in part)
the total time spent asleep. Factors related to the benefits
of immobilization may also influence total sleep time.
Overall, blackout sleep may simply be the safest way to
deal with the fact that neural units must be taken off-line
for maintenance and restoration in an interconnected and
interdependent brain (multiplicative interactions are un-
doubtedly the rule in a real brain). It may indeed be the
case that the need for restoration ultimately is responsible
for blackout sleep, but simply demonstrating the need for
restoration will not, in itself, be sufficient to explain the
blackout. As noted by Moorcroft (1995), Rechtschaffen
(1998) and others, many of the proposed restorative
functions of sleep could conceivably be performed to
a large extent while awake. Our simple model suggests,
however, that such a ‘waking’ strategy of sleep organiza-
tion is unlikely to be favoured by natural selection. Finally,
we note that our ideas are compatible with many of the
proposed restorative functions of sleep (Rechtschaffen
1998; Siegel 2003), provided that such restoration can be
done more effectively if neural units are taken off-line.
Thinking about sleep under the risk of predation has
usually focused on the architecture of sleep. By ‘architec-
ture’, we mean the states of sleep, the distribution of such
states within sleep bouts, length of bouts, timing of sleep
across the day, and so forth. Regardless of whether a pre-
dation-risk perspective will ultimately shed much light on
the function of sleep, predation risk has many important
implications for the architecture of sleep.
The Sentinel Hypothesis and REM Sleep
In an early paper, Snyder (1966) proposed the idea that
REM sleep serves a sort of sentinel function because it is in
many respects a state of much neural activity. The basic
idea is that episodes of REM sleep allow an animal to be
ready to escape should an attack occur, and that the brief
arousals to wakefulness that may precede or follow REM
sleep also serve an antipredator function. Voss (2004) has
recently recast the sentinel idea to include other sleep
states and their distribution over time. She suggests that
the arousals associated with REM sleep establish a ‘pro-
tective field’ around the (human) sleeper.
The sentinel hypothesis is appealing in many respects.
Mammals in REM sleep are clearly more physiologically
prepared for wakefulness than are those in deep SWS
(Tolaas 1978). A recent study also suggests that rats are in
a relatively alert state when awakened from REM sleep
(Horner et al. 1997; see also Van Twyver & Garrett 1972).
Furthermore, humans (Nashida et al. 2000; Cote 2002;
Perrin et al. 2002) and rats (Maho & Hennevin 1999) can
detect and process information to a surprising degree
during REM sleep (Bastuji & Garcia-Larrea 1999; Nashida
et al. 2000; Cote et al. 2001; see also below). Paradoxically,
however, arousal thresholds in REM sleep tend to be
higher than in SWS (Dillon & Webb 1965; Van Twyver
& Garrett 1972; Amlaner & McFarland 1981; Neckelmann
& Ursin 1993); hence, time in REM sleep could be
relatively dangerous (see below). In addition, the brief
arousals sometimes associated with REM sleep (Van
Twyver & Garrett 1972) may not be frequent enough to
have much benefit from an antipredator perspective (see
Bednekoff & Lima 1998).
Overall, the validity of the sentinel hypothesis is not yet
established. This hypothesis has never really been tested,
and its predictions are not obvious. One simple prediction
might be that REM sleep would become more prominent
in species that suffer higher predation, or when the
perceived risk of predation increases. As we will show in
the next section, however, this prediction does not hold.
Perhaps a more direct prediction would be that an increase
in predation risk would lead to shorter cycles of REM sleep
and thus, more arousals to wakefulness (see also Voss
Comparative Perspectives on Sleep
Allison & Van Twyver (1972) and Allison & Cicchetti
(1976) published early papers that defined some impor-
tant aspects of the architecture of mammalian sleep under
the risk of predation (also see Meddis 1983). These authors
sought insight into the evolution of sleep by relating key
structural features of sleep, such as total sleep time and
percentage of time in REM sleep, to constitutive traits of
animals such as brain and body mass. In addition to these
constitutive variables, Allison and colleagues also included
key features of the environment, including the degree to
which the mammals in question are vulnerable to pred-
ators and the determinants of relative exposure to pred-
ators. Their analyses indicated a strong role for the
predatory environment as a determinant of sleep archi-
tecture, a role that was roughly as important as factors
such as brain mass and body size. In general, mammals
living in relatively exposed environments had short sleep
times and a low density of REM sleep in both absolute and
relative measures. Predatory mammals were most likely to
show large amounts of REM sleep. These results apparent-
ly reflect the fact that arousal thresholds tend to be higher
in REM sleep than in other states, making REM sleep
a relatively dangerous form of sleep. Adding to the
problem with REM sleep is the fact that large animals like
ungulates must lay down during REM sleep (due to loss of
muscle tone), which might advertise their vulnerable state
to predators (Ruckebusch 1972). The predatory environ-
ment also appears to influence sleep patterns in birds
(Amlaner & Ball 1983). However, the relative lack of
detailed information on REM sleep and SWS in birds
limits inferences about the effects of predators on avian
sleep architecture.
None of the comparative studies conducted to date have
made any inferences about changes in the relative in-
tensity of SWS with increasing predation risk, because
such data are unavailable for many species. Presumably,
however, species with less REM sleep would also show
a less intense form of SWS as well (see below).
These comparative studies also demonstrate some im-
portant points about the states of sleep and REM sleep in
particular. First, the amount of REM sleep and SWS can
each vary greatly across the phylogenetic spectrum of
mammals, and thus, there are no fixed requirements for
either state (at least in an evolutionary sense). Other
comparative studies of sleep architecture (Zepelin &
Rechtschaffen 1974; Elgar et al. 1988) also reported much
interspecific flexibility in the distribution of SWS and REM
sleep. Second, given the apparent dangers associated with
REM sleep, it appears that mammals require at least some
REM sleep (but see Lyamin et al. 2000 on the possible lack
of REM sleep in cetaceans). Lastly, these comparative
studies suggest that the dangers imposed by certain states
of sleep are largely unavoidable constraints on vertebrate
sleep. That is, natural selection has not ‘found a way’ to
circumvent these constraints except through altering the
time spent in certain dangerous states like REM sleep.
An important matter regarding quantitative compara-
tive work on sleep is the lack of analyses that explicitly
take into account the phylogenetic relationships among
animals (Harvey & Pagel 1991). Virtually all existing
comparative work on sleep simply takes each species as
statistically independent of all others, but this approach is
largely unacceptable given the current understanding of
such analyses (Martins 2000). Only Elgar et al. (1988)
provide any sort of phylogenetically based comparative
study on mammalian sleep. We do not wish to imply that
existing comparative studies are necessarily wrong or
misleading, but it is possible that some of the inferences
derived from such analyses will not hold up under the
scrutiny of modern techniques.
Plasticity in Sleep under the Risk of Predation
Distribution of sleep states
The literature on behavioural plasticity under the risk of
predation is replete with examples of graded behavioural
responses to changes in the perceived risk of predation
(Sih 1987; Lima & Dill 1990). Furthermore, many studies
show that the behaviour of feeding animals represents
a trade-off (or adaptive compromise) between the benefit
of energy intake and the obvious fitness costs of an early
death (Lima 1998). Analogous trade-offs also apply to
mating and other forms of reproductive behaviour (Sih
1994). So far, however, this ‘trade-off’ concept has largely
bypassed the study of sleep.
There are many ways in which sleep responses to
changes in the perceived risk of predation might be
adaptive. Under an increase in the perceived risk of
predation, we might expect less time spent sleeping and
a disproportionate decrease in REM sleep. We might also
expect more frequent arousals, less time in deep SWS,
more time in unihemispheric sleep (if possible), and
perhaps a reallocation of sleep to different times of day.
