Fluid Accumulation, Migration and Anaerobic Oxidation of Methane Along a Major Splay Fault at the Hikurangi Subduction Margin (New Zealand): A Magnetic Approach

Understanding the locus of fluid flow along thrust and splay faults is important to understand the hydraulic properties of accretionary systems and fault mechanics. Here, we use rock magnetic techniques in combination with backscattered electron imaging to depict the locus of enhanced magnetic mineral alteration within the Pāpaku fault, an active splay fault of the subduction interface at the northern Hikurangi Margin. The Pāpaku fault was cored at Site U1518 during Expedition 375 of the International Ocean Discovery Program and we report room temperature magnetic parameters, complemented by first‐order reversal and thermomagnetic curves in the depth interval 250–400 m below seafloor (mbsf). The ∼60‐m wide Pāpaku fault zone comprises two main slip zones, referred to as the upper main brittle (304–321 mbsf) and lower subsidiary (351–361 mbsf) fault zones, and an intervening zoned, termed the lower ductile deformation zone. Two narrow zones, at the top of the main brittle fault zone, and one in a sand‐rich interval above the subsidiary fault zone, experienced enhanced magnetic mineral diagenesis, which resulted in the recrystallization of ferrimagnetic greigite to paramagnetic pyrite. We propose that secondary magnetic mineral diagenesis was driven by anaerobic methane oxidation within these intervals, which occurs in the presence of methane and sulfate. We relate the observed changes to the fault parallel transport of fluids which is restricted to two damage zones. Overlying compacted and clay‐rich sediments likely act as a barrier to upward advective flow through the fault zone and into the hanging wall.

. Rock magnetic experiments are rapid, and require little sample material only. So, they can be conducted in higher resolution than more conventional geochemical investigations. Next to detrital iron (oxy) (hdyr)oxides, authigenically formed magnetic minerals, particularly greigite (Fe 3 S 4 ), prevail in rapidly accumulated sediment. En route to pyrite (FeS 2 ) formation, the intermediate sulfides (including greigite) may be preserved in case of a favorable iron to sulfur balance. Sufficient sulfide to complete pyrite formation is often not available in rapidly accumulating sediment (e.g., Roberts et al., 2011). The magnetic minerals thus constitute a probe of past temperature and fluid flow because they are sensitive to heating and/or prevailing redox conditions. These conditions are modified during slip and fluid flux (Chen et al., 2019).
In this paper, we discuss meter scale anomalies in the magnetic behavior of sediments sampled throughout the Pāpaku fault zone, which is a frontal splay fault near the deformation front of the Hikurangi Subduction Margin, New Zealand ( Figure 1). The sediments were recovered at Site U1518 during Expedition 375 of the International Ocean Discovery Program (IODP). The site is located in a region that is known to host shallow slow-slip earthquakes (Wallace et al., 2016), although it is yet unclear how shallow thrust faults contribute to the dissipation of seismic energy (Wallace et al., 2016). The results of the present study provide evidence for the migration of methane-bearing fluids along spatially limited pathways, thus providing crucial information needed to better understand the permeability structure, mechanic properties, and evolution of shallow thrust faults.

