Water column profiles

Over the two cruises, we constructed 21 water column profiles covering seabed depths ranging from 55 to 5320 m. Oxygen declined steeply from an average of 193 μmol l−1 in the top 20 m, to 6.4 μmol l−1 at 80 m and then slowly until it went below our LOD (1.4 μmol l−1) at 230 m (Figure 1). Below 800 m, oxygen returned to detectable concentrations and reached 100 μmol l−1 (30% saturation) at 2500 m where it remained stable until the seabed (>5000 m). The baseline nitrite concentrations were 0.05 (LOD) to 0.2 μmol l−1 but within the OMZ, at 275–600 m, there was a large, secondary nitrite maximum of up to 1.8 μmol l−1 at 345 m and in the epipelagic waters, a smaller, primary maximum at around 50 m (max 1.37 μmol l−1) in most profiles (Figure 1). The true anoxic core of the OMZ was where oxygen was below the LOD and a clear secondary nitrite maximum was present (230–600 m).

Figure 1 Water column profiles of (a) oxygen, (b) methane and (c) nitrite were constructed from conductivity–temperature–depth (CTD) deployments with the top 800 m shown in main panels and all data (0–5000 m) shown in inset plots. The shaded grey segment indicates the core of the OMZ (230–600 m) where oxygen is below LOD (~1.4 μmol l−1) and methane and nitrite are accumulating. Black line indicates atmospheric equilibration of methane at the depth-specific temperature and salinity. Full size image

Methane was supersaturated relative to the atmosphere throughout the water column and there was a clear peak around 250–600 m (Figure 1). The maximum concentration (102 nmol l−1) was measured at 368 m on the continental shelf where the seabed depth was 506 m, although very close to the sediments (<15 cm) 254 nmol l−1 was measured. Outside the core of the OMZ, the methane was consistently above atmospheric equilibration at 3–5 nmol l−1, except for a small epipelagic methane peak (Figure 1), which was only found in some of the profiles (maximum concentration 25 nmol l−1 at 65 m).

Methane flux, methanogenesis and the methanogen community

In the intact, anoxic core incubations, methane flux averaged 262±65.5 nmol m−2day−1, peaking at 1007 nmol m−2day−1 at 550 m and with the slowest of 162 nmol m−2day−1 measured at 300 m (Figure 2a). Although all sediment cores were degassed to remove oxygen (that is, optimal conditions for methanogenesis), methane efflux was greater in sediments from locations where the OMZ intersected the shelf (indicated by shaded area on Figure 2a) compared with those with oxygenated (2–4.6 μmol l−1) bottom-water (X2 (1) =13.261, P<0.0001). The efflux of methane from the sediments was positively correlated with the concentration of methane in the bottom-water (X2 (1) =23.233, P<0.0001), which ranged from 6 to 254 nmol l−1 (Figure 2b). Further, there was a strong, non-linear inverse relationship between the concentration of methane and oxygen in the bottom-water (Figure 2b, inset, LOD~1.4 μM O 2 ).

Figure 2 Spatial variation in (a) methane flux from sediment mini-cores and (b) methane concentration in the bottom, water measured above the cores (<15 cm from sediment surface), plotted against seabed depth. The shaded area indicates where the conductivity–temperature–depth (CTD) profiles measured anoxic bottom water (~10 m from seabed). The Gaussian peak, 4-paramter regression line (black), 95% confidence (blue) and prediction (red) intervals are shown in a. Inset, oxygen concentration and methane concentration in the bottom water for the points shown in b. (c) Depth profiles of methanogenic potential (displayed on a logarithmic scale) measured in slurries at five discrete sediment layers from 0 to 25 cm in cores from 550 m (grey) and 650 m (black). As a result of this strong depth decay, all further experiments focused on the top 5 cm only. Hydrogen sulphide profile for a core from 550 m (650 m data unavailable) overlain in unfilled triangles. Full size image

Incubating anoxic sediment slurries from discrete depth intervals from two locations revealed that the bulk of the methanogenic potential was in the surface sediments (Figure 2c) and so all further experiments were focused on this layer (Table 1). Methanogenesis was also detected in the overlying water (0–5 cm above sediment) and in sediments down to 25 cm, but in the uppermost layer the rate was at least an order of magnitude higher than any other depth (Figure 2c). The greatest potential was measured in sediment from 750 m (88 nmol g−1 day−1, Table 1). Hydrogen sulphide was detected in two out of the six cores in which it was measured; at 550 m the concentration reached 59 μmol l−1 at 25 cm (Figure 2c) and in a core from 350 m it reached 219 μmol l−1 at 23 cm. The porewater profiles (Supplementary Figure S3) revealed that in the top 2 cm of sediment, where methane production was most active, sulphate (23 mmol l−1), nitrite (2.8 μmol l−1) and nitrate (86 μmol l−1) were similar to, or above, bottom-water concentrations.

