Seep-carbonate lamination controlled by cyclic particle flux

Authigenic carbonate build-ups develop at seafloor methane-seeps, where microbially mediated sulphate-dependent anaerobic oxidation of methane facilitates carbonate precipitation. Despite being valuable recorders of past methane seepage events, their role as archives of atmospheric processes has not been examined. Here we show that cyclic sedimentation pulses related to the Indian monsoon in concert with authigenic precipitation of methane-derived aragonite gave rise to a well-laminated carbonate build-up within the oxygen minimum zone off Pakistan (northern Arabian Sea). U–Th dating indicates that the build-up grew during past ~1,130 years, creating an exceptional high-resolution archive of the Indian monsoon system. Monsoon-controlled formation of seep-carbonates extends the known environmental processes recorded by seep-carbonates, revealing a new relationship between atmospheric and seafloor processes.


Results and Discussion
Structural build-up properties and dominant carbon source. Numerous millimetre to submillimetre thin convex-up shaped laminae comprise the build-up framework, overall displaying a stromatolitic fabric. Laminae are internally consistent and generally less than 1 mm thick, except for centimetre scale mottled intervals in the middle and top part (Fig. 2a). A total of 204 laminae were counted visually; considering the mottled intervals and potential underestimation due to the limited optical resolution, a 10% error margin is applied to the visual counts (i.e. 204 ± 20). The framework is mainly composed of 13 C-depleted fibrous and clotted aragonite cement (Figs 1d and 2a,b,c; Supplementary Table S1). Exclusively negative δ 13 C aragonite values (− 55.1 to − 41.3‰ VPDB) point to seepage methane (δ 13 C methane = − 70‰ VPDB; ref. 15) as the dominant carbon source and agree with sulphate-dependent anaerobic oxidation of methane (AOM: CH 4 + SO 4 2− → HCO 3 − + HS − + H 2 O; e.g., ref. 17) -the biogeochemical key process inducing carbonate precipitation at methane seeps (e.g., refs 1-4). Inside the build-up, upward flow of gaseous methane was facilitated by the porous carbonate framework (permeability ~1.088 * 10 −9 m 2 ; porosity between 5-16 vol.%; Fig. 3; Supplementary Fig. S2).
Timing of build-up growth. The temporal context for build-up growth is provided by mean U-Th isochron ages which were obtained for the lower (1.13 ± 0.28 ka), central (0.14 ± 0.04 ka), and top part (0.07 ± 0.02 ka; Table 1). Despite the uncertainties inherent to U-Th dating of seep-carbonates 18 , the ages indicate relatively fast  (dive 196) showing four active gas emission sites covered by white and orange patches of microbial mats at the sampling site; arrows point out the domal geometry of bacterial mats caused by carbonate growth below the mats; note the colour difference between black sediment underneath the microbial mats and the surrounding olive-grey seafloor. (c) ROV photograph of the studied columnar carbonate build-up which was not covered by a bacterial mat as the other active methane emission sites (dive 196; sample GeoB12353-11). (d) Colour scan of polished slab revealing internal lamination (reflected light). Scales in (b,c) are approximate.
Scientific RepoRts | 6:37439 | DOI: 10.1038/srep37439 growth (i.e. ~470 mm ka −1 ). Theoretical simulation of seep-carbonate growth showed that the development of a 10-20 mm thick carbonate crust takes about 310 years 4 (i.e. approximately 48 mm ka −1 ). A field study yielded similar values (~50 mm ka −1 ) for methane-derived aragonite 18 . Apparently, the Makran build-up had grown about one order of magnitude faster. Fast growth also agrees with, and was favoured by the high sedimentation rate for the Makran OMZ (~1.2 mm year −1 ) 12 . If the sedimentation rate would have exceeded build-up growth, the structure would have been buried with time 4 . Before methane seepage, and hence carbonate accretion, had ceased for an unknown reason ~70 (± 20) years ago, the build-up had grown above the sediment-water interface.
