This preview shows page 1. Sign up to view the full content.
Unformatted text preview: Behavioral Ecology Vol. 11 No. 6: 624–632 Living with the enemy: avoidance of hyenas
and lions by cheetahs in the Serengeti
Sarah M. Durant
Institute of Zoology, Zoological Society of London, Regent’s Park, London, NW1 4RY, England
Predator avoidance is likely to play a strong role in structuring species communities, even where actual mortality due to predation
is low. In such systems, mortality may be low because predator avoidance is effective, and if the threat of predation is lifted then
entire community structures may be altered. Where competition is intense, then competitor avoidance may have a similar impact
on communities. Avoidance behaviors have been documented for a wide range of species, but this is the ﬁrst attempt to
document avoidance behavior within a large carnivore community. Audio playback techniques are used to examine the risk
perceived by cheetahs from their two main competitors that are also their main predators, lions and hyenas. The results from
these experiments show that cheetahs actively moved away from lion and hyena playback experiments, compared with dummy
playbacks where no sound was played. Cheetahs showed no differences in their responses to playbacks dependent on their sex
or reproductive status, suggesting they were responding principally to a competition rather than a predation threat. However,
cheetahs were much less likely to hunt after competitor playbacks than after dummy playbacks, and this resulted in a lower kill
rate after competitor playbacks, demonstrating that the perceived presence of competitors had a noticeable impact on the
foraging rate of cheetahs. Furthermore, while cheetahs moved just as far following lion playbacks as after hyena playbacks, they
spent signiﬁcantly more time looking at the loudspeaker and were less likely to make a kill after lion playbacks, suggesting that
cheetahs perceive lions to be a greater threat than hyenas. Key words: anti-predator behavior, carnivores, competition, foraging
strategy, kleptoparasitism, playback experiments, predator avoidance, predation risk. [Behav Ecol 11:624–632 (2000)] P redator avoidance, whereby prey avoid encounters with
predators, is one means by which prey are able to reduce
the chance of predation. An avoidance behavior can be deﬁned as any behavioral strategy that enhances the survivorship
of prey by reducing the probability that they occur within the
foraging range of potential predators (Brodie et al., 1991).
Predator avoidance is likely to play a strong role in structuring
species communities, even where actual mortality due to predation is low (Lima and Dill, 1990; Turner and Mittelbach,
1990). In such circumstances, mortality may be low precisely
because predator avoidance is effective and, if the threat of
predation is lifted, then entire community structures may be
altered. Avoidance of predation has attracted increasing attention over recent years, however predation is not the only
negative interspeciﬁc interaction which a species may want to
avoid; competitor avoidance behaviors may also be important
as a mechanism for shaping communities. Competitor avoidance can be deﬁned similarly to predator avoidance, except
enhancement of survival is through indirect rather than
through direct mechanisms.
Both predator and competitor avoidance are likely to carry
costs. For example, a reduction in activity or an increase in
the use of refuges due to a perceived risk of predation can
lead to a reduction in foraging rate (Kennedy et al., 1994;
Ward et al., 1997). The relative balance between the costs and
beneﬁts of predator avoidance may differ between different
species and between different age and reproductive classes
within species, and can result in different avoidance strategies,
even when the predator is identical (Peckarsky, 1996; Sih,
1992). In addition, some species may evolve an ability to adapt
their avoidance tactics in response to a perceived predation
risk (Loose and Dawidowicz, 1994; McIntosh and Townsend, Address correspondence to S. M. Durant. E-mail: email@example.com.
Received 8 June 1998; revised 17 December 1999; accepted 24 January 2000.
2000 International Society for Behavioral Ecology 1994; Peckarsky, 1996). The evolution of such a ﬂexible response will depend on the costs associated with gathering information about a potential predation threat (Dill, 1986; Sih,
1987). Selection will only favor ﬂexibility in predator avoidance tactics when the cost of gathering information about the
predation threat is relatively low, when the risk of predation
ﬂuctuates unpredictably, and when there are reliable cues for
detecting predation risk (Harvell, 1986, 1990). Competitor
avoidance is likely to be subject to weaker selective forces than
predator avoidance, since the cost of not avoiding a competitor will generally be lower than the cost of not avoiding a
predator, as the latter carries a risk of direct mortality.
To date most studies of predator avoidance have concentrated on aquatic and small mammal communities, while communities of large mammals have been neglected (but see
Bshary and Noe, 1997). This article addresses this gap in our
knowledge by examining avoidance of the two main predators, which are unusually also the main competitors, of the
cheetah (Acinonyx jubatus)—lions (Panthera leo) and spotted
hyenas (Crocuta crocuta). In cheetahs, offspring survival is
strongly affected by lion and hyena predation. Over 90% of
cheetah cubs die before reaching independence, predominantly due to predation (Laurenson, 1994). Adult cheetahs
may also lose their kills to these predators (Caro, 1994; Schaller, 1972), and may also be killed by lions (Durant SM, unpublished data). There are no published records of adult
cheetahs being killed by hyenas, although this possibility cannot be completely discounted. The impact of lions on cheetah
populations results in a negative relationship between cheetah
population size and lion density both across and within different protected areas (Durant et al., under review; Laurenson, 1995b).
Since cheetahs have small jaws and a light build, a mother
cannot defend her cubs or kills against lions and hyenas. However, she can reduce predation of her cubs by adapting her
behavior and adopting an avoidance strategy (Caro, 1994).
