Density and distribution of megafauna at the Håkon Mosby Mud Volcano ( the Barents Sea ) based on image analysis

Introduction Conclusions References


Introduction
Mud volcanoes are formations created by geo-excreted fluids.To date, more than 1700 mud volcanoes have been identified, of which some 800 are located offshore (Dimitrov, 2002).It has been estimated that more than 10 000 mud volcanoes exist on continental shelves and abyssal plains (Ivanov et al., 1996;Milkov, 2000;Pimenov et al., 2000;Fleysher et al., 2001;Dimitrov, 2002;Kopf, 2002;Milkov et al., 2003).The processes leading to their formation include mud diapirism, tectonic compression, dehydration of clay minerals at depth or a rapid deposition of mass flows (Kohl and Roberts, 1994;Kopf, 2002;Huguen et al., 2004).Mud volcanoes are often associated with petroleum deposits, tectonic subduction zones and orogenic belts.They may form domes up to several kilometres in diameter and a height of several tens of meters above the adjacent seafloor (Vogt et al., 1999).Manifestations of mud volcanism are characterised by active seepages of fluids, gas plumes, enhanced heat flow, deposits of gas hydrates and unique chemoautotrophic communities (Damm and Budeus, 2003;Sauter et al., 2006).They are considered a significant source contributing to the marine methane budget (Jerosch et al., 2007) and, as such, of greenhouse emissions to the hydrosphere and atmosphere (Milkov et al., 2003).
The Håkon Mosby mud volcano (HMMV) was discovered in 1989 during a regional side-scan sonar mapping survey (Crane and Solheim, 1995;Vogt et al., 1997;Milkov et al., 1999).To date, the HMMV is the only active mud volcano known from high northern latitude beyond the polar circle.Centred at ca. 72 • 00.25 4 N, 14 • 43.50 E, it is situated on the Norwegian continental slope in the SW Barents Sea (Fig. 1) at ca. 1250 m water depth with a maximum elevation of 10 m and covers an area of ca. 2 km 2 (Krupskaya et al., 2001).
A massive 6 km thick sediment pile that started to accumulate 30-40 million years ago after the continental break-up into what is now Greenland and Eurasia (Vogt et al., 1997;Hjelstuen et al., 1999), lies underneath the volcano.The formation of the volcano might have coincided with a submarine landslide during the late Pleistocene, 330 000-200 000 yr ago (Vogt et al., 1999).Today, fluids rise from a sediment depth of 2-3 km through a conduit below the HMMV (Vogt et al., 1999).The age of the carbonate structures inside the modern caldera has been estimated as 11 400 yr BP (Lein and Ivanov, 2009).
During the R/V Håkon Mosby cruise in 1995, the mud volcano was mapped by 3.5 kHz profiling, sediment coring and heat-flow measurements (Vogt et al., 1997).A gravity core from the flank of the structure contained gas and gas hydrate.High heat flow values (337 mWm −2 ) were recorded.Together with observations of microbial methane (CH 4 ), hy- drogen sulphide gas and CH 4 gas hydrate, it was inferred that the structure was an active gas seep on the seafloor (Vogt et al., 1997).
Subsequent investigations using deep-towed side-scan sonar, sediment coring, CTD and heat flow measurements -8.8 kHz profiling and seafloor photography during the expedition of the R/V Professor Logachev in 1996 -confirmed that the HMMV oozed mud and seeped gas (Lein et al., 1998(Lein et al., , 1999;;Eldholm et al., 1999).This expedition brought the first evidence of the presence of high benthic microbial activity and symbiotrophic fauna (two species of pogonophorans) associated with the mud volcano caldera (Pimenov et al., 1999;Smirnov, 2000).
The HMMV site was revisited in 1998 during the 40th cruise of the R/V Akademik Mstislav Keldysh with a joint team of scientists from Norway, Germany, the USA and Russia.Two Mir submersibles were used for six dives (Bogdanov et al., 1999).Detailed information about the geological features of mud volcano, gas hydrate distribution, biogeochemical processes in the sediment, water and biota were obtained (Lein et al., 2000;Pimenov et al., 2000).The general features of benthic communities were reported, based on camera observations and samples taken by submersibles (Gebruk et al., 2003).It was shown that the megabenthic community is dominated by two species of symbiotrophic siboglinid tube worms (pogonophorans): Sclerolinum contortum, later transferred to the genus Archeolinum (Smirnov, 2008), whose biomasses reached 435 g m −2 (Gebruk et al., 2003) and Oligobrachia haakonmosbiensis (Smirnov, 2000), later re-identified as O. webbi (Meunier et al., 2010) reached 350 g m −2 (Gebruk et al., 2003).Associations of benthic fauna with each pogonophoran species were described.The background community outside the caldera was dominated by the ophiuroids Ophiocten gracilis and Ophiopleura borealis.Pycnogonids (Colossendeis proboscidea), buccinid gastropods and asteroids were also present.The trophic relationships between species were discussed based on the δ 13 C data (Gebruk et al., 2003).
During the expedition ARK XIX/3b of the German research icebreaker R/V Polarstern, with the French remotely operated vehicle (ROV) Victor 6000, in 2003, the microtopography of HMMV was determined, video observations of the seafloor were made and the sediment and water column was sampled (Klages et al., 2004).High-resolution bathymetric data, still photographs and video mosaics were integrated in a geographic information system (GIS).This allowed a detailed analysis of spatial patterns and quantification of biogeochemical habitats by geostatistical methods (Jerosch et al., 2007;Portnova et al., 2010).High concentrations of hydrogen sulphide and methane in the sediment at HMMV and other offshore mud volcanoes offer a specific habitat for microbiological and macrofaunal associations (Sibuet and Olu, 1998;Pimenov et al., 1999;Boetius et al., 2000Boetius et al., , 2002;;Hinrichs and Boetius, 2002;Soltwedel et al., 2005;De Beer et al., 2006;Sauter et al., 2006).
In previous studies three main habitat types have been identified in the HMMV: bacterial mats, pogonophoran fields and plain light-coloured sediments (Gebruk et al., 2003;Jerosch et al., 2007).Preliminary data were given on macroand megafauna associated with these habitats (Gebruk et al., 2003) and on meiofauna (Soltwedel et al., 2005;Portnova et al., 2010).Based on a photographic survey of the HMMV with a towed camera system, we compare composition, density and diversity of megafauna in habitats inside the volcano caldera with different density of seafloor coverage by bacterial mats and pogonophoran tube worms, and in habitats with different combinations of bacterial mats and pogonophorans, to identify an effect of a habitat type on the structure of benthic communities.Also we compare composition, diversity and density of megafauna inside and outside the volcano caldera, to determine an effect of the mud-volcano environment on benthic communities and to identify possible specialists of the HMMV.

