Cold-seep assemblages on a giant pockmark off West Africa: spatial patterns and environmental control


K. Olu-Le Roy, Département Etude des Ecosystèmes Profonds (DEEP), IFREMER Centre de Brest, BP70-29280 Plouzane, France.


A giant pockmark colonised by dense cold-seep assemblages near 3160 m depth along the Congo-Angola margin has been surveyed by the ROV Victor 6000. The quantitative distribution of chemosynthetic communities was mapped along the dive tracks from a video study using GIS and image mosaicking. Several types of faunal assemblages, either dominated by bivalves of the families Mytilidae (Bathymodiolus sp.) or Vesicomyidae (Calyptogena sp., ‘Vesicomya’ aff. chuni), or by Siboglinidae polychaetes (Escarpia southwardae) were mapped over the 800-m diameter pockmark area and sampled for fauna, water and sediment. The isotopic analyses (δ13C) of tissues from symbiont-bearing species were within the range typical of nutrition via symbiosis using methane for mussels and sulphide for vesicomyids and siboglinids. The living chemosynthetic communities were distributed on a SW-NE axis, corresponding to the expression at the sediment surface of a main buried channel providing fluids to the pockmark. The site was characterised by a more active central part in a depression with abundant carbonate concretions where high-density clusters of siboglinids and mytilids dominate. Large fields of dead and live vesicomyids with a lower mean density were observed in the external areas. The mean coverage of each of the three symbiotic taxa in these two contrasted areas was estimated from mosaic analysis and was up to 30% in the central area dominated by E. southwardae bushes (23%). Symbiont-bearing species distribution was consistent with methane concentrations in seawater that were generally higher in mytilid beds than in the vicinity of siboglinids and vesicomyids. A Principal Component Analysis performed on environmental factors at the ten sampling sites revealed that 37% of the observed variance in the distribution of symbiont-bearing species may be explained by variation in both methane and oxygen concentrations, while a Canonical Redundancy Analysis selected methane concentration as the only variable which explains symbiont-bearing species densities. This spatial distribution of chemosynthetic species at the pockmark scale may reflect temporal patterns of succession of both substrate and fauna, and may be related to different individual pockmarks visible on the microbathymetry mapped using ROV data.


There is an extensive spatial heterogeneity of microbes and fauna at seeps (Levin, 2005). The species richness of the symbiotic megafauna is highly variable among sites and may be explained partially by depth (Sibuet & Olu-Le Roy, 2002), but is also likely to be the consequence of habitat heterogeneity. Mytilid, vesicomyid and lucinid bivalves, and siboglinid polychaetes have been described from the Louisiana slope and the Barbados prism, but at different sites. It has been suggested that their distribution may be controlled by geology, including the geometry of seeps, the pattern and volume of fluid flow, and the occurrence of hydrates (MacDonald et al., 1990; Aharon, 1994; MacDonald et al., 2003). Likewise, local geology can regulate the chemistry of seeps (e.g. methane and oxygen concentrations), the presence of brines, hydrocarbon composition (Sassen et al., 1994; Bergquist et al., 2005), and related faunal composition. The distribution of different Mytilidae species was also related to fluid emission intensity through mud volcanoes (Olu et al., 1996). The distribution of Vesicomyidae and other infaunal bivalves has been related to sulphide concentrations and gradients in the sediment (Barry et al., 1997; Sahling et al., 2002; Levin et al., 2003). Although several cold-seep communities have been described in the Western Atlantic, the only described cold-seep site in the Eastern Atlantic is the Haakon Mosby Mud Volcano, which is covered by extensive bacterial mats and small Siboglinidae of the Monilifera group (Milkov et al., 1999; Gebruk et al., 2003). The occurrence of seeps off Mauritania is assumed from the lucky collection by fishermen of the new large Bathymodiolus mauritanicusCosel, 2002, and in the Gulf of Cadiz from sampling of large solemyid bivalves Acharax sp. and bathymodiolids on mud volcanoes (M. Cuhna, personal communication). The first seep community described in the South East Atlantic was discovered in 2001 on a giant pockmark ‘REGAB’ on the Gabon continental margin (Ondréas et al., 2005).

Pockmarks are sedimentary depressions caused by the seepage of fluids through the seabed. They are found worldwide along both passive and active margins, but mostly on continental shelves (Barber et al., 1986; Hovland & Judd, 1988). However, a few cold-seep communities have been described on pockmarks. One was described on a 150-m deep pockmark in the North Sea (Dando et al., 1991), but the best known example is the Bush Hill site in the Northern Gulf of Mexico, comprising a large mussel bed spread around a brine-filled pockmark at 650 m depth (MacDonald et al., 1990).

In a recent paper, thanks to a combination of video and side scan sonar mosaics collected by submarine, it has been possible to estimate the areal coverage of some cold-seep communities and the areas occupied by different components of the community at ‘Bush Hill’ (MacDonald et al., 2003). But the quantitative distribution of seep communities has, until now, rarely been included in ecological studies. Sibuet & Olu-Le Roy (2002) proposed a classification of cold seeps with respect to the density of clusters in the active area and organisms in the clusters and the size of the active area. Variation in the density or biomass of chemosynthetic fauna at seeps has been related to the intensity of fluid flow (Henry et al., 1992; Olu et al., 1997; Sibuet & Olu-Le Roy, 2002; Olu-Le Roy et al., 2004; Mau et al., 2006), and to the age of the structure or mud flow extent on mud volcanoes (Olu et al., 1997). The spatial distribution of symbiont-bearing species on mud volcanoes has also been related to the structure type (diapirs/mud pies). Like volcanoes, pockmarks also have different origins, associated with fault zones, salt diapirs or a Bottom Simulating Reflector (BSR) (Gay et al., 2003) and may differ in terms of fluid flow.