Cheetahs make use of various techniques in order to minimize
direct interactions with lions and hyenas. They reduce visual
and audio cues by killing silently by asphyxiation after a short Durant • Living with the enemy chase, hunting during the day when many of their competitors are inactive, and dragging kills immediately into cover to
avoid attracting vultures to carcasses (Caro, 1994). These behaviors minimize contact with competitors and may reduce
rates of kleptoparasitism and cub mortalities. Cheetahs can
further minimize direct contact by making use of any spatial
heterogeneity in the distributions of lions and hyenas and
seeking out ‘‘competition refuges’’—areas with low densities
of their competitors (Durant, 1998).
The risk of predation and kleptoparasitism to cheetahs is
likely to be inﬂuenced by a number of predictable factors,
therefore a ﬂexible avoidance response may have evolved in
this species. Cheetahs are more likely to be noticed by predators when they are active and hunting, when they are also
vulnerable to kleptoparasitism, and so avoidance should be
particularly marked at these times. Moreover, if cub vulnerability is the predominant cause of avoidance, avoidance
should be more marked in females than in males, since the
former are more frequently accompanied by dependent cubs
and hence might perceive predators to be a greater threat.
Female cheetahs should also show greater avoidance when
they are accompanied by dependent cubs than when solitary.
Finally, since young cubs are more vulnerable to predation
than older cubs (Caro, 1994), the strength of avoidance by a
mother should depend on the age of her cubs.
Direct observation of causes of mortality of cheetahs are
rare (Caro, 1994), making it difﬁcult to directly ascertain the
extent of the relative threat that hyenas and lions pose to
cheetahs. Moreover, observation of interactions between cheetahs and their competitors are highly variable in terms of the
distance of nearest approach of the competitors and the time
of day at which interactions occur (Caro, 1994). If more reproductively successful cheetahs are better at avoiding lions
and hyenas (Durant, 2000), then data gathered by ﬁeld observation alone would be biased towards less successful cheetahs, since these would most often be seen near other competitors. In this article I employ audio playback techniques to
examine avoidance behavior by cheetahs. Such experiments
allow the manipulation and standardization of naturally rare
I tested the following predictions in order to ascertain the
extent of avoidance tactics by cheetahs and whether cheetahs
adapt avoidance behavior according to the relative risk of predation or competition. First, I tested whether cheetahs avoid
hyenas and lions. Second, I tested whether reactions are ﬂexible, and are affected by various cues such as the hunger state
of the cheetah in the experiment and the presence of prey.
Third, I tested whether responses are dependent on sex or
reproductive state. Finally, I tested whether avoidance has a
real cost to cheetahs in terms of a reduced foraging rate.
Henceforth lions and hyenas are referred to as ‘‘competitors,’’
however it should be borne in mind that they are also known
predators of cheetah cubs and, in the case of lions, adult cheetahs.
The hypotheses were tested on data from a long-term study
population of cheetahs in the Serengeti National Park in Tanzania. The study area covers 2200 km2 in the south east of the
Serengeti (for full description see Caro, 1994). Within this
area the habitat ranges from open woodland, dissected with
rivers, in the north and west, through the long grass plains to
short grass plains in the south east. The entire area is scattered
with rocky outcrops known as ‘‘kopjes’’ which provide havens
for trees and bushes and are often the only cover available 625 out on the open plains. The climate is seasonal with a wet
season that starts in November and ends in June (Sinclair,
Cheetahs on the Serengeti plains have been studied intensively since the mid-1970s (Caro, 1994). Each cheetah can be
individually recognized by distinctive spot patterns on its face
and haunches (Caro and Durant, 1991). Cheetahs were located in the early morning and late afternoon when they are
most active, by driving to high points with good visibility and
scanning through binoculars (10 50 magniﬁcation). Active
cheetahs can be seen up to and occasionally beyond a distance
of 3 km. Upon approach each cheetah was identiﬁed and its
hunger state estimated by its belly size according to an increasing 14 point scale (Caro, 1994). This scale has proven to
be accurate and has a high reliability between different observers (Caro, 1994). Cubs were aged when they were ﬁrst
seen, when their age could be estimated to within an accuracy
of 1 month (Caro, 1994).
Audio playback techniques were used to quantify competitor
avoidance behavior, in a series of controlled experiments. Between May 1993 and September 1996 lion and hyena vocalizations were played to individual cheetahs and their reactions
recorded. Recordings of lion roars and hyena whoops were
obtained within the study area using either a Panasonic V250,
Sony TCD D3 or Sony TCD D7 digital audio tape recorder
linked to a Sennheiser MKH816 microphone. Recordings
were made from within 30 m of a single individual vocalizing
lion or hyena. Lion recordings were of adult females, while
the sex of the caller in the hyena recordings was not determined. Playbacks were played through a Sony XM 4020 Sony
ampliﬁer linked to a Sony TCD D3 digital audio tape recorder
and a Martin Audio CT2 studio monitor loudspeaker. The
volume of the playback experiments was standardized across
experiments so that calls were played at natural sound pressure levels. Experiments were conducted in the early morning
between 0630 and 0930 when cheetahs are most active and
are likely to hear lion and hyena vocalizations (East and Hofer, 1991; Kruuk, 1972; Schaller, 1972).
The speaker was placed approximately 200 m from the
cheetah and observations made from a vehicle at a distance
of 100 m. The speaker was placed upwind of the cheetah to
ensure that the sound carried with the wind. Since cheetahs
rely principally on sight and hearing, rather than smell, it is
unlikely that the fact that they were unable to smell a competitor despite being able to hear one would have inﬂuenced
their reaction. The location of the loudspeaker and initial position of the cheetah were noted using a Global Positioning
system (GPS) which was accurate to within 100m. This allowed
veriﬁcation of the initial distance between the speaker and
the cheetah (mean and standard error 232 8m). Wherever
possible the speaker was hidden from the cheetah either in
long grass, in a dip, behind a termite mound, or behind a
bush. All playbacks were conducted in open habitat on the
short or long grass plains, or at the woodland edges. Cheetahs
in this study were well habituated to vehicles. A total of 45
individuals were used in the analyses presented in this article.