Photographic survey
A photographic survey was carried out during the expedition ARK-XVIII/1b of the research icebreaker R/V Polarstern in August 2002 to the HMMV.Different habitats inside the volcano caldera and outside were photographed using a towed camera system (Ocean Floor Observation System, OFOS).Four OFOS transects were carried out in total: three across the caldera (I-III) and one outside the caldera (IV) to the south of volcano (Fig. 2).The length of each transect was ca. 2 km (Table 1).
The OFOS was equipped with a still camera (Benthos Inc.) and a black and white video camera (Deep-Sea Power and Light), two high-intensity discharge (HID) lights of 250 W each, strobes (600 W s −1 ) and three laser pointers.A triangle laser scale with 50 cm between lasers was used to determine the camera's footprint.The still camera was mounted to the frame in a vertical position to the seafloor.It was triggered automatically every 30 s.Additional manually triggered images were taken when features of particular interest occurred in the field of view of the video camera, which was aligned with the still camera to help the winch operator adapting to bottom topography.The still camera was loaded with a 30 m Kodak Ektachrome 100 ASA film, allowing ≤800 exposures (for more details see Soltwedel et al., 2009).The OFOS was towed at approximately 0.5 knots and a target altitude of 1.5 m.From earlier OFOS deployments, this altitude has proven to be the optimal distance to the seafloor to achieve the best illumination and resolution of the images.The area of seafloor on the images ranged from 1 to 8 m 2 owing to the ship's movement, in most cases it was 3-4 m 2 .Start and end positions of OFOS transects (from GPS fixes) and water depths along transects (from echo soundings) were obtained from the ship's data acquisition and management system.

