Controls on amorphous organic matter type and sulphurization in a Mississippian black shale

Paleoredox proxies (Fe speciation,traceelement and δ 34 S py ) integratedwith sedimentological and palynological observations link the distribution and type of particulate organic matter (OM) preserved to hydrocarbon source rock potential. In the Mississippian Bowland Shale Formation (Lancashire, UK), particulate OM is dominated by “ heterogeneous ” amorphous OM (AOM), primarily “ sharp-edged, pellet-like ” (AOMpel) and “ heterogeneous, granular ” (AOMgr) types. AOMpel is abundant in muds deposited under anoxic and moderately to highly sulphidic conditions and most likely represents the fecal minipellets of zooplankton and/or pellets of macro-zooplankters. We recognize two intervals, “ A ” and “ B, ” which exhibit S org /TOC N 0.04, suggesting a bulk Type II-S kerogen composition. The Interval A palynofacies is typi ﬁ ed by pyritized AOMpel (AOMpyr) particles that contain high-relief organic spheres surrounding individual pyrite framboids, within each AOMpyr particle. These textures are interpreted as sulphurized OM local to pyrite framboids (S org-PF ). S org-PF is rarely observed in Interval B, and absent in all other samples. Redox oscillation between ferruginous and euxinic conditions during early diagenesis of Interval A likely promoted S cycling in microenvironments surrounding pyrite framboids, which generated reactive S species and reactive OM required for sulphurization. Early diagenetic redox oscilla- tion processes were apparently triggered by relative sea level fall, associated with an increased supply of Fe HR from adjacent shelves into the basin. Interval B represents deposition during the late stages of basin in ﬁ ll and transition from anoxic to (sub)oxic bottom waters, where AOMpel is replaced by AOMgr as the dominant type of AOM. A large particle diameter at the limit of the mesh size (500 μ m), sheet-like, fragmented character, and presenceofcandidateorganicsheathssuggestsAOMgratleastpartiallyrepresentfragmentsofbenthicmicrobial mats, probably as sulphide-oxidizers. A ternary plot of AOMpel + AOMpyr versus AOMgr versus spores + phytoclastslinkstheobservedpalynofaciestobottomandporewaterredoxconditions,watercolumnproductiv-ityandproximityto ﬂ uvial(deltaic)supplyofsporesandphytoclasts.Thesevariablesweremoderatedbychang-ing basin accommodation, driven primarily by eustatic sea level ﬂ uctuation. A sequence-stratigraphic control on AOMtypeandsulphurizationisimportantforunderstandingthelinkbetweensourcerockheterogeneityandthe timing of hydrocarbon generation and expulsion from this source rock.


Introduction
The upper unit of the Mississippian Bowland Shale Formation (Upper Bowland Shale; herein "Bowland Shale") is a potential target for unconventional hydrocarbon exploration and an important conventional hydrocarbon source rock in the UK (e.g., Andrews, 2013;Clarke et al., 2018) and time-equivalent units across Europe (Kerschke and Schulz, 2013;Nyhuis et al., 2015). The spatial and temporal distribution and type of organic matter (OM) in organic-rich mudstones, such as the Bowland Shale, is an important control on hydrocarbon prospectivity (e.g., Jarvie, 2012). Palynological assessment of particulate OM type yields important data relating to hydrocarbon source rock generative potential, particularly when supplemented with organic geochemical analyses such as RockEval pyrolysis (e.g., Espitalie et al., 1977). Palynofacies analysis can be used to quantify the proportions of highly aliphatic ("Type I"), moderately aliphatic ("Type II"), humic ("Type
Hemipelagic sediments in the Bowland Shale include discrete, macrofauna-bearing, calcareous sedimentary packages ("marine bands"; e.g., Ramsbottom, 1977;Fig. 1c) that are typically linked to starved siliciclastic input during the maximum rate of transgression (Posamentier et al., 1988) and/or at maximum marine flooding (Martinsen et al., 1995). Marine band cyclicity was likely a response to far-field ice-sheet volume on Gondwana (Veevers and Powell, 1987).