However, without a clear understanding of the function of
sleep, we cannot be very specific about either the trade-
offs involved or the predicted changes in sleep behaviour
(such trade-offs are much more straightforward in the
study of feeding animals; Lima & Dill 1990). We could
base quantitative predictions on empirical demonstrations
of lowered performance as a result of differing degrees of
sleep deprivation (which might occur given any of the
above responses to increased risk), but such studies are not
easily done with animals. At this point, we can say that
sleep seems to serve a vital function, and that there are
negative consequences of losing sleep short of some sort
of ‘target’ (Rattenborg et al. 2004). However, we would not
expect most aspects of sleep to be absolutely ‘protected’
any more than energy intake is absolutely protected in
feeding–predation trade-offs.
As mentioned earlier, this sort of predation-risk ‘trade-
off’ perspective is rare in the study of plasticity in sleep
architecture. Lendrem (1984) provides a unique exception
to this rule. He studied changes in the sleep of doves,
Streptopelia risoria, in response to the brief, controlled
appearance of a predator (ferret, Mustela putorius). Follow-
ing their brief encounter with the predator, the doves
showed a marked increase in the rate at which sleep was
interrupted by ‘peeking’ (eyes open) to scan the local
environment. These doves also spent less overall time
sleeping after the encounter with the predator. Further-
more, doves in the safer setting of a group spent more time
sleeping with lower rates of peeking while asleep. Lendrem
(1983) also found that mallards, Anas platyrhynchos, sleep-
ing in relatively vulnerable places showed increased rates of
sleep interruptions (peeking), and that mallards in larger
groups showed longer bouts of sleep. Similar effects have
been observed in other ducks (Gauthier-Clerc et al. 1998,
2000, 2002) and a shorebird (Dominguez 2003).
Studies demonstrating plasticity in sleep architecture
are most valuable when they involve the manipulation of
predation risk in conjunction with EEG recordings of sleep
states. We know of only a few such studies. Sanford et al.
(2001) examined the distribution of sleep states in rats
after electric shock conditioning; although not predation
per se, electric shocks probably mimic many aspects of it.
The main effect on sleep was a substantial reduction in the
prevalence of REM sleep for several hours (also see Sanford
et al. 2003). The functional consequences (if any) of this
long-term reduction in REM sleep were not studied. One
effect of severe REM sleep deprivation may be a general
decrease in fear (Hicks & Moore 1979; Martı
et al. 2004), but this degree of deprivation is probably not
normally observed in free-living animals.
A study of unihemispheric sleep in mallards (Rattenborg
et al. 1999a, b) demonstrated a different sort of plasticity
in sleep architecture under variable risk. Ducks, like most
birds, sleep either unihemispherically or bihemispheri-
cally (Amlaner & Ball 1994). When sleeping in a relatively
risky situation (on the edge of a group versus within the
group’s interior), mallards engage in less overall sleep and
show a large relative increase in the prevalence of USWS
(Fig. 4;Rattenborg et al. 1999a, b), although bihemi-
spheric sleep is still the dominant form of sleep. This
suggests that bihemispheric sleep is more valuable than
USWS, but that the latter is used to aid in predator
detection under an increased threat. During USWS, the
open eye (contralateral to the awake hemisphere) is
readily able to detect incoming threats (Rattenborg et al.
1999a, b). The REM state is a minimal component of
overall sleep in mallards and is not significantly affected
by an individual’s position in the group. As with the study
by Sanford et al. (2001), the functional consequences of
a decrease in the riskier (i.e. higher-quality) form of sleep
were not determined.
Although there are few studies that both manipulate
risk and record EEGs, there is a fairly large literature on
sleep states under various forms of stress (in both humans
and rats; Van Reeth et al. 2000) that might shed light on
predation-related trade-offs (also see Voss 2004). Human
patients suffering from high levels of stress or anxiety
show sleep patterns that are much diminished in deep
SWS and REM sleep (Fuller et al. 1997; Voss 2004). An
increase in stress hormones in rats causes a fragmentation
of sleep and a decrease in the deeper stages of SWS
(Dugovic et al. 1999). These sorts of studies usually do
not deal with life-threatening stimuli such as predators
(but see Lavie et al. 1991), but the general phenomenon of
stress-induced changes in sleep almost certainly forms the
basis for at least some predator-induced changes in sleep
Timing of sleep
The timing or ‘phasing’ of sleep (that is, the distribution
of sleep over the daily cycle) should also be influenced by
trade-offs between the quality of sleep and the risk of
predation (also see Ball 1992), but there are no studies on
the topic. As a possible example of this phenomenon,
Fenn & Macdonald (1995) showed a dramatic shift in the
activity patterns of free-living rats from nocturnal to
diurnal in response to changes in activity of red foxes,
Vulpes vulpes (who were, in turn, responding to changes in
the activity patterns of humans). This shift in activity
presumably also resulted in a switch to a more nocturnal
sleep pattern in the rats. Similar sorts of shifts in sleep
have been demonstrated in laboratory rats in response to
changes in feeding schedules (see Tobler 1989).
We agree with Tobler’s (1989) suggestion that animals
with polyphasic sleep patterns and an ability to feed at
any time probably have the greatest flexibility in sleep
phasing (e.g. the rats in Fenn & Macdonald 1995). We also
suggest that prey facing generalist predators (such as
foxes) are most likely to show major changes in the
temporal patterning of sleep with changing risk. Specifi-
cally, animals that are mainly incidental prey of generalist
predators are most likely to benefit from a major change in
sleep times. A predator that specializes on a given prey
species will most likely match any change in the prey’s
activity pattern (also see Kotler et al. 2002); hence, major
changes in prey sleep phasing in specialized predator–prey
systems are unlikely.
Where to sleep
An animal’s choice of a sleep site is probably one of the
most important determinants of predation risk experi-
enced during the sleep period, and thus sets the stage for
many of the issues discussed so far. As might be expected
from the preceding discussion, however, there are few
studies on plasticity in the choice of sleep site. There is
virtually no experimental work, or even systematic obser-
vational work, on the topic. There currently exist only
anecdotal observations of sleep-site selection in dangerous
situations (e.g. Skutch 1989). Such observations from
primates, however, suggest that safety is paramount in
the choice of where to sleep (Anderson 1998). Ramak-
rishnan & Coss (2001) provide a particularly detailed
treatment of the topic in a macaque.
We naturally would expect animals to choose safe sites
for sleeping, but the picture is undoubtedly more compli-
cated. The choice of sleep site will depend on (among
other things) the safety of available sites, choices made by
others (degree of crowding, attraction of predators to
groups, etc.), proximity to profitable feeding areas, as well
as thermodynamic considerations (Bakken 1992). ‘Shell
game’ considerations, in which animals frequently shift
sleeping sites to avoid predators (Mitchell & Lima 2002;
also see Day & Elwood 1999), may also apply. The
expected quality of sleep in a given location could itself
be an important consideration in choice of a site.
It is likely that most animals routinely sleep in relatively
secure, thermodynamically favourable places. However,
the safest sites will not necessarily allow for deep sleep.
Under some circumstances, the safest available sites might
be exposed spots where predators have trouble approach-
ing unseen (e.g. starlings roosting on open power lines).