Image analysis and identification of megafauna
In total 1604 film images were examined using a stereo microscope.Of these, 1045 images were used for statistical analyses.Among the analysed images, 894 were taken along the three transects across the caldera and 151 images along the transect outside the volcano.Images of unsatisfactory quality (with sediment clouds, too strong or low illumination, large or small distance from the bottom) were excluded from the analysis.All images within the volcano were taken in the soft-sediment environment.Part of images along the transect outside the volcano were taken on hard substrata (boulders, carbonate rocks and crust).Those images also were excluded from the analysis because the aim of our study was to compare fauna from similar sediment environments.All visible nonsymbiotrophic megafauna were counted and identified to the lowest possible taxonomic level.Taxonomic identifications were made with the assistance of experts (see Acknowledgements).We used the category "morphospecies" for megafauna identified only to high taxonomic level.Recurring biological objects that could not be assigned to any taxa were indicated as "morphotypes".It is possible that some morphotypes (and morphospecies) include several species.Overall, 31 categories (29 species/morphospecies and 2 morphotypes of megafauna) were identified and quantified and used in multivariate statistical analyses and estimations of megafaunal density (Table 2).Density was not determined for the two pogonophorans, Oligobrachia webbi and Archeolinum contortum.For these and for bacterial mats we estimated the degree of seafloor area coverage using the laser scale.Identification and ground-truthing was aided by previously collected material and samples taken outside the caldera using an Agassiz trawl and inside the caldera using a box corer, ROV, trap lander and baited traps.For details on gears and deployments see Table 1 in Supplement and Pimenov et al. (1999), Bergmann et al. (2009) and Hildebrandt  (2011).For a complete list of species found inside and outside the HMMV caldera see Table 2 in Supplement.

Data analysis
The number of organisms on each image was converted to individuals per m 2 (density).Non-metric, multidimensional scaling (MDS) plots were generated for each transect inside the caldera based on the Bray-Curtis similarity coefficient using the PRIMER v6 programme and square root transformed species density matrices.Square root transformation of densities was used to buffer the influence of most abundant taxa.Analysis of similarity (ANOSIM) was used to assess the significance of differences among images from areas with different percent cover of bacterial mats, pogonophorans or images with different combinations of percent cover of bacterial mats and pogonophorans.Contribution of species to similarity of different groups of images was calculated using the SIMPER subroutine of PRIMER v6 (Clarke and Gorley, 2006).ANOSIM and SIMPER were used for comparing images taken on the plain light-coloured sediment inside the caldera and images outside the caldera.
Combined data from three OFOS transects over the caldera were used for the calculation of the total megafaunal density ( ± standard deviation) and the mean taxa densities (± standard deviation) in areas with different coverage by bacterial mats or pogonophorans.The relative contributions of the most abundant taxa to the total density in different areas were calculated.Additionally the mean taxa densities were estimated for areas with different combinations of bacterial mats and pogonophorans.
The following diversity indices were applied to combined data of the three OFOS transects to describe megafaunal assemblages: Pielou's evenness (J ), Shannon-Wiener diversity (H ) (log 2 ) and Simpson's diversity (1-λ) in areas with different coverage by bacterial mats or pogonophorans.Similar calculations were performed for transect IV taken outside the caldera.Taxa richness in different areas was compared based on an equal number of randomly selected images.
We compared total megafaunal densities, mean taxa densities and diversity on images with different coverage by bacterial mats or pogonophorans overall using a Kruskal-Wallis test (Minitab 16 software).Pairwise comparisons of different areas were performed using the Mann-Whitney U Test and Bonferroni corrections of p = 0.05 n −1 , where n is number of comparisons (comparison of bacterial mat sites: n = 10, p = 0.005; comparison of pogonophoran sites: n = 6, p = 0.00833; comparison of area of plain lightcoloured sediments and pogonophoran sites: n = 10, p = 0.005) (Minitab 16).Comparisons of diversities were done for the equal number of randomly selected images from each area because diversity indices are based on the taxa richness, which is affected by the number of images analysed.Similarly, images from outside the caldera were compared with images from the light-coloured sediments using the Mann-Whitney U Test.
Spearman's rank correlations were used to compare taxa richness, total megafaunal densities, diversity indices with coverage of bacterial mats or pogonophorans.

