University of Plymouth
PEARL https://pearl.plymouth.ac.uk
Faculty of Science and Engineering School of Biological and Marine Sciences
2011-08-01
In situ observations of fish associated
with coral reefs off Ireland
Soffker, M
http://hdl.handle.net/10026.1/1329
10.1016/j.dsr.2011.06.002
Deep-Sea Research Part I: Oceanographic Research Papers
All content in PEARL is protected by copyright law. Author manuscripts are made available in accordance with
publisher policies. Please cite only the published version using the details provided on the item record or
document. In the absence of an open licence (e.g. Creative Commons), permissions for further reuse of content
should be sought from the publisher or author.
In situ observations of fish associa ted with coral reefs off Ireland
M. S
¨
offker
a,
n
, K.A. Sloman
b
, J.M. Hall-Spencer
c
a
College of Life and Environmental Sciences, School of Biosciences, University of Exeter, Hatherly Laboratories, Prince of Wales Road, Exeter EX4 4PS, Devon, UK
b
School of Science, University of the West of Scotland, Paisley, Scotland, UK
c
School of Marine Science and Engineering, University of Plymouth, Plymouth, Devon, UK
a r t i c l e i n f o
Article history:
Received 18 April 2011
Received in revised form
1 June 2011
Accepted 13 June 2011
Available online 8 July 2011
Keywords:
Cold-water corals
Continental Slope
Fish abundance
Fish community composition
Lepidion eques
Lophelia pertusa
Northeast Atlantic
Porcupine Bank
Remotely Operated Vehicle (ROV)
observations
a b s t r a c t
The abundance and behaviour of fish on and around coral reefs at Twin Mounds and Giant Mounds,
carbonate mounds located on the continental shelf off Ireland (600–1100 m), were studied using two
Remotely Operated Vehicle (ROV) dives. We recorded 30 fish taxa on the dives, together with three
species of Scleractinia (Lophelia pertusa, Madrepora oculata and Desmophyllum cristagalli) and a diverse
range of other corals (Antipatharia, Alcyonacea, and Stylasteridae). Stands of live coral provided the
only habitat in which Guttigadus latifrons was observed whereas Neocyttus helgae was found
predominantly on structural habitats provided by dead coral. Significantly more fish were found on
structurally complex coral rubble habitats than on flatter areas where coral rubble was clo gged with
sand. The most common species recorded was Lepidion eques (2136 individuals), which always occurred
a few cm above bottom and was significantly more active on the reefs than on sedimentary habitats.
Synaphobranchus kaupii (1157 indiv.), N. helgae (198 indiv.) and Micromesistius poutassou (116 indiv.)
were also common; S. kaupii did not exhibit hab itat-related differences in behaviour, whilst N. helgae
was more active over the reefs and oth er structured habitats whereas M. poutas sou was more active
with decreasing habitat complexity. Trawl damage and abandoned fishing gear was observed at both
sites. We conclude that Irish coral reefs provide complex habitats that are home to a diverse
assemblage of fish utilising the range of niches occurring both above and within the reef structure.
& 2011 Elsevier Ltd. All rights reserved.
1. Introduction
Some cold-water corals live in such dense aggregations that they
provide structurally complex habitats which support a diverse
assemblage of associated invertebrate life (Le Danois, 1948; Burdon-
Jones and Tambs-Lyche, 1960; Buhl-Mortensen and Mortensen, 2004;
Metaxas and Davis, 2005; Mortensen et al., 2008) and fish (Auster,
2005; Costello et al., 2005; Reed et al., 2006; Ross and Quattrini, 2007;
Roberts et al., 2008; Harter et al., 2009). Cold-water coral reefs have
been classified as ‘Essential Fish Habitat’ (EFH) in the western Pacific
(Witherell and Coon, 2000). While it is debatable whether cold-water
corals are ‘essential’ to fish, as the fish recorded to date may survive
well in other habitats, it is clear that corals provide an important
source of three dimensional structure in the predominantly sedimen-
tary habitats of deeper waters (Auster, 2005, Roberts et al., 2008;
Harter et al., 2009). A comparison of visual surveys at eight locations
from Ireland to Norway led Costello et al. (2005) to conclude that
reefs formed by Lophelia pertusa attract elevated concentrations of
fish in the NE Atlantic. These authors tentatively related the elevated
abundance of fish to the occurrence of reefs in waters rich in
zooplankton (benefiting zooplanktivorous fish) at sites with increased
availability of benthic invertebrates (benefiting invertivores). This
combination of elevated zooplankton and benthic invertebrates
provides appropriate habitat for at least some of the fish recorded
(rockfish, Sebastes spp.). Faunal abundance and diversity on coral sites
is consistently higher compared to near-by non-coral sites
(Mortensen et al., 2008; Roberts et al., 2008; D’Onghia et al., 2010;
Du Preez and Tunnicliffe, 2011), suggesting corals may function as
refuge; for some species, the provided structure is a significant factor
(Du Preez and Tunnicliffe, 2011). Husebø et al. (2002) conclude that in
Norway the L. pertusa reefs were used by the tusk Brosme brosme for
feeding and for physical shelter by the ocean perch Sebastes marinus,
and Mediterranean L. pertusa and Madrepora sp. reefs were nurseries
e.g. for Helicolenus dactylopterus (rockfish), Merluccius merluccius
(European hake), Micromesistius poutassou (blue whiting), or Phycis
blennoides, of which some are commercially exploited species
(D’Onghia et al., 2010).