Site Details and Shipboard Investigations
The Pāpaku fault zone is a roughly westward dipping, shallow subduction thrust which traces Earth's surface only a few kilometers away from the deformation front of the northern Hikurangi subduction margin in north-east New Zealand (Figure 1a). It roots into the main plate interface décollement ca. 2 km further to the west (Figure 1b) (Bell et al., 2014;Fagereng et al., 2019;Saffer et al., 2017;Wallace et al., 2019). Hole U1518F of IODP Expedition 375 (38°51.57'S, 178°53.76'E, water depth: 2,626.1 mbsl) cored hanging wall, fault zone and footwall sequences up to a maximum depth of 492.3 meters below sea floor (mbsf). The average core recovery in the studied intervals is ca. 43%. The core recovery in the fault zone and footwall is somewhat lower (Figure 2a). In these intervals, less cohesive sand-rich intervals and fractured material led to increased destruction of core material during the drilling process. Data sets from shipboard investigations, including geochemical analysis, core description and physical properties measurements provide a wealth of information on fault zone architecture and sediment mechanical properties. Shipboard core descriptions GREVE ET AL.  Barker et al. (2018). The location of IODP Site U1518 and trace of the Pāpaku fault are indicated. Figure modified from (Fagereng et al., 2019;Greve et al., 2020). IODP, International Ocean Discovery Program.
included the determination of porosity ( Figure 2b) and the estimation of the fracture intensity (Figure 2c), which quantifies the density of open fractures within the core-sections recovered (Wallace et al., 2019).
Shipboard investigations also included the analysis of key pore-water solutes, including sulfate (Figure 2d), chloride, and dissolved metal ions (Wallace et al., 2019). For these analyses ca. 20-40-cm long segments were removed from every 5-10 m of core recovered through the hanging-and footwall. Through the fault zone interval one sample every 7-20 m was processed. Pore-water was squeezed out of the sediment using a hydraulic press, and key solutes measured. Additional gas chromatography was conducted on two headspace samples per core to measure hydrocarbon gases including methane (cf. Figure 2e; ca. one sample every 5 m, refer to Wallace et al., 2019 for details).

Lithostratigraphy, Fracture Density, Porosity, and Pore-Fluid Chemistry
Hole U1518F cored the Pāpaku fault zone between 304.5 and 361.7 mbsf. The roughly 60-m thick fault zone forms the active, structural boundary between consolidated and heavily deformed hanging wall sediments which are overthrust over younger and less consolidated footwall sediments (Fagereng et al., 2019;Greve et al., 2020;Morgan et al., 2020;Wallace et al., 2019).  . Shipboard downcore properties. Displayed in (a) are the core IDs and recovery of core-material (black intervals: core recovered); the main structural domains (cf., Wallace et al., 2019). UFZ-upper main brittle fault zone, LDZ-lower ductile deformation zone, LFZ-lower subsidiary fault zone. The shaded intervals in yellow correspond to intervals in which anomalous rock magnetic properties were measured (cf., Figure 4, top: Anomaly A, bottom: Anomaly B). (b) Downcore variations of discrete sample porosity; (c) fracture intensity, an index number which describes the relative abundance of fractures over 10 cm intervals, averaged per core section (1: long unfractured section of core, 5: intensely brecciated zone); (d) sulfate concentration (uncorrected for drilling fluid contamination); and (e) methane concentration. For data, details and methods the reader is referred to Wallace et al. (2019).

(a) (b) (c) (d) (e)
The sediments recovered throughout hanging wall, fault zone, and footwall consist of Pleistocene mudstone sequences that are sparsely interbedded with thin graded silt-and sandstone layers. The frequency of sand-and siltstone beds increases near the bottom of the fault zone and through the footwall, where contorted, sand-rich intervals have been interpreted as gravity driven mass-transport deposits that contain coarser grained clastic and volcaniclastic material (Figure 3f). The lower hanging wall is pervasively deformed. It yields variable bedding tilts and localized zones of overturned bedding. Folding is accompanied with brittle fracturing, and the fracture intensity peaks in a fold hinge zone at ca. 275 mbsf ( In contrast, the deformation decreases throughout an intervening zone between 322 and 351 mbsf, referred to as lower ductile deformation zone. A second zone of higher intensity deformation, here referred to as the lower subsidiary fault zone, was identified between 351 and 361 mbsf (Wallace et al., 2019).
Biostratigraphic investigations identified an age inversion near the top of the upper main brittle fault zone suggesting that the majority of the fault zone interval is located within the younger footwall rock (Wallace et al., 2019). This observation is coherent with the finding of magnetic fabric analyses (Greve et al., 2020), which indicate that strain is decoupled between hanging-and footwall sequences near the top of the upper main fault.