Bacterial mat sites
We used the following five categories for the seafloor area coverage by bacterial mats: 0 %, 0-10 %, 10-50 %, 50-80 % and 80-100 %.ANOSIM revealed significant differences in megafaunal composition and density of images with no bacterial mats at all (0 %) from images with bacterial mats  present and did not reveal significant differences in megafaunal composition and density between images with different coverage of bacterial mats (Global R = 0.538, p = 0.001).SIMPER indicated that five species contributed most to the separation of the mat coverage groups: Ophiocten gracilis, lysianassid amphipods, Nymphon macronix, Metacaprella horrida and Lycodes squamiventer (Fig. 3).The mean densities of these species in areas of different coverage by bacterial mats are presented in Table 4.The most notable differences in density were found in O. gracilis and L. squamiventer.Ophiocten densities decreased significantly from 37.9 ± 38.7 ind.m −2 in the absence if mats to 1.3 ± 5.3 ind.m −2 when bacterial mats (< 10 % coverage) appeared (M-W : W = 271840.0,p < 0.00001).By contrast, the density of L. squamiventer increased from 0.1 ± 0.3 to 1.2 ± 0.6 ind.m −2 when bacterial mats were present (coverage > 80 %).Fish densities on bacterial mats were significantly different between bacterial mats > 80 % and all other bacterial mat categories (M-W test, p < 0.00001) with highest densities recorded in areas of 80 % bacterial coverage and lowest in areas 10 % coverage.
Total megafaunal densities sharply decreased from 43.7 ± 44.3 ind.m −2 to 4.7 ± 8.0 ind.m −2 when the coverage by bacterial mats increased from 0 % to 10 %.The Kruskal-Wallis test indicated significant differences in total megafaunal densities in areas of different coverage by bacterial mats (K-W : H = 417.90,p < 0.0001, DF = 4).The pairwise Mann-Whitney U Test revealed significant differences in the total density of megafauna between images from areas with the following bacterial mat coverage: 0 % and < 10 % (M-W : W = 269903.5,p < 0.00001).The test did not reveal significant differences between images from ar-eas with bacterial mat coverage > 10 % (p > 0.005).Total megafaunal densities were significantly and negatively correlated with coverage by bacterial mats (Table 5).

Area without bacterial mats and pogonophorans (plain light-coloured sediments)
ANOSIM revealed significant differences in the species composition of images taken from light-coloured sediments and pogonophoran settlements (Global R = 0.315, p = 0.001).SIMPER indicated two species contributing most to the separation of images from these areas: O. gracilis and lysianassid amphipods.The mean Ophiocten densities were significantly lower on the plain light-coloured sediment (35.7 ± 21.7 ind.m −2 ) than in areas with a coverage by pogonophorans exceeding 50 % (M-W : W = 108228.5,p < 0.00001), but higher than in areas with a coverage of 0-50 % (M-W : W = 92895.0-100705.5,p < 0.00001).
SIMPER revealed that five species contributed most to the separation of groupings of images with different combination of areas with bacterial mats and pogonophorans: O. gracilis, lysianassid amphipods, N. macronix, M. horrida and L. squamiventer (Fig. 3).
Relative contributions of the most abundant taxa to the total density in areas with different coverage by bacterial mats and tube worms are shown on Fig. 6.Fish accounted for 86 % of the total density (0.8 ± 0.6 ind.m −2 ) in areas with extensive bacterial mats (> 50 %).By contrast, the relative contribution of fish was 12 % (0.2 ± 0.4 ind.m −2 ) in areas of extensive pogonophoran settlements (> 50 %) and 6 % (0.1 ± 0.3 ind.m −2 ) in areas without bacterial mats and tube worms.The relative contribution of ophiuroids increased from 0.8 % (0.1 ± 0.4 ind.m −2 ) in areas with extensive bacterial mats to 39 % (40.2 ± 62.3 ind.m −2 ) in areas of extensive pogonophoran fields and up to 87 % (34.4 ± 21.6 ind.m −2 ) in areas without bacterial mats and tube worms (on light-coloured sediments).The highest relative contribution of lysianassids (26 %) and caprellids (0.9-2 %) was found in the pogonophoran area.

