Content uploaded by Michael Heithaus
Author content
All content in this area was uploaded by Michael Heithaus
Content may be subject to copyright.
Oecologia (2007) 153:1031–1040
DOI 10.1007/s00442-007-0802-3
123
BEHAVIORAL ECOLOGY
Fear factor: do dugongs (Dugong dugon) trade food
for safety from tiger sharks (Galeocerdo cuvier)?
Aaron J. Wirsing · Michael R. Heithaus ·
Lawrence M. Dill
Received: 7 November 2006 / Accepted: 18 June 2007 / Published online: 17 July 2007
© Springer-Verlag 2007
Abstract Predators can inXuence plants indirectly by
altering spatial patterns of herbivory, so studies assessing
the relationship between perceived predation risk and habi-
tat use by herbivores may improve our understanding of
community organization. In marine systems, the eVects of
predation danger on space use by large herbivores have
received little attention, despite the possibility that preda-
tor-mediated alterations in patterns of grazing by these ani-
mals inXuence benthic community structure. We evaluated
the relationship between habitat use by foraging dugongs
(Dugong dugon) and the threat of tiger shark predation in
an Australian embayment (Shark Bay) between 1997 and
2004. Dugong densities were quantiWed in shallow (puta-
tively dangerous) and deep (putatively safe) habitats (seven
survey zones allocated to each habitat), and predation haz-
ard was indexed using catch rates of tiger sharks (Gale-
ocerdo cuvier); seagrass volume provided a measure of
food biomass within each zone. Overall, dugongs selected
shallow habitats, where their food is concentrated. Foragers
used shallow and deep habitats in proportion to food avail-
ability (input matching) when large tiger sharks were scarce
and overused deep habitats when sharks were common.
Furthermore, strong synchrony existed between daily mea-
sures of shark abundance and the extent to which deep hab-
itats were overused. Thus, dugongs appear to adaptively
manage their risk of death by allocating time to safe but
impoverished foraging patches in proportion to the likeli-
hood of encountering predators in proWtable but more dan-
gerous areas. This apparent food-safety trade-oV has
important implications for seagrass community structure in
Shark Bay, as it may result in marked temporal variability
in grazing pressure.
Keywords Community structure · Foraging ·
Herbivory · Predator intimidation · Shark Bay
Introduction
Foraging by herbivores can lead to marked changes in plant
biomass, distribution, and diversity (Crawley 1983). Thus,
predators may aVect plants indirectly by altering spatial
patterns of herbivory (Abrams 1995; Schmitz 2003; and see
review by Schmitz et al. 2004). Predation can inXuence the
distribution of herbivores lethally if individuals are
removed diVerentially across space, or sublethally (noncon-
sumptively) if a positive correlation between resources and
danger prompts individuals to trade access to proWtable
foraging patches for safety (Sih 1980; McNamara and
Communicated by Peter Peterson.
Electronic supplementary material The online version of this
article (doi:10.1007/s00442-007-0802-3) contains supplementary
material, which is available to authorized users.
A. J. Wirsing · L. M. Dill
Behavioural Ecology Research Group,
Department of Biological Sciences,
Simon Fraser University, Burnaby,
BC, Canada, V5A 1S6
A. J. Wirsing
Honorary Research Associate, School of Animal Biology,
University of Western Australia, 35 Stirling Highway,
Crawley, WA 6009, Australia
A. J. Wirsing (&) · M. R. Heithaus
Department of Biological Sciences, Marine Biology Program,
Florida International University, Biscayne Bay Campus MSB,
3000 NE 151 St, North Miami, FL 33181, USA
e-mail: wirsinga@Wu.edu
1032 Oecologia (2007) 153:1031–1040
123
Houston 1987; Peacor and Werner 2000). An increasing
amount of evidence suggests that changes in herbivore
space use driven by sublethal eVects of predators (i.e.,
intimidation) are especially likely to trigger trophic cas-
cades within communities (e.g., Peacor and Werner 2001;
Preisser et al. 2005). Consequently, studies focused on
these changes should improve our understanding of ecosys-
tem organization (Werner 1998). While the link between
predator intimidation and space use by terrestrial herbi-
vores is established (Lima and Dill 1990; Lima 1998;
Brown et al. 1999; Laundré et al. 2001; Verdolin 2006), the
extent to which large, marine herbivores (e.g., sea turtles,
sirenians) trade food for safety has received little attention
(Dill et al. 2003).
Exchanges of food for safety can be explored using con-
sumer time allocation patterns and ideal free distribution
(IFD) theory (Fretwell and Lucas 1970). According to IFD
theory, consumer populations should be distributed across
habitats in proportion to food supply. Thus, forager densi-
ties in all habitats should be equal after dividing by the food
supply in each habitat. If predation risk inXuences habitat
selection, however, then fewer consumers should be found
in dangerous habitats than predicted by food availability,
causing food-corrected forager densities in these habitats to
fall below those in safer ones (van Baalen and Sabelis
1993). Therefore, spatial diVerences between forager densi-
ties that appear when antipredator defense is induced and
are not explained by variance in food supply can be used to
quantify the food foragers sacriWce by avoiding dangerous
habitats (i.e., “hazardous duty pay”, Brown and Kotler
2004).
Using this theoretical framework, we asked whether per-
ceived predation risk from tiger sharks (Galeocerdo cuvier)
aVects habitat use decisions of dugongs (Dugong dugon) in
Shark Bay, Western Australia. Dugongs are seagrass spe-
cialists, and must forage for much of the day to oVset meta-
bolic costs (Marsh et al. 1982). In Shark Bay, seagrass
grows primarily in shallow habitats (·4.5 m in depth;
Walker et al. 1988; Travers and Potter 2002; Heithaus
2004a), but these habitats also are used preferentially by
tiger sharks (Heithaus et al. 2002), the dugong’s major local
predator (Heithaus 2001; Simpfendorfer et al. 2001). Con-
sequently, dugongs choosing between deep and shallow
foraging patches may face a trade-oV between energy
acquisition rate and the risk of encountering and being
killed by sharks. Tiger shark densities vary seasonally in
Shark Bay, however, peaking in the austral warm season
(January–February) and reaching a nadir during the cold
season (July) (Heithaus 2001; Wirsing et al. 2006), so the
magnitude of this trade-oV
should vary temporally.
Accordingly, we tested whether dugongs minimize their
risk of mortality by using safer, but energy-poor, foraging
habitats to a greater extent as overall predator abundance
increases (predation risk sensitivity hypothesis). This
hypothesis predicts that, after adjustment for spatial and
temporal diVerences in food supply, the degree of inequal-
ity between foraging dugong densities in safe and hazard-
ous habitats should be inXuenced by predator numbers.
