Ocean acidification alters fish – jellyfish

rspb.royalsocietypublishing.org symbiosis
Ivan Nagelkerken1, Kylie A. Pitt2, Melchior D. Rutte1
and Robbert C. Geertsma1
Research 1
Southern Seas Ecology Laboratories, School of Biological Sciences and The Environment Institute,
The University of Adelaide, DX 650 418, Adelaide, South Australia 5005, Australia
2
Cite this article: Nagelkerken I, Pitt KA, Rutte Griffith School of Environment and Australian Rivers Institute (ARI)—Coast and Estuaries,
MD, Geertsma RC. 2016 Ocean acidification Griffith University, Gold Coast campus, Queensland 4222, Australia
alters fish – jellyfish symbiosis. Proc. R. Soc. B
Symbiotic relationships are common in nature, and are important for
283: 20161146. individual fitness and sustaining species populations. Global change is
http://dx.doi.org/10.1098/rspb.2016.1146 rapidly altering environmental conditions, but, with the exception of coral–
microalgae interactions, we know little of how this will affect symbiotic
Downloaded from https://royalsocietypublishing.org/ on 06 June 2023

relationships. We here test how the effects of ocean acidification, from rising
anthropogenic CO2 emissions, may alter symbiotic interactions between juven-
Received: 24 May 2016 ile fish and their jellyfish hosts. Fishes treated with elevated seawater CO2
Accepted: 15 June 2016 concentrations, as forecast for the end of the century on a business-as-usual
greenhouse gas emission scenario, were negatively affected in their behaviour.
The total time that fish (yellowtail scad) spent close to their jellyfish host in a
choice arena where they could see and smell their host was approximately
three times shorter under future compared with ambient CO2 conditions. Like-
Subject Areas: wise, the mean number of attempts to associate with jellyfish was almost three
behaviour, ecology times lower in CO2-treated compared with control fish, while only 63% (high
CO2) versus 86% (control) of all individuals tested initiated an association at
Keywords: all. By contrast, none of three fish species tested were attracted solely to jelly-
global change, jellyfish blooms, juvenile fish, fish olfactory cues under present-day CO2 conditions, suggesting that the
altered fish–jellyfish association is not driven by negative effects of ocean acid-
commensalism, behaviour
ification on olfaction. Because shelter is not widely available in the open water
column and larvae of many (and often commercially important) pelagic
species associate with jellyfish for protection against predators, modification
Author for correspondence: of the fish–jellyfish symbiosis might lead to higher mortality and alter species
Ivan Nagelkerken population dynamics, and potentially have flow-on effects for their fisheries.
e-mail: ivan.nagelkerken@adelaide.edu.au

1. Introduction
Symbiotic relationships are ubiquitous in the terrestrial and marine realms
and sustain species diversity and various important ecological processes [1,2].
For positive symbiotic interactions, such as mutualism or commensalism, at
least one of the species involved benefits from the interaction. Such interactions
are important because they mediate the persistence of other species through facili-
tation, particularly in conditions of high disturbance, stress or predation [3].
Mutualism and commensalism provide benefits in terms of resources, such as
food, habitat and shelter, and such positive interactions are important drivers of
species population dynamics, species distributions and diversity, and community
organization [4].
Perhaps due to the vastness of the ocean, symbiotic relationships in the sea—
other than coral–zooxanthellae symbiosis—are less well studied and understood
than on land [5]. However, they occur globally in tropical, temperate and polar
regions, and cover a wide range of benthic and pelagic species [6]. A well-
known marine symbiosis is that between anemones and clown-fishes (e.g. [7]),
Electronic supplementary material is available popularized through the animated movie Finding Nemo. However, fishes also
form symbiotic relationships with other taxa such as jellyfish, the pelagic relatives
at http://dx.doi.org/10.1098/rspb.2016.1146 or
of anemones. Fish–jellyfish interactions are complex as they include predation by
via http://rspb.royalsocietypublishing.org. jellyfish on larval fish, predation by fishes on jellyfish, competition for

& 2016 The Author(s) Published by the Royal Society. All rights reserved.
 Table 1. Summary of the water chemistry parameters. Average (+s.e.) temperature (T), pH, salinity, total alkalinity (TA) and pCO2 (calculated using CO2sys), 2
with their respective sample sizes (n). kgSW ¼ kilograms of seawater.

rspb.royalsocietypublishing.org
TA pCO2
treatment T (88C) pHNBS salinity n (mmol kgSW21) (matm) n

yellowtail control 23.7 (0.0) 7.95 (0.01) 35.0 (0.0) 14 2331 (7) 740 (29) 4
scad high CO2 23.7 (0.0) 7.61 (0.01) 35.0 (0.0) 14 2369 (6) 1794 (31) 4
barramundi control 26.3 (0.1) 8.07 (0.01) 38.6 (0.1) 120 2515 (12) 673 (10) 4
high CO2 26.2 (0.1) 7.72 (0.01) 38.6 (0.1) 120 2521 (15) 1464 (31) 4
mulloway control 22.6 (0.1) 8.02 (0.01) 38.3 (0.1) 104 2532 (8) 734 (15) 4