Comparison of megafauna from three zones inside the volcano caldera
Inside the HMMV caldera the following three main habitats can be distinguished: bacterial mats, pogonophoran settlements and light-coloured sediments (Fig. 7).The habitats are arranged in zones around the volcano centre in a more or less concentric pattern, however, boundaries between zones are not always clear because bacterial mats and pogonophoran colonies may co-occur in different proportions (Fig. 8).Variations in the mean density of selected taxa and area coverage by bacterial mats and pogonophorans in different zones are shown on Fig. 8. Zone I.A zone of bacterial mats is most clearly distinguished around the caldera centre.The zone is characterised by extensive bacterial mats on grey or yellowish sediment.Megafaunal densities ranged from 0.1 to 52.7 ind.m −2 between individual images (mean 3.3 ± 6.4 ind.m −2 ).The fish Lycodes squamiventer reached a density of 0.6 ind.m −2 in this zone.
Zone II.The zone of settlements of siboglinid tube worms (pogonophorans) Archeolinum contortum and Oligobrachia webbi is located further away from the centre of the caldera.Most of the megafauna is absent from this zone, whereas the density of pycnogonids, N. macronix, caprellids, M. horrida, lysianassid amphipods and bivalve molluscs clearly increased on siboglinid fields.Total megafaunal densities in this zone ranged from 0.1 to 271.2 ind.m −2 between individual images (mean 34.0 ± 56.0 ind.m −2 ).Ophiuroid densities ranged from 0 to 248 ind.m −2 between individual images (mean 24.9 ± 48.1 ind.m −2 ).
Zone III.The zone of plain light-coloured sediments devoid of bacterial mats and pogonophorans forms the periphery of the caldera.This zone was dominated by the ophiuroid O. gracilis.Most of the megafauna taxa, first of all the echinoderms, were abundant in this zone.Total megafaunal densities ranged from 0.1 to 116.Some of the megafauna (mainly relatively large mobile crustaceans and rays) were present throughout the caldera with no clear association to any particular habitat type.
The highest density of non-symbiotrophic megafauna was recorded in the zone of light-coloured sediments, followed by the zone of pogonophoran settlements.The lowest densities were recorded from the bacterial mat zone (Fig. 9).
ANOSIM revealed significant differences in megafaunal composition and density on images taken inside the caldera on light-coloured sediments devoid of bacterial mats and pogonophorans and images outside the caldera (Global R = 0.378, 0 = 0.001).SIMPER showed that the following three species contributed most to this separation of images: O. gracilis, sabellid polychaetes and S. borealis.
The composition of the dominant taxa in the background area differed from that inside the caldera.Only two taxa, O. gracilis and lysianassid amphipods, were abundant in both areas (Table 7).In the background area, other abundant taxa included the sabellids, S. borealis and unidentified sedentary organisms absent or found in low numbers inside the caldera.The mean density of O. gracilis was significantly lower in the background area (22.3 ± 21.8 ind.m −2 ) compared to the zone of light-coloured sediments inside the caldera (35.7 ± 21.7 ind.m −2 ) (M-W : W = 29635.0,p = 0.00001).Similarly, the mean density of lysianassid amphipods was significantly lower in the background (0.3 ± 0.7 ind.m −2 ) compared to the zone of light-coloured sediments inside the caldera (0.5 ± 1.6 ind.m −2 ) (M-W : W = 91731.5,p = 0.0236) (Table 4).
In total, 24 taxa were found inside the volcano caldera and 26 outside on the OFOS images.In both areas 19 taxa were present (Table 2).However, sedentary biota, such as sponges, cnidarians and some unidentifiable morphotypes were only recorded in the background area.Based on the equal number of randomly selected images, taxa richness was higher outside the caldera (26) than in the area of light-coloured sediments inside the caldera (21).