Under the assumption that shallow habitat is relatively dan-
gerous, then, foragers should distribute themselves in pro-
portion to food supply when sharks are scarce and use deep
habitats more often than predicted by food availability
when sharks are most abundant. If predation danger does
not inXuence habitat use by foraging dugongs, then we
should observe proportional forager densities in the two
habitat types (i.e., input matching) throughout the year
(food quantity hypothesis). To test our understanding of
spatial variation in predation risk, we also asked whether
dugongs select deep habitats while resting, when vulnera-
bility to predation presumably is elevated and there is no
beneWt to using dangerous areas (Lima and Dill 1990).
Materials and methods
Study site
This study was undertaken in the Eastern Gulf of Shark Bay
(»25°45⬘S, 113°44⬘E; Fig. 1). Listed as a World Heritage
Area in 1991, Shark Bay features a mosaic of embayment
plains (6.0–15 m deep), swift-current channels (6.0–12 m),
shallow banks (<4.5 m), and sandy Xats (generally <2.5 m
and intertidal). Approximately one-third of its area
(»4,000 km
2
) is covered by seagrass meadows (Walker
et al. 1988), which support 10,000–14,000 dugongs (Marsh
et al. 1994; Preen et al. 1997; Gales et al. 2004).
Fig. 1 This study was conducted in the Eastern Gulf of Shark Bay
(»25°45⬘S, 113°44⬘E), Western Australia, between 1997 and 2004.
Survey zones (i.e., 400-m sighting belts; n = 14) were stratiWed across
shallow (2.5–4.5 m in depth) and deep (>6 m in depth) habitats
throughout the 160 km
2
study area. Land is portrayed in black, while
shades of gray depict water depth in categories ranging from <2 to
>9 m deep
Oecologia (2007) 153:1031–1040 1033
123
Food availability
We quantiWed food availability (biomass) in 14 survey
zones, allocated evenly to deep (¸6 m in depth) and shal-
low (2.5–4.5 m deep) habitats (Fig. 1c). The survey zones
did not incorporate intertidal sandy Xats. Each zone con-
sisted of a central transect line surrounded by a 200-m
buVer, yielding an average sampling area of 141.27 ha (SD
26.17 ha). Sampling stations were positioned at 200 m
intervals along the transect line, as well as along parallel
lines 100 and 200 m to either side (75–120 stations per sur-
vey zone). At each station, seagrass species coverage and
composition were estimated within a 1 m
2
quadrat by a
diver. Seagrass height (cm) was also measured within the
quadrat; though within-quadrat height variability was rela-
tively consistent, the point of measurement was selected
randomly to minimize bias. Sampling occurred during the
late austral winter of 2003 (August–September), when her-
bivore densities were relatively low, in order to minimize
the confounding inXuence of herbivory and shark hazard to
divers.
Food biomass was expressed as above-ground seagrass
volume (area covered £ height; m
3
); measures from sam-
pling points were pooled to generate overall values for each
survey zone. Volume measures for survey zones consisted
of two seagrass species: Amphibolis antarctica
(mean = 82.7% of total volume) and Posidonia australis
(mean = 17.3% of total volume). While A. antarctica is a
food source for dugongs in the study area during both cold
(Anderson 1986) and warm months (A. Wirsing, personal
observation), the value of P. australis as a dietary item for
dugongs is not known. The removal of this species from
consideration, however, does not aVect the results. We may
also have incurred bias by failing to measure rhizomal (i.e.,
subsurface) biomass (de Iongh et al. 1995). Rhizomes of
the dominant species (A. antarctica) are not available to
dugongs as food (Anderson 1986), however, so such bias
was likely modest. Finally, we were unable to quantify the
biomass of tropical seagrass species since they are scarce
during the winter in our study area. Nevertheless, we
assumed that our biomass measure adequately captured
diVerences in food availability between deep and shallow
habitats because tropical species are always spatially con-
cordant with A. antarctica and P. australis (Walker et al.
1988; D. Burkholder, unpublished data).
Dugong density and habitat use
We assessed patterns of dugong abundance using transect
passes through the survey zones from 1997 to 2004 (2000–
2001 excluded). To maintain sampling consistency, tran-
sect eVort was allocated evenly across days (n =218;
mean = 7.52 passes day
¡1
, SD 2.42), months (mean =
44.17 § 18.24 passes month
¡1
; note, however, that surveys
were not conducted during the months of November to Jan-
uary), and habitats (shallow survey zones: mean = 3.79
§ 1.64 passes day
¡1
; deep survey zones: 3.86 § 1.65
passes day
¡1
). Survey zones within shallow and deep habi-
tats were selected to ensure that all portions of the study
area were monitored evenly, and no survey zone was vis-
ited more than once per day. Similarly, to reduce the eVects
of tidal and diel variation, the order, and direction in which
transects were driven each day were haphazard. Transects
were conducted only in Beaufort wind conditions ·2 to
minimize sighting bias caused by poor visibility in bad
weather.
Transect passes were conducted using a small (4.5 m)
vessel driven at 6–9 km h
¡1
. When dugongs were sighted at
the surface within a survey zone (i.e., a 400-m sighting
belt) before being passed by the boat, we recorded their
position with a GPS and the water depth and substrate at
their location. Between 2002 and 2004 (n = 114 days), we
also determined the behavioral state (foraging, resting, and
traveling) of all dugongs sighted at the surface based on
direct observation and diagnostic surface behavior (Ander-
son 1986; Chilvers et al. 2004). Individual dugongs were
distinguished using scarring patterns (Anderson 1995) and
counted only once per day; individuals were rarely resam-
pled during each year of the investigation.
Predator abundance
Our assumption that tiger sharks are active predators of
dugongs is based on several pieces of evidence. One, the
frequency with which dugong remains occur in the stom-
achs of tiger sharks appears to be too high to be explained
by scavenging alone: dugong tissue was found in 15 of 84
tiger sharks caught along the northwestern coast of Austra-
lia (Simpfendorfer et al. 2001) and in all six large sharks
(>3.0 m) caught in our study area from which complete
stomach contents were collected (Heithaus 2001; unpub-
lished data). Two, some adult dugongs bear scars from
unsuccessful attacks by sharks (Anderson 1995; A. Wir-
sing, personal observation). Three, tiger sharks have been
sighted harassing, attacking, killing, and consuming dug-
ongs of various ages, including adults, and healthy individ-
uals (A. Wirsing, personal observation; D. Charles,
Western Australia Department of Conservation and Land
Management, personal communication). Four, when sharks
are abundant, microhabitat shifts by dugongs (Wirsing et al.