Throughout the world, cold-water coral habitats are highly
vulnerable to the impacts caused by current fishing practices.
These impacts are long-lasting and visible even after many years
of closure to fisheries (Althaus et al., 2009). Bottom-trawling in
particular can quickly damage coral structures that take thou-
sands of years to develop, so an understanding of the ecological
role of cold-water coral reefs is important for the sustainable
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/dsri
Deep-Sea Research I
0967-0637/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.dsr.2011.06.002
n
Corresponding author. Tel.: þ 1 392 723786.
E-mail address: mks204@exeter.ac.uk (M. S
¨
offker).
Deep-Sea Research I 58 (2011) 818–825
management of fisheries (Hall-Spencer et al., 2002; Koslow,
2007). In the NE Atlantic, complaints from long-line fishermen
led to a ban on trawling within selected coral-rich provinces off
Norway (Foss
˚
a and Alvsv
˚
ag, 2003). In 2003, an area called the
‘Darwin Mounds’ off NW Scotland became the first cold-water
coral protected area in the European Union (Commission Regula-
tions EC 1475/2003 and 263/2004, Davies et al., 2007) and was
followed in 2007 by further bans on trawling by the EU and the
North East Atlantic Fisheries Commission on coral-rich areas
around Hatton and Rockall Banks west of Scotland (NEAFC
Recommendation IX-2007). Currently there is no protection of
coral-rich provinces within the Irish Exclusive Economic Zone but
the present study contributes to a rapidly increasing body of
evidence concerning the importance of coral rich areas off Ireland
(De Mol et al., 2002; Hall-Spencer et al., 2002; Freiwald and
Roberts, 2005; Roberts et al., 2008; Dorschel et al., 2009).
Extensive trawling surveys have provided a detailed overview
of the benthic fish assemblages of the NE Atlantic (e.g. Haedrich
and Merrett, 1988; Merrett et al., 1991; Gordon and Bergstad,
1992; Merrett, 1992; Gordon et al., 1996; Bergstad et al., 1999)
with studies focussed mainly on commercially important families
such as the gadoids and macrourids (e.g. Mauchline and Gordon,
1984; Kelly et al., 1997; Coggan et al., 1999). Baiting experiments
have provided information about population densities, swimming
speeds and activity levels of some deepwater fish (Armstrong
et al., 1992; Collins et al., 1999), and in recent years behavioural
observations of benthic deepwater fish communities have been
made using submersibles and ROVs (Johnson et al., 2003; Widder
et al., 2005; Lorance and Trenkel, 2006; Ross and Quattrini, 2007).
For example, the behavioural ecology of seven benthic fish found
in the North Eastern Atlantic was described (Lorance et al., 2002;
Uiblein et al., 2002, 2003) along with consideration of how fish
species associate with specific habitat types and how habitat type
influences fish behaviour (Uiblein et al., 2003).
Here, we present an analysis of ROV footage to compare the
benthic fish communities on and around coral reefs on the
Porcupine Bank in the NE Atlantic. Descriptions of the study area
and some preliminary expedition results have been published
within Myers and Hall-Spencer (2004) and Wheeler et al. (2005).
We assess the association of fish species with different substrate
types and note and compare evidence of fishing activity around
the reefs. As little is known about the deep-sea fish we were able
to observe, we describe in detail the behaviour of the three most
abundant species across a range of habitat types.
2. Methods
We studied fish associations on and around coral reefs on Twin
Mounds and Giant Mounds which are carbonate mounds situated
9 km apart on Porcupine Bank in the NE Atlantic (Fig. 1, Table 1).
Geological and physical features of Twin Mounds and Giant
Mounds were recently described by Dorschel et al. (2009). Video
was collected during Expedition ARK-XIX/3a in June 2003 using
ROV ‘Victor 6000’ (operated by IFREMER) aboard the RV ‘Polar-
stern’ (operated by the Alfred Wegener Institute).