Discussion
The Håkon Mosby mud volcano has received considerable attention over the last decade as one of the target areas of ambitious EU-funded programmes such as HERMES and HERMIONE and various nationally funded projects.However, studies on macro-and megafaunal communities inhabiting the HMMV are still very limited and our knowledge of the volcano background community remains poor.
We were able to identify 40 taxa in total (including species, "morphospecies" and "morphotypes") based on images and samples taken inside the volcano caldera.Gebruk et al. (2003) reported 46 macrofaunal benthic taxa from this area, Decker, Olu (2012) and Decker et al. (2012) reported 28 macrofaunal benthic taxa.A large portion of taxa listed by these authors included small and infaunal forms, such as polychaetes and amphipods that are not easily detected on seafloor photographs and were missing in our records.Based on combined data of Gebruk et al. (2003), Decker, Olu (2012), Decker et al. (2012) and our present study, the species inventory of the HMMV comprises at least 80 taxa.In the present study, we added at least 20 species/morphospecies to the inventory: Cnidaria (4), Annelida (4), Sipunculida (1), Cephalorhyncha (1), Crustacea (1), Echinodermata (6), Pisces (3) (Table 2 in Supplement).
Taxa richness did not differ noticeably inside and outside the caldera: 40 vs. 36, although the number of species outside the caldera could be higher if three transects (as inside the caldera) had been taken or the sampling effort had been more intense.A photographic survey at the HAUSGARTEN observatory west of Svalbard (∼1300 m depth) revealed lower taxa richness (12 taxa/morphotypes) (Soltwedel et al., 2009).Thus, Vinogradova et al. (1996) reported 200 benthic taxa based on 22 trawls at the Russian nuclear submarine Komsomolets wreck site at 1700 m depth in the Bear Island area, approximately 120 miles north of the HMMV on the continental slope.The benthic community at this site (based on the biomass) was termed the "Bathybiaster-Ophiocten-Amphipoda-Oweniidae" assemblage.Similar to the HMMV, benthos at the Komsomolets site was dominated by echinoderms (primarily asteroids and ophiuroids), amphipods and polychaetes.In our study, in trawl catches outside the caldera ophiuroids and asteroids were also dominant.
Two species of siboglinid tube worms (perviate pogonophorans), Archeolinum contortum and Oligobrachia webbi, dominated inside the caldera.Both genera are common at depths of the Norwegian-Greenland Basin (NGB) and are usually associated with reducing biotopes.They were found at methane seeps at 1230 m depth on the Vestnesa Ridge (Gebruk et al., 2003) and on the Norwegian slope at the Nyegga site at ∼ 720 m depth (Krylova et al., 2011).Archeolinum contortum was also abundant at the recently discovered deep-sea hydrothermal vent site 'Loki's Castle' at a depth of 2350 m at the junction of the Mohn and Knipovich ridges (Pedersen et al., 2010).The species O. webbi was reported from the shallower hot vent site on the Mohn Ridge north of Jan Mayen (500-750 m water depth; Schander et al., 2010).Both siboglinid species apparently are highly specialized to reducing habitats and can inhabit their different types: methane seeps and hydrothermal vents, sunken wood and other decomposing organic material (Gebruk et al., 2003).
According to our data, the density and proportions of taxonomic groups noticeably varied between habitats inside and outside the caldera.The megafaunal species observed along OFOS transects could be grouped according to their occurrence in different habitats.Many species abundant outside the caldera and at the inactive periphery of the volcano were either completely absent or very rare in the pogonophoran and bacterial mat zones.Among such species were echinoderms (echinoids, starfish, crinoids and holothurians), polychaetes, sponges and nemertea.An exception was the ophiuroid O. gracilis, which was quite abundant in the dense settlements of pogonophorans (> 50 %) without bacterial mats.The mean density of O. gracilis in this habitat was higher than in the background: 63.6 ± 67.8 ind.m −2 and 22.3 ± 21.8 ind.m −2 , respectively.A photographic survey at the HAUSGARTEN observatory west of Svalbard (∼1,300 m depth), conducted also on the ARK XIX/3 cruise, revealed slightly lower densities of O. gracilis (16.8 ± 3.7 ind.m −2 ) and an aggregated occurrence (Soltwedel et al., 2009).Ophiocten gracilis occurs in a wide range of habitats; this species appears to be very adaptive and opportunistic.In hydrothermal vent habitats on the Mid-Atlantic Ridge ophiuroids (Ophiocten acies) reached densities as high as 80 ind.dm −2 (Gebruk et al., 2000) in the Logatchev area.
Thus, the effect of the mud-volcano environment on benthic communities is expressed in increasing of biomass, changing of taxa composition (absence of some groups known outside the volcano and emergence of some groups unknown outside the volcano) and changing of proportions of most taxonomic groups.The most special groups for HMMV environment excluding symbiotrophic megafauna were pisces Lycodes squamiventer, lysianassid amphipods, pycnogonid Nymphon macronix, caprellid Metacaprella horrida and ophiuroid Ophiocten gracilis.
In the HMMV area both pogonophorans are restricted to the volcano and the nearest surrounding (Gebruk et al., 2003).Populations of the more abundant A. contortum cover thousands of square meters of the seabed.The georeferenced video mosaics and detailed spatial analyses by Jerosch et al. (2007) illustrate particularly dense populations of these tube worms in the so-called 'hummocky periphery' of the caldera and estimated the total area covered by tube worms at 275 958 m 2 , by dense bacterial mats (density > 50 %) at 31 435 m 2 and the central caldera part at 115 165 m 2 .Our data indicate that the highest densities of A. contortum are in areas with bacterial mat coverages of less than 50 % of the sea floor.Our results based on the analysis of 894 images from HMMV disagree with conclusion of Niemann et al. (2006) that sulphide-oxidizing mats and tubeworm colonies are mutually exclusive.A great number of images showed both biota.
