Faunal and carbonate distribution
The mosaic and the distribution map of the megafaunal communities give a very detailed view and full coverage of the entire study area. The results show that the megafauna at Regab is concentrated mainly in three distinct areas of high faunal presence, separated by areas of relatively lower presence. Such distribution indicates that the chemical fluxes that are required to sustain these chemosynthetic communities are heterogeneous over the study area. Indeed, the distribution of the main faunal assemblages showed a concentric spatial zonation pattern, starting from mussel beds in the middle to tubeworms and finally fields of vesicomyids towards the outside. In our study this spatial zonation pattern from mussels to vesicomyid clams was observed, to various extents, around the three main mussel areas (M1, M2/M3, M4).
A model presenting a concentric pattern has been proposed previously for the Regab pockmark (Gay et al. 2006) but it considered both mussels and tubeworms as methane-dependent species inhabiting the same carbonate-dominated facies (Olu-Le Roy 2006). Although Bathymodiolus aff. boomerang contains both methanotrophic and thiotrophic symbionts (Duperron et al. 2005) and, for the populations living in the Regab pockmark, is known to rely on methane as dominant energy source (Olu et al. 2009; Duperron et al. 2011), tubeworms are known to host sulfur-oxidizing symbionts (Dubilier et al. 2008) and to have very high demands in terms of sulfide supply (Cordes et al. 2003).
We postulate that the observed distribution is controlled by the strength of fluid advection and related methane fluxes, and propose a model in which the megafaunal distribution at Regab is structured by the presence of discrete zones of intense fluid advection and methane fluxes under the mussel beds (Fig. 9). The distribution of the other communities would therefore be related to decreasing advection rates with distance from the mussel beds. The existence of such localized pathways of high fluid advection rate is compatible with the current understanding that the center of the pockmark is composed of several smaller pockmarks (Ondréas et al. 2005). In this section we discuss the concepts of this model, and compare them with more detailed observations of the faunal distribution and of the presence/absence of carbonate crusts.
Figure 9. Summary schematic model (not to scale). The main aggregations are distributed in concentric patterns with the mussels in the middle, then the tubeworms on thick concretions, and finally the vesicomyid clams in the surrounding sediments. Mussels are present in an area of intense flux with significant release of methane to the water column. A transition zone is observed where mussels are present at the bottom and on the tubes of the vestimentiferans. Vestimentiferans are present on carbonate concretions but reach the sediments with their roots. Through sulfate release, they maintain the AOM and the sulfide production. Juvenile tubeworms are distributed near the limit of the crusts where the sulfide fluxes from the sediments are likely higher. The presence of dark, reduced sediments around the concretions indicate that part of the methane and sulfide fluxes are redirected from under the crusts towards more sulfate-rich zones where AOM occurs. Populations of vesicomyid clams occur in the surrounding sediments. Their patchy distribution suggests that this is controlled by discrete and transient fluxes from below.
Download figure to PowerPoint
Our observations show that the main mussel beds occur in areas where carbonates form blocs of indurated sediments and concretions within slight depressions, and that the main tubeworm aggregations occur in areas with extensive, continuous and prominent carbonate crusts. The formation of authigenic carbonates is a byproduct of the anaerobic oxidation of methane (AOM) in the sediment (Boetius et al. 2000; Aloisi et al. 2002) and is an indicator of methane fluxes and microbial activity within the sediments. However, the formation of continuous carbonate crusts impacts the porosity and permeability of the sediments and hence reduces the possible pathways for methane- and sulfide-rich fluid escapes and for sulfate-rich seawater infiltration (Hovland 2002; Luff et al. 2004). Therefore, areas of mussel occurrence are likely to be characterized by higher fluxes between the sediments and the bottom water than the encrusted areas where tubeworms occur.
