Methane cold seeps as biological oases in the high‐Arctic deep sea

Cold seeps can support unique faunal communities via chemosynthetic interactions fueled by seabed emissions of hydrocarbons. Additionally, cold seeps can enhance habitat complexity at the deep seafloor through the accretion of methane derived authigenic carbonates (MDAC). We examined infaunal and megafaunal community structure at high‐Arctic cold seeps through analyses of benthic samples and seafloor photographs from pockmarks exhibiting highly elevated methane concentrations in sediments and the water column at Vestnesa Ridge (VR), Svalbard (79° N). Infaunal biomass and abundance were five times higher, species richness was 2.5 times higher and diversity was 1.5 times higher at methane‐rich Vestnesa compared to a nearby control region. Seabed photos reveal different faunal associations inside, at the edge, and outside Vestnesa pockmarks. Brittle stars were the most common megafauna occurring on the soft bottom plains outside pockmarks. Microbial mats, chemosymbiotic siboglinid worms, and carbonate outcrops were prominent features inside the pockmarks, and high trophic‐level predators aggregated around these features. Our faunal data, visual observations, and measurements of sediment characteristics indicate that methane is a key environmental driver of the biological system at VR. We suggest that chemoautotrophic production enhances infaunal diversity, abundance, and biomass at the seep while MDAC create a heterogeneous deep‐sea habitat leading to aggregation of heterotrophic, conventional megafauna. Through this combination of rich infaunal and megafaunal associations, the cold seeps of VR are benthic oases compared to the surrounding high‐Arctic deep sea.


Introduction
Marine environments in the high-Arctic are characterized by intense seasonality, sub-zero bottom water temperatures, and extended periods of overlying sea ice. These polar features set a framework for regulation of communities and ecosystems, with intense, episodic pulses of fresh organic matter interspersed among long periods of food limitation Wassmann and Reigstad 2011;Boetius et al. 2013;Meyer et al. 2013). The deep sea (bathypelagic zone > 1000 m depth) also experiences a framework of food limitation, where photosynthetically derived organic matter (e.g., particulate organic matter, marine snow, fecal pellets) from the euphotic zone has been extensively processed during its transit through the water column, arriving at the seafloor highly degraded (Southward and Southward 1982;Graf 1989;Gage and Tyler 1991). These processes lead to what has been termed a biological desert with respect to macrofaunal and megafaunal communities, with low abundance and biomass, but with sometimes high species diversity (Sanders and Hessler 1969;Rex 1981). Instead, deep-sea benthic communities are usually dominated by meiofauna (< 0.5 mm) (Clough et al. 1997;Vanreusel et al. 2000;Hoste et al. 2007), with microbial activity playing a substantial role in carbon processing and remineralization (Wheeler et al. 1996;Boetius et al. 2013).
The composition of deep-sea benthos is also regulated by factors including sediment characteristics and heterogeneity at the seabed (Etter and Grassle 1992;Carney 2005). The expansive, relatively featureless, soft-bottom plains prevalent in deep-sea environments favor suspension and detrital feeders while at the same time largely excludes epifauna that require hard substrate (Levin et al. 2001;Carney 2005). Habitats with mixed substrates, with a high heterogeneity containing a mosaic of soft and hard bottom, generally support the highest diversity and biomass of benthic communities (Gage and Tyler 1991;Qu eric and Soltwedel 2007;Buhl-Mortensen et al. 2012).
Cold seeps are environments where emissions of methane, sulfide, or other reduced chemicals occur at the seafloor and these are known from both deep sea and shallow ocean shelves worldwide (Vanreusel et al. 2009;Olu et al. 2010;Levin et al. 2016). Cold seeps are commonly associated with seafloor features like pockmarks, craters, carbonate mounds, or underwater pingos (Dando et al. 1991;Lammers et al. 1995;Hovland and Svensen 2006;Ritt et al. 2011;Zeppilli et al. 2012). Pockmarks are circular depressions, formed where upward seepage of gas causes a collapse of sediment, and are common features where gas pockets are present in near-surface sediments (Cathles et al. 2010). Long-term seepage of methane can be accompanied by the precipitation of methane derived authigenic carbonates (MDAC) (Bohrmann et al. 1998;Hovland et al. 2005), leading to sometimes extensive outcrops of MDAC in the vicinity of cold seeps (Vanreusel et al. 2009;Bowden et al. 2013). These combinations of features result in seafloor habitat heterogeneity fundamentally different from the predominant monotypic soft-bottom environment in the deep sea (Rex 1981).
Cold seeps, in general, exhibit a wide range of seep fauna, i.e., chemo-obligate species Levin et al. 2016), and usually support macrofaunal communities with high abundances and biomass but low diversity compared to surrounding non-seep habitats (Levin 2005;Tarasov et al. 2005). Seep-associated organisms may rely directly (chemoautotrophic symbionts) or indirectly (trophic relationships) on anaerobic oxidation of methane (AOM) and sulfate reduction as an alternative energy source, instead of, or in addition to, photoautotrophy (Levin and Michener 2002;Boetius and Suess 2004;Levin 2005;Niemann et al. 2013;Thurber et al. 2013). Habitat heterogeneity and chemosymbiotic foundation species associated with cold seep environments (e.g., siboglinid tubeworms, clams, and mussels) may attract epifauna and vagrant mobile organisms from surrounding habitats, and thereby increase the local diversity of these deep-sea ecosystems (Sellanes et al. 2008;Levin et al. 2016).
Few studies document macrobenthic communities at seeps in polar regions (i.e., Domack et al. 2005;Å str€ om et al. 2016). One of the most well-studied deep cold seep and chemosynthetic systems within the Arctic is the Håkon Mosby mud volcano (HMMV) located at the border to the Barents Sea (728 N, 148 E) at 1250 m water depth (Vogt et al. 1997;Lein et al. 1999). HMMV has been a focus of biogeochemical and geophysical studies documenting the fate of venting methane from the seabed (Milkov et al. 1999;L€ osekann et al. 2008;. The benthic environment around the caldera of HMMV consists of three main habitats; microbial mats, siboglinid (pogonophoran) worm fields and plain light-colored sediments, each possessing different faunal community patterns influenced by seafloor methane emissions (Gebruk et al. 2003;Rybakova et al. 2013). Megafaunal densities and taxa richness varied significantly in relation to these different habitats, (Rybakova et al. 2013) and methane derived carbon is incorporated into the faunal communities via trophic interactions (Gebruk et al. 2003;. Å str€ om et al. (2016) described macrofaunal benthic communities associated with cold seeps around western Svalbard and the northwest Barents Sea shelf (75-798 N). They found distinct seep associated faunal assemblages, novel bivalve species (Å str€ om et al. 2017), and higher biomass at seeps compared to nearby non-seep areas, but with highly localized effects reflecting strong environmental gradients associated with individual seeps.
Discoveries of extensive methane seepage areas around the high-Arctic Svalbard archipelago have been described both from shallow ocean shelves (Solheim and Elverhøi 1993;Westbrook et al. 2009) and from the deep sea (Vogt et al. 1994;Hustoft et al. 2009). Of particular focus in the present context is an active methane venting region at the Vestnesa Ridge (VR), along the continental slope in Fram Strait at 798 N and > 1200 m water depth. Numerous pockmarks along this ridge are associated with sub-seabed methane hydrate reservoirs (Vogt et al. 1994;B€ unz et al. 2012;Plaza-Faverola et al. 2015) and origin of the gas is both microbial and abiotic/thermogenic . Plaza-Faverola et al. (2015) documented the history of methane seepage for the last 2.7 My along VR identifying multiple historical events of seepage. There have also been methane seepage events identified in the stratigraphic record through analysis of fossil marine fauna (Ambrose et al. 2015;Sztybor and Rasmussen 2016). Hong et al. (2016) document vigorous biogeochemical processing and transformations in the surface sediments at VR consistent with high methane consumption via microbial AOM. There has not, however, been a previous formal study of the distribution and abundance of benthic organisms at methane-rich deep-sea pockmarks at such northerly latitudes.
We investigated faunal community patterns of active cold seeps in pockmarks at VR, focusing on infaunal and megafaunal assemblages. We assessed species associations, ecological structure, and diversity of macrofauna by comparing deep-sea infaunal samples, sediment characteristics, and methane concentrations at active seeps and a nearby inactive region. In addition to infaunal communities, we describe benthic megafaunal composition associated with pockmarks from VR from analyses of seafloor images. By combining results of analyses from infaunal and megafaunal communities, this study provides new insights into faunal community patterns in a high-Arctic deep-sea methane seep.