The macrofaunal body fossils present, particularly ammonoids, form the basis of a high-resolution biostratigraphic framework (Ramsbottom and Saunders, 1985). The ammonoid biozones E 1a1 , E 1b1 , E 1b2 and E 1c1 are recognized in the Upper Bowland Shale (e.g., Brandon et al., 1998) (Fig. 1c), with an average periodicity of 111 ka (Waters and Condon, 2012). Multiple flooding surfaces are recognized for E 1a1 (a, b and c) and E 1b2 (a and b) marine bands (Fig. 1c), which may represent sub-100 ka precession or obliquity forcing (Waters and Condon, 2012).
The Bowland Shale was followed by deposition of delta-top facies on the Askrigg Block and a submarine turbidite fan complex in the basin (Fraser and Gawthorpe, 1990;Kirby et al., 2000), as the Pendleton Formation (Waters et al., 2007) (Fig. 1c). Variscan inversion of the Craven Basin resulted in the development of a set of north-east south-west trending folds, thrust-folds and monoclines, collectively defined as the Ribblesdale Fold Belt (Arthurton, 1984;Gawthorpe, 1987) (Fig. 1b).

Palynology
34 palynology samples were processed using standard preparation methods (Wood et al., 1996), using hydrochloric (36% vol/vol) and hydrofluoric (HF) (40% vol/vol) acid to remove carbonate and silicate phases, respectively. Kerogen residues were sieved at 500 and 10 μm. Residues were spiked with Lycopodium clavatum spores (batch no. 3862, MPA66949-66953, 2 tablets in 5 g sample; MPA67171-68150, 3 tablets in 5 g sample) in order to calculate absolute abundances using the marker grain method (Maher, 1981;Stockmarr, 1971). Since AOM was the primary target in this study, and highly abundant compared to spores, this precluded calculation of absolute abundances using the marker grain method. A much larger spike (i.e., N 10 tablets) could potentially enable quantification of spore absolute abundances (but not AOM) in this material. A subset of residues were oxidized for 7 min using fuming nitric acid.
Each sample residue was strew-mounted onto two slides and set in low fluorescing Elvacite™ resin. Microscopic observations were conducted using an Axio Imager M2 m fitted with a motorized stage under transmitted light (brightfield and differential interference contrast) and incident green light. Green-light emission was attained using a 100 V tungsten filament (HVP) lamp at maximum brightness passed through the Zeiss 00 Propidium Iodide (PI) filter set (λ 530-585 nm excitation). A subset of slides were also analyzed under a mercury HXP light source. Under green-light, the spectral irradiance for tungsten filament lamps and mercury lamps is similar (Lin and Davis, 1988). × 10, × 40 and × 100 (oil immersion) EC Plan Neofluor objectives were utilized. Emitted light was red-light filtered (530-585 nm λ beam-splitter, filtering emission above λ 615 nm).
Green light fluorescence (coupled with red light filter) was used as a proxy for moderately quenched (i.e., polymerized) structures and highly "delocalized" double bonds (Lin and Davis, 1988). Aromatic structures contain highly delocalized bonds, rather than acyclic compounds which may contain isolated double-bonds (e.g., alkenes, acyclic terpenoids). Thus green light fluorophors represent phenols and heteroatomic aromatic moieties contained within aliphatic polymers, and some aromatic (e.g., lignin-containing) substances. Highly polymerized substances that contain abundant delocalized bonds (e.g., humic compounds) promote quenching and are therefore typically non-fluorescent under all wavelengths (Lin and Davis, 1988).
Images were captured with an AxioCam MR R3 camera connected to a PC with Zen 2012 (Blue Edition) software. All images were captured in a dark room and under the same ambient light conditions. Particles were described and counted using an automatic stage, based on visual assessment and semi-automated analysis of extracted particle shape and fluorescence parameters using the Fiji ImageJ platform (Schindelin et al., 2012). Total AOM and AOMgr abundances are therefore relative and corrected for particle area. Palynofacies abundance data are provided in Appendix A.

