Behavioral sleep in the giraffe (Giraffa Camelopardalis) in a zoological garden

Abstract

Behavioural sleep was assessed for 152 nights in 5 adult, 2 immature and 1 juvenile giraffes at a zoological garden, using continuous time-lapse video recording. Sleep occurred while the giraffes were standing (SS) and in recumbency (RS). Paradoxical sleep (PS) was recognized by the peculiar positioning of the head on the croup and by phasic events. The 24-h sleep profile had a main bimodal nocturnal sleep period between 20.00 and 07.00 hours, with a trough between 02.00 and 04.00 hours, and several short naps between 12.00 and 16.00 hours. Total sleep time (TST), excluding the juvenile, was 4.6 h, whereby PS comprised only 4.7%. TST was not age dependent, but the lowest amount of RS and the highest amount of SS occurred in the oldest and the two oldest animals, respectively. Sleep was fragmented, as indicated by the predominance of RS episodes lasting less than 11 min. Sleep cycle duration was very variable with most values between 1 and 35 min (when no waking or RS was allowed within PS episodes), or 6-35 min (when the criteria for ending a PS episode allowed 1-2 min interruptions by RS). There were several indications for sleep regulation: (i) RS and SS complemented each other to yield a relatively stable daily value of TST; (ii) sleep was redistributed on nights following a day when the giraffes spent a few hours in an outside enclosure. The first peak of the bimodal sleep profile was absent and RS was more prominent in the second half of the night compared with nights following days spent in the barn; and (iii) napping was followed by a minor reduction of RS and an increase in SS in the subsequent night compared with nights following days without naps.

Full text

J. Sleep Res. (1996) 5, 21–32
Behavioural sleep in the giraffe(Giraffa camelopardalis)ina
zoological garden
I. TOBLER andB. SCHWIERIN
Institute of Pharmacology, University of Zu
¨rich, Zu
¨rich, Switzerland
Accepted 10 January 1996; received 17 November 1995
ABSTRACT
Behavioural sleep was assessed for 152 nights in 5 adult, 2 immature and 1 juvenile
giraffes at a zoological garden, using continuous time-lapse video recording. Sleep
occurred while the giraffes were standing (SS) and in recumbency (RS). Paradoxical
sleep (PS) was recognized by the peculiar positioning of the head on the croup and
by phasic events. The 24-h sleep profile had a main bimodal nocturnal sleep period
between 20.00 and 07.00 hours, with a trough between 02.00 and 04.00 hours, and
several short naps between 12.00 and 16.00 hours. Total sleep time (TST), excluding
the juvenile, was 4.6 h, whereby PS comprised only 4.7%. TST was not age dependent,
but the lowest amount of RS and the highest amount of SS occurred in the oldest
and the two oldest animals, respectively. Sleep was fragmented, as indicated by the
predominance of RS episodes lasting less than 11 min. Sleep cycle duration was very
variable with most values between 1 and 35 min (when no waking or RS was allowed
within PS episodes), or 6–35 min (when the criteria for ending a PS episode allowed
1–2 min interruptions by RS).
There were several indications for sleep regulation: (i) RS and SS complemented
each other to yield a relatively stable daily value of TST; (ii) sleep was redistributed
on nights following a day when the giraffes spent a few hours in an outside enclosure.
The first peak of the bimodal sleep profile was absent and RS was more prominent
in the second half of the night compared with nights following days spent in the barn;
and (iii) napping was followed by a minor reduction of RS and an increase in SS in
the subsequent night compared with nights following days without naps.
 age, behavioural sleep, giraffe, nap, sleep regulation
INTRODUCTION sleep quotas are the body mass index, brain weight and
metabolic rate (Zepelin and Rechtschaffen 1974; Allison and
Little is known about the sleep behaviour of large herbivores, Cicchetti 1976; Elgar et al. 1988; Zepelin 1994). The amount
especiallythosewhicharenotdomesticated.Zoologicalgardens of sleep decreases as a function of increasing body mass, or
provide an environment where observations of animals in semi- brain weight, but the latter is a better predictor of daily sleep
wild conditions are possible. In addition, modern technology quotas than either body weight or metabolic rate. ‘Danger’
allows continuous observation with light-sensitive cameras of variables based on asubjective index consisting ofthe predation
animals in zoo enclosures, so that the animals are undisturbed risk were also found to be better predictors of total sleep time
by the observer or by additional light. thanbody or brain weight(Meddis 1983). Correlationsbetween
Total sleep time exhibits a large species variation (Zepelin the amount of paradoxical sleep (PS) and the danger of
and Rechtschaffen 1974; Campbell and Tobler 1984). The predation showed that animals with a small predation index
amount of sleep in ungulates is of special interest because some (e.g. small rodents, large carnivores) exhibit more PS, whereas
of the constitutional variables which correlate best with daily the ungulates, particularly the undomesticated species, exhibit
little PS (Allison and Cicchetti 1976). The domesticated
herbivores seem to have retained this feature as they also
Correspondence: Prof. Irene Tobler Ph.D. Institute of Pharmacology
exhibit little sleep. However, the sleep data of large animals
University of Zu
¨rich Winterthurerstrasse 190 CH-8057 Zu
¨rich,
Switzerland. Tel.: +41/ 1-257-2957; Fax: +41/ 1-257-5707
such as the giraffe and the elephant (which contribute to
21
1996 European Sleep Research Society

