Degraded Environments Alter Prey Risk Assessment.

0.69 MB·English

by  Lonnstedt, Oona M

Degraded Environments Alter Prey Risk Assessment
Oona M. Lo¨nnstedt1, Mark I. McCormick1 & Douglas P. Chivers2
1ARCCentreofExcellenceforCoralReefStudies,andSchoolofMarineandTropicalBiology,JamesCookUniversity,Townsville,Qld4811,
Australia
2DepartmentofBiology,UniversityofSaskatchewan,Saskatoon,SKS7N5E2,Canada
Keywords Abstract
Chemicalalarmcues,coraldegradation,coral
reefs,Pomacentrusamboinensis,predation, Elevated water temperatures, a decrease in ocean pH, and an increasing preva-
Pseudochromisfuscus. lence of severe storms have lead to bleaching and death of the hard corals that
underpin coral reef ecosystems. As coral cover declines, fish diversity and abun-
Correspondence dance declines. How degradation of coral reefs affects behavior of reef inhabit-
OonaM.Lo¨nnstedt,ARCCentreof ants is unknown. Here, we demonstrate that risk assessment behaviors of prey
ExcellenceforCoralReefStudies,andSchool
are severely affected by coral degradation. Juvenile damselfish were exposed to
ofMarineandTropicalBiology,JamesCook
visual and olfactory indicators of predation risk in healthy live, thermally
University,Townsville,Qld4811,Australia.
bleached, and dead coral in a series of laboratory and field experiments. While
Phone:[phone];Fax:[phone]
1570;Email:[email protected] fish still responded to visual cues in all habitats, they did not respond to olfac-
tory indicators of risk in dead coral habitats, likely as a result of alteration or
FundingInformation degradation of chemical cues. These cues are critical for learning and avoiding
ThisstudywasfundedthroughanAustralian predators, and a failure to respond can have dramatic repercussions for survival
ResearchCouncilCentreofExcellencefor
and recruitment.
CoralReefStudiesResearchGrantandwas
conductedunderJCUethicsapprovalA1593
andA1720.
Received:3August2012;Revised:29August
2012;Accepted:3September2012
EcologyandEvolution2013;3(1):38–47
doi:10.1002/ece3.388
Kavaliers and Choleris 2001). It is the decisions that indi-
Introduction
viduals make under the threat of predation that decide
Global Environmental Change (GEC) is having major their fate and the genes they hold, in this way indirectly
impacts on all of the world’s ecosystems and is viewed as shaping prey community composition (Abrams 2000).
one of the biggest threats to the natural world (Meehl Predators and their prey must continuously react and
et al. 2007). The earth’s climate is warming at a far adapt to their environment, but in today’s changing
greater rate than at any time during the past 10,000 years, world, we know very little about how climate-induced
in part, due to greatly increased emissions of atmospheric habitat change will affect the intricate, and at times
CO (Walther et al. 2002). On a population level, GEC is subtle, relationships between predators and their prey
2
expected to reduce both species abundance and diversity, (Ferrari et al. 2011a).
in some cases resulting in local or even global extinctions Impacts of GEC on the marine ecosystem include rising
(Hughes 2000; Williams et al. 2003; Munday 2004; Par- seasurface temperatures, changing hydrodynamic regimes,
mesan 2006). In addition to human-induced threats, and altered ocean chemistry (Munday et al. 2009; Roessig
animals are continually exposed to a broad array of risks et al.2004).Intheocean,coralreefsareamongthosehabi-
and dangers in their natural environment. The number tats that are most likely to be adversely affected by climate
of dangers an animal will face throughout its life are change (Hughes et al. 2003; Hoegh-Guldberg et al. 2007).
numerous and varied (e.g., parasites, bacterial infections, Coralreefenvironmentsrepresent oneoftheworld’smost
con- and hetero-specifics), but one threat that may end biologically diverse ecosystems; however, very little is
in instant death if ignored is predation (Sih 1984; knownoftheinteractionsbetweenpredatorsandtheirprey
38 ª2012TheAuthors.PublishedbyBlackwellPublishingLtd.ThisisanopenaccessarticleunderthetermsoftheCreative
Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original
workisproperlycited.
O.M.Lo¨nnstedtetal. DegradedEnvironmentsAlterPreyRiskAssessment
that have shaped this astonishing biodiversity. Although increased perception of colorful prey fishes against the
these habitats have become popular systems for examining white background of the coral (Coker et al. 2009;
variousaspectsoftheeffectofclimatechangeonbehavioral McCormick 2009, 2012).
interactions, the subject is still very much in its infancy The goal of this study was to determine how predator
(e.g., McCormick 2009; Dixson et al. 2010; Munday et al. risk assessment abilities of a na¨ıve coral reef fish prey
2010; Ferrari et al. 2011a,b). Decreases in ocean pH along (Pomacentrus amboinensis) were affected by three different
withincreases inwatertemperatures andtheprevalence of coral reef habitats, which represent a cline from healthy
severe storms have lead to bleaching and death of the live to degraded coral. Specifically, we undertook laboratory
hard corals that underpin coral reef ecosystems (Hughes and field experiments to examine whether three different
et al. 2003). As coral reefs degrade from live healthy coral stages of coral (live healthy, thermally bleached, or
to rubble, fish diversity and abundance declines (Graham degraded algae-covered dead coral) affected prey
et al. 2006). The majority of adult reef fishes are not responses to: (1) conspecific damage-released chemical
directly dependent on live corals for survival (Pratchett cues, (2) visual cues of a predator, and (3) a combination
et al.2008).Despitethis,wholefishcommunitieshaveseen of visual and chemical cues. Further experiments
dramaticdeclinesfollowinglossofcoralcoversuggestinga addressed the mechanisms responsible for the impaired
widespread reliance on the coral reef habitat (Jones et al. chemosensory responses in degraded coral habitats. Evi-
2004).Thewidereffectsofcoralbleachingonfishcommu- dence suggests that the process of coral degradation will
nities and, in particular, on the complex interrelationships not only affect prey directly through changes in their
betweenpredatorsandpreyremainpoorlyunderstoodand resource base, but indirectly through modifications of the
research is required to identify the underlying behavioral cues they use to assess predation risk.
processes that are driving the declines in abundance of
fishes.