Geochemistry
Samples selected for geochemical analysis were crushed and agatemilled to a fine powder. 193 samples were selected for pyrolysis, conducted on finely powdered samples in a Rock-Eval 6 pyrolyser (Vinci Technologies) in standard configuration and operated using the Bulk-Rock method (e.g., Słowakiewicz et al., 2015). The generated total organic carbon (TOC) and inorganic C (MINC) data were previously reported by Emmings et al. (2019).
Total carbon (C) and sulphur (TS) concentrations were measured on 110 samples using a LECO CS 230 elemental analyser. Major and trace element concentrations were measured on fused beads (109 samples) and powder briquettes (108 samples) with a PANalytical Axios The Bowland Shale Formation is exposed as part of the Ribblesdale Fold Belt, including Westphalian structural elements (Fraser and Gawthorpe, 1990) and study sites Hind Clough (outcrop) and borehole Marl Hill 4 (MHD4). Outcrop extent data are based on DigMapGB-625, published with permission of the British Geological Survey. (c) Generalized Lower Namurian stratigraphy and Craven Basin composite after Brandon et al. (1998). Marine band areal extent data from Waters and Condon (2012), including diagnostic ammonoid fauna from Riley et al. (1993). Mesothems from Ramsbottom (1973). HS = Hind Sandstone Member. PG = Pendle Grit Member. Advanced X-Ray Fluorescence (XRF) spectrometer using default PANalytical SuperQ conditions. "Excess Si" (sensu. Sholkovitz and Price, 1980) was calculated by Emmings et al. (2019) and interpreted primarily as a pelagic and biogenic (radiolarian) signal that did not migrate significantly during diagenesis. Trace element enrichment factors (EFs) (e.g., Tribovillard et al., 2006) were utilized to normalize to abundances to the detrital fraction (Eq. 1), using Post-Archaean Average Shale (PAAS) (Taylor and McLennan, 1985). X is the element concentration (major elements; wt.% and trace elements; ppm).
The pyrite (Fe py ) fraction was estimated via extraction of sulphide S liberated by boiling chromous chloride solution, and titrated as Ag 2 S (Canfield et al., 1986). This followed extraction of HCl-soluble (acid-volatile, Fe AVS ) sulphide S, in boiling HCl (Canfield et al., 1986), although in all samples Fe AVS was only present as a trace component (below the limit of determination). Fe HR /Fe T (Fe T ; total Fe) and Fe py /Fe HR are compared with established thresholds for redox (Poulton and Canfield, 2011;Poulton and Raiswell, 2002;Raiswell and Canfield, 1998). Facies H-I Fe HR is presented on a Fe mag -free basis, due to the likely input of detrital Fe mag (see Emmings, 2018).
37 sample powders (~1 g) were washed for 24 h in 200 ml 10 wt/vol % NaCl in order to leach free sulphate (Kampschulte et al., 2001). Total S measured on NaCl-washed residues was compared with the total S of untreated powders, yielding an estimate for the sulphate S (S sul ) fraction, after correction for the mass loss assuming leaching of pure CaSO 4 (Appendix B). In most cases, estimated S sul was within or close to ±0.08 wt.% of the total (untreated) S (i.e., the long-term analytical precision). Therefore S sul is negligible for most samples. S org content was estimated by subtraction of S py (Fe py *1.15) and S sul from TS (Eq. 3) (e.g., Tribovillard et al., 2001). Propagating the precision of TS, S py and S sul in quadrature yields a precision estimate of ±0.12 wt.% S org .

Palynofacies and paleoredox
The palynofacies assemblage in sedimentary Facies A-G (see Fig. 2) is dominated by AOM, while spores and phytoclasts are a minor component (b20%), and marine palynomorphs are absent ( Fig. 2; Table 1). In Facies H-I at Cominco S9, spores and phytoclasts are dominant ( Fig. 3; Table 1). Nearly all AOM is "heterogeneous" (e.g., Tyson, 1995) and subdivided into two main types; "sharp-edged, pellet-like" AOM (AOMpel; Plate I) and "granular" AOM (AOMgr; Plate II). AOMgr is rare in Facies A-F and is the dominant type of AOM in Facies G, whereas AOMpel is dominant Facies A-F and rare in Facies G.