The spatial patterns of chemosynthetic communities are thus assumed to vary according to the type or activity of pockmarks. Gay et al. (2006) proposed the first model of seafloor features, including fauna distribution, related to methane flow, at the REGAB site on the basis of a single video transect. However, we think that the model simplifies the spatial distribution of organisms (Olu-Le Roy, 2006).

The present ecological study combines measurements of the quantitative distribution of megafauna and environmental parameters at different spatial scales to estimate reliably the spatial distribution and coupled environmental and biological patterns at the REGAB chemosynthetic community. We mapped the community using GIS-based software and mosaicking, allowing a quantitative estimate of the extent, cover, and distribution of fauna assemblages on a giant pockmark. The distribution of the three symbiont-bearing species – Mytilidae and Vesicomyidae bivalves and large Siboglinidae polychaetes – was analysed in the context of geological and geochemical patterns in this spatially variable cold-seep site. Biotic and abiotic variables including faunal density and biomass, substratum observations and chemical measurements in water and sediment close to the fauna, were collected simultaneously at different assemblages dominated by either Mytilidae, Vesicomyidae or Siboglinidae, to compare different micro-environments and the response of the chemosynthetic fauna. The major interest of this pockmark is the unusual occurrence of these three symbiotic taxa; each is more typically the sole dominant taxon at seep sites, and their biotopes are often studied separately.

Material and methods

Study site and symbiont-bearing species

The 800-m wide, 15–20 m deep pockmark REGAB was first observed as a dark spot on multibeam imagery where it appeared as a circular-shaped pockmark, apparently undisturbed by bottom currents (Ondréas et al., 2005; Fig. 1). This pockmark is located over the oceanic crust (Moulin et al., 2002, Ondréas et al., 2005) <8 km north of the Zaïre canyon (Fig. 1). It is in an area where other fluid escape features have been observed (Kasten et al., 2001; Gay et al., 2003), including another large pockmark discovered c. 60 km to the north (G. Borhmann, personal communication). The REGAB pockmark is linked to a deep channel acting as a reservoir for superficial fluids (Ondréas et al., 2005) by a 300-m long rooting pipe visible on seismic profiles, along which gas escapes. Small-scale faults, which may drive fluids upward, dip towards the centre of the feature, and develop within a tulip-shaped anomaly visible on seismic profiles (Gay et al., 2006). Although a BSR was not observed on the profile, gas hydrates were sampled in gravity cores from the surface to a depth of 6 m in the sediment (Charlou et al., 2004) and hydrate outcrops were observed by ROV on the seabed (Fig. 2). Analyses of gas hydrate samples indicate that methane is the major chemical component of the emitted fluids (Charlou et al., 2004). The shallow burial of the paleochannel (<300 mbsf) suggests a biogenic origin of the fluids, confirmed by isotopic analyses on hydrates and fluid (Charlou et al., 2004; D. Levaché, personal communication) and the occurrence of a hydrocarbon reservoir linked to the pockmark is unlikely. The term ‘pockmark’ is used throughout the paper to name the whole REGAB site, which is a depression of 15–20 m depth. Nevertheless, it is composed of several individual pockmarks <100 m in diameter which were identified on a microbathymetric map (Ondréas et al., 2005) and REGAB can thus be considered as a ‘pockmark cluster’. Dense communities dominated by large symbiont-bearing species including the Vesicomyidae bivalves ‘Vesicomya’ aff. chuni and Calyptogena n. sp. (R. von Cosel, personal communication), the Mytilidae Bathymodiolus sp. aff. boomerang (K. Olu-Le Roy et al., unpublished data) and the Siboglinidae polychaetes Escarpia southwardae (Andersen et al., 2005) were observed on the pockmark. All these species are associated with symbiotic chemoautotrophic bacteria (Duperron et al., 2005; Ondréas et al., 2005).

Figure 1.

 Location of the REGAB site along the Gabon continental margin. The giant pockmark is located 8 km north of the Zaire canyon.

Figure 2.

 (a) Methane hydrate outcrop; (b) gas bubbles escaping among a mussel bed.

Video survey and analysis

The REGAB pockmark was first surveyed during two dives of the ROV Victor 6000 (ZAIROV cruise, 2000) and five subsequent dives during the BIOZAIRE 1 and 2 cruises were dedicated to faunal communities (Fig. 3). Megafauna assemblages were primarily classified according to the dominant megafaunal species: the mytilid Bathymodiolus sp., the siboglinid Escarpia southwardae, and the vesicomyids. It was impossible to distinguish between ‘Vesicomya’ aff. chuni and Calyptogena n. sp. using video images. The distribution of assemblages was mapped along the dive tracks from the video study of seven ROV dives and using the Adelie extension for ArcGIS 8.0 (©ESRI) developed at Ifremer ( The first dive track included 10 N-S transects over the pockmark, and two transects crossing the pockmark along a SW–NE axis (Fig. 3). Subsequent dives were undertaken along this SW–NE direction where the living cold-seep communities and the concretions were observed. A point on the GIS plotted each chemosynthetic faunal assemblage during the video processing. Points with similar identifications (similar taxa) separated by <10 m were associated, that is to say the width of the forward-viewing video image during exploration dives at about 3 m above the bottom. As the video survey covered a large proportion of the active zone, the areas covered by each assemblage could be visualised and their surface estimated. The substratum was characterised simultaneously and classified as soft bottom or carbonate concretions.

Figure 3.

 Location of the seven dive tracks and main sampled site on the REGAB site.

