Journal of

Plankton Research academic.oup.com/plankt

J. Plankton Res. (2017) 00(00): 1–9. doi:10.1093/plankt/fbw096

Community cascades in a marine pelagic

food web controlled by the non-visual
apex predator Mnemiopsis leidyi
PETER TISELIUS1* AND LENE FRIIS MØLLER2

DEPARTMENT OF BIOLOGICAL AND ENVIRONMENTAL SCIENCES–KRISTINEBERG, UNIVERSITY OF GOTHENBURG, SE- FISKEBÄCKSKIL, SWEDEN AND

DANISH SHELLFISH CENTRE, DTU-AQUA, TECHNICAL UNIVERSITY OF DENMARK,  NYKØBING MORS, DENMARK

*CORRESPONDING AUTHOR: peter.tiselius@bioenv.gu.se

Received September 26, 2016; editorial decision December 8, 2016; accepted December 10, 2016

Corresponding Editor: Marja Koski

Trophic cascades are a ubiquitous feature of many terrestrial and fresh-water food webs, but have been difficult to
demonstrate in marine systems with multispecies trophic levels. Here we describe significant trophic cascades in an
open coastal planktonic ecosystem exposed to an introduced top predator. The ctenophore Mnemiopsis leidyi was
monitored for an 8-year period concurrent with measures of the food web structure of the plankton and strong
trophic cascades were evident. In the 5 years when M. leidyi were found, their target prey (grazing copepods) were
reduced 5-fold and the primary producers doubled their biomass when released from the grazing pressure. The
increased phytoplankton biomass could unequivocally be assigned to grazing release since concurrent measurements
of primary production did not differ between years with or without M. leidyi. Copepod biomass prior to the mass
occurrence of the ctenophore was important. The years without M. leidyi had significantly higher biomass of cope-
pods in July, the month preceding the outburst of the ctenophore. The profound changes of the pelagic ecosystem
faced with a non-selective apex predator shows that marine communities are not exceptions from trophic cascade
mechanisms.

KEYWORDS: trophic cascades; predation; ctenophore; phytoplankton; copepods; ciliates; Mnemiopsis leidyi

INTRODUCTION
Predation is the strongest selective force acting on popu-
lations and it is therefore not surprising that the most
visible controlling factor in food webs is the presence or
absence of predators. Numerous studies have shown
top-down control by apex predators and the resulting
trophic cascades can propagate all the way to the bot-
tom of the food webs. There are two fundamentally dif-
ferent cascades, those among single species and those
among trophic levels where the levels are considered to

available online at academic.oup.com/plankt

© The Author 2017. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
 JOURNAL OF PLANKTON RESEARCH j VOLUME 
be communities of similarly feeding animals (Polis et al.,
2000). With selective predators feeding on single prey
species, cyclic systems will evolve alternating between
top-down and bottom-up control. The numerous exam-
ples of trophic cascades in lakes involve one or a few
species at each trophic level and they are also observed
because lakes have limited immigration or emigration
(Sommer, 2008). In the marine environment, many spe-
cies typically occupy each trophic level and this redun-
dancy often prevents observation of trophic cascades
over more than two trophic levels (Shurin et al., 2002).
In order to show cascades in the marine pelagic environ-
ment, the predation needs to be strong and non-
selective such that the predator can supress an entire
trophic level (Polis et al., 2000; Baum and Worm, 2009).
The time-span of observation also needs to be shorter
than in lakes due to the transient nature of the environ-
ment. The top-down effects may in fact be as strong as
in a lake, but when averaged over a long time they are
obscured by advection. Finally, it must be remembered
that the definition of a trophic level is not always clear-
cut. Part of the difference between lakes and the sea
might be the better definition of distinct trophic levels in
lakes (e.g. vertebrate predators), but a weaker definition
of lower trophic levels in general, and in particular mar-
ine pelagic food webs (Angus Atkinson, pers. com.).
There are a number of studies showing community
cascades in laboratory or mesocosm experiments from
the marine environment (e.g. Granéli and Turner,
2002; Stibor et al., 2004; Sommer and Sommer, 2006).
Their strength is that they are causative since they are
manipulated in only one factor, typically addition or
removal of top predators. The design of the experiments
can be said to mimic lakes in that advection is removed
and the predation is high. Therefore, it is not surprising
that results convincingly show trophic cascades over two
or three trophic levels. However, even in manipulative
experiments it is difficult to show effects over three
trophic levels (Micheli, 1999). Predation on more than
one trophic level (omnivory) or redundancy within each
trophic level will obscure the relationships. Perhaps
most convincing support for this effect is the observation
of a switch between three- or four-level food webs
caused by induced behaviours of the grazers (Stibor
et al., 2004). Field evidence of trophic cascades in marine
plankton is rare and in principle restricted to systems
with strong jellyfish predation. In Narragansett Bay and
in Danish fjords, time series of jellyfish, copepods and
chlorophyll have shown a strong predation effect and
also signs of cascades (e.g. Møller and Riisgård, 2007;
Sullivan et al., 2007). They resemble lake ecosystems in
their long residence times and species level interactions.
The cascades are, however, not statistically tested and