Methods
Coral reef fishes have complex life histories incorpo-
rating a widely dispersive larval phase, lasting from
Study species and collections
weeks to months, followed by settlement to the benthic
reef environment. During this larval–juvenile transition, Experiments were conducted at Lizard Island (14°40′S,
mortality rates are extremely high, primarily driven by 145°28′E), northern Great Barrier Reef (GBR), Australia
predation (more than 50% are preyed upon in the first from October to November 2010. The ambon damselfish,
48 h; Almany and Webster 2006). Successful identifica- Pomacentrus amboinensis, were used as a model prey
tion of predators requires the newly settled larvae to species in all experimental trials. P. amboinensis is a
detect olfactory and visual signs of danger all within a common fish within coral reef fish communities in the
highly complex environment containing numerous dif- Indo-Pacific (especially on the GBR) and settle to a wide
ferent stimuli. Olfaction is particularly important at range of habitats, but are found in highest densities in
night when the larvae settle and in the highly complex shallow sandy areas on live corals (McCormick et al.
habitats of coral reefs that limit visual abilities and assist 2010). Their pelagic larval phase lasts between 15 and
cryptic predators (McCormick 2009; Vail and McCor- 23 days and the new recruits are readily collected over-
mick 2011). At this time, chemical alarm cues from the night with light traps that have been moored just outside
damaged skin of prey play an important role in the the reef (see Meekan et al. 2001 for design). Fish used
identification and avoidance of predators (Leduc et al. for the present studies were all caught in light traps and
2010; Lo¨nnstedt et al. 2012). Recent studies have sug- brought back to the Lizard island research station at
gested that GEC is threatening to perturb the delicate dawn and placed in 60-L flow-through seawater holding
balance between predators and their prey (Ferrari et al. tanks (densities of ~50 fish/tank). Fish were fed twice
2011b). Munday et al. (2010) found that newly settled daily with newly hatched Artemia sp. nauplii ad libitum
damselfish (Pomacentrus wardi) that had their olfactory to allow for recovery from the stress of capture. Juvenile
sense disrupted through exposure to increased CO lev- Apogon doederleini were used as control fish for adding
2
els had a five to ninefold increase in mortality than con- the skin extract cue of a heterospecific fish into the
trol fish when placed on the reef. Similarly, Ferrari et al. aquarium. These fish are phylogenetically and ecologi-
(2011b) showed a five to sevenfold increase in mortality cally distant from P. amboinensis, thus being an ideal
for another damselfish (P. chrysurus) exposed to elevated control fish. Apogonids were collected on the reef using
CO . Furthermore, it has been suggested that coral hand nets.
2
dwelling damselfish (family Pomacentridae) are more One of the most common and abundant predators on
susceptible to predation in bleached coral as the ability new settlers during the recruitment season is the dotty-
of prey fish to camouflage is diminished due to the back Pseudochromis fuscus (Feeney et al. 2012). As na¨ıve
©2012TheAuthors.PublishedbyBlackwellPublishingLtd. 39
DegradedEnvironmentsAlterPreyRiskAssessment O.M.Lo¨nnstedtetal.
prey fish have been found to have an innate fright reac- seawater until the commencement of trials (see Appendix
tion to the sight of this predator (unpublished data), it S1 in supporting information). The tanks were set up, so
was used as a model predator species to expose fish to they were continuously fed seawater from three separate
in the various habitats. As a control fish for the visual reservoirs (60 L) that either contained four coral heads
cues, we used the herbivorous goby (Amblygobius phala- (10 9 15 9 12 cm) of live healthy, live thermally
nea), which are of similar size and shape as adult dotty- bleached, or dead algae-covered coral habitat of the com-
backs. Both of these species are found in large numbers mon bushy hard coral Pocillopora damicornis. One of the
around Lizard Island and were collected using hand nets three types of coral habitat (live healthy, live bleached, or
and a dilute solution of clove oil anesthetic. Gobies and dead coral) was placed along the short side of the aquaria
dottybacks were brought back to the research station and creating vertical shelters (18 9 20 9 4 cm). All corals
placed individually in 13-L aquaria and fed daily with were replaced every 2 days and used coral was returned
fish food pellets. to the field. Na¨ıve P. amboinensis (n = 15–17/treatment)
Live healthy and dead algae-covered hard coral (Pocil- were placed individually in the aquaria and allowed to
lopora damicornis) were collected from the fringing reefs acclimate overnight. Prior to the start of the trial, the
around Lizard Island and placed in well aerated 500-L water flow was stopped and 5 mL of Artemia sp (~ 550
flow-through seawater holding tanks. The process of Artemia) was added to the aquaria to stimulate feeding.
bleaching involves the expulsion of symbiotic zooxanthel- The behavior of a single P. amboinensis was recorded for
lae algae when the coral is under stress. This can happen a 4-min pre-stimulus period. Immediately following the
when water temperatures reached >1°C above the pre-stimulus period, a further 5 mL of Artemia was added
summer maximum (Anthony et al. 2007). Pocillopora and fish were exposed to the relevant cue treatment and
damicornis colonies bleach in about 10 days and will die the behavior of the fish was then recorded for a further
in 2–3 weeks if the temperature remains consistently 4 min.
high, after which they get rapidly colonized by various To prepare the damage-released cues, we sacrificed one
algal and invertebrate species. In this study, healthy colo- recruit per trial using cold shock. The flank of each
nies were thermally bleached over a 12-day period using recruit was then superficially cut six times. The total cue
the protocol of McCormick et al. (2010). After colonies area was rinsed with 10 mL of seawater that had been
had expelled their zooxanthellae and were visibly collected from the test aquaria and was then filtered
bleached, but not dead, temperatures were once again through filter paper (47 mm Ø) prior to being used in
lowered to the ambient 28°C. the experiment. The behavioral response to experimental
treatments was quantitated by recording: total number of
successful feeding strikes, total time spent inside of shelter
Experimental outline
(s), and activity (quantitated as the number of times a
Weconductedthreeseparateexperiments,twointhelabo- fish crossed a line on the grid (3 9 3 cm) that had been
ratory and one in the field. All experiments were designed drawn on the side of the tank).