2007) mirror those of other species subject to tiger shark
predation (bottlenose dolphins, Tursiops sp., Heithaus and
Dill 2006; green turtles, Chelonia mydas, Heithaus et al.
2007). Importantly, although other predators have been
observed attacking dugongs in the Bay’s Western Gulf
(killer whales, Orcinus orca, Anderson and Prince 1985),
1034 Oecologia (2007) 153:1031–1040
123
tiger sharks are the dugong’s only potential predator in the
study area.
The extent of tiger shark predation on dugongs in Shark
Bay is unclear, and kill rates are unlikely to be high. Tiger
sharks nevertheless have the potential to inXuence dugong
behavior. Indeed, long-lived species like the dugong should
invest in anti-predator behavior even if the risk of being
killed is low (Warner 1998). For example, bottlenose dol-
phins (Tursiops sp.) are almost never found in the stomachs
of tiger sharks in Western Australia (Simpfendorfer et al.
2001), but shift their habitat use at multiple spatial scales in
response to tiger shark predation risk (Heithaus and Dill
2002, 2006).
Catch rates provide a reliable index of tiger shark abun-
dance in Shark Bay (Heithaus 2001; Wirsing et al. 2006).
Sharks were caught on drumlines equipped with a single
hook, baited primarily with Australian salmon (Arripis trut-
taceus), and deployed at dawn in six Wshing zones through-
out the study area; all sharks caught throughout the day
were brought next to the vessel, measured for total length
(TL), tagged, and released (for details, see Heithaus 2001;
Wirsing et al. 2006). Because of an ontogenetic shift in the
tiger shark’s diet (Lowe et al. 1996), tiger sharks under
3.0 m TL are much less likely to pose a threat to dugongs.
Thus, only sharks >3.0 m TL were used to generate daily
catch rates. Note, however, that the inclusion of sharks fall-
ing below this size threshold does not change the results of
the analyses (see “Electronic supplementary material”).
Although our Wshing eVort was intensive and daily Wsh-
ing eVort varied little (Wirsing et al. 2006), Wshing fre-
quency (»6 Wshing days month
¡1
) did not allow for a
continuous evaluation of the relationship between daily
dugong habitat use and predator numbers. Thus, we used a
sinusoidal function with a period of one year to predict the
annual trend in daily catch rates (sharks h
¡1
; Fig. 2). The
catch data used in the model were combined for the years
2002–2004 because interannual variation was not detected
after accounting for seasonal eVects (F
(2,117)
=0.97,
P = 0.38). We Wt the model using maximum-likelihood
under the assumption of a Poisson error distribution since
our data consisted of integer values for the number of
sharks caught per day (we rarely caught more than one
large shark per Wshing day). The model was characterized
by a good Wt (i.e., tight conWdence intervals; see Fig. 2), so
we are conWdent that it furnished reliable daily estimates of
large tiger shark abundance. Nevertheless, to assess the
validity of the model’s predictions and test whether pooling
years introduced bias, we carried out a companion analysis
involving running two-week averages of daily tiger shark
catch rates centered on each day on which dugong density
estimates were calculated; the results of this analysis did
not diVer from those presented below (see “Electronic sup-
plementary material”).
Statistical analyses
Dugong densities for individual survey zones were calcu-
lated by dividing the number of animals sighted by the area
searched (ha). Densities for all survey zones visited on a
given day were pooled into shallow and deep categories;
daily densities for the study area were derived by summing
the densities for shallow and deep habitats, weighted by the
proportional coverage of each habitat category. Between
1997 and 1999, only dugongs sighted within 100 m of the
central transect line (a 200-m sighting belt) were recorded.
Thus, densities over the six years of the investigation were
calculated using this sighting area to facilitate annual com-
parison. Factors potentially aVecting dugong density were
assessed using a generalized linear model with a negative
binomial error distribution because the dependent variable
consisted of non-negative integer values with a mode of
zero, a low mean, and large variance (White and Bennetts
1996); day-of-year (DOY), year, habitat (shallow versus
deep), and sea surface temperature (°C) (a putative driver
of dugong movements in Shark Bay; Anderson 1986;
Marsh et al. 1994) were included as independent variables.
A concurrent study involving hour-long focal animal fol-
lows (n = 120) revealed that surfacing rates of dugongs
diVer in shallow (0.96 surface trips min
¡1
) and deep
(0.84 trips min
¡1
) habitats (t
118
=2.19, P = 0.03) (A. Wir-
sing et al., unpublished data). Thus, to ensure that dugong
density estimates reXected actual patterns of habitat use, we
multiplied deep habitat densities by the ratio between these
two rates (1.14).
Foraging dugong densities for survey zones were
expressed as the number of feeding animals sighted (i.e.,
counts within the sighting belt; here, the entire 400-m belt
was used to maximize sample size) divided by the volume
of seagrass surveyed (m
3
). This measure assumes that for-
aging animals should be distributed across habitats in pro-
portion to food availability (van Baalen and Sabelis 1993).
Fig. 2 Maximum likelihood model of tiger shark abundance based on
daily catch rates (Wlled circles) pooled over three years (95% CI shown
in gray lines). The model was Wt using a sinusoidal function with a pe-
riod of one year and a Poisson error distribution given the strong sea-
sonal pattern and heterogeneity manifest in the raw catch data
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0
50 150 200 250 300 350
Day-of-year, DOY
Tiger shark catch rate (no.h
-1
)
100
Oecologia (2007) 153:1031–1040 1035
123
Thus, after division by food supply, asymmetry between
forager densities in two patch types serves as a proxy for
the degree to which habitat choice is inXuenced by other
factors, including perceived predation danger. Note that use
of this habitat use metric requires that the analysis be
restricted to foraging individuals; thus, only survey data
from 2002 to 2004 were used to generate forager densities,
as activity states were not determined between 1997 and
1999. Foraging dugong densities for survey zones visited
on a given day were pooled into shallow and deep catego-
ries, and deep habitat densities were adjusted using the con-
version (1.17) between the surfacing rates of foraging
dugongs (n = 74) in the two habitats. We sighted ten
mother–calf pairs engaged in foraging while conducting
transects. The behavior of dugong calves mirrors that of
their mother, so we treated these pairs as one individual for
purposes of analysis; removing them from consideration
had no measurable eVect.