Seep-carbonate build-ups in the Black Sea basin and in the eastern Mediterranean Sea have grown in a similar fashion in oxygen-restricted environments, causing them to project above the sediment-water interface 1,2 . However, none of the previously described build-ups show regular lamination indicative of a cyclic formation mode. Given that the Makran build-up grew into the water column, AOM-induced carbonate accretion was modified by seasonal sedimentation events, which are common in the Arabian Sea 11,12,19 . Assuming constant methane flow and AOM-induced carbonate precipitation, build-up lamination was likely controlled by cyclic sedimentation events. The reported sedimentation peaks, according to monsoonal particle flux 11,12 , suggests that two laminae formed per year (i.e. the 204 ± 20 laminae were formed during ~102 years). A growth period of ~102 years is, however, is stark contrast with U-Th ages in the lower and middle portions, which indicate much slower growth (Table 1). Assuming constant growth without periods of seepage quiescence, two distinct growth phases can be identified based on the isochron U-Th ages: The lower portion comprises about 102 (± 10) laminae that formed between 1.13 ± 0.28 and 0.14 ± 0.04 ka, corresponding to a growth rate of approximately one lamina every ten years (~1000 years/~102 laminae). In contrast, the remaining 102 laminae formed between 0.14 ± 0.04 and 0.07 ± 0.02 ka, equivalent to approximately one lamina per year (72 years/102 laminae). This indicates a marked increase of build-up growth rate. Alternatively, build-up growth may not have been constant due to periods of seepage quiescence and periods of methane seepage fed from reservoirs below. Changes in methane flux are known from pore water data of sediment cores at the Makran margin, and are potentially triggered by earthquake shaking of the seafloor 20 . Such changes in growth rates, however, cannot be unravelled on the basis of the three isochron ages reported here.
Origin of lamination. The laminated Makran OMZ sediments have been shown to reflect a wide range of different cyclicities, ranging from < 10 to 750 years 12,21 . Interestingly, high-frequency cycles (< 10 years) have been interpreted to reflect monsoonal sedimentation events 12 , which can be overprinted by lunar tide effects 21 . It is therefore likely that the build-up lamination reflects a combination of both, tidal and seasonal forcing, with the latter one becoming more prominent for build-up growth since at least 0.14 ka. It is noteworthy that a lacustrine stromatolite, which was not affected by monsoonal circulation, was found to reveal a similar growth periodicity of one laminae per four to six years over a growth period of ~1000 years 22 . This lamination pattern was interpreted to reflect mainly climate forcing linked to Pacific sea surface temperature and El Niño-Southern Oscillation (ENSO) dynamics 22 . However, the influence of ENSO dynamics on sedimentation patterns of the Makran OMZ was rendered insignificant 12,19 and its influence on the Indian monsoon is not straightforward 23 . We therefore argue that ENSO was not a major factor for the cyclic growth of the build-up, but put forward that the observed cyclicity reflects monsoonal sedimentation patterns.
Coverage by microbial mats or biofilms is considered significant for seep-carbonate build-up formation above the sediment surface. Whereas methane oxidation with oxygen (CH 4 + 2 O 2 → CO 2 + 2 H 2 O) inhibits carbonate precipitation 24 , reducing conditions beneath mats of sulphide oxidizing bacteria supported AOM even in oxygenated waters on the Cascadian margin 25 . Moreover, microbial mats baffle and bind settling particles, thereby contributing to build-up lamination. The main build-up component, fibrous AOM-induced aragonite, is locally covered by brownish organic matter, which is interpreted as remains of microbial biofilms and mats (see Supplementary Fig. S1). LA-ICP-MS analyses revealed variations in elemental compositions between the clear, pure variety of aragonite, and biofilm-covered aragonite (Fig. 2c,d,e; Supplementary Table S2): In contrast to pure aragonite, biofilm-associated aragonite yielded higher silicon, aluminium, magnesium, titanium, and zirconium contents, the latter two being indicative of terrigenous sediment on the Makran continental margin 26 . Similarly, yttrium to holmium ratios of biofilm-associated aragonite (14 < Y/Ho < 30) are similar to upper continental crust material (Y/Ho ≈ 27), but significantly lower than seawater (47 < Y/Ho < 77; e.g., ref. 27) and lower than the ratios found for pure aragonite (49 < Y/Ho < 217).
Although the provenance of terrigenous sediment cannot be deduced from the LA-ICP-MS data alone, it likely derived from the Makran hinterland. During winter (NE) monsoons, enhanced river runoff transports loads of suspended sediment onto the shelf. Subsequently, terrigenous fines are transported farther down-slope by agitation and redeposition 12,19 . Remarkably, winter monsoon irrigation in the Makran hinterland increased within the past 1,000 years 28 , possibly impacting build-up growth within the OMZ. The baffling and binding effect of microbial mats covering the build-up is also apparent from biogenic fines on laminae surfaces. The most abundant constituents were coccoliths of surface-water dwelling calcareous nannoplankton species, including Gephyrocapsa oceanica, Emiliania huxleyi, Calcidiscus leptoporus, Umbilicosphaera sibogae ( Supplementary Fig. S1) Fig. S3); activity ratios were corrected for mass fractionation and spike contributions; n. d. = not determined; δ 234 U (0) represents the measured activity ratios at time zero (today) expressed in delta notation (δ 234 U = [(( 234 U/ 238 U) − 1) * 1,000]). *230 Th/U ages were calculated using the Th230Age spreadsheet function of Isoplot. ** Isochron ages were derived from Osmond isochrons calculated as linear 3-dimensional projection, using the measured activity ratios.