to test the effects of coral degradation on antipredator
response of fishes to predation cues. The first experiment,
Experiment 1: Does coral degradation
conducted in the laboratory, examined responses of
influence prey risk assessment in the
damselfish to visual, chemical, and combined visual and
laboratory?
chemical cues that indicate risk. The second experiment,
conducted in the field, focused solely on responses to Na¨ıve fish placed individually within aquaria containing
chemicalinformationandwasundertakentodeterminethe one of three coral habitats (live healthy, bleached, or dead
extent to which the findings of the first study were algae-covered coral) were exposed to one of seven differ-
pertinent to natural populations. The final laboratory ent cue treatments and their behavior was recorded as
experiment tested if seawater that contained, or had been above (n = 15–16). Chemical cue treatments included: (1)
in contact with, dead algae-covered coral caused a modifi- damage-released chemical cue of injured conspecifics; (2)
cation (alteration or degradation) of conspecific chemical control cues from injured heterospecifics, A. doederleini;
alarm cues or simply masked (i.e., overwhelmed) alarm and (3) saltwater control. Visual treatments included: (4)
cuesfrombeingdetected. a transparent bag filled with water; (5) a transparent bag
that contained a herbivorous goby, A. phalanea; (6) a
transparent bag that contained a predatory dottyback, P.
Design of laboratory experiments
fuscus. The seventh treatment included a combination of
All behavioral observations were conducted in transparent a pairing of treatment one and six, as we reasoned that
15-L aquaria (38 9 24 9 27 cm) with a constant flow of fish would have a stronger response when both sources of
40 ©2012TheAuthors.PublishedbyBlackwellPublishingLtd.
O.M.Lo¨nnstedtetal. DegradedEnvironmentsAlterPreyRiskAssessment
risk cues were available (e.g., Lima and Steury 2005; (bite rate), which also relates strongly to survival in the
McCormick and Manassa 2008). wild at this life stage (McCormick and Meekan 2010).
Distance from shelter for these recently settled fishes has
also been found to be closely related to survival in the
Experiment 2: Does coral degradation
first few days after settlement to the reef (McCormick
influence the antipredator response to
2009, 2012; McCormick and Meekan 2010; Munday
chemical indicators of risk in the field?
et al. 2010).
Our laboratory studies indicated that coral degradation
influences the responses of damselfish to chemical cues
Experiment 3: Does dead coral mask or
that indicate risk. This experiment aimed to determine
modify chemical indicators of predation
whether there was evidence of environmental masking or
risk?
alteration of damage-released cues in the field under nat-
ural conditions. All experimental trials were conducted Here, we attempted to identify a possible mechanism
within a sand patch surrounded by hard coral reef responsible for the impaired responses that we observed
(composed of a typical diversity of live and dead coral for fish exposed to alarm cues in dead coral habitats.
habitats) using SCUBA at depths between 4 and 8 m. Specifically, we tested whether the impaired chemosenso-
Small patch reefs (25 9 15 9 20 cm) of either live ry responses in dead coral likely resulted from (1) a
healthy P. damicornis, thermally bleached P. damicornis, chemical alteration/degradation of the cue (i.e., a struc-
or dead algal-covered P. damicornis were assembled in the tural change in the chemical cues that are not revers-
sandy area adjacent to the reef (see Appendix S2 in sup- ible), or (2) odor masking, whereby the lack of a
porting information). To avoid any contamination behavioral response in dead coral occurs as a result of a
between patch reefs, there was a minimum of 3 m high level of background odor that overwhelms the fish’s
between patches and we moved in an up-current direc- olfactory sense making the cues hard to discern. To
tion when doing the experiment. A single juvenile accomplish this, individual na¨ıve fish (n = 16–19) were
P. amboinensis was placed onto each patch reef and placed in tanks containing one of two habitats (live or
allowed to acclimate for a minimum of 30 min before dead hard coral) and left to acclimate. Fish in each hab-
behavioral observations commenced. A 2-m plastic tube itat were then exposed to conspecific skin extracts that
was attached up-current at the edge of the patch reef had been prepared (as above) with water from two dif-
using metal skewers. The behavioral response of na¨ıve P. ferent sources: (1) water that had flown past dead corals
amboinensis to three different treatments was tested: (1) (from a 60-L flow-through tank containing four dead,
skin extracts from damaged conspecifics; (2) skin extracts algae-covered colonies of P. damicornis
from damaged heterospecifics; and (3) saltwater (blank [10 9 15 9 12 cm]); or (2) water that had flown past
control) (n = 15). The behavior of focal fish was quanti- live healthy P. damicornis (four colonies in a 60-L tank).
fied for 3 min before (pre-stimulus period) and 3 min Their behavior was recorded before and after the injec-
after (post-stimulus period) the addition of a stimulus tion of the stimulus as above (c). In accordance with
(skin extract or saltwater). the previous experiments, we predicted impairment in
To prepare skin extracts underwater, light trap caught behavioral responses for fish exposed to alarm cues pre-
P. amboinensis fish were brought underwater in pared in healthy coral water, but tested in dead coral
75 9 125-mm click seal bags, which were filled with ~ habitats, and for fish exposed to alarm cues prepared in
40 mL of sea water. Fish were euthanized by a quick dead coral water and tested in dead coral habitats. We
blow to the brain case and the epidermis of the fish was predicted fish exposed to alarm cues prepared from
lightly scratched using a scalpel blade that had been healthy coral water and then tested in healthy coral hab-
placed in the bag. A disposable syringe equipped with a itat would display antipredator responses. If alarm cues
fine needle was used to perforate the bag and extract are altered/degraded by chemicals released from the dead
30 mL of the prepared stimulus. Behavior of the fish coral and these changes are not reversible, then fish
was assessed by a SCUBA diver positioned at least 1.5 m tested in healthy live coral environments should fail to
away from the patch reef. Four aspects of activity and respond to alarm cues prepared in dead coral water. In
behavior were estimated for each 3-min sampling period: contrast, if fish exposed to alarm cues prepared in dead
bite rate (successful and unsuccessful strikes), average coral water and tested in the presence of live coral
distance from shelter (cm), maximum distance from respond normally, then this would be regarded as evi-
shelter (cm), and time spent in shelter (s). Three min- dence that the dead coral water simply masks the odor
utes has previously been found to be sufficient to obtain of the alarm cues, as the effect is reversible with dilution
a representative estimate of an individual’s behavior into the tank.