We used information-theoretic methodology (Burnham
and Anderson 1998) to evaluate the relationship between
daily use of shallow and deep habitats by foraging dugongs
and predator abundance (predicted large tiger shark catch
rate, sharks day
¡1
). This approach ranks models of the rela-
tionship between dependent and explanatory variables
according to Wt, while accounting for diVerences in com-
plexity and varying degrees of freedom, and therefore facil-
itates rigorous evaluation of competing hypotheses
(Burnham and Anderson 1998; Anderson et al. 2000). Four
primary competing models of food-corrected forager den-
sity were evaluated: (1) a model based on large tiger shark
abundance (S), to test whether variation in predator num-
bers was accompanied by changes in foraging dugong den-
sity but neither a habitat shift nor a habitat preference (i.e.,
to challenge the food quantity hypothesis); (2) a model
based on both shark abundance and habitat category (deep
versus shallow; H), to assess whether foraging dugongs
consistently overused one habitat after accounting for
changes in overall dugong and predator abundance (i.e., to
challenge the food quality hypothesis); (3) a model incor-
porating shark abundance and its interaction with habitat
category (H £ S), to test whether Xuctuation in predator
numbers was accompanied by changes in dugong density
and a habitat shift (i.e., to test for an exchange of food for
safety and thereby challenge the predation risk sensitivity
hypothesis); and (4) a full model including shark abun-
dance, its interaction with habitat, and habitat type to test
whether foragers evinced a threat-sensitive shift but never-
theless maintained an overall habitat preference. In our
study area, sea surface temperature (T) and shark abun-
dance covary (Heithaus 2001). Thus, we also evaluated
four secondary models, each replacing shark abundance
with temperature (i.e., T, T + H, T + H £ T, and
T+H£ T + H), to ensure that any statistical relationships
between predator abundance and foraging dugong habitat
use were not a spurious consequence of correlation between
shark numbers and temperature. Given that the dependent
variable, daily forager counts divided by the volume of sea-
grass surveyed, consisted of non-negative integer values
with a mode of zero, a low mean, and large variance, mod-
els were Wt using maximum-likelihood under the assump-
tion of a negative binomial error distribution (White and
Bennetts 1996). We evaluated the strength of each model
using Akaike’s Information Criterion, corrected for small
sample size (AIC
c
; Burnham and Anderson 1998; Ander-
son et al. 2000). Akaike weights (w), which index the likeli-
hood that a particular model is the best among a set of
competitors, and the change in AIC
c
between models (
AIC
c
), were used to assess model uncertainty (Burnham
and Anderson 1998). The predictive strengths of individual
explanatory variables were evaluated using coeYcient esti-
mates with 95% conWdence intervals (Burnham and Ander-
son 1998).
Note that nonlinear transformation of the continuous
independent variables in our analyses of dugong and forag-
ing dugong density (DOY, shark catch rate, temperature)
did not improve the performance of any of the linear mod-
els considered ( AIC
c
·2, Burnham and Anderson 1998).
Thus, only linear models are presented.
Results
Dugong density
We sighted 528 dugongs while conducting transects. Most
sightings (320 of 398) were of solitary animals, and group
size averaged 1.33 (§0.09, 95% CI) individuals, facilitating
robust estimation of survey zone (and overall) densities.
The density of dugongs in the study area varied temporally,
with warm season estimates (September–May) exceeding
those for the cold season (June–August) (Fig. 3a). The
retention of the DOY variable in the generalized linear
model, coupled with the exclusion of all interaction terms
(all P ¸ 0.15; Table 1), indicates that the observed annual
trend in dugong abundance was both continuous and con-
served across years. However, the magnitude of numerical
change showed considerable annual variation (Table 1;
Fig. 3b). After blocking for spatial (i.e., survey zone)
eVects, average seagrass biomass estimates for quadrat
samples from shallow habitats (0.17 m
3
§ 0.01, 95% CI)
greatly exceeded those for deep areas (0.01 m
3
§ 0.003)
(F
(1,1809)
= 1042.58, P < 0.001). Not surprisingly, therefore,
dugong densities were consistently higher in shallow than
in deep habitats (Table 1; Fig. 3c). Finally, after accounting
for temporal and habitat eVects, sea surface temperature
was a signiWcant predictor of dugong density (Table 1).
1036 Oecologia (2007) 153:1031–1040
123
Although the relationship between water temperature and
dugong abundance was positive, the highest density esti-
mates coincided with intermediate temperatures (19–20 °C)
(Fig. 3d).
Foraging dugong habitat use
Between 2002 and 2004, we sighted 151 foraging dugongs
while conducting transects. Most encounters (83 of 113)
involved solitary animals, and group size averaged 1.34
(§0.33) individuals, facilitating reliable estimation of for-
ager densities for survey zones, and across habitat catego-
ries. The best model of foraging dugong density
incorporated large tiger shark abundance (S) and the interac-
tion between shark abundance and habitat (H £S; Table 2);
this model explained 21% of the observed daily variation in
forager density across the two habitat types (Table 2). The
Akaike weight (w) for this model (0.86) suggested that its
probability of being superior to the others under consider-
ation was high; indeed, its closest competitor was approxi-
mately eight times less likely to be preeminent (Table 2).
Furthermore, the 95% conWdence intervals for the coeYcient
estimates of shark abundance (7.40–23.86) and its interac-
tion with habitat (18.23–109.51) did not encompass zero,
suggesting that the relationships between these parameters
and forager density were statistically signiWcant. The coeY-
cient estimate for shark abundance was positive (15.63),
indicating that foraging dugong density increased with shark
numbers. The inclusion of the interaction between shark
abundance and habitat in the best model, however, means
that the degree to which food-adjusted forager densities and
predator abundance were associated diVered as a function of
habitat type: increases in shark abundance corresponded
with marked elevation in the use of only deep habitat by for-
aging dugongs (Fig. 4). On days when foragers were sighted
(n = 69), the extent to which use of deep survey zones
exceeded that of shallow survey zones following food
adjustment (i.e., preference for deep patches) correlated pos-
itively with tiger shark abundance (linear regression,
r
2
=0.17, t
67
= 3.73, = 6.41, P < 0.001; Fig. 5). By impli-
cation, input matching occurred when sharks were scarce,
while the tendency of foragers to overuse deep habitats was
greatest when sharks were most abundant (Figs. 4, 5).