j NUMBER  j PAGES – j 
there is still a clear need for unequivocal field evidence
of trophic cascades in the marine plankton.
Increasing management of nature by manipulation of
predators calls for a better knowledge of the response of
the pelagic food to apex predators (Baum and Worm,
2009). Ecosystem-based management is a leading prin-
ciple, but it has still to show its applicability to marine
coastal waters. Since it is generally implemented by
altering the fishing pressure of one selected species, it
relies on a cascading effect of the same sort as in lakes,
i.e. a species trophic cascade, and it is implicit that com-
munity cascades follow the same principle. The critical
test is whether a changing predation regime in the field
will affect the assembly of zooplankton and result in a
significant effect on the field assembly of phytoplankton.
To test the hypothesis of community cascades, we can
use the natural variation of predators on zooplankton
and study periods of high predatory abundance and
contrast them with periods of few predators. This, how-
ever, requires a careful and long-time series of relevant
sampling at several trophic levels. The approach may be
more relevant than mesocosm experiments for the situ-
ation in the field (Atkinson et al., 2014), but is also very
costly (Baum and Worm, 2009).
To test the hypothesis that community cascades may
control a marine plankton assemblage, we investigated
an open coastal planktonic ecosystem exposed to an
introduced top predator, the ctenophore Mnemiopsis leidyi.
Mnemiopsis leidyi is one of the best-studied species
involved in top-down control in marine planktonic eco-
systems. Estuaries along the American east coast are the
native areas for the species and life cycles (Purcell et al.,
2001; Costello et al., 2006) and predator-prey interac-
tions (e.g. Kremer,1979; Durbin and Durbin, 1981;
Purcell and Decker, 2005; Sullivan et al., 2007) have
been studied in depth. Since the introduction of the spe-
cies to Scandinavian waters in 2006 (Hansson, 2006),
the ctenophore has been intensively studied due to con-
cern about detrimental effects on the pelagic food web
(e.g. Jaspers et al., 2011; Haraldsson et al., 2013).
The sampling was done at a station at the entrance of
the Gullmar Fjord, a permanently stratified fjord with a
maximum depth of 120 m and a water depth of 37 m at
the sampling station. The upper mixed layer of the fjord
has a residence time of 16 days (Arneborg, 2004), and
advection is important for the plankton composition
(Lindahl and Perissinotto, 1987). Calanus finmarchicus and
Calanus helogolandicus are important for the total zoo-
plankton biomass (Lindahl and Hernroth, 1988), but
smaller species dominate numerically (Tiselius et al.,
2016). The predation impact from chaetognaths (Sagitta
setosa) is significant (Tönnesson and Tiselius, 2005), but
confined to the period September–October; up to
 P. TISELIUS AND L.F. MØLLER j COMMUNITY CASCADES IN A MARINE PELAGIC FOOD WEB
48% d−1 on Pseudocalanus sp., 29% d−1 on Paracalanus
parvus and 7% d−1 on Acartia clausi. In the deeper part of
the fjord, Sagitta elegans can remove 5–18% of the
Calanus sp. population daily.
In this study, M. leidyi was monitored for an 8-year
period concurrent with measures of the food web struc-
ture of the plankton and strong trophic cascades were
evident. We discuss the development of the M. leidyi
population in relation to zooplankton and point to the
dual controlling factors for the initiation of rapid growth
of the predator population. We conclude that a top
predator may control even the productive waters of the
Swedish west coast.
METHOD
Data for this study comes from sampling at the perman-
ent sampling station at the entrance of the Gullmar Fjord
(58°15,6′N, 11°27,2′E). The data for primary production
and chlorophyll were sampled every 2 weeks and the zoo-
plankton and M. leidyi at variable frequencies from every
second day to once each month from September 2007 to
December 2014. The measurement of primary produc-
tion was part of the national monitoring programme
described in detail by Tiselius et al. (2016). Briefly, incorp-
oration of 14C-labelled carbonate was recorded in
100 mL glass bottles filled with water from 0, 1, 2, 3, 4,
6, 8, 10 and 15 m and suspended at the same depths for
4 h during midday. After retrieval, the samples were acid-
ified and remaining carbonate removed by gentle bub-
bling and the incorporated 14C analysed on a Beckmann
scintillation counter. We used the full water fraction,
which include released dissolved organic matter during the
incubation, and primary production (mg C m−2 d−1) was
estimated after subtraction of dark bottles at 0 and
15 m. Chlorophyll a (mg m−3) was analysed by filtering
100 mL of water from the same depths through 0.7 µm
GF/F filters, extracted in ethanol for 24 h and measured
on a Turner 10-AU fluorometer. Taxonomic analysis of
phytoplankton was done through the national environ-
mental monitoring programme and data were retrieved
from the SMHI database Sharkweb (http://www.smhi.
se/klimatdata/oceanografi/havsmiljodata). Monthly av-
erages of total carbon concentration of diatoms (mg C m−3)
were retrieved for samples collected at the station in
September and October each year.
Zooplankton was sampled using duplicate vertical
tows (0–20 m) with a 90 µm WP-2 net (57 cm diameter)
equipped with a flow metre and preserved in 4% buf-
fered formaldehyde. After subsampling, the plankton
was analysed to groups or, in case of copepods, to spe-
cies level. 100–300 copepods were counted and 10