Some megafaunal species at HMMV were clearly associated with pogonophorans.Among them were the caprellid amphipod Metacaprella horrida, the pycnogonid Nymphon macronix and lysianassid amphipods.The density of burrowing bivalve molluscs also increased on the pogonophoran fields.Total megafaunal densities of pogonophoran fields would increase dramatically if the tube worms themselves were added to our calculations.However, on the seafloor images it is almost impossible to determine even approximate numbers of tubes, especially when they are curled as in A. contortum.Also many tubes in dense populations are empty (A. Gebruk, personal communication, 1998).
Overall, total megafaunal densities clearly increased on the pogonophoran fields (> 50 % coverage) compared to the background habitat (25.3 ± 21.0 ind.m −2 ) and dropped to minimum values on the bacterial mats (> 50 %).The limited number of megafauna species able to live in a "moderate" reducing environment are apparently capable of taking advantage of low competition in partitioning of resources and reach high population densities.Models based on temperature profiles indicate lower fluid flow rates (0.4 m yr −1 ) in the pogonophoran fields compared to those of the centre or bacterial mat zone (Kaul et al., 2006).The increased habitat complexity provided by pogonophorans together with a more stable environment may result in higher megafaunal densities.Furthermore, Niemann et al. (2006) concluded that microbe-invertebrate symbioses have an advantage over free-living microbial populations as they can engineer their environment to increase access to both electron donors and acceptors by special migratory behaviour, mining and pumping.
The bacterial mat zone was characterized by the lowest megafaunal density and species richness.The opposite trend was shown for the density of meiofauna at the HMMV it was the highest in the bacterial mat zone compared to pogonophoran fields and the background (Portnova et al., 2010).In our study, megafaunal density decreased sharply even at low bacterial coverage (< 10 %).The only species with higher densities on bacterial mats was the fish Lycodes squamiventer.This species had significantly higher growth rates at the HMMV compared with conspecifics from HAUS-GARTEN caught at comparable depth (Hildebrandt et al., 2011) indicating favourable local growth conditions at the mud volcano.Furthermore, stable isotope analyses showed that the δ 13 C and δ 15 N in L. squamiventer muscle tissues from the HMMV were significantly lower than those from HAUSGARTEN indicating that the fish from HMMV relied on food linked to the methane-based food web (Gebruk et al. 2003;Bergmann et al. 2009, and unpubl. data).Zoarcidae is a fish family often encountered at hydrothermal vents and cold seeps, with some specialised species that are likely to have adapted to the toxic environment (Desbruyères et al., 2006).
The density of megafauna in our study was relatively high.For example, on the mud volcanoes in the eastern Mediterranean the density of non-symbiotrophic megafauna were two or three orders of magnitude lower (Olu-Le Roy et al., 2004).However, the depth of these volcanoes is greater, approximately from 1700 to 2000 m.
The three habitats (zones) distinguished at the HMMV are arranged in a more or less concentric pattern: the central zone, around it pogonophoran fields and the light-sediment zone in the periphery.The latter two zones harbour a complicated mosaic of bacterial mats.Such a concentric structure of a habitat and community pattern is more typical of hydrothermal vents, centred around hot fluid emissions (Galkin, 2002).For example, Hessler and Kaharl (1995) distinguished "the vent opening, the near-field and the periphery".This obvious simplification points at pronounced gradients of environmental parameters that are very characteristic of hydrothermal vents and other habitats with focused emission of heat and reduced compounds, one of such habitats being mud volcanoes.At typical cold methane seep sites the pattern of fauna distribution is usually more heterogeneous and is first of all controlled by fluid flow (Sahling et al., 2002;Levin, 2005).
The central zone of the volcano appears to be devoid of megafauna most likely because of the instability of the environment.Indeed, Feseker et al. (2008) reported an exceptionally high level of mud volcanic activity with temperature gradients of more than 40 • C m −1 within the upper half meter of the sediments close to the geometric centre of the caldera, frequent mud eruptions of small volume and limited local extent.
The variety of reducing habitats known at present in the Norwegian-Greenland Basin (NGB) includes two "cold" methane seep sites -Nyegga at 710-730 m depth on the Norwegian slope in the Storrega slide area (Ivanov et al., 2007(Ivanov et al., , 2010) ) and the methane seep area at 1230 m depth on the Vestnesa Ridge (Gebruk et al., 2003), which includes the relatively hot and active mud volcano (HMMV), a shallow-water hydrothermal vent north of Iceland, Kolbenisey (∼ 90 m depth) (Fricke et al., 1989), two upper bathyal hydrothermal vents, Grimsey (400 m) north of Iceland (Hannington et al., 2001) (no biological investigations yet) and the area north of Jan Mayen on the Mohn Ridge with the fields Trollveggen (500-560 m), Galionella Garden (∼ 615 m), Soria Moria I and II (700-750 m) (Schander et al., 2010), and one recently discovered hydrothermal vent site, Loki's Castle, at 2350 m depth, at the junction of the Mohn and Knipovich Ridges (Pedersen et al., 2010).According to published data, most of the macro-and megafauna of methane seeps and hydrothermal vents in the NGB, including the HMMV, is not specialised to these habitats, it comprises primarily regular background fauna (with exception of two species of pogonophoran tube worms, opportunists of reducing environments).This feature is unusual: in other areas, reducing habitats are dominated by highly specialized megafaunal species restricted in their distribution to a specific reducing habitat or rarely to several of them.In the NGB the only species clearly obligate to methane seeps is the bivalve Isorropodon nyeggaensis (Krylova et al., 2011), representative of the family Vesicomyidae, one of the iconic taxa of reducing environments (Levin, 2005;Desbruyères et al., 2006).The lack of vent/seep specialists in the NGB may result from a young age of fauna: the recent deep-sea benthic fauna north of the Faroe-Iceland Ridge shows signs of origin after the end of the last glacial period (Golikov, 1985;Svararsson et al., 1993).

