Manganese enrichments near a large gas‐hydrate and cold‐seep field: a record of past redox and sedimentation events

The spatial distribution, mineralogy, and origin of manganese enrichments surrounding a large gas hydrate and cold seep field (Mississippi Canyon 118, Gulf of Mexico) are investigated in this study, to better constrain their biogeochemical context in deep‐sea sediments and to assess how gas hydrates may alter such records. Manganese depth profiles from 10 sediment cores, documented using centimetre‐scale X‐ray fluorescence core scanning, display highly‐enriched 1 to 10 cm thick layers. These manganese‐rich layers are more numerous, but of lower concentration, in close proximity to the field, and show no consistent relationship with sedimentology (clay vs. carbonate content) or the established chronostratigraphic framework at the site. X‐ray diffraction and sequential dissolution procedures indicate that the manganese enrichments are authigenic carbonates, which formed along a palaeo redox boundary during periods of prolonged steady‐state conditions. The hypothesis that spatial heterogeneity of this manganese record is linked to the nearby gas hydrate and cold seep field, by influencing redox conditions and/or sedimentation processes, is investigated here. Results are consistent with more frequent interruption of steady‐state sedimentation in closer proximity to the salt‐tectonic induced bathymetric mound, which contains the active cold seeps and gas hydrate deposits. Thus, spatial mapping of manganese enrichment horizons provides a tool to reconstruct sedimentation surrounding these volatile sea bed features, yielding a measure of past activity of gas hydrates and cold seeps.


INTRODUCTION
Marine gas hydrates have received much attention due in part to the vast amount of carbon contained within these deposits (Kvenvolden, 1988;Dickens, 2001;Milkov, 2004), and their potential linkages to slope destabilization and past climate events (Nisbet, 1990;Paull et al., 1996Paull et al., , 2003Haq, 1998;Maslin et al., 1998Maslin et al., , 2004Mienert et al., 2005). These aspects have motivated detailed biogeochemical and geophysical investigations of modern gas hydrates (Sassen et al., 2004;Castellini et al., 2006;McGee, 2006;Brunner, 2007;Lapham et al., 2008Lapham et al., , 2010McGee et al., 2009;Macelloni et al., 2012Macelloni et al., , 2013Simonetti et al., 2013;Feng et al., 2014;Martens et al., 2016), and the development of palaeoceanographic proxy approaches to assess the stability of hydrates in Earth's past (Dickens et al., 1995;Dickens, 2003). Unique opportunities to understand marine gas hydrates are provided by integrated studies that seek to link modern and palaeo-perspectives, through the investigation of detailed stratigraphic, sedimentological and geochemical records surrounding present-day gas hydrates (Castellini et al., 2006;Brunner, 2007;Ingram et al., 2010Ingram et al., , 2013. Late Quaternary sedimentation and detailed geochemical records are considered in this study, through an analysis of deep-sea cores recovered from the first National Gas Hydrate Seafloor Observatory (McGee, 2006), located on the northern Gulf of Mexico slope within offshore Federal Lease Block Mississippi Canyon 118 (MC118). A special emphasis is placed on the genesis and interpretation of anomalous Mn enrichments ('Mn-layers') that occur at the MC118 sitedocumented for the first time in this study. These Mn-rich layers are investigated for their potential to reconstruct sedimentation and biogeochemical changes at the site, including linkages to the gas hydrate field. The Mn enrichments are revealed in exceptional detail through high-resolution elemental analysis (centimetre-scale) using X-ray fluorescence (XRF) scanning techniques (Richter et al., 2006), applied to a network of 10 deep-sea sediment cores across the MC118 site. The Mn-layers are further characterized via chemical dissolution/extraction procedures and X-ray diffraction (XRD) analysis. This suite of analytical approaches allows a detailed assessment of the spatial distribution of the Mn-layers at the MC118 site, determination of their mineral composition, and evaluation of possible linkages to past redox cycling and/or sedimentation processes. Interpretation of the Mn record is aided by a well-established stratigraphic framework, which includes previous chronostratigraphic (Ingram et al., 2010) and sedimentological (Ingram et al., 2013) studies of the same suite of cores.