Study regions
In order to distinguish between cold seep and conventional deep-sea benthic infaunal communities, we compared an active cold seep region, VR, and an adjacent inactive control region, Svyatogor Ridge (SvR) in the Fram Strait. Both regions are located in the high-Arctic  at the northeastern North Atlantic continental shelf margin (Fig. 1). The Fram Strait is the opening between Greenland and Svalbard and it is the main passage for the exchange of both intermediate and deep-sea water between the Arctic and North Atlantic (Rudels et al. 2000). VR (Fig. 1), is an approximately 100 km long ultraslow spreading sediment-drift ridge, (798 N, > 1200 m water depth)  located south of the Yermak Plateau and north of the Molly transform fault. We investigated two pockmarks at Vestnesa, both with infaunal benthic sampling and with seafloor imagery (Fig. 2). The pockmarks are approximately 500 m wide or long and around 10-15 m deep and named "Lomvi" and "Lunde" (Fig. 2). Multiple methane bubble plumes have been acoustically detected in the water column above the pockmarks rising up to 800 m above the seafloor (B€ unz et al. 2012). Both pockmarks support MDAC outcrops precipitated at the seafloor. These are rock-like formations coupled to the presence of methane and gas hydrates in the sediment (Bohrmann et al. 1998;Cr emière et al. 2016).
Approximately 80 km south of VR, another ridge complex, SvR (788 N) (Fig. 1), is located on the northwestern tip of the Knipovich Ridge. VR and SvR once belonged to the same ridge complex, before being separated by the Molloy transform fault . Seismic profiles at SvR indicate paleo-seep features including chimneys (acoustic blankings), pockmarks, and sub-surface gas, but no active venting of methane or hydro acoustic plumes have been observed from the region. Å str€ om et al.
Cold seeps in the high-Arctic deep sea

Benthic sampling
We sampled the benthos at the two active seeping pockmarks, Lunde and Lomvi, on VR and at the inactive control region, SvR (Figs. 1, 2; Table 1) in June 2014 and May 2015 aboard the RV Helmer Hanssen. Range finding and bathymetry were conducted with ship-mounted multi-beam and threedimensional (3D) seismic surveys. Locations of active hydrocarbon seepages were selected based on acoustic reflections from flares detected on a keel mounted single beam echo sounder (Simrad EK 60 frequencies 18 KHz and 38 KHz) and on information from previous geophysical surveys in the region (B€ unz et al. 2012;Smith et al. 2014;Plaza-Faverola et al. 2015). Benthic sampling in 2014 at SvR and VR was conducted using a multicorer (ø 10 cm) or by subcoring (ø 10 cm) from a box core. Sampling in 2015 at VR was carried out where characteristic seep features such as flares, microbial mats, and MDAC (hereafter referred to as carbonate outcrops) were identified through seafloor imagery. Images were captured with a Tow Cam, a camera guided multicorer (cores ø 10 cm) and conductivity, temperature, depth (CTD) sampler. This towed camera system was developed through Multidisciplinary Instrumentation in Support of Oceanography (MISO) at Woods Hole Oceanographic Institution (http:// www.whoi.edu/main/instruments/miso). The Tow Cam system was equipped with a 16 megapixel still digital camera that transferred data from the camera and CTD in real time to the research vessel, which allowed a guided sampling effort. Additionally, we noted large-scale seafloor characteristics such as depressions, cracks, and rock slabs in the cruise log. Vertical CTD (SBE 9 plus sensor) profiles of seawater hydrography were taken at each location. Water collected from the CTD was used to measure water column methane concentrations. We also collected qualitative core samples in order to analyze sediment characteristics such as porosity, grain size, total organic carbon (TOC), benthic chlorophyll a (Chl a) pigments, and sediment methane concentrations.