representative animals for each species was measured
for cephalothorax length and converted to carbon con-
tent (µg C ind.−1) using appropriate length–weight
regressions (Mauchline, 1998). Ctenophores were
sampled by duplicate oblique tows (0–20 m) with a
450 µm WP-3 net (113 cm diameter) equipped with a
flow metre. All animals in the sample were analysed
live, typically within 1–2 h and the oral–aboral length
(L, mm) measured on each individual. Biomass (wet
weight (WW) g) was then estimated for each individual
using a length–weight regression established from 32
individuals of various lengths:
WW = 0.0146*L1.9, R 2 = 0.92.
Predation impact of M. leidyi was estimated based on
reported gut contents and digestion times from the
Gullmar Fjord in September 2008 (Granhag et al.,
2011). The gut content method alleviates the negative
effect of container size that all incubation methods have
and should therefore give a more correct estimate of
predation. We used their regression relating clearance
rate (FC, L ind.−1 h−1) to oral–aboral length of M. leidyi
(LOA, mm) with Acartia sp. as prey
FC = 0.0022 L OA 2.33.
The average size M. leidyi at each sampling was used
for the regressions and the estimated individual clear-
ance was subsequently multiplied by total abundance of
ctenophores (ind. m−3) to yield a population clearance
rate (L m−3) expressed as the fraction of the water vol-
ume from 0 to 20 m cleared of calanoid copepods per
day (d−1). This approach to estimate predation rate is
affected by temperature and prey density, but we used
the estimate as an average for the autumn in the area
since Granhag et al. (2011) performed their study at the
same station and period as our sampling.
Statistical analysis
The data consisted of time series of primary production
(mg C m−2 d−1), chlorophyll (average 0–10 m, mg m−3),
biomass of calanoid copepods (copepodites and adults
only, mg C m−3) and biomass of M. leidyi (WW, g m−3)
for the time period July 2007–December 2014. We con-
sidered years as replicates and the time series was
divided into years with M. leidyi (2007–2010, 2014) and
years without (2011–2013). Many factors may differ
between years and increase the variance of the esti-
mates. We therefore consider our method of using
ANOVA as conservative. Since M. leidyi only occurred
in the autumn, we used data from August–December
 JOURNAL OF PLANKTON RESEARCH
and compared monthly averages for the years with M.
leidyi with years without using two-factor ANOVA (fac-
tors: month and the presence of M. leidyi). We also com-
pared biomass of M. leidyi between years for those years
when it occurred using two-factor ANOVA (factors:
year and month). The effect of copepods on early devel-
opment of M. leidyi was tested using two-sample t-tests,
comparing copepod biomass in July (the month preced-
ing the first appearance of M. leidyi in the samples) in
years with and without M. leidyi. Copepod grazing
effects on diatom biomass were tested using two-sample
t-tests, comparing years with M. leidyi (low copepod bio-
mass) with years without M. leidyi (high copepod bio-
mass). Data were log transformed to homogenize
variances and Student-Newman–Keuls post-hoc tests
were used when significant effects were detected by the
ANOVA. Significance was assumed at P ≤ 0.05.
RESULTS
Mnemiopsis leidyi was only observed during autumn in
the Gullmar Fjord and reached their highest biomass
Fig. 1. (A) Biomass of M. leidyi (g WW m−3, filled symbols) and of calanoid copepods (mg Cm−3, open symbols). (B) Specific predation impact
on calanoid copepods from M. leidyi (d−1). Error bars = SE.
j VOLUME  j NUMBER  j PAGES – j 
in September (Fig. 1A). There was a significant differ-
ence in biomass between years (ANOVA, F4,120 = 2.4,
P < 0.001) with elevated biomass in 2009 and 2014, (up
to 80 and 58 g WW m−3, respectively), intermediate bio-
mass in 2008 and significantly lower biomass in 2007
and 2010. Between February 2011 and July 2014, no
animals were sampled by our net, except on 26 October
and 5 Decemeber 2013 when single animals were
observed from the ship and 0.2 g WW m−3 was esti-
mated. For the statistical analysis and the categorization,
however, we have considered also 2013 as a year with-
out M. leidyi.
The occurrence of calanoid copepods showed a sea-
sonal variation reaching the highest biomass in July or
September (160 mg C m−3) and an occasional peak in
late spring. The autumn biomass was severely reduced
in years of M. leidyi occurrence, and consistently higher
in the autumns of 2011–2013, the years without M. leidyi.
Since the appearance of M. leidyi was restricted to the
autumn, only those data have been used for the further
comparisons.
Mnemiopsis leidyi was never recorded in July, but already
in August the average biomass reached a high level