We propose that the three main mussel areas present in our study area are located on focused zones where the seepage activity is the strongest, and where fluid flow is intense enough for the methane fluxes to reach the sediment/water interface. Indeed, the distribution of mussels mainly in dense circular beds suggests the presence of localized areas of intense methane fluxes. This hypothesis was mentioned previously from the results of the BIOZAIRE mosaic (Ondréas et al. 2005; Olu-Le Roy et al. 2007a). It also is strongly supported by recent biogeochemical analyses in various Regab habitats, according to which extensive seepage of gaseous and dissolved methane was observed only under the mussel habitat (Pop Ristova et al. 2012). This is also in accordance with previous studies in other areas that showed that release of methane to the water column is indeed facilitated in areas of high flux (Boetius & Suess 2004; de Beer et al. 2006; Niemann et al. 2006).
Co-occurrence of mussels and tubeworms is commonly observed at the transition between the two populations. In such zones, numerous mussels are observed on the tubes of the vestimentiferan aggregations that directly border the mussel beds. This could be the result of space limitations within the mussel beds, which would constrain mussels to invade neighboring tubeworm aggregations. Indeed, the dual symbiosis of Bathymodiolus aff. boomerang allows this species to use both methane and sulfide, similarly to several other seep and vent mussels of the Bathymodiolus genus (Duperron et al. 2005). However, unlike in the gills of Bathymodiolus azoricus at Mid-Atlantic hydrothermal vents, symbionts in seep mussel gills at Regab and at Gulf of Mexico seeps are predominantly methanotrophic; the thiotrophic symbionts are likely limited by the absence or low level of sulfide in the seawater (Duperron et al. 2011). Moreover, the fact that mussel/tubeworm co-occurrence zones are mostly close to the main mussel beds may also support the hypothesis that methane fluxes are higher in those areas than in the tubeworm aggregations located farther from the mussel beds. This is in accordance with previous studies that indicate that mussel beds at Regab are located in areas with the highest concentrations of methane in the water (Charlou et al. 2004; Olu-Le Roy et al. 2007a). Such behavior was also observed for methanotrophic mussel populations in the Southern Barbados prism (Olu et al. 1996b) and along the Costa Rica margin (Mau et al. 2006). In the Gulf of Mexico, mussels associated with brine seeps with high methane concentration grow faster and are in better physiological condition than those from petroleum sites with low methane but high sulfide concentrations (Bergquist et al. 2004).
The distribution of the tubeworm aggregations around the mussel beds could reflect lower fluid advection rates than under mussel beds. Niemann et al. (2006) and de Beer et al. (2006) suggested that by preventing downward fluxes of sulfate-rich water into the sediments, intense fluid advection rates can hinder the efficiency of AOM. This is in accordance with results from Pop Ristova et al. (2012), who calculated the proportion of upward diffusing methane that is removed by AOM at Regab to be only 6–20% under a mussel habitat compared with 47–97% in lower flux areas such as under clam habitats. Such processes could result in different environmental conditions inside and outside the mussel beds, and partly control the relative distribution of tubeworms and mussels. Biotic interactions are known to occur between mussels and tubeworms at hydrothermal vents (Johnson et al. 1994; Desbruyères 1998; Lenihan et al. 2008). For instance, mussels of the species Bathymodiolus thermophilus are able to disperse fluids laterally (Johnson et al. 1994), which could make fluids unavailable for tubeworms, and to inhibit recruitment of other vent species (Lenihan et al. 2008). Such interactions could also apply in dense B. aff. boomerang aggregations. However, unlike vent tubeworms, which take up sulfide from the seawater (Arp et al. 1985), seep tubeworms such as Escarpia southwardae take up sulfide directly from the sediments through their roots (Julian et al. 1999; Andersen et al. 2004) and are unlikely to be affected by fluid dispersion by the mussels. Furthermore, sulfide is almost absent in the seawater overlying mussel beds at Regab (Olu-Le Roy et al. 2007a; Duperron et al. 2011). Therefore, the absence of tubeworms in mussel beds is more likely due to competition for space or reduced AOM efficiency rather than to biotic interactions.
