Marine Biology

, Volume 151, Issue 4, pp 1471–1478

Benthic bioluminescence in the bathyal North East Atlantic: luminescent responses of Vargula norvegica (Ostracoda: Myodocopida) to predation by the deep-water eel (Synaphobranchus kaupii)


    • OceanlabUniversity of Aberdeen
  • N. J. King
    • OceanlabUniversity of Aberdeen
  • B. D. Wigham
    • OceanlabUniversity of Aberdeen
  • A. J. Jamieson
    • OceanlabUniversity of Aberdeen
  • P. M. Bagley
    • OceanlabUniversity of Aberdeen
  • L. Allan
    • OceanlabUniversity of Aberdeen
  • O. Pfannkuche
    • Leibniz-Institut für Meereswissenschaften (IFM-GEOMAR)
  • I. G. Priede
    • OceanlabUniversity of Aberdeen
Research Article

DOI: 10.1007/s00227-006-0587-7

Cite this article as:
Heger, A., King, N.J., Wigham, B.D. et al. Mar Biol (2007) 151: 1471. doi:10.1007/s00227-006-0587-7


Using two autonomous lander systems, one equipped with a high sensitivity intensified silicon intensifier target (ISIT) video camera and a second with a 4.1-megapixel digital stills camera, observations were made of bioluminescent emissions and fauna attracted to artificial food falls at ca. 1,000 m depth in the carbonate mound provinces of the Porcupine Seabight and Rockall Bank, North East Atlantic. On the Galway coral mound (Belgica Mound Province), seven bioluminescent events per hour were observed, whereas on an area of sediment at the base of the Kiel Mount on the Rockall Bank 133 events per hour were observed. This increase in bioluminescent activity was associated with the presence of the deep water eels Synaphobranchus kaupii, and the ostracods, Vargula norvegica. Captured ostracods luminesced readily in response to mechanical stimulation and were also observed emitting a luminous secretion. We hypothesise that V. norvegica, attracted to bait, luminesce as a defence response against the predatory activity of S. kaupii that compete for bait but also feed on the ostracods. Benthic bioluminescence in the carbonate mound provinces was not directly linked to the presence of corals.


Bioluminescence is a well-known phenomenon in the surface layers of the oceans and many studies using bathyphotometers and imaging devices have described the distribution of pelagic luminescent organisms (Widder et al. 1999; Cussatlegras et al. 2001). In contrast, despite numerous descriptions of luminescent potential of benthic marine organisms (Herring 1990), few in situ studies have been carried out on the occurrence of benthic bioluminescence in the Deep Sea. During bathyphotometer profiling studies, Clarke and Hubbard (1959) stated that the frequency of bioluminescent events does not increase near the Deep Sea floor. Herring et al. (2001) first reported aggregations of luminescent organisms in the Deep Sea at artificial food falls using a downward facing ultra-low-light intensified silicon intensifier target (ISIT) video camera at 3,200 m off Cape Verde in the North Atlantic. In a detailed analysis of these data, Priede et al. (2006) observed 66 events over 1 h consisting of repeated flashes from small mobile fauna as well as the release of luminescent material into the bottom current. Subsequently, Gillibrand et al. (2006) undertook a survey in the temperate North East Atlantic Ocean deploying baits from 4,800 m depth on the Porcupine Abyssal Plain to 1,000 m depth in the Porcupine Seabight. They found a strong relationship with depth with less than one event per hour in the abyss and a mean of 58 per hour at 1,000 m. However, they also found that at this shallower depth in the region of the Belgica carbonate mounds (De Mol et al. 2002) the mean was 191 events per hour with a maximum of 482 events per hour. These events often persisted for tens of seconds so that the whole bait was illuminated, causing features in the surrounding area to be visible. The high levels of bioluminescence were correlated with the presence of the Kaup’s arrowtooth eel, Synaphobranchus kaupii, but this species does not have luminescent capability. It was hypothesised that the luminescence was generated by organisms too small to be resolved by the conventional PAL video format (625 lines, ca. 0.4 megapixels, equivalent to pixel size ca. 1 mm−2). In this study we revisited the Belgica carbonate mound province and another mound area in the Rockall Bank to investigate the source of this benthic bioluminescence. We hypothesised that cold water coral areas, which are known to be biodiversity hotspots, may also yield higher levels of benthic bioluminescence. The ISIT camera system (Priede et al. 2006) was mounted on a GEOMAR lander (Pfannkuche and Linke 2003), which could be positioned precisely using a video-guided launcher to make observations directly on coral mounds. A second lander, RObust BIOdiversity (ROBIO) (Jamieson and Bagley 2005), equipped with a digital stills camera (4.1 megapixels) was also deployed in parallel, to identify candidate luminescent scavenging organisms at small food falls.