 P. TISELIUS AND L.F. MØLLER j COMMUNITY CASCADES IN A MARINE PELAGIC FOOD WEB

(11 g WW m−3, Fig. 2A) increasing further in September
to a highest average of 19 g WW m−3. Biomass then
declined and the last observations occurred in December.
The copepod biomass responded dramatically to the
presence of M. leidyi. Without the ctenophore, biomass was
highest in September–November reaching 45 mg C m−3,
with lower values both in August and December. When
M. leidyi was present, biomass of calanoid copepods was
one order of magnitude lower, 3–6 mg C m−3, and rela-
tively constant throughout the autumn.
Chlorophyll a concentrations were relatively constant
in years without M. leidyi, ranging from 1.2 to 2.1 mg m−3.
Concurrent with the reduced biomass of copepods in
years with M. leidyi, concentrations were twice as high,
ranging from 1.8 to 4.3 mg m−3. Primary production
declined steadily from 800 mg C m−2 d−1 in August to
<35 mg C m−2 d−1 in December, and there was no dif-
ference between years with and without M. leidyi.
presence/absence (“predator”) as factors. The statistical
analysis confirmed the pattern in Fig. 2, where copepods
were significantly reduced by M. leidyi (Table IA) and
chlorophyll a was significantly higher when copepod
grazers were removed (Table IB). Since M. leidyi did not
have an effect on primary production (Table IC), it can
be concluded that the reason for the increased chloro-
phyll a in the presence of M. leidyi is due to decreased
grazing. In fact, the chlorophyll levels continued to
increase despite a significantly decreasing primary pro-
duction (Fig. 2), further supporting grazing as a signifi-
cant regulatory factor for the phytoplankton.
Mnemiopsis leidyi also had an indirect effect on the
phytoplankton composition through the removal of
copepod grazers (Fig. 3). In years with M. leidyi, the dia-
tom biomass was three times higher (two-sample t-test,
t14 = 2.5, P = 0.025), corroborating a selective grazing
effect of the copepods.

Trophic cascades
In contrast to the general practice of using correlation
approaches to show negative relations between adjacent
trophic level biomasses, our data allow a stringent statis-
tical test of the effect of M. leidyi on the food web. Using
log-transformed data, we carried out regular two-factor
ANOVAs with months (“month”) and M. leidyi the
DISCUSSION
The recurring mass occurrence of M. leidyi resulted in a
trophic cascade spanning three trophic levels. One pre-
requisite for a cascade to occur is that the predation
pressure is strong enough to significantly affect the prey
population. Estimated specific predation ranged from

A B

C D

Fig. 2. (A) Biomass of M. leidyi (g WW m−3), (B) Biomass of calanoid copepods (mg C m−3), (C) Chlorophyll-a concentrations (average 0–10 m,
mg m−3), (D) Primary production (mg C m−2 d−1). Open symbols: data from years without M. leidyi (2011–2013), filled symbols: data from years
with M. leidyi (2007–2010, 2014). Error bars = SE.


 JOURNAL OF PLANKTON RESEARCH j VOLUME  j NUMBER  j PAGES – j 

Table I: Trophic cascades. ANOVA of 180

the effect of M. leidyi (“Predator”) and 160
month on calanoid copepods, chlorophyll a and
primary production. All variables are log- 140