Our observations show that large tubeworm aggregations around the main mussel beds correspond to areas where the carbonate crusts are most prominent, and likely the thickest. Carbonate precipitation likely reduced methane flux, thus enhancing AOM and sulfide production favoring tubeworm settlement and dominance over mussels. Moreover, as observed at Gulf of Mexico seeps, tubeworms of the species Lamellibrachia luymesi were shown to release sulfate through their roots into the sediments (Cordes et al. 2005a), preventing a potential sulfate-depletion of the sediments. This characteristic is believed to allow adult L. luymesi to fuel or even enhance the AOM (Cordes et al. 2005a; Dattagupta et al. 2008) to maintain their supply of sulfide. As a result, the vestimentiferan population contributes to the formation of carbonates, which is supported by our observations that tubeworms are present where concretions form continuous and prominent crusts.
Juvenile tubeworms consistently occur near or at the limit between carbonate crusts and bare sediments. Sulfide fluxes and concentrations are likely to be higher in such areas with unsealed sediment/water interface than in areas covered by thick crusts, and thus provide a suitable environment for the larvae to settle until they can self-maintain their supply of sulfide. According to Bergquist et al. (2002), the recruitment of new tubeworms of the species L. luymesi and Seepiophila jonesi is time-constrained and ceases in older aggregations, due to the presence of thick carbonate pavements and low concentrations of sulfide in the water (Bergquist et al. 2003a). Sulfide concentrations have indeed been reported to be higher around aggregations of juvenile rather than within older aggregations (Bergquist et al. 2003b).
This interpretation is further supported by the occurrence of bands of black reduced sediments along the limits of the vestimentiferan-hosting carbonate concretions (Fig. 5), indicating that AOM and sulfide release occur in those areas. These features also suggest that part of the methane fluxes trapped beneath the carbonates could be redirected toward the sides of the carbonate crusts.
The distribution of the vesicomyid population is likely related to even lower advection rates than the tubeworm aggregations. Vesicomyid clams require soft substrata to access sulfide through their foot and, thus, are generally excluded from encrusted areas. Results showed that the living vesicomyid clams are indeed mainly located in the soft sediment areas surrounding the tubeworms and mussel aggregations. However, according to our interpretation, those areas are also where methane and sulfide fluxes are lower. In particular, Olu-Le Roy et al. (2007a) observed that vesicomyid clams at Regab are located in areas with relatively low methane concentrations, in comparison with tubeworm and mussel habitats. This is consistent with the current understanding that methane/sulfide availability shapes the structure of the microbial and megafaunal communities (Olu et al. 1996a,b, 1997; Sahling et al. 2002; Sibuet & Olu-Le Roy 2002; Levin et al. 2003; Levin 2005; Ritt et al. 2011; Pop Ristova et al. 2012).
The vesicomyid clam environment is sometimes proposed as being a precursor to a tubeworm/carbonate environment (Sahling et al. 2008). This is partly supported by our observations of dead shell on some bare carbonate concretions. The presence of dead shells within the concretions has been reported previously (Pierre & Fouquet 2007). These observations suggest that areas with vesicomyids transfer to carbonated crust formations, either with or without vestimentiferans. However, we have also observed large patches of dead vesicomyid shells within the sediments, sometimes almost buried, which would suggest that seepage activity in these areas decreased or stopped and that living populations either died or moved away. Such areas might never turn into tubeworm/carbonate environments due to too low or too transient fluxes. Thus, vesicomyid populations might not be restricted to one particular successional stage of colonization, but be present in a range of areas representing different development stages of the seeping activity, and the observed patterns of distribution may reflect the spatial heterogeneity of fluid flux.
Bergquist et al. (2003b) and Cordes et al. (2005b) suggested that the relative distribution of mussels (Bathymodiolus childressi) and tubeworms (L. luymesi and S. jonesi) could be related to different stages of succession, and explain that mussel beds indicate an earlier stage of colonization that would later be replaced by tubeworms when the formation of carbonate concretions reduces the methane supply to the water column.