Rock Magnetic Analysis
Samples for rock magnetic analyses were extracted from working-half sections by pushing 7 cm 3 "Natsuhara-Giken" sample cubes into the sediment, while avoiding intervals that were affected by localized deformation or mass transport deposits (i.e., sand, silt-rich intervals, and volcanic ash flows). For this study 63 samples were collected from the interval 250-400 mbsf in Hole U1518F, using a spacing of about 1 m of core recovered. Anisotropy of magnetic susceptibility was measured using an AGICO KLY-4 Kappa Bridge at the Japan Agency for Marine-Earth Science and Technology in Yokosuka (Japan) and mean magnetic susceptibility (χ) extracted. These results were previously reported in Greve et al. (2020). All other magnetic measurements were conducted at the Center for Advanced Marine Core Research in Nankoku (Japan). Samples were first measured for their natural remanent magnetization (NRM), subsequently demagnetized by static alternating fields (AF) up to 80 mT and afterward given an anhysteretic remanent magnetization (ARM) in a direct current (DC) bias field of 50 µT with a peak AF of 80 mT. Acquired GREVE ET AL.
10.1029/2020JB020671 4 of 15 ARM was measured using a 2G SQUID cryogenic pass-through magnetometer. A selection of samples was subjected to partial ARM acquisition (pARM) in AF fields that were increased in 5 mT steps to a peak field of 80 mT, while applying a 50 µT DC bias field.
Further analyses included the measurement of isothermal remanent magnetization (IRM) using a Natsuhara Giken spinner magnetometer following exposure of the samples to a 1.2 T applied field using a MMPM1 pulse magnetizer. At this field samples were considered saturated and the acquired remanence will be referred to as saturation isothermal remanent magnetization (SIRM) in the following text. Subsequently, after the measurements on the cubes, 150-300 mg of material was carefully removed from each cube, covered, dried overnight in air at room temperature and packed into gelatin capsules for additional analyses. Magnetic hysteresis, backfield curves and IRM acquisition curves were measured using a Princeton Measurements Corporation vibrating sample magnetometer with a 1 T maximum field. Hysteresis and backfield remanence parameters (saturation magnetization M s , saturation remanent magnetization M rs , coercivity B c , and remanent coercivity B cr ) were calculated following automated slope correction (saturation is assumed at 70% of the saturating field). These parameters effectively describe the shape of hysteresis loops, which is linked to dominant magnetic mineralogy and magnetic grain-size in a sample. We summarize all four parameters into the ratio D JH = (M rs /M s )/(B cr /B c ), which has been shown to increase with increasing iron-sulfide content in mixed assemblages of iron oxides and diagenetically produced iron sulfides (Housen & Musgrave, 1996).
To further investigate magnetostatic interactions we measured first-order reversal curves (FORCs; Pike et al., 1999;Roberts et al., 2014Roberts et al., , 2000 on a selection of samples from the hanging wall, the fault zone and the footwall sequences. Each measurement sequence included the measurement of 150 FORCs with an averaging time of 100 ms and a field increment of 2 mT. The FORC diagrams were processed using the GREVE ET AL. 10.1029/2020JB020671 5 of 15  (Housen & Musgrave, 1996). (g) The ratio between the anhysteretic remanent magnetization (ARM) divided by the saturation isothermal remanent magnetization (SIRM). (h) The ratio between SIRM and χ.

Backscattered Electron Imaging
Seven samples including two from Anomalies "A" and "B" each, and three samples from the hanging wall, lower ductile deformation zone and the footwall, respectively, were further investigated using backscattered electron (