Mechanisms for sulphurization
Intervals "A" (within Facies F) and "B" (Facies G) exhibit S org /TOC N 0.04, suggesting a bulk Type II-S kerogen composition (Orr, 1986). A microbial mat origin to AOMgr may explain the exceptionally high S org content in Interval B (Facies G). Modern microbial mats oxidize H 2 S using O 2 or NO 3 − as an electron acceptor across a strong redox gradient at the seabed (Bailey et al., 2009;Canfield and Teske, 1996;Grunke et al., 2011;Sievert et al., 2007;Wirsen et al., 2002), and are consortia of several different types of bacteria. AOMgr is probably most comparable to the colorless sulphur bacterium Arcobacter (phylum Proteobacteria) that produces a "cotton-ball"-like structure (Grunke et al., 2011;Wirsen et al., 2002) via fixation of S org in filaments   Fig. 2 for key to sedimentary facies. **Facies H-I Fe HR is presented on a Fe magfree basis, due to the likely input of detrital Fe mag (see Emmings, 2018 for discussion). (Steudel, 1989). Arcobacter mats are present at cold seeps on the Nile Deep Sea Fan (Grunke et al., 2011), for example. AOMgr may also be comparable to Thiopoloca, a sulphide-oxidizing denitrifier that is widespread beneath the upwelling Peru-Chile upwelling region (Fossing et al., 1995). Alternatively, the high S org content in Interval A may be explained by direct reaction of H 2 S with OM (e.g., Sinninghe Damsté and De Leeuw, 1990a;Amrani, 2014). Regardless of S fixation mechanism, stagnation or advection of sulphidic porwaters near seabed implies Plate I. "Sharp-edged" heterogeneous amorphous organic matter (AOMpel) example transmitted light microphotographs. Scale bars = 50 μm unless otherwise stated. 1-4. MPA68143. 5. MPA68144. 6. MPA66949. 7-9. MPA68145.
relatively vigorous and sustained rates of SO 4 -reduction (as the likely source of sulphide) and/or minimal buffering by reactive Fe in underlying sediments.
Interval A in Facies F overlying the E 1a1 marine band exhibits relatively high S org content yet contains minimal AOMgr ( Fig. 2; Table 1). This package is distinctive because the palynological fraction is dominated by AOMpel that is coated by fine, microcrystalline pyrite ("AOMpyr"; Plate I, 6). AOMpel in Interval A also contains abundant orange, high-relief organic spheres within each particle (Plate III). Rarely, these spheres are also present within AOMgr in Facies G (Interval B; Fig. 2; Plate III, 6-9).
The spheres exhibit a finely reticulated texture in transmitted light that mimics pyrite framboids (Plate III, 3), and are best observed in oxidized slides following removal of the microcrystalline pyrite coatings on AOM. In some cases, each sphere contains a single pyrite framboid (Plate III, 1). The diameter of organic spheres in Interval A (Fig. 6) is similar to pyrite framboids in Facies F (Emmings et al., 2019). The orange color of the organic spheres suggests sulphurized OM (e.g., Tribovillard et al., 2001), possibly comparable to orange gel-like AOM "drops" (Aycard et al., 2003). AOM sulphurization is consistent with the relatively high S org content in Interval A. The catalytic effect of S radicals associated with S org during the thermal decomposition of kerogen (Lewan, 1998) may also explain the relatively low Rock-Eval T max through Interval A. T max through Interval A is approximately 15°C lower than the adjacent mudstone packages (Figs. 2, 7; Table 1). Such T max inversions are not necessarily diagnostic of Type II-S OM, however, because a similar reduction in T max is observed through the E 1c1 marine band at Cominco S9 despite very low S org content (Fig. 3). This suggests T max is also influenced more broadly by OM composition (Type II versus III) and/or mineral matrix effects.