Materials and methods

Data were collected during the first leg of the FS Meteor cruise 61, 19 April–5 May 2004, where four lander deployments were conducted (Table 1). Deployments 1 and 2 were in the Belgica mound province, Porcupine Seabight. The ISIT-GEOMAR lander was positioned on the North flank of the Galway Mound, amongst a coral community, and the ROBIO lander was deployed at the periphery of the Pollux Mound, a coral-covered mound 4–5 km from the Galway Mound. Deployments 3 and 4 were located 2 km apart, both in the south western Rockall Bank on an area of soft sediment at the base of the carbonate Kiel Mount (Fig. 1).
Table 1

Deployment logs for the ISIT-GEOMAR and ROBIO landers in April 2004






Depth (m)


Belgica mound province, PSB











Rockall Bank









Fig. 1

Chart of the deployment sites in the NE Atlantic. Symbols indicate the two study locations


The ISIT camera (OE1325: Kongsberg Simrad, UK, faceplate sensitivity 5 × 10−6 Lux) was mounted on the GEOMAR lander and positioned to view a bait on the seafloor (Fig. 2a). The bait, two mackerel (Scomber scombrus) with a total weight ca. 0.5 kg, was attached to a stainless steel arm, resting ca. 50 mm above the seafloor. The lander was accurately positioned on the seabed using the Oktopus video launcher system (Pfannkuche and Linke 2003). Depth, current velocity and direction, temperature and conductivity were logged at 60 s intervals by an ultrasonic current meter (UCM-60; Sensortec, Norway). For each deployment a total of 1 h of ISIT camera output was recorded on a mini digital video (DV) tape coupled to a custom-built control and logging system (Priede et al. 2006). The system was programmed to record 13 sequences, each of 4 min duration, at 45-min intervals, enabling time-lapse recording over a 9-h period on the seafloor. During the last 15 s of each sequence, a 20 W tungsten light (Deep Multi-lite; DSP&L, USA) illuminated the scene to aid identification of any active fauna.
Fig. 2

Geometry of cameras viewing bait on the seafloor. a ISIT-GEOMAR lander: ISIT camera at 50° angle (field of view is ca. 500 mm). ROBIO lander: digital camera mounted horizontally

ROBIO lander

The ROBIO lander was equipped with a 4.1-megapixel digital stills camera (OE14-028, Kongsberg Maritime Ltd, UK), flash unit (OE11-242, Kongsberg Maritime Ltd, UK) and acoustic Doppler current meter (AquaDopp; Nortec, Norway). The camera was mounted horizontally with its axis 235 mm above the sea floor viewing an arm baited with five mackerel in deployment 2 and nine in deployment 4. The bait was positioned 950 mm in front of the camera lens (Fig. 2b). Images were taken at 60 s intervals over a 6 h period, with an average of 365 images per deployment. In addition, scavenging organisms attracted to the bait were captured using two baited (mackerel head and tail) 1-l funnel traps (25 mm wide opening) attached on the underside of the ROBIO lander, ca. 200 mm above the seafloor.