Proc. R. Soc. B 283: 20161146

high CO2 22.6 (0.1) 7.65 (0.01) 38.1 (0.2) 104 2527 (12) 1757 (38) 4

zooplankton prey between jellyfish and fish, jellyfish acting as attain 100 m – 3 [20]. The three species of fish were selected for
intermediate hosts for fish parasites, and commensal relation- their variety of responses to jellyfish. Juveniles of the yellowtail
ships in which fishes receive protection from potential scad (Trachurus novaezelandiae) frequently associate with C. mosai-
Downloaded from https://royalsocietypublishing.org/ on 06 June 2023

predators [8–10]. Most fish species that associate with jellyfish cus [9], and the species was selected to determine whether
exposure to CO2 affected the ability of fish to interact with the jelly-
form a temporary association as juveniles and hence have not
fish. Mulloway (Argyrosomus japonicas) do not associate with
evolved elaborate protective mechanisms against the veno-
jellyfish, and barramundi (Lates calcarifer) actively avoid water con-
mous tentacles of jellyfish [8]. This association is particularly
taining the olfactory cues of jellyfish [21]. These last two species
advantageous for larval and juvenile fish species during their were selected to determine whether their ability to recognize
oceanic life stage, because any type of pelagic habitat that olfactory cues of jellyfish would be affected by elevated CO2.
provides protection from predation (e.g. driftwood, jellyfish, To assess the prevalence of jellyfish–fish associations, during a
floating seaweed) can minimize early-life mortality [11,12]. jellyfish bloom event we counted the number of fish associa-
Jellyfish frequently form very dense blooms, predominantly ted with 62 C. mosaicus in southern Moreton Bay, Queensland
during spring and summer [13], which often coincides with (17 October 2014). Jellyfish and their associated fish were captured
the timing of the pelagic larval phase of fish species [14]. using a dip net (70 cm diameter, 1 mm mesh). Jellyfish were
Many of the fishes that associate with jellyfish are commercially removed from the net and their bell diameter was measured
prior to them being released. The average bell diameter of
important, including pollock, jacks and trevallies [9,15]. By
the jellyfish sampled was 17.9 + 1.8 cm (s.d.), with a range of
hosting these species during their early life-history stages, jelly-
12.5–24.0 cm. During the field collections, only T. novaezelandiae
fish may thus benefit commercial fisheries. However, there is
were found associated with the jellyfish. All fish captured with
a gap in our understanding of how climate stressors might the jellyfish were transferred to a polystyrene container of seawater
affect mutualistic or commensal relationships in general [16], that was aerated continuously during transport to the laboratory
and fish–jellyfish associations in particular. Ocean warming and used for subsequent vision and olfaction experiments.
and acidification resulting from increasing anthropogenic CO2
emissions alter a wide array of critical behaviours in marine
vertebrates and invertebrates [17,18], but, with the exception (b) CO2 manipulation of yellowtail scad
of coral–zooxanthellae relationships, their effects on symbioses After capture from the field, fishes were directly transferred to a
have not been established. Here we test whether ocean acidifica- constant-temperature room set at 248C at Griffith University.
Fishes were randomly allocated to each of four 60 l plastic bins
tion can alter the symbiotic association between juvenile fish
filled with seawater (salinity 36 units) collected from the Gold
and their jellyfish host through visual as well as olfactory sen-
Coast Seaway on rising tide and passed through a sand filter.
sory modalities, as these senses are known to be affected by The fishes were acclimated in ambient seawater (i.e. ‘high pH’
high CO2. We show that fishes are not attracted to jellyfish olfac- and ‘low pCO2’, representing present-day conditions) in their
tory cues, but that ocean acidification reduces the combined respective bins for approximately 9 days before the CO2 treat-
visual/olfactory attraction of fishes towards their host jellyfish, ment (i.e. ‘low pH’ and ‘high pCO2’, representing future
causing them to spend more time away from the jellyfish’s pro- conditions) started. Two bins were allocated to each of two pH
tective umbrella and oral arms. Such altered interactions that treatments: approximately 7.9 ( present day) and approximately
increase predation risk may have consequences for early-life 7.6 (future scenario; table 1). The low-pH treatment was based
survival of pelagic species. on the RCP 8.5 emission scenario (business-as-usual scenario)
with a predicted mean (+s.d.) decrease in ocean surface pH of
approximately 0.33 + 0.003 units by 2100 compared with the
1990s [22]. The two bins of each treatment shared a 200 l
2. Material and methods header tank. In the low-pH treatment, CO2 was gradually elev-
ated in the header tank from ambient to future conditions over
(a) Study area and species a period of 2 days. pH was manipulated by bubbling pure CO2
We studied the effects of ocean acidification on interactions into a 10 l bucket, after which the enriched water was mixed
between the blue blubber jellyfish Catostylus mosaicus (Scyphozoa; (with a 200 l h21 pump) with that in the low-pH header tank
Rhizostomeae) and three species of fish. Catostylus mosaicus is a until the desired pHNBS of 7.6 was attained, as measured with
large medusa (maximum bell diameter approx. 35 cm) that a Mettler Toledo Five Go 2. During this mixing process, the out-
forms conspicuous blooms in estuaries and coastal waters of east- flow from the header tank to the fish holding bins was
ern Australia. During blooms, densities of adult jellyfish can interrupted for a few minutes. Addition of CO2-enriched water
exceed 1 m – 3 [19] and peak densities of small medusae can to the low-pH header tank was done twice daily (9.00 and
 19.00 h) during the duration of the experiment; the seawater in (e) Combined visual/olfactory choice tests 3
the ambient header tank was also replenished but with ambient The association between juvenile yellowtail scad and jellyfish was
seawater. A floating lid was placed on top of the water in the