Fig. 3a–d a Daily estimates of
overall dugong density (open
circles, number per ha
¡1
) in rela-
tion to day-of-year (DOY;
three years pooled); the cold sea-
son (June–August) is marked
with a gray line. b Annual trends
in dugong density (each trend
line is labeled with a year). c
Daily estimates of dugong den-
sity in shallow (open circles,
gray line) and deep (Wlled cir-
cles, black line) habitat. d Daily
estimates of dugong density in
relation to sea surface tempera-
ture (°C). Trend lines were gen-
erated using distance-weighted
least squares smoothing (DWLS,
tension 1.0, SYSTAT 10.2)
a
0.03
0.00
0.01
0.02
cold season
0
100 200 300
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
Day-of-year, DOY
Day-of-year, DOY
c
Dugong density (no.ha
-1
)
0
100 200 300
0.000
0.004
0.008
0.012
0.016
0.020
b
1997
1998
1999
2002
2003
2004
10 15 20 25 30
0.00
0.01
0.02
0.03
d
Temperature, °C
Table 1 Generalized linear model of dugong density (day
¡1
) in Shar
k
Bay, Western Australia, as a function of day-of-year (DOY, expressed
continuously), year, habitat (shallow versus deep), and sea surface
temperature (°C)
For all independent variables, coeYcient estimates () are accompa-
nied by standard errors and 95% conWdence intervals (CI). All interac-
tions were excluded from the Wnal model (all P ¸ 0.15)
Factor df SE ()95% CI P
DOY 1 0.009 0.002 0.006 to 0.012 <0.001
Year 5 ¡0.062 0.027 ¡0.115 to ¡0.009 0.001
Habitat 1 ¡0.859 0.143 ¡1.139 to ¡0.578 <0.001
Temperature 1 0.270 0.036 0.199 to 0.340 <0.001
Oecologia (2007) 153:1031–1040 1037
123
Resting dugong habitat use
We sighted 31 resting dugongs while conducting transects,
all of which were solitary. Only four individuals (12.9%)
were observed resting in shallow habitat, indicating a
strong tendency to select deep (safe) habitats among indi-
viduals engaged in this potentially dangerous behavior
(logistic regression, t
229
= 3.67, and P < 0.001).
Discussion
Consumers are predicted to match their resources when
predation risk is low and to trade food for safety by avoid-
ing dangerous habitats when risk is high (Power 1984a,
1984b; Abrahams and Dill 1989; van Baalen and Sabelis
1993). Habitat use patterns of this nature are diYcult to
document in the Weld, however, especially when the con-
sumer of interest is highly mobile and manipulative experi-
mentation is implausible. Here, using predictable numerical
variation in predator abundance to conduct a natural experi-
ment, we show that the habitat use decisions of a large,
wide-ranging marine herbivore match this theoretical
expectation. After adjustment for food supply, foraging
dugong densities were not proportional in shallow and deep
habitats throughout the year (rejecting the food availability
hypothesis). Rather, the degree of similarity between
Table 2 Competing models of the density of foraging dugongs across
two habitat types (shallow and deep)
Forager densities within each habitat were expressed as counts within
survey zones (n = 14, replicated seven times per habitat category) di-
vided by food supply (seagrass volume, m
3
). Models were generated
using linear combinations of habitat (H), an estimate of large (>3 m)
tiger shark abundance (S), sea surface temperature (°C) (T), the inter-
action between habitat category and shark abundance (H £ S), and the
interaction between habitat and water temperature (H £ T), and then
ranked using Akaike’s Information Criterion, corrected for small sam-
ple size (AIC
c
). For each model, K is the number of parameters in the
model + 1, is the change in AIC
c
between the model and the “best”
model (i.e., the model with the lowest AIC
c
; highlighted in bold), w is
the Akaike weight (i.e., the likelihood of pre-eminence), and R
2
L
is
the R
2
analog for models analyzed using maximum-likelihood
Model K wR
2
L
S+(H£ S) 3 0.000 0.864 0.208
S+H+(H£ S) 4 4.124 0.110 0.204
S + H 3 7.692 0.018 0.196
T+(H£ T) 3 10.392 0.005 0.191
T+H+(H£ T) 4 11.870 0.002 0.192
T + H 5 13.994 0.001 0.186
S 3 36.609 0.000 0.147
T 5 48.093 0.000 0.129
Constant 1 127.643 0.000 –
Fig. 4a–b Relationships between day-of-year (DOY) and a daily esti-
mates of foraging dugong counts divided by seagrass volume (m
3
) in
shallow (open circles, gray line) and deep (Wlled circles, black line)
habitats and b predicted large (>3 m) tiger shark catch rate. Trend lines
created using DWLS smoothing (tension 1.0)
0.00000
0.00001
0.00002
0.00003
0.00004
0.00005
0.00006
0.00000
0.00001
0.00002
0.00003
0.00004
0.00005
0.00006
0.00000
0.00001
0.00002
0.00003
0.00004
0.00005
0.00006
0
100 200 300
0.00
0.01
0.02
0.03
0.04
0.05
0
100 200 300
0.00
0.01
0.02
0.03
0.04
0.05
Shark catch rate (no.h
-1
)
Day-of-year, DOY
*
Foraging dugongs (no.m
3
seagrass)
a
b
Fig. 5 Residual counts of foragers in deep habitat (observed counts—
expected counts based on food supply) versus daily large tiger shar
k
catch rate (open circles, no. caught h
¡1
). Residuals were log(x +1)-
transformed to homogenize variances. Equality between observed and
expected deep counts (i.e., residual values equaling 0; dashed line) sig-
niWes proportional habitat use by foragers, after adjustment for food
biomass (input matching). Large, positive residual values signify over-
use of deep habitat
-0.3
-0.1
0.1
0.3
0.5
0.7
0 0.01
Shark catch rate (no. h
-1
)
Deep habitat residuals
0.02 0.03 0.04
1038 Oecologia (2007) 153:1031–1040
123
forager densities relative to food in shallow and deep habi-
tats was correlated with large tiger shark abundance: we
observed input matching (i.e., an ideal free distribution)
when sharks were scarce, overuse of deep habitats when
sharks were common, and strong association between daily
measures of shark abundance and the extent to which deep
habitats were overused. Thus, dugongs apparently (1) are
sensitive to variation in predation risk across habitats, (2)
perceive deep patches to be safe relative to shallower
patches, and (3) manage their probability of death by allo-
cating time to inherently safe but impoverished foraging
patches in proportion to the overall danger associated with
their surroundings.