Diatom biomass, mg C m–3

transformed 120
Source df MS F p
100
Calanoid copepods
Predator 1 10.9 51 <0.001
80
Month 4 0.44 2.0 0.088
Predator × month 4 0.33 1.5 0.187
Error 154 0.21 60
Chlorophyll a
Predator 1 0.56 10.1 0.002 40
Month 4 0.25 4.5 0.003
Predator × month 4 0.028 0.51 0.73 20
Error 56 0.056
Primary production 0
Predator 1 0.084 0.75 0.39 With Without
Month 4 4.49 40.4 <0.001 Mnemiopsis Mnemiopsis
Predator × month 4 0.028 0.51 0.91
Error 56 0.11
Fig. 3. Biomass of diatoms (mg C m−3) in September and October in
years with and without M. leidyi. Symbols indicate average concentra-
tion of all samples of one month and the horizontal bar is the median
of the data for each period.
10 to 100% of the water column cleared of calanoid
copepods daily in 2009 and 2014, 5–50% in 2008 and
1–20% in 2007 and 2010 (Fig. 1B). This is a strong prey for larval ctenophores and since copepods are also
predation pressure compared with the specific growth strong predators on ciliates, there could be competition
of copepods in the autumn, which has been reported to and even a critical threshold biomass of copepods above
5–40% d−1 (Calliari and Tiselius, 2009) and 5–11% d−1 which ciliates will be limiting for ctenophore population
(Kiørboe and Nielsen, 1994) based on egg production. growth. The tipping point hinges on the copepods and if
It is therefore evident that the ctenophore predation was some other predator rapidly reduces the copepod bio-
able to control the biomass of copepods in years with mass, this may initiate a M. leidyi mass occurrence.
M. leidyi. Declines in copepod biomass before the M. leidyi
Preceding the first occurrence of M. leidyi, the cope- increase have been observed previously. Kremer (1979)
pod biomass declined rapidly which cannot be was the first to suggest an uncoupling between cteno-
explained by ctenophore predation. In our study, years phore predation and copepod declines. She observed
with the strongest decline and lowest biomass of cope- that copepods declined, perhaps due to food limitation
pods in July coincided with a rapid growth of the M. leidyi, or some other predator, well in advance of the M. leidyi
while years with a considerable biomass of copepods in outbreak. Using historical data, Sullivan et al. (2007)
July never developed mass occurrences. Indeed, the bio- showed that in the majority of observations, the cope-
mass of copepods was significantly higher in July in years pods declined in advance of the M. leidyi outbreaks.
when M. leidyi did not develop (two-sample t-test on log- Javidpour et al. (2009) observed an increase of M. leidyi
transformed biomass, t38 = −2.6, P = 0.013). We suggest larvae after a copepod decline and a ciliate increase,
that the biomass of copepods is a crucial factor for the supporting this mechanism. McNamara et al. (2013)
onset of rapid population growth of M. leidyi, and further- identified microzooplankton as important for M. leidyi
more, that the ciliates play an important role as suggested larvae development and a decline in copepods pre-
by McNamara et al. (2013). ceded the ctenophore mass occurrence. Collectively,
If the copepods are very abundant, the ciliate biomass these studies and our data suggest that an indirect effect
is kept low and there is a shortage of food for M. leidyi of decreased copepod predation on ciliates could lead
larvae (Fig. 4A). If the copepod biomass is reduced by to abundant microzooplankton food for ctenophore lar-
predation from other predators than M. leidyi or from vae and a rapid growth of the M. leidyi population,
starvation (Durbin and Durbin, 1981), the predation which then would control the copepod biomass. We do
pressure on ciliates decreases and they may increase their not know why M. leidyi appear in the fjord only in some
populations allowing a larger food source for ctenophore years, but food web interactions and feedback loops
larval growth (Fig. 4B). Ciliates are one of the prime may be important factors in addition to advection.


 P. TISELIUS AND L.F. MØLLER j COMMUNITY CASCADES IN A MARINE PELAGIC FOOD WEB

A Without Mnemiopsis B With Mnemiopsis

Adult Mnmeiopsis

Mnemiopsis larvae,
copepods

Diatoms,ciliates

Nano-flagellates

Picoplankton

Fig. 4. A suggested scenario how copepods and M. leidyi may affect the lower trophic levels of the food web. (A) In years without M. leidyi the
copepods are abundant controlling the ciliate biomass and reducing the larger phytoplankton. This creates a flagellate dominated base for the
food web. (B) When M. leidyi removes the copepods, ciliates thrive and reduce flagellates while supporting M. leidyi larval growth. In order to start
the rapid reproduction of M. leidyi, a sufficient fraction the copepod biomass needs to be removed to relieve ciliates from their predation pressure.
Green arrows indicate changes in phytoplankton biomass, blue arrows indicate changes in zooplankton biomass.