rspb.royalsocietypublishing.org
tested at Griffith University in a glass aquarium. An aquarium of
header tank to minimize the exchange of gases with the atmos- 50  20  40 cm (L  W  H) was filled with either ambient or
phere and to maintain a constant pH. Water from the header CO2-treated water, of similar values as those in the fish holding
tanks was gravity-fed into the respective fish holding bins at a tanks to avoid measuring potential shock effects in the fish due
rate of approximately 120 ml min21 (approx. 173 l per day). to transfer to water of different chemistry. The jellyfish were not
Fishes were exposed to low- and high-pH treatment conditions, pre-adapted to the high-CO2 water, but observations did not
respectively, for 7–8 days, based on a minimum exposure of 4 reveal distinct differences in the (tethered) jellyfish behaviour
days to elicit effects of high CO2 on behaviour [23]. pHNBS inside between choice arenas for the different treatments. The water in
all fish bins was measured daily at 14.00 h (roughly in-between the choice tank was also of the same temperature as that of the
the morning and evening water refreshments). Oxygen concen- fish holding tanks (248C). One of two medusae (bell diameter
trations in the fish-holding tanks were measured twice daily 16.5 and 17.5 cm, respectively) was tethered in the top corner of

Proc. R. Soc. B 283: 20161146

using an optical dissolved oxygen sensor (Mettler Toledo the tank using a thin rope that was fixed loosely around the pedun-
OptiOx) that was calibrated daily. Oxygen saturation of the cle of the bell (electronic supplementary material, figure S1). If the
water exceeded 80% throughout the experiment. Fish were fed jellyfish did not pulsate actively during the experiment, then it was
daily with live Artemia as well as dried commercial fish feed. replaced with another jellyfish. A single fish (approx. 5 cm total
Excess food and debris was removed from tanks each morning. length) was introduced to the opposite lower corner of the tank
using a vertical tube and its movements recorded from the side
Downloaded from https://royalsocietypublishing.org/ on 06 June 2023

using a camcorder (HF R406 Legria, Canon, Japan) during

10 min. The seawater in the choice tank was replaced with fresh
(c) CO2 manipulation/larval rearing of barramundi and water (either of ambient or high CO2, depending on the treatment
mulloway being tested) for every second fish. Each fish was used only once
Fertilized mulloway and barramundi eggs were obtained from the and the number of replicates per treatment is reported in figure 2.
South Australian Research and Development Institute (first-
generation broodstock) and from a commercial hatchery (Robarra,
seventh-generation broodstock), respectively. Both species were (f ) Olfactory choice tests
reared simultaneously (in separate tanks) at the University of Ade- Olfactory response to jellyfish chemical cues was tested for yellow-
laide under current and future CO2 levels (table 1). Larvae were tail scad at Griffith University, and for barramundi and mulloway
exposed to their respective treatments until they had undergone at the University of Adelaide. All fish tested were approximately
metamorphosis and approached their settlement stage, which 1 cm in total length. Control yellowtail scad were deterred by jelly-
was 24–25 days post-hatching (dph) for mulloway and 19– fish olfactory cues (see Results) and therefore olfactory cues were
20 dph for barramundi. For each species and each of their two not considered as an important driver of the fish– jellyfish associ-
CO2 treatments, larvae were reared in four replicate 60 l food- ations. Owing to relatively low numbers of yellowtail scad
safe rearing tanks. Seawater was sourced from Gulf St Vincent caught and animal ethics considerations we used the remainder
(approx. 10 m depth, several kilometres offshore) and recirculated of the fish to perform the combined visual/olfactory choice test
in a closed system, with each rearing tank connected to its own 20 l (representative of what the fish would experience in situ) in
sump containing a biological filter, a protein skimmer (WG-308, which we did observe a high-CO2 effect on the fishes.
Boyu, China) and a UV sterilizer (UView, Blue Planet, China). All jellyfish used for the trials were collected from Moreton
Fishes were reared at ambient temperatures of their environment Bay, Queensland. For trials at Griffith University, jellyfish were
(table 1), and therefore differed between the two species as barra- collected the day before the trials and maintained in 50 l tanks.
mundi is a tropical species while mulloway is a temperate species. For trials at the University of Adelaide, four blue blubber jellyfish
Fish were fed ad libitum with rotifers for the first 12 days post- C. mosaicus were directly flown to Adelaide where they were
hatching, then with Artemia nauplii and a dry feed (Otohime, maintained in a 150 l aquarium and fed Artemia nauplii daily.
Japan) of increasing granule size as development progressed. Jellyfish olfactory cues were prepared immediately before each
High-CO2 treatment conditions were achieved by bubbling choice test by immersing one jellyfish in fresh seawater in a
CO2-enriched air into the rearing tanks and their sumps with a 20 l bucket for 1 h. The jellyfish cue water was then filtered
Pegas 4000 MF gas mixer (Columbus). pHNBS was measured through a 10 mm filter bag to remove stinging cells.
daily with a SG2-ELK SevenGo pH probe (Mettler Toledo) The responses of the three fish species to jellyfish olfactory cues
calibrated with a three-point calibration. were tested using a two-channel choice-flume (13  4 cm), devel-
oped by Gerlach et al. [27]. The flume allows pairwise choice
experiments in which individual fish can move freely between
water originating from two different sources that continuously
(d) Analysis of carbonate chemistry flows from one end to the other end of the flume. Dye tests were
The CO2 partial pressure in the seawater was calculated for all three conducted at least once per day to ensure distinct and parallel
fish species using measured values of temperature, salinity, pHNBS water flows in each flume channel, without any turbulence or
and total alkalinity (TA) in the fish holding tanks (table 1), using the eddies. Dixson et al. [28] observed that using either acidified or con-
software CO2SYS [24] with constants K1 and K2 from Mehrbach trol water in flume choice tests did not affect behavioural choices
et al. [25] and refit by Dickson & Millero [26]. Fewer salinity and thus water of ambient pH was used throughout all olfactory
measurements (table 1) were taken for yellowtail scad than for bar- choice trials. For each trial, water from the two different treatments
ramundi/mulloway, owing to the shorter treatment exposure time (control and jellyfish olfactory cues) was gravity-fed from buckets
in the former. Alkalinity was measured by Dynamic Endpoint into the choice-flume at a rate of approximately 100 ml min21. The
Titration using an 888 Titrando (Metrohm) titrator. During the water in the buckets was of the same temperature as that of the fish
course of the study, alkalinity values were within 1% accuracy of holding tanks. Trials were performed following Dixson et al. [28],
certified standards (reference materials from Dr A. Dickson, Scripps where an individual fish was placed at the downstream end of
Institution of Oceanography). Seawater samples for alkalinity were the flume and left to acclimate for 2 min while exposed to
analysed immediately upon collection. Salinity was measured with both water flows with different cues and able to explore both
a SR6 refractometer (Vital Sine). sides of the flume. After acclimation, the position of the fish was
 video-recorded for 2–2.5 min (depending on the species). This was 4
followed by a 1 min rest period during which the water cue sides
were switched (controlling for any side preferences) after which