Infaunal samples
In total, we sampled 20 quantitative core samples, nine from the active VR (three stations) and 11 from the inactive SvR (three stations) in order to characterize macro-infaunal communities. In this survey, we have targeted macrofauna ( 500 lm) that live mainly inside the sediment and are hereafter referred to as "infauna." The samples were sieved on board with a mesh size of 500 lm. Material retained on the sieve was fixed in formaldehyde (4%), mixed with rosebengal for staining living tissues, and the solution was buffered with borax (sodium tetra-borate decahydrate). Samples were sorted and identified to the lowest possible taxon and stored in 80% ethanol. This procedure followed the ISO 16665:2014 fieldwork protocols to ensure consistency and quality control of benthic faunal surveys. Organisms were first separated into five main phyletic groups: Crustacea, Echinodermata, Mollusca, Polychaeta, and Diverse (containing members of Brachiopoda, Nemertea, Oligochaeta, and Sipuncula). Each individual was counted and weighed (aggregated wet weight in phyletic groups). Planktonic taxa were excluded from analysis as were Foraminifera and Nematoda since such taxa are not properly retained on a 500 lm mesh size.

Megafaunal communities and seafloor images at Vestnesa pockmarks
In order to characterize epifaunal megafauna from VR, we took a continuous series of seafloor images every 15 s along two transects from the outside of the active pockmarks moving toward the center of the feature. Images were taken at an altitude of approximately 2.5-5 m above the seafloor and were analyzed for the distribution and abundance of megafauna. We assigned images into three spatial categories (hereafter referred to as "locations") relative to the center of the pockmark: "Outside," "Edge," and "Inside." These designations were determined based on habitat changes observed from the images in relation to ship/camera location at the pockmark. A total of 144 images were analyzed from two transects that moved progressively from the outside to the inside of the two pockmarks (one transect at each pockmark; Fig. 2). Each image was manually analyzed, and the presence or absence of visible epifaunal megafauna taxa was noted (hereafter visible epifaunal megafauna is referred to as "megafauna"). The resolution of seafloor images did not always allow identification of taxa to species-level. Instead, taxa were grouped into morphologically different faunal groups based on higher taxonomical ranks (i.e., phylum, class, and order), appearance, and size.

Methane measurements in sediment and water column
For compositional analyses of methane in water and sediments at each region (VR and SvR), a conventional headspace sampling preparation technique was applied. Bulk sediments (5 mL), collected from sediment cores, were subsampled with a plastic syringe. The sediments were transferred into 20 mL headspace glass vials containing 5 mL of 1-molar NaOH solution and two glass beads. Vials were immediately capped with rubber septa, sealed with aluminum crimp caps and shaken. Seawater was collected with 5 L Niskin bottles mounted on a 12-bottle rosette for water column vertical profiling. Immediately after recovery of the rosette, 60 mL plastic syringes were flushed three times and filled with water aliquots from Niskin bottles. Five milliliters of pure nitrogen gas was introduced into each bottle as a conventional headspace and the syringe was shaken for 2 min to allow the headspace nitrogen to equilibrate with the dissolved methane in the water sample. Sediment and water samples were stored at 28C prior to analysis, and were analyzed within 1-2 h.
Methane and other hydrocarbon concentrations were determined with a gas chromatograph (GC) ThermoScientific FOCUS GC equipped with a flame ionization detector (FID). Only methane was separated in water samples at 1708C and with the isothermal oven temperature set to 408C. To separate methane and other hydrocarbons in sediment samples, temperature was altered between 408C, 708C, and 1208C. Hydrocarbon gases were separated on a column RESTEK HS-Q 80/100, 2 mm using hydrogen as the carrier gas. The system was calibrated with external standards of 2 ppm and 30 ppm (Air Liquide). For methane concentrations in sediments, porosity data from benthic sediment sampling was used to convert gas chromatograph (GC)-flame ionization detector (FID) results between ppm and nmol. Methane concentrations in water samples presented here in nmol L 21 were calculated according to Wiesenburg and Guinasso (1979) with consideration of salinity, sample temperature, and ambient atmospheric pressure.

Benthic pigment and sediment analysis
Sediment samples were collected to measure benthic Chl a and phaeopigments (PhP), as indicators of photosynthetically based organic material at the two regions, VR (active) and SvR (inactive). Sediment Chl a indicates the fresher, relatively recently produced material settled at the seafloor, whereas PhP are a degradation product of Chl a. Sediment pigment concentrations from the two regions (VR and SvR) were analyzed by fluorometry in accordance with Holm-Hansen et al. (1965). Chl a and PhP samples were extracted with acetone for 24 h in the dark, centrifuged, decanted, and measured for fluorescence in a Turner Design Model 10 AU fluorometer before and after acidification with hydrogen chloride (HCl). The measured concentrations were corrected for sediment porosity.
Porosity of sediment samples from both regions was determined by using a wet-dry method where pre-weighed vials of known volume were filled with sediment, reweighed and later dried at 608C until all water evaporated (Zaborska et al. 2008). The density of the sediment was calculated by using the basis from the wet weight of sediment and water combined.
Sediment grain size (fraction of pelite < 0.63 lm) and TOC were determined by subsampling core samples (minimum 50 g) from downcore profiles from the two regions.
Grain size was determined according to Bale and Kenny (2005). The TOC samples were analyzed with a Shimadzu SSM TOC 5000 and Elementar Vario TOC Cube.