Chlorophyll concentrations were two times higher
in years with M. leidyi indicative of trophic cascades
mediated through the copepods. The chlorophyll esti-
mate is a composite of all phytoplankton larger than
0.7 µm and include pico-, nano- and microphytoplank-
ton and because the food web of organisms smaller
than copepods include both autotrophs and hetero-
trophs, we cannot trace a cascade through the smaller
sized organisms in our study. Ciliates are the domin-
ant grazers on autotrophic and heterotrophic nano-
plankton, and also a main prey for copepods, and
ciliates should be more abundant during M. leidyi years
along with the autotrophic microplankton. Our study
did not include the microzooplankton, however, and
we can only speculate on this. The trophic pathway
copepods-ciliates-autotrophic nanoflagellates should
act to dampen the effect that the M. leidyi had on total
chlorophyll concentrations (sensu Boyce et al., 2015),
but this was apparently overwhelmed by the positive
effect on the larger diatoms. The grazing by copepods
seems to have affected the diatoms, which were three
times more abundant when the copepods were reduced
during M. leidyi years (Fig. 3). If we apply carbon spe-
cific grazing rates based on gut fluorescence (Calliari
and Tiselius, 2009) to the copepod biomass in this
study, a substantial grazing pressure on the phyto-
plankton is evident in years without M. leidyi (Table II).
119% of the primary production and 30% of the
chlorophyll a could be grazed daily by the copepods,
but this was reduced to 11% and 2%, respectively,
Table II: Grazing rate (±SE) of copepods on
the phytoplankton during years with and with-
out M. leidyi
With Without
Mnemiopsis Mnemiopsis
Primary production, mg C m−3 d−1 26 ± 12 19 ± 8
Chlorophyll a, mg C m−3 140 ± 25 77 ± 5
Copepod grazing, mg C m−3 d−1 3 ± 0.3 23 ± 1.5
Weight specific grazing rate was set to 0.55 ± 0.12 d−1 (µg C µg C−1,
±95% confidence interval, n = 12), using data from October at the same
station (Calliari and Tiselius, 2009). Average biomasses and production
rates (±SE) from Fig. 2 was used, and chlorophyll a was converted to car-
bon using a C:Chl a ratio = 40.
when M. leidyi had reduced the copepods. We conclude
that the copepods were able to control the biomass of
phytoplankton cropping more than the primary pro-
duction in years without ctenophores, while phyto-
plankton population growth was independent of
copepod grazing in M. leidyi years. Similar trophic cas-
cades have been observed in Narragansett Bay, where
diatom blooms occurred as a response to strong cteno-
phore predation on the copepods (Durbin and Durbin,
1981; Deason and Smayda, 1982) and in nearby
Limfjord (Denmark), where scyphozoan and ctenophore
predation resulted in increased chlorophyll a levels
(Møller and Riisgård, 2007; Riisgård et al., 2012).
Community cascades spanning three trophic levels
have rarely been shown in marine pelagic food webs
(but see Casini et al., 2008), and most marine examples


 JOURNAL OF PLANKTON RESEARCH j VOLUME 
are from experimental manipulations of mesocosms
(e.g. Granéli and Turner, 2002; Stibor et al., 2004;
Sommer and Sommer, 2006). In a meta-analysis of
both manipulation experiments and field time series,
Micheli (1999) only found weak links to support the
trophic cascades hypothesis. Failure to observe cascades
in community approaches may be due to size variations
at each trophic level (Boyce et al., 2015) and can be
stronger when two complimentary grazers co-occur
allowing a wider prey size spectrum (Sommer, 2008). In
our study, M. leidyi acts like a generalist predator remov-
ing an entire functional group from the food web and
this may be the reason why we could statistically show
cascading effects in the highly variable coastal environ-
ment. As an apex predator, M. leidyi occupies a similar
role as humans do in the fishery; they sweep clear an
entire trophic level with dramatic consequences for the
ecosystem.
As a final consideration, our practice of using years as
replicates can be discussed. Most evidence of trophic cas-
cades in marine pelagic environment comes from experi-
ments in mesocosms limited in volume and time. As
advection is removed, the species diversity gradually
declines and cascading effects start to appear as the system
gradually turns from community interactions into species
interactions. In contrast, the field approach requires sev-
eral years of repeated introduction of the predator to
reduce the confounding factors. Fishery data spanning
decades have shown species cascades (e.g. Worm and
Myers, 2003; Casini et al., 2008), but there are no similar
data on community cascades in the plankton. In our
study, 5 years with predators were contrasted with 3 years
without, and we were able to show significant trophic cas-
cades. In conclusion, we suggest that there should be no
principal difference between controlling factors in marine
and limnic environments, the reason why they are hard to
observe in the sea is that the diversity at each trophic level
is higher (Pershing et al., 2015) and that advection is stron-
ger, both acting to dampen the direct predation effects.
Consequently, we have underestimated the importance of
top-down control in marine pelagic systems and con-
founding effects in time series leave bottom-up control as
a more important mechanism. This has lead to a domin-
ance of resource control in management practices, such as
mitigation of nutrient discharges, when in fact the entire
ecosystem should be considered for restoration.
CONCLUSIONS
Community trophic cascades were observed in a coastal
pelagic ecosystem following the introduction of the apex
predator M. leidyi. Using environmental monitoring data