Thirty-four of these were females and 11 males. In 55 out of
the 79 playbacks made to females, females were accompanied
by dependent cubs at the time of the experiment, 14, 20, and
21 of these were to females in dummy, hyena, and lion experiments respectively.
Lion playbacks consisted of a single bout of roars (McComb
et al., 1994), where hyena playbacks involved two bouts of the
same whoop recording separated by a 5 min interval. Since
hyenas often whoop when moving (Kruuk, 1972), playing two Behavioral Ecology Vol. 11 No. 6 626 Table 1
Number of each playback type played to individual cheetahs
participating in playback experiments Cheetah Gender Dummy Hyena Lion Location
Total Subjects M
12 Playbacks 0
26 Location and hunting data supplemented data obtained from
dummy playbacks in analyses of distance moved and hunting
behaviour (see Methods). bouts was intended to give the impression that a hyena is stationary in the area. Whooping is a long distance contact call
and, once away from the den, is most commonly given by
solitary hyenas (Kruuk, 1972), therefore it is unlikely that a
cheetah, upon hearing two whoops, would perceive two hyenas as being present. Playbacks alternated between two different hyena recordings and ﬁve different lion recordings. A total of 36 hyena playbacks and 37 lion playbacks were conducted (Table 1). In addition 17 ‘‘dummy’’ playbacks where
the equipment was set up but no sound was played were used
for controlled comparisons. No responses by lions or hyenas
were recorded in any of the experiments reported here. The presence of prey (Thomson’s or Grant’s gazelle) within
1 km of the cheetah at the start of the experiment was noted.
For 1 h after starting to play the ﬁrst recording, instantaneous
scans were taken every 30 s to record whether the cheetah
was looking at the speaker and whether it was moving. The
number of scans where a cheetah was looking at the speaker
or moving was divided by the total number of scans when the
cheetah was visible to give a proportion of time spent looking
or moving. The latency or time to ﬁrst movement was also
recorded, if no movement occurred, as was the case in 20%
of experiments, latency was set to 1 h, the duration of observation after experiments. In addition, for all except nine experiments, data were collected for at least 5 min before the
start of each experiment. These data were used in paired t
tests to assess changes in behavior before and after the three
types of playbacks.
During the 1-h observation period the presence of any
hunting activity was noted. A cheetah was deﬁned to stalk if
it placed its head below shoulder height, stared ﬁxedly at prey
and walked a minimum of two steps towards prey. A cheetah
was deﬁned to chase prey if it broke into a fast run after prey.
At the end of the hour the cheetah’s location was noted. Repeat lion and hyena playbacks of the same type to the same
individual were made a maximum of three times with a minimum of 1 month between repeats. Data from dummy playback experiments were further supplemented by location and
hunting data obtained on 16 occasions from cheetahs that
were ﬁrst sighted between 0630 and 0930am and subsequently
followed for an hour (Table 1). These data were used in analyses of distance and hunting behavior.
A combination of paired t tests, ANCOVAs, and generalized
linear models were used for analyses. The initial distance of
the loudspeaker, hunger state, presence of prey and time at
the start of the experiment were controlled for in all analyses
except paired t-tests and those investigating hunting behavior.
In the latter analyses these terms were initially included in the
model and were then deleted stepwise by dropping the least
signiﬁcant terms one by one until all remaining terms had
probability values less than 0.1. This process was also adopted
in analyses investigating restricted data sets, that is those estimating effects of cub presence, cub age and litter size, and
explains small differences in the degrees of freedom in reported statistics. Paired t tests implicitly controlled for all these
In paired t tests, whenever an individual was involved in
more than one experiment of the same type, responses were
averaged for that individual. All other analyses, except where
indicated, controlled for individual identity as a categorical
variable whenever it exerted a signiﬁcant effect. Terms were
judged as statistically signiﬁcant when probabilities were less
than 5%, however all probabilities of less than 10% are reported. Because a number of statistical analyses are conducted
on the same data set, Bonferroni statistical signiﬁcance is indicated next to reported p values in the text and in Tables 3–
5 by way of the following symbols: I for p
.1; * for p .05;
** for p
.01; and *** for p
.001 (Rice, 1989). It should
be noted that Bonferroni corrections give an overly conservative result (Hsu and Nelson, 1998; Samuel-Cahn, 1996).
Three types of model were used:
1. Paired t tests and ANCOVAs for analyses of proportionate
measures of cheetah activity (looking at the speaker or
moving). Proportions were converted to a normal distribution before analysis using an arcsin transformation after
ﬁrst correcting for zero values (Sokal and Rohlf, 1981).
2. ANCOVAs for analyses of the latency of ﬁrst movement and Durant • Living with the enemy 627 Table 2
Effect of playback type on transformed measures of vigilance and movement during the ﬁrst 30 min
and the second 30 min after playback experiments in comparison with pre-playback levels Playback type PreFirst 30 min
t-test Looking at speaker
Lion playbacks 0.212
Lion playbacks 0.531
t23 Second 30 min
Signiﬁcance Mean t-test Signiﬁcance 0.14
p 0.015 Means reported are from transformed data. Bonferroni corrections were not performed on these
analyses as they are partially independent of later analyses. the straight line distance from the speaker 1 h after the
start of the playback. Since the logarithm of both latency
and distance were distributed normally (goodness of ﬁt test
on the logarithm of latency and distance respectively: 52
6.54, ns and 72
8.18, ns), these variables were log
transformed before analysis.
3. A logistic regression model for analyses of the binomial
variate of the presence or absence of hunting behavior during the experiment.