Based on the analyses outlined above it is postulated that the Mn-layers at MC118 record the duration of steady-state redox conditions, along a palaeo-redox boundary between oxic and post-oxic sediments, and that movement of this boundary is linked to the gas hydrate and cold seep field through its influence on sedimentation. As will be shown, the distribution and concentration of Mn within the discrete Mn-layers is consistent with more frequent interruption of steady-state conditions closer to the field. This observation reveals a potential new method to evaluate past redox/sedimentation conditions across a wide range of time scales, through detailed characterization of solid-phase Mn profiles, by employing XRF scanning to 'map' movement of past redox boundaries. In addition, results presented here indicate that the MC118 field has influenced sea floor morphology and sedimentation, but has not contributed to a catastrophic slope failure during the last 14 000 years, consistent with prior studies (Ingram et al., 2010).

The MC118 site
The offshore MC118 site includes a large sea floor mound with active cold seeps and gas hydrates (Fig. 1), and is the focus of ongoing geophysical and geochemical monitoring (McGee, 2006;Brunner, 2007;Lapham et al., 2008Lapham et al., , 2010McGee et al., 2009;Macelloni et al., 2012Macelloni et al., , 2013Simonetti et al., 2013). The field itself is centred at 28Á8523°N and 88Á4920°W and lies at approximately 890 m water depth completely within the offshore federal lease block (Fig. 1). Regionally, the northern Gulf of Mexico slope, including this study site, is influenced by salt diapirism beneath the sea floor (Diegel et al., 1995;Jackson, 1995;Galloway et al., 2000). The field is underlain by a salt diapir 200 to 300 m below the sea floor (mbsf), which contributes to the formation of the sea floor mound itself Sleeper et al., 2006). Moreover, the cold-seep mound is the epicentre for the migration and release of hydrocarbons from the sea floor. This supply of hydrocarbons (natural gas and petroleum) supports active biological seep communities  and microbial chemolithotrophy in the vicinity of the active seep. The upward migrating hydrocarbons are rapidly cycled and oxidized in shallow sediments, also documented at similar sites in the Gulf of Mexico (Castellini et al., 2006). While intense redox cycling occurs immediately over the MC118 mound, it decreases markedly in sediments outside the field (Lapham et al., 2008). This is important, as the cores used in the present study are positioned mostly outside the area with active seeps, with the exception of Core PEL-15 ( Fig. 1). Visible outcroppings of gas hydrates, faulted carbonate 'hardgrounds', authigenic carbonates and pockmark features are present in places across ca 1 km 2 of the sea floor in the vicinity of the mound Sleeper et al., 2006;Feng et al., 2014). Marine gas hydrates are a major feature at this research site, and worldwide they represent a massive reservoir of light hydrocarbons (Milkov, 2004). The vast majority of this carbon occurs in shallow deep-sea sediments along continental slopes (Ginsburg, 1998), such as at MC118. In the Gulf of Mexico, gas hydrates often form alongside cold seeps and supply hydrocarbons to the sea floor, driving substantial changes in both pore-water and sediment chemistry through various redox processes (Sassen et al., 2004Paull et al., 2005;Castellini et al., 2006). The present work documents authigenic Mn in deep-sea sediments complicated by nearby gas hydrates and cold seeps, and the cores investigated in this study reflect ambient sea floor conditions as well as those influenced by the mound (Fig. 1).

Core collection and processing
A total of 10 gravity cores were collected from surface ships, five onboard the R/V Hatteras in August, 2007 and five by the R/V Pelican in April, 2008 (Sleeper & Lutken, 2008;Ingram et al., 2010;Fig. 1). Cores were transported to and processed at a shore-based laboratory at the University of North Carolina, Chapel Hill, where they were prepared for geochemical analyses (see Ingram et al., 2010, for detailed description of coring and processing methods).
X-ray fluorescence (XRF) core scanning Cores were analysed using an Avaatech-XRF scanner (2nd generation) with a rhodium target X-ray source, and a Canberra X-PIPS detector. Continuous down-core XRF scanning was conducted at a resolution of 1 cm along the archive-half split core surface (Richter et al., 2006;Ingram et al., 2010). Manganese was scanned using a 10 kV acceleration voltage, 1000 mA, with a cellulose filter, and 90 sec measurement time, with duplicate scans every 10 cm, which yielded highly reproducible results ( Table 1). The XRF scanning 'count' data are calibrated (least-squares linear fits; Table 1) to Mn concentration data for selected cores (HAT-03, PEL-04, PEL-07; 54 samples), using Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES; SGS Laboratory method ICP-AES 40B, see Table 1 for error analysis).