Statistical analysis Infaunal data
Infaunal abundances from core samples were used to calculate community diversity parameters including species richness (S), Evenness (J 0 ), Shannon Wiener Diversity (H 0 log e ). Single square-root transformation and standardization were carried out on infaunal abundances to balance the impact of both highly abundant and rare taxa in the same dataset. Abundance data was also used to conduct a principal component analysis (PCA) (non-parametric test, Primer V C 6; Clarke and Gorley 2006). To test differences in community structure and biomass, an analysis of variance (ANOVA) (non-parametric single-factor ANOVA: Kruskal-Wallis test on ranks) was used, with "methane seepage" (i.e., active vs. inactive) as the dependent variable. All pairwise comparisons were made using Dunn's test with an overall significance level of p 0.05 using SigmaPlot v.12.5.

Megafaunal composition
The presence/absence data of faunal groups from the seabed photos were analyzed with a two-factor ANOVA (dependent variables "pockmark" and "location") to detect differences between the two pockmarks (Lunde and Lomvi) and among the three assigned locations ("Outside," "Edge," "Inside"). Analysis by two-factor ANOVA, after testing the conformity of the dataset for the assumptions of ANOVA (normality of distributions and homogeneity of variances) was performed on square-root transformed data using Sigma-Plot v12.5.
Species richness data from presence/absence of taxa was assembled in Primer V C 6, in a Bray-Curtis similarity matrix based on the number of taxa represented at each image. This was followed by a two-way similarity percentage analysis (SIMPER) in order to identify the dissimilarity in species richness between the groups based on "location" and "pockmark" and a PCA in order to identify the faunal taxa contributing most to the variance of each group (i.e., "location" and "pockmark").

Environmental statistical data-sediment and methane
We used various statistical tests to differentiate environmental and sediment characteristics of the two regions; active VR and inactive SvR. Downcore profiles (0-30 cm, 10 samples) of sediment grain size (% pelite) and TOC satisfied the conditions of normality and equal variance and were tested with a Student's t-test to identify significant differences between VR and SvR. Sediment Chl a and benthic pigment concentrations between VR and SvR were compared with a single-factor ANOVA after the data were examined for normality and equal variances. The power of the tests was below the desired < 0.8 due to small sample size per

Environmental characteristics
Both regions, active VR and inactive SvR, exhibit oceanographic characteristics of a typical Arctic deep-sea habitat located below 1200 m of water depth, possessing dense bottom water with high salinity (34.9 psu), sub-zero temperatures (20.88C), and relatively high oxygen content (5.4 mL L 21 ).
Chl a and PhP concentrations in the upper 0-2 cm surface sediment from VR and SvR were not significantly different, p > 0.05, (F(1,4) 5 0.64, p 5 0.47), (Fig. 3). The ratio between the amount of fresh production (Chl a) and degraded pigments (PhP) was 17% at VR and 13% at SvR showing that VR had a slightly higher portion of "fresh," recently produced Chl a.
The main environmental difference between the regions, active VR and inactive SvR, was the presence of methane. In the bottom water at VR, the methane concentration was 76.4 nmol L 21 , whereas at SvR the concentration was 2.2 nmol L 21 . Methane concentrations were significantly higher p < 0.05, (F(1) 5 232.6, p < 0.001) in the sediment at VR compared to SvR. Downcore profiles of methane in the sediment at VR varied between 262.0 ppm at the sediment surface and 1911.5 ppm at 60 cm below the seafloor (Fig. 4); with maximum value for an individual sample of 9219 ppm at 39 cm below seafloor level (bsfl). At SvR, the methane concentration in the sediment ranged from 2.1 ppm to 11.3 ppm from 0-60 cm bsfl, with a maximum sample value of 13.7 ppm at 54 cm bsfl. We recovered pieces of gas hydrate from sediments at VR collected from gravity cores.

Seabed features at Vestnesa pockmarks
Different habitats at VR are observed when moving along transects into the depression of the two active pockmarks Lunde and Lomvi. Inside these pockmarks, hard rock features, identified as carbonate outcrops compose reef-like 3D structures among scattered patches of soft-bottom sediment covered by microbial mats and worm tufts (chemosymbiotic siboglinid polychaetes) (Figs. 5,6). Toward the edge of each pockmark, carbonate outcrops disappear, and soft-bottom sediments with patches of microbial mats and worm tufts predominate. Outside pockmarks, the seabed is dominated by relatively homogenous and featureless deep-sea soft-bottom plains, interrupted by ice rafted-debris such as drop stones (Fig. 5).

Megafaunal patterns at active pockmarks
The large-scale structural habitat differences (see previous section) at the two pockmarks clearly influences the composition of megafauna (Figs. 6, 7) at the three main locations ("Outside," "Edge," "Inside"). "Inside" the pockmark, the community was composed of a combination of hard-and soft-bottom living organisms. Foliose and calcareous bryozoans, stalked hydroids, small gastropods, different species of sponges and large pycnogonids (possibly Colossendeis sp.) occurred on the carbonates (Fig. 5f), occasionally with gadiform fishes (rockling-like morphotype). Between carbonate outcrops, soft-bottom patches were colonized by microbial mats and siboglinid worm tufts which were often associated   siboglinid tufts remain common (Fig. 5c,d) indicating a diffuse methane flux at this locality. The microbial mats and the siboglinid tufts present at the "Edge" are mixed with conventional deep-sea fauna including ampharetid polychaetes, anemones, pycnogonids, gastropods, and amphipods. Along the "Edge," microbial patches and siboglinid tufts gradually disappear and the seafloor becomes densely populated by ophiuroids. A two-way ANOVA tested the factors "pockmark" (Lunde or Lomvi) vs. "location" (Inside, Edge, Outside) for megafaunal composition. We found significant differences between pockmarks (p < 0.001) and among locations (p < 0.001) ( Table 3). "Inside" megafaunal species richness was significantly higher than "Outside" at both pockmarks, however, pockmark Lunde had significantly higher overall taxon richness compared to Lomvi. The "Edge" locations differed between the two pockmarks and was significantly different from each other in terms of faunal composition (Fig. 6). The faunal community at location "Edge" was either similar to the "Outside" community (for Lomvi) or more similar to the "Inside" community (Lunde). This pattern explains the significant interaction term in the ANOVA (Table 3).
The SIMPER analysis of presence/absence of megafaunal taxa identified from the seafloor images demonstrate that average dissimilarity between pockmark "Lomvi" and "Lunde" across all locations (Inside, Edge, Outside) was 49.6%, where "Anemone" (soft) and "Bivalvia" represented the largest dissimilarities. This separation between the two pockmarks is observed in the PCA-plot of the presence/ absence of identified megafauna and where "Lunde" and "Lomvi" are separated along the y-axis, PC2 (Fig. 7a). Between locations for both pockmarks, the highest average dissimilarity is seen for locations "Inside" and "Outside" (75.4%) and the largest differences is recorded for the faunal taxa "Ophiuroidea" and "Siboglinidae," comprising a dissimilarity of 12.2% and 11.1%, respectively. These differences between "Outside" and "Inside" locations are indicated in the PCA (Fig. 7b) where the taxa "Ophiuroidea" and "Siboglinidae" are split along the x-axis (PC1).