All analyses were conducted using the GENSTAT 5 version 3.1
statistical package (Payne et al., 1987). Results from ANCOVAs
were tested using the t and F distributions, while those from
logistic regression models were tested using a 2 distribution
(Sokal and Rohlf, 1981).
Cheetahs did not respond differently to different recordings
of the same competitors in levels of vigilance (measured here
as the proportion of time they spent looking at the speaker),
movement patterns or hunting behavior. Cheetahs also did
not respond signiﬁcantly differently according to whether the
speaker was visible to or was hidden from the cheetah at the
start of the experiment. Finally, distance moved and hunting
behavior after dummy playbacks did not differ according to
whether the equipment was set up or not, allowing the use of
these data in supplementing experimental data in subsequent
Reactions to predators
Cheetahs changed their behavior when a sound of a lion or
hyena vocalization was played, but not when no sound was
played (Table 2). Cheetahs were signiﬁcantly more likely to
look at the loudspeaker and spent signiﬁcantly more time
moving during the ﬁrst 30 min after both lion and hyena
playbacks than during the observation period prior to the experiments (Table 2). In the second half hour of observation
vigilance was also signiﬁcantly higher after both lion and hyena playback experiments than during the pre-playback period, but cheetahs only spent signiﬁcantly more time moving
after lion experiments. There was no difference in vigilance
or movement before and after dummy experiments.
Comparing across the three different playback types over
the entire 1-h observation period after experiments, cheetahs
were signiﬁcantly more likely to look at the loudspeaker after
lion or hyena playbacks than after dummy playbacks (Figure
1, Table 3). Cheetahs looked at the loudspeaker most after lion playbacks, and least after dummy playbacks. The latency
to ﬁrst movement did not differ between playback types (Table 4), however cheetahs spent signiﬁcantly more time moving
after lion or hyena playbacks than after dummy playbacks,
moving most after a lion playback and least after a dummy
playback (Figure 1, Table 4). In addition, at the end of the
hour cheetahs were signiﬁcantly farther from the loudspeaker
after lion and hyena playbacks than after dummy playbacks
(Figure 1, Table 4), suggesting that they were moving in a
consistent direction and moved more quickly after competitor
Despite an increase in activity, cheetahs were much less likely to hunt after a lion or hyena playback than after a dummy
playback (Figure 1, Table 5). Cheetahs were also less likely to
chase and kill prey after competitor playbacks than after dummy playbacks (Table 5). Of those cheetahs which did hunt,
cheetahs were no less likely to chase prey after playbacks of
competitors (effect of playback type, 2
0.80, ns), but they
were less likely to kill prey after playbacks of lions than after
dummy or hyena playbacks (effect of playback type: coefﬁcients for dummy, hyena and lion playbacks respectively 0.0,
0.94, 9.7, 22
.004*). This suggests that once
cheetahs started hunting, they were just as likely to chase prey
after lion playbacks as after dummy playbacks, but they may
not have invested as much energy into the chase, resulting in
a lower kill rate, and indicating an avoidance of investing energetic effort into hunts once started.
Are reactions strongest to lions or hyenas?
Cheetahs were signiﬁcantly more likely to look at the loudspeaker after lion playbacks than after hyena playbacks (effect
of playback type, F1,67 5.73, p .019). However the latency
to ﬁrst movement did not differ between lion and hyena playbacks (effect of playback type, F1,68
1.71, ns), neither did
cheetahs spend more time moving (effect of playback type:
1.47, ns), nor move farther (effect of playback type,
0.74, ns). Cheetahs were also no less likely to hunt or
chase prey after a lion playback than after a hyena playback
(effect of playback type on hunting, 12
1.25, ns, and on
chasing: 12 0.70, ns), however they were less likely to make
a kill after a lion playback (effect of playback type, 12 4.05,
Factors affecting reaction
Cheetahs were not signiﬁcantly more vigilant after playback
experiments when they were hungry than when they were well
fed (Table 3). However their latency to ﬁrst movement was
signiﬁcantly greater if they were well fed (Table 4), and they 628 Behavioral Ecology Vol. 11 No. 6 Figure 1
Responses of cheetahs over 1 h
following dummy, hyena and
lion playback experiments. Upper left, proportion of time
spent looking at the loudspeaker; upper right, proportion of time spent moving; lower left, distance from loudspeaker; and lower right, the
chance of hunting. Filled bars,
dummy playback; hatched
bars, hyena playback; open
bars, lion playback. Means
were calculated across all experiments, bars indicate standard errors. spent signiﬁcantly less time moving (Table 4), moved signiﬁcantly less far (Table 4), were less likely to hunt, and were
marginally less likely to chase prey, but were not signiﬁcantly
less likely to make a kill (Table 5).
If there were no prey present at the start of the experiment,
cheetahs were marginally less vigilant (Table 3), spent significantly more time moving (Table 4), but were only marginally
signiﬁcantly farther from the loudspeaker at the end of the
experiment than if prey were present (Table 4). They were
no more likely to hunt if prey were initially present than if
prey were absent (Table 5). This was partly because either
prey would move into or cheetahs move out of the area during
the 1-h observation, and partly because cheetahs often hunted
prey, such as hares or gazelle fawns, which were initially not
visible to the observer.