rspb.royalsocietypublishing.org
the entire process including acclimation was repeated. The fish
movements during the trials were recorded from above using a
camcorder (HF R406 Legria, Canon, Japan) to avoid any observer’s
effects. Each fish was used only once and the number of replicates
per treatment is reported in figure 2.

(g) Analysis of fish behaviour

For the olfactory choice tests, the location of the head of the fish
(left versus right flume channel) was visually determined from

Proc. R. Soc. B 283: 20161146

the video recordings every 5 s for the period before as well as
after the side switch in cue water. Fish that did not explore
both channels of the chamber during the acclimation period
were not included in the analysis as these were not deemed to Figure 1. Association between juvenile fish (concentrated within circle) and
have assessed the choices available (n ¼ 1 control yellowtail jellyfish (Pseudorhiza haeckeli). Photo credit: James Brook.
scad; n ¼ 1 high-CO2 mulloway).
For the combined visual/olfactory choice test, the position of
Downloaded from https://royalsocietypublishing.org/ on 06 June 2023

the fish in the choice chamber was continuously and automatically of the chance of a fish occupying a zone (e.g. 25 : 25 : 25 : 25% in
tracked from the 10 min video recordings using ETHOVISION XT10 case of a fish being released in the centre of an aquarium at the
(Noldus Information Technology, Wageningen, The Netherlands). position where all four zones intersect when they are located at
The combined use of remote recordings and automated tracking equal distances from one another). Therefore, we simply tested
eliminated the risk of observer bias and external influences on be- against a 0.5 probability based on a behavioural outcome of
haviour caused by the presence of the observer. ETHOVISION either willing or unwilling to associate with the jellyfish.
calculated the amount of time the fish spent in the ‘jellyfish zone’
and the number of times the fish approached the jellyfish
(i.e. entered the jellyfish zone from zone 2). The jellyfish zone was
defined as a rectangle (side view, electronic supplementary
material, figure S1), with the width defined as the distance between
3. Results
the tank wall and the most distal part of the bell when the jellyfish Fish –jellyfish associations (figure 1) were observed in the
tether was stretched maximally, and height as the distance between field for 32% of the 62 jellyfish sampled. Jellyfish harboured
the water surface and the lowest point of the arms when the jellyfish 0–3 fish with a mean density of 0.5 + 0.7 (s.d.) fish per jelly-
tether was stretched maximally. Three other zones, located at fish. All associated fish were juvenile (size range approx.
increasing distances from the jellyfish, were also included (electro- 1–5 cm total length, with 89% less than 2.5 cm) yellowtail
nic supplementary material, figure S1), and all four zones were of scad (Trachurus novaezelandiae). There was no relationship
equal volume. between jellyfish bell size and number of associated fishes
(linear regression: R 2 ¼ 0.00, p ¼ 0.975).
(h) Statistical analyses Total time spent close to jellyfish in the choice arena was
For each of the two CO2 treatments, percentage of time spent at the approximately three times shorter (t-test, t31 ¼ 22.364, p ¼
flume side with jellyfish olfactory cues was compared to a random 0.025) for fish from the high-CO2 treatment compared with
50% distribution (i.e. equal distribution at both sides in case there the control treatment (figure 2a). Although fish in the high-
was no response to the cue) using either a one-sample t-test or a CO2 and control treatments only spent 2.6 + 1.4% and 8.0 +
Wilcoxon signed-rank test, depending on whether the data were 2.2%, respectively, of the total observation time in the jellyfish
normally distributed as tested with a Kolmogorov–Smirnov test zone, this difference was still significant (t31 ¼ 22.170, p ¼
of normality. 0.038) when standardized to total observation time. Time
For the combined visual/olfactory choice experiment, total spent in the other zones of the choice arena did not differ
time spent in each of the four zones of the choice arena was com-
between the two treatments (adjusted t-test, t26.4  1.347, p .
pared between fish from the ambient and elevated CO2 treatments
0.189). Nevertheless, there was a trend of a higher percent-
using an independent-samples t-test on log-transformed data. In
cases where variances were not homogeneous, the output of an age of time spent in the zone near the jellyfish by the
adjusted t-test was used, based on lower degrees of freedom (auto- high-CO2-treated fish compared with the control fish
matically calculated within IBM SPSS v. 20). The same statistical (electronic supplementary material, figure S2), suggesting
analysis was used to test for the number of fish transitions from that these fish showed some attraction to jellyfish but failed
zone 2 to the jellyfish zone. Percentage of time spent in the jellyfish in most cases to successfully bond.
zone and number of jellyfish approaches represent different behav- The mean number of attempts to associate with jellyfish was
ioural traits because a fish could, for example, approach the also approximately three times lower (t-test, t31 ¼ 23.173, p ¼
jellyfish once and remain associated with it for an extended 0.003) for CO2-treated compared with control fish (figure 2b).
period, or approach it multiple times but associate only for a Of all fish, only 63% of the individuals from the high-CO2 treat-
short period, both of which scenarios could lead to a similar
ment associated with jellyfish (bionomial test, p ¼ 0.096),
total time spent in the jellyfish zone.
whereas in the control treatment 86% ( p ¼ 0.006) associated
In addition, the percentage of all individuals from each treat-
ment that entered the jellyfish zone was calculated and tested with jellyfish.
with a binomial test. Because the four zones were not located Yellowtail scad held in ambient seawater conditions were
at equal distances from one another and because the fish were deterred by jellyfish olfactory cues in the flume choice test
not released at an equal distance from each zone (electronic sup- (figure 2c; p ¼ 0.002). The same was true for high-CO2-treated
plementary material, figure S1), there was no a priori expectation larval mulloway (Argyrosomus japonicus) ( p , 0.001) and for
 (a) time in jellyfish zone (b) jellyfish approaches CO2 through the same neurological pathway, leading to 5
modified behavioural responses, such as altered risk percep-
60 20