Downhole Trends in NRM, χ, SIRM, and Hysteresis Parameters
Downcore diagnostic magnetic parameters from discrete sample measurements are displayed in Figure 4. The NRM ranges from 2.1 × 10 −7 up to 1.4 × 10 −5 Am 2 /kg ( Figure 4b) although almost all samples yield NRM intensities around 3.2 × 10 −6 Am 2 /kg (average and standard deviation of hanging-and footwall: 3.21 × 10 −6 ± 3.6 × 10 −6 Am 2 /kg, N = 31), equivalent to 4.9 × 10 −3 A/m when NRM is volume-normalized. The peak intensity was measured on a single sample at ca. 296.3 mbsf. This sample was collected from a more sand-and silt-rich interval, where tectonic deformation resulted in a mixing between hemipelagic sediments and sand/silt-rich gravity deposits. The higher remanence intensity measured on this sample presumably reflects a higher concentration of detrital magnetic minerals. In contrast, in two narrow intervals, between ca. 304-312 mbsf and ca. 334-351 mbsf, sample measurements display remarkably low NRM intensities. These apparent anomalies are reflected in all magnetic measurements discussed throughout this paper. We thus refer to "Anomaly A" for the upper interval between ca. 304-312 mbsf and "Anomaly B" for the lower interval between ca. 334-351 mbsf. All other intervals will be referred to as "base-line" in the remaining discussion.
Magnetic susceptibility χ (Figure 4c) mimics the variations in NRM. It ranges from 5.2 to 13.7 × 10 −8 m 3 / kg, with the lowest values measured within Anomalies "A" and "B," and the peak value measured on the isolated sample taken from the sand-rich interval. The magnetic coercivity (B c ), the coercivity of remanence (B cr ), and the D JH parameter are not affected by the magnetic mineral concentration. Peaks and troughs measured within the anomalous intervals appear much sharper than those recorded by the concentration dependent proxies discussed above (i.e., NRM and χ). Hysteresis loops from the host sediment are open (Figure 5b) with B c values measured on samples from the hanging-and footwall averaging 52.5 ± 4.7 mT (N = 39) (Figure 4d). In contrast, hysteresis loops from Anomalies A and B are much skinnier and display coercivities as low as 13 mT (Figures 5b and 4d). B cr (Figure 4e) averages 70 ± 3.2 mT (N = 39) in hanging-and footwall and shows values as low as 37.9 mT within the anomalous intervals. Hanging wall and footwall samples display D JH (Figure 4f) ratios of 0.39 ± 0.06 (N = 39), while Anomalies "A" and "B" display two distinct troughs with values as low as 0.056. The ARM/SIRM ratio, sensitive to the magnetic grain-size (Q. Liu et al., 2012;Peters & Dekkers, 2003), displays "base-line" values of 4.9 × 10 −4 ± 1.2 × 10 −4 (N = 31) and two prominent peaks at 1.7 × 10 −3 and 3.4 × 10 −3 that correspond to Anomalies A and B, respectively ( Figure 4g). The SIRM/χ ratio is sensitive to grain-size and has also shown to increase with the introduction of greigite in diagenetically altered marine sediments (e.g., Larrasoaña et al., 2007;Musgrave et al., 2019;Snowball, 1991). It mostly mimics the variations in NRM intensity and χ described above (Figure 4h). Spikes in SIRM/χ ratio, which are associated with sand-and/or silt-rich intervals obscure the signal of the anom-GREVE ET AL.

FORC Diagrams, ARM Acquisition, and Thermomagnetic Curves
Differences in the magnetic properties between "base-line" lithology and the two anomalous intervals, Anomalies "A" and "B," are also reflected in FORC diagrams and thermomagnetic curves. Samples from the hanging wall, footwall and most of the fault zone display FORC density distributions that have closed and concentric contours with a coercivity in between 40 and 90 mT; the highest density is at ∼70 mT. This is typical of the behavior often observed in samples that contain single domain (SD) greigite (Fe 3 S 4 ) (Roberts et al., 2006; see also Greve et al., 2020). ARM acquisition curves from these horizons display a wide distribution of coercivities that peak above 40 mT (Figure 5d, see also Greve et al., 2020). Thermomagnetic curves ( Figure 6b) display slight change in slope at heating temperatures between 250°C and 400°C, which, following heating to 350°, is irreversible. Heating curves display a second decrease in slope at temperatures higher than ca. 525°C. The first is probably caused by the progressive decomposition of greigite upon heating to fine-grained magnetite. The latter loss in magnetization is explained by the oxidation of the newly formed magnetite to hematite (Chang et al., 2008;Dekkers et al., 2000;Krs et al., 1990;Roberts, 1995;van Baak et al., 2016;Vasiliev et al., 2008). Samples from Anomalies "A" and "B" display markedly different FORC and thermomagnetic behavior. FORC diagrams are significantly noisier and display narrow contours along the Bc axis that tail up to 50 mT, and diverge along Bu axis in the low coercivity (0-10 mT) range, typical for mixtures of vortex-state and multidomain sized iron-oxide minerals (Pike et al., 1999;Roberts et al., 2014). pARM acquisition curves also indicate a predominance of lower coercivity magnetic mineral phases. All thermomagnetic curves measured on samples from Anomalies "A" and "B" (e.g., Figure 6a) display a noteworthy and irreversible increase in magnetization at temperatures higher than 400°C, followed by a rapid decay in magnetization upon heating to temperatures between 500°C and 580°C. This behavior has been explained as the oxidation of nonmagnetic pyrite to ferrimagnetic magnetite during heating in past literature (Passier et al., 2001;van Baak et al., 2016;Van Baak et al., 2013). Further oxidation of magnetite to hematite during progressive heating results in lower magnetization measured on cooling curves (Figure 6a). A slightly steeper slope on the heating curves between 250°C and 400°C, which would provide evidence for the coexistence of pyrite and greigite, is not visible. This contrasts markedly with "base-line" samples where such an irreversible decay is clearly present (Figure 6b).