Thus, the safety of such places may be reflected in a bias
towards the more shallow sleep states. Animals choosing
to sleep in such situations will tend to form groups for
safety (Krause & Ruxton 2002). Such exposed spots are
often not the most thermodynamically favourable sites,
and this fact may dictate changes in sleep irrespective of
antipredator considerations. Since endotherms cannot
thermoregulate well in REM sleep (Heller et al. 1983;
Parmeggiani 2003), diminished REM sleep is expected
under thermodynamically stressful conditions. Thus, ther-
modynamic and antipredator considerations may interact
Centre Periphery
% Time
Bihemispheric sleep
Unihemispheric sleep
Position in
Figure 4. Percentage of time spent in bihemispheric or unihemi-
spheric sleep by mallard ducks (as determined by electroencephalo-
gram power analyses, see Rattenborg et al. 1999a, b). Ducks slept in
linear groups of four birds such that two were in the centre and two
were on the periphery of the group. Data are averaged over 16
ducks. Time not spent in one of these two states of sleep was spent
to determine choice of sleep site and its effect on sleep
States of Vigilance while Sleeping: Some
Conceptual Issues
The various states or stages of sleep (and wakefulness)
are referred to as ‘states of vigilance’ by sleep scientists.
Compared to the literature on antipredator vigilance
(Elgar 1989; Bednekoff & Lima 1998), the sleep literature
uses the term ‘vigilance’ more in reference to awareness of
the environment in general than attention directed
towards predators, but there is an obvious connection
between the two usages. As implied above, deep SWS and
REM sleep are low states of vigilance (high arousal thresh-
olds; Dillon & Webb 1965; Van Twyver 1969; Neckelmann
& Ursin 1993), whereas drowsiness might be considered
the highest state of vigilance while still sleeping to some
extent (Tobler 2000). Similarly, in birds, bihemispheric
sleep would be considered a lower state of vigilance
relative to unihemispheric sleep. Quiet wakefulness (deep
rest) is also considered a high state of vigilance, and active
wakefulness is considered the most vigilant state.
Here we address two related questions: (1) why are there
different states of vigilance/sleep? and (2) why not always
sleep at maximum intensity? Regarding the latter ques-
tion, it seems clear that some deep (intense) sleep is
essential in the long term, and it is in many ways
a superior form of sleep. Furthermore, following signifi-
cant sleep deprivation, the sleep debt is often mitigated by
a substantial increase in deep SWS (Horne 1988; Tobler
2000). It therefore seems possible that all or most of the
functions of SWS, at least, could be accomplished during
a relatively short but intense period of deep sleep.
However, a considerable amount of time is actually spent
in the shallower states of SWS (see Fig. 2), at least in
humans, and deep SWS may represent only a relatively
minor portion of overall sleep (Horne 1988; Borbe
´ly &
Achermann 2000).
There are at least three reasons why an animal might not
sleep at the maximum intensity. First, perhaps the less
intense forms of sleep are states through which an animal
must pass when progressing through the sleep cycle. For
instance, the less intense forms of sleep may unavoidably
occur when passing from deeper SWS to the more neuro-
logically active REM sleep and vice versa. In humans, there
is a tendency to pass into REM sleep from stage 2 non-REM
sleep (Fig. 2;Borbe
´ly & Achermann 2000), but the pre-
dominance of stage 2 sleep seems much greater than one
would expect if it were simply a transitional sleep state.
Second, perhaps different aspects of the function of sleep
are accomplished in the different stages such that all stages
are necessary to realize the full benefits of sleep.
A third possible reason concerns the antipredatory
consequences of sleep outlined so far: the continuum of
sleep/vigilance states may allow sleep to be tailored to the
prevailing risk of predation and current sleep debt such
that survival (fitness), rather than the benefits of sleep per
se, is maximized. In nature, bouts of high risk are episodic
(as per Lima & Bednekoff 1999), and the use of the
vigilant states of sleep during such periods would allow
sleep to proceed to some extent (as long as some or most
of the functions of sleep can be served by the relatively
high-vigilance states of sleep). Furthermore, if the sleep
debt is minimal, then extensive reliance on deep SWS may
not be necessary.
There is considerable evidence for adaptive changes in
sleep states in response to changes in the risk of predation.
We have already discussed several ways in which changes
in the perceived risk of predation (or perceived threats in
general) influence the distribution of sleep states (espe-
cially REM sleep). Observations on the effects of stress/fear
on human sleep are also consistent with the idea that
sleep states can be tailored to prevailing perceptions of risk
(also see Voss 2004).
Further support for the ‘tailoring-of-sleep’ hypothesis
comes from empirical observations on the distribution of
sleep states (or the intensity of sleep) across the sleep
period. Distribution of sleep states across time has not
been studied in many animals, so we focus our attention
on humans and rats. During a long bout of sleep, there is
typically a strong bias towards deep sleep early in the sleep
period, followed by less deep (more vigilant) states and an
increase in REM sleep as the end of the bout approaches
(Fig. 2; also see Horne 1988; Borbe
´ly & Achermann 2000).
This general temporal pattern in changing sleep (vigi-
lance) states is analogous to changes in antipredator
vigilance in feeding animals as their energetic states
change (McNamara & Houston 1986). As we argue below,
the similarity in these two vigilance-related phenomena
suggests that deep sleep is high-quality sleep, and that
a major function of sleep is restorative in nature (as per
Rechtschaffen 1998; Siegel 2003).
Our reasoning here is based directly on dynamic models
of antipredator vigilance (e.g. McNamara & Houston
1986; Clark & Mangel 2000). In these models, energetic
state and the passage of time combine to determine
antipredator vigilance. These models assume that feeding
functions to redress the energetic deficit incurred during
the overnight fast. Feeding can be interrupted at any time
during the day for unknown lengths of time due to
encounters with predators or bad weather. The fastest
way to erase this energetic deficit is to feed at the maximal
rate, but this option precludes any antipredator vigilance
(vigilance and feeding are often mutually exclusive activ-
ities). These models predict that a hungry animal should
devote much effort to feeding and relatively little to
vigilance. As its energetic reserves increase, the animal
should devote increasing effort to vigilance at the expense
of feeding rate. In other words, as the energetic deficit is
eliminated, the animal takes fewer risks by engaging in
greater vigilance. Changes in antipredator vigilance allow
the behavioural sequence to be tailored to changing
energetic states such that overall fitness is maximized.
Different states of vigilance in sleeping animals may play
the role of antipredator vigilance in the above conceptual
scenario, with the sleep deficit as the analogue of the
energetic deficit. Assume further that unpredictable events
(e.g. predators, general disturbance) can disrupt sleep for
much of the night. Assume that sleep is essential for some
sort of restorative function, and that deep SWS can erase
this deficit faster than the more vigilant states of SWS. In
a dynamic model of sleep, an animal would accept the risk
of relatively intense SWS early in the sleep period when the
sleep deficit is greatest (e.g. Fig. 2); this ensures that enough
high-quality sleep takes place on most nights, even if
interruptions occur occasionally. A shift towards less-deep
sleep will occur as high-quality sleep accumulates. As
mentioned earlier, the observation that intense SWS occurs
early in the sleep period is consistent with the idea that
sleep serves a restorative function. The tailoring-of-sleep
hypothesis also seems similar to the ‘changing priorities’
idea presented by Voss (2004) to explain changes in sleep
states over time. This sort of dynamic model of sleep would
predict a greater degree of intense SWS following sleep
deprivation, which is commonly observed in mammals
(Tobler 2000). The corresponding phenomenon in birds of
less unihemispheric sleep and more bihemispheric sleep in
response to sleep deprivation has recently been observed
(Boerema et al. 2003).
The above analogy may help to explain the existence of
various states of SWS sleep, but REM sleep presents
a problem for which there is no real analogue in the
above feeding–vigilance trade-off. The late portion of the
sleep period is indeed dominated by less-intense SWS, but
there is also a relative increase in REM sleep as well (Fig. 2).
As mentioned earlier, arousal thresholds are higher in
REM sleep than for most or all intensities of SWS. Thus, an
overall theory for the distribution of sleep states over time
must address the issue of why REM sleep is most prevalent
late in the sleep period. Perhaps there is some antipredator
value to ‘packaging’ REM sleep with relatively vigilant
forms of SWS sleep (Voss 2004). The relatively late
appearance of REM sleep may also be explained by
assuming that deep SWS and REM sleep are antagonistic
states, and that the function of REM sleep is not as
important as that of SWS (at least on a short-term basis).