Infaunal community structure
There are clear differences in infaunal composition and community structure between the active VR and inactive SvR regions. There is no overlap among the top five most dominant taxa at VR stations compared to those at SvR (Table 4). For the entire survey (active and inactive samples combined), the top five most dominant taxa contributed to 61.6% of total infaunal composition: for VR, they contributed to 70.1% and for SvR the contribution is 75.4%. The PCA (Fig. 9) illustrates the separation between active and inactive stations and replicates with infaunal taxa clearly separated along the x-axis (PC1), distinguishing active seep samples from inactive controls. Among individual replicates (Fig. 9), the SvR samples are more dispersed along the y-axis (PC2) compared to VR replicates. This indicates a larger faunal variation among samples at the inactive SvR relative to the active VR samples.
Total infaunal abundance tested in a one-way ANOVA for the factor "methane seepage" shows a significant difference (p 0.05) between the VR and SvR samples (Table 5). Aggregated total faunal abundance (group mean of all stations, separated by location, i.e., active VR or inactive SvR) is more  (Fig. 10a). The highest and lowest mean biomass per station is 5.07 g ww 0.1 m 22 at Sta. VR1 and 0.47 g ww 0.1 m 22 at Sta. SvR2, respectively (Table 6). Total infaunal biomass is five times higher at VR compared to SvR (Fig. 10a). This difference between the two locations is significant (one-way ANOVA; p < 0.05) ( Table 5). Species richness (S) and diversity (H 0 ) are significantly different between VR and SvR (one-way ANOVA; p < 0.05) ( Table 5; Fig. 10b). Comparing species evenness (J 0 ), the difference between active and inactive locations is not significant (one-way ANOVA; p > 0.05).
In the overall survey, encompassing 20 replicate cores, 74 taxa were identified, distributed among seven phyla. The phylum Annelida (class Polychaetea) contributes more than half of the 74 identified taxa (41) and represents 32.8% of the total relative faunal abundance. The second largest taxonomic group is Mollusca, divided among 13 taxa and Siboglinidae. indet ( and Svyatogor (inactive). Taxon names are displayed for faunal groups where at least 20% of the variance is explained on the first two PC-axes.
There is a clear separation on PC1 (x-axis) between taxa found at the active VR stations compared to taxa recorded at the inactive SvR controls.  (Table  6). Polychaetes dominate the total infaunal biomass, contributing more than 50% to total relative biomass, second largest is the group "Diverse," contributing 17.5% to the total relative infaunal biomass in the overall survey.

Discussion
Methane-an energy source for benthos at VR VR and SvR are two high-Arctic regions located at water depths greater than 1200 m. Both VR and SvR exhibit deepsea characteristics (Sanders and Hessler 1969) regulated by sub-zero temperatures and high-Arctic seasonal variations. There were no differences in water temperature, salinity, oxygen concentration, grain size, or sedimentary pigment concentration (photosynthetically produced organic matter) between VR and SvR. The main extrinsic difference between these two regions is methane seepage. We recorded sediment concentrations of methane up to 100-1000 times higher in upper surface sediment layers (0-60 cm bsfl) and 50-70 times higher in bottom water at VR compared to SvR (Fig. 4). Bottom water methane concentration at SvR (2.2 nmol L 21 ) is considered background concentrations for marine environments (Rehder et al. 1999;Gentz et al. 2014).
Marine environments in the Arctic are characterized by strong seasonality with respect to input of photosynthetically derived organic matter during a short and intense productivity   (Sakshaug 2004;Søreide et al. 2006;Ardyna et al. 2013). Photosynthetically derived organic matter produced in the photic zone and transported to the seafloor via trophic interactions and organic matter sedimentation is the energy source for conventional deep-sea ecosystems (Sanders and Hessler 1969;Rice et al. 1985;Gooday et al. 1990). The short productive season and the long transit of organic matter through the mesopelagic zone can result in high-Arctic deep-sea environments being food-limited over extended periods (Graf 1989). Compounds such as methane and sulfide, through microbial processes, can serve as an alternative local energy source for organisms that are able to utilize chemoautotrophic production in addition to photosynthetically-derived organics from the euphotic zone (Carney 1994). The similarity in all measured environmental parameters at active VR and inactive SvR, with the exception of methane, strongly suggests that methane emissions provide an alternative food source for conventional heterotrophic consumers at the deep VR seeps. Methane emissions result in a heterogeneous environment comprised of microbial mats, chemosynthetic worm tufts, and carbonate outcrops, providing 3D-structure and hard substrate at an otherwise relatively featureless and homogenous soft-bottom seafloor (Fig. 11).