Sphere fluorescence under green light (Plate III,5,7,9) suggests the presence of aromatic compounds within a moderately polymerized structure (Lin and Davis, 1988), potentially including S heteroatoms such as thiophenes (e.g., Eglinton et al., 1990). Since AOMgr in Interval B lacks fluorescence under green light and generally lacks S org-PF , this Fig. 4. "APP" ternary plot of heregeneous AOM (defined as AOMpyr + AOMpel + AOMgr) versus phytoclasts + homogenous AOM (the latter is typically negligible in samples in this study) versus spores (terrestrial palynomorphs). Fields and interpreted processes are from Tyson (1995). Reprinted/adapted by permission from Springer Nature Customer Service Centre GmbH: Springer Nature, Sedimentary Organic Matter: Organic Facies and Palynofacies by Tyson, R. V. © 1995. See Fig. 2 andEmmings et al. (2019) for sedimentary facies description. Palynofacies abundance data for the Morridge Formation in the Edale Gulf (borehole Karenight 1) and Widmerpool Gulf (borehole Carsington DRC3), from Hennissen et al. (2017), are also plotted for comparison.
The orange spherical texture in AOM is termed "S org-PF ," and is comparable to the texture originally described by Love (1957Love ( , 1962. Pyrite framboids are present in all sedimentary facies (Emmings et al., 2019), yet S org -PF is only present in interval A, and rarely interval B. Clearly the microenvironment surrounding pyrite framboids within OM promoted sulphurization. Sulphurization of OM requires access to compounds prone to complexation with S (Adam et al., 2000) and available reactive S species in the local microenvironment. O contained within carbonyl groups, in particular, is prone to replacement by reduced S (e.g., Adam et al., 2000). Carbohydrates are also susceptible to sulphurization (Van Kaam-Peters et al., 1998). Reactive S species include H 2 S, polysulphides, sulphites, or other S of intermediate oxidation states (Aizenshtat et al., 1995;Amrani, 2014;Amrani and Aizenshtat, 2004;Rickard, 2012;Sinninghe Damsté and Leeuw, 1990b;Wasmund et al., 2017).
Early diagenetic redox oscillation within Interval A is consistent with the Mo and U and δ 34 S py record (Table 1; Fig. 2). Firstly, high Mo relative to U in Interval A (Facies F mean Mo EF~76 ± 86, U EF~8 ± 3) suggests the presence of a relatively weak and unstable water column chemocline and development of "particulate shuttle" conditions (Algeo and Tribovillard, 2009). This contrasts with moderate to high Mo and U contents in "marine band" Facies A-C (mean Mo EF~32 ± 39, U EF~11 ± 11), which suggests relatively stable, moderately sulphidic conditions in bottom waters. Secondly, strongly negative (close to -40‰) δ 34 S py in Interval A suggests pyrite precipitation under long-lived open-system conditions and/or large S fractionations via intermediate S cycling (e.g., Nissenbaum et al., 1972;Mossmann et al., 1991;Canfield et al., 1992;Canfield et al., 2010).
H 2 S that migrated into surrounding porewaters was likely oxidized to produce sulphuric acid (e.g., Soetaert et al., 2007). This explains the lack of carbonate (Fig. 2), despite the presence of skeletal moulds (Emmings et al., 2019), in Intervals A and B. Therefore the contacts between OM microenvironments and adjacent porewaters exhibited an intermittently high redox-gradient (Fig. 8). This gradient was likely strongest during periods of redoxcline deepening, where porewaters were flooded by Fereducers. S cycling across these micro-redox fronts, local to each framboid, provided the reactive S species required for sulphurization.
Pyrite nucleation within AOM likely formed initially as FeS, via reaction with dissolved reduced Fe, sourced via dissimilatory Fe 3+ reduction or direct reduction of particulate FeOOH by H 2 S (Fig. 8). FeS subsequently reacted with H 2 S and/or polysulphides to produce pyrite (e.g., Rickard, 2012). Microcrystalline pyrite coatings on OM (Plate I, 6) are therefore interpreted as a relict redox front, defined by reaction of H 2 S generated local to OM, and Fe 3+ and/or FeOOH present in the surrounding porewaters.