Our conclusion that dugongs perceive deep foraging
areas as havens from predation is corroborated by the ten-
dency of animals at rest (a high-risk activity providing no
beneWt when undertaken in dangerous habitat) to use deep
habitat almost exclusively, the fact that individuals sur-
prised in the shallows by boats (a predator proxy, Frid and
Dill 2002) usually Xee to deeper water while those
approached in deep areas rarely respond (A. Wirsing,
unpublished data), and the tendency of other prey species
of tiger sharks (bottlenose dolphins, Tursiops sp.; pied cor-
morants, Phalacrocorax varius) to also shift into deep for-
aging habitats when sharks are abundant (Heithaus and Dill
2002; Heithaus 2005). We recognize, however, that the
existence of a trade-oV between food and safety also
requires that excess foragers in deep habitat, following cor-
rection for food biomass, are unable to compensate nutri-
tionally. For dugongs in our study area, deep water foraging
likely entails a substantial nutritional cost. Given that the
rhizomes of A. antarctica are inaccessible to dugongs
(Anderson 1986), and that aboveground seagrass biomass
was extremely low in deep survey zones, it is highly likely
that individuals feeding in the deep habitats we surveyed
experience depressed intake rates. Furthermore, deep sub-
strates in our study area are composed primarily of sand
and silt (Travers and Potter 2002; Heithaus 2004a), and are
largely devoid of tropical seagrass species (Walker et al.
1988; A. Wirsing, personal observation). Thus, dugongs
foraging in these habitats presumably are also faced with
higher search times and a lower quality food supply than
those feeding in the shallows. Finally, dugongs in deep hab-
itats undoubtedly expend more energy per unit time diving
to the ocean Xoor to acquire food than do individuals in
shallow patches.
The predator-mediated habitat shifts reported here could
have been the product of territoriality if individuals were
excluded from desirable (i.e., shallow) feeding areas during
periods of peak dugong abundance (when sharks also hap-
pen to be most common) (Ward et al. 2000). We consider
this scenario to be unlikely since territorial behavior has not
been observed in our study area and foraging dugongs often
congregate in preferred feeding areas (Preen 1995). Fur-
thermore, foraging dugongs adjusted their use of shallow
and deep habitats on a continuous basis, responding to
changes in predator abundance even during periods of low-
dugong density when the inXuence of any territoriality
would have been minimal.
The distribution of tropical seagrass species is consistent
with that of A. antarctica, but changes in tropical seagrass
availability in the shallows also could have contributed to
the observed habitat shift. This scenario lacks support
because the availability of tropical species in the shallows
decreases during the winter in our study area (D. Burk-
holder, unpublished data). If dugongs adjust their relative
use of deep and shallow habitats in response to tropical sea-
grass biomass, then we would have expected them to
reduce, rather than increase, their relative use of shallow
banks as tiger shark numbers dropped.
Interestingly, while ambient thermal conditions did not
aVect the patch choices of foraging dugongs within the study
area, they did inXuence the extent to which dugongs used the
study area itself (i.e., larger-scale space use). SpeciWcally,
the periodic pattern of overall dugong abundance we
observed was inversely related to sea surface temperature.
This trend is consistent with previous studies of dugong dis-
tribution in Shark Bay, which have shown that dugongs
using our study site during the warm season tend to shift
northward and/or to the Western Gulf during the cold season
(e.g., Marsh et al. 1994; Preen et al. 1997; Gales et al. 2004).
Yet, the highest densities we observed coincided with water
temperatures between 19 and 20 °C, which approach the
value proposed as the lower physiological threshold for dug-
ongs (19 °C; Anderson 1986), rather than peak temperatures
(26 °C), and we sighted dugongs with some regularity even
when water temperatures were as low as 16 °C. Thus, while
dugong densities apparently are inXuenced by water temper-
ature over large spatial scales, individuals can use particular
areas where ambient temperatures are 19 °C or less (see also
Lanyon et al. 2005; Sheppard et al. 2006).
The possibility that large marine predators like sharks
exert top–down eVects on marine communities has received
some empirical support (e.g., Myers et al. 2007) but remains
contentious (Heithaus 2004b). Previous work in Shark Bay
has revealed that increases in the threat of tiger shark preda-
tion elicit reductions in the use of shallow habitats by pisci-
vores (bottlenose dolphins and pied cormorants), which may
in turn provide a temporary reprieve for seagrass Wshes
(Heithaus and Dill 2002; Heithaus 2005). Dugongs, which
represent an attractive prey resource for tiger sharks, are
found primarily in shallow areas. Thus, it is possible that
their presence leads sharks to hunt predominantly in the
shallows and initiates this indirect relationship between
sharks and teleosts (Dill et al. 2003). The results from the
present study suggest that tiger sharks elicit a similar habitat
Oecologia (2007) 153:1031–1040 1039
123
shift by dugongs. Although the impact of dugongs on the
temperate seagrasses of Shark Bay is not known, grazing by
dugongs can alter the composition and structure of seagrass
meadows, seagrass nutrient content, and detrital cycles (de
Iongh et al. 1995; Preen 1995; Nakaoka and Aioi 1999;
Masini et al. 2001; Aragones et al. 2006). Hence, this shift
may in turn aVect the distribution and abundance of other
species (e.g., invertebrates, teleosts) that rely on seagrass for
shelter and subsistence and the quality of forage available to
mesograzers. We can speculate, then, that tiger sharks may
not only exert an indirect inXuence on seagrass meadows
(e.g., by sheltering preferred but dangerous areas from her-
bivory and/or preventing seagrass species requiring distur-
bance from gaining a foothold among more persistent types)
but also trigger trophic cascades within seagrass communi-
ties. Collectively, these results suggest that apex predators
such as tiger sharks may help to structure marine ecosys-
tems, and further that overexploitation of these predators
may have consequences for prey distributions and ecosys-
tem dynamics. In Shark Bay, the scope of the tiger shark’s
top-down role will of course remain hypothetical until stud-
ies quantifying the eVects of predator-mediated changes in
foraging by its prey on Wsh and seagrass biomass and com-
munity structure are conducted.
Finally, our results help to underscore the potentially
broad scope of intimidation by predators in marine environ-
ments. Despite wide interest in the implications of fear for
ecological communities (Brown and Kotler 2004), to date its
eVects on consumers in marine systems have rarely been
demonstrated (Dill et al. 2003). One notable example
involves several zooplankton taxa, which appear to shift
from shallow to deep, relatively nutrient-poor strata of the
water column during the day as a means of minimizing their
vulnerability to visually orienting predators (Hays 2003).
Our studies in Shark Bay show analogous exchanges of food
for safety in bottlenose dolphins (Heithaus and Dill 2002,
2006), sea birds (Heithaus 2005), and now dugongs. In all of
these cases, observed prey behavioral adjustments (i.e., hab-
itat shifts) were substantial, and consistent with expectations
based on spatial and temporal patterns of food distribution
and predation risk. Thus, we conclude that (1) the inXuence
of predator intimidation in marine environments is appar-
ently widespread, and that its implications should be consid-
ered even for species, like the dugong, that are long-lived
and do not suVer heavy predation rates, and (2) spatial
responses by marine prey to fear in at least some communi-
ties can be predicted using a combination of habitat charac-
teristics and the Wrst principles of behavioral ecology.