Distinct infaunal taxa at methane enriched environments
There is little overlap in infaunal community composition between active VR and inactive SvR ( Fig. 9; Table 4). None of the top five taxa from VR stations occurs at any of the SvR stations and the top five most abundant taxa from SvR are a minor part of the total relative abundance at VR. We attribute these differences to the methane activity at the cold seep. Four of the top five infaunal taxa at VR seeps (Table 4) are recognized from reduced environments including cold seeps Blazewicz-Paszkowycz and Bamber 2011;. The most numerically dominant taxa among these is Tanaidacea that occurred en masse at Sta. VR3. Tanaid crustaceans are commonly observed in deep-sea macrobenthos (Bluhm et al. 2011) and have previously been identified at deeper shelf systems around Svalbard Soltwedel et al. 2015) and cold seep environments along the Norwegian shelf (Blazewicz-Paszkowycz and Bamber 2011; Decker and Olu 2012). Tanaidacea can exploit both chemo-, and photo-autotrophic energy sources and have been suggested to feed on microbial mats, (Sellanes et al. 2011;Levin et al. 2016), which offer alternative food resources during periods of low organic matter input. The high densities of tanaids seen at VR are comparable to densities of tanaids recorded both in cold seep habitats along the Norwegian shelf and from New Zealand. At the HMMV, tanaid densities near 20,000 ind. m 22 were recorded  in chemosymbiotic siboglinid fields of Oligobrachia haakonmosbiensis Smirnov, 2000, (also referred to as O. webbi in Meunier et al. 2010). Similarly, at Nyegga pockmark at the Norwegian shelf, 648 N,  documented high occurrences of tanaids in fields of Sclerolinum cf. contortum Smirnov, 2000. High densities of Tanaids are also found at Svalbard shelf seeps (Å str€ om et al. 2016). Presumably, the high densities of tanaids from the VR cold seep in this survey exemplifies the benefits of exploiting alternative food resources in deep-sea environments.
Oligochaetes are also taxa recognized from chemosynthetic habitats where some species (gutless oligochaetes) rely on microbial endosymbionts for nutrition (Blazejak et al. 2005;Dubilier et al. 2008). While we observed oligochaetes from several VR replicates, they were not present at SvR, and we can only speculate whether these oligochaetes rely on symbiosis with microbes for nutrition. The most abundant mollusk at VR is Thyasira dunbari Lubinsky, 1976, a high-Arctic species recorded from a wide range of circumpolar-Arctic habitats from both shallow bays and the deep sea. Thyasirids are a family of bivalves commonly known from reduced habitats, although they are not exclusively dependent on mutualistic trophic interactions with endosymbionts, and there is a wide range of nutritional dependence on symbionts within this family (Payne and Allen 1991;Dufour 2005;Taylor and Glover 2010). T. dunbari seems to be a highly adaptive species based on its habitat preferences. For example, Włodarska- Kowalczuk and Pearson (2004) documented high abundances of T. dunbari from a shallow, inner glacial bay community (< 100 m Kongsfjord, Svalbard) where the bay community was strongly influenced by glacial induced disturbance and high sedimentation rates. Furthermore, T. dunbari was the dominant macrofaunal species from deeper slopes > 1500 m in the Fram Strait representing over 20% of the total macrofaunal community (Soltwedel et al. 2015). The only obligate chemosynthetic infaunal taxa found in the present study is siboglinid worms, a chemoobligate group of polychaetes that hosts microbial endosymbionts known to utilize methane and sulfur sources (Pleijel et al. 2009). We document high densities of siboglinid worms (2930 ind. m 22 ) from one station at Vestnesa (VR1). Underwater photographs (Figs. 5, 8) reveal large, dense fields of siboglinid worms adjacent to seepage features including carbonate crusts and microbial mats inside the pockmarks. Due to difficulties in the taxonomy of this group and poorly preserved samples, identification to species level has not yet been possible.
In our survey, we see a number of taxa at active VR seeps that are directly (Siboglinidae) or partially dependent on or related to chemoautotrophy (Tanaidacea, Oligochaeta, Thyasiridae) while these species are absent at inactive SvR. We also document higher TOC contents in sediments at VR compared to SvR (Table 2) with no differences in sedimentary pigment concentrations. This indicates that there is an additional source of organic carbon at VR that we suggest originates from chemoautotrophy. Hence, the influence of chemosynthesis on the infaunal community at VR is a viable explanation for the large observed regional differences in abundance, biomass, and diversity between the inactive SvR and active VR.