Detailed bathymetry

The original map of the REGAB site was derived from four dives by Ondréas et al. (2005) using Victor positioning and depth data (the addition of vehicle elevation above the seafloor and immersion). We improved the map for this paper by adding data from three new dives targeting the exploration and sampling of the central active part of the site. These new data revealed the bathymetric relief of the site and allowed more detailed comparisons with the small-scale distribution of the fauna. The map was processed using the geostatistical analyst extension of ArcGIS 8.0 (© ESRI) and the ordinary kriging method for interpolation (spherical model of a variogram).

Video mosaic analysis

Video mosaicking was used to allow the detailed analysis of faunal distributions. Two areas were selected for mosaic creation. Mosaic 1 (70 × 50 m) is located in the largest vesicomyid field at the South-West part of the pockmark (soft bottom area) and Mosaic 2 (200 × 40 m) is positioned along the SW–NE axis colonised mostly by mytilids and siboglinid bushes with a dominance of hard bottom (Fig. 3). These two boxes were covered by video imagery, using the vertically mounted Tri-CCD camera of the ROV Victor. Transects of the mosaic were reconstructed using the MATISSE system developed at Ifremer (Allais et al., 2004). This system produces geo-referenced mosaics from a video input and navigation data. Each mosaic was about 3 m wide, corresponding to the mean altitude of the ROV during the survey, and was from 10 to 30 m long (500–1000 images per mosaic). As many as 120 mosaics of 30 m2 on average and 80 mosaics of about 100 m2 were analysed for mosaic 1 and 2 respectively, using ADOBE Photoshop® to draw faunal patches. SIGMA SCAN PRO® was used to estimate the patch size for faunal groups and subsequently the total coverage of each assemblage type, either dominated by Bathymodiolus sp., Vesicomyidae, or E. southwardae.

Density and biomass estimations

According to the first map of assemblage distribution, eleven assemblages (or ‘sites’) were selected for fauna, water and sediment sampling all over the pockmark (Figs 3–5). The assemblages dominated by Bathymodiolus sp. (M1, M2 and M3) have different cluster sizes. M3 is associated with erect E. southwardae (Fig. 5D) and M1 with recumbent ones. The selected vesicomyid clusters have different proportions of living and dead individuals and are either isolated in the active centre of the pockmark (VC and V3), in the large vesicomyid field (VA, VB and V1) or in the external, less active area (Vext). Two erect bushes of the escarpiid E. southwardae were sampled, one of large adult individuals about 1.5 m long (EA, Fig. 5E), and one with smaller tubes of <0.5 m long, presumably juvenile individuals (EB, Fig. 5F). Bivalve density was estimated from close-up views using four laser points spaced 23 cm from each other or from known size sampling tools for surface estimation. When possible, density was averaged from several images (6–16) or from complete mosaics of the sampling site. Tubeworm density was estimated using close-up images of the tubes in the vertical plane, at different levels above the seafloor for the same bush and the surface was estimated at the base of the bush, using lasers. However, it is likely that this method underestimates siboglinid density as all the tubes in the 3D bushes could not be enumerated.

Figure 4.

 Detailed bathymetry of the active central part of the REGAB site and location of the sampling sites. White dots indicate were gas bubbling was observed during the dives.

Figure 5.

 Assemblages dominated by symbiotic megafauna species. (a) Vesicomyidae cluster V1, (b) close-up view on Vesicomyidae at V1, (c) Bathymodiolus sp. bed M2, (d) Bathymodiolus bed and Escarpia southwardae M3, (e) E. southwardae bush, (f) young E. southwardae bush EB.

Individual wet weights were estimated from specimens collected by the ROV grab or a slurp gun and preserved in formalin, and dry weights after 48 h at 60 °C. Ten individuals of various sizes of Bathymodiolus sp. from M1, vesicomyids from VB and V3 and E. southwardae from EA and EB were available. Bivalve shells and siboglinid tubes were removed before wet or dry weight determination. Wet and dry weights were positively correlated (Pearson correlation coefficient = 0.98, P = 0.001).

Chemical analyses in water and sediment

Fluids were sampled using the ROV water sampling pump and collected in 200 ml titanium bottles at the different sites (Table 1). A suction cannula mounted on the ROV allowed accurate sampling of water among mytilids and vesicomyids a few centimetres above the sediment–water interface (bottom), just above the bivalve anterior part (top) (Fig. 5B), at the base of the escarpiid tubes and near the worm plume. Additional water samples were obtained a few decimetres above or around the vesicomyid clusters, to highlight vertical and horizontal gradients and confirm that these clusters were associated with active methane outflows. Three sampling points from the centre to the edge of the 2-m diameter patch were obtained in the M3 mussel bed. Water pH measurements were performed on board with a pH-meter linked to a glass electrode (metrohm). Dissolved oxygen was determined by the Winkler method adapted for small volumes, with an error of 4%. Hydrogen sulphide was measured by the colorimetric method (Fonselius, 1983, error of ±5%). For sulphide analyses, 5 ml of ZnCl2 (0.2 mol·l−1) was pre-loaded into titanium sample bottles to precipitate the sulphide in each seawater sample as zinc sulphide before analysis. Methane and total carbon dioxide concentrations were measured on board using the headspace technique coupled with a gas chromatograph GC (HSS-GC) equipped with a thermal-conductivity detector (TCD) and a flame-ionisation detector (error of 4%) (Sarradin & Caprais, 1996).

Table 1.   Sampling sites.
Sampling siteCruiseFauna/substrateMeasured parameters in waterMeasured parameters in sediment
  1. O.C., organic carbon; PAHs, polycyclic aromatic hydrocarbons; THCs, total hydrocarbons.