22 I. Tobler and B. Schwierin
the correlations), are based on direct observations of few ‘deep sleep’ was considered as sleep. Since drowsiness was
individuals and the reliability of the data can be questioned. difficult to define it was not quantified (Immelmann 1958).
The exclusion of large grazing animals considerably weakened A subsequent investigation which comprised 20 nights of
the correlation between body size and total sleep time (Meddis continuous observation and 20 nights of observations until
1983), but was again strong if smaller grazing species were also after midnight, found that adults did not engage in
excluded (Elgar et al. 1988). recumbency during the day, whereas the 4-month old juvenile
Several factors may contribute to the small amounts of sleep lay down 2–3 times at mid-day for ≈10 min but did not
reported for ungulates. In a zoological garden they are rarely sleep (Immelmann and Gebbing 1962). The adults became
seen sleeping, possibly due to the frequent disturbances by the quiet at this time, either ruminating or standing completely
public. In addition, observations at night, when it is more still.
probable that the animals are sleeping, are influenced by the These observations were complemented by a 3-week
presence of the observer. Wild animals do have the capacity continuous observation of two captive females, 3 y and 9 y
tofurther reduce theirsleep whenthe circumstances areadverse old (the latter was shortly before parturition) in the Buffalo
(Hediger 1983, 1985). Zoological Garden (Kristal and Noonan 1979). ‘Deep’ sleep
An additional factor contributing to the short sleep (assumed to correspond to fast-wave sleep) was short (1–10
duration in ungulates (compared, e.g. with carnivores) is the min) and occurred only 1–2 times per night, whereas 3–8
vegetarian diet which necessitates large amounts of food reclining episodes lasting 3–75 min were present, with the
intake. In the wild, an adult giraffe bull consumes ≈37kg longer ones later in the night. Cud was chewed during most
of food per day (Moss 1975). Thus a large amount of the of the time spent in recumbency. Most of the reclining time
24h is spent in feeding and ruminating. EEG recordings in was spent with the neck in a vertical position, but occasionally
domesticated herbivores have shown that rumination and S-sleep (assumed to correspond to slow-wave or light sleep),
sleep are not exclusive behaviours. Non-REM sleep, but not occurred in episodes lasting from 5 to 30 min. S-sleep was
PS can occur during rumination (Ruckebusch and Bell 1970; characterized by the absence of cud chewing and a relaxation
Bell and Itabisashi 1972; Ruckebusch and Dougherty 1974). of the neck, which remained in a vertical position. D-sleep
In giraffe calves, complete rumination, which begins when began only after a long period of reclining rest or S-sleep. In
suckling is ended, occurred at the age of 6–8 months D-sleep the neck and head were lowered until the head rested
(Langman 1977). on the hip or thigh (‘sleeping swan’) and the eyes were closed.
Grzimek (1956) described the sleeping behaviour of four A similar sleep posture was reported for giraffes observed in
well-adapted giraffes observed for 14 nights in the Frankfurter the wild (Mejia in Moss 1975). During the first two days of its
Zoo, and disproved the common belief that giraffes do not life the calf spent ≈25% of 24 h sleeping with 90% in the D-
sleep at all. The opinion was that animals which are in danger sleep posture (Kristal and Noonan 1979).
of predation must compromise between the need for In the wild, i.e. in Kenya, Zaire, Tanzania and South Africa,
recuperation and avoidance of predators. He published the only day-time activities have been quantified for the giraffe
first photograph of an adult giraffe sleeping with its neck bent (Leuthold and Leuthold 1978; Foster; Mejia, Innis, Backhaus,
backwards resting the tip of its head on the ground, behind in Moss 1975). Most of these studies did not include night
the laterally extended hind limb (reproduced in Zepelin 1994; observations or the data were collected on full moon nights.
see also Fig. 1). Earlier, Hediger (1955) had published a photo Often all giraffe in a herd were seen standing or lying in one
of a young giraffe in a similar position, where the head tip place, and chewing. Rumination comprised 3–5 h per day. In
rests on the croup and the hind leg is extended towards addition, 2–3 h were spent lying down without chewing (males
the front in parallel with the abdomen. These postures were sometimes during daytime, females usually only at night).
considered typical for ‘deep sleep’, and are not unique to the These animals lay on the brisket, with legs curled under them
giraffe. Okapi, the closest relative of the giraffe, antelope and and the neck and head held upright. ‘Deep sleep’, with the
cow also bend their neck backwards, but not only the tip of head resting onthe flank,lasted only about1 min,and occurred
the head but the entire head can be placed on the croup (the for no more than 5–30 min per day. Only this position was
long neck of the adult giraffe does not allow a positioning on considered to correspond to sleep. Sometime after midnight
the croup). This posture in giraffes was later described again, the animals stopped feeding and lay down and ruminated for
being referred to as D-sleep (‘deep’ sleep) and assumed to 2–3 h. At this time they slept for very short intervals. An age
represent fast-wave sleep (Kristal and Noonan 1978). dependence of the amount of time spent lying was observed
Grzimek (1956) reported that adult giraffes spent ≈6.5h (Langman 1977). Time in recumbency decreased from infants
in recumbency with five ‘deep sleep’ episodes lasting 2.5–6 (0–60 d) to juveniles (60d–1.5 y) and was least in theimmature
min (total ‘deep sleep’: 21 min in the adult, 63–70 min in (1.5–3 y female, 1.5–4 y male).
the 4-month old juvenile). The same data analysed later in The aims of this study were to describe and quantify sleep
more detail resulted in 7–9 h per night in recumbency, with behaviour, assess its 24-h pattern and to investigate the effects
2–3 recumbent episodes of 11 min to 3h interrupted by ≈of age and daytime access to an outdoor pen in a group of
1 hour of waking spent in feeding and defaecation. Most
of recumbency was described as a drowsy state and only giraffes kept in the Emmen Zoo.
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32