Fig. 1 .
Fig. 1.Location of the Håkon Mosby mud volcano in the south-west Barents Sea.

Fig. 2 .
Fig. 2. Positions of the OFOS transects I, II and III taken across the Håkon Mosby mud volcano caldera area and the reference transect IV taken outside the volcano caldera.

Fig. 5 .
Fig. 5. Mean densities (ind.m −2 ) of selected taxa in areas with combinations of different coverage by bacterial mats (BM) and pogonophorans (P).Relative location within the volcano caldera is shown.

Fig. 6 .
Fig. 6.Relative contribution of the most abundant taxa to total density in areas with different bacterial mat (BM) and pogonophoran (P) cover.Data of three OFOS transects (I, II and III) combined.

Fig. 9 .
Fig. 9. Mean densities (ind.m −2 ) of selected taxa in different zones of the Håkon Mosby mud volcano.Data of three OFOS transects (I, II and III) combined.

Table 1 .
Details of OFOS transects.Transects I, II and III -inside the caldera, transect IVa and b -reference site.

Table 2 .
Megafauna and other categories recorded on OFOS transects.

Table 3 .
Indices of megafaunal diversity and megafaunal density (ind.m −2 ) in areas of different coverage by bacterial mats and pogonophorans inside the HMMV caldera (OFOS transects I, II and III combined) and the background (OFOS transect IV).

Table 4 .
Mean density (ind.m 2 ) and standard deviations (SD) of the most abundant taxa in areas of different coverage (%) by bacterial mats and pogonophorans.Data from OFOS transects I, II and III combined and from the background transect IV.

Table 5 .
Spearman's rank correlations between taxa richness, total megafaunal densities and diversity indices with coverage of bacterial mats or pogonophorans.* p < 0.05.