S org-PF is abundant in Interval A but rare in Interval B (compare Plate III, 2 and 8), perhaps because pyrite framboids are also rare in Interval B (Emmings et al., 2019). This suggests that H 2 S and/or polysulphide in porewaters seldom attained the critical supersaturation conditions required for framboid nucleation (Ohfuji and Rickard, 2005;Rickard, 2012). Together, this suggests that the steepest redox gradient during deposition of Interval B was located near or at (and parallel to) the seabed, likely utilized by microbial mats, and not local to OM. Thus pyrite nucleation and growth in Interval B was uncoupled to sites of in situ H 2 S production driven by OM degradation, but instead driven by reaction of reduced Fe with upward-diffusing H 2 S. It is also possible AOMgr, as the dominant type of particulate OM in Interval B, was relatively refractory and therefore relatively resistant to hydrolysis and degradation by microbes. This limited the production of H 2 S local to OM required for framboid (and therefore S org-PF ) growth.

Controls on organic matter distribution
The distribution of key palynofacies categories through the Upper Bowland Shale in the Craven Basin is predictable and linked to changing bottom water redox conditions and changing basin accommodation. Facies A-C muds deposited during marine transgressions (Figs. 2-4, Table 1) lack S org-PF likely because the redoxcline bounding zones of SO 4 and Fe reduction was fixed in the water column during deposition (Fig. 9a). Marine maximum flooding is associated with a high AOMpel/AOMgr and low abundance of spores and phytoclasts.
Intervals A and B overlie "marine band" packages and were therefore likely deposited during sea level fall (Emmings et al., 2019). An enhanced shelf-to-basin "Fe shuttle" during falling sea level (Lyons and Severmann, 2006) increased the Fe HR supply to the basin (Emmings, 2018). This process promoted buffering of H 2 S by Fe HR via pyrite formation (including nucleation on AOM to generate AOMpyr), weakening and destabilizing the chemocline, stimulating early diagenetic redox oscillation and triggering S org-PF formation (Interval A; Fig. 9b). Therefore S org-PF is a proxy for redox oscillation between sulphidic and ferruginous anoxic micro-environments during early diagenesis. Pendle delta progradation during the E 1b2 biozone triggered euphotic zone desalination, which sufficiently reduced the export of autochthonous (marine) OM to seabed and therefore promoted bottom water oxygenation (Emmings, 2018). Bottom water ventilation promoted colonization of the seabed by sulphide-oxidizing microbial mats (Fig. 9c). Therefore the transition from anoxic (Facies F) to oxic/ suboxic conditions (Facies G) exhibits a "cross-over" between AOMpel and AOMgr ( Fig. 2; Table 1). The proportionality of AOMpel and AOMgr may therefore delineate ancient redox fronts at or near seabed. Palynological assessment of AOM types should be coupled with detailed sedimentological characterization, however, in order to determine whether AOMgr particles are present in situ or as rip-up clasts. . Rock-Eval pyrolysis T max versus organic S/C, with point size and color mapped to the relative abundance of AOMgr, and with Type II-S field (S/C N 0.04) from Orr (1986). Several turbidites, debrites and hybrid flow deposits exhibit moderate S/C and contain AOMgr interpreted as rip-up clasts (Emmings et al., 2019). Interval A is possibly positioned at the end of a mixing line defined by the catalytical effect of S radicals during hydrocarbon maturation (Lewan, 1998).  