Scientific RepoRts | 6:37439 | DOI: 10.1038/srep37439 is a cosmopolitan species, the other abundant species (G. oceanica, C. leptoporus, and U. sibogae) represent an assemblage typical for nutrient rich surface water conditions during the monsoon 29 . It needs to be stressed that the varved pattern described for Makran OMZ sediments 12,13,19,21 is not observed within the examined build-up. The OMZ sediments comprise alternations of dark (i.e. rich in organic matter and coccoliths, interpreted as winter monsoon deposit) and light layers, interpreted as summer monsoon deposits (rich in terrigenous material). In contrast, the light-dark lamination of the build-up comprises pure authigenic aragonite cement, and aragonite partly covered with biofilm remains that bound and baffled settling particles ( Fig. 2; Supplementary Fig. S1).
Consequently, an attribution of summer or winter monsoon particles based on the compositional data alone remains ambiguous despite the evident influence of particle flux on laminae formation.

Conclusions
In summary, this study reveals that the lamination of a methane-derived carbonate build-up was mainly controlled by cyclic sediment influx. Age constraints on build-up growth indicate that the Indian monsoon possibly in combination with tidal forcing gave rise to the delicate lamination, in concert with carbonate precipitation induced by anaerobic oxidation of methane. This study provides compelling evidence for the influence of the global weather and astronomical forcing on seep-carbonate formation -a previously unrecognized relationship.
In particular, the Indian monsoon is a crucial parameter that steers the formation of laminated carbonates at seeps on the Makran continental margin within the OMZ.

Methods
Seafloor sampling, splitting, and colour scanning. The sample was taken from the seafloor using the manipulator arm of the ROV MARUM-QUEST 4000 m during R/V METEOR expedition M74/3 in 2007 30 . Once on board, the sample was dried and stored for further studies onshore. The build-up was cut in two halves using a wet-cutting rock saw. From one half a ~2 cm thick slab was cut off and embedded in epoxy resin in order to preserve the internal fabric. The slab surface was polished and subsequently scanned with a smartCIS 1600 Line Scanner equipped with a Triple Line Scan Camera SK12240GKOC-LB at 500 dpi resolution. One half was used to prepare subsamples for thin sections and high-resolution X-ray computer tomography (HRXCT).

Mineralogy, petrography, and stable isotope analysis.
Semi-quantitative X-ray diffraction analysis was performed on one bulk-rock powder sample obtained with a handheld micro drill from the cut surface, using a Phillips X'Pert Pro MD X-ray diffractometer equipped with a Cu-Kα -tube (λ = 1.541; 45 kV, 40 mA). Large thin sections (100 × 150 mm) were prepared from cut slabs and examined using standard petrographic microscopy on a Zeiss Axioskop 40 A microscope. Carbonate carbon and oxygen stable isotopes were analysed on powders reacted with 100% phosphoric acid at 75 °C. The released carbon dioxide was injected into a Finigan MAT 251 mass spectrometer. Carbonate δ 13 C and δ 18 O values are reported in per mill (‰) relative to the Vienna PeeDee Belemnite (VPDB) standard. Reproducibility was checked by replicates of an internal standard calibrated against the NBS19 standard. Long term standard deviation for carbonate δ 13 C and δ 18 O analyses is less than 0.05‰ (δ 13 C) and 0.07‰ (δ 18 O).
Digital and scanning electron microscopy. Subsamples of individual laminae were taken using a stainless steel tweezer and photographed with a VHX-5000 digital microscope under reflected light at ZMT (Bremen, Germany). The surface of individual laminae was scanned with a ZEISS field emission scanning electron microscope (University of Bremen, Germany) at 3.0 kV beam current and ~5 mm scanning distance.