M1BZ2Bathymodiolus sp.CH4, CO2, O2, pH, H2SO.C., O2 profile, PAHs, THcs
M2BZ2Bathymodiolus sp.CH4, CO2, O2, pH, H2SO.C., THcs
M3BZ2Bathymodiolus sp.CH4, CO2, O2, pH, H2SO.C., O2 profile, THcs
VABZ1Vesicomyidae (living)CH4, CO2, O2, pHO.C.
VBBZ1Vesicomyidae (dead)CH4, CO2, O2, pH 
VCBZ1VesicomyidaeCH4, CO2, O2, pH 
V1BZ2VesicomyidaeCH4, CO2, O2, pH, H2SO.C., PAHs, THcs
V3BZ2VesicomyidaeCH4, CO2, O2, pH, H2SO.C., O2 profile, PAHs THcs
EABZ1Adult Escarpia sp.CH4, CO2, O2, pH 
EBBZ1Young Escarpia sp.CH4, CO2, O2, pH 

Sediments were collected as near as possible to assemblages by push cores 30 cm long and 5.5 cm in diameter. The sediment cores were split on board and frozen for later analysis. Dry sediment reduced in powder was analysed for total C content with the auto-analyser LECO CS-125, and for organic carbon using the auto-analyser LECO WR 12 after sediment decarbonation with 6 N HCl. Total hydrocarbon measurements were performed by infra-red spectrometry according to the French AFNOR norm (NF T90-114, AFNOR, 1979). Polycyclic aromatic hydrocarbons (PAHs) were analysed by gas chromatography coupled with mass spectrometry (GC/MS). Oxygen profiles were performed in the sediment using an oxygen minisensor OX 100 coupled to a picoammeter (PA 2000, Unisense) and a micro-manipulator using a profix data acquisition software (Unisense). The spatial resolution of the sensor was <200 μm.

Multivariate analyses

A principal component analysis (PCA) was performed on environmental factors – methane and oxygen concentrations and substratum – at the ten sampling sites (M1, M2, M3, VA, VB, VC, V3, V1, EA and EB). The substratum was binary coded for presence or absence of soft sediment and carbonate concretions. A canonical redundancy analysis (RDA) was also run to analyse species distributions among the different sampling sites, according to environmental factors. The first matrix included the table densities of the three symbiont-bearing species at nine sampling sites and the second matrix included environmental factors at the same sampling sites. Both PCA and RDA were performed using the ‘rda’ function of the vegan R-language library and the R package© (R Development Core Team, 2004). We used a procedure to select the environmental variables of the RDA that contribute significantly to modelling the faunal densities at the different sampling sites. This procedure was developed by S. Dray and is available in the vegan R-language library.

Isotopic measurements

Mytilids, vesicomyids and siboglinids from the different sites were dissected on board and the bivalve gills and mantles and the siboglinid trophosomes, obturacula and vestimenta were frozen. After vacuum lyophilisation, 23 individuals of Bathymodiolus sp., nine of both species of vesicomyids and 12 adults and juveniles of E. southwardae were analysed in triplicate to work out the carbon and nitrogen isotopic ratios using a DELTA Plus (thermo Finnigan) isotopic mass spectrometer (LPTC, Bordeaux, France).


Distribution of symbiotic megafauna assemblages at the pockmark scale

Chemosynthetic assemblages were observed throughout most dive tracks in the circular area (800 m diameter) identified previously by its high acoustic backscatter in Simrad EM12 imagery (Ondréas et al., 2005). However, most live seep fauna was concentrated along a SW–NE axis, crossing the centre of the pockmark (Fig. 6). Scattered aggregations (c. 1 m in diameter) of vesicomyid clams (mostly dead) were observed in small depressions along the first dive track in the northern and the southern part of the pockmark. The ‘active’ section of the pockmark consisted of large Escarpia southwardae bushes, beds of modiolid mussels (Bathymodiolus sp.), and clusters of vesicomyid clams. E. southwardae bushes formed a dense cover (up to 20 m by 30 m) and were generally associated with small concretions or carbonate pavements. Bathymodiolus sp. beds covering one to several square meters (up to 50 m by 60 m) were observed in small depressions, with or without visible concretions. Vesicomyids typically occurred in clusters c. 1 m in diameter, and were often found in reduced black sediment at the edge of carbonate pavements. Microbial mats were also observed in this area. Mytilid beds and siboglinid bushes were only observed in the central section of the SW–NE axis, together with carbonate concretions. Mytilids were occasionally observed on the anterior end of siboglinid tubes, and siboglinids were seen in some mussel beds (Fig. 5D). Although vesicomyids were separated from mytilids by as little as 1 m, both families were not observed together within a cluster. The largest vesicomyid fields covering (50 × 100 m) were located in the eastern and western margins of the active area and included live individuals and empty shells (Fig. 6). These fields, located in depressions c. 100–150 m in diameter, may correspond to individual pockmarks (Ondréas et al., 2005). The most south-western vesicomyid field where VA, VB and V1 were located was visible on the micro-bathymetry map (Fig. 4). In contrast, the central area, a roughly circular feature (from site VC to H1) showed a positive relief, which was filled by methane-derived carbonate concretions surrounded by a depression. Most of the mytilid beds were observed in this area, which was also covered by a wide E. southwardae field. Methane was observed at several sites within the pockmark (Figs 2 and 4). Massive hydrates were observed on the seabed (Fig. 2A), and at other places hydrate particles floated from the seafloor. Methane gas was observed bubbling from mussel beds (Fig. 2B) and adjacent to tubeworm clusters.