However, we propose that the situation at Regab is more complex, as the observed patterns of faunal distribution could also be related partly to spatial heterogeneity of the fluid advection regime. A particular feature of this site is the coalescence of several pockmarks within the Regab site, which may differ in their fluid flow regimes or their evolution stages (Ondréas et al. 2005; Gay et al. 2006). If the relative mussel/tubeworm distribution were solely related to different colonization stages, the mussel population would be expected to be observed mainly together with, or in the vicinity of, juvenile tubeworm aggregations. Although we observed juveniles around some small mussel clusters, the larger mussel beds present in the study area were predominantly bordered by large adult tubeworm aggregations. Considering the extremely slow growth rate of tubeworms (Fisher et al. 1997; Bergquist et al. 2000), this indicates that fluid advection in those areas has been going on for a relatively long time, but that recruitment of juvenile has not occurred or has been hindered.
The comparison of the maps of faunal distribution and of the computed areas reveals that the size of the areas of faunal occurrence has remained globally the same between the BIOZAIRE (2001) and the WACS (2011) cruises. We consider the discrepancies in computed values and mapped areas to be caused largely by uncertainties in the method. First, images for each mosaic have been taken with different camera and lighting setups, which results in different resolutions and visibilities between the BIOZAIRE and WACS mosaics. Furthermore, small perspective distortions can, in places, impact the precision of the relative geo-referencing of the mosaics, or make a feature look larger on one mosaic than on the other. Finally, the delineation process is a manual step that strongly depends on the interpretation and precision of the observer. For all these reasons, mapped features may look different and discrepancies in the computed areas may arise that are difficult to evaluate. Nevertheless, considering all these possible sources of uncertainty and the large size of the study area, the computed areas are remarkably consistent between the two mosaics. Indeed, quantitative results showed that the changes in coverage per category are lower than 2% of the overlap area, which suggest very little change between the two mosaics. However, based on qualitative observations described below, we consider that the calculated areas are impacted by errors of the method and cannot be used to analyze further such small variations in the areas of faunal cover.
For instance, our observations of the mosaics confirm that no change occurred in the population of tubeworms within the overlap area over the past 10 years. Indeed, an increase of the area covered by tubeworms would signify that recruitment occurred. However, we did not observe new juvenile aggregations, possibly because the observation period was too short and that juveniles are still too small to be seen, or because the recruitment in the area covered by the mosaics was somehow limited. Overall, the absence of changes in the vestimentiferan population is in accordance with the findings that some tubeworms (Lamellibrachia luymesi and Seepiophila jonesi) may be very slow-growing and long-living (Fisher et al. 1997; Bergquist et al. 2000). Nevertheless, Lessard-Pilon et al. (2010) observed evidence of tubeworm recruitment (Lamellibrachia spp. or Escarpia laminata) within discrete seep communities in the Gulf of Mexico over a 15-year period, confirming that changes within tubeworm communities could be observed on such time-scales. Their work was based on a comparatively small study area (23.4 m2) and observations were likely more detailed than in our study (4605 m2); indeed, they used photo datasets, which provided higher definition images and closer views of the sea floor. Hence, tubeworm recruitment may have occurred within our study area but it was not evident from our data.
Conversely, some changes in size of individual beds of living and dead mussels were observed which suggests that localized variations of methane fluxes or carbonate precipitation may have occurred. For instance, small-scale visual observations suggest that the dense mussel bed in M2/M3 might contain, in the overlap area, a greater abundance of mussels in 2011 than in 2001, and that some minor mussel beds at M1 might have disappeared. This could indicate that the intensity of fluxes increased in M2/M3 and decreased in M1. A decrease in activity in M1 would be consistent with the findings of Olu-Le Roy et al. (2007a), who also hypothesized that a decreasing methane flux occurs in this area, based on the lower density of the mussel beds. Alternatively, the mussel abundance might have been stable and the observed changes could reflect a rearrangement of the mussels. In either case, distribution changes are likely the result of local variations in environmental conditions.