Observations were made throughout each dive using only the
main colour camera, pointing at a 301 angle downwards-forward.
A four point parallel laser system (distance between two opposite
points¼ 23 cm) was available to assess object size, although this was
only accurate when the target was perpendicular to the laser beams.
Although most of each dive was spent filming a 2.5 m wide strip of
seabed with the ROV on average 2.65 m (all data collected was
observed o 5 m) above bottom, the data do not represent a true
transect. As there was some variation in altitude and each site was
passed over only once, no true replication of observations was
possible. Consequently, our data are more qualitative than quantita-
tive. However, acknowledging the problems associated with observing
a
b
Fig. 1. (a) Study sites at Twin Mounds and Giant Mounds at Porcupine Bank, Northeast Atlantic. (b) Dive tracks at Twin Mounds (13 km) and Giant Mounds (14 km).
M. S
¨
offker et al. / Deep-Sea Research I 58 (2011) 818–825 819
deep sea fish communities by ROV (Adams et al., 1995; Trenkel et al.,
2004a, 2004b), it was possible to reliably estimate fish abundance and
to make some quantitative measurements of fish behaviour. Due to
the patchiness and high spatial and area heterogeneity of substrates,
the most reliable and robust method was to calculate average fish
abundance over an entire substrate (see below).
Footage was only considered less than 5 m from the sea bed
and at ROV speeds below 1.3 knots. Footage was excluded if the
ROV was stationary for a period of time. From footage suitable for
data collection, a still image was taken every minute and the
substrate type below the ROV was noted. The substrate type was
divided into four categories (coral, sand, bedrock and three-
dimensional (3D) structured, which consisted of patchy or dense
dropstones and rocky outcrops). The coral habitat was further
divided into the following sub-categories; (1) live coral, (2) coral
rubble, (3) mixed living and dead coral and (4) sand-clogged coral,
based on the expedition cruise report (Klages et al., 2004) which is
similar to other published studies (Dorschel et al., 2007; Dorschel
et al. 2009). This allowed the total time of footage over a particular
substrate type to be calculated for each dive, using a total of 2029
still images. The footage was then analyzed again; this time, each
fish encountered was recorded and for each fish observed the fish
species, body position in relation to sea bed, behaviour (Table 2),
substrate type and depth, and any evidence of fishing were noted.
Combining the data on time spent over each substrate and fish
occurrences, we adopted the ‘fish counts per minute’ method of
Costello et al. (2005) to compare fish abundance between habitats,
i.e. the total number of fish observed in each habitat was divided
by the minutes spent filming in each habitat.
Evidence of fishing at the sites was noted in each of the 2029
still images and frequency of trawling damage occurring at each
site and on each substrate type calculated from these.
To compare abundance of fish on each substrate type (as fish
per minute counts), a Fisher’s exact test for count data in
contingency was used to test the difference between the observed
and the exact probability (expected) of abundance (Fisher, 1922).
Within a site, we tested whether the observed abundance of fish
across substrates differed significantly from that expected if they
were equally distributed across all substrates. The behavioural
patterns of five fish species were considered in detail. For each
species, firstly, within one habitat, the frequency a specific beha-
viour was observed for was compared to see if some behaviours
were performed more frequently than would be expected by
chance. Secondly, across different habitats, the performance of
each specific behaviour was compared to understand if that
behaviour was performed equally in each habitat. Fisher’s exact
test was also applied to test whether trawling damage was
independent of substrate type, assuming equal probabilities of
trawling damage on all substrates. All statistical analysis was
carried out using the software package R (Crawley, 2007).
3. Results
The amount of time spent over each substrate is shown in
Table 3. The live coral framework was dominated by L. pertusa and
Madrepora oculata, (although Desmophyllum cristagalli was also
seen), but solitary antipatharian and anthozoan corals, as well as
anthozoan reefs were also observed. The adjacent coral rubble
zones supported a rich assemblage of sessile and vagile inverte-
brate fauna. Large particle aggregates (‘marine snow’) and swarms
of planktonic copepods were observed for the majority of the dives
and were abundant near the seabed. Fish abundance and commu-
nity composition differed between sites (Table 4) although the
two commonest species, the North Atlantic codling Lepidion eques
and the northern cutthroat eel Synaphobranchus kaupii, were
recorded in similar abundances at each site. At Twin Mounds the
false boarfish Neocyttus helgae was the third most common species
observed, but this species was not seen at Giant Mounds. Con-
versely, the M. poutassou was the third most abundant species at
Giant Mounds, but was not observed at Twin Mounds.