Figure 6.

 Distribution map of megafauna associations dominated by mytilid or vesicomyid bivalves or by siboglinid tubeworms, along the dive tracks.

Quantitative distribution at small scale

Image mosaics were prepared to compare two areas. Mosaic 1 is located in a section of the large SW vesicomyid field. Mosaic 2 is in the central part of the site (Fig. 3). Figure 7 shows examples of individual mosaics, drawn up in both areas. A quantitative analysis of the transects covering mosaic 1 (3500 m2 = 50 m × 70 m) and mosaic 2 (8000 m2 = 200 m × 40 m) allowed comparisons of the cover and mean density of chemosynthetic metazoan species in these areas (Table 2). Aggregations with at least some live vesicomyids covered 14% of mosaic 1 (Table 2) but aggregations in which most individuals were live (at least 80%) covered only 2% of the vesicomyid field. The mean density of live individuals was 51.5 ind·m−2 (SD = 36.7) and 10.02 ind·m−2 (SD = 22.6) in living and ‘mixed’ clusters, respectively. Considering these estimates, the mean vesicomyid density was c. 2.28 ind·m−2 for the whole mosaic area (about 8000 ind). About 30% of the central pockmark area (mosaic 2) was estimated to be colonised by chemosynthetic communities, including E. southwardae bushes (23%), Bathymodiolus sp. beds (4.5%), and living vesicomyids clusters (2%). Carbonate concretions were visible on 16% of the exposed surface, but were undoubtedly more extensive, as both Escarpia and Bathymodiolus require hard substratum for settlement. Inclusion of areas occupied by Escarpia and mytilids increases the estimated cover of concretions and pavements to 46%.

Figure 7.

 Two mosaics drawn using MATISSE application, in the SW vesicomyid field (mosaic 1, a) and in the central area of REGAB (mosaic 2, b.). Video images were used to create each mosaic.

Table 2.   Cover and mean cluster sizes of symbiont-bearing species aggregates in the mosaic areas of REGAB.
MosaicCluster type Area (m2) Cover (%)Mean cluster size (SD; n) (m2)
  1. For location see Fig. 4.

1Vesicomyidae, living732.10.6 (1.1;85)
Vesicomyidae, living/dead418124.5 (3.5;40)
Vesicomyidae, shells47313.64.2 (4.9; 99)
Total living49114.11.85 (2.8;125)
Total dead47313.64.2 (4.9; 99)
2Escarpia, dense144418.25.66 (9.1;255)
Escarpia + Bathymodiolus2673.49.21 (12.5;29)
Escarpia, young430.540.93 (1.6;45)
Escarpia, sparse57.60.738.23 (5.4;7)
Escarpia, recumbent1962.55.03 (7.1;39)
Escarpia, senescent8013.5 (9.1 ;23)
Mytilidae, dense333.64.23.1 (4.8 ;108)
Mytilidae, sparse27.60.353.06 (2.3;9)
Mytilidae shells27.50.351.1 (0.4;3)
Vesicomyidae, living280.350.83 (0.8;34)
Vesicomyidae living/dead15722.57 (4.1;60)
Vesicomyidae shells1401.81.83 (1.9;75)
Total living235829.84.3 (7.6;547)
Total dead4435.653 (4.5;140)
Concretions1240166.7 (9;187)

Assemblage description, density and biomass at sampling sites

Assemblages of chemosynthetic species dominated by Bathymodiolus sp. varied greatly in size, mussel density (Table 3) and presence of siboglinids. Siboglinids were typically large on concretions surrounding M1 and small among mussels at M3. The vesicomyid assemblages differed by their size, density and the proportion of live and dead individuals (Table 3). The EA escarpiid bush included c. 30 individuals of similar (1.5 m) tube length, and EB was mostly smaller individuals 0.15–0.45 m long, which were presumed to be juveniles.

Table 3.   Density and biomass of mytilid, vesicomyid clusters and siboglinid bushes sampled in the REGAB site.
SiteCluster size (m2)Percentage of living individuals Mean density (ind·m−2) (SD; n)Biomass (kg·m−2), wet weight to dry weight
Bathymodiolus sp./ VesicomyidaeE. southwardaeBathymodiolus sp./ VesicomyidaeE. southwardae
  1. See Fig. 4 for location.

  2. *Densities from mosaic covering the whole cluster.

M1400c. 100591.8 (116; 8)250.0*7.84–1.972.31–1.09
M21c. 100958.9 (283.5; 6)012.71–3.19 
M34c. 100870.1 (196.4; 11)251.1*11.53–2.92.32–1.10
V14>80586.4 (109; 16) 7.00–1.55 
VC0.75>80607.5* 7.25–1.61 
V30.35>80484.5* 5.78–1.28 
VA0.4c. 50173.0* 2.06–0.46 
Vext0.45c. 5066.7* 0.79–0.17 
EA<0.5c. 100 557.3 (335.5; 10) 5.14–2.43
EB<0.25c. 100 250.0 (69.7; 7) 0.53–0.18

The highest densities of chemosynthetic taxa were found in mussel beds. Their densities were the highest (958 ind·m−2) at the smaller M2 assemblage without siboglinids. The lowest densities were observed at the largest M1 site (Table 3). Density was also high in some vesicomyid clusters found in the central area (VC and V3), in the external field (V1), and in adult tubeworm bushes. The highest biomass of symbiont-bearing species (about 3 kg·m2 dry weight) was found in the M2 and M3 mussel beds (Table 3). A rough estimate of the mean density and biomass of symbiont-bearing species in each mosaic area was calculated by combining the percentage cover of each assemblage type with the density/biomass estimates of individual clusters. The density of symbiont-bearing species was 2.28 and 163.2 ind·m−2 in mosaic 1 and 2, respectively. Contrast was even more evident in biomass, with 0.8 g·m−2 in the vesicomyid field and 460 g·m−2 in the central area.