BSE Observations
We found framboidal aggregates of iron-sulfide minerals in all samples from BSE observations (Figure 7). In samples from "base-line" lithology iron sulfide occurrence is limited to few, small (<5 µm) circular framboids, which were difficult to identify. Each framboid consists of small (<0.5 µm), octahedral, or irregular shaped iron-sulfide grains which display a bright backscatter (Figures 7a-7c). In contrast, the samples from Anomalies "A" and "B" contain a significantly larger abundance of iron-sulfide minerals. These exist as large (up to 400 µm diameter) polyframboidal aggregates (Figures 7d-7f), in which the individual crystals of the various framboids have different grain sizes. Recrystallization and formation of euhedral iron sulfides with sizes < 5 µm (probably pyrite) is often visible along the rims of individual aggregates (Figures 7h  and 7i). Larger (50-150 µm), irregular shaped iron-sulfide grains (pyrite) which probably formed during progressive recrystallization of preexisting aggregates (Figures 7j-7l). Recrystallization of framboids and formation of euhedral iron-sulfide grains (probably pyrite) in samples from Anomalies "A" and "B" indicate that secondary mineral diagenesis occurred within these intervals. Besides the recrystallization and neoformation of iron sulfides, this observation would have resulted in the alteration of accessory greigite, originally present in "base-line" lithology, as evidenced from thermomagnetic experiments, to pyrite (see, e.g., Roberts & Weaver, 2005, refer also to discussion). The removal of ferrimagnetic greigite from "base-line" lithology during secondary mineral diagenesis, means that the weak NRM in the anomalous intervals is carried most likely by a few residual iron oxides, for example such that exist as inclusions in dissolution-resistant silicate minerals (Just et al., 2012). This finding provides an explanation for the lower NRM and SIRM values in samples from Anomalies "A" and "B" (Figure 4), and the larger effective magnetic grain-size of residual iron-oxides as evidenced from coercivity, ARM/χ (Figure 4), FORC diagrams ( Figure 5) and the predominance of lower coercivity magnetic mineral phases found in pARM acquisition curves.