However, severe REM sleep deprivation may well cause
problems with memory consolidation (Walker & Stickgold
2004; but see Vertes 2004) and perhaps a shift in REM
sleep to the earlier portions of a sleep bout. Alternatively,
Voss (2004) suggests that REM sleep occurs late in an
undisturbed period of sleep because that is when (hu-
mans) are most secure in their sleeping arrangements
(reflecting the lack of disturbance).
Drowsiness as a state of vigilance
Drowsy animals are usually immobile, relativel y unrespon-
sive, and usually have their eyes at least partially opened.
Drowsiness is taxonomically widespread, and is known to
occur in many species of mammal s (Allison & Cicchetti 1976;
Meddis 1983)andbirds(Amlaner & Ball 1983). Many workers
have grappled with the question of whether to categorize this
state as a form of sleep or quiet wakefulness (e.g. Allison &
Cicchetti 1976; Meddis 1983; Campbell & Tobler 1984;
Tobler 1995; Zepelin 2000), and all have chosen not to treat
it as a form of sleep. However, drowsiness is in many ways
a particularly vigilant form of sleep (Makeig et al. 2000; Noser
et al. 2003), and might be an effective way of dealing with the
problem of predator detection.
If drowsiness has a significant antipredator function,
then we might intuitively expect more drowsiness in
a risky environment (but not when dealing with an acutely
risky situation with a predator present). More generally,
the degree of drowsiness should respond to changes in the
risk of predation. The degree of drowsiness might even
influence the distribution and density of various states of
sleep during ‘true’ sleep. However, these issues surround-
ing drowsiness will be unresolved until the sleep-related
benefits (if any) of drowsiness can be determined.
Alert and awake animals use several cues and sensory
modalities to gain information about the local risk of
predation (Lima & Steury, in press). Is a similar range of
information also available to sleeping animals? Are in-
formation-rich predator cues more likely to cause an
arousal from sleep than are other types of cues? Are
arousal thresholds in the various stages of sleep a function
of the information content of particular cues? We know
that a sleeping animal is monitoring the environment for
relevant stimuli to some extent (Velluti 1997; Coenen &
Drinkenburg 2002), and that the magnitude of stimuli
necessary to awaken an animal varies with different sleep
states. Beyond these simple facts, however, little is known
about the monitoring and assessment of potential preda-
tory threats by sleeping animals.
The degree to which information-rich cues might alter
arousal thresholds is a particularly important question for
our purposes. It seems likely that meaningful predatory
stimuli would have a markedly greater effect than would
other stimuli that may trigger arousals. It is also conceiv-
able that important cues may be linked to unexpectedly
low arousal thresholds in deep sleep. It has long been
known that rats do indeed show greater arousal thresholds
in deeper sleep even when the cue in question is a tone
associated with foot shocks (Van Twyver & Garrett 1972),
but the relevance of this result to their natural predator
cues is unclear. It does seem likely that arousal thresholds
are greater in deep sleep, but arousal thresholds for key
predatory stimuli are likely to be lower than expected
based on current available work. There are almost certainly
specific neural structures devoted to monitoring the
predatory environment during sleep (Lang et al. 2000,
¨hman & Mineka 2001; also see Lima & Steury, in press),
making these lower thresholds a real possibility.
Of course, arousal from sleep reflects a process of not only
information detection but also a degree of information
processing. Thus, a lack of arousal from sleep does not
necessarily reflect a lack of information available to the
sleeping animal; some sensory information can undoubt-
edly be received without overt arousals. Separating these
two processes is not easy, but techniques designed to detect
an evoked or event-related potential (ERP) may yield some
insights. These techniques allow one to record neural
responses to stimuli that are detected but do not necessarily
lead to arousals or wakefulness (see Bastuji & Garcia-Larrea
1999 for a review). They can be used to show that the
sleeping human brain can (during shallow SWS and REM
sleep) differentiate the subject’s own name from general
speech (Bastuji et al. 2002). Analyses of ERPs have shown
that the sleeping avian brain can also detect and ‘acknowl-
edge’ song playbacks (Nick & Konishi 2001). During REM
sleep, humans can detect odd items in a string of auditory
stimuli without awaking (Nashida et al. 2000; Cote 2002;
Perrin et al. 2002). In some respects, ERPs during sleep can
be similar to those evoked by the same stimuli during
wakefulness (but this is often not the case; Cote 2002).
Humans and rats show ERPs that are much diminished
or absent during deep SWS (Bastuji & Garcia-Larrea 1999;
Nashida et al. 2000; Cote et al. 2001). This result suggests
that during deep SWS (1) information detection or pro-
cessing (or both) is greatly limited, and (2) interactions
between the vertebrate sleeping brain and the external
environment are stemmed early in the detection and
processing of information (Voss 2004). Such limitations
may well apply to the detection and processing of predator
cues, but this remains to be shown. It would be valuable to
apply this ERP technique and more recent imaging techni-
ques (Portas et al. 2000) to the detection and processing of
predatory stimuli in a diverse array of animals.
The work on ERPs illustrates an apparent paradox
regarding the detection of threatening stimuli while in
REM sleep. As noted earlier, EEG patterns during REM sleep
are more like those in awake subjects than those in SWS.
Thus, it may not be surprising that many sorts of stimuli
evoke neural responses during REM sleep (in addition to
cardiac responses in rats; Maho & Hennevin 1999). This
apparent neural interaction with the external environ-
ment, along with the fact that animals aroused from REM
sleep are more alert and ready to deal with threats (Horner
et al. 1997), is suggestive of the sentinel hypothesis (see
above) for the function of REM sleep. The only problem for
this hypothesis is the fact that arousal thresholds are high
during REM sleep. Portas et al. (2000) suggested that these
relatively high arousal thresholds reflect a deactivation of
the higher-level processing of information during REM
sleep. Future work should establish whether this ‘REM
paradox’ really applies to the detection of predatory
stimuli under ecologically relevant conditions.
Sleep (broadly defined) appears widespread in the animal
world. The functions of sleep remain elusive but one fact is
clear: sleep renders an animal more vulnerable to predatory
attack than just about any other behaviour. However,
sleeping animals are not helplessly turned-off, and certain
states of sleep and ways to sleep are safer than others. We
are certain that a research programme focused on pre-
dation-related consequences of various sleep patterns will
yield new insights into the nature of sleep in general.
Relatively little work has been conducted on sleeping
under the risk of predation, despite several early studies
focused on this topic. All of the areas that we covered need
a great deal more attention, and many areas are virtually
unstudied. Regarding comparative work, the classic stud-
ies of Allison & Cicchetti (1976), Meddis (1983) and others
need to be updated using modern phylogenetically based
statistical techniques. These studies also need to be
expanded where possible to cover nonmammalian taxa.
Comparative work on sleep in ectothermic vertebrates and
invertebrates would be particularly valuable. Indeed, there
is a great need to simply characterize sleep in a wide
variety of invertebrates. From ecological and evolutionary
perspectives, it is essential to expand the current view of
sleep beyond the mammalian paradigm.
We suspect that the most rapid advances will be made in
the study of the flexibility in sleep architecture. Experimen-
tal work focusing on the dynamic changes in sleep states or
intensity over the sleep period and under various levels of
predation risk should prove valuable. Thresholds of arousal
from various states of sleep must also be characterized
relative to important predator-based stimuli. Flexibility in
the phasing of sleep over the circadian cycle and the choice
of sleep sites under the risk of predation should prove
relatively easy to study. We also need to address the issue of
the degree to which drowsiness is a state of vigilant sleep.