Laboratory studies on captured animals

Post recovery, the contents of each trap from the ROBIO lander were transferred to plastic aquaria filled with filtered seawater in a dark room held at 10°C. Observations were made by dark-adapted, naked eye and recorded using a silicon intensifying target camera (SIT, OE1324: Kongsberg Simrad, UK; faceplate sensitivity 2 × 10−4 Lux), connected to a control unit, monitor and a Sony DV recorder. Following quiescent observations, pieces of mackerel were added to the tanks followed by a period of mechanical disturbance to assess the bioluminescent potential of the catch. Organisms were then sorted and separate taxa were placed in 100 ml transparent plastic jars filled with seawater. These jars were then placed ca. 150 mm in front of the SIT camera. Experimental manipulation of luminescent species allowed observations from (1) quiescent conditions; (2) following addition of food, and (3) mechanical stimulation. Density dependent effects on the intensity of produced light were also investigated. Luminescent species were subsequently photographed and preserved in 70% ethanol for further identification.

Data analysis

All videos from the ISIT-GEOMAR deployments were viewed frame by frame using proprietary software (Final Cut Pro, v2.0, Apple Computers Inc., USA). Initial analysis provided data on the number of bioluminescent events per minute following touchdown of the lander vehicle. The number of events observed in each 4-min segment was recorded together with notes on the shape and duration of each event. A bioluminescent event was defined as a single pulse (or sequence) of light that appeared as a discrete event, therefore assumed to be from one potential organism. Sequential events in the same position were assumed to be from the same organism and scored as a single event. The various scavengers observed in the field of view during the illuminated periods were counted and identification was confirmed by comparison with digital images and voucher specimens from the ROBIO lander. In the ISIT video sequences the maximum number of Synaphobranchus kaupii was recorded for each illuminated 15-s period, values of 20 and above may be subject to error owing to difficulty of discriminating shadows and moving S.kaupii in relatively low resolution images. Each image obtained from the ROBIO lander was assessed for the presence of S.kaupii and the number of eels visible in each image was recorded. All statistical analysis was undertaken using SPSS v14.0 (SPSS Inc., USA).


In situ observations

Current meter data from the ISIT-GEOMAR lander indicated a tidal periodicity in water velocity and direction in both deployments. The current speed at the Galway Mound site ranged between 4 and 125 mm s−1 (mean 42 ± 26) and at the base of the Kiel Mount current speeds were 34–92 mm s−1 (mean 58 ± 12). Mean water temperatures at the two sites were 9.52°C (SD = 0.21) and 7.07°C (SD = 0.07), respectively.

Belgica Mound Province

At the Galway Mound (deployment 1) the first evidence of bioluminescence was seen 250 min after lander touchdown. A total of seven events were observed over 1 h of recording distributed in segments during 9 h on the sea floor (Fig. 3a). Most events consisted of static spots of light (apparent diameter <1 cm) or a combination of streaks and spots lasting <5 s with the exception of one 27-s duration static event. During the 30-s illuminated periods, scavenging animals were observed at the bait. One small eel, Synaphobranchus kaupii, and an unidentified fish were observed at the start of the deployment. A crab, probably Bathynectes sp. and other unidentifiable decapods were partially in view during the remaining illuminated sequences. Small, mobile organisms <5 mm length were visible intermittently throughout the deployment. However, the resolution of the ISIT camera prevented their accurate identification. By the time of the second illuminated sequence, 70 min after touchdown, the bait had been completely consumed. The seafloor in the vicinity was covered with coral colonies of Madrepora oculata and Lophelia pertusa.
Fig. 3

ISIT-GEOMAR lander: time course of observations after touchdown. Paired observations of the number of bioluminescent events per minute (closed histogram bar) during dark periods and the number of eels (open histogram bar) represents the maximum number of Synaphobranchus kaupii visible during 15 s light periods consecutive to the dark observations. a Galway Mound. b Kiel Mount

Digital images from the ROBIO deployment on the Pollux Mound (dep. 2) revealed the main scavengers to be fishes, including the eel S. kaupii and the morid Mora moro, in addition to lysianassid amphipods. The mean number of S. kaupii per image was three (SD = 2.74) with a maximum of 13 at 69 min after touchdown. Other species photographed investigating the bait included the teleost fishes Lepidion eques and Phycis sp. and two elasmobranch species Galeus melastomus and Galeus murinus. Benthic invertebrate crustaceans observed were Bathynectes sp. and Munida tenuimana. Sessile epibenthic species observed were the living polyps of Madrepora oculata and a hexactinellid sponge Aphrocallistes bocagei.