Behavioural sleep in the giraffe 23
Table 1 Time of day of main sleep period
Gender Age (y) N Sleep onset Wake onset
(RS or SS)
Twanja f 0.2 juvenile 6 19.00 (7.4) 06.00 (9.3)
Raisa f 2.1 adolescent 25 20.06 (7.7) 06.18 (11.8)
Hans m 2.2 adolescent 25 20.36 (8.6) 05.54 (14.5)
Naomi f 3.4 adult 22 19.30 (10.7) 06.18 (14.1)
Augusta f 3.5 adult 25 20.36 (10.3) 06.12 (8.3)
Tiny f 3.6 adult 18 21.24 (13.2) 06.36 (9.6)
Taos f 13 mother 6 19.54 (14.9) 06.42 (8.7)
Cornelia f 20 adult 25 21.24 (16.7) 06.30 (8.6)
Mean (n=8) 20.18 (18.1) 06.18 (5.9)
Mean (n=7)
a
20.30 (16.7) 06.24 (6.0)
Time of day is the mean time (±SEM in minutes), over all days per individual or
a
mean value
excluding Twanja; n=number of animals. RS=recumbent sleep, SS=standing sleep; N=
number of days.
METHODS nights of Group 2, Naomi (7 nights) or Tiny (1 night) remained
inGroup1andcouldnot be assessed. Recordings were obtained
The behaviour of 8 giraffes (Giraffa camelopardalis) (age 0.2 for 25d (Group 1) and 6 d (Group 2).
months–20 y; 5 adult females, 2 adolescents, a male and female
and one juvenile female; Table 1) was recorded continuously RESULTS
on two time-lapse video recorders (Panasonic AG-6720 A) for
6–25d during the hours when the animals were in the animal Behavioural states
house of a zoological garden. The recordings were performed Waking was scored when the animals were standing or
with two infrared sensitive cameras (CCD, with automatic recumbent, but moving. Waking behaviour during the night
time-lapse shutter) and infrared illuminators (Sennheiser, SZI consisted largely of feeding. Sleep occurred in a recumbent
1019A, 950 nm), allowing data collection in the undisturbed (RS) and standing posture (SS). SS was scored when the
animals. The cameras were mounted on opposite sides of the animals were motionless(including theears, which wereusually
enclosure (at a height of ≈8–10m) to ensure that the entire directed backwards), and the neck position was at a narrower
enclosure was visible. The giraffes were kept in one of several angle relative to the ground, than was the case during waking.
enclosures (9.60×5.85m) within a barn. When the weather SS usually alternated with short waking episodes, characterized
was mild the animals were allowed to go outside into a large by righting of the neck and head lifting accompanied by ear
park next to the giraffe house, shared with animals of several movements. SS closely resembled the standing sleep reported
other African species, for 2–6 h between 09.40 and 16.30 hours. earlier in elephants (Tobler 1992). In recumbency the animals
Thereafter they were returned to adjacent pens in the giraffecould either be awake or sleeping. The posture in RS was
house, where food was available. An artificial LD cycle typical for ungulates (Fig. 1): lying on the brisket and the
provided light from 06.40 to 18.40 hours; light intensity, 95–110 abdomen or flank with the legs folded under and slightly
lx. In addition several windows provided natural daylight. The displaced to the sides; the neck bent forward at an angle of
recordings were performed in spring (March–April) in Holland
≈70°from the ground (sometimes even as little as 30°), i.e.
(53°15 min latitude). in a position less vertical than that of waking; and the neck
The video systems simultaneously recorded 24 h of real time and head immobile. The neck angle relative to the ground
on a 3-h tape. The tapes were played back at normal speed could not be determined reliably due to the placement of the
(50 half-frames per second) and the behaviour of the animals cameras above the animals. Therefore, the changes in angle
was visually scored for 1-min epochs. Epochs lasting <30s could not be used for scoring the behavioural states. The
were not considered, whereas epochs lasting between 30 and presence or absence of chewing was not considered a criterion
59s were rounded to 1 min. For scoring of the individual for RS or waking (uninterrupted chewing could occur for long
animals, the two tapes recorded simultaneously from different intervals, i.e. ≈30 min, while no other motor activity was
angles of the barn, were compared. Missing values due to lack evident). PS was scored when in RS the animals suddenly bent
of visibility of the giraffes occurred only in one giraffe (Naomi) the neck backwards and rested the head on the flank of a rear
in 3 of the 22d (missing: 36, 73 and 140 min). leg (Fig. 1). During these PS episodes phasic events were
The giraffes were kept in 2 groups: Group 1 consisted of 5 sometimes apparent, consisting of twitches of the ears and
females and 1 male, Group 2 consisted of a mother and her neck.
female juvenile progeny. The two groups were kept in adjacent The transition from locomotion to RS was gradual. Thus
enclosures. Either Group 1 or Group 2 was placed into the
enclosurewherethe cameras were installed. On severalrecorded individuals had a preferred sleeping site within the enclosure
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32

24 I. Tobler and B. Schwierin
Figure 1. Behavioural changes leading to postures typical for the vigilance states recumbent sleep (bottom left) and PS (bottom right).
which they would approach several times before finally taking an adult and the immature male, are illustrated in Figs 2 and
3. On several days napping episodes were evident. On mosta recumbent posture. The transition from recumbent
wakefulness to RS could occur several times before a PS days the main sleep period was between 20.00 and 07.00 hours.
episode could be seen. The occurrence of SS was not related
to the subsequent RS. Sleep profile
Sleep rarely occurred outside the barn (our own occasional The mean 24-h sleep profile is illustrated in Fig. 4. It exhibited a
observations and personal communication from the similar bimodal pattern in all the individuals with RS occurring
caretakers). The animals were always returned to the barn between 20.00 and 07.00 hours, with a small trough between
before 17.00 hours, and thereafter spent several hours feeding. 02.00 and 04.00 hours. SS occurred more frequently towards
Sleep onset was defined as the time of day when sleep occurred the end of the sleep period (ANOVA factor interval from 18.00
for at least 3 consecutive minutes. Sleep onset was at 20.18 to 08.00 hours, N=6, d.f.=13, SS F=4.15; RS F=8.54, PS
hours (±18.1 min; N=8) and the end of the sleep period was F=4.12, all vigilance states P<0.0001, except PS/TST which
at 06.18 hours (±5.9 min). The time of day when sleep began was n.s.). In the comparison of the first and second half of the
did not differ between nights following a day with access to night, SS was significantly more abundant in the second half
the outside enclosure (20.6 hours±23.8 min) vs. nights (P<0.02, two-tailed paired t-test). PS was evenly distributed
following days spent inside (20.6 hours±22.0 min).
Thevigilancestateson8consecutivedaysfortwoindividuals, across the entire night.
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32