Methane, sulphide and oxygen concentration in water

Methane levels varied both within and among assemblages, but were generally the highest in mussel beds, and much lower within vesicomyid or escarpiid aggregations. Vertical gradients in methane were measured in all assemblages, as a decrease in methane from the bottom to the top of individuals (Table 4). Horizontal gradients were also observed, with methane decreasing from the centre to the margin of vesicomyid clusters, the M3 mussel bed, and young escarpiid bush EB (Table 4). Methane concentrations were higher above living vesicomyid clusters in the centre of the pockmark (V3 and VC) or within young escarpiid bush EB than in external areas (VA, VB and V1). Hydrogen sulphide concentrations in seawater were very low or undetectable. No significant anomalies or variations in carbon dioxide and pH were detected. Oxygen levels close to mytilid clusters were homogeneous, but more variable in the vicinity of vesicomyids and escarpiids inside the sediment.

Table 4.   Chemical characteristics of the water in bivalve clusters and siboglinid bushes sampled in REGAB.
Sampling siteCH4 (μm) bottom/top CO2 MM H2S μm O2 μmpH POC (%)Substratum
  1. Values between brackets were taken a few decimetres outside the clusters.

  2. Bottom samples were taken between the bivalves or at the base of the tubeworms the closest to the sediment, top samples at the anterior edge of the shells or at the level of the siboglinid plumes.

  3. Co., concretion; Co./sed., concretions surrounded by soft sediment; S. sed., soft sediment.

M1 centre1.6–0.7n. e.<0.12407.75.755Co.
M2 centre33.7–23.2n. e.<0.12307.73.09Co./sed.
M3 centre23.5–12.5n. e.<0.12307.83.665Co./sed.
M3 middle9.4–8.7n. e.<0.12337.8  
M3 periphery11.8–6.3n. e.<0.12307.8  
VA0.99–0.46 (0.3)2.2n. e.211–2387.83.0S. sed.
VB (dead)0.42–0.222.1n. e.175–2337.8 S. sed.
VC3.9–1.62.3n. e.2207.7 S. sed.
V10.9–0.4n. e.<0.12217.82.26S. sed.
V34.4–3.4n. e.n. e.2367.82.23S. sed.
EA (adult)1.11–0.632.4n. e.218–2317.8 Co.
EB (young)4.45–2.43 (0.41)2.2n. e.218–2327.8 Co.

Oxygen, organic carbon and hydrocarbon in the sediment

The oxygenated zone in surficial sediments was very thin (1.5–3.5 mm) in areas of fluid seepage. This was consistent with the high organic carbon content of surface sediments, which varied (from 2% to 6%) close to aggregations of symbiont-bearing species, to a maximum of 5.7% within mytilid beds (Table 4). The total hydrocarbon concentration in the upper 4 cm of sediment was very low at vesicomyid sites and was more variable at mytilid sites. Except at M1 (89 ppm) and M2 (720 ppm), these measurements did not indicate unusual hydrocarbon contents (<70 ppm). High levels of perylene (93–97%) identified during the analysis of PAHs by CG/MS and HPLC suggest that the hydrocarbon content is related to hopans of bacterial origin.

Fauna-habitat relationships at the sampling sites

Multivariate analyses indicated that methane and oxygen were the main factors explaining the distribution of each of the three symbiotic taxa among the 10 sampling sites (Fig. 8). The first two PCA axes explained 37% of the variance observed (23% for the first axis and 14% for the second) in faunal distributions among the environmental variables included. Canonical redundancy analyses (RDA) run with the densities of the three symbiont-bearing species at each sampling site or environmental factors for the same sites suggested that methane and oxygen could control the distribution of mytilids, whereas substratum (concretions or soft sediment) seemed to influence the distribution of siboglinids and vesicomyids (Fig. 8B). The first two axes explained 42% of the variance found in fauna densities. Depending on the procedure used, it was possible to indicate methane concentrations as the only variable explaining significantly the densities of symbiont-bearing species (P = 0.22).

Figure 8.

 Ordination graphs showing the first two axes of PCA (a) and RDA (b). Arrows indicate direction of increase in environmental parameters.

Isotopic measurements in symbiont-bearing species

Bathymodiolus sp. tissues were more depleted in δ13C than other symbiont-bearing species studied (Table 5). Mussels from the M2 site had significantly lighter δ13C values, particularly the gill tissues (−67.05‰), than for mussels from M1 and M3 (Kruskal Wallis test, P = 0.001). Isotopic ratios for carbon also varied slightly among mytilid tissues (P = 0.11). δ13C in mytilid gills was significantly heavier than mantle tissues only at M2 (P = 0.001). Isotopic ratios for vesicomyid tissues from V1 and V3 did not differ significantly and no distinction could be made between gills and mantle. Adult escarpiid tissues differed significantly among sites with the least depleted δ13C at M1 where mytilids dominated, and the most enriched δ13C at EA. Juveniles sampled at EB also differed from adults. No differences between tissues, trophosomes and vestimenta, were detected.

Table 5.   Isotopic values of symbiont-bearing species collected at different sampling site on the REGAB pockmark.
Species/sitesδ13C (‰) mantleδ13C (‰) gillδ15N(‰) mantleδ15N (‰) gillδ13C(‰) mean*δ15N(‰) mean*
  1. Number of measured individuals indicated in brackets.