Downhole Rock Magnetic Variations as a Proxy for Anaerobic Oxidation of Methane Within the Pāpaku Fault
The results of magnetic properties measurements, together with BSE imaging provide compelling evidence that SD sized greigite is the primary remanence carrier throughout most of the studied interval, that is, in the "base-line" sediment. Exceptional are two anomalous intervals (approximately between 304 and 312 mbsf and 334 and 351 mbsf, respectively) within the Pāpaku fault zone that yield a rock magnetic signature distinctively different from the surrounding lithology. This is linked to the enhanced precipitation of iron-sulfide phases (presumably mostly pyrite) and reduction of existing, ferrimagnetic greigite to paramagnetic pyrite in these zones. Alteration of magnetic minerals in shear zones has previously been linked to frictional heating and/or fluid migration (Yang et al., 2020 and references herein). In the following, we discuss possible mechanisms as explanation for the observed changes. Magnetic studies on fault rocks mostly focused on sites that hosted large megathrust earthquakes, such as the Mw 9.0 Tohoku-oki earthquake in 2011 (Yang et al., 2018(Yang et al., , 2016 and the Mw 7.9 Wenchuan earthquakes in 2008 (D. Liu et al., 2016), where localized frictional heating temperatures on the principal slip zone are argued to have exceeded 500°C.
Authigenic greigite and pyrite, which were identified here, break down at temperatures as low as 250°C-350°C and 500°C, respectively (Yang et al., 2020). They usually form at low burial temperatures following early sediment deposition at the sulfate-methane transition zone (SMTZ) where the reaction between dissolved sulfide (mostly from anaerobic methane oxidation [AOM]) and iron (either dissolved or not) results in the progressive reduction of iron-oxide minerals to paramagnetic pyrite (FeS 2 ) (see, e.g., Chen et al., 2019;Fu et al., 2008;Kars & Kodama, 2015;Larrasoaña et al., 2007;Roberts et al., 2011;Roberts & Weaveret al., 2005;Rowan et al., 2009). Greigite (Fe 3 S 4 ) is an intermediate mineral phase in the pyrite formation chain that remains in the sediment if dissolved sulfate is depleted before greigite's further reduction to pyrite. Rapid sedimentation rates and an associated upward shift of the SMTZ at continental margins favor the preservation of greigite over pyrite formation. However, a change in redox conditions has been widely described to drive enhanced or secondary mineral diagenesis (e.g., Housen & Musgrave, 1996;Larrasoaña et al., 2007;Musgrave et al., 2019). Localized enhanced magnetic mineral diagenesis in active deformation zones have previously been linked to fluid flux, methane trapped in high permeability fracture zones (e.g., Larrasoaña et al., 2007;Musgrave et al., 2019), methane accumulating below low porosity intervals and gas hydrate occurrences Kars & Kodama, 2015). Faults are assumed to be the main transport pathway for surface gas seeps along the Hikurangi margin (Faure et al., 2010;Greinert et al., 2010;Lauer & Safffer, 2015;Pettinga, 2003;Watson et al., 2019) and continental margins worldwide (Crutchley et al., 2014;Geersen et al., 2016;Mau et al., 2017;Suess, 2010). Site U1518 lies within the hydrate stability zone, and widespread but cm-thick hydrate accumulations identified within sand-and silt-rich layers led to the conclusion that the vertical flux of methane is largely controlled by diffusive migration (Cook et al., 2020). Gas hydrate formation is a slow and exothermic process that depends on temperature, pressure, methane concentration and solubility in existing pore-water, and geometry of the pore-space available Ruppel & Waite, 2020;You et al., 2019). The diffusive transport of methane probably results in its accumulation in brittle, more porous structures within the damage zone of the Pāpaku fault from where it is more favorable to travel in a fault parallel direction. Dissolution in deeper pore-water seeping along high permeability zones would reduce the overall concentration of methane in the pore-fluid which may become undersaturated, thus hampering hydrate formation (e.g., Collett et al., 2019). Continuous and slow seepage of methane rich fluids along the permeable structure of the Pāpaku fault may provide a self-sustaining transport mechanism. For AOM to occur beneath the SMTZ, methane and sulfate are required. Traditionally thought to be sourced from seawater only, increasing evidence suggests that the microbially driven oxidation of pyrite can result in the production of sulfate (see, e.g., Bottrell et al., 2000;Holmkvist et al., 2011;Torres et al., 2015). At Site U1518, an alternative source of sulfate could be from fresh porewater from the underthrust footwall. The SMTZ is located at ∼20 mbsf within the upper meters of unconsolidated seafloor sediment in the hanging wall (Wallace et al., 2019), and it thus appears more likely that these soft and unconsolidated sediments were offscraped and frontally accreted rather than incorporated into the accretionary system.