rspb.royalsocietypublishing.org
tion of poisonous jellyfish tentacles or reduced recognition of
50
* 15 jellyfish as a potential host. Alternatively, elevated CO2 could
40 make fishes less fearful of potential predators [36,39],

number
time (s)

*
30 10 venturing further away from protective shelter [23], and
20 thereby weakening the close association with the jellyfish
5
10 and its protective tentacles.
0 0 The fish –jellyfish association does not appear to be
(19/14) (19/14) driven by olfactory cues under ambient ( present-day) pCO2
(c) conditions. Our control flume choice experiment revealed
70 control that none of the three fish species tested were attracted

Proc. R. Soc. B 283: 20161146

high-CO2 towards jellyfish olfactory cues, and in most cases were
60
time in jellyfish cue (%)

deterred by the cue. This is consistent with previous obser-

50 vations for larval barramundi (deterrence) and for mud
* *
40 * * crab larvae and juvenile shrimp (both: no response) towards
olfactory cues of the same jellyfish species [21]. Together,
30
these results suggest that larval fish can detect jellyfish olfac-
Downloaded from https://royalsocietypublishing.org/ on 06 June 2023

20 tory cues but are not attracted by them. By contrast, in the

10 choice test where fish could see (and smell) the jellyfish, yel-
lowtail scad repeatedly approached and associated with the
0 jellyfish. This suggests that jellyfish detection or association
scad mulloway barramundi in nature is not driven by olfaction, but rather by other sen-
(32/0) (40/39) (33/33)
sing such as vision, thigmotaxis or negative phototropism [8].
Figure 2. Response of larval/juvenile fishes from control and high-CO2 treat- Altered fish–jellyfish symbiosis could potentially have
ments towards visual and olfactory jellyfish cues. (a) Mean (þs.e.) time important consequences for the population dynamics of the
juvenile scad spent close to jellyfish and (b) number of times scad respective fish species. Fish –jellyfish associations occur in
approached jellyfish, tested in an aquarium test arena. (c) Time spent by temperate as well as tropical ecosystems across more than
three fish species in jellyfish cue water, tested in a flume choice chamber. 80 species pairs and serve as a mechanism to reduce preda-
The dashed line shows the ‘no choice’ value based on a 50% chance of pres- tion on associated fish [9]. We found about a third of the
ence at either side of the flume, and asterisks show a significant choice as jellyfish in our study area to harbour one or more juvenile
tested with a one-sample t-test or Wilcoxon signed-rank test. Numbers fishes. It is typically juveniles of pelagic species such as
between brackets show number of replicate fish tested in control and scads, trevallies, medusafishes, driftfishes and butterfishes
high-CO2 treatments, respectively. that associate with jellyfish [8]. These fish species have a pelagic
lifestyle and therefore do not strongly associate with benthic
habitats. In an open water environment with little protection
control ( p ¼ 0.003) as well as CO2-treated ( p ¼ 0.002) larval
from predators, jellyfish aggregations can provide pelagic habi-
barramundi (Lates calcarifer). Mulloway larvae from the con-
tat that decreases mortality during the sensitive early-life stage.
trol treatment did not show any preference or deterrence to
Weakening of this symbiosis might thus lead to increased
the cues (figure 2c; p ¼ 0.411).
juvenile mortality, which might be particularly detrimental
for fish species or localities where this fish–jellyfish association
is common. However, further insights are needed to evaluate
the degree to which the relatively small decrease in time
4. Discussion spent near jellyfish by high-CO2-treated fish will result in
Our study is the first to show that non-coral symbiotic relation- increased mortality from predation compared with non-
ships in the ocean have the potential to be altered by ocean affected fish. Yellowtail scad and several other fishes (e.g.
acidification. In a choice experiment, fish that commonly walleye pollock, Atlantic bumpers, trevallies) that commonly
associate with jellyfish in nature approached their jellyfish associate with jellies are commercially harvested, and high-
host less frequently and spent less time close to their host CO2 disruption of fish–jellyfish associations could therefore
under future high-CO2 conditions than under present-day con- have flow-on effects for fisheries. Disruption of this symbiosis
ditions. Most fishes associating with jellyfish do not have any could even have cascading effects on higher trophic levels, as
immunity against jellyfish toxins [8], and therefore need to con- some seabird species specifically target fishes associated with
tinuously adjust their position relative to the pulsating jellyfish jellyfish [40], with attack rates increasing as a function of fish
host to avoid contact with the cnidocytes on their oral arms or density among jellyfish tentacles [41].
tentacles [15,29]. Ocean acidification can reverse or alter a wide Increases in jellyfish populations could offset the negative
range of animal behaviours mediated by vision, olfaction or effects of a weakening in the commensal relationship between
audition in fishes as well as invertebrates [18,30] by interfering fish and jellyfish. However, while claims of a global increase
with the brain neurotransmitter function [31,32]. The poten- in jellyfish populations have pervaded the scientific literature
tial for acclimation and adaptation in fishes towards ocean (e.g. [42]), a recent quantitative global meta-analysis showed
acidification effects remains elusive [33,34]. Visual acuity [35], that there was no difference in the number of jellyfish popu-
visual risk assessment [36], visual habitat association [37] lations that were increasing versus decreasing [43]. Indeed, if
and sheltering behaviour [38] are all altered by elevated CO2. fish associate with a species of jellyfish that is in decline, then
Hence, fish–jellyfish associations may be affected by high the effects of an impaired ability of fish to associate with
 jellyfish would be exacerbated. Much more information adds a magnitude to this response by showing the degree to 6
about the specificity of fish –jellyfish relationships is required, which fishes return to their jellyfish host after the first inspec-