  2. *For Escarpia southwardae, mean δ13C and δ15N calculated from the three tissues analysed (trophosome, vestimentum and obturaculum) are presented because no significant differences were detected among tissue types.

Bathymodiolus sp.
M1−63.3 ± 3.5 (6)−63.7 ± 2.1−0.5 ± 0.8 (6)−0.6 ± 0.3  
M2−64.5 ± 0.8 (5)−67.0 ± 0.80.3 ± 0.5 (3)−0.7 ± 0.4  
M3−63.7 ± 1.5 (6)−62.4 ± 0.9−1.9 ± 0.9 (12)−3.1 ± 1.0  
Vesicomyidae sp.
V1−35.6 ± 1.6 (5)−35.7 ± 1.12.9 ± 2.0 (3)−1.0 ± 2.1  
V3−36.0 ± 0.8 (4)−36.0 ± 0.153.5 ± 0.3 (5)2.9 ± 2.0  
E. southwardae
M1    −24.2 ± 5.5 (5)2.5 ± 1.6 (6)
EA (adult)    −36.22 ± 3.3 (4)2.9 ± 0.6 (5)
EB (young)    −29.1 ± 3.4 (3)


The REGAB site is the first cold-seep site described in the South East Atlantic. It appears as a giant pockmark formed by the association of several individual pockmarks, resulting in a large area covered by a chemosynthetic community. The live chemosynthetic assemblage is located in an area of at least 80,000 m−2 along a 70° N axis, which probably corresponds to the main buried channel providing fluids to the pockmark (Ondréas et al., 2005). Other small (c. 1 m diameter) secondary pockmarks can be seen throughout the structure, filled mainly with vesicomyid shell material. The spatial distribution of chemosynthetic assemblages obtained from GIS mapping at the REGAB site and mosaic analyses indicates clear differences in the structure (dominant species) and density of the assemblages. The central area of the pockmark is characterised by high densities of live chemosynthetic assemblages, covering 34% of the mosaic area analysed. This area is dominated by E. southwardae bushes, with small depressions containing dense mytilid beds and occasional dense aggregations of vesicomyid clams. The south-western and north-eastern margins of the pockmark were dominated by large fields of vesicomyid bivalves. Although vesicomyid shell densities were high in these fields, clusters dominated by live clams were sparse. This spatial pattern was first related to the methane concentrations in seawater and to the substratum heterogeneity observed at the REGAB site (Charlou et al., 2004, Ondréas et al., 2005). In fact, methane levels measured 3–4 m above the sea-floor were highly variable (5.8 μmversus 1.1 μm) with the highest concentrations in the central area.

Chemical measurements in this study were made close to chemosynthetic species and also indicated that methane is an important factor in the distribution of the symbiont-bearing species. The highest methane concentrations were recorded in mytilid-dominated assemblages (Table 4). Methane concentration decreased with density in mussel beds from M2 (highest density and methane) to M3 and to the least dense M1 site. The low methane concentrations at M1 could be explained by temporal variability in methane emissions. In fact, turbulent flows of solid methane hydrate particles have been suggested to explain the high heterogeneity of methane concentration in the water column above REGAB (Charlou et al., 2004). M1 also differed from the other sites by the larger the mussel bed, lower mussel density, presence of large recumbent Escarpia and presence of actinians. We hypothesise that the assemblage at M1 is older than M2 and M3, and now has a decreasing methane flux. Differences in biotope chemistry among several cold-seep mussel beds has also been reported for B. childressi in the Gulf of Mexico (Bergquist et al., 2005). Moreover dual-symbiosis mussels may thrive in a wider range of fluid chemistry than solely methanotrophic mussels, owing to their sulphide-oxidising symbionts (Olu et al., 1996; Levin, 2005). The isotopic carbon values of mussel tissues (about −62‰), suggest that methanotrophy dominates the nutrition of this species, despite dual symbiosis with both methanotrophic and sulphide-oxidising symbionts reported by Duperron et al. (2005). Assuming negligible isotopic fractionation during methanotrophy (Conway et al., 1994), isotopic values were within the range of those of methanotrophic mussels (−40 to −70‰) (Kennicutt et al. 1992, Conway et al., 1994; MacAvoy et al., 2005) and are consistent with the methane δ13C of −67‰ to −70‰ recorded in seawater from REGAB (D. Levaché, personal communication). The most depleted δ13C values, suggesting a higher contribution of methanotrophic symbionts, were from the M2 site mytilids where the highest methane concentrations were recorded. However, no variation in the relative abundance of sulphide- or methane-oxidizers in Bathymodiolus sp. gills was observed among the different sites at REGAB (Duperron et al., 2005), as has been suggested for the vent mussel Bathymodiolus azoricus (Colaço et al. 2002; Trask & Van Dover, 1999) under variable sulphide and methane levels.

Methane levels above assemblages are a consequence of the methane flux and of the efficiency of anaerobic methane oxidation (AMO) by the microbial consortia (Boetius et al., 2000). The total density of the microbial consortia (ANME/SRB) integrated over 0–15 cm within the sediment column was twice as high at the mussel site M3 (18 × 106 aggregates per ml of sediment) than at vesicomyid site V1 (9.9 × 106) or 10 times that at a microbial mat (1.6 × 106) (Nadalig et al., 2002). Despite microbial consortia activity and methane uptake by mytilid symbionts, methane concentration available for the fauna at the sediment–water interface was higher above mytilids than above vesicomyids or siboglinids. Fluid flow may therefore be higher in mytilid beds. The highest sediment carbon content was also measured in mussel beds, indicating high biological production in sediments.