Behavioural sleep in the giraffe 25
Figure 2. Vigilance states of an individual adult female giraffe (Augusta, age 3.5 years) recorded for eight consecutive nights. W=waking while
in locomotion or moving, WR=waking recumbent, SS=standing sleep, RS=recumbent sleep, PS=paradoxical sleep.
The mean 24-h-values of the vigilance states are presented unchanged when the data were restricted to days when the
animals were not allowed outside.in Table 2 and Fig. 5. Total sleep time expressed as a percentage
of24 h varied between 14.4and 22.4% with sleep ina recumbent
position between 6.4 and 20.4% of 24h. A one-way ANOVA Sleep episode duration
factor ‘age’, for the amount of each of the vigilance states was The PS episodes were short, with ≈24% lasting less than 1
significant for all stages including TST (d.f.=7, SS F=21.31, min (Fig. 6, Table 3). The longest PS episodes occurred in the
RS F=20.72, PS F=8.41, TST F=7.37, all vigilance states juvenile, which exhibited also the largest variability in PS
P<0.0001). While no age relationship was evident for TST in duration. The inclusion of a maximum interruption criterion
the post-hoc analysis, the lowest amount of RS was present in of 2 min RS to interrupt PS did not have a large effect on the
the oldest giraffe and SS was most prominent in the oldest two duration of PS episodes (mean duration in minutes: 4.3±0.3,
giraffes (comprising 7.5% of 24 h; Fig. 5, Duncan multiple N=8; 4.1±0.23, N=7; Table 3). The mean duration of RS
range test, P<0.05). The juvenile never exhibited SS. An age episodes (excluding PS) varied between 8.1 and 13.2 min,
dependence became evident for the total time in recumbency, although single RS episodes could last for up to 100 min. Of
irrespective of whether the animals were asleep or in all RS episodes 58.7%±2.4 (N=7) were shorter than 11 min
wakefulness (Table 2). The oldest animal exhibited the lowest (data not shown). No evident age relationship was present for
amount and the youngest the highest amount of recumbency
(Duncan multiple range test, P<0.05). This result was episode duration of any vigilance state, with the exception
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32

26 I. Tobler and B. Schwierin
Figure 3. Vigilance states of an individual immature male giraffe (Hans, age 2.2y) recorded for eight consecuitve nights. W=waking while in
locomotion or moving, WR=waking recumbent, SS=standing sleep, RS=recumbent sleep, PS=paradoxical sleep.
that the juvenile exhibited the shortest RS episodes (Duncan Access to outdoor pen
multiple range test, P<0.05; N=8). Only Group 1 had access to the outdoor pen (mean duration
of time spent outdoors 5.0 h±1.4 min, N=5; range 2.1–6.7 h;
Sleep cycle duration on N=10d the entire group remained inside due to bad
weather, and the male (Hans) was outside on only two of the
A PS cycle was defined as the interval between the onset of 24 d and was excluded from the analysis). Since the juvenile
two consecutive PS episodes. When no interruptions of the female (Twanja) was still too young to be allowed outside in
cycle were allowed by other vigilance states, there was a large the early spring weather, the mother (Taos) and the juvenile
peak of PS cycles of 1–5 min duration and an even distribution always remained inside the barn. The comparison of the sleep
of cycles of 6–35 min duration (Fig. 7, top left). The inclusion states and their distribution on nights (18.00–08.00 hours)
ofacriterion allowing <5 min of wakingassingle or consecutive following a day spent outdoors with nights following upon
1-min epochs within a PS cycle did not have a major effect on days when the animals remained in the barn resulted in a
this distribution (Fig. 7, top right). Allowing 1 or 2 min RS significant redistribution of TST (Fig. 8; ANOVA factor
within a PS episode with or without the waking criterion interaction ‘condition ∗interval’ for 1-h intervals, d.f.=13,
reduced the frequency of PS cycles Ζ5 min, whereas the
distribution of cycles between 6 and 35 min remained similar. F=2.43, P<0.007). The biphasic sleep pattern was less
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32

Behavioural sleep in the giraffe 27
compriseatleast 5 min between 08.00 and17.00 hours. Daytime
sleep was usually seen between 12.00 and 16.00 hours in some
of the giraffes (Figs 2, 3, 4, Table 2). The total duration of
daytime sleep is illustrated in the frequency histogram (Fig. 9).
The amount of daytime sleep was small since most naps
were short, comprising 5–50 min, but occasionally reaching a
cumulative value of 50–90 min.
The comparison of sleep duration in the nights following a
day with or without daytime sleep (N=3–15 vs. N=10d with
daytime sleep in 4 giraffes; daytime sleep < than 10 min was
excluded from this analysis) resulted in a significantly lower
RS value (48.1 min) after daytime sleep and an increase in
SS (15.2 min; P<0.04, two-tailed paired t-test); whereas TST
remained unchanged. The results were not affected when the
exclusion criterion of <10 min for daytime sleep was eliminated
(N=1–17 vs. N=4–10 days with daytime sleep, in 5 giraffes).
The duration of all daytime sleep episodes did not correlate
significantly with TST on the subsequent night, or with the
amount of RS alone (r
2
=0.06 for TST and for RS).
DISCUSSION
The animals obtained a relatively low amount of sleep per
24h, although we did include SS, which previously has never
been quantified in giraffes, and RS which due to the frequent
rumination bouts was not considered as sleep. However, the
4.6 h are more than usually reported for the giraffe, since most
other studies were influenced by the notion that giraffe sleep
only in the ‘deep sleep’ posture, which corresponds to the
behavioural PS scored in our study. Kristal andNoonan (1979)
did describe S-sleep (behavioural sleep in recumbency), but did
not report its amount in the adults. In contrast to their study
we chose to exclude chewing in recumbency as a criterion for
waking. EEG recordings in cattle and sheep have shown that
nonREM sleep occurs during cud chewing (Ruckebusch et al.
Figure 4. Daily distribution of the vigilance states. The lines connect
1974; Bell and Itabisashi 1993).
mean hourly values. Top panel: mean values per hour (N=6
giraffes) recorded for 18–25d, middle panel: a 13-y old female,
SS has also been observed in the elephant (Tobler 1992), the
mother of the juvenile female (3-month old, lower panel). The
behaviour during this state being very similar in elephants and
mother and the juvenile were simultaneously recorded for 6d. Black
giraffes. However, in the giraffes it did not precede RS as was
area, paradoxical sleep (PS), dashed area, recumbent sleep (RS),
often the case in the elephants, and is was not interrupted by
white area, standing sleep (SS). Total area under the curve represents
total sleep time.
short waking bouts characterized by swaying movements of
the body and head (and flapping the ears) as in the elephants.
It has been shown on the basis of EEG recordings for many
ungulates that sleep can occur in a standing position (e.g.
prominent when the animals had been outside, since the lower Ruckebusch and Bell 1970). On the basis of such data it can
values between 02.00 and 04.00 hours were absent. While be assumed that the behaviourobserved in the standing giraffes,
TST over the entire night or the 24h was unaffected, it was scored here as SS, could indeed be considered sleep. The
significantly decreased in the second half of the night following inclusion of SS did not have a large effect on the TST reported
an indoor day vs. a day with access to the outside enclosure here since SS comprised only 47 min of 24 h. In general, RS
(mean values in minutes after days: inside 131.2±5.4, outdoors episodes were short (58.7% RS episodes were shorter than 10
150.3±3.2; N=5; P<0.008, two-tailed paired t-test). min; in the juvenile over 70% were shorter than 10 min). Thus
sleep in giraffe is very fragmented. Possibly the long legs and
Effect of daytime sleep neck contribute to a difficult recumbent sleeping posture which
does not allow quick fleeing as a response to external stimuli.
Napping (SS, RS and PS) was scored on those days when the
animals remained indoors. A daytime sleep episode had to The difficulties in raising from the ground could contribute to
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32