Thin section laser ablation inductively coupled plasma mass spectrometry. Laser ablation analyses were carried out at the Geoscience Department, University of Bremen, using a NewWave UP193 laser connected in line to a ThermoFinigan Element2 inductively coupled plasma mass spectrometer. The laser was operated at a 5 Hz pulse rate at approximately 1.3 GW/cm −2 irradiance, using a beam diameter of 100 μ m. Samples were analysed for 90 seconds, using low mass resolution mode. For each spot analyses, a pre-ablation step with 120 μ m beam diameter (approximately 0.3 GW/cm −2 , 1 Hz) was done to avoid surface contamination. Data were calibrated against the NIST610 glass standard with contents reported by 31  Analytical procedures for U-Th measurements and isochron dating. Fragments of individual layers were sampled with a stainless steel tweezer and pulverized using agate mortar and pestle. Between 14 and 93 mg of material was dissolved in 7.5 NHO 3 before adding 200 μ l of a mixed 236 U/ 229 Th spike and subsequent digestion in concentrated HNO 3 on a hot plate. Dissolved U and Th were concentrated using Fe-oxyhydroxide co-precipitation and isolated by ion exchange chromatography. Uranium and Th concentrations and isotopes were analysed with the Neptune multi collector-inductively coupled plasma-mass spectrometer at the Pôle Spectrométrie Océan, IFREMER. Measured isotope ratios were corrected for mass discrimination by standard bracketing, using IRMM-184 (U) and IRMM-035 (Th) standard solutions. Precision of 238 U/ 234 U and 229 Th/ 230 Th measurements was generally better than 50‰ and 5‰, respectively. Apparently co-genetic laminae were dated (Table 1) with an isochron approach; ages were calculated in three dimensional mode using the Isoplot software (Version 3.75; ref. 33). Seep-carbonates contain variable amounts of detrital material (e.g., clay minerals), representing significant inherited 230 Th (detrital fraction) in addition to the 230 Th (authigenic fraction) from uranium decay within the carbonate lattice. The application of isochron dating methods helps circumvent this problem. Corrected isochron ages can be calculated by analysing multiple sub-samples of the same age, comprising various proportions of carbonate and detrital fractions. It was assumed that the sub-samples collected from the upper (n = 3), middle (n = 2), and lower (n = 2) parts of the build-up are co-genetic, and were thus used to calculate distinct average isochron ages for each part. The measured activity ratios ( 232 Th/ 238 U, 230 Th/ 238 U, and 234 U/ 238 U) and the nuclide half-lives ( 238 U, 234 U, 232 Th, and 230 Th) of ref. 34 were used for calculations. Individual sample 230 Th/U ages were obtained with the 230Th/U Isoplot spreadsheet function (Table 1). Because the error of the isochron age using all activity ratios are unrealistically small, an error margin of ± 25% was applied which corresponds to the external U-Th isotopic measurement reproducibility obtained from repeated analyses of an in-house seep-carbonate standard (see refs 2, 35).
Quantitative porosity measurements. Before cutting the build-up it was scanned using the medical X-ray computer tomography (CT) facility at the hospital Klinikum Bremen Mitte (Bremen, Germany). The X-ray source voltage was 120 kV at a current of 600 mA. Scan resolution was 1 mm in x-y-z directions. The CT data were processed with the Amira software (VGS, Burlington, U.S.A.). Two materials, i.e. pore space and carbonate, were defined by using the Amira semiautomatic grey-scale segmentation editor. Porosity was quantified for each CT-scan by calculating the percentile area of pore space per tomographic slice.
High-resolution CT scan and gas flow simulation. A representative subsample was cut from the epoxy-resin embedded working half for high-resolution X-ray computer tomography (HRXCT) scanning with the CT-Alpha scanner (ProCon X-ray, Garbsen, Germany) at the Institute of Geosciences, Johannes-Gutenberg-University Mainz, Germany. The device is equipped with a micro focus X-ray tube featuring a diamond coated target and a 2048*2048 pixels detector. A three dimensional tomogram was assembled from multiple 2D projections with the Amira software. Based on the 3D tomograms the gas flow through a representative segment from the upper central part of the build-up towards the top was simulated using the GeoDict software. The permeability for the segment was 1.088*10 −9 m 2 . Open pore space was defined empty (permeable) and carbonate layers as filled space (impermeable). Pore filling material was characterized with the properties shown in Supplementary Table S4. Gas inflow was limited to the bottom part of the simulation, applying an external overpressure of 0.35 bar. Outflow towards the sides from open pores was blocked by digitally coating the outer surface with impermeable material (= filled pore space).