©2012TheAuthors.PublishedbyBlackwellPublishingLtd. 41
DegradedEnvironmentsAlterPreyRiskAssessment O.M.Lo¨nnstedtetal.
cues in the laboratory (Pillai’s Trace = 0.19, df = 8, 266,
Statistics
P < 0.005). Univariate ANOVAs that examined the
To test whether the behavior of fish differed (in both change in behavior after exposure to the various threat
the field and the laboratory) among the three different cues showed that there was a significant difference in bite
habitats (healthy, bleached, and dead coral), and whether rate, activity, and time spent in shelter depending on
fish had been given olfactory indicators of risk (conspe- which habitat the fish occupied (P < 0.01; Fig. 1e,f). Fish
cific skin extract, heterospecific skin extract, or a saltwater in healthy and bleached habitats strongly reduced both
control), visual indicators of risk (visual predator, visual activity levels and bite rates to visual and chemical threat
herbivore, or none), or a combination of visual (preda- cues, and there was an additive effect when both cue
tor) and chemical indicators of risk (conspecific chemical sources were present. Prey fish did not respond stronger
alarm cue), a multivariate analysis of variance (MANO- to the simultaneous exposure of both sources of risk in
VA) was employed. A two-way MANOVA tested whether the dead coral compared with both the live and bleached
the behavior of fish differed between the background habitats (P > 0.05; Fig. 1e,f).
habitat (live or dead coral) or how the skin extract cue The habitat fish were on affected their response to
had been prepared (mixed with water in that had been damage-released chemical alarm cues in the field (MA-
in contact with live healthy coral or dead coral) and NOVA: Pillai’s Trace = 0.35, df = 4,123, P < 0.001;
whether behavior was affected by the interaction between Fig. 2). Univariate statistics indicate that prey fish were
these two factors. All data were analyzed as the differ- negatively impacted in dead coral habitats when assessing
ence between the magnitude of behaviors before an predation risk by olfaction. In the healthy habitats, fish
experimental stimulus and after exposure to a stimulus responded to chemical cues by retreating to shelter and
(post-pre). Variables included in the analysis were as fol- reducing their foraging compared with the controls (Tu-
lows: bite rate, activity level, distance from shelter, and key’s HSD tests: P < 0.05; Fig. 2). Although fish
time spent in shelter. Time spent in shelter was responded to damage-released cues when in the bleached
log (x + 1) transformed to meet assumptions of nor- coral and fish spent less time inside the habitat, their
10
mality. Univariate analysis of variance (ANOVA) was behavior did not significantly differ from the two controls
employed to examine the nature of the significant differ- (Fig. 2b; P < 0.05). In the dead coral habitat, fish did not
ence found by MANOVAs. Significant ANOVAs were significantly change their behavior when exposed to dam-
further explored using unequal sample Tukey’s HSD age-released cues compared with the controls (Tukey’s
tests. A reduction in activity and foraging and movement HSD tests: P > 0.05; Fig. 2).
into or close to the shelter are common antipredator Therewasastronginteractiveeffectofbackgroundhab-
responses of damselfish to risk in both the laboratory itat and the type of water that the cue was prepared with
and field (Lo¨nnstedt et al. 2012). on the behavior of na¨ıve fish (Pillai’s Trace = 0.4,
df = 3,59, P < 0.001; fig. 3a,b). This was caused by the
combination of a live healthy coral background and skin
Results
extract cues prepared with water that had been in contact
The way fish changed their behavior in response to con- with live coral differing from all the other treatments,
specific damage-released cues differed among habitats in which in turn, did not differ from one another (fig. 3a,b).
the laboratory compared with the two controls (MANO- Univariate ANOVAs on each behavioral variable revealed
VA: Pillai’s Trace = 0.19, degrees of freedom [df] = 8, that na¨ıve fish in tanks with a background of healthy live
266, P < 0.001). Fish exhibited a significant decrease in coral responded with a reduction in activity, bite rate, and
bite rate when exposed to chemical cues in both the distancefromshelter(F = 17.7,P < 0.001;F = 11.9,
1,61 1,61
healthy and bleached coral habitats (Tukey’s HSD tests: P (cid:1) 0.001;F = 18.7,P < 0.001)whenexposedtocon-
1,61
P < 0.05; Fig. 1a). Fish in the dead coral habitat did not specificskinextracts that hadbeenprepared with seawater
significantly change their bite rate compared with the that had only been in contact with live healthy coral
controls when exposed to damage-released chemical cues (fig. 3a,b). Contrastingly, fish with a background habitat
(Tukey’s HSD tests: P > 0.05; Fig. 1a). Fish decreased for- consisting of dead, algae-covered coral did not respond to
aging and activity in the three different habitats when any skin extracts (regardless of how they had been pre-
exposed to the sight of a predator compared with the two pared). Similarly, fish inlive healthycoral habitats didnot
visual controls (MANOVA: Pillai’s Trace = 0.17, df = 8, respond to conspecific skin extracts prepared with water
266, P < 0.0001; Fig. 1c,d). that had been in contact with degraded coral habitats. It
Habitat type strongly influenced the response of fish to appears as though seawater that is, or has been in contact
chemical (conspecific skin extract), visual (visual preda- with, dead algae-covered coral may alter the structure of
tor), or a combination of chemical and visual predator conspecificchemicalalarmcues.
42 ©2012TheAuthors.PublishedbyBlackwellPublishingLtd.