A total of 126 303 mcrA gene sequences from the top 2 cm were retrieved and clustered into 16 operational taxonomic units (OTUs), hereafter named as ETNP_MG. The majority of sequences (50% of sequences represented by ETNP_MG2 and 35% represented by ETNP_MG1) were 96–97% similar to Methanococcoides sp. (Supplementary Table S2). Most of the OTUs (13 out of 16) clustered within the order Methanosarcinales, whereas two OTUs (ETNP_MG3 and ETNP_MG5) clustered within the order Methanomicrobiales and one (ETNP_MG14) within the order Methanocellales (Figure 3a). The Methanococcoides-like species dominated all five seabed samples, which displayed a similar level of intra-sample diversity (assessed by Shannon and Simpson indices, Supplementary Table S2). Principal coordinate analysis indicates that most of the variation (91.2%) in the methanogen community is explained by the first two principal coordinates (58.5% of the variation explained by axis 1 and 32.7% by axis 2, Figure 3b). The most separated communities, by the first coordinate, are those from 650 to 222 m, because of unique Methanomicrobiales (ETNP_MG5) and Methanocellales (ETNP_MG14) sequences at 650 m and Methanosarcinales at 222 m. The community at 342 m is also somewhat separated from the others, as it is the only one with ETNP_MG3 Methanomicrobiales-like sequences.

Figure 3 Methanogen community at five points across the seabed sediment. (a) Maximum likelihood tree of representative partial mcrA sequences (about 362 bp). Collapsed branches are indicated by a polygon. The mcrA gene of Methanopyrus kandleri (MKU57340) was used as the outgroup. Asterisks indicate local support values over 75%. The number in parenthesis following the ETNP_MG sequences indicates the % relative abundance of each OTU over the total number of OTUs (see Supplementary Table S2 for details). The bar represents 0.1 average nucleotide substitutions per base. (b) Principal coordinate analysis (PCoA) plot of the methanogen community based on the mcrA sequences and a maximum likelihood tree, and, constructed with the weighted Unifrac metric. Full size image

Methane oxidation and methanotrophs

Methane oxidation was measured, through the accumulation of 13C-DIC, in short-term (10–15 days) time series incubations with water from the uppermost margins of the OMZ (Figure 4a and Supplementary Figure S2); 13C-DIC was produced at a rate of 3.0–5.9 nmol l−1 day−1, and, after 15 days, 26% of the 13C-CH 4 had been oxidised to 13C-DIC. In the epipelagic zone, we could not measure any methane oxidation at 47 m but we did at 65 m (Table 2). Our dose-response experiments indicate that the methanotrophs can oxidise methane at concentrations much higher than the ambient concentration (Figure 4b). Water from both the epipelagic (65 m) and mesopelagic waters (200 m, where oxygen was below detection) oxidised increasing amounts of 13C-CH 4 to 13DIC with increasing initial methane spike (Figure 4b) and there was a good correlation between methane oxidised and 13C-DIC produced (R2=0.96). This relationship between 13C-DIC produced and starting methane concentration was linear (R2 (200m) =0.94 and R2 (65m) =0.54) for the range of concentrations tested (85–760 nmol l−1) and the slope (b1 (200m) =0.0055 and b1 (65m) =0.0057) of the relationship was similar for the two different water samples (Figure 4b).

Figure 4 Methane oxidation measured as 13C-DIC accumulation, (a) over a 13-day time series with water from 200 to 226 m, on the upper margin of the OMZ, where oxygen is close to LOD (~1.4 μmol l−1) and methane is rising above background concentrations, and (b) over 5 months with varying initial concentrations of methane. Full size image

Long-term (5 months) incubations from eight different locations (47–228 m), yielded mixed results, with methane oxidation being undetectable in some vials (Table 2). The greatest amount of 13C-DIC produced was at 200 m where, following a 3.3 nmol spike of 13C-CH 4 , 0.6 nmol 13C-DIC was measured in the water after 5 months. In the shorter incubations, water from the same location, produced a similar amount in only 10 days (Figure 4a), which indicates methane oxidation did not continue linearly during the 5 months. For comparison, if a total of 0.6 nmol 13C-DIC in the vial accumulated linearly, after 150 days, it would equate to 0.42 nmol l−1 day−1, which is 14 times slower than that measured in the short-term incubations.