The trade-off approach to studying flexibility that we
advocate will be greatly enhanced by an increased un-
derstanding of the link between the quality of sleep and
daytime performance. Collectively, this work on sleep
architecture will illustrate the dynamic environmental
challenges under which sleep evolved, and extend the
striking patterns in sleep seen in the early comparative work.
In summary, we advocate an ecologically and evolu-
tionarily based view of sleep and its consequences. In
doing so, we echo the suggestion of Horne (1988), Tobler
(1989) and others that more work is needed on sleep in
ecologically realistic settings. The vast majority of work on
sleep behaviour is done in highly simplified laboratory
environments that may not capture all of the salient
features of sleep. We have little doubt, however, that
many productive collaborations between behavioural
ecologists, sleep scientists and neurophysiologists will be
forged in studying sleep under the risk of predation.
Neurobiological information and insights will be key to
the success of this new paradigm. This is particularly
apparent when determining the sorts of information
available to sleeping animals and the mechanisms with
which the brain processes information about risk while
sleeping. Perhaps such collaborations will ultimately help
forge an understanding of the functions of sleep itself.
Mark Opp kindly provided the EEG information in Fig. 1.
Support for this work was provided by the Department of
Ecology & Organismal Biology and the School of Graduate
Studies at Indiana State University.
Allison, T. & Cicchetti, D. V. 1976. Sleep in mammals: ecological
and constitutional correlates. Science,194, 732–734.
Allison, T. & Van Twyver, H. 1972. The evolution of sleep. Natural
History,79, 56–65.
Amlaner, C. J. & Ball, N. J. 1983. A synthesis of sleep in wild birds.
Behaviour,87, 85–119.
Amlaner, C. J. & Ball, N. J. 1994. Avian sleep. In: Principles and Practice
of Sleep Medicine. 2nd edn (Ed. by M. H. Kryger, T. Roth & W. C.
Dement), pp. 81–94. Philadelphia: W. B. Saunders.
Amlaner, C. J. & McFarland, D. J. 1981. Sleep in the herring gull
(Larus argentatus). Animal Behaviour,29, 551–556.
Anderson, J. R. 1998. Sleep, sleeping sites, and sleep-related
activities: awakening to their significance. American Journal of
Primatology,46, 63–75.
Bakken, G. S. 1992. Measurement and application of operative and
standard operative temperatures in biology. American Zoologist,
32, 194–216.
Ball, N. J. 1992. The phasing of sleep in animals. In: Why We Nap
(Ed. by C. Stampi), pp. 31–49. Boston: Birkha¨user.
Bastuji, H. & Garcia-Larrea, L. 1999. Evoked potentials as a tool
for the investigation of human sleep. Sleep Medicine Reviews,3,
Bastuji, H., Perrin, F. & Garcia-Larrea, L. 2002. Semantic analysis
of auditory input during sleep: studies with event
related potentials. International Journal of Psychophysiology,46,
Bednekoff, P. A. & Lima, S. L. 1998. Randomness, chaos, and
confusion in the study of anti-predator vigilance. Trends in Ecology
and Evolution,13, 284–287.
Berger, R. J. & Phillips, N. H. 1995. Energy conservation and sleep.
Behavioural Brain Research,69, 65–73.
Boerema, A. S., Riedstra, B. & Strijkstra, A. M. 2003. Decrease in
monocular sleep after sleep deprivation in the domestic chicken.
Behaviour,140, 1415–1420.
´ly, A. A. & Achermann, P. 2000. Sleep homeostasis and
models of sleep regulation. In: Principles and Practice of Sleep
Medicine. 3rd edn (Ed. by M. H. Kryger, T. Roth & W. C. Dement),
pp. 377–390. Philadelphia: W. B. Saunders.
Campbell, S. S. & Tobler, I. 1984. Animal sleep: a review of sleep
duration across phylogeny. Neuroscience and Biobehavioral Re-
views,8, 269–300.
Cirelli, C., Gutierrez, C. M. & Tononi, G. 2004. Extensive and
divergent effects of sleep and wakefulness on brain gene
expression. Neuron,41, 35–43.
Clark, C. W. & Mangel, M. 2000. Dynamic State Variable Models in
Ecology. Oxford: Oxford University Press.
Coenen, A. M. & Drinkenburg, W. H. 2002. Animal models for
information processing during sleep. International Journal of
Psychophysiology,46, 163–175.
Cote, K. A. 2002. Probing awareness during sleep with the auditory
odd-ball paradigm. International Journal of Psychophysiology,46,
Cote, K. A., Etienne, L. & Campbell, K. B. 2001. Neurophysiolog-
ical evidence for the detection of external stimuli during sleep.
Sleep,24, 791–803.
Day, R. T. & Elwood, R. W. 1999. Sleeping site selection by
the golden-handed tamarin Saguinus midas midas: the role of
predation risk, proximity to feeding sites, and territorial defense.
Ethology,105, 1035–1051.
Dillon, R. F. & Webb, W. B. 1965. Threshold of arousal from
activated sleep in the rat. Journal of Comparative and Physiological
Psychology,59, 447–449.
Dominguez, J. 2003. Sleeping and vigilance in black-tailed godwit.
Journal of Ethology,21, 57–60.
Dugovic, C., Maccari, S., Weibel, L., Turek, F. W. & Van Reeth, O.
1999. High corticosterone levels in prenatally stressed rats predict
persistent paradoxical sleep alterations. Journal of Neuroscience,
19, 8656–8664.
Elgar, M. A. 1989. Predator vigilance and group size in mammals
and birds: a critical review of the empirical evidence. Biological
Reviews,64, 13–33.
Elgar, M. A., Pagel, M. D. & Harvey, P. H. 1988. Sleep in mammals.
Animal Behaviour,36, 1407–1419.
Fenn, M. G. P. & Macdonald, D. W. 1995. Use of middens by red
foxes: risk reverses rhythms of rats. Journal of Mammalogy,76,
Flanigan, W. F. 1972. Behavioral states and electroencephalograms
of reptiles. In: The Sleeping Brain: Perspectives in the Brain Sciences
(Ed. by M. H. Chase), pp. 14–18. Los Angeles: Brain Information
Service/Brain Research Institute, University of California Los Angeles.
Flanigan, W. F., Wilcox, R. H. & Rechtschaffen, A. 1973. The EEG
and behavioral continuum of the crocodilian, Caiman sclerops.
Electroencephalography and Clinical Neurophysiology,34, 521–538.
Flanigan, W. F., Knight, C. P., Hartse, K. M. & Rechtschaffen, A.
1974. Sleep and wakefulness in chelonian reptiles. I. The box turtle,
Terrapene carolina.Archives Italiennes de Biologie,112, 227–252.
Fuller, K. H., Waters, W. F., Binks, P. G. & Anderson, T. 1997.
Generalized anxiety and sleep architecture: a polysomnographic
investigation. Sleep,20, 370–376.
Gauthier-Clerc, M., Tamisier, A. & Ce
´zilly, F. 1998. Sleep-vigilance
trade-off in green-winged teals (Anas crecca crecca). Canadian
Journal of Zoology,76, 2214–2218.
Gauthier-Clerc, M., Tamisier, A. & Ce
´zilly, F. 2000. Sleep-vigilance
trade-off in gadwall during the winter period. Condor,102, 307–313.
Gauthier-Clerc, M., Tamisier, A. & Ce
´zilly, F. 2002. Vigilance while
sleeping in the breeding pochard Aythya ferina according to sex
and age. Bird Study,49, 300–303.
Hartmann, E. L. 1973. The Functions of Sleep. New Haven,
Connecticut: Yale University Press.