O.M.Lo¨nnstedtetal. DegradedEnvironmentsAlterPreyRiskAssessment
Healthy coral Bleached coral Dead coral
(a) 10 (b) 10
5 5
0 b b b b 0
Bite rate––11–505 b b b Activity––11–505 b b b b b b b
–20 a –20 a
–25 –25
a
–30 –30 a
Conspecific skin Heterospecific skin Saltwater control Conspecific skin Heterospecific skin Saltwater control
extract extract extract extract
(c) 10 (d) 15
5 10
0 5
–5 b b 0
–10 b b b b –5 b b b b b b
Bite rate–––221505 Activity––––22115050
–30 –30
–35 –35
a
–40 a –40 a
–45 a a –45 a
–50 –50
Predator Herbivorous goby Empty bag control Predator Herbivorous goby Empty bag control
(e) 0 (f) 0
–5 –5
–10 a –10 a
–15 –15
Bite rate–––––4332205050 b b bc bc Activity–––––4332205050 b b bc bc bc
–45 bc bc –45 bc
–50 c –50
–55 c –55 c c
–60 –60
Conspecific skin Predator Conspecific skin Conspecific skin Predator Conspecific skin
extract extract & predator extract extract & predator
Figure1. Coral degradation affects assessment of predation risk by prey in the laboratory; with the column on the left showing the mean
change in biterate, and thecolumn onthe right displays the meanchange inactivity of fishto thedifferent treatments. Change in behavioral
responsesofcoralreeffishexposedtochemicalalarmcues(a,b),apredator(c,d),andapairingofthetwo(e,f).Biteratesandactivitylevelsare
significantlydecreasedwhenexposedtothreatcuesinbothhealthyandbleachedcoralhabitats.Whenexposedtothevisualsightofapredator
activity, levels andbiterateswere stronglyreduced regardless ofthebackgroundhabitat. Inthehealthy andbleached habitats,thesebehaviors
wereintensifiedwhenfishwerepresentedwiththesightofapredatorpairedwithachemicalcue.LettersaboveorbelowbarsrepresentTukey’s
HSDgroupingofmeans(a=0.05).
coral habitats, prey occupying degraded habitats did not
Discussion
show a stronger response when given the combined cue
Weshowedthatcoraldegradationhadaprofound influence sources. Failing to respond to an olfactory indicator of
on the behavioral responses of fish to cues that indicate risk greatly increases the likelihood of being preyed upon
predation risk. Fish in live healthy and bleached coral fed (Munday et al. 2010; Ferrari et al. 2011b). Fish with
above the colony and reduced swimming, ignored food, impaired olfactory abilities are also less likely to find a
and sought refuge when exposed to either chemical or suitable settlement sites and potential mates (Curtis et al.
visual indicators of risk. Prey in the dead, algae-covered 2001; Dixson et al. 2010; Munday et al. 2009, 2010;
coral habitat showed a similar antipredator response when Devine et al. 2012).
exposed to the sight of a predator, but when presented We know from previous studies that small bodied coral
with damage-released alarm cues of conspecifics, they did dwelling damselfish (family Pomacentridae) decline in
not visibly change their behavior in either the laboratory abundance following coral bleaching and reef degradation
or the field. While a pairing of olfactory and visual threat (Wilson et al. 2006; McCormick 2009). As they are not
cues had an additive effect on the prey response in live obligate corallivores, it has been not clear why they exhi-
©2012TheAuthors.PublishedbyBlackwellPublishingLtd. 43
DegradedEnvironmentsAlterPreyRiskAssessment O.M.Lo¨nnstedtetal.
(a) Healthy coral Bleached coral Dead coral (a) Dead Dead Live Live Background coral
10 c c Dead Live Dead Live Water for skin
0 extract
5 c
bc bc bc –5
0 e b b
at –10 b
ate –5 Bite r –15
e r–10 ab –20
Bit–15
–25
–20 a
–30
a
–25 a (b) 5
–30 0
–5
(b) 0.1
–10 b b
e 0.05 c bc bc bc bc bc Activity ––2105 b
c
n 0 –25
a
st –30 a
di
ve –0.05 bc –35 Dead Live Dead Live Background coral
elati –0.1 Dead Dead Live Live Wexatrtaecrt for skin
R ab
–0.15 (c) 0.1
a
0.05
–0.2
0
Conspecific skin Heterospecific skin Saltwater control c
extract extract e–0.05
c
an –0.1 b b
Figure2. Mean change of naı¨ve fish when exposed to various Dist–0.15
olfactory cues in the field. (a) Bite rate is strongly reduced in both
healthyandbleachedcoralwhenexposedtoconspecificskinextracts –0.2 a
whilenotwhenexposedtoheterospecificskinextractsorasaltwater –0.25 Background coral
Dead Live Dead Live
control.(b)Whenexposedtochemicalalarmcuesofconspecifics,fish Water for skin
Dead Dead Live Live extract
strongly reduced their distance from shelter in the live healthy coral,
but tended to retire to shelter less in both bleached and dead coral Figure3. Comparison of the behavior of Pomacentrus amboinensis
habitats.LettersaboveorbelowbarsrepresentTukey’sHSDgrouping
in the laboratory that had been exposed to conspecific skin extracts
ofmeans(a=0.05).
prepared with water containing either live healthy coral or dead
bit such strong reductions in abundance following large algae-coveredcoralinoneoftwobackgroundhabitats(livehealthyor
dead algae-covered). Behaviors are the change between the 4-min
scale bleaching events. It was initially believed to be due
pre-andpost-stimulusperiodin(a)biterate,(b)activitylevel,and(c)
to a decline in coral cover and the subsequent reduction
average distance from the coral shelter. Letters above or below bars
in the structural complexity of the coral reef framework representunequalTukey’sHSDgroupingofmeans(a=0.05).