Enhanced infaunal community structure at methane enriched environments
There were large significant differences in infaunal community indices between the active VR and inactive SvR stations, with total abundance and biomass approximately five times higher at VR compared to SvR (Fig. 10a). The number of taxa (species richness) and Shannon Wiener diversity were also significantly higher at VR compared to SvR (Fig. 10b). The total infaunal abundance from all three VR stations , demonstrating that VR has an elevated overall total faunal abundance. Deep-sea environments are usually recognized as systems with low biomass, although the number of species and faunal diversity can be high (Sanders and Hessler 1969;Rex 1981). Cold seeps generally represent habitats of high species abundance and high biomass (Vanreusel et al. 2009;Cordes et al. 2010). High biomass has also been reported from cold seeps at the Svalbard shelf (Å str€ om et al. 2016). The average biomass from VR stations (2.68 g ww 0.1 m 22 ) is slightly higher than the biomass recorded by  at 1500 m in Fram Strait, where they reported much higher values compared to previous studies over the deep-sea Arctic basins (Paul and Menzies 1974;Kr€ onke 1998). The total biomass from seep Sta. VR1 (Table 6) is similar in magnitude to troughs and depressions around the Svalbard shelf (< 500 m deep) Cochrane et al. 2012;Å str€ om et al. 2016) and > 10 times higher than the shallowest stations along a latitudinal transect in Fram Strait 2300 m (Vedenin et al. 2016). This indicates that infaunal biomass at our active deep-sea seep is comparable to adjacent shelf stations in Svalbard, and therefore suggests a substantial enhancement to the infaunal community from a localized energy source.
Species richness is, on average, over 2.5 times higher at the VR stations compared to SvR and Shannon Wiener diversity is also significantly higher at VR relative to SvR (Fig.  10b). The diversity indices from VR are, however, not unique in relation to similar studies, either from conventional deepsea environments, or shallower shelf or fjords in Svalbard (Włodarska-Kowalczuk et al. 2004, 2012Renaud et al. 2007;Vedenin et al. 2016). Cold seep systems or other chemosynthetic environments exhibiting strong chemical gradients are usually known to have high faunal abundance and high biomass but low diversity due to chemical stress from compounds such as hydrogen sulfide (Vismann 1991;Warwick and Clarke 1995;Bernardino et al. 2012). We observed the opposite, with the infaunal community at the VR active seep being more diverse than the non-seep control location. This could be a result of higher productivity via chemoautotrophy at the seep or due to an effect of deep-sea heterogeneity, where the SvR stations are comparatively barren, homogenous, and food-limited. Species evenness is the only faunal index that did not show significant differences between active and inactive locations. In general, evenness is uniformly high in all stations, ranging from 0.70 to 0.98 with the exception of a single replicate with mass occurrence of tanaids (J 0 5 0.26). The species evenness values from our study are slightly higher than was documented from Svalbard shelf seeps (Å str€ om et al. 2016). The relatively high J 0 in this survey may be related to the patchy distribution of organic matter in deep-sea environments (Rex 1981;Gage and Tyler 1991) and the relatively small abundance per sample of many different taxa.