28 I. Tobler and B. Schwierin
Table 2 Duration of vigilance states, recumbency and daytime sleep
Age (y) N SS RS PS PS+RS PS/TST TST N Nap REC
Twanja 0.2 5 0.0 (0.0) 15.0 (0.6) 1.4 (0.3) 16.4 (0.4) 8.6 (1.8) 16.4 (0.4) 1 0.6 40.7 (1.6)
Raisa 2.1 25 1.5 (0.4) 17.3 (1.0) 0.7 (0.1) 18.1 (1.0) 3.7 (0.4) 19.6 (0.9) 3 1.0 (0.3) 25.2 (1.1)
Hans 2.2 25 0.8 (0.2) 18.3 (1.0) 1.3 (0.1) 19.6 (1.0) 7.0 (0.8) 20.4 (1.0) 17 1.9 (0.3) 30.7 (0.8)
Naomi 3.4 22 0.9 (0.4) 20.4 (1.0) 1.1 (0.1) 21.5 (1.1) 4.9 (0.5) 22.4 (1.1) 6 4.0 (0.5) 32.0 (1.2)
Augusta 3.5 25 2.1 (0.4) 17.5 (0.9) 1.1 (0.1) 18.6 (0.9) 5.6 (0.6) 20.7 (0.8) 7 1.0 (0.2) 26.7 (0.9)
Tiny 3.6 18 2.9 (0.6) 12.2 (0.9) 0.8 (0.1) 13.0 (1.0) 4.9 (0.5) 16.0 (0.9) 0 0.0 (0.0) 18.4 (1.2)
Taos 13 6 7.5 (2.0) 11.8 (2.1) 0.7 (0.2) 12.5 (2.1) 3.4 (0.9) 20.1 (0.7) 0 0.0 (0.0) 16.4 (2.7)
Cornelia 20 25 7.5 (0.8) 6.4 (1.0) 0.4 (0.1) 6.9 (1.1) 3.1 (0.7) 14.4 (1.1) 0 0.0 (0.0) 11.1 (1.5)
Mean (n=8) 2.9 (1.1) 14.9 (1.6) 1.0 (0.1) 15.8 (1.7) 5.1 (0.7) 18.7 (1.0) 25.1 (3.4)
Mean (n=7)
a
3.3 (1.1) 14.9 (1.8) 0.9 (0.1) 15.7 (1.9) 4.7 (0.5) 19.1 (1.1) 22.9 (3.0)
Sleep was subdivided into standing (SS), recumbent (RS) and paradoxical sleep (PS). REC=total time spent in recumbency. Total sleep (TST)
includes the three subdivisions of sleep. Mean values in percentage of 24h (±SEM) except for PS/TST; age in years; n=number of animals;
N=number of days or number of days with daytime sleep; Nap=TST between 12:00 and 16:00 hours;
a
mean value excluding Twanja.
Figure 5. Vigilance states represented as a
function of age. The bars represent mean
values±s.e.m. as percentage of 24h for total
sleep, standing sleep and recumbent sleep, and as
percentage of total sleep time for paradoxical
sleep (note the different scale for paradoxical
sleep). The age of the animals is indicated below
the bars.
the short duration of sleep and maybe to the development of that PS is less abundant and of shorter duration in animals
with a large predatory index (Allison and Cicchetti 1976). Thethe capacity to obtain sleep in a standing posture. Elephants
havedifficulties in rising easily fromrecumbency,also; however, question remains open as to whether the PS described here
(which others have quantified as ‘deep sleep’) corresponds tothe mean duration of their RS (which included PS) was 61–77
min (Tobler 1992). It is probable that elephants are less REMsleep found in othermammals; this needsto beconfirmed
by electroencephalography. The remarkable posture, thesusceptible to predation than giraffe.
The PS episodes were brief, usually below 3 min and they concomitant phasic events and the EEG findings in species
such as horse, cattle, sheep and piglets showing that PS isonly amounted to 13 min per 24h or 4.7% of TST. These
values are similar to those for ‘deep sleep’ episodes which attained only when in recumbency and the head can be rested
(Ruckebusch et al. 1974), are a strong indication that thelasted 1–10 min in 1–2 episodes per night reported by Kristal
and Noonan (1979) and the 2.5–6 min-episodes found by behavioural PS defined in the giraffe is a correlate of PS as
described by EEG patterns in other mammals. In summary,Grzimek (1956). Thus the present data support the hypothesis
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32