Other variables such as the time of day and the initial distance of the loudspeaker from the cheetah had some effect
on responses. Cheetahs were less vigilant and had a greater
latency to ﬁrst movement when experiments were started later
in the day (Tables 3 and 4), and were farther from the loudspeaker at the end of the experiment if they were farther from
it at the beginning (Table 4). Individual identity did not have
a signiﬁcant effect on vigilance or movement patterns, but did
affect hunting behavior (Tables 3–5). Sex and reproductive status
Female cheetahs did not react more strongly to competitor
playbacks than male cheetahs. Although females were significantly more likely to look at the loudspeaker than males (effect of sex, t71
.011I), this effect was probably a
result of a generally higher level of vigilance for females compared with males, since there was no signiﬁcant interaction
between sex and playback type (effect of interaction between
0.09, ns). Females were not
sex and playback type: F1,71
signiﬁcantly more likely to move than males (effect of sex, t72
0.09, ns, interaction between sex and playback type, F1,72
1.24, ns), neither did they move farther (effect of sex, t86
1.29, ns, interaction between sex and playback type: F2,88
1.61, ns) and were no more likely to hunt, chase or kill after
playback experiments (effect of sex for hunt, chase and kill
0.02, ns, 12
0.01, ns and 12 0.00, ns,
interaction between sex and playback type, 22 0.53, ns, 22
0.60, ns and 22
0.45, ns). It should be noted that only
two males were involved in hyena and only one in dummy
playback experiments (but location and hunting data were
available for a further four males—see Table 1), while eight
males were involved in lion experiments. Therefore these
analyses were largely dependent on a difference between the
sexes in response to lion rather than hyena playback experiments. Durant • Living with the enemy 629 Table 3
Factors affecting vigilance during the full hour of observation after
Lion Statistic 0.00
0.22 Hunger state
Proportion of time
Distance from speaker
at start of
Presence of prey
— F2,83 Signiﬁcance 11.37 p 0.001*** t81
p 0.033 t81 4 2.03 p 0.046 t81
ns F39,81 Table 4
Factors affecting movement and distance moved during the full
hour of observation after playback experiments 0.096 The proportion of time spent looking at the loudspeaker was ﬁtted
as the dependent variate in an ANCOVA after an arcsin
transformation. The model explained 31.4% of the total variance.
Probability symbol after Bonferroni corrections: I for p
.1; * for p
.05; ** for p
.01; and *** for p
.001 (Rice, 1989).
a Not included in ﬁnal model. Female cheetahs did not react more strongly to competitor
playbacks if they had cubs than if they were solitary. Solitary
females were marginally less likely to look at the loudspeaker
after competitor playbacks than females with cubs (effect of
cub presence, t67
.086), but this effect was independent of playback type (effect of interaction between cub
presence and playback type, F1,66
0.15, ns) and therefore
probably reﬂected higher overall levels of vigilance for cheetah mothers (Laurenson, 1995a). Females with cubs were not
signiﬁcantly more likely to move after experiments than were
females without cubs (effect of cub presence, t65
interaction between cub presence and playback type, F1,64
0.40, ns), neither did they move farther (effect of cub presence, t76
0.58, ns, interaction between cub presence and
playback type, F2,74
0.83, ns). Furthermore, females were
not less likely to hunt, chase or kill prey after playback experiments if they had cubs (effect of cub presence, 12
ns; 12 0.09, ns and 12 0.04, ns, and interaction between
cub presence and playback type: 22 1.78, ns; 22 2.39, ns
0.32, ns on hunts, chases, and kills respectively).
Reactions of females with cubs did not depend on the cubs’
age or number. Females did not look at the speaker more
often, spend more time moving, move farther or hunt more
frequently as their litter size increased or the age of their cubs
decreased, regardless of playback type (Table 6).
These results demonstrate that cheetahs actively moved away
from lions and hyenas once they perceived them to be present
through an auditory stimulus. Perception of the presence of
a competitor through playbacks also had a measurable impact
on foraging rates of cheetahs, since cheetahs were much less
likely to hunt after competitor playbacks than after dummy
playbacks, resulting in a lower kill rate. Cheetahs appeared to
perceive lions as a greater threat than hyenas since they were
signiﬁcantly more vigilant and were less likely to make a kill
after lion playbacks than after hyena playbacks, although they
did not move signiﬁcantly farther. These differences are likely
to be particularly marked given that two calls were played dur- Coefﬁcient
Latency to ﬁrst movementa
Distance from speaker
at start of
Presence of prey
Distance from speaker
at start of
Presence of prey
— Statistic Signiﬁcance F2,101 0.81 ns t99
ns F2,85 4.54 p 0.013I t83
ns 0.001*** t83
0.029 F40,83 2.06
1.15 F2,101 6.51 p 0.002* t99
ns 0.001** t99
0.05* c Distance from speaker
Distance from speaker
at start of
Presence of prey
— F44,99 Probability symbol after Bonferroni corrections: I for p
.1; * for p
.05; ** for p
.01; and *** for p
.001 (Rice, 1989).
a The time to ﬁrst movement was ﬁtted as the dependent variate to
an ANCOVA after a logarithmic transformation. The model
explained 10.4% of the total variance.
b The proportion of time spent moving was ﬁtted as the dependent
variate to an ANCOVA after an arcsin transformation. The model
explained 21.6% of the total variance.
c The distance from the loudspeaker at the end of the 1-h
observation period was ﬁtted as the dependent variate after a log
transformation to an ANCOVA. The model explained 24.3% of the
d Not included in ﬁnal model. ing hyena experiments while only one call was played during
It could be argued that the results obtained in this study
might have been obtained if a cheetah avoids any loud sound
regardless of whether it is the sound of a competitor. However,
although some might argue that the sounds of ungulates such
as wildebeest might be a more appropriate control than that
used here, it might have also been argued that cheetahs may
have been attracted to these sounds, and hence any differences between controls and predator playbacks were driven
by prey attraction rather than competitor avoidance. In fact, Behavioral Ecology Vol. 11 No. 6 630 Table 5
Factors affecting hunting behaviour during the full hour of
observation after playback experiments
Stalking or chasing preya
Distance from speaker
at start of
Presence of prey
Distance from speaker
at start of
Presence of prey
Distance from speaker
at start of
Presence of prey Statistic Signiﬁcance 2
2 14.65 p 0.001* 2
ns 0.021 0.20
p 0.009I 2
2 6.01 p 0.050 2
2 11.48 p 2
44 0.003* The binomial variate of whether a cheetah exhibited hunting
behaviour during the 1-h observation period after playback
experiments was ﬁtted as the dependent variate in a logistic
regression model. The ﬁnal model includes all factors which
affected the model with p
.1 and leaves out those terms which do
not. Probability symbols after Bonferroni corrections: I for p
.05 (Rice, 1989).