Hartse, K. M. 1994. Sleep in insects and nonmammalian verte-
brates. In: Principles and Practice of Sleep Medicine. 2nd edn (Ed. by
M. H. Kryger, T. Roth & W. C. Dement), pp. 95–104. Philadelphia:
W. B. Saunders.
Harvey, P. H. & Pagel, M. D. 1991. The Comparative Method in
Evolutionary Biology. Oxford: Oxford University Press.
Hediger, H. 1969. Comparative observations on sleep. Proceedings
of the Royal Society of Medicine,62, 153–156.
Heller, H. C., Graf, R. & Rautenberg, W. 1983. Circadian and
arousal state influences on thermoregulation in the pigeon.
American Journal of Physiology,245, R321–R328.
Hicks, R. A. & Moore, J. D. 1979. REM sleep deprivation diminishes
fear in rats. Physiology & Behavior,22, 689–692.
Horne, J. A. 1988. Why We Sleep: the Functions of Sleep in Humans
and Other Animals. Oxford: Oxford University Press.
Horner, R. L., Sanford, L. D., Pack, A. I. & Morrison, A. R. 1997.
Activation of a distinct arousal state immediately after spontane-
ous awakening from sleep. Brain Research,778, 127–134.
Huber, R., Ghilardi, M. F., Massimini, M. & Tononi, G. 2004. Local
sleep and learning. Nature,430, 78–81.
Kaiser, W. & Steiner-Kaiser, J. 1983. Neuronal correlates of sleep,
wakefulness and arousal in a diurnal insect. Nature,301, 707–709.
Kohsaka, M., Fukuda, N., Honma, K., Honma, S. & Morita, N.
1992. Seasonality in human sleep. Experientia,48, 231–233.
Kotler, B. P., Brown, J. S., Dall, S. R. X., Gresser, S., Ganey, D. &
Bouskila, A. 2002. Foraging games between gerbils and their
predators: temporal dynamics of resource depletion and appre-
hension in gerbils. Evolutionary Ecology Research,4, 495–518.
Krause, J. & Ruxton, G. D. 2002. Living in Groups. Oxford: Oxford
University Press.
Krueger, J. M. & Oba
´l, F., Jr. 1993. A neuronal group theory of sleep
function. Journal of Sleep Research,2, 63–69.
Krueger, J. M. & Oba
´l, F., Jr. 2002. Function of sleep. In: Sleep
Medicine (Ed. by T. L. Lee-Chiong, M. J. Sateia & M. A. Carskadon),
pp. 23–30. Philadelphia: Hanley & Belfus.
Lang, P. J., Davis, M. & O
¨hman, A. 2000. Fear and anxiety: animal
models and human cognitive psychophysiology. Journal of
Affective Disorders,61, 137–159.
Lavie, P., Carmeli, A., Mevorach, L. & Liberman, N. 1991.
Sleeping under the threat of the SCUD: war-related environ-
mental insomnia. Israel Journal of Medical Sciences,27,
Lendrem, D. W. 1983. Sleeping and vigilance in birds. I. Field
observations of the mallard (Anas platyrhynchos). Animal Behav-
iour,31, 532–538.
Lendrem, D. W. 1984. Sleeping and vigilance in birds. II. An
experimental study of the barbary dove (Streptopelia risoria).
Animal Behaviour,32, 243–248.
Lesku, J. A., Rattenborg, N. C. & Amlaner, C. J. In press. The
evolution of sleep: a phylogenetic approach. In: Encyclopedia of
Sleep Medicine (Ed. by T. Lee-Chiong). Hoboken, New Jersey:
J. Wiley.
Lima, S. L. 1998. Non-lethal effects in the ecology of predator–prey
interactions. BioScience,48, 25–34.
Lima, S. L. 2002. Putting predators back into behavioral predator–
prey interactions. Trends in Ecology and Evolution,17, 70–75.
Lima, S. L. & Bednekoff, P. A. 1999. Temporal variation in danger
drives antipredator behavior: the predation risk allocation hypoth-
esis. American Naturalist,153, 649–659.
Lima, S. L. & Dill, L. M. 1990. Behavioural decisions made under the
risk of predation: a review and prospectus. Canadian Journal of
Zoology,68, 619–640.
Lima, S. L. & Steury, T. D. In press. Perception of risk: the
foundation of non-lethal predator–prey interactions. In: The Ecology
of Predator–Prey Interactions (Ed. by P. Barbosa & I. Castellanos).
Oxford: Oxford University Press.
Lyamin, O. I., Manger, P. R., Mukhametov, L. M., Siegel, J. M. &
Shpak, O. V. 2000. Rest and activity states in a gray whale. Journal
of Sleep Research,9, 261–267.
McNamara, J. M. & Houston, A. I. 1986. The common currency for
behavioral decisions. American Naturalist,127, 358–378.
Maho, C. & Hennevin, E. 1999. Expression in paradoxical sleep of
a conditioned heart rate response. Neuroreport,10, 3381–3385.
Mahowald, M. W. & Schenck, C. H. 1992. Dissociated states of
wakefulness and sleep. Neurology,42, 44–51.
Makeig, S., Jung, T. P. & Sejnowski, T. J. 2000. Awareness during
drowsiness: dynamics and electrophysiological correlates. Canadian
Journal of Psychology,54, 266–273.
´lez, D., Obermeyer, W., Fahy, J. L., Riboh, M.,
Kalin, N. H. & Benca, R. M. 2004. REM sleep deprivation induces
changes in coping responses that are not reversed by amphet-
amine. Sleep,27, 609–617.
Martins, E. P. 2000. Adaptation and the comparative method.
Trends in Ecology and Evolution,15, 296–299.
Meddis, R. 1975. On the function of sleep. Animal Behaviour,23,
Meddis, R. 1977. The Sleep Instinct. London: Routledge & Kegan Paul.
Meddis, R. 1983. The evolution of sleep. In: Sleep Mechanisms and
Functions (Ed. by A. Mayes), pp. 57–95. Berkshire: Van Nostrand
Mitchell, W. A. & Lima, S. L. 2002. Predator–prey shell games: large
scale movement and its implications for decision-making by prey.
Oikos,99, 249–259.
Mitchell, W. A. & Valone, T. J. 1990. The optimization research
program: studying adaptations by their function. Quarterly Review
of Biology,65, 43–52.
Moorcroft, W. H. 1995. The function of sleep: comments on the
symposium and an attempt at synthesis. Behavioural Brain
Research,69, 207–210.
Nashida,T., Yabe, H., Sato, Y.,Hiruma, T., Sutoh, T.,Shinozaki, N. &
Kaneko, S. 2000. Automatic auditory information processing in
sleep. Sleep,23, 821–828.
Neckelmann, D. & Ursin, R. 1993. Sleep stages and EEG power
spectrum in relation to acoustical stimulus arousal threshold in the
rat. Sleep,16, 467–477.
Nick, T. A. & Konishi, M. 2001. Dynamic control of auditory activity
during sleep: correlation between song response and EEG.
Proceedings of the National Academy of Sciences, U.S.A.,98,
Nitz, D. A., van Swinderen, B., Tononi, G. & Greenspan, R. J.
2002. Electrophysiological correlates of rest and activity in
Drosophila melanogaster.Current Biology,12, 1934–1940.
Noser, R., Gygax, L. & Tobler, I. 2003. Sleep and social status in
captive gelada baboons (Theropithecus gelada). Behavioural Brain
Research,147, 9–15.
¨hman, A. & Mineka, S. 2001. Fears, phobias, and preparedness:
toward an evolved module of fear and fear learning. Psychological
Review,108, 483–522.
Parmeggiani, P. L. 2003. Thermoregulation and sleep. Frontiers in
BioScience,8, S557–S567.
Perrin, F., Bastuji, H. & Garcia-Larrea, L. 2002. Detection of verbal
discordances during sleep. Neuroreport,13, 1345–1349.
´ron, H. 1913. Le Proble
`me Physiologique du Sommeil. Paris: Masson.