Rockall Bank

A total of 133 bioluminescent events were observed at the base of the Kiel Mount (dep. 3) with the first event occurring within 60 s of touchdown, followed by a steady increase in bioluminescent activity peaking at 88–123 min after touchdown, with 11.48 events per minute (Fig. 3b). Subsequently, the number of luminous events decreased over time and the overall average was 2.89 events per minute. The events, classified into functional groups, were (1) streaks, averaging 2.9 s in duration (SD = 1.79) and moving quickly across the field of view; (2) static spots with various degrees of intensity and longevity with an average duration of 10.3 s (SD = 9.75), and (3) a combination of streaks and static spots averaging 4.2 s in duration (SD = 1.8). A total of six large events were observed during the peak period of luminescence, with duration averaging 52.8 s (SD = 9.75). These large events released clouds of bright luminous material that glowed or pulsed before fading (Fig. 4).
Fig. 4

ISIT-GEOMAR video images of bait at the base of the Kiel Mount (deployment 3). Arrows indicate current direction, mean current velocity 58 mm s−1 (SD = 12). a Bioluminescence 105 min after touchdown. Note the aggregation of bioluminescence down-current of the bait. The bait can be seen illuminated by the bioluminescence on the left hand part of the images. b The same field of view during the subsequent illuminated sequence 132 min after touchdown

Synaphobranchus kaupii were observed in every illuminated sequence reaching a peak abundance of 50 individuals, 132 min after touchdown (Fig. 3b). The maximum frequency of bioluminescent events was generally associated with the maximum number of S. kaupii active in the field of view but at the actual time of peak luminescent activity, the number of eels decreased. The eels exhibited a slow swimming motion, as described by Uiblein et al. (2002), just sufficient to stem the bottom current. However, several individuals occasionally undulated the head in a very rapid, large amplitude sideways motion, cycle period 0.4 s, of the same order of rapidity as sprint fast starts in this species (Bailey et al. 2005). These twitches clearly had no locomotor or predatory function and appeared to be the result of an unseen external irritation. Additional invertebrate fauna observed included amphipods that were seen in all sequences, ophiuroids (after 40 min) and a gastropod mollusc, Colus sp. (first seen at 260 min). Other invertebrates may have been present but were not resolved in the video.

Analysis of the ROBIO photographs from the Kiel Mount (dep. 4) showed that the dominant scavengers attracted to bait were S. kaupii, lysianassid amphipods, several species of elasmobranchs, including Centrophorus squamosus, Deania calceus and Galeus melastomus, and the teleost fishes Mora moro and Phycis blennoides. Invertebrate fauna observed in the area surrounding the bait included a gastropod mollusc (Colus sp.), pagurid crabs, a spider crab and several ophiuroids. The number of S. kaupii was highest in the first 3 h after touchdown, with a maximum number of 104 eels observed at 18 min post-touchdown. The number of S. kaupii recorded was significantly higher than at the Pollux Mound (ANOVA: F = 2.945, df = 70, P = 0.000) with a mean of 27.8 (SD = 18.37) individuals per image. Ostracods were first observed at 24 min into the deployment and were recognisable from their shape and size with identity confirmed by voucher specimens caught in traps. Numerous ostracods swarmed around the bait and S. kaupii appeared to be actively feeding upon them (Fig. 5). We believe that ostracods also attempted to graze on the flanks of the S. kaupii, some of which responded with an apparent self-grooming behaviour, turning their heads towards the body and biting themselves (Fig. 6). The bait was completely consumed after 50 min on the seafloor.
Fig. 5