Acknowledgments We thank R. Abernethy, V. Alla, L. Barre, F.
Bretos, S. Buchannan, T. Bujas, J. Burghardt, S. Burghardt, C. Chow,
M. Davis, H. Finn, C. Genrich, P. Green, A. Greenley, K. Harper, L.
Heithaus, M. Kerford, S. Kowalewsky, A. Krickan, J. Lasky, L. Mar-
shall, K. Martin, J. McLash, R. McPhie, B. Stalvey, J. Wilder, and K.
Wirsing for Weld assistance. This material is based in part upon work
supported by the National Science Foundation under Grant Number
0526065. Other grants and support were provided by Australian Geo-
graphic, Humminbird, Mercury Marine Australia, Monkey Mia Dol-
phin Resort, Monkey Mia Wildsights (Shotover), National Geographic
Society Expeditions Council, National Geographic Remote Imaging,
NSERC Canada grant A6869 to L.M. Dill, PADI Foundation, Shake-
speare Electronics, Shark Bay Fish Factory, University of Western
Australia, and public donations. We are grateful to I. Anderson, B. Bar-
ton, C. Beck, K. Crane, A. Fraser, I. Gordon, D. Rose, R. Swann, and
D. Witt for logistical support. Special thanks go to D. Charles for ad-
vice, D. Capewell, and the Yadgalah Aboriginal Corporation for infor-
mation about dugongs, D. Massey for extra Weld housing, H. Raven for
temperature data, E. Elle, J. Estes, G. Hays, P. Peterson, B. Sargeant,
R. Ydenberg, and anonymous reviewers for helpful comments, B.
Smith, and T. Steury for statistical help, B. Black, J. Heyman, and R.
Holst for their hospitality, and the Dill lab for helpful suggestions. This
research was conducted under Fisheries Western Australia permits 67/
97 and 08/01, Western Australia Department of Conservation and
Land Management permits NE001808, SF002347, SF003818,
SF004228, and SF004542, SW008085, and renewals, and Simon Fra-
ser University Animal Care permits 639B and 653B, and complied
with the current laws of the country in which it was performed. This is
contribution number 27 of the Shark Bay Ecosystem Research Project.
References
Abrahams MV, Dill LM (1989) A determination of the energetic
equivalence of the risk of predation. Ecology 70:999–1007
Abrams PA (1995) Implications of dynamically variable traits for
identifying, classifying and measuring direct and indirect eVects
in ecological communities. Am Nat 146:112–134
Anderson DR, Burnham KP, Thompson WL (2000) Null hypothesis
testing: problems, prevalence, and an alternative. J Wildl Manage
64:912–923
Anderson PK (1986) Dugongs of Shark Bay, Australia—Seasonal migra-
tion, water temperature, and forage. Natl Geogr Res 2:473–490
Anderson PK (1995) Scarring and photoidentiWcation of dugongs (Dug-
ong dugon) in Shark Bay, Western Australia. Aq Mam 21:205–211
Anderson PK, Prince RIT (1985) Predation on dugongs: attacks by
killer whales. J Mammal 66:554–556
Aragones LV, Lawler IR, Foley WJ, Marsh H (2006) Dugong grazing
and turtle cropping: grazing optimization in tropical seagrass sys-
tems? Oecologia 149:635–647
Brown JS, Kotler BP (2004) Hazardous duty pay and the foraging cost
of predation. Ecol Lett 7:999–1014
Brown JS, Laundré JW, Gurung M (1999) The ecology of fear: optimal
foraging, game theory, and trophic interactions. J Mammal
80:385–399
Burnham KP, Anderson DR (1998) Model selection and inference: a
practical information-theoretic approach. Springer, New York
Chilvers BL, Delean S, Gales NJ, Holley DK, Lawler IR, Marsh H,
Preen AR (2004) Diving behaviour of dugongs, Dugong dugon. J
Exp Mar Biol Ecol 304:203–224
Crawley MJ (1983) Herbivory. University of California Press, Los
Angeles, CA
de Iongh HH, Wenno BJ, Meelis E (1995) Seagrass distribution and
seasonal biomass changes in relation to dugong grazing in the
Moluccas, East Indonesia. Aq Bot 50:1–19
Dill LM, Heithaus MR, Walters CJ (2003) Behaviorally mediated indi-
rect interactions in marine communities and their conservation
implications. Ecology 84:1151–1157
Fretwell SD, Lucas HL (1970) On territorial behavior and other factors
inXuencing habitat distributions in birds: I. Theoretical develop-
ments. Acta Biotheor 19:16–36
1040 Oecologia (2007) 153:1031–1040
123
Frid A, Dill LM (2002) Human-caused disturbance stimuli as a form of
predation risk. Cons Ecol 6(1):11 (http://www.consecol.org/vol6/
iss1/art11)
Gales N, McCauley RD, Lanyon J, Holley D (2004) Change in abun-
dance of dugongs in Shark Bay, Ningaloo and Exmouth Gulf,
Western Australia: evidence for large-scale migration. Wildl Res
31:283–290
Hays GC (2003) A review of the adaptive signiWcance and ecosystem
consequences of zooplankton diel vertical migrations. Hydrobio-
logia 503:163–170
Heithaus MR (2001) The biology of tiger sharks, Galeocerdo cuvier,
in Shark Bay, Western Australia: sex ratio, size distribution, diet
and seasonal changes in catch rates. Environ Biol Fish 61:25–36
Heithaus MR (2004a) Fish communities of seagrass meadows and
associated habitats in Shark Bay, Western Australia. B Mar Sci
75:79–99
Heithaus MR (2004b) Predator-prey interactions. In: Carrier JC, Mu-
sick J, Heithaus MR (eds) The biology of sharks, skates, and rays.