j NUMBER  j PAGES – j 
on primary production, chlorophyll and zooplankton
biomass, years with and without the ctenophore were
compared.
• Mnemiopsis leidyi caused significant reductions in cala-
noid copepod biomass, which resulted in increased
chlorophyll levels.
• Diatoms dominated the increased phytoplankton
biomass.
• Mnemiopsis leidyi had no effect on primary production,
which proved that the phytoplankton increase was
caused by released grazing pressure.
Copepods seemed to be involved in the initiation of
the M. leidyi outbreaks. In years with a high copepod
biomass in July, no M. leidyi populations developed. A
scenario is suggested where strong predation on ciliates
by the copepods reduces the food for the M. leidyi lar-
vae, thus keeping them from rapid growth.
ACKNOWLEDGEMENTS
Numerous people contributed to the time series through
the years, but we particularly thank Pia Engström, Hans
Olsson and Lars Ljunqvist for collecting the data on pri-
mary production. We thank the crew onboard R/V
Oscar von Sydow and R/V Skagerrak and Fayez
Alsaleh for analysing zooplankton samples. Comments
from Angus Atkinson and two anonymous reviewers sig-
nificantly improved the manuscript.
FUNDING
The Swedish Agency for Marine and Water Management
(contract number 1256-14) ; the Swedish Environmental
Protection Agency (contract number 899-12); the Swedish
Research Council for Environment; Agricultural Sciences
and Spatial Planning (contract numbers 2007-1105,
217-2008-917, 2008-1882 and 2008-1889). The long-
term primary production series has been supported by
the Swedish Environmental Protection Agency; the County
Administrative Board of Västra Götaland and Bohuskustens
Vattenvårdsförbund. The Wåhlströms Foundation;
Wilhelm and Martina Lundgren Foundation and the
Adlerbertska Research Foundation are acknowledged
for their support.
REFERENCES
Arneborg, L. (2004) Turnover times for the water above sill level in
Gullmar Fjord. Cont. Shelf Res., 24, 443–460.
 P. TISELIUS AND L.F. MØLLER j COMMUNITY CASCADES IN A MARINE PELAGIC FOOD WEB
Atkinson, A., Hill, S., Barange, M., Pakhomov, E., Raubenheimer, D.,
Schmidt, K., Simpson, S. and Reiss, C. (2014) Sardine cycles, krill
declines, and locust plagues: revisiting ‘wasp-waist’ food webs. Trends
Ecol. Evol., 29, 309–316.
Baum, J. K. and Worm, B. (2009) Cascading top-down effects of chan-
ging oceanic predator abundances. J. Anim. Ecol., 78, 699–714.
Boyce, D. G., Frank, K. T. and Leggett, W. C. (2015) From mice to
elephants: overturning the ‘one size fits all’ paradigm in marine
plankton food chains. Ecol. Lett., 18, 504–515.
Calliari, D. and Tiselius, P. (2009) Organic carbon fluxes through the
mesozooplankton and their variability at different time-scales in the
Gullmarsfjord, Sweden. Estuar. Coast. Shelf Sci., 85, 107–117.
Casini, M., Lövgren, J., Hjelm, J., Cardinale, M., Molinero, J. C. and
Kornilovs, G. (2008) Multi-level trophic cascades in a heavily
exploited open marine ecosystem. Proc. Biol. Sci., 275, 1793–1801.
Costello, J., Sullivan, B. and Gifford, D. (2006) A physical–biological
interaction underlying variable phenological responses to climate
change by coastal zooplankton. J. Plankton Res., 28, 1099–1105.
Deason, E. E. and Smayda, T. J. (1982) Ctenophore-zooplankton-
phytoplankton interactions in Narragansett Bay, Rhode Island,
USA, during 1972-1977. J. Plankton Res., 4, 203–217.
Durbin, A. and Durbin, E. (1981) Standing stock and estimated pro-
duction rates of phytoplankton and zooplankton in Narragansett
Bay, Rhode Island. Estuaries, 4, 24–41.
Granéli, E. and Turner, J. (2002) Top-down regulation in ctenophore-
copepod-ciliate-diatom-phytoflagellate communities in coastal
waters: a mesocosm study. Mar. Ecol. Prog. Ser., 239, 57–68.
Granhag, L., Møller, L. F. and Hansson, L. J. (2011) Size-specific
clearance rates of the ctenophore Mnemiopsis leidyi based on in situ
gut content analyses. J. Plankton Res., 33, 1043–1052.
Hansson, H. (2006) Ctenophores of the Baltic and adjacent Seas–the
invader Mnemiopsis is here! Aquat. Invasions, 1, 295–298.
Haraldsson, M., Jaspers, C., Tiselius, P., Aksnes, D. L., Andersen, T. and
Titelman, J. (2013) Environmental constraints of the invasive
Mnemiopsis leidyi in Scandinavian waters. Limnol. Oceanogr., 58, 37–48.
Jaspers, C., Møller, L. F. and Kiørboe, T. (2011) Salinity gradient of