ROBIO observations on the Kiel Mount (deployment 4). a 206 min after touchdown: Eels, Synaphobranchus kaupii, ostracods probably Vargula norvegica (O), and lysianassid amphipods (A). b 336 min after touchdown, ostracods probably Vargula norvegica (O) moving up into the water column and amphipods (A) grazing on the bait arm. The end of the tubular probe supporting the bait is visible at the bottom of the field of view
Fig. 6

Putative grooming behaviour in Synaphobranchus kaupii observed by the ROBIO camera on the Kiel Mount (deployment 4); a 74 min after touchdown, b 256 min after touchdown

Traps and shipboard observations

Baited traps from the ROBIO deployment in the Belgica Province (dep. 2) recovered ca. 5,000–6,000 lysianassid amphipods (based on counting sub-samples in a grid superimposed on a digital photograph of the entire sample spread out on a flat surface). They ranged from 4 to 14 mm in length and were mostly Tmetonyx cicada. Many of the amphipods were still motile but no bioluminescence was observed during any of the laboratory experiments. The contents of the traps from the Kiel Mount (dep. 4) contained 1,899 lysianassid amphipods consisting mostly of Stephonyx biscayensis, as well as 1,365 ostracods identified as Vargula norvegica. There was no evidence of luminescent activity when these organisms were left undisturbed or when food (mackerel flesh) was added to the aquaria. However, bioluminescence was recorded when mechanical stimulation (manual agitation) was applied. Upon separation of the major taxa, and subsequent laboratory trials, the amphipods were found not to be bioluminescent. No bioluminescence was observed when the ostracods were left quiescent or following the addition of food. There was a positive response to mechanical stimulation (shaking), the ostracods were observed ejecting a viscous substance that emitted a pale blue light visible to the naked eye. When mechanically agitated in transparent bottles in front of the SIT camera, the amount of light produced varied according to the number of ostracods in the bottle, with ten individuals sufficient to illuminate 100 ml of seawater (Fig. 7).
Fig. 7

Ostracods, Vargula norvegica trapped and placed in tubes on board ship. SIT video images of luminescence generated by different numbers of ostracods following mechanical stimulation (shaking). a n = 1, b n = 3, c n = 6, d n = 10, e n = 20


The discovery of a new bioluminescence hotspot at the Kiel Mount (Rockall Bank), a location 720 km from the original bioluminescence hotspot in the Porcupine Seabight identified by Gillibrand et al. (2006) indicates that benthic bioluminescence is not an isolated phenomenon in the temperate NE Atlantic Ocean. It is highly probable that the ostracod V. norvegica was the source of much of the luminescent activity at both sites. None of the fish species observed are known to be capable of bioluminescence. Some amphipods can produce faint light but not of the brightness produced by ostracods and showed no evidence of luminescence in our shipboard experiments. Ostracods were observed in high numbers at the bait in the ROBIO lander images. The shipboard experiments (Fig. 7) showed that in response to mechanical stimulation such numbers of Vargula norvegica were capable of releasing luminescent material of sufficient volume to account for illumination of the entire bait as observed at the Kiel Mount and by Gillibrand et al. (2006) in the Porcupine Seabight. The small size of V. norvegica (mean length 3.3 mm, SD = 0.41 mm, n = 100) places it just at the limits of resolution of the ISIT camera, but they were recognisable in the digital camera images, with species level identification aided by trap-captured specimens. There was a clear association between the amount of bioluminescence observed and the presence of Synaphobranchus kaupii at the Kiel Mount site. Gordon and Mauchline (1996) and Marques (1998) described the diet of S. kaupii in the Rockall Trough and in the Porcupine Seabight, respectively. Stomach content analyses revealed fragments of both crustaceans and ostracods. We propose that it is the mechanical stimulation by the eel activity, which may prey on V. norvegica that stimulates the release of bioluminescent material as an anti-predation defence mechanism. Morin (1986) observed tropical shallow-water Vargula sp. in situ and noticed that, when attacked by predators, they produced an extensive cloud of luminescence that glowed for about 1 min. During some of these attacks, the predator regurgitated the ostracod and it was suggested that the luminous secretion startled and probably temporarily blinded the predator.