Fault Permeability Structure
Our results suggest that the migration of methane-and/or sulfate/sulfide and to a minor extent frictional heating at the drill site played a role in the precipitation of pyrite and secondary reduction of existing greigite. Recent results from biomarker thermal maturity analysis indicate elevated paleo-heating temperatures in four discrete, and mm thick intervals through the Pāpaku fault zone, two of which are located within the intervals of the magnetic property anomalies discussed in this paper (Coffey et al., 2021). Short-lived slip events, in association with frictional heating and the migration of co-seismic fluids may thus have resulted in the mobilization of iron and sulfur at the drill site, but were restricted to a narrow, millimeter to centimeter wide zone immediately adjacent to the principal slip zone and thus cannot explain to observed anomalies. In addition, BSE images showed the presence of anhedral pyrite grains and large poly-framboidal growths of iron sulfides within the anomalous intervals. We infer from a mass-balance point of view that their formation is more likely caused by slow AOM that occurs predominantly during interseismic periods.
The two magnetic properties anomalies are located at the top of the upper main brittle fault of the Pāpaku fault zone (Anomaly "A"), and immediately above the lower subsidiary fault (Anomaly "B"). They are separated by the central part of the fault zone, which is an interval with mostly intact, subhorizontal bedding, that is superimposed with few brittle and ductile deformation structures (Figure 8, Wallace et al., 2019). In clay-rich sediments, permeability anisotropy caused by the compaction of clay-rich sediments, fracturing and grain-size differences, influences the direction and locus of fluid flow (Saffer & Tobin, 2011). The majority of advective flow is thought to be concentrated along narrow pathways on thrust and secondary splay-faults (Lauer & Saffer, 2015;Saffer & Tobin, 2011).
We propose that the zonation between sediments that experienced enhanced secondary diagenesis and those that have not, reflect on differences in hydraulic permeability through the fault zone. Anomaly "A" coincides with the upper 8 m of the upper main brittle fault zone, which experienced intensive brecciation, thus forming a more permeable conduit. While at its top Anomaly "A" displays a sharp peak from baseline values to its maximum expression, in its lower portion the magnetic properties more gradually return to base-line values with depth, indicating a progressive increase in greigite content. We suggest that the highest degree of diagenesis coincides with the most prolonged accumulation of solutes at the top of the main brittle fault zone, where fluids are trapped beneath less permeable hanging wall sediment and forced to migrate in a fault parallel direction (see also Musgrave et al., 2019). At first sight it appears contradictory that Anomaly "A" is capped at the top of the fault zone, even though the lower hanging wall forms a wider and fractured damage zone (Figure 2c). It is noteworthy, however, that the hanging wall preserves bedding, with a near-bedding parallel alignment of clay-fabric (Greve et al., 2020;Wallace et al., 2019), which probably provides a barrier to limit advective flow out of the more permeable main brittle fault zone. The top of the main fault also forms the boundary between overriding hanging-and footwall, which have experienced significantly differing strain (Greve et al., 2020;Fagereng et al., 2019;Morgan et al., 2020). Subhorizontally compacted and ductilely deformed sediments within the center of the fault zone likely provide a barrier that limits the migration of fluid across the fault. Anomaly "B" is located above the lower subsidiary fault zone, and in a sand-and silt-rich interval that, based on shipboard core description, was described as having experienced low intensity deformation only (e.g., Figure 3d). It coincides with the locus of a slip zone identified from thermal maturity analysis (Coffey et al., 2021). Similar to our interpretation of Anomaly "A", it is possible that shearing was accompanied by the flow and/or accumulation of reducing fluids that produced conditions favorable for the crystallization of iron-sulfide minerals. While it remains difficult to quantify and determine the source of methane and sulfate, the time-scale across which these reactions take place, and thus whether fluids were sourced within the upper accretionary system or whether they traveled from the main décollement, analysis of the results from the deployment of observatories (Wallace et al., 2019) in the fault zone may provide further insight into these processes.

Conclusions
The ca. 60 m thick Pāpaku fault zone forms a major and active frontal thrust near the deformation front of the Hikurangi Subduction Margin, New Zealand. It was cored during Expedition 375 of the IODP. Magnetic property measurements in combination with electron microprobe imaging were conducted on one to two samples per meter of core recovered. Two prominent magnetic anomalies in two narrow intervals between ∼304 and 312 mbsf and ∼334-351 mbsf were identified. Within these intervals secondary diagenesis resulted in the alteration of ferrimagnetic greigite (Fe 3 S 4 ) to paramagnetic pyrite (FeS 2 ). This is most likely caused by the accumulation and drainage of fluids, including dissolved methane, and sulfate along high permeability fracture zones. Focused and fault parallel advective fluid flow appears to be limited to the top and brecciated fault zone, and to a slip zone near the subsidiary fault zone, where overlying compacted sediments act as seal that limits further upward fluid migration.