a The ﬁnal model explained a deviance of 85.62 with 47 df out of a
total deviance of 115.80 with 105 df.
b The ﬁnal model explained a deviance of 69.47 with 4 df out of a
total deviance of 93.34 with 105 df.
c The ﬁnal model explained a deviance of 11.48 with 2 df out of a
total deviance of 66.24 with 105 df. There was insufﬁcient variation
to be able to ﬁt individual identity as a factor in this analysis. a prior study is necessary to ﬁnd a suitable control, and this
was not feasible within the constraints of time and resources
in this study. However while it is not possible to completely
discount the loud noise hypothesis, it is most likely that the
cheetahs were responding to the perceived presence of a competitor, especially given that cheetahs responded differently in
terms of vigilance and kill rates to lion compared with hyena
The stronger reaction of cheetahs to playbacks of lion calls
than to hyena calls could have one of two explanations. First,
cheetahs may perceive lions to be a greater threat than hye- nas. Second, cheetah responses may vary because lions and
hyenas have different hunting strategies, as lions are stalking
and hyenas coursing predators. Unfortunately it is difﬁcult to
distinguish between these two explanations. Previous studies
have shown that prey are more vigilant and have a longer
ﬂight distance (i.e., retreat further) to stalking predators compared with coursing predators (FitzGibbon and Lazarus,
1995). However whether this is because the stalking predator
is a greater threat than a coursing predator, or is due to a
different response pattern to alternative predator tactics is difﬁcult to determine.
Were cheetahs responding to a perceived predation or a
competition threat? Cheetahs did not adapt their behavior to
their reproductive status. This result is surprising, given the
high cub mortality due to predation (Laurenson, 1994), and
could result from one of three explanations. First, cheetahs
may be more at risk when they have dependent cubs but, if
the cost of avoidance is low, selection may be too weak for
cheetahs to adapt their behaviors to the relative predation
risk. Second, cheetahs may themselves be at risk with or without dependent cubs, and by responding to predators they are
avoiding a real threat to themselves as well as their cubs.
Third, cheetahs may continue to respond to predators even
when they have no cubs, because they are still vulnerable to
kleptoparastism. Since this study showed that cheetahs suffered a measurable reduction in foraging intake when they
perceived other competitors to be present, the ﬁrst explanation is unlikely. Distinguishing between the last two explanations depends on determining the relative threat that lions
and hyenas pose to adult cheetahs.
While adult cheetahs have been known to be killed by lions
within the Serengeti (Durant SM, unpublished data), there
are no records of predation by hyenas within this ecosystem.
Therefore, while there might be no reason for cheetahs to
adapt their behavioral responses to lions according to the
presence of dependent cubs, there may be a strong reason for
cheetahs to adapt their responses to hyenas, yet this pattern
was not reﬂected in the results reported here. Alternatively,
since cheetahs lose kills to both lions and hyenas (Caro,
1994), they are vulnerable to kleptoparastism from both these
competitors. Hence there are probably indirect costs to not
responding to the presence of a competitor even when an
individual’s survival is not directly threatened. Given this, it
seems likely that cheetahs would beneﬁt from avoiding both
these competitors regardless of their reproductive status,
through an avoidance strategy. Further evidence for this hypothesis is provided by the results obtained in this study,
where cheetahs moved farther from competitor playback experiments when they were hungry than when they were well
fed and were least vulnerable to kleptoparasitism.
A number of other studies have demonstrated behavioral
avoidance of predators (Chivers and Smith, 1995; Dickman
and Doncaster, 1984; Flowers and Graves, 1997; Hileman and
Brodie, 1994; Holomuzki and Short, 1988; Kiesecker et al.,
1996; Li and Li, 1979; Loose and Dawidowicz, 1994; Main,
1987; Peckarsky, 1996; Semlitsch and Reyer, 1992; Ward et. al.,
1997; Werner, 1991), however, in most of these examples
avoidance is due to a visual, scent or chemical stimulus. There
are few examples of avoidance due to an auditory stimulus
(but see Bshary and Noe, 1997). Avoidance of infanticide, a
behavior that is directly related to avoidance of cub predation,
has been documented for lions, where female lions are able
to recognize strange male lions that pose an infanticidal
threat to her cubs, and move away from playbacks of their
calls (McComb et al., 1992). Examples of avoidance of kleptoparastism are rare.
Avoidance through active movement away from a potential
predation or competitive threat is only likely to be worthwhile Durant • Living with the enemy 631 Table 6
Effect of playback type, cub age and litter size on measures of vigilance, patterns of movement and hunting behaviour during the full hour
of observation after playback experiments
Effects Statistic Looking at speaker
Distance from speaker
0.00 Cub age playback type Litter size Litter size playback type Signiﬁcance Statistic Signiﬁcance Statistic Signiﬁcance Statistic Signiﬁcance ns
2 2 0.11
2 2 0.02 t50
1.10 Statistics are reported from the ﬁnal model, which includes all factors which affected the model with p
0.1 and leaves out those terms
which did not. F statistics were obtained from ANCOVAs, while 2 statistics were obtained from logistic regression models. Dash indicates
insufﬁcient variation for inclusion in the model. if predators or competitors are rare or aggregated (Colegrave,
1997). Thus prey, by moving away from a predator once seen,
are more likely than not to be moving to an area where predators are at lower densities. This, in effect, means the prey is
making use of a spatio-temporal refuge, in a similar way to a
physical refuge ﬁxed in time and space. Both lions and hyenas
have an aggregated distribution (Durant, 1998), and so a
cheetah which moves away from these competitors once seen
has a better than average chance of moving into an area with
lower competitor densities.