Water incubated with 13CH 4 and 15NO 2 − did produce both 13DIC (0.9–15.4 nmol) and 29+30N 2 (8.9–80.5 nmol), and the relative proportions produced varied across depths with high rates of 29+30N 2 production at 235 and 264 m, indicative of nitrite reduction alongside methanotrophy (Table 2 and Supplementary Figure S2).

Methanotrophic bacteria were targeted in waters offshore (30–1250 m depth) and closer to the coast (200 and 228 m). Analysed aerobic pmoA sequences in the offshore samples (6202 in total) were clustered into six OTUs, hereafter called ETNP_Offshore_MO, and in the inshore samples (363 816 in total) were clustered into six OTUs, hereafter called ETNP_Inshore_MO. The sequences from both the offshore and inshore samples were highly similar (97–100% BLAST similarity) to uncultured bacteria from marine environments (Supplementary Table S3, Figure 5a). The vast majority of the offshore sequences are represented by two OTUs, that is, ETNP_Offshore_MO1 (44.16% of sequences) and ETNP_Offshore_MO2 (47.19% of sequences). Similarly, ETNP_Inshore_MO1 represents the majority (89.43%) of the analysed inshore sequences. Phylogenetic analysis shows that all the OTUs cluster within known type I methanotrophs (Figure 5a). Among them, three OTUs of the offshore samples (ETNP_Offshore_MO1/MO3/MO5) sit within a sub-cluster of the family Methylococcaceae including Methylococcus and Methylomonas species.

Figure 5 Aerobic and anaerobic methanotroph community of the water column. (a) Maximum likelihood tree of representative partial pmoA sequences (about 347 bp). Collapsed branches are indicated by a polygon. The amoA gene of Nitrosococcus oceani ATCC 19707 (U96611) was used as the outgroup. Asterisks indicate local support values over 75%. The number in parenthesis following the ETNP_Inshore_MO, ETNP_Offshore_MO and ETNP_NDAMO sequences indicates the % relative abundance of each OTU over the total number of Inshore, Offshore and N-DAMO OTUs, respectively (see Supplementary Table S3 for details). The bar represents 0.1 average nucleotide substitutions per base. (b) Principal coordinate analysis (PCoA) plot of the methanotroph community based on the pmoA sequences and a maximum likelihood tree, and, constructed with the weighted Unifrac metric. Circles indicate inshore samples and triangles indicate offshore samples. MO stands for aerobic methanotrophs and N-DAMO for anaerobic methanotrophs. MO_Inshore and N-DAMO communities are shown by overlapping green and purple circles, respectively. Full size image

The diversity (based on Shannon and Simpson indices) within all the analysed samples and particularly of the inshore ones is small, with the most diverse sample being that of 290 m offshore (Shannon=1.19, Simpson=0.64; Supplementary Table S3). The principal coordinate analysis plot also shows a very close proximity of all the inshore samples (that is, green circles on Figure 5b practically overlap), whereas there is some variance among the offshore samples, as indicated by their good separation along the second principal coordinate, that is, axis 2, explaining 26.9% of the observed variance (triangles in Figure 5b).

However, most of the principal coordinate analysis variance is explained by the first principal coordinate (axis 1, explaining 70.8% of the variance), which is mainly driven by the divergence of the anaerobic methanotroph community (overlapping purple circles on Figure 5b) and, to a lesser extent, by the divergence of two offshore aerobic methanotroph samples (30 and 645 m; blue triangles on Figure 5b). The diversity within the two OTUs of the anaerobic methanotrophs (ETNP_NDAMO1 and ETNP_NDAMO2) is minimal (see Shannon and Simpson indices, Supplementary Table S3). Indeed, phylogenetic analysis placed both of these OTUs into a separate and well-defined cluster, related to the Candidatus Methylomirabilis oxyfera anaerobic methanotroph (Figure 5a).