CRC Press, Boca Raton, FL, pp 487–521
Heithaus MR (2005) Habitat use and group size of pied cormorants
(Phalacrocorax varius) in a seagrass ecosystem: possible eVects
of food abundance and predation risk. Mar Biol 147:27–35
Heithaus MR, Dill LM (2002) Food availability and tiger shark preda-
tion risk inXuence bottlenose dolphin habitat use. Ecology
83:480–491
Heithaus MR, Dill LM (2006) Does tiger shark predation risk inXuence
foraging habitat use by bottlenose dolphins at multiple spatial
scales? Oikos 114:257–264
Heithaus MR, Dill LM, Marshall GJ, Buhleier BM (2002) Habitat use
and foraging behavior of tiger sharks (Galeocerdo cuvier) in a
subtropical seagrass ecosystem. Mar Biol 140:237–248
Heithaus MR, Frid A, Wirsing AJ, Dill LM, Fourqurean JW, Burk-
holder D, Thomson J, Bejder L (2007) State-dependent risk-tak-
ing by green sea turtles mediates top-down eVects of tiger shark
intimidation in a marine ecosystem. J Anim Ecol (in press).
doi:10.1111/j.1365-2656.2007.01260.x
Lanyon J, Johns T, Sneath HL (2005) Year round presence of dugongs
in Pumicestone Passage, south-east Queensland, examined in re-
sponse to water temperature and seagrass distribution. Wildl Res
32:361–368
Laundré JW, Hernández L, Altendorf KB (2001) Wolves, elf, and bi-
son: reestablishing the “landscape of fear” in Yellowstone Na-
tional Park, USA. Can J Zool 79:1401–1409
Lima SL (1998) Stress and decision making under the risk of preda-
tion: recent developments from behavioral, reproductive, and
ecological perspectives. Adv Study Behav 27:215–290
Lima SL, Dill LM (1990) Behavioral decisions made under the risk of
predation: a review and prospectus. Can J Zool 68:619–640
Lowe CG, Wetherbee BM, Crow GL, Tester AL (1996) Ontogenetic
dietary shifts and feeding behavior of the tiger shark, Galeocerdo
cuvier, in Hawaiian waters. Environ Biol Fish 47:203–211
Marsh H, Channells PW, Heinsohn GE, Morrissey J (1982) Analysis
of stomach contents of dugongs from Queensland. Aust Wildl Res
9:55–67
Marsh H, Prince RIT, Saalfeld WK, Shepherd R (1994) The distribu-
tion and abundance of the dugong in Shark Bay, Western Austra-
lia. Wildl Res 21:149–161
Masini RJ, Anderson PK, McComb AJ (2001) A Halodule-dominated
community in a subtropical embayment: physical environment,
productivity, biomass, and impact of dugong grazing. Aq Bot
71:179–197
McNamara JM, Houston AI (1987) Starvation and predation as factors
limiting population size. Ecology 68:1515–1519
Myers RA, Baum JK, Shepherd TD, Powers SP, Peterson CH (2007)
Cascading eVects of the loss of apex predatory sharks from a
coastal ocean. Science 315:1846–1850
Nakaoka M, Aioi K (1999) Growth of the seagrass Halophila ovalis at
the dugong trails compared to existing within-patch variation in a
Thailand intertidal Xat. Mar Ecol Prog Ser 184:97–103
Peacor SD, Werner EE (2000) Predator eVects on an assemblage of
consumers through induced changes in consumer foraging behav-
ior. Ecology 81:1998–2010
Peacor S, Werner E (2001) The contribution of trait-mediated indirect
eVects to the net eVects of a predator. Proc Natl Acad Sci USA
98:3904–3908
Power ME (1984a) Habitat quality and the distribution of algae-graz-
ing catWsh in a Panamanian stream. J Anim Ecol 53:357–374
Power ME (1984b) Depth distributions of armored catWsh: predator-in-
duced resource avoidance? Ecology 65:523–528
Preen A (1995) Impacts of dugong foraging on seagrass habitats:
observational and experimental evidence for cultivation grazing.
Mar Ecol Prog Ser 124:201–213
Preen AR, Marsh H, Lawler IR, Prince RIT, Shepherd R (1997) Distri-
bution and abundance of dugongs, turtles, dolphins, and other
megafauna in Shark Bay, Ningaloo Reef and Exmouth Gulf,
Western Australia. Wildl Res 24:185–208
Preisser EL, Bolnick DI, Benard MF (2005) Scared to death? The
eVects of intimidation and consumption in predator-prey interac-
tions. Ecology 86:501–509
Schmitz OJ (2003) Top predator control of plant biodiversity and pro-
ductivity in an old-Weld ecosystem. Ecol Lett 6:156–163
Schmitz OJ, Krivan V, Ovadia O (2004) Trophic cascades: the primacy
of trait-mediated indirect interactions. Ecol Lett 7:153–163
Sheppard JK, Preen AR, Marsh H, Lawler IR, Whiting SD, Jones RE
(2006) Movement heterogeneity of dugongs, Dugong dugon (Mül-
ler), over large spatial scales. J Exp Mar Biol Ecol 334:64–83
Sih A (1980) Optimal behavior: can foragers balance two conXicting
demands? Science 210:1041–1043
Simpfendorfer CA, Goodreid AB, McAuley RB (2001) Size, sex, and
geographic variation in the diet of the tiger shark, Galeocerdo cu-
vier, from Western Australian waters. Environ Biol Fish 61:37–
46
Travers MJ, Potter IC (2002) Factors inXuencing the characteristics of
Wsh assemblages in a large subtropical marine embayment. J Fish
Biol 61:764–784
van Baalen M, Sabelis MW (1993) Coevolution of patch selection
strategies of predator and prey and the consequences for ecologi-
cal stability. Am Nat 142:646–670
Verdolin JL (2006) Meta-analysis of foraging and predation risk trade-
oVs in terrestrial systems. Behav Ecol Sociobiol 60:457–464
Walker DI, Kendrick GA, McComb AJ (1988) The distribution of sea-
grass species in Shark Bay, Western Australia, with notes on their
ecology. Aq Bot 30:305–317
Ward JF, Austin RM, Macdonald DW (2000) A simulation model of
foraging behaviour and the eVect of predation risk. J Anim Ecol
69:16–30
Warner RR (1998) The role of extreme iteroparity and risk-avoidance
in the evolution of mating systems. J Fish Biol 53(Suppl A):82–
93
Werner EE (1998) Ecological experiments and a research program in
community ecology. In: Resetarits WJ, Bernardo J (eds) Experi-
mental ecology: issues and perspectives. Oxford University Press,
Oxford, UK, pp 3–26
White GC, Bennetts RE (1996) Analysis of frequency count data using
the negative binomial distribution. Ecology 77:2549–2557
Wirsing AJ, Heithaus MR, Dill LM (2006) Tiger shark (Galeocerdo
cuvier) abundance and growth in a subtropical embayment: evi-
dence from 7 years of standardized Wshing eVort. Mar Biol
149:961–968
Wirsing AJ, Heithaus MR, Dill LM (2007) Dugongs prefer to forage in
microhabitats that allow escape from rather than avoidance of pre-
dators. Anim Behav 74:93–101