Reductive and acidification dissolutions
Five Mn-rich layers were sampled from Core PEL-07 to operationally determine the mineral phases associated with the Mn enrichment. Sediment from each depth interval (Mn-layer) was first pulverized into 100-mesh powder. The procedure from Chun et al. (2010) was then Table 1. XRF-ICP Mn Calibration Data: Calibration equations (leastsquares linear fits) relating concentration from ICP-AES method to counts from X-ray fluorescence method, Pearson correlation coefficients (r 2 value) and number of samples used for calibration. Average coefficient of variation (CV) ratios for Mn XRF counts (XRF CV) are based on duplicate scans analysed every ca 10 cm, and select duplicates from ICP-ES sent to SGS Labs (SGS Laboratory method ICP-AES 40B). Average errors (CV) for XRF counts of Mn are slightly higher than other elements previously measured (Ingram et al., 2010), likely a result of lower concentration of Mn in these marine sediments. used to remove oxides and oxyhydroxides by reductive dissolution, followed by an acid dissolution step (sodium acetate, pH = 3Á96) to remove carbonates, with the remaining residue comprising 'insoluble' clays and/or fine-grained silicates. Manganese and Ti concentrations were measured after each step, using ICP-AES (SGS Laboratory method ICP-AES 40B) on the same split sample from each fraction of the procedure, allowing estimation of: (1) untreated (total Mn), (2) Mn oxide, (3) Mn carbonate, and (4) insoluble residue, which represents the unreactive aluminosilicates (clay content) and/or silicates (Schenau et al., 2002;Chun et al., 2010). Manganese enrichment factors (EF), EF = (metal/ Ti) sample /(metal/Ti) crust were calculated to normalize Mn to Ti, a conservative detrital input. 'Excess' Mn (authigenic) was calculated using the following equation [Mn excess = Mn total À (Ti sample *(Mn/Ti) crust )] from Chun et al. (2010), and the bulk crustal Mn/Ti ratio (mol/mol) of 0Á156 from Rudnick & Gao (2003). Enrichment factors and excess Mn calculations are a means to correct for inadvertent loss of mass during the sequential extraction procedures. These values are used to better constrain the fraction of Mn contained in carbonates versus oxides (Table 2).

X-ray diffraction (XRD) for select Mn-layers
A total of five Mn-layers from Core PEL-07 were analysed by X-ray diffraction. Powder samples were analysed using a Rigaku Rapid II X-ray diffractometre with a 2D detector and Molybdenum Ka radiation operated at 50 kV, 50 mA. Samples were ground to a fine powder and then mounted within capillary glass tubes, which yields improved XRD spectra with lower background compared to standard mounts.

Stratigraphy and XRF scanning results
Ten shallow gravity cores collected around the MC118 field comprise 38Á6 m of total recovered sediment ( Fig. 2), and reveal a detailed Mn record documented by centimetre-scale XRF core scans. The occurrence of Mn-layers is described in the context of the previously established chronostratigraphy at the site. Detailed sedimentological, stratigraphic and chronostratigraphic information can be found in Ingram et al. (2010Ingram et al. ( , 2013) (For the chronostratigraphic framework, see figs 3-6 in Ingram Table 2. Mn concentrations from ICP-AES determined for sediment samples from authigenic Mn-layers within Core PEL-07. Sediment samples were selected from 5 depths, and subjected to a two-step chemical dissolution to first remove oxide/oxyhydroxides, and then carbonates (see Material and Methods). Mn enrichment factor (Mn-EF) and 'excess' Mn (E. Mn) is also determined for each layer (see Methods; Chun et al., 2010), and is not applicable (NA) for insoluble clay, as 'excess' Mn is defined as the amount which exceeds the siliciclastic fraction, predominantly clay here. The insoluble (clay) fraction is the remaining material after dissolutions and is fine-grained, consisting of clay with minor silts possible. The sample was measured for Mn concentration following each step, and for untreated samples. The Mn concentrations for oxides, carbonates and insoluble (clay), are inferred using the following scheme: oxides = untreatedstep 1; carbonates = step 1step 2; insoluble (clay) = step 2. The percentage of the total (far right column) is a ratio of Mn in each fraction, oxides, carbonates and insoluble (clay), relative to the untreated sample or total Mn.