rspb.royalsocietypublishing.org
as well as comprehensive understanding of jellyfish popu- tion (on average 12 times in control fish). Because this
lation trajectories to fully assess the implications for fish behavioural response was also reduced under high CO2 (only
populations of a potential weakening in the fish –jellyfish four times), the results for these two factors suggest an ecologi-
symbiosis. cally relevant response to jellyfishes by control fishes, and a
Experimental test arenas such as aquaria are often poor negative effect under high-CO2 conditions. However, how
representatives of natural environments and this is reflected this retarded response will scale up to population level and
in our results. Control fish only spent 8% of their time in fishes in a natural environment needs further consideration.
the jellyfish zone and were often swimming around. At By serving as hosts for a diverse suite of vertebrate and
least two potential factors might explain this behaviour: invertebrate organisms, jellyfish maintain an important role
(i) the absence of a natural predator, which could have in sustaining pelagic biodiversity. Our study suggests that

Proc. R. Soc. B 283: 20161146

resulted in lack of a trigger to seek shelter near the jellyfish; changing ocean conditions may impair the ability of sym-
and (ii) the fact that the fish were present in an artificial, con- bionts to associate with their jellyfish hosts, with potential
fined environment where there was likely an increased negative effects on pelagic biodiversity in general, and for
motivation to explore the novel environment and/or search fish in particular.
for a way out. Hence, the percentage of time associated with
the jellyfish is not the ecologically most relevant response vari-
Downloaded from https://royalsocietypublishing.org/ on 06 June 2023

able to evaluate the effect of high CO2 on fish behaviour. By Ethics. All experiments were performed under animal ethics approvals
ENV/16/14/AEC (Griffith University) and S-2013-153 (University of
contrast, (i) number of attempts to associate with the jellyfish Adelaide) and according to the universities’ animal ethics guidelines.
and (ii) the percentage of all individuals that initiated an associ- Data accessibility. Data are available from the Dryad Digital Repository:
ation are ecologically more relevant because both reflect the http://dx.doi.org/10.5061/dryad.9008g.
ability of fish to show regular association behaviour. Percen- Authors’ contributions. I.N. and K.A.P. conceived and designed the study.
tages individuals were measured on a binary scale (yes/no) M.D.R., R.C.G. and K.A.P. collected the data. I.N., M.D.R. and R.C.G.
and are therefore less fraught to test arena biases than per analysed the data. I.N. wrote the main manuscript and all authors
contributed to the writing of the article.
cent time (continuous scale). Of our control fish tested in the
aquarium (and which were all caught in the field in association Competing interests. The authors declare no competing interests.
with a jellyfish) 86% re-established their association with the Funding. This study was supported by an Australian Research Council
Future Fellowship (grant no. FT120100183) to I.N.
jellyfish showing a strong response reflective of natural behav-
Acknowledgements. We thank Tullio Rossi, Jennifer Pistevos and Wayne
iour. Of the high-CO2-treated fish only 63% of the individuals Hutchinson for help with larval rearing and measurement of water
re-associated with the jellyfish, suggesting a true effect of ocean chemistry parameters in Adelaide. Shannon Klein provided logistic
acidification. Number of attempts to associate with jellyfish support in the field and laboratory at Griffith University.