and pyrite microtextures described by Emmings (2018) and Emmings et al. (2019). A ternary plot of AOMpel + AOMpyr versus AOMgr versus spores + phytoclasts (Fig. 11) links palynofacies abundances to bottom and pore water redox conditions, water column productivity and proximity to fluvial (deltaic) supply of spores and phytoclasts. Field I delineates AOMgr-rich samples located in the paralic Widermerpool Gulf (Hennissen et al., 2017). Field I is a mixing line between AOMgr and spores and phytoclasts. Lack of AOMpel suggests relatively low productivity water column conditions. Dominance of AOMgr in Field I may indicate widespread occurrences of candidate microbial mats existed in these paralic basins, supported by a long-lived, restriction-driven redox gradient near or at seabed. Alternatively, it is possible AOMgr is Fig. 10. Summary of key observations and interpretations spanning palynology and organic geochemistry (this study) and sedimentology (Emmings et al., 2019). Backscattered electron microphotographs are also reported by Emmings (2018). All scale bars = 50 μm. Generalized geochemistry includes mean Mo and U EFs with uncertainty quantified as two standard deviations. generated via multiple pathways; perhaps AOMgr also includes a component of bacterially modified terrestrial OM (TOM). Field II is interpreted to indicate low to moderate water column productivity, dominantly oxic or suboxic bottom water conditions and sulphidic conditions near or at seabed (Figs. 11, 12a). The atomic S org /TOC in Field II (Interval B) exceeds 0.04 (Table 1; Fig. 12b), the threshold for definition of Type II-S kerogen (Orr, 1986). Field III represents moderate to high rates of water column productivity and autochthonous OM export to seabed, linked to anoxic and at least intermittently sulphidic bottom water conditions (Figs. 11;12a). This field includes Type II-S kerogen (S org-PF ) generated via early diagenetic redox oscillation (Fig. 9b). S org-PF exhibits a possible catalytic effect on T max (Lewan, 1998). The majority of Bowland Shale samples from the Craven Basin and contemporaneous mudstones from the Edale Gulf (Hennissen et al., 2017) plot within this field. This suggests productivity was relatively widespread, and perhaps stimulated by nutrient loading from the nearby Pendle delta system (Figs. 1a, 9a-b) (Emmings, 2018). Field V represents oxygenated conditions and/or close proximity to the supply of TOM (Figs. 9, 12a). Facies H-I in Cominco S9, located proximal to the Pendle delta system and deposited during reduced basin accommodation, are sited in this field. Thus Field IV likely represents transitional settings defined by fluctuating oxic/suboxic and anoxic conditions, or significant supply of spores and phytoclasts into anoxic bottom waters (Fig. 11).
The mixing line between fields III and V (and therefore through Field IV; Fig. 11) represents increasing proximity to fluvial sources and/or bottom and pore water ventilation that was sufficiently gradual or diffuse. This inhibited development of a high redox gradient at seabed. Candidate microbial mats (AOMgr) were unable to colonize seabed in these settings. The mixing line between fields III and I represents rapid bottom water ventilation and/or high frequency redox fluctuation, coupled to persistently stagnant or advective, sulphidic porewaters near seabed. This configuration supported colonization by the candidate microbial mats.