Core
Depth Dissolution  , 2010). Additionally, the previously established stratigraphy is also provided here for visual context of the sedimentological units (Fig. 2). Shallow sediments of Unit I are informally broken into Units IA and IB to better describe the occurrence of Mn-layers within the larger stratigraphic Unit I (Fig. 2). Sediments of Unit IA are late Holocene in age (ca 2300 calendar years BP to present) calcareous nannofossil silty clays, and are laterally discontinuous across the study area. Sediments of underlying Unit IB are also calcareous nannofossil silty clays, but are more nannofossil-rich than the overlying Unit IA. Across all studied cores, the more calcareous Unit IB is generally thicker than Unit IA (Fig. 2), and accumulated over a longer time period (ca 2300 to 9500 calendar years BP). Sediments of Unit IB are also lighter in colour and exhibit fewer sedimentary structures (Ingram et al., 2010).
With these sedimentological units defined, the occurrence of Mn-layers is presented in stratigraphic context (Figs 2 and 3). The shallowest Unit IA exhibits very few Mn-layers, with some exceptions. Core HAT-04 displays a shallow Mn-layer completely within Unit IA, and to a lesser extent PEL-04 shows an increased Mn concentration at the top of the core (Fig. 3). With the exception of core HAT-04, all other recovered cores lack discernable Mnlayers within this shallowest stratigraphic interval. Unit IB below also contains few authigenic Mn-layers, with one notable exception, Core HAT-03, which displays two nearly synchronous layers at the base of the unit with Mn content well above 'background' levels ( Fig. 3).
Unit II beneath the shallower carbonate-rich interval (Fig. 2) is a mottled, hemipelagic nannofossil silty clay (Ingram et al., 2010). Sediments here are markedly more clay rich with substantially lower carbonate content. This unit appears visually darker with sedimentary structures that are disturbed by burrows in places. Sediments are early Holocene to late Pleistocene in age ranging from 9500 to 14 000 calendar years from the top to the base of the unit (Fig. 2). This interval contains more Mn-layers than any other unit (at least 15 identified layers) with highly pronounced layers in Cores HAT-01, -05 and PEL-02 (Fig. 3). XRF scans of these same three cores also display less numerous Mn-layers, and in places exhibit only one or two, while cores that contain many layers generally yield lower concentrations for the individual layers. Calibration of XRF data indicates these 'more numerous' layers are ca 1000 p.p.m. Mn less concentrated than the single or double highly discrete layers (Fig. 3).
A continuation of clay-rich sediments comprises Unit III, a well-laminated hemipelagic nannofossil silty clay, which is differentiated from overlying sediments by numerous red-brown layers (Fig. 2). Sediments of this lowermost unit are chemically and lithologically similar to the unit above, yet with slightly higher lithogenic inputs (Ingram et al., 2013). The top of this unit is marked by a prominent red-brown band ('red band'), also observed by previous investigators Brunner, 2007;Sleeper & Lutken, 2008;Ingram et al., 2010). This marker bed is coincident with reworked pre-Quaternary nannofossils deposited during Melt Water Pulse 1A (MWP-1A; Marchitto & Wei, 1995), and is dated between 14 000 and 15 000 calendar years BP (Ingram et al., 2010). The unit lacks a defined base, and thus extends to the greatest depth recovered. Based on accumulation rates and radiocarbon dating (Ingram et al., 2010), sediments are latest Pleistocene in age and certainly older than the MWP-1A reworked nannofossil horizon, which defines the top of the unit (Fig. 2).
Manganese profiles spanning units II and III may appear correlative, such as in Cores PEL-04, -08 and -07, yet upon close examination, Mn layers are not chronostratigraphically equivalent (PEL-04, PEL-07 and -08; Fig. 3). It is also apparent that Unit III mostly lacks the highly concentrated Mn-layers, with one exception in Core PEL-08 (near the top of the unit; Fig. 3), yet recovery of this unit was incomplete using shallow gravity coring. The Pelican transect recovered more of Unit III than the previous cruise and yields several metres of late Pleistocene-aged sediments in some cores. While this unit lacks the discrete highly concentrated Mn-layers, it does display many 'minor' layers (Fig. 3).