Portas, C. M., Krakow, K., Allen, P., Josephs, O., Armony, J. L. &
Frith, C. D. 2000. Auditory processing across the sleep–wake
cycle: simultaneous EEG and fMRI monitoring in humans. Neuron,
28, 991–999.
Pravosudov, V. V. & Lucas, J. R. 2000. The costs of being cool:
a dynamic model of nocturnal hypothermia by small food-caching
birds in winter. Journal of Avian Biology,31, 463–472.
Ramakrishnan, U. & Coss, R. G. 2001. Strategies used by bonnet
macaques (Macaca radiata) to reduce predation risk while
sleeping. Primates,42, 193–206.
´n, F., Herna
´n, J., Nguyen, B. & Bullock, T. H.
2004. Slow wave sleep in crayfish. Proceedings of the National
Academy of Sciences, U.S.A.,101, 11857–11861.
Rattenborg, N. C. & Amlaner, C. J. 2002. Phylogeny of sleep. In:
Sleep Medicine (Ed. by T. L. Lee-Chiong, M. J. Sateia & M. A.
Carskadon), pp. 7–22. Philadelphia: Hanley & Belfus.
Rattenborg, N. C., Lima, S. L. & Amlaner, C. J. 1999a. Half-awake
to the risk of predation. Nature,397, 397–398.
Rattenborg, N. C., Lima, S. L. & Amlaner, C. J. 1999b. Facultative
control of avian unihemispheric sleep under the risk of predation.
Behavioural Brain Research,105, 163–172.
Rattenborg, N. C., Amlaner, C. J. & Lima, S. L. 2000.
Behavioral, neurophysiological and evolutionary perspectives on
unihemispheric sleep. Neuroscience and Biobehavioral Reviews,24,
Rattenborg, N. C., Mandt, B. H., Obermeyer, W. H., Winsauer,
P. J., Huber, R., Wikelski, M. & Benca, R. M. 2004. Migratory
sleeplessness in the white-crowned sparrow (Zonotrichia leu-
cophrys gambelii). PLoS Biology,2, 924–936.
Rechtschaffen, A. 1998. Current perspectives on the function of
sleep. Perspectives in Biology and Medicine,41, 359–390.
Rechtschaffen, A. & Bergmann, B. M. 2002. Sleep deprivation in
the rat: an update of the 1989 paper. Sleep,25, 18–24.
Ruckebusch, Y. 1972. The relevance of drowsiness in the circadian
cycle of farm animals. Animal Behaviour,20, 637–643.
Sanford, L. D., Silvestri, A. J., Ross, R. J. & Morrison, A. R. 2001.
Influence of fear conditioning on elicited ponto-geniculo-occipital
waves and rapid eye movement sleep. Archives Italiennes de
Biologie,139, 169–183.
Sanford, L. D., Tang, X., Ross, R. J. & Morrison, A. R. 2003.
Influence of shock training and explicit fear-conditioned cues on
sleep architecture in mice: strain comparison. Behavior Genetics,
33, 43–58.
Sauer, S., Kinkelin, M., Herrmann, E. & Kaiser, W. 2003. The
dynamics of sleep-like behavior in honey bees. Journal of
Comparative Physiology A,189, 599–607.
Sauer, S., Herrmann, E. & Kaiser, W. 2004. Sleep deprivation in
honey bees. Journal of Sleep Research,13, 145–152.
Schuppe, H. 1995. Rhythmic brain activity in sleeping bees. Wiener
Medizinische Wochenschrift,145, 463–464.
Shaw, P. J., Cirelli, C., Greenspan, R. J. & Tononi, G. 2000.
Correlates of sleep and waking in Drosophila melanogaster.Science,
287, 1834–1837.
Shaw, P. J., Tononi, G., Greenspan, R. J. & Robinson, D. F. 2002.
Stress response genes protect against lethal effects of sleep
deprivation in Drosophila.Nature,417, 287–291.
Siegel, J. M. 2003. Why we sleep. Scientific American,(Nov) 289,
Sih, A. 1987. Predator and prey lifestyles: an evolutionary and
ecological overview. In: Predation: Direct and Indirect Impacts
on Aquatic Communities (Ed. by W. C. Kerfoot & A. Sih), pp. 203–
224. Hanover, New Hampshire: University of New England Press.
Sih, A. 1994. Predation risk and the evolutionary ecology of
reproductive behaviour. Journal of Fish Biology,45, 111–130.
Skutch, A. F. 1989. Birds Asleep. Austin: University of Texas Press.
Snyder, F. 1966. Towards an evolutionary theory of dreaming.
American Journal of Psychiatry,123, 121–136.
Szymczak, J. T., Kaiser, W., Helb, H. W. & Beszczynska, B. 1996. A
study of sleep in the European blackbird. Physiology & Behavior,
60, 1115–1120.
Tobler, I. 1985. Deprivation of sleep and rest in vertebrates and
invertebrates. In: Endogenous Sleep Substances and Sleep Regulation
(Ed. by S. Inoue
´& A. A. Borbe
´ly), pp. 57–66. Tokyo: Japan
Scientific Societies Press.
Tobler, I. 1989. Napping and polyphasic sleep in mammals. In:
Chronobiological, Behavioral, and Medical Aspects of Napping (Ed. by
D. F. Dinges & R. J. Broughton), pp. 9–30. New York: Raven Press.
Tobler, I. 1995. Is sleep fundamentally different between mamma-
lian species? Behavioural Brain Research,69, 35–41.
Tobler, I. 2000. Phylogeny of sleep regulation. In:Pr inciples and Pract ice
of Sleep Medicine. 3rd edn (Ed. by M. H. Kryger, T. Roth & W. C.
Dement), pp. 72–81. Philadelphia: W. B. Saunders.
Tolaas, J. 1978. REM sleep and the concept of vigilance. Biological
Psychiatry,13, 135–148.
Tononi, G. & Cirelli, C. 2003. Sleep and synaptic homeostasis:
a hypothesis. Brain Research Bulletin,62, 143–150.
Van Reeth, O., Weibel, L., Spiegel, K., Leproult, R., Dugovic, C. &
Maccari, S. 2000. Interactions between stress and sleep: from basic
research to clinical situations. Sleep Medicine Reviews,4, 201–219.
Van Twyver, H. 1969. Sleep patterns of five rodent species.
Physiology & Behavior,4, 901–905.
Van Twyver, H. & Garrett, W. 1972. Arousal threshold in the rat
determined by ‘‘meaningful’’ stimuli. Behavioural Biology,7, 205–
Velluti, R. A. 1997. Interactions between sleep and sensory
physiology. Journal of Sleep Research,6, 61–77.
Vertes, R. P. 2004. Memory consolidation in sleep: dream or reality?
Neuron,44, 135–148.
Voss, U. 2004. Functions of sleep architecture and the concept of
protective fields. Reviews in the Neurosciences,15, 33–46.
Walker, M. P. & Stickgold, R. 2004. Sleep-dependent learning and
memory consolidation. Neuron,44, 121–133.
Webb, W. B. 1975. Sleep: the Gentle Tyrant. Englewood Cliffs, New
Jersey: Prentice Hall.
Zepelin, H. 1970. Sleep of the jaguar and the tapir: a prey–predator
contrast. Psychophysiology,7, 305–306.
Zepelin, H. 1989. Mammalian sleep. In: Principles and Practice of
Sleep Medicine (Ed. by M. H. Kryger, T. Roth & W. C. Dement), pp.
30–49. Philadelphia: W. B. Saunders.
Zepelin, H. 2000. Mammalian sleep. In: Principles and Practice of
Sleep Medicine. 3rd edn (Ed. by M. H. Kryger, T. Roth & W. C.