It is interesting to note that on Kiel Mount (dep. 3 and 4) when the bioluminescence reached its peak the number of S. kaupii decreased from the maximum observed. This may indicate that the mass release of bioluminescence was effective as a defence mechanism. ISIT footage and ROBIO images also revealed unusual behaviours in S. kaupii, with the eels exhibiting fast twitching and self-grooming action. Vannier et al. (1998) observed feeding in myodocopid ostracods and reported that they will attack and feed on live animals. In their study, small swarms (tens of ostracods) were seen to consume small live annelids within a few minutes. We believe it is highly probable that V. norvegica opportunistically attempt to feed on S. kaupii attracted to a food fall and that the latter then exhibited twitching and putative grooming behaviours.

Morin (1986) has described two different kinds of luminescent displays by Vargula shulmanae, V. contragula and V. graminicola in shallow waters of the Caribbean; firstly, the anti-predator release of a bomb-like cloud of luminescence, and secondly, complex trains of precisely spaced pulses emitted mainly by displaying males at lekking sites. These latter displays are species-specific and Morin (1986) drew parallels with mating aggregations of various species of insects such as fireflies. It is not known if V. norvegica conduct such displays but it is possible that some of the lesser flashes we observed may be attributed to fragments of spontaneous displays, which could not be reproduced in the shipboard experiments.

Gillibrand et al. (2006) observed that the amount of bioluminescence within the Belgica Mound Province was unpredictable, with three deployments producing 51, 54 and 65 events per hour and another three producing 239, 256 and 482 events per hour. This dichotomy in the results could be explained by the need for both eels and ostracods to be present in numbers above a critical abundance to produce the mass release of luminescent material. A further possibility is that Vargula occur only in large numbers at leks, which would account for specific localities with very high luminescent activity.

The depth of observations of mass bioluminescence at 950 m at the Kiel Mount and down to 1,019 m by Gillibrand et al. (2006) is close to the maximum previously reported depth limit of 957 m for V. norvegica (Poulsen 1962). It is very unlikely that this species can account for the luminescent activity recorded by Priede et al. (2006) at 3,200 m in the tropical NE Atlantic and furthermore we cannot assume that all the luminescence observed in the Porcupine Seabight is generated by V. norvegica.

The lack of any significant bioluminescence on the coral-topped Galway Mound (dep. 1), indicates that the presence of benthic bioluminescence is not a characteristic of cold-water corals or their directly associated fauna. Uiblein et al. (2002) showed that S. kaupii preferred a habitat with temperatures of 7.18–8.61°C and variable current velocities, 0.1–0.5 m s−1. In the Porcupine Seabight and Rockall Bank, corals occur in areas with strong bottom currents (De Mol et al. 2002). Occurrence of V. norvegica and S. kaupii in soft sediment areas in coral mound provinces may be linked to the fact that the trophic and current regimes that enable coral polyps to feed successfully also provide a suitable environment for these two species that feed using odour plumes carried on bottom currents.


We thank the officers and crew of the RVMeteor as well as M. Thurston and M. Angel, National Oceanography Centre in Southampton for the identification of the amphipods and ostracods, respectively. Special thanks also to Dr E. Jones of the Fisheries Research Services, Marine Laboratory in Aberdeen for the loan of camera equipment. This research was conducted as part of the EU HERMES project. A. Heger was partially funded by the Luxembourgish Ministry of Culture, Higher Education and Research. N. King was supported by NERC studentship NER/S/A/2003/11190.

Copyright information

© Springer-Verlag 2007