the Baltic Sea limits the reproduction and population expansion of
the newly invaded comb jelly Mnemiopsis leidyi. PLoS. ONE., 6,
e24065.
Javidpour, J., Molinero, J. C., Peschutter, J. and Sommer, U. (2009)
Seasonal changes and population dynamics of the ctenophore
Mnemiopsis leidyi after its first year of invasion in the Kiel Fjord,
Western Baltic Sea. Biol. Invasions, 11, 873–882.
Kiørboe, T. and Nielsen, T. G. (1994) Regulation of zooplankton bio-
mass and production in a temperate coastal ecosystem. 1.
Copepods. Limnol. Oceanogr., 39, 493–507.
Kremer, P. (1979) Predation by the ctenophore Mnemiopsis leidyi in
Narragansett Bay, Rhode Island. Estuaries, 2, 97–105.
Lindahl, O. and Perissinotto, R. (1987) Short-term variations in the
zooplankton community related to water exchange processes in the
Gullmar fjord, Sweden. J. Plankton Res., 9, 1113–1132.
Lindahl, O. and Hernroth, L. (1988) Large-scale and long-term varia-
tions in the zooplankton community of the Gullmar fjord, Sweden,
in relation to advective processes. Mar. Ecol. Prog. Ser., 43, 161–171.
Mauchline, J. (1998) The biology of copepods. Adv. Mar. Biol., 33,
1–710.
McNamara, M., Lonsdale, D. and Cerrato, R. (2013) Top-down con-
trol of mesozooplankton by adult Mnemiopsis leidyi influences micro-
plankton abundance and composition enhancing prey conditions for
larval ctenophores. Estuar. Coast. Shelf Sci., 133, 2–10.
Micheli, F. (1999) Eutrophication, fisheries, and consumer-resource
dynamics in marine pelagic ecosystems. Science, 285, 1396–1398.
Møller, L. and Riisgård, H. (2007) Population dynamics, growth and
predation impact of the common jellyfish Aurelia aurita and two
hydromedusae, Sarsia tubulosa, and Aequorea vitrina in Limfjorden
(Denmark). Mar. Ecol. Prog. Ser., 346, 153–165.
Pershing, A. J., Mills, K. E., Record, N. R., Stamieszkin, K., Wurtzell,
K. V., Byron, C. J., Fitzpatrick, D., Golet, W. J. et al. (2015)
Evaluating trophic cascades as drivers of regime shifts in different
ocean ecosystems. Philos. Trans. R. Soc., 370, 1–9.
Polis, G., Sears, A., Huxel, G., Strong, D. and Maron, J. (2000) When
is a trophic cascade a trophic cascade? Trends Ecol. Evol., 15,
473–475.
Purcell, J. E. and Decker, M. B. (2005) Effects of climate on relative
predation by scyphomedusae and ctenophores on copepods in
Chesapeake Bay during 1987–2000. Limnol. Oceanogr., 50,
376–387.
Purcell, J. E., Shiganova, T. A., Decker, M. B. and Houde, E. D.
(2001) The ctenophore Mnemiopsis in native and exotic habitats:
US estuaries versus the Black Sea basin. Hydrobiologia., 451,
145–176.
Riisgård, H., Madsen, C., Barth-Jensen, C. and Purcell, J. E. (2012)
Population dynamics and zooplankton-predation impact of the indi-
genous scyphozoan Aurelia aurita and the invasive ctenophore
Mnemiopsis leidyi in Limfjorden (Denmark). Aquat. Invasions, 7,
147–162.
Shurin, J. B., Borer, E. T., Seabloom, E. W., Anderson, K.,
Blanchette, C. A., Broitman, B., Cooper, S. D. and Halpern, B. S.
(2002) A cross-ecosystem comparison of the strength of trophic cas-
cades. Ecol. Lett., 5, 785–791.
Sommer, U. (2008) Trophic cascades in marine and freshwater plank-
ton. Int. Rev. Hydrobiol., 93, 506–516.
Sommer, U. and Sommer, F. (2006) Cladocerans versus copepods: the
cause of contrasting top-down controls on freshwater and marine
phytoplankton. Oecologia., 147, 183–194.
Stibor, H., Vadstein, O., Diehl, S., Gelzleichter, A., Hansen, T.,
Hantzsche, F., Katechakis, A., Lippert, B. et al. (2004) Copepods act
as a switch between alternative trophic cascades in marine pelagic
food webs. Ecol. Lett., 7, 321–328.
Sullivan, B., Costello, J. and Van Keuren, D. (2007) Seasonality of the
copepods Acartia hudsonica and Acartia tonsa in Narragansett Bay, RI,
USA during a period of climate change. Estuar. Coast. Shelf Sci., 73,
259–267.
Tiselius, P., Belgrano, A., Andersson, L. and Lindahl, O. (2016)
Primary productivity in a coastal ecosystem: a trophic perspective
on a long-term time series. J. Plankton Res., 38, 1092–1102.
Tönnesson, K. and Tiselius, P. (2005) Feeding of the chaetognaths
Sagitta setosa and S. elegans in relation to prey abundance and vertical
distribution. Mar. Ecol. Prog. Ser., 289, 177–190.
Worm, B. and Myers, R. A. (2003) Meta-analysis of cod-shrimp inter-
actions reveals top-down control in oceanic food webs. Ecology., 84,
162–173.