Table 1
Details of the study sites Twin Mounds and Giant Mounds on the Porcupine Bank, Northeast Atlantic, during the ARKTIS XIX/3 survey (additional information can be found
in the cruise report under http://epic.awi.de/Publications/BerPolarforsch2004488.pdf and in Dorschel et al., 2009).
Twin Mounds Giant Mounds
Depth (m) 750–1120 650–1060
Date and time of survey 12.06.2003 16:03:44 to 13.06.2003 09:43:04 14.06.2003 08:05:28 to 15.06.2003 03:14:28
Start coordinates 53105
0
25
00
N 14148
00
05
00
W 53110
00
16
00
N 14143
00
37
00
W
End coordinates 53106
0
13
00
N 14155
00
19
00
W 53111
0
33
00
N 14147
0
06
00
W
Dive track length (km) 13 14
Studied footage time (h) 16.5 17.0
Velocity range (knots) 0.0–1.3 0.0–1.3
Table 2
Behavioural categories used to classify observed fish behaviour during each dive.
Behaviour Description
‘stationary’ Fish sitting on the seabed or showing no locomotion movements.
‘station holding’ Fish swimming sufficiently against the current to hold position over the substrate.
‘associated’ Fish observed in or associated with 3D structures.
‘active forward’ Fish swimming actively forward (displacement).
‘disturbed’ Abrupt changes in direction, accelerating swimming speed, sudden increased fin beating or apparent disorientation.
Table 3
Percentage of total time each substrate category was filmed during each dive.
Substrate Twin Mounds (%) Giant Mounds (%)
Coral 48.4 50.3
-live coral 5.9 15.5
-dead coral 36.5 6.9
-mixed dead and living coral 1.7 14.6
-sand-clogged coral 4.3 13.2
Bedrock 1.0 9.5
Sand 35.9 30.0
3D structured 14.6 10.2
M. S
¨
offker et al. / Deep-Sea Research I 58 (2011) 818–825820
At both sites, more fish were found in 3-D structured habitats
than in coral habitats (Fig. 2A), although only at Giant Mounds
was the association of fish with specific habitats significantly
different from random (Fisher’s exact test, p¼0.011). The abun-
dance of fish was, as a trend, associated with specific coral
habitats (although not significantly, Fisher’s exact test, p¼0.09)
at Giant Mounds, with the highest levels of fish found on rubble
habitats, fewer on live coral thickets and fewer still in areas of
decreasing habitat complexity; although a similar pattern was
seen at Twin Mounds (Fig. 2B), the association was not significant
(Fisher’s exact test, p¼0.12).
The behaviour of the most common fish found on and around
the reefs differed between species and some species showed clear
differences in behaviour between different habitats (Fig. 3). Bed-
rock habitats comprised less than 1% of footage at Twin Mounds
and, therefore, were excluded from behavioural analyses at both
sites. Sixty to eighty percent of observed S. kaupii, a scavenging eel,
were performing active behaviour regardless of habitat (Fig. 3A).
Lepidion eques was always recorded within a few centimetres
above bottom and displayed a larger variety of behaviours. There
was a tendency for L. eques to be found associated with structures
that were emergent from the seabed, such as boulders, drop
stones and solitary corals. L. eques were more stationary and less
active on sandy substrata at both sites (Fisher’s exact test,
p¼ o 0.001 at both sites; Fig. 3B) with more individuals associated
with a 3-D structure (Fisher’s exact test, p¼0.006 for Giant
Mounds, p¼0.008 at Twin Mounds). Neocyttus helgae, a laterally
flattened fish, was most active above coral 3-D structured habitats
(Fisher’s exact test, p¼ 0.007) whereas M. poutassou, a fusiform
fish, appeared to be more active above sandy substrata than coral
(Fig. 3C), although this was not statistically significant (Fisher’s
exact test, p¼0.71).
Interestingly, two species of fish were particularly associated
with coral habitats. N. helgae presence was tied to coral habitat
(Fisher’s exact test, po 0.001) and it was found most frequently
above dead coral rubble (Fisher’s exact test, p¼0.06) whereas
Guttigadus latifrons was mainly found amongst dense living coral
thickets (Fisher’s exact test, p¼ 0.006), and was not observed on
non-coral habitat. The behaviour of G. latifrons differed between
dead and living coral habitat (
Fig. 4). In particular, G. latifrons
spent more time active and within living coral (Fisher’s exact test,
p¼ 0.08) whereas when it was sighted in association with dead
coral it tended to be more stationary and less hidden. On one
occasion on Giant Mounds, a fish strongly resembling Gaidopsarus
sp. was found among living pink L. pertusa. However, we cannot
be certain about the identification of this individual.