Dement), pp. 82–92. Philadelphia: W. B. Saunders.
Zepelin, H. & Rechtschaffen, A. 1974. Mammalian sleep,
longevity and energy metabolism. Brain Behavior and Evolution,
10, 425–470.
... Alternatively, the risk of inactivity, especially sleep [67,68], is possibly equal to that of activity and thus may create opposing patterns that mask any relationship between foraging and predation that does exist. Although inactivity such as hibernation has been reported to increase prey survival potentially through reduced predation risk [69], sleep requires individuals to enter a state of reduced neural activity and suspension of consciousness that results in a lack of awareness of, and responsiveness to, the environment [68]. ...
... Alternatively, the risk of inactivity, especially sleep [67,68], is possibly equal to that of activity and thus may create opposing patterns that mask any relationship between foraging and predation that does exist. Although inactivity such as hibernation has been reported to increase prey survival potentially through reduced predation risk [69], sleep requires individuals to enter a state of reduced neural activity and suspension of consciousness that results in a lack of awareness of, and responsiveness to, the environment [68]. Some organisms minimize the dangers of sleep by relying on communal resting [70,71] or using safe places or refuges for sleep [72], but not all environments provide refuges that completely eliminate risks. ...
Full-text available
The assumption that activity and foraging are risky for prey underlies many predator-prey theories and has led to the use of predator-prey activity overlap as a proxy of predation risk. However, the simultaneous measures of prey and predator activity along with timing of predation required to test this assumption have not been available. Here, we used accelerometry data on snowshoe hares (Lepus americanus) and Canada lynx (Lynx canadensis) to determine activity patterns of prey and predators and match these to precise timing of predation. Surprisingly we found that lynx kills of hares were as likely to occur during the day when hares were inactive as at night when hares were active. We also found that activity rates of hares were not related to the chance of predation at daily and weekly scales, whereas lynx activity rates positively affected the diel pattern of lynx predation on hares and their weekly kill rates of hares. Our findings suggest that predator-prey diel activity overlap may not always be a good proxy of predation risk, and highlight a need for examining the link between predation and spatio-temporal behaviour of predator and prey to improve our understanding of how predator-prey behavioural interactions drive predation risk.
... If survival occurs, it is commonly assumed that the animal resumes its normal activities [9] and physiological equilibrium is restored. However, studies on commercial strains of rats and mice have shown that non-lethal, long-lasting effects of predators, such as fear, anxiety, or post-traumatic responses, can have a significant impact on individual morphology, behavior, and reproductive success [10][11][12][13][14]. A fundamental question is thus the extent to which experimental findings using a commercial strain of rats or mice can be generalized to real-world situations. ...
Full-text available
In the wild, animals face a highly variable world full of predators. Most predator attacks are unsuccessful, and the prey survives. According to the conventional perspective, the fear responses elicited by predators are acute and transient in nature. However, the long-term, non-lethal effects of predator exposure on prey behavioral stress sequelae, such as anxiety and post-traumatic symptoms, remain poorly understood. Most experiments on animal models of anxiety-related behavior or post-traumatic stress disorder have been carried out using commercial strains of rats and mice. A fundamental question is whether laboratory rodents appropriately express the behavioral responses of wild species in their natural environment; in other words, whether behavioral responses to stress observed in the laboratory can be generalized to natural behavior. To further elucidate the relative contributions of the natural selection pressures influences, this study investigated the bio-behavioral and morphological effects of auditory predator cues (owl territorial calls) in males and females of three wild rodent species in a laboratory set-up: Acomys cahirinus; Gerbillus henleyi; and Gerbillus gerbillus. Our results indicate that owl territorial calls elicited not only “fight or flight” behavioral responses but caused PTSD-like behavioral responses in wild rodents that have never encountered owls in nature and could cause, in some individuals, enduring physiological and morphological responses that parallel those seen in laboratory rodents or traumatized people. In all rodent species, the PTSD phenotype was characterized by a blunting of fecal cortisol metabolite response early after exposure and by a lower hypothalamic orexin-A level and lower total dendritic length and number in the dentate gyrus granule cells eight days after predator exposure. Phenotypically, this refers to a significant functional impairment that could affect reproduction and survival and thus fitness and population dynamics.
... LHD is the typical REM sleep posture and thus can be used to estimate REM sleep. This is based on the fact that due to postural atonia, the animal's head needs to be rested in REM sleep (Lima et al., 2005;Zepelin et al., 2005). It is a common method for estimating REM sleep by the LHD position; recent studies using this method include studies of Cetartiodactyla, like the common eland (Zizkova et al., 2013), the giraffe (Seeber et al., 2012), the dromedary camel (El Allali et al., 2022), and cattle (Ternman et al., 2014). ...
Full-text available
Introduction The nocturnal behavior of many ungulate species has currently not been sufficiently studied. However, the behavioral patterns of large herbivores vary greatly between day and night, and knowledge about species’ behavior is not only scientifically interesting, but also required for successful animal management and husbandry. Material and methods In the current study, the nocturnal behavior of 196 individuals of 19 ungulate species in 20 European zoos is studied, providing the first description of the nocturnal behavior of some of the species. The importance of a wide range of possible factors influencing nocturnal behavior is discussed. Specifically, the behavioral states of standing and lying were analyzed, evaluating the proportion and number of phases in each behavior. The underlying data consist of 101,629 h of video material from 9,239 nights. A deep learning-based software package named Behavioral Observations by Videos and Images Using Deep-Learning Software (BOVIDS) was used to analyze the recordings. The analysis of the influencing factors was based on random forest regression and Shapley additive explanation (SHAP) analysis. Results The results indicate that age, body size, and feeding type are the most important factors influencing nocturnal behavior across all species. There are strong differences between the zebra species and the observed Cetartiodactyla as well as white rhinos. The main difference is that zebras spend significantly less time in a lying position than Cetartiodactyla. Discussion Overall, the results fit well into the sparse existing literature and the data can be considered a valid reference for further research and might help to assess animal's welfare in zoos.
... By switching to a predominantly nocturnal pattern of activity in human-dominated landscapes, wild boars chose to expose their resting phase to a higher risk of anthropic disturbance. This likely is a successful strategy for risk mitigation, as resting is a well-known and efficient anti-predator strategy (Lima et al., 2005). However, this also means that the daily choice of a resting site is critical for minimizing the risk of encountering people. ...
Full-text available
Many animals living in anthropized landscapes try to avoid encountering people by being active at night. By doing so, however, they risk being disturbed while at rest during the day. To mitigate this risk, diurnally resting species may be highly selective about where they rest. Here, we used GPS and activity sensors to study how wild boars (Sus scrofa) might adjust their resting site selection and revisitation patterns to the risk of disturbance by people. We evaluated the probability of daytime relocation to assess the efficacy of wild boars' resting strategy in reducing the risk of human encounter while at rest. We attempted to identify the cause of some relocations using audio recordings. Generally, we found that wild boars did not specifically avoid resting near villages or roads, that is, where the risk of encountering people is higher, if they could find sites with suitable vegetation cover. The risk of disturbance by people was low, even near villages. Resting sites located close to villages were visited more repeatedly than those located further away, suggesting that focusing on a few familiar and quiet resting sites was a successful strategy for resting undisturbed in an anthropized landscape.
... In addition, predator avoidance is a factor underlying the formation of sleeping clusters in primates (Zhang et al., 2011), which increase cohesion and are generally large in primates (Anderson, 1984), promoting predator detection and group defense (Brividoro et al., 2021). In response to predators, primates have developed specialized sleep-wake mechanisms, frequently waking during the night to maintain vigilance and ensure their safety (Eban-Rothschild et al., 2017;Lima et al., 2005). Diurnal animals often awaken or move within their sleeping sites at night to avoid predators (Pruetz, 2018). ...