Predator avoidance is also only likely to be a useful strategy
if anti-predator tactics are ineffective, since individuals will not
choose to bear the energetic costs of avoidance when they are
unlikely to be at any real risk. Anti-predator tactics are more
likely to be employed by social species, since they are more
effective when a large number of individuals are involved and
individual risks of predation can be spread across the group
(Stanford, 1995). In addition, social species are more likely to
be the target of predator attacks because of the increased conspicuousness of large groups (Cowlishaw, 1997; Lima and Dill,
1990). These principles are also valid when an individual faces
a kleptoparastism threat rather than a predation threat. Cheetahs are generally solitary or in small groups and are unlikely
to beneﬁt from anti-predator behavior, since they are much
smaller than lions and hyenas and are nearly always displaced
by these competitors at kills (Caro, 1994; Durant SM, personal
If a cheetah is spotted by a lion or hyena it will often be
approached (Durant SM personal observation). At this point
a cheetah can stay put and risk attack, or ﬂee and risk ﬂeeing
into the path of another competitor, since both hyenas and
lions live in social groups. Both these options are risky, and
so it is likely to be better for a cheetah to move away before
the threat becomes real, preferably before it is sighted by other competitors. In this study cheetahs generally moved a few
hundred m from competitors (Figure 1), distances easily covered by lions and hyenas, but which make the difference between detection or no detection of a resting cheetah by competitors. However, the activity associated with hunting is more
easily detectable over these distances, which might explain the
reduction in hunting activity after competitor playbacks.
Does avoidance have implications for the distribution of
cheetahs within the ecosystem? A previous study has shown
that whenever cheetahs are found near high densities of lions
or hyenas they are less likely to be hunting and more likely
to be moving than at low densities (Durant, 1998). Furthermore, both lions and hyenas are found near high densities of
gazelle, the main prey of cheetahs on the Serengeti plains
(Caro, 1994; FitzGibbon, 1990a), where cheetahs are more
frequently found near low densities of gazelle, while avoiding areas with no gazelle at all (Durant, 1998). By avoiding competitors, cheetahs might move away from areas with high prey
densities to areas of lower prey densities, where they are able
to survive because of their higher hunting success on small
groups or isolated individuals (FitzGibbon, 1990b). The mobility of cheetahs and their ability to avoid direct competition
in an ever-changing landscape of competitors and prey may
be the key to their coexistence with lions and hyenas.
I thank Tanzanian National Parks and the Tanzania Wildlife Research
Institute (TAWIRI) for permission to conduct this study. This research
was funded by the National Geographic Society, the Royal Society,
Frankfurt Zoological Society, and the Institute of Zoology. I also thank
M. Borner, S. and S. Tham, J. Ole Kwai, H. van Lawick and his team,
and all my colleagues at TAWIRI for logistical support. K. McComb
provided lion roar playback tapes, inspiration and help with the experimental design. G. Cowlishaw, T. Jones, R. Pettifor, S. Semple, and
two anonymous referees have all improved this manuscript. REFERENCES
Brodie EDJ, Formanowicz DRJ, Brodie EDI, 1991. Predator avoidance
and antipredator mechanisms: distinct pathways to survival. Ethol
Ecol Evol 3:73–77.
Bshary R, Noe R, 1997. Anti-predation behaviour of red colobus monkeys in the presence of chimpanzees. Behav Ecol Sociobiol 41:321–
Caro TM, 1994. Cheetahs of the Serengeti Plains: group living in an
asocial species. Chicago: University of Chicago Press.
Caro TM, Durant SM, 1991. Use of quantitative analyses of pelage
characteristics to reveal family resemblances in genetically monomorphic cheetahs. J Hered 82:8–14.
Chivers DP, Smith RJF, 1995. Chemical recognition of risky habitats is
culturally transmitted among fathead minnows, Pimephales promelas
(Osteichthyes, Cyprinidae). Ethology 99:286–296.
Colegrave N, 1997. Can a patchy population structure affect the evolution of competition strategies? Evolution 51:483–492.
Cowlishaw G, 1997. Trade-offs between foraging and predation risk
determine habitat use in a desert baboon population. Anim Behav
Dickman CR, Doncaster CP, 1984. Responses of small mammals to
red fox (Vulpes vulpes) odour. J Zool Lond 204:521–531.
Dill LM, 1986. Animal decision-making and its ecological consequences: the future of aquatic ecology and behaviour. Can J Zool 65:803–
Durant SM, 1998. Competition refuges and coexistence: An example
from Serengeti carnivores. J Anim Ecol 67:370–386.
Durant SM, 2000. Predator avoidance, breeding experience and reproductive success in endangered cheetahs (Acinonyx jubatus).
Anim Behav 60:121–130.
Durant SM, Kelly MJ, Caro TM, under review. Factors affecting life
and death in Serengeti cheetahs: Environment, age and sociality.
Behav Ecol. 632 East ML, Hofer H, 1991. Loud-calling in a female dominated mammalian society. I. Structure and composition of whooping bouts of
spotted hyaenas, Crocuta crocuta. Anim Behav 42:637–649.
FitzGibbon CD, 1990a. Mixed species grouping in Thomson’s and
Grant’s gazelles: the antipredator beneﬁts. Anim Behav 39:1116–
FitzGibbon CD, 1990b. Why do hunting cheetahs prefer male gazelles? Anim Behav 40:837–845.