Mn-layer mineralogy
A multi-step reductive dissolution and acid digestion procedure was used to operationally determine probable mineral phases associated with the authigenic Mn ( Fig. 4; Table 2). From five sediment samples taken within the Mn-layers, just over 80 percentage of the total Mn content is comprised of Mn-rich calcite (Ca-Mn(CO 3 ) 2 ) and/ or rhodochrosite (MnCO 3 ) ( Table 2). Manganese oxides account for slightly less than 10 percent, and the 'insoluble' clay fraction accounts for ca 10% (Table 2). Thus,  Table 1 for calibration of XRF counts to ICP-AES Mn concentration; all other cores are presented here as XRF counts. Down-core profiles display Mn counts (1000x) with lines drawn across the profiles representing stratigraphic boundaries. The black lines separate Unit I from Unit II, which mark a substantial shift in carbonate content. The thin-dashed blue line separates informal units IA and IB, defined by a change in carbonate content (Ingram et al., 2010(Ingram et al., , 2013. The light-red line is the depth to the 'red band', a chronostratigraphic marker (Ingram et al., 2010), which defines the top of Unit III. Mn within the concentrated layers is predominantly associated with carbonates, followed by aluminosilicates and minor amounts of oxides. Some change in concentration may be an artefact of the total change in sample mass associated with dissolution of carbonates. This is a concern for sediments with very high carbonate content, however, the samples investigated here (selected for sequential dissolution in PEL-07) are from clay-rich Units II and III with less than 13 wt.% CaCO 3 on average. To more rigorously address this issue, Mn enrichment factors and 'excess' Mn are also calculated (Table 2). Manganese enrichment and 'excess' Mn (defined as the non-clay fraction) are largely associated with carbonates and not oxides (Table 2). This is shown by determining the percentage of 'excess' Mn in each respective fraction (oxide verses carbonate) where the clay fraction has already been removed (via calculation of 'excess' Mn). The following expression: ððuntreated excess À carbonate excess =untreated excess ÞÃ100Þ yields only 6Á8% of Mn as potentially associated with oxides or oxyhydroxides, based on the average of all five layers (Table 2). Hence, 'excess' Mn from within the Mn-layers is dominantly associated with carbonates (over 90% on average). Six untreated samples were also analysed using XRD (PEL-07-layer1, PEL-07-layer2, PEL-07-layer3, PEL-07-layer4, PEL-07-layer5 and PEL-04 at 5 cm depth; Fig. 5). All six XRD diffraction results show the existence of the following phases but in different proportions: quartz, calcite, feldspar, dolomite, rhodochrosite and clay minerals. Manganese is present as carbonate or Mn-Fe carbonate, which may form a complete solid solution (Fig. 5). Manganese carbonate in most samples has a d 104 peak at ca 2Á82 A, except sample PEL-07-layer 4 that has a value of 2Á827 A. The d 104 peak for the pure Mn end member (rhodochrosite) is 2Á84 A, and that for the pure Fe end member (siderite) is 2Á79 A. The d 104 of the samples indicates a Mn-rich (>50% Mn) carbonate. One acid treated sample taken from PEL-07 (Mn-layer3; Fig. 5B) yielded no carbonates in the diffraction pattern; this result indicates that the dissolution procedure used here and by Chun et al. (2010) effectively removes Mn-carbonate minerals.
To summarize, the X-Ray diffraction and reductive dissolution/acidification results reveal that Mn within the Mn-layers is largely contained in the carbonate phase. Given sample heterogeneity, preferred orientation and different proportions of mineral phases, intensities of the rhodochrosite d 104 peak is not expected to exactly correlate with the XRF/ICP-AES derived Mn concentrationfor example, Mn-layer3 is not the most concentrated based on the XRF profile (Figs 4 and 5). However, Mnlayers 1, 2 and 3 exhibit larger rhodochrosite peaks than Mn-layers 4 and 5 (Fig. 5), which is generally consistent with Mn concentrations from XRF and ICP-AES ( Fig. 4; Table 2).