Behavioural sleep in the giraffe 29
min in a 4-month old, but similar to the 21 min in the adults
(Grzimek 1956). Although no age effect was found for PS, the
juvenile did exhibit the largest variation in the duration of PS
episodes, but as in many other species PS/TST was most
prominent in the youngest animal. In addition, RS episodes
were shortest in the juvenile, indicating that her sleep was the
most fragmented. SS was never observed in this animal, which
is reminiscent of the late appearance of SS in the developing
young elephant (Tobler 1992).
An additional factor which may explain the larger difference
between young and adults in both elephants and giraffes is
that young giraffes have only a weak bond with their mother
except during the first few days after birth, in contrast to the
strong bond lasting for several years in the elephant. However,
Langman (1977) who investigated the cow–calf relationship in
giraffes observed cases of a strong relationship between the
giraffe cow and calf which lasted until the cow’s next calving.
Giraffe calf usually suckle only until approx. one month of
age, but up to one year has been seen, whereas elephants suckle
3–4 years. A giraffe calf stays with its mother till 16–18 months
(usually in a ‘kindergarten’, i.e. nursery) when it becomes
independent. Sexual maturation is reached at 4 years. Our
Figure 6. Frequency distribution of paradoxical sleep episodes. Bars
juvenile was never observed suckling, which indicates a
represent mean values±s.e.m. computed for 10-min bins of N=7
relatively large independence from the mother.
giraffes (top panel) and the single juvenile female (bottom panel).
An age dependence in the amount of time spent lying was
Data are expressed as percentage of the total amount of episodes for
seen in giraffes in the wild, but no details are given (Langman
each individual. Note the different scale of the top and the lower
1977). The adult male was seen only once sleeping for 10 min
panels.
motionless in a standing position with his eyes closed. In the
zoo total time spent in recumbency (7–9h) did not differ
between the adults and the single juvenile (Grzimek 1956).
the present data suggest that sleep duration in the giraffe has The juvenile did get up more frequently but returned into
been underestimated. recumbency more rapidly than the adults. In the present study
It is well known that for many mammalian species the the juvenile exhibited the largest amount of total recumbency
amounts of each sleep stage depends on the age and level of and the oldest giraffe the smallest amount (Table 2). Also,
development at birth (rat, cat, guinea pig: Jouvet-Mounier et another ungulate exhibited an age-dependent amount of time
al. 1969; piglet: Kuipers and Whatson 1979). Due to the age spent in recumbency: foals (7.1% of 24 h), stallion (2.4%), and
distribution in our giraffes such aspects could be addressed. mares (3.6%) (Bubenik 1978). The oldest giraffes exhibited the
However, the 3-month old juvenile borders on to the next age lowest amount of RS and largest amount of SS. Assuming that
category of ‘immature’, and cannot be expected to reflect the the reclining position involves considerable effort (Fig. 1), it
sleep states of a young animal. The amount of PS in the 3-
month old female was 20 min which was less than the 63–72 could explain why these older animals obtained less RS and
Table 3 Duration of vigilance state
episodes
Age (y) PS N RS N SS N
Twanja 0.2 5.6 (1.0) 18 8.1 (0.4) 152 0.0 (0.0) 0
Raisa 2.1 2.9 (0.2) 94 13.2 (0.5) 473 5.7 (0.6) 94
Hans 2.2 4.9 (0.2) 99 11.1 (0.4) 591 5.4 (0.7) 53
Naomi 3.4 3.8 (0.3) 92 12.0 (0.5) 536 5.3 (0.9) 52
Augusta 3.5 3.5 (0.2) 118 10.9 (0.4) 578 6.7 (0.6) 112
Tiny 3.6 3.8 (0.4) 54 10.0 (0.5) 316 6.9 (0.6) 111
Taos 13 3.2 (0.4) 19 11.4 (1.1) 90 6.0 (0.5) 109
Cornelia 20 4.2 (0.4) 38 9.9 (0.6) 234 5.2 (0.2) 517
Mean (n=8) 4.0 (0.3) 10.8 (0.5) 5.1 (0.8)
Mean (n=7)
a
3.7 (0.3) 11.2 (0.4) 5.9 (0.3)
Mean values in min (±SEM) of all days separately for each animal and over different groups
of animals. N=total number of episodes per giraffe; n=number of animals;
a
mean value
excluding Twanja.
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32

30 I. Tobler and B. Schwierin
Figure 7. Frequency distribution of PS-cycle
duration in 1-min bins. The frequency of each
bin was expressed as percentage of the total
number of PS cycles per individual. The bars
represent mean values±s.e.m. (N=7, i.e.
excluding the juvenile). Left panels: no waking
was allowed within the cycle. Right panels:
waking <5 min as single or consecutive 1-min
epochs was allowed within a cycle. Top panels:
no interruption of PS episodes was allowed.
Bottom panels: a maximum of 2 single or
consecutive 1-min RS epochs (but no waking)
was allowed within a PS episode.
Figure 9. Frequency distribution of total duration of the cumulated
daytime sleep episodes within single days. Bars represent the
frequency of daytime sleep duration classified into 5-min bins for
N=6 giraffes.
Figure 8. Effect of access to an outdoor pen during the day on sleep
distribution in the subsequent night. The lines connect mean hourly
values of total sleep time (±s.e.m., N=5), after days with (outdoor)
and without (indoor) access to an outdoor pen. Comparison of 1-h
(Zepelin and Rechtschaffen 1974; Allison and Cichetti 1976).
intervals outdoor vs. indoor, ∗∗P<0.007, ∗P<0.02, two-tailed paired
The daily sleep quota cited by Zepelin (1994) for the Asiatic
t-test).
elephant was 3.9h (African elephant 3.3 h) and for giraffes
1.9h, which contrasts with both our TST values in elephants
(circus without sleep ad lib: 3.5h; zoo – summer 5.9 h, winter
6.7h; Tobler 1992) and in giraffes, at 4.6 h (evenexcluding SS,compensated with an increase in SS. Also in the elephants the
oldest cow exhibited the least RS (Tobler 1992). which is arbitrary and could be considered as drowsiness this
would give values of: giraffe: 3.8h, elephants: circus 3.3h, zooIn summary, the smaller amount of sleep and the shorter
duration of RS and PS episodes in giraffes may be a summer 3.7 h, winter 4.5 h).
It has been shown for many species that sleep is regulatedconsequence of a larger predatory risk, but this does not
confirm the negative correlation withbody size or brain weight as a function of prior wakefulness (reviewed by Borbe
´ly 1994).
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32