Methane derived carbonate outcrops-a substrate for megafauna
The environment inside and outside pockmarks can vary substantially due to strong gradients in physical and environmental drivers such as currents, sedimentation, substrate and geochemistry, structuring faunal communities (Dando et al. 1991;Hammer et al. 2009;Webb et al. 2009a). Pockmarks from shallow shelves and slopes worldwide are known to attract aggregations of motile megafauna regardless of seepage activity, where local heterogeneity attracts "background" (conventional) fauna even when there is no gas seepage (Hovland and Judd 1988;MacDonald et al. 2010;Zeppilli et al. 2012). Regardless of depth and seep activity, pockmarks can also act as a refuge for slow growing species such as corals, cnidarians, and sponges as well as for fish populations especially in regions impacted by intense trawling pressure (Webb et al. 2009b;MacDonald et al. 2010;Clark et al. 2016).
There is a paucity of information on the ecology of deepsea pockmarks (deeper than 1000 m), particularly, the association of seep communities and local, conventional fauna (Olu et al. 2009;Ritt et al. 2011). Our analysis of seafloor images reveals large changes in megafaunal composition along transects from outside and into the pockmarks. A key driver of this faunal change is bottom substrate. Outside the pockmark, the environment is relatively homogenous, dominated by vast expanses of soft-bottom substrate. This outer locality ("Outside") of the pockmarks shows a megafaunal pattern with an even distribution of ampharetids and brittle stars occasionally interrupted by the presence of soft-bottom anemones and motile organisms such as starfish, skates, bivalves, and amphipods. Burrowing tracks (or "lebenspuren") in the sediment were commonly seen on the images, indicating activity of motile megafauna. This activity also creates micro-scale heterogeneity in the soft bottom environment (Qu eric Taylor et al. 2016). Epifaunal overgrowth and various fishes were commonly associated with drop stones on the surface seafloor, highlighting the importance of 3D structures and hard substrate on deep-sea soft bottom plains (Schulz et al. 2010;Meyer et al. 2014). The densities of ophiuroids recorded outside the VR pockmarks in this study (mean 49 ind. m 22 ) are higher than those reported by Soltwedel et al. (2009) from nearby locations in Fram Strait (mean 16.7 ind. m 22 ). Likewise, Meyer et al. (2013) reported similar densities (mean 16.5-19.2 ind. m 22 ) of ophiuroids as Soltwedel et al. (2009) at 798 N, 1200 m in Fram Strait in the first 2 yr of an inter-annual study of megabenthos; whereas during the last year, the density was significantly higher, (mean 49.6 ind. m 22 ), and comparable to densities in this study. Generally, echinoderms are the dominant megafauna in the Arctic (Bluhm et al. 2011;Piepenburg et al. 2011) and brittle stars are considered to be the most prominent megafaunal group in Svalbard waters . Furthermore, it has been reported that the background community at HMMV, outside the caldera, is dominated by ophiuroids, mainly Ophiocten gracilis (Sars G.O., 1871) and Ophiopleura borealis, Danielssen and Koren, 1877 (Gebruk et al. 2003). In our study, there is not a single record of ophiuroids inside the pockmarks in relation to the carbonate outcrops, microbial mats nor the siboglinid worm tufts. Moving along transects from the "Outside" locality where ophiuroids are dominant toward the "Edge," brittle stars gradually disappear and become completely absent in images where microbial mats and worm tufts appear. This observation could indicate that the brittle stars are sensitive to chemical compounds associated with the source location of the active seepage and thus only occupy habitats at the periphery of the pockmark. From Lau Basin in the Pacific Ocean, Sen et al. (2016) reported the presence of ophiuroids only from peripheral vent sites and attribute this to sulfide sensitivity since no detectable concentrations of hydrogen sulfide were recorded in the outer zones of the vents. Organisms occupying peripheral habitats may also benefit from increased productivity close to the seep because advection from seafloor emissions may influence the amount of particulate organic matter in adjacent areas. Increased vertical mixing can enhance water column productivity, supporting nearby benthos and suspension feeders in the periphery of a seep or vent (Sorokin et al. 2003;Levin et al. 2016).
At the edge of the pockmark, the megafaunal composition changes, and all brittle stars are absent whereas siboglinid worm tufts and microbial mats become present. The overriding bottom type remains soft bottom with sporadic ice rafted drop stones and thus the main habitat difference at the "Edge" is the presence of microbial mats and tufts. Motile megafauna are frequently observed in association with tufts and mats, either adjacent to them or lying or sitting directly within them. This aggregation around specific biological seep features is mainly observed with sea spiders, zoarcids, snails, and amphipods within the "Edge" and the "Inside" locations. The "Edge" community is either similar to the "Outside" community or the "Inside" depending on the individual pockmark and its seafloor morphology and associated habitat gradients. This pattern is logical since photos were taken along a transect. We expect to find gradual changes in the habitat from both abiotic structures (carbonate crusts) and biotic structures (burrowing tracks and foundation organisms such as the siboglinids and microbial mats). At pockmark Lunde, there is a slightly higher mean species richness at the "Edge" megafaunal community compared to the community "Inside" (mean 5 7.46 vs. 6.94) but this difference is not significant (p > 0.05). The phenomenon of higher species diversity at an edge-ecotype, however, is known as an "edge-effect" and implies that at the boundary of two shifting habitats there will be a mix of species from both habitats, possibly generating greater complexity and biodiversity (Livingston 1903;Harris 1988).
Large changes in habitat heterogeneity occur moving toward the "Inside" of the pockmark. Here, carbonate outcrops rise up to several meters above the surrounding seafloor. The "Inside" locality combines both hard and soft bottom substrates, allowing colonization by organisms with different habitat requirements or preferences. Correspondingly, we notice a large variety of visible megafauna (Fig. 6). We observed aggregations of zoarcid fishes primarily on or at the edges of microbial mats and worm tufts, lying on the seafloor in softer sediments between carbonate outcrops, whereas sea spiders and small gastropods were seen on top the carbonate outcrops. Various epifaunal taxa including sponges, hard-bottom anemones, bryozoans, and hydroids were attached to the carbonate structures. The occurrence of various organisms aggregating around reefs, outcrops, and other 3D structures is a well-known phenomenon from several studies comparing natural and artificial reef structures (Stone et al. 1979;Bohnsack 1989;Baine 2001). Habitat complexity, the physical substrate and shelter to avoid predation, are all believed to be important factors in attracting organisms to such structures (Stone et al. 1979;Wilson and Elliott 2009;Ashley et al. 2014). The physical difference between the soft bottom plains outside the pockmarks and the carbonate concretions inside is apparent and is reflected in the megafaunal composition with significantly different taxon richness between the "Inside" and "Outside." This pattern highlights the importance of such natural structures in an otherwise non-complex environment, namely, the deep-sea soft bottom plains.
The "oasis-effect" at active seeps The pattern of megafauna at the methane seeping pockmarks is characterized by strong spatial differences in faunal composition and large aggregations of both low-and hightrophic level taxa. It is most likely related to two factors: (1) the reef-like MDAC provide a 3D structure and add complex heterogeneity to the deep sea, offering shelter and substrate to both sessile epifauna and motile fauna and (2) increased food availability from a local chemosynthetic source that supports a diverse community including aggregations of larger and higher trophic-level organisms (Fig. 11). These characteristics of the pockmark serve to attract organisms from the surrounding environment that interact with the chemosynthetic community.
Visual observations of the seafloor at VR indicate that methane emissions create seafloor heterogeneity unique to a cold seep system resulting in large habitat variability within the pockmark area. This heterogeneity, represented as a patchwork of microbial mats, worm tufts and the carbonate outcrops, likely drives the pattern of megafaunal species distribution along the transect. The presence of large, hightrophic level organisms around carbonate outcrops or drop stones exemplifies the importance of structural objects in deep-sea environments (Meyer et al. 2014). However, the presence of aggregated biomass and/or predatory organisms may also be related to food availability. Although we did not measure the isotopic composition (d 13 C), indicative of overriding carbon source of organisms, there is little doubt that there is an additional source of nutrition supporting the infaunal community at Vestnesa seeps which is absent from the community at the SvR. Gebruk et al. (2003) reported highly depleted d 13 C of Lycodes squamiventer (251.9&) from a site at 76807 0 N, 6810 0 E, (referred to as VR in Gebruk et al. 2003 but located further south along Knipovich Ridge compared to sites investigated in this study), where they suggested that the diet of the zoarcids includes chemosymbiotic siboglinids (Sclerolinum sp.). Also, paleo-communities of chemosymbiotic vesicomyid bivalves ( 17,000 yr B.P.) from VR pockmarks exhibited depleted shell organic and inorganic d 13 C, indicating partial nutritional dependency on chemoautotrophic production (Ambrose et al. 2015). We suggest that the infaunal community at VR is supported by chemoautotrophic production in the sediment in addition to the detrital energy derived via the conventional, photosynthetic carbon cycle. The enhanced community of infaunal organisms likely serves as a food source for larger megafauna aggregating around the carbonate structures. This hypothesis is also supported by the suggestion that siboglinid polychaetes are included in the diet of zoarcids around seeps (Gebruk et al. 2003).

Summary
We have shown that active methane seepage strongly influences deep-sea benthos in the high-Arctic, resulting in enhanced infaunal abundance, diversity, and biomass compared to an inactive control region. We attribute these differences to the presence of methane and its utilization in the marine biosphere. Active hydrocarbon seepage and chemoautotrophic production at Vestnesa seeps influence benthos via mutualistic relationships and/or trophic interactions from microbes to megafauna. Thriving infaunal communities and habitat heterogeneity, both utilized by megafauna, produce a pronounced seep related "oasis effect" (Carney 1994;Levin et al. 2016). The Arctic deep seafloor oasis effect identified in this study may be the result of unique high-latitude environmental drivers such as strong seasonality and episodic productivity in the euphotic zone and sub-zero temperatures where seeps provide a refuge for conventional organisms compared to the relatively featureless, homogenous, and food limited deep-sea surroundings.