At Giant Mounds the reefs appeared not to be heavily impacted
by trawling, although one monofilament net was seen, filled with
fresh coral rubble and broken live coral fragments. No trawl tracks
could be seen in the proximity of the net, suggesting that this was
a lost gill-net. At Twin Mounds there was widespread evidence of
the impacts of fishing where recently broken coral, lost gear and
trawl tracks seen during 14% of the recording time (as opposed to
4% at Giant Mounds). The occurrence of fishing evidence at Twin
Mounds was linked to substrate type (Fisher’s exact test,
p¼ 0.0008), positively correlated with patchy live coral (boulders
on sand, overgrown with coral such as drop stones) and a
negatively correlated with sandy coral rubble and dead coral.
4. Discussion
Dense aggregations of scleractinian corals were discovered in
deep waters off SW Ireland during dredging surveys aboard HMS
Porcupine (Duncan, 1873) but the invertebrate diversity of these
habitats did not begin to be fully appreciated until the more
detailed dredge and grab studies of Le Danois (1948). Tradition-
ally, scientific fish trawls were avoided on these grounds, due to
Table 4
The abundance of fish taxa recorded at Twin Mounds (53106
0
N 14155
0
W, 750–1120 m depth) and Giant Mounds (53112
0
N 14146
0
W, 650–1060 m depth) on Porcupine Bank,
Irish Continental Slope, 13–15 June 2003 by the ROV ‘‘Victor 6000’’. Percentage values are the abundance of a species as a percentage of the total number of fish seen.
Twin Mounds Giant Mounds
Fish taxa Individuals % Fish taxa Individuals %
Lepidion eques G
¨
unther, 1887 1286 45.46 Lepidion eques G
¨
unther, 1887 850 42.99
Synaphobranchus kaupii Johnson, 1862 652 23.05 Synaphobranchus kaupii Johnson, ,1862 504 25.49
Neocyttus helgae Holt and Byrne, 1908 198 7.00 Micromesistius poutassou Risso, 1827 116 5.87
Unidentified Macrouridae 120 4.24 Unidentified Macrouridae 42 2.12
Guttigadus latifrons Holt and Byrne, 1908 53 1.87 Helicolenus dactylopterus Delaroche, 1809 35 1.77
Molva dypterygia Pennant, 1784 26 0.92 Mora moro Risso, 1810 22 1.11
Molva sp. 24 0.85 Phycis blennoides Br
¨
unnich, 1768 21 1.06
Mora moro Risso, 1810 22 0.78 Coryphaenoides rupestris Gunnerus, 1765 18 0.91
Chimaera monstrosa Linnaeus, 1758 20 0.71 Notacanthus bonapartei Risso, 1840 16 0.81
Notacanthus bonapartei Risso, 1840 16 0.57 Deania calcea Lowe, 1839 12 0.61
Coryphaenoides rupestris Gunnerus, 1765 15 0.53 Nezumia aequalis G
¨
unther, 1878 12 0.61
Nezumia aequalis G
¨
unther, 1878 12 0.42 Chimaera monstrosa Linnaeus, 1758 10 0.51
Scyliorhinidae 12 0.42 Lophius sp. 7 0.35
Lophius sp. 11 0.39 Arctozenus risso Bonaparte, 1840 7 0.35
Rajidae 9 0.32 Hoplostethus atlanticus Collett, 1889 5 0.25
Unidentified Moridae 9 0.32 Molva dypterygia Pennant, 1784 5 0.25
Hoplostethus atlanticus Collett, 1889 8 0.28 Molva sp. 5 0.25
Molva molva Linnaeus, 1758 7 0.25 Centrophorus sp. 1 0.05
Phycis blennoides Br
¨
unnich 1768 7 0.25 cf. Gaidropsarus sp. 1 0.05
Unidentified Gadiform (possibly Merluccius merluccius Linnaeus, 1758) 6 0.21 Molva molva Linnaeus, 1758 1 0.05
Helicolenus dactylopterus Delaroche 1809 5 0.18 Rajidae 1 0.05
Deania calcea Lowe, 1839 4 0.14 Trachyscorpia cristulata Goode and Bean, 1896 1 0.05
Trachyscorpia cristulata Goode and Bean, 1896 4 0.14
Arctozenus risso Bonaparte, 1840 2 0.07
Centrophorus sp. 1 0.04
Squalidae 1 0.04
unidentified 299 10.81 unidentified 285 14.42
Total 2829 Total 1977
M. S
¨
offker et al. / Deep-Sea Research I 58 (2011) 818–825 821