Manganese carbonate layers at MC118
A detailed deep-sea sedimentary record is documented at MC118 in an effort to better understand accumulation of highly enriched Mn-layers and possible mechanisms for their formation. Foremost, it is clear that Mn-layers occur in sediments of different age and generally do not follow lithostratigraphic or chronostratigraphic trends. Hence, their characterization as authigenic deposits, which formed in situ through redox processes (Figs 2 and 3), representing diagenetic palaeo-redox horizons. Multi-step chemical dissolution procedures combined with XRD analyses confirms that the Mn-layers are primarily carbonates.
The documentation of such a highly detailed spatiotemporal record of recurring authigenic Mn carbonate layers as observed at MC118 is somewhat unique considering water depth and proximity to the seabed with nearby cold seeps (Canfield et al., 1993;Chun et al., 2010). However, Mn enrichments are often observed in ancient marine sediments or mudstones, along with other elements associated with authigenic deposits, and have been linked to changes in past climate or oceanographic conditions (Dickens & Owen, 1994;Schenau et al., 2002;Tribovillard et al., 2006). The MC118 results indicate that substantial variability in Mn enrichment is present in Recent (Late Pleistocene-Holocene) deep-sea sediments over a relatively small area of the sea floor (ca 2 km 2 : Fig. 1). This has important implications for interpreting chemostratigraphic records from similar deep-sea depositional environments, as it implies that modern and ancient sedimentary Mn records may reflect localized conditions from depositional heterogeneity, rather than more widespread regional or 'global' events. Thus, subtle variation in depositional or redox environments in deepmarine continental slope sediments may drive noticeable differences in the accumulation of authigenic minerals. The MC118 record shows that authigenic Mn can vary considerably within continental slope sediments over a small area of the sea floor, and suggest linkages to the local gas-hydrate and cold-seep field. This includes sea  Fig. 4). The mineral diffraction peaks labelled from left to right are quartz, albite, calcite, dolomite and rhodochrosite. The rhodochrosite diffraction peak is present in all samples (Mn-layers), and is largest (most concentrated) in Mn-layer 3. (B) X-ray diffraction spectra for Mn-layer 3, before acid dissolution (black curve), and after acid dissolution (red curve). The other diffraction peaks appear larger, as the carbonate fraction was removed by the acid dissolution, thereby concentrating siliciclastic minerals in the treated sample.

Biogeochemistry of the Mn-carbonate layers
Nearly all of the Mn-layers occur at depths well below expected oxygenated pore waters for this depositional setting, consistent with a Mn-carbonate phase preserved in post-oxic sediments. Previous studies have demonstrated that oxygen depletion occurs within 5 cm of the sea floor for similar settings such as the western Gulf of Mexico Shelf (Hu et al., 2011) and the Mississippi Canyon region (Diaz & Trefry, 2006). More recent studies have reported Mn-enrichments (oxides) within 10 cm of the sea floor at similar water depths and close to the MC118 site (Brooks et al., 2015;Hastings et al., 2016). In contrast, most Mn layers at MC118 are found more than a metre below the sea floor, thus they are considered much too deep for active formation of Mn-oxides, which develop near the oxygen depletion depth (Burdige & Gieskes, 1983;Kalhorn & Emerson, 1984;Heggie et al., 1986;Aller, 1990Aller, , 1994Shaw et al., 1990;Reimers et al., 1992). However, it should also be noted that Mn carbonates can develop a protective crust around oxides formed earlier, thereby diminishing their dissolution within post-oxic pore waters (Burdige, 1993). This protective crust may account for the small fraction (under 10%) of Mn-oxide observed in some layers (Table 2).
Repeated Mn cycling (the 'Mn pump ';Sageman & Lyons, 2003) can concentrate dissolved Mn within alkaline pore waters to form Mn-rich carbonates. Thus, Mn enrichment may form along palaeo-redox boundaries, related to the carbonate-Mn pump, where dissolved Mn concentrated in pore waters leads to precipitation of Mncarbonate minerals under sufficiently alkaline conditions. Mn-oxide dissolution may occur contemporaneously with formation of Mn-carbonates just beneath an active redox front, followed by subsequent burial and preservation of the palaeo-redox boundary (Pedersen & Price, 1982;Burdige, 2006). The rapid burial of highly concentrated Mnoxides into deeper dysoxic/anoxic and alkaline pore waters, due to a (pulsed) increase in sedimentation rate, would also promote the preservation of Mn-carbonate layers. More generally, sedimentation can influence redox processes through erosion and/or deposition.