Behavioural sleep in the giraffe 31
In particular, sleep intensity is increased after prolonged the two conditions were responsible for the redistribution of
sleep and its increase in the second half of the night.
wakefulnessorreduced in the night following anappingepisode In conclusion, this study demonstrates that behavioural
(Feinberg 1982; Werth et al. 1996). In several species TST is observations in undisturbed animals can contribute to our
increased and motor activity during sleep is decreased after a understanding of sleep mechanisms inspecies where performing
period of sleep deprivation (reviewed by Tobler 1985; Deboer EEG recordings is more difficult.
et al. 1994). Sleep regulation has been little investigated in
ungulates. Ruckebusch (Ruckebusch and Bell 1970;
Ruckebusch et al. 1974; Ruckebusch 1975) performed selective ACKNOWLEDGEMENTS
deprivation of REM sleep and partial sleep deprivation in cows The study was supported by the Swiss National Science
by preventing recumbency for 14–22h per day for 2–4 weeks. Foundation, grant 31.32574.91 and 3100.042500.94. The
Both nonREM sleep and REM sleep were significantly recordings were obtained in the Dierepark Emmen, Holland
increasedduringrecovery.During recovery,drowsiness, defined with the permission of H. Hiddingh. We thank Mr H. Scheeve
as a wakefulness with a mixture of high and low voltage and the caretakers of the giraffes for their cooperation and for
synchronized EEG activity, usually encompassing about 30% the daily changing of the video tapes, Mr K. Wu
¨thrich for
of wakefulness, was not affected. Similarly, cattle deprived technical assistance and P. Achermann and A.A. Borbe
´ly for
from lying for 3h increased the duration of lying in the critical reading of the manuscript.
subsequent recovery period (Metz et al. 1984/1995). Data from
the present study reflect several aspects of sleep regulation in
the giraffe. Whereas experimental manipulations such as sleep REFERENCES
deprivation in wild animals in a zoo are not without risks the
Adams, J. and Berg, J.K. Behavior of female African elephants
continuousobservationof animals does allow one toinvestigate
(Loxodonta africana) in captivity. Appl. Anim. Ethol., 1980, 6:
257–276.
the effects of napping episodes on the subsequent night. The
Allison, T. and Cicchetti, D.V. Sleep in mammals: ecological and
frequent occurrence of napping allowed the analysis of nights
constitutional correlates. Science, 1976, 194: 732–734.
following naps vs. nights not preceded by napping. If sleep in
Balch, C.C. Sleep in ruminants. Nature Lond., 1955, 175: 940–941.
the giraffe is finely regulated and if SS is a less intense sleep
Bell, F.R. and Itabisashi, T. The electroencephalogram of sheep and
stage than RS we would expect a decrease of RS in the nights
goats with special reference to rumination. Physiol. Behav., 1973,
11: 503–514.
following daytime naps. A significant reduction of RS was
Benedict, F.G. The Physiology of the Elephant. Carnegie Institution of
present, while SS was not affected. These results may be similar
Washington, Washington, 1936.
to the reduction of EEG slow-wave activity in humans on the
Borbe
´ly, A.A. Sleep homeostasis and models of sleep regulation. In:
night following a napping episode (Feinberg 1982; Werth et al.
M.H. Kryger, T. Roth, and W.C. Dement (Eds) Principles and
1996).
Practice of Sleep Medicine, 2nd edn. Saunders, Philadelphia, 1994:
309–320.
It is difficult to establish the relationship of SS to RS in the
Campbell, S.S. and Tobler, I. Animal sleep: a review of sleep duration
absence of EEG recordings. Thus it is not clear whether SS is
across phylogeny. Neurosci. Biobehav. Rev., 1984, 8: 269–300.
a less intense sleep state than RS. The relationship between SS
Dallaire, A. Rest behavior. Vet. clin. north. amer. (Equine Pract.), 1986,
and drowsiness as defined by the EEG by Ruckebusch (1972)
2: 591–607.
is also not clear. There are several indications that SS is a form
Dallaire, A., Toutain, P.L. and Ruckebusch, Y. Sur la pe
´riodicite
´du
sommeil paradoxal: faites et hypothe
`ses. Physiol. Behav., 1974, 13:
of ‘light’ sleep. Moreover, the present results show that these
395–400.
two sleep states were finely regulated. Thus both SS and RS
Deboer, T., Franken, P. and Tobler, I. Sleep and cortical temperature
showed a prominent intra- and inter-individual variation, but
in the Djungarian hamster under baseline conditions and after sleep
complemented each other, resulting in a relatively constant
deprivation. J. Comp. Physiol. A., 1994, 174: 145–155.
value for TST. This feature has also been observedin elephants
Deboer, T. and Tobler, I. Shortening of the photoperiod affects sleep
distribution, EEG and cortical temperature. J. Comp. Physiol. A.,
(Tobler 1992). A remarkable stability of TST hasbeen reported
in press.
for other species on the basis of EEG recordings, e.g. under
Elgar, M.A., Pagel, M.D. and Harvey, P.H. Sleep in mammals. Anim.
shortand long photoperiods. Despite themarked redistribution
Behav., 1988, 36: 1407–1419.
of sleep in the two photoperiods, TST remained constant (rat:
Feinberg,I., March, J.D., Floyd,T.C.,Jimison,R., Bossom-Demitrack,
Franken et al. 1995; Djungarian hamster: Deboer and Tobler
L. and Katz, P.H. Homeostatic changes during post-nap sleep
maintain baseline levels of delta EEG. Electroenceph. Clin.
1996).
Neurophysiol., 1985, 61: 134–137.
Furthermore, the access to the outdoor pen affected the sleep
Franken, P., Tobler, I. and Borbe
´ly, A.A. Varying the photoperiod in
profile in the subsequent night. Thus the bimodal sleep pattern
the laboratory rat: profound effect on the 24-h sleep pattern but no
was less prominent and TST significantly enhanced in the
effect on sleep homeostasis. Am. J. Physiol., 1995, 269: R691–R701.
second half of the night, on nights following access to the
Grzimek, B. Schlaf von Giraffen und Okapi. Naturwiss., 1956, 17: 406.
Hediger, H. Tiere im Schlaf, 2. Documenta Geigy, J.R. Geigy S.A.
outdoor enclosure. Since sleep onset was not affected, and TST
Basel, 1955: 1–9.
in the first part of the night did not differ between nights after
Hediger, H. Wie Tiere schlafen. Med. Klin., 1959, 20: 938–946.
hours outside vs. nights spent indoors, it is improbable that
Hediger, H. Comparative observations on sleep. Proc. R. Soc. Med.,
the result was a consequence of the activities in the outside
1969, 62: 153–156.
Hediger, H. Natural sleep behaviour in vertebrates. In: M. Monnier
enclosure. It is possible that different feeding patterns under
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32