References
1. Boucher DH. 1985 The Biology of mutualism: Pomacentridae). Mar. Ecol. Progr. Ser. 188, collapse? In Jellyfish blooms (eds AK Pitt, HC Lucas),
ecology and evolution. Oxford, UK: Oxford University 207 –218. (doi:10.3354/meps188207) pp. 79 –103. Dordrecht, The Netherlands: Springer.
Press. 8. Mansueti R. 1963 Symbiotic behavior between 14. Russell B, Anderson G, Talbot F. 1977 Seasonality
2. Connor RC. 1995 The benefits of mutualism: a small fishes and jellyfishes, with new data on that and recruitment of coral reef fishes. Mar. Freshwater
conceptual-framework. Biol. Rev. Camb. Phil. Soc. between the stromateid, Peprilus alepidotus, and Res. 28, 521– 528. (doi:10.1071/MF9770521)
70, 427–457. (doi:10.1111/j.1469-185X.1995. the seyphomedusa, Chrysaora guinguecirrha. Copeia 15. Brodeur RD. 1998 In situ observations of the
tb01196.x) 1963, 40 –80. (doi:10.2307/1441273) association between juvenile fishes and
3. Hacker SD, Gaines SD. 1997 Some implications of 9. Purcell JE, Arai MN. 2001 Interactions of pelagic scyphomedusae in the Bering Sea. Mar. Ecol.
direct positive interactions for community species cnidarians and ctenophores with fish: a review. Progress Ser. 163, 11 –20. (doi:10.3354/
diversity. Ecology 78, 1990 –2003. (doi:10.1890/ Hydrobiologia 451, 27 –44. (doi:10.1023/ meps163011)
0012-9658(1997)078[1990:SIODPI]2.0.CO;2) a:1011883905394) 16. Gilman SE, Urban MC, Tewksbury J, Gilchrist GW,
4. Bertness MD, Callaway R. 1994 Positive interactions 10. D’Ambra I, Graham WM, Carmichael RH, Hernandez Holt RD. 2010 A framework for community
in communities. Trends Ecol. Evol. 9, 191 –193. FJ. 2015 Fish rely on scyphozoan hosts as a primary interactions under climate change. Trends Ecol. Evol.
(doi:10.1016/0169-5347(94)90088-4) food source: evidence from stable isotope analysis. 25, 325 –331. (doi:10.1016/j.tree.2010.03.002)
5. Bronstein JL. 1994 Our current understanding Mar. Biol. 162, 247–252. (doi:10.1007/s00227- 17. Briffa M, de la Haye K, Munday PL. 2012 High CO2
of mutualism. Q. Rev. Biol. 69, 31 –51. (doi:10. 014-2569-5) and marine animal behaviour: potential
1086/418432) 11. Kingsford MJ. 1993 Biotic and abiotic structure in mechanisms and ecological consequences. Mar.
6. Hay ME, Parker JD, Burkepile DE, Caudill CC, Wilson the pelagic environment: importance to small Pollut. Bull. 64, 1519–1528. (doi:10.1016/j.
AE, Hallinan ZP, Chequer AD. 2004 Mutualisms and fishes. Bull. Mar. Sci. 53, 393–415. marpolbul.2012.05.032)
aquatic community structure: the enemy of my 12. Bonaldo RM, Krajewski JP, Sazima I. 2004 Does the 18. Nagelkerken I, Munday PL. 2016 Animal behaviour
enemy is my friend. Annu. Rev. Ecol. Evol. Syst. association of young fishes with jellyfishes protect shapes the ecological effects of ocean acidification
35, 175–197. (doi:10.1146/annurev.ecolsys.34. from predation? A report on a failure case due and warming: moving from individual to
011802.132357) to damage to the jellyfish. Neotrop. Ichthyol. 2, 103– community-level responses. Glob. Change Biol. 22,
7. Arvedlund M, McCormick MI, Fautin DG, Bildsoe M. 105. (doi:10.1590/S1679-62252004000200008) 974–989. (doi:10.1111/gcb.13167)
1999 Host recognition and possible imprinting in 13. Pitt KA, Budarf AC, Browne JG, Condon RH. 2014 19. Pitt KA, Kingsford MJ. 2003 Temporal variation in
the anemonefish Amphiprion melanopus (Pisces: Bloom and bust: why do blooms of jellyfish the virgin biomass of the edible jellyfish, Catostylus
 mosaicus (Scyphozoa, Rhizostomeae). Fish. Res. 63, of reef fish larvae. Proc. Natl Acad. Sci. USA 104, GABA(A) receptors. J. Exp. Biol. 217, 323 –326. 7
303–313. (doi:10.1016/s0165-7836(03)00079-1) 858 –863. (doi:10.1073/pnas.0606777104) (doi:10.1242/jeb.092478)