Hydrogen index (HI), a key measure for hydrocarbon source rock potential, is highest near Field I and III apices for all immature to early oil-mature data plotted ( Fig. 12c; including data from Hennissen et al., 2017). Absence of inorganic geochemical data for Field I precludes assessment of the mechanism for high HI in this field. However, assuming paralic basins were subject to long-lived and stable redoxclines near seabed, it is plausible these conditions promoted condensation and preservation of relatively labile (aliphatic) OM. High HI in Field III is best explained by enhanced preservation of labile OM under sulphidic bottom water conditions beneath a stable redoxcline. Changing basin accommodation during deposition of the Bowland Shale is considered at least partially equivalent to eustatic sea level systems tracts (Posamentier et al., 1988). A sequence-stratigraphic control on OM type is important for understanding the link between source rock heterogeneity and timing of hydrocarbon generation and expulsion from this source rock. The effect of changing sea level for biozones E 1a1 to E 1c1 is summarized using the ternary plot of AOMpel + AOMpyr versus AOMgr versus spores + phytoclasts (Fig. 13). Key intervals deposited during falling sea level contain bulk Type II-S OM, which likely entered the oil window at a relatively low temperature (e.g., Dembicki, 2009). Understanding the distribution and type of OM is important for exploring this unconventional hydrocarbon resource in the UK (e.g., Andrews, 2013;Clarke et al., 2018), especially if Type II-S intervals are laterally extensive. The possibility that sulphideoxidizing microbial mats colonized seabed, and across several basins, suggests nutrient and inorganic S and C cycling in epicontinental Missississippian seaways likely operated in a vastly different way compared to modern oceans.  Fig. 11 for details), with (a) Fe HR /Fe T ; (b) organic S/C, and; (c) hydrogen index (HI) mapped to each data point. Palynofacies abundance data for the Morridge Formation in the Edale Gulf (borehole Karenight 1) and Widmerpool Gulf (borehole Carsington DRC3), from Hennissen et al. (2017), are also plotted with HI. See text for discussion. Fig. 13. Ternary plot of AOMpel + AOMpyr versus AOMgr versus spores + phytoclasts (see Fig. 11 for details), with organic S/C mapped to each data point and interpreted basin accommodation pathways from Emmings et al. (2019). Basin accommodation pathways are at least partially equivalent to the eustatic sea level systems tracts of Posamentier et al. (1988). TST = transgressive systems tract; HST = highstand systems tract; FSST = falling stage systems tract; LST = lowstand systems tract.

Conclusions
Geochemical and palynological data were integrated through the Upper Bowland Shale unit in the Craven Basin (Lancashire, UK), a basin with ongoing hydrocarbon exploration. Fe-speciation, trace element geochemistry and δ 34 S py analyses were utilized in order to assess syngenetic and early diagenetic redox conditions. These data were integrated with sedimentological and palynological observations, in order to understand the controls on OM sulphurization and the distribution of AOM.
Particulate OM in the Upper Bowland Shale is dominated by two types of AOM; "homogenous" AOM (AOMpel) and "heterogeneous, granular" AOM (AOMgr). On the basis of textural observations, AOMpel most likely represent the fecal minipellets of zooplankton and/or pellets of macro-zooplankters. On the transition from anoxic to oxic bottom waters, AOMgr replaces AOMpel as the dominant type of AOM (Interval B). A large particle diameter (likely N 500 μm), sheet-like, fragmented character, and presence of candidate organic sheaths suggests AOMgr at least partially represent fragments of benthic microbial mats, likely as sulphide-oxidizers.
Abundant orange, high-relief organic spheres are recognized within each AOM particle, particularly in one key interval (A) overlying the E 1a1 marine band. These textures are associated with a high S org content and are therefore interpreted as sulphurized OM local to pyrite framboids (S org-PF ). Sulphurization is linked to early diagenetic redox oscillation processes. Whilst the precise mechanism for sulphurization is unclear, we propose redox oscillation promoted sulphurization in two ways. Firstly, redox oscillation enhanced the degradation of OM. This produced organic compounds prone to sulphurization. Secondly, redox oscillation also promoted S cycling across micro-redox fronts local to each framboid. Intermediate, and therefore reactive, S species were thus available for complexation with OM.
S org-PF formed primarily under anoxic conditions during periods of reduced sea level (Interval A), via an increased supply of Fe HR from adjacent shelves. An increased supply of Fe HR stimulated redox oscillation between ferruginous and euxinic conditions, which promoted acidification of porewaters near seabed. Redox oscillation was associated with S cycling required to generate reactive S species.
Both intervals A and B exhibit S org /TOC N 0.04 and are therefore interpreted as Type II-S kerogen. A ternary plot of AOMpel + AOMpyr versus AOMgr versus spores + phytoclasts links the observed palynofacies and sulphurization (intervals A and B) to bottom and pore water redox conditions, water column productivity and proximity to fluvial (deltaic) supply of spores and phytoclasts. These variables were moderated by changing basin accommodation, driven primarily by eustatic sea level fluctuation. This is important for understanding the link between source rock heterogeneity and timing of hydrocarbon generation and expulsion from this source rock.