FitzGibbon CD, Lazarus J, 1995. Antipredator behavior of Serengeti
ungulates: Individual differences and population consequences. In:
Serengeti II: Dynamics, management and conservation of an ecosystem (Sinclair ARE, Arcese P, eds). Chicago: University of Chicago
Flowers MA, Graves BM, 1997. Juvenile toads avoid chemical cues
from snake predators. Anim Behav 53:641–646.
Harvell CD, 1986. The ecology and evolution of inducible defenses
in a marine bryozoan: cues, costs and consequences. Am Nat 128:
Harvell CD, 1990. The ecology and evolution of inducible defenses.
Q Rev Biol 65:323–340.
Hileman KS, Brodie ED, 1994. Survival strategies of the salamander
Desmognathus ochrophaeus: interaction of predator-avoidance and
anti-predator mechanisms. Anim Behav 47:1–6.
Holomuzki JR, Short TM, 1988. Habitat use and ﬁsh avoidance behaviors by the stream-dwelling isopod Lirceus fontinalis. Oikos 52:
Hsu JC, Nelson B, 1998. Multiple comparisons in the general linear
model. J Comp Graph Stat 7:23–41.
Kennedy M, Shave CR, Spencer HG, Gray RD, 1994. Quantifying the
effect of predation risk on foraging bullies no need to assume an
ifd. Ecology 75:2220–2226.
Kiesecker JM, Chivers DP, Blaustein AR, 1996. The use of chemical
cues in predator recognition by western toad tadpoles. Anim Behav
Kruuk H, 1972. The spotted hyaena: a study of predation and social
behavior. Chicago: University of Chicago.
Laurenson MK, 1994. High juvenile mortality in cheetahs (Acinonyx
jubatus) and its consequences for maternal care. J Zool Lond 234:
Laurenson MK, 1995a. Behavioural costs and constraints of lactation
in free-living cheetahs. Anim Behav 50:815–826.
Laurenson MK, 1995b. Implications of high offspring mortality for
cheetah population dynamics. In: Serengeti II: Dynamics, management and conservation of an ecosystem (Sinclair ARE, Arcese P,
eds). Chicago: University of Chicago Press; 385–399.
Li JL, Li HW, 1979. Species-speciﬁc factors affecting predator-prey
interactions of the copepod Acanthocyclops vernalis with its natural
prey. Limnol Oceanogr 24:613–626.
Lima SL, Dill LM, 1990. Behavioral decisions made under the risk of
predation: a review and prospectus. Can J Zool 68:619–640. Behavioral Ecology Vol. 11 No. 6 Loose CJ, Dawidowicz P, 1994. Trade-offs in diel vertical migration by
zooplankton: the costs of predator avoidance. Ecology 75:2255–
Main KL, 1987. Predator avoidance in seagrass meadows: prey behavior, microhabitat selection, and cryptic coloration. Ecology 68:170–
McComb K, Packer C, Pusey A, 1994. Roaring and numerical assessment in contests between groups of female lions, Panthera leo.
Anim Behav 47:379–387.
McComb K, Pusey A, Packer C, Grinnell J, 1992. Female lions can
identify potentially infanticidal males from their roars. Proc R Soc,
Lond B 252:59–64.
McIntosh AR, Townsend CR, 1994. Interpopulation variation in mayﬂy antipredator tactics: differential effects of contrasting predatory
ﬁsh. Ecology 75:2078–2090.
Payne RW, Lane PW, Ainsley AE, Bicknell KE, Digby PGN, Harding
SA, Lech PK, Simpson HR, Todd AD, Verrier PJ, White RP, Gower
JC, Tunnicliffe Wilson G, Paterson LJ, 1987. GENSTAT 5 Reference
manual. Oxford: Oxford University Press.
Peckarsky BL, 1996. Alternative predator avoidance syndromes of
stream-dwelling mayﬂy larvae. Ecology 77:1888–1905.
Rice WR, 1989. Analyzing tables of statistical tests. Evolution 43:223–
Samuel-Cahn E, 1996. Is the Simes improved Bonferroni procedure
conservative? Biometrika 83:928–933.
Schaller GB, 1972. The Serengeti Lion: A study of predator-prey relations. Chicago: University of Chicago Press.
Semlitsch RD, Reyer H-U, 1992. Modiﬁcation of anti-predator behaviour in tadpoles environmental conditioning. J Anim Ecol 61:353–
Sih A, 1987. Predators and prey lifestyles: an evolutionary and ecological overview. In: Predation: direct and indirect impacts on
aquatic communities (Kerfoot CW, Sih A, eds). Hanover: University
Press of New England; 203–224.
Sih A, 1992. Prey uncertainty and the balancing of antipredator and
feeding needs. Am Nat 139:1052–1069.
Sinclair ARE, 1979. The Serengeti environment. In: Serengeti: Dynamics of an ecosystem (Sinclair ARE, Norton-Grifﬁths M, eds).
Chicago: University of Chicago Press; 31–45.
Sokal RR, Rohlf FJ, 1981. Biometry. New York: W. H. Freeman and
Stanford CB, 1995. The inﬂuence of chimpanzee predation on group
size and anti-predator behavior in red colobus monkeys. Anim Behav 49:577–587.
Turner AM, Mittelbach GG, 1990. Predator avoidance and community
structure interactions among piscivores, planktivores, and plankton.
Ward JF, Macdonald DW, Doncaster CP, 1997. Responses of foraging
hedgehogs to badger odour. Anim Behav 53:709–720.
Werner EE, 1991. Nonlethal effects of a predator on competitive interactions between two anuran larvae. Ecology 72:1709–1720. ...
View Full Document
- Fall '08