Methane levels near chemosynthetic taxa were the unique variable among those tested which explained the densities of symbiont-bearing species at the sampling sites in the RDA analysis, taking into account 42% of the variance. Previous studies in western Atlantic seeps suggested that the distribution of methane-dependent mussels or sponges versus sulphide-dependent species may also be controlled by flux intensities and methane concentrations (Aharon, 1994; Olu et al., 1996, 1997). A recent study of chemosynthetic communities on the Costa Rica margin suggested that mytilids were associated with higher fluid flow rates than vesicomyids (Mau et al., 2006). MacDonald et al. (2003) indicated the occurrence of mytilids and hydrate outcrops in small areas affected by a rapid gas flux while tubeworm distribution is much more scattered. In contrast, long fluid flow duration or relict seeps may favour vesicomyids or siboglinids (Sassen et al., 1994).

Although vesicomyids were found in areas with low methane concentrations, they were not entirely independent of methane fluxes. Methane levels decreased from the centre to the outside of vesicomyid clusters, and generally increased with vesicomyid density (Tables 3 and 4).

Methane concentrations measured at the adult Escarpia bush (site EA) were low, but methane levels may be variable in both time and space. Methane bubbles were observed on video in some bushes. Higher methane levels were measured in young colonies than in adult ones but carbonate formation may slow methane flux through the sediment. Such a decrease of reduced compound levels with age or size of siboglinids has been suggested by Julian et al. (1999) for interstitial sulphide concentrations beneath aggregations. Bergquist et al. (2003) also reported that, despite high variability within individual aggregations, sulphide is commonly present in the water surrounding the tubes of younger aggregations but is rare around the tubes of older aggregations.

Sulphide as a second energy source is also likely to play a great role in the control of sulphide-dependent symbiont-bearing species taking sulphide up from sediment pore-water, as shown for vesicomyids at other sites (Sahling et al., 2002; Levin et al., 2003). Siboglinid distributions must also be influenced by sulphide concentrations in the sediment owing to their sulphide-oxidising symbionts. These obligate symbiont-bearing species may live for decades without active methane fluxes, thanks to sulphide accumulation in the sediment (Bergquist et al., 2002). On the Haakon Mosby Mud Volcano, anaerobic oxidation of methane and sulphide production occurs only in older mud flows, and not in recent ones affected by the highest methane fluxes that preclude sulphate penetration in the upper sediment column (de Beer et al., 2006; Niemann et al., 2006). High methane fluxes may thus not favour settlement of vesicomyids or siboglinids.

As shown by the distribution of the symbiotic taxa at the REGAB site and results of the rda analysis, substratum is also likely to play a role in the structure of these communities. Vesicomyids are excluded from areas dominated by carbonate concretions that are favourable for siboglinids and mytilids, which require a hard substratum for larval settlement. However, the range of substratum inhabited by the mytilid Bathymodiolus sp. was wider than for E. southwardae. A better knowledge of the distribution of small size concretions and pavements under the sediment surface should help assess the relative role of substratum, compared with methane fluxes, in the structuring of the REGAB community.


The 800 m-diameter giant pockmark REGAB is characterised by high spatial variability in the structure of chemosynthetic communities, with different assemblages dominated by three key symbiotic taxa (Mytilidae, Vesicomyidae and Siboglinidae). Although mytilids were only observed in the centre of the pockmark, this area is dominated by invertebrates with sulphide-oxidising symbionts, and thus, does not support the preliminary hypotheses given by the Gay et al. (2006) model. Spatial variation in faunal structure is linked to the wide range in environmental conditions at the pockmark scale and also highlights small-scale habitat heterogeneity. The substratum is variable, ranging from areas of soft sediment to large carbonate pavements. Methane fluxes, characterised by methane concentrations a few meters above the seafloor (Charlou et al., 2004) and close to the fauna (this paper) are also highly variable. Methane levels at the sediment-water interface appear as one of the key factors that regulate the distribution of each of the three dominate taxa comprising the chemosynthetic assemblages at the REGAB site.

The distribution of living fauna may also underline the variable activity of several individual pockmarks. The REGAB site appears to be a ‘pockmark’ cluster, composed of pockmarks at different stages of activity and successional status as observed for mud volcanoes (Neurauter & Roberts, 1994). The spatial variability of the fauna may also reflect temporal variability in the environmental conditions, as well as biotic interactions, e.g. competition between dominant symbiotic species. As suggested by Levin (2005), further coordinated studies at seeps with in situ multidisciplinary measurements over extended periods are required to understand the biological response to the high spatial and temporal variability in fluid flow and attendant microbial activities.

The REGAB chemosynthetic community is also inhabited by non symbiotic mega-and macrofauna, particularly Alvinocaridae shrimps (Komai & Segonzac, 2005), Synaptidae holothurians and several gastropods. These seep-associated megafaunal taxa are relatively constant between assemblage types with dominance of shrimps and gastropods (Olu-Le Roy et al., 2005). However, some differences appeared in isotopic signatures of several species that could be related to differences in habitat or assemblage type.


Many thanks to the crew of the N.O. L'Atalante and VICTOR 6000 team. ZAIROV cruise (chief scientist H. Ondréas) was the last cruise of the ZAIANGO program supported by the Total oil company and Ifremer (P.I. Bruno Savoye, Ifremer). Biozaire 1 and 2 cruises were partially funded by the Total company (Biozaire program, P.I. M. Sibuet, Ifremer). Rudo von Cosel's expertise on bivalves was essential. We are indebted to D. Levaché, Total for methane isotopic measurements, Philippe Crassous for sediment carbon content measurements. We are indebted to the two anonymous reviewers and to Lisa Levin and Jim Barry, editors, for their helpful comments on the manuscript.