(making them more susceptible to predators), but bleach-
ing does notnecessarily equate to a loss in habitat structure the smell of dying tissue may force recruit stage fish away
in the short term (Pratchett et al. 2008). During bleach- from bleached coral, leading to higher vulnerability. Our
ing, the density of zooxanthellae (photosynthetic algae results suggest that the mechanism underlying the move
within the coral tissue) is reduced either through the away from degraded coral habitats may be their reduced
expulsion or death of the minute algal cells, thus not ability to identify the olfactory cues that are innately asso-
affecting the structure, but only the pigmentation of the ciated with predation threat (the chemical alarm cues).
coral. It is the subsequent death and erosion that results The information on which they base their decision has
in the loss of coral structure (Booth and Beretta 2002). changed, affecting their perception of where they should
Hence, it is the live coral in itself that offers some sort of best sit along the axis of risk from shelter (and reduced
advantage to fish. This study demonstrates that fish foraging opportunities) to open water (and increased
appear unwilling to retreat back into bleached or foraging opportunities).
degraded coral when exposed to threat cues, spending less The relative context in which a threat stimulus is
time in shelter compared with when occupying the live receivedcaninfluenceboththequalityandeffectivenessof
healthy coral colonies. McCormick (2009) suggested that asignalascertainenvironmentalconditions,or“background
44 ©2012TheAuthors.PublishedbyBlackwellPublishingLtd.
O.M.Lo¨nnstedtetal. DegradedEnvironmentsAlterPreyRiskAssessment
noise”, can alter the signals perceived form (Endler 1992). In contrast, the mechanism described in this study is
Thephenomenonofodormaskinghasbeenwellstudiedin external to the animal, and involves the modification of
terrestrial environments (for a comprehensive review see the cue, such that it is either not recognized or inappro-
Schroder and Hilker 2008), but the focus in this literature priately categorized. Our data suggest that all individuals
is often background odors masking resource indicating where similarly impacted, suggesting a limited ability to
cues. For instance, certain plants produce an odor that adapt to the loss of this important sensory cue. As coral
repels insects, or hides the odors of their host plants death and degradation becomes increasingly prevalent
(Mauchlineet al.2005).Theybenefittheplantsbyallowing (Wilkinson 2004), further research is required to
them to effectively hide from consumers in a complex determine the extent to which the risk assessment of
chemicalenvironment.Inourstudy,wetestedwhetherthe other species may be affected by the same mechanism
background odor of dead coral masked or modified the and the community wide repercussions.
scent of alarm cues, reducing the response of prey to Due to GEC, coral reefs all over the world are declin-
threats. Once a coral is dead and overgrown by algae, a ing in health and what once were fields of live coral are
wholenewcommunitysettlesintoitandallthesedifferent now low lying rubble beds. Fish living within these
life forms (together with the algae) may overpower other changed environments are more likely to become
odors in the environment (such as the scent of wounded stressed as the coral degrades, both as a result of the loss
conspecifics). However, we found no evidence for odor of refuge space and from a change in their olfactory
masking, as fish exposed to alarm cues prepared in dead environment. Is it plausible that the fish are so stressed
coral water did not elicit a response in water containing by their new surroundings that they fail to respond to
healthycoral.Wepreparedthealarmcuein10 mLofwater predator threats? This seems very unlikely as the fish still
and injected the cue into a tank containing 15 L of water. responded to the sight of a predator when in the dead
Despite this huge dilution effect, the “unmasked cue” did coral. Predation is one of the most important processes
notelicitafrightresponseinthefish.Fishhavebeenshown shaping coral reef fish communities. Our current find-
to havearemarkableability todifferentiate between threat ings suggest that coral bleaching and coral death will
cues even when presented together (Mitchell et al. 2011), impact the crucial interactions between fish predators
which also suggests that odor masking is unlikely. As an and their prey. Bleached and dead coral patches appear
alternative to odor masking, our results support the to interfere with olfactory cues critical for the assessment
hypothesis that dead coral rapidly alters or degrades the of risk by prey. Without detecting the olfactory signposts
chemicalalarmcue.Wheneverthealarmcueswereincon- of risk, prey are unable to identify the early signs of dan-
tactwithdeadcoral(eitherpreparedindeadcoralwateror ger and are more likely to fall prey to hungry predators
injected into a tank containing dead algae-covered coral), (McCormick 2009; Munday et al. 2010). Biologists and
fishfailedtoelicitnormalantipredatorresponses.Assuch, managers wishing to predict the long-term consequences
our results resemble the responses of salmonid fishes in of global environmental change on reef fish assemblages
freshwater systems, whereby the alarm cues are rendered will need to understand the repercussions of this crucial
inactivewhenthepHdropsto6.0(Leducet al.2004).The developmental bottleneck (Lo¨nnstedt et al. 2012; Ferrari
proximatechemicalmechanismresponsibleforthischange et al. 2012a,b).
inoursystemremainsunknown,butlikelyisnotaresultof
a pH change, as this was not altered in the study systems Acknowledgments
giventhatmarinesystemsdonotshowlargechangesinpH
We thank L. Vail, A. Hogget, and R. Duffy for logistic
(Gaglianoet al.2010).
support. This study was funded through an Australian
The impact on the olfactory sense due to degraded
habitat is different from the recently documented impacts Research Council Centre of Excellence for Coral Reef
of elevated dissolved CO on the olfactory sense. Dis- Studies. Research was conducted under JCU ethics
2
solved CO elevated above 900 latm has been shown to approval A1593 and A1720.
2
alter the function of neurotransmitters in fish (Nilsson
et al. 2012), leading to the reduced discrimination of per- Conflict of Interest
tinent sensory cues (Dixson et al. 2010; Ferrari et al.
None declared.
2012b) and the negation of learning processes associated
with the correct identification of chemical alarm cues
References
(Ferrari et al. 2012a,b). Luckily, there may be sufficient
variability in the physiological response at low CO2 con- Abrams,P.A.2000.Theevolutionofpredator-prey
centrations (700 latm) within populations for fish to interactions:theoryandevidence.Annu.Rev.Ecol.Syst.
adapt to this CO -rich world through ecological selection. 31:79–105.
2
©2012TheAuthors.PublishedbyBlackwellPublishingLtd. 45
DegradedEnvironmentsAlterPreyRiskAssessment O.M.Lo¨nnstedtetal.
Almany,G.R.,andM.S.Webster.2006.Thepredation oceaniccoralreefecosystems.Proc.Nat.Acad.Sci.
gauntlet:earlypost-settlementmortalityinreeffishes.Coral 103:8425–8429.