Other factors that drive the movement of the Mnredox boundary include, but are not limited to, (1) changes in organic carbon delivery and reactivity, (2) bottom water oxygen content, or (3) intensity of bioturbation/bioirrigation. All of the above factors influence preservation of 'relict' Mn-peaks (Burdige, 2006). Most of these factors are not explicitly constrained in this study, although previous efforts quantify sedimentation rates, and organic matter accumulation/composition (Ingram et al., 2013). There is a shift towards more reactive organic matter (Type II) within the shallow Holocene sediments, which is expected to drive the redox boundary upwards, yet this is also balanced by slower sedimentation rates (Ingram et al., 2010). Regardless, there are too many discrete Mn-layers to explain with one stepwise change in sedimentation. Furthermore, while organic matter accumulation is variable with time (Ingram et al., 2013), it should be more uniform spatially in a deep-sea settingunless it is also influenced by the presence of the seeps. Thus, dynamic sedimentation associated with the mound is hypothesized as a more viable driver to explain spatiotemporal trends of the Mn-peaks across the network of MC118 cores, which suggest the field's influence.
Mapping Mn-carbonate layers: a tool to delineate palaeo-redox conditions The analysis presented here suggests that the high-resolution Mn record at MC118 primarily reflects the duration of steady-state conditions and the frequency with which they are interrupted, through changes in sedimentation that are driven by local effects from the presence of the elevated bathymetric mound. Relatively more stable sedimentation is expected for core sites farther away from the area of active cold seepage and gas hydrate formation (centred over the bathymetric mound). Such distal sites should be characterized by prolonged periods of steadystate conditions, and hence a more stable palaeo-redox boundary along which more concentrated Mn-layers can form. Conversely, frequent interruption of steady-state conditions is expected in closer proximity to or downslope from the field yielding less-concentrated and morenumerous Mn-layers (Fig. 2). The distribution of Mn-layers at the site is consistent with this hypothesis, as only one or two very-concentrated authigenic Mn-layers (Cores HAT-01, HAT-05 and PEL-02) are observed at locations distal from the field, while a greater abundance of less-concentrated layers are observed closer to the field (Core HAT-03 and HAT-04, Fig. 3).
Authigenic Mn enrichments are well-documented in marine sediments (Burdige, 1993(Burdige, , 2006, however, it is the remarkable detail of the MC118 record in both time and space that makes this study unique, and reveals past movement of the palaeo-redox boundary at various core sites. This record is made possible through the application of high-resolution XRF scanning techniques. The integration of such spatio-temporal data with quantitative biogeochemical models for Mn-cycling should provide a powerful new tool for evaluating redox cycling and/or sedimentation events over a wide range of time scales, for example, quantification of the duration of steady state conditions required to generate a Mn-layer of given concentration (Froelich et al., 1979;Burdige & Gieskes, 1983;Finney et al., 1988;Burdige, 1993;Price, 1998), and deconvolution of local versus regional/global influences on the observed Mn record.

CONCLUSIONS
The results presented here demonstrate that multiple discrete Mn-layers observed at MC118 are authigenic deposits, preserved as carbonates and formed within shallow sediments along a transient palaeo-redox boundary. The Mn-layers occur independently from established lithostratigraphic and chronostratigraphic horizons at the site, and results from sequential extraction procedures, along with XRD analysis, reveal that layers are mostly rhodochrosite (MnCO 3 ). It is not known if the Mn-layers formed initially in shallow sediments as oxides/oxyhydroxides or as carbonates, however, they are presently carbonate minerals preserved within post-oxic conditions.
The spatio-temporal heterogeneity of this Mn record is linked to the MC118 gas-hydrate and cold-seep field, including the related salt diapirism beneath the mound. This sea floor feature and associated release of hydrocarbons in turn influenced palaeo-redox cycling and/or sedimentation processes that impacted the duration of steady-state conditions in the past. Thus, the concentrations and frequency of Mn-layers across the site is a fingerprint of the palaeo-redox boundary and its spatial expression with time. While numerous factors drive changes in redox conditions, it is suggested that variability in sedimentation rate, caused by the presence of the sea floor mound, is a viable mechanism to explain the complexity of the Mn record.