However, the observed changes were localized and, overall, the mussel population showed very little variation. Despite the scale difference between the two studies, these findings are consistent with observations of Lessard-Pilon et al. (2010). Both studies agree that only little change was observed on the total area covered by foundation fauna. However, Lessard-Pilon et al. (2010) observed significant small-scale changes in the distribution of mussel populations (Bathymodiolus brooksi and Bathymodiolus heckerae), and reported that about 50% of the area originally covered by living mussels at one site had been, after a period of 15 years, either replaced by dead mussel shells or colonized by tubeworms (Lamellibrachia spp. or Escarpia laminata). Interestingly, as we observed at Regab, the hydrothermal vent mussel population of Bathymodiolus azoricus was described to be stable on a decadal scale in terms of overall percentage of colonization (Cuvelier et al. 2011), whereas small fluctuations occurred on shorter time scales and on smaller spatial scales for seep mussels of the Gulf of Mexico (Lessard-Pilon et al. 2010).
In our study, no change was observed in the tubeworm aggregations surrounding these areas where mussel distribution varied. Mussel populations are expected to be more dynamic and to respond faster to environmental changes than tubeworm populations (Lessard-Pilon et al. 2010), which may be insensitive to small variations of seepage activity. Alternatively, the increased occurrence of dead mussels in M1 could reflect a late stage of the successional model developed by Bergquist et al. (2003b), characterized by a decrease of mussel population due to a decrease of methane and maybe of sulfide in the water column. Indeed, mussels are associated with areas of vigorous seepage and high methane concentration (Nix et al. 1995; Bergquist et al. 2005). However, the model proposed by Bergquist et al. (2003b) is true for long time-scales and it is uncertain whether such a trend is detectable within a 10-year period.
Changes in the population of vesicomyid clams were more frequently observed than for the mussel population. Although we cannot conclude if the total living population did change globally, the location of the aggregates of living individuals shows relatively more differences in comparison to the other populations studied. Indeed, several patches of living individuals observed in the 2011 mosaic did not exist in 2001. Conversely some patches of vesicomyid clams that existed in 2001 did not exist anymore in 2011. Also, in some cases, old patches of dead clams were re-colonized by living clams. However, changes in the distribution of patches of living/dead clams are difficult to detect since the relative proportion of living and dead clams cannot be estimated from the images. Overall, the distribution of living vesicomyid clams is very patchy and is difficult to understand.
Sahling et al. (2008) proposed a model for pockmarks of the Congo fan (100 km north of Regab), in which the distribution of tubeworm (E. southwardae) and clam (of similar species than at Regab) assemblages is controlled by the depth of the gas hydrate deposits. The model considers that gas hydrates deposits act as ‘capacitors’ (Dickens 2003) that buffer the transient influxes of methane from below and that ensure a more stable diffusion of methane into the porewater above, thus sustaining long-living seep communities. We think that a similar control mechanism occurs at Regab. Indeed, the presence of gas hydrates at Regab is known both from direct observation of outcrops on the sediment surface and from sediment cores (Charlou et al. 2004; Ondréas et al. 2005).
However, although this model is supported by the presence of such large populations of long-living seep communities and by their spatial patterns of distribution, it does not fully explain the temporal changes observed within the clam populations. At Regab, most aggregates of living individuals are indeed located within patches of black sediments, indicating the occurrence of AOM; clams are mobile fauna, hence changes in areas of distribution must somehow reflect changes of sulfide availability. One possible explanation could be that gas hydrate deposits under clam communities are either absent or too thin to buffer the transient methane fluxes over such a time period. The observed changes in clam distribution would therefore be the response to the transient release of methane and subsequent transient sulfide production. Clearly, some monitoring of sulfide concentrations and some geological sampling under the clam aggregates would be required to further refine this interpretation. However, this is in accordance with other studies that suggest that vesicomyid clams are supported by diffuse or transient fluxes (Olu et al. 1996a,b).