rspb.royalsocietypublishing.org
20. Kingsford MJ, Pitt KA, Gillanders BM. 2000 28. Dixson DL, Munday PL, Jones GP. 2010 Ocean 36. Ferrari MCO, McCormick MI, Munday PL, Meekan
Management of jellyfish fisheries, with special acidification disrupts the innate ability of fish to MG, Dixson DL, Lonnstedt O, Chivers DP. 2012
reference to the Order Rhizostomeae. In detect predator olfactory cues. Ecol. Lett. 13, 68 – Effects of ocean acidification on visual risk
Oceanography and marine biology, vol. 38 (eds RN 75. (doi:10.1111/j.1461-0248.2009.01400.x) assessment in coral reef fishes. Funct. Ecol 26,
Gibson, M Barnes), pp. 85 –156. London, UK: 29. Bailey KM. 1984 Comparison of laboratory rates of 553–558. (doi:10.1111/j.1365-2435.2011.01951.x)
Taylor & Francis Ltd. predation on 5 species of marine fish larvae by 3 37. Devine BM, Munday PL. 2013 Habitat preferences of
21. Carr EF, Pitt KA. 2008 Behavioural responses of planktonic invertebrates—effects of larval size on coral-associated fishes are altered by short-term
zooplankton to the presence of predatory jellyfish. vulnerability. Mar. Biol. 79, 303–309. (doi:10.1007/ exposure to elevated CO2. Mar. Biol. 160,
J. Exp. Mar. Biol. Ecol. 354, 101–110. (doi:10.1016/ bf00393262) 1955– 1962. (doi:10.1007/s00227-012-2051-1)
j.jembe.2007.10.012) 30. Leduc A, Munday PL, Brown GE, Ferrari MCO. 2013 38. Rossi T, Nagelkerken I, Simpson SD, Pistevos JCA,

Proc. R. Soc. B 283: 20161146

22. Bopp L et al. 2013 Multiple stressors of ocean Effects of acidification on olfactory-mediated Watson S-A, Merillet L, Fraser P, Munday PL,
ecosystems in the 21st century: projections with behaviour in freshwater and marine ecosystems: Connell SD. 2015 Ocean acidification boosts larval
CMIP5 models. Biogeosciences 10, 6225 –6245. a synthesis. Phil. Trans. R. Soc. B 368, 20120447. fish development but reduces the window of
(doi:10.5194/bg-10-6225-2013) (doi:10.1098/rstb.2012.0447) opportunity for successful settlement. Proc. R. Soc. B
23. Munday PL, Dixson DL, McCormick MI, Meekan M, 31. Nilsson GE, Dixson DL, Domenici P, McCormick MI, 282, 20151954. (doi:10.1098/rspb.2015.1954)
Ferrari MCO, Chivers DP. 2010 Replenishment of fish Sorensen C, Watson SA, Munday PL. 2012 Near- 39. Nagelkerken I, Russell BD, Gillanders BM, Connell
Downloaded from https://royalsocietypublishing.org/ on 06 June 2023

populations is threatened by ocean acidification. future carbon dioxide levels alter fish behaviour by SD. 2016 Ocean acidification alters fish populations
Proc. Natl Acad. Sci. USA 107, 12 930 –12 934. interfering with neurotransmitter function. Nat. indirectly through habitat modification. Nat. Clim.
(doi:10.1073/pnas.1004519107) Clim. Change 2, 201–204. (doi:10.1038/ Change 6, 89– 93. (doi:10.1038/nclimate2757)
24. Pierrot DL, Wallace R. 2006 MS Excel program nclimate1352) 40. Duffy DC. 1988 Predator–prey interactions between
developed for CO2 system calculations. Oak Ridge, 32. Watson SA, Lefevre S, McCormick MI, Domenici P, common terns and butterfish. Ornis Scand. 19,
TN: Carbon Dioxide Information Analysis Center, Oak Nilsson GE, Munday PL. 2014 Marine mollusc 160–163. (doi:10.2307/3676466)
Ridge National Laboratory, US Department of predator-escape behaviour altered by near-future 41. Sato NN, Kokubun N, Yamamoto T, Watanuki Y,
Energy. carbon dioxide levels. Proc. R. Soc. B 281, Kitaysky AS, Takahashi A. 2015 The jellyfish buffet:
25. Mehrbach C, Culberson CH, Hawley JE, Pytkowicx 20132377. (doi:10.1098/rspb.2013.2377) jellyfish enhance seabird foraging opportunities by
RM. 1973 Measurement of the apparent dissociation 33. Munday PL, Warner RR, Monro K, Pandolfi JM, concentrating prey. Biol. Lett. 11, 20150358.
constants of carbonic acid in seawater at Marshall DJ. 2013 Predicting evolutionary responses (doi:10.1098/rsbl.2015.0358)
atmospheric pressure. Limnol. Oceanogr. 18, to climate change in the sea. Ecol. Lett. 16, 42. Richardson AJ, Bakun A, Hays GC, Gibbons MJ. 2009
897–907. (doi:10.4319/lo.1973.18.6.0897) 1488 –1500. (doi:10.1111/ele.12185) The jellyfish joyride: causes, consequences and
26. Dickson AG, Millero FJ. 1987 A comparison of the 34. Nagelkerken I, Connell SD. 2015 Global alteration of management responses to a more gelatinous future.
equilibrium constants for the dissociation of ocean ecosystem functioning due to increasing Trends Ecol. Evol. 24, 312– 322. (doi:10.1016/j.tree.
carbonic acid in seawater media. Deep Sea Res. Part human CO2 emissions. Proc. Natl Acad. Sci. USA 112, 2009.01.010)
A. Oceanogr. Res. Pap. 34, 1733 –1743. (doi:10. 13 272– 13 277. (doi:10.1073/pnas.1510856112) 43. Condon RH et al. 2013 Recurrent jellyfish blooms
1016/0198-0149(87)90021-5) 35. Chung WS, Marshall NJ, Watson SA, Munday PL, are a consequence of global oscillations. Proc. Natl
27. Gerlach G, Atema J, Kingsford MJ, Black KP, Miller- Nilsson GE. 2014 Ocean acidification slows retinal Acad. Sci. USA 110, 1000–1005. (doi:10.1073/pnas.
Sims V. 2007 Smelling home can prevent dispersal function in a damselfish through interference with 1210920110)