Reefs25:19–22. Hoegh-Guldberg,A.,P.J.Mumby,A.J.Hooten,R.S.Steneck,
Anthony,K.R.N.,S.R.Connolly,andO.Hoegh-Guldberg. P.Greenfield,E.Gomez,et al.2007.Coralreefsunderrapid
2007.Bleaching,energetics,andcoralmortalityrisk:effects climatechangeandoceanacidification.Science318:1737–1742.
oftemperature,light,andsedimentregime.Limnol. Hughes,L.2000.Biologicalconsequencesofglobalwarming:is
Oceanogr.52(2):716–726. thesignalalreadyapparent?TrendsEcol.Evol.15:56–61.
Booth,D.J.,andG.A.Beretta.2002.Changesinafish Hughes,T.P.,A.H.Baird,D.R.Bellwood,M.Card,
assemblageafteracoralbleachingevent.Mar.Ecol.Prog. S.R.Connolly,C.F.R.Grosberg,et al.2003.Climate
Ser.245:205–212. change,humanimpacts,andtheresilienceofcoralreefs.
Coker,D.J.,M.S.Pratchett,andP.L.Munday.2009.Coral Science301:929–933.
bleachingandhabitatdegradationincreasesusceptibilityto Jones,G.P.,M.I.McCormick,M.Srinivasan,andJ.V.Eagle.
predationforcoral-dwellingfishes.Behav.Ecol. 2004.Coraldeclinethreatensfishbiodiversityinmarine
20:1204–1210. reserves.Proc.Nat.Acad.Sci.U.S.A.101:8251–8253.
Curtis,J.T.,Y.Liu,andZ.Wang.2001.Lesionsofthe Kavaliers,M.,andE.Choleris.2001.Antipredatorresponses
vomeronasalorgandiruptmating-inducedpairbondingin anddefensivebehaviour:ecologicalandethological
femaleprairievoles(Microtusochrogaster)Brain.Res. approachesfortheneurosciences.Neurosci.Behav.Rev.
901:167–174. 25:577–586.
Devine,B.M.,P.L.Munday,andG.P.Jones.2012.Homing Leduc,A.O.H.C.,J.M.Kelly,andG.E.Brown.2004.
abilityofadultcardinalfishisaffectedbyelevatedcarbon Detectionofconspecificalarmcuesbyjuvenilesalmonids
dioxide.Oecologia168:269–276. underneutralandweaklyacidicconditions:laboratoryand
Dixson,D.L.,P.L.Munday,andG.P.Jones.2010.Ocean fieldtests.Behav.Ecol.139:318–324.
acidificationdisruptstheinnateabilityoffishtodetect Leduc,A.O.H.C.,J.W.Kim,C.J.MacNaughton,and
predatorolfactorycues.Ecol.Lett.13:68–75. G.E.Brown.2010.Sensorycomplementmodelhelpsto
Endler,J.A.1992.Signals,signalconditions, andthedirection predictdielalarmresponsepatternsinjuvenileAtlantic
ofevolution.Am.Nat.139:125–153. salmon(Salmosalar)undernaturalconditions. Can.J.Zool.
Feeney,W.E.,O.M.Lonnstedt,Y.Bosiger,J.Martin, 88:398–403.
G.P.Jones,R.J.Rowe,et al.2012.Highrateofprey Lima,S.L.,andT.D.Steury.2005.Pp.166–188The
consumptioninasmallpredatoryfishoncoralreefs.Coral perceptionofpredatorrisk-thefoundationofnon-lethal
Reefs31:909–918. predator-preyinteractions.InEcologyofPredator/Prey
Ferrari,M.C.O.,D.L.Dixson,P.L.Munday,M.I. Interactions(eds.BarbosaP.,CastellanosI.).Oxford,
McCormick,M.G.Meekan,A.Sih,et al.2011a.Intrageneric OxfordUniversityPress.
variationinanti-predatorresponsesofcoralreeffishesto Lo¨nnstedt,O.M.,M.I.McCormick,M.G.Meekan,
oceanacidification:implicationsofprojectingclimatechange M.C.O.Ferrari,andD.P.Chivers.2012.Learnandlive:
onmarinecommunities.Glob.ChangeBiol.17:2980–2986. theroleofpredatorexperienceininfluencingprey
Ferrari,M.C.O.,M.I.McCormick,P.L.Munday,M.G. behaviourandsurvival.Proc.R.Soc.Lond.BBiol.Sci.
Meekan,D.L.Dixson,O.M.Lonnstedt,et al.2011b. 279:2091–2098.
PuttingpreyandpredatorintotheCO2equation- Mauchline,A.L.,J.L.Osborne,A.P.Martin,G.M.Poppy,
quantitativeandqualitativeeffectsofoceanacidificationon andW.Powell.2005.Theeffectsofnon-hostplantessential
predator-preyinteractions.Ecol.Lett.14:1143–1148. oilvolatilesonthebehaviourofthepollenbeetleMeligethes
Ferrari,M.C.O.,R.Manassa,D.L.Dixson,P.L.Munday,M. aeneus.Entomol.141:181–188.
I.McCormick,M.G.Meekan,et al.2012a.Effectsofocean McCormick,M.I.2009.Behaviourallymediatedphenotypic
acidificationonlearningincoralreeffishes.PLoSONE7 selectioninadisturbedcoralreefenvironment.PLoSONE
(2):e31478.doi:10.1371/journal.pone.0031478. 4(9):e7096.doi:10.1371/journal.pone.0007096.
Ferrari,M.C.O.,P.L.Munday,M.I.McCormick,M.G. McCormick,M.I.2012.Lethaleffectsofhabitatdegradation
Meekan,D.L.Dixson,O.M.Lonnstedt,et al.2012b.Effects onfishesthroughchangingcompetitiveadvantage.Proc.
ofoceanacidificationonvisualriskassessmentincoralreef R.Soc.Lond.B.Sci279:3899–3904.
fishes.Funct.Ecol.26:553–558. McCormick,M.I.,andR.Manassa.2008.Predationrisk
Gagliano,M.,M.I.McCormick,J.A.Y.Moore,andM. assessmentbyolfactoryandvisualcuesinacoralreeffish.