32 I. Tobler and B. Schwierin
and M. Meulders (Eds) Functions of the Nervous System. Elsevier, Ruckebusch, Y. Sleep deprivation in cattle. Brain Res., 1974, 78:
495–499.
Amsterdam, 1983: 105–130. Ruckebusch, Y. and Toutain, P.L. La phyloge
´ne
`se du sommeil.
Hediger, H. Schlafsto
¨rungen. In: V. Faust (Ed.), Compendium Confront. Psychiat., 1977, 15: 9–48.
Psychiatricum. Hippokrates, Stuttgart, 1985: 48–54. Ruckebusch, Y. and Morel, M.T. Etude polygraphique du sommeil
Immelmann, K. Vom Schlaf der Giraffe. Umschau, 1958, 12: 356–357. chez le porc, C.R. Acad. Sci. Paris, 1968, 1346–1354.
Immelmann, K. and Gebbing, H. Schlaf bei Giraffiden. Z. Tierpsychol., Ruckebusch, Y. The relevance of drowsiness in the circadian cycle of
1962, 19: 84–92. farm animals. Anim. Behav., 1972, 20: 637–643.
Jouvet-Mounier, D., Astic, L. and Lacote, D. Ontogenesis of the states Ruckebusch, Y. The hypnogram as an index of adaptation of farm
of sleep in the rat, cat, and guinea pig during the first post-natal animals to changes in their environment. Appl. Anim. Ethol., 1975,
month. Devel. Psychobiol., 1969, 2: 216–239. 2: 3–18.
Kristal,M.B.andNoonan,M.Noteonsleepincaptivegiraffes(GiraffaRuckebusch, Y., Dougherty, R.W. and Cook, H.M. Jaw movements
camelopardalis reticulata). S. Afr. J. Zool., 1979, 14: 108. and rumen motility as criteria for measurement of deep sleep in
Kuipers, M. and Whatson, T.S. Sleep in piglets:an observational study. cattle. A.J. Vet. Res., 1974, 35: 1309–1312.
Appl. Anim. Ethol., 1979, 5: 145–151. Tobler, I. Deprivation of sleep and rest in vertebrates and invertebrates.
Langman, V.A. Cow-calf relationships in giraffe (GiraffaIn: S. Inoue
´and A.A. Borbe
´ly (Eds) Endogenous Sleep Substances
camelopardalis giraffa). Z. Tierpsychol., 1977, 43: 264–286. and Sleep Regulation. VNU Science Press BV, Utrecht (Taniguchi
McKay, G.M. Behavior and ecology of the asiatic elephant in Symposia, Series no. 8), 1985: 57–66.
southeastern Ceylon. Smithson. Contr. Zool., 1973, 125: 1–113. Tobler, I. Napping and polyphasic sleep in mammals. In: D. Dinges
Meddis,R. The evolutionofsleep. In: A.Mayes(Ed.) Sleep Mechanisms and R. Broughton (Eds) Napping: Biological, Psychological, and
and Functions in Humans and Animals: an evolutionary perspective. Medical Aspects. Raven Press, 1990: 9–30.
Van Nostrand Reinhold press (UK), Berkshire 1983: 57–106. Wert, E., Dijk, D.J., Achermann, P. and Borbe
´ly, A.A. Dynamics of
Metz, J.H.M. The reaction of cows to a short-term deprivation of the sleep EEG after an early evening nap: experimental data and
lying. Appl. Animal Behav. Sci., 1984/85, 13: 301–307. simulations. A. J. Physiol., in press.
Moss, C. Portraits in the Wild. Houghton Mifflin Co., Boston, 1975: Zepelin, H. and Rechtschaffen, A. Mammalian sleep, longevity, and
39–61. energy metabolism. Brain Behav. Evol., 1974, 10: 425–470.
Ruckebusch, Y. and Bell, F.R. Etude polygraphique et Zepelin, H. Mammalian sleep. In: M.H. Kryger, T. Roth, and W.C.
comportementale des e
´tats de veille et de sommeil chez la vache Dement (Eds) Principles and Practice of Sleep Medicine. WB
Saunders Company, 1989: 30–49.(Bos taurus). Ann. Rech. ve
´te
´r., 1970, 1: 41–62.
1996 European Sleep Research Society, J. Sleep Res.,5, 21–32