Depczynski.2010.Thebasicsofacidification:baseline CoralReefs27:105–113.
variabilityofpHonAustraliancoralreefs.Mar.Biol. McCormick,M.I.,andM.G.Meekan.2010.Theimportance
157:1849–1856. ofattitude:theinfluenceofbehaviouronsurvivalatan
Graham,N.A.J.,S.K.Wilson,S.Jennings,N.V.S.Polunin, ontogeneticboundary. Mar.Ecol.Prog.Ser.407:
J.P.Bijoux,J.Robinson,et al.2006.Dynamicfragilityof 173–185.
46 ©2012TheAuthors.PublishedbyBlackwellPublishingLtd.
O.M.Lo¨nnstedtetal. DegradedEnvironmentsAlterPreyRiskAssessment
McCormick,M.I.,J.A.Y.Moore,andP.L.Munday.2010. inducedcoralbleachingoncoral-reeffishesecologicaland
Influenceofhabitatdegradationonfishreplenishment. economicalconsequences.Oceanogr.Mar.Biol.Annu.Rev.
CoralReefs29:537–546. 46:251–296.
Meehl,G.A.,T.F.Stocker,P.Collins,P.Friedlingstein, Roessig,J.M.,C.M.Woodley,J.C.Joseph,andL.J.Hansen.
A.T.Gaye,andJ.M.Gregory,et al.2007.Globalclimate 2004.Effectsofglobalclimatechangeonmarineand
projection.inS.Solomon,D.Qin,M.Manning,Z.Chen, estuarinefishesandfisheries.Rev.FishBiol.Fish.14:251–275.
M.Marquis,K.B.Averyt,et al.eds.ClimateChange2007: Schroder,R.,andM.Hilker.2008.Therelevanceof
ThePhysicalScienceBasis.ContributionofWorkingGroup backgroundodorinresourcselocationbyinsects:a
itotheFourthAssessmentReportontheIntergovernmental behaviouralapproach.BioSci.58:308–316.
PanelonClimateChange,CambridgePress,Cambridge Sih,A.1984.Thebehaviouralresponseracebetweenpredator
Meekan,M.G.,S.G.Wilson,A.Halford,andA.Retzel.2001. andprey.Am.Nat.123:143–150.
Acomparisonofcatchesoffishesandinvertebratesbytwo Vail,A.,andM.I.McCormick.2011.Larvalreeffishavoid
lighttrapdesigns,intropicalNWAustralia.Mar.Biol. predatorscentwhenchoosingahome.Biol.Letdoi:10.1098/
139:373–381. rsbl.2011.0380.
Mitchell,M.,M.I.McCormick,M.C.O.Ferrari,and Walther,G.R.,E.Post,P.Convey,A.Menzel,C.Parmesan,
D.P.Chivers.2011.Coralreeffishrapidlylearntoidentify T.J.C.Beebee,et al.2002.Ecologicalresponsestorecent
multipleunknownpredatorsuponrecruitmenttothereef. climatechange.Nature416:389–395.
PLoSONE6(1):e15764.doi:10.1371/journal.pone.0015764. Wilkinson,C.R.2004.Statusofcoralreefsoftheworld2004,
Munday,P.L.2004.Habitatloss,resourcespecialization,and vols1and2.AustralianInstituteofMarineScience,
extinctiononcoralreefs.Glob.ChangeBiol.10:1642–1647. Townsville,Australia.
Munday,P.L.,D.L.Dixson,J.M.Donelson,G.P.Jones,M. Williams,S.E.,E.E.Bolitho,andS.Fox.2003.Climate
S.Pratchett,G.V.Devitsina,et al.2009.Oceanacidification changeinAustraliantropicalrainforests:animpending
impairsolfactorydiscrimination andhomingabilityofa environmentalcatastrophe.Proc.R.Soc.Lond.BBiol.Sci.
marinefish.Proc.Nat.Acad.Sci.U.S.A.106:1848–1852. 270:1887–1892.
Munday,P.L.,D.L.Dixson,M.I.McCormick,M.G. Wilson,S.K.,N.A.J.Graham,M.S.Pratchett,G.P.Jones,
Meekan,M.C.O.Ferrari,andD.P.Chivers.2010. andN.V.C.Polunin.2006.Multipledisturbancesandthe
Replenishmentoffishpopulationsisthreatenedbyocean globaldegradationofcoralreefs:arereeffishesatriskor
acidification.PNAS.107:12930–12934. resilient?Glob.ChangeBiol.12:2220–2234.
Munday,P.L.,M.I.McCormick,andG.E.Nilsson.inpress
a.ImpactofglobalwarmingandrisingCO oncoralreef Supporting Information
2
fishes:whathopeforthefuture?J.Exp.Biol.
Additional Supporting Information may be found in the
Munday,P.L.,M.I.McCormick,M.G.Meekan,D.L.Dixson,
online version of this article:
S.Watson,M.C.O.Ferrari,et al.inpressb.Selectionfor
CO2toleranceinmarinefishes.OceanAcidification. Appendix S1. Outline of the experimental set-up in the
Nilsson,G.E.,D.L.Dixson,P.Domenici,M.I.McCormick,
laboratory
C.Sørensen, S.Watson,et al.2012.Near-futureCO2levels Appendix S2. Outline of the experimental set-up in the
alterfishbehaviourbyinterferencewithneurotransmitter
field
function.NatureClimateChange2:201–204.
Parmesan,C.2006.Ecologicalandevolutionaryresponsesto Please note: Wiley-Blackwell are not responsible for the
recentclimatechange.Annu.Rev.Ecol.Evol.Syst. content or functionality of any supporting materials sup-
37:637–66. plied by the authors. Any queries (other than missing
Pratchett,M.S.,P.L.Munday,S.K.Wilson,N.A.J.Graham, material) should be directed to the corresponding author
J.E.Cinner,D.R.Bellwood,et al.2008.Effectsofclimate- for the article.
©2012TheAuthors.PublishedbyBlackwellPublishingLtd. 47