Ironstone is a Phanerozoic marine biochemical sedimentary rock type with an Fe content of greater than 15 wt% (Van Houten and Bhattacharyya, 1982; Van Houten and Arthur, 1989; Young et al., 1989; Pufahl, 2010). Most ironstone is aluminous, phosphate-rich, and contemporaneous with major intervals of phosphorite accumulation (Young, 1992). Phosphorite, an important fertilizer ore, is also a marine biochemical sedimentary rock containing at least 18 wt% P₂O₅ (Filippelli, 2008; Pufahl, 2010). The two most prominent episodes of coincident ironstone-phosphorite deposition occurred in the Ordovician and the Jurassic (Van Houten and Arthur, 1989; Young, 1992). Both periods generally correspond to intervals of increased tectonic activity, elevated hydrothermal input, globally warm climate, and biologic radiation in the marine realm (Figure 1; Van Houten and Bhattacharyya, 1982; Van Houten and Arthur, 1989; Sturesson et al., 2000; Servais et al., 2010).

The interplay of these processes ultimately controls the delivery, cycling, and sedimentation of Fe and P in the ocean (Pufahl, 2010; Föllmi, 2016). Although the deposition of phosphatic sediment is generally well understood (e.g., Föllmi, 1996; Pufahl and Groat, 2017), processes governing ironstone accumulation remain poorly constrained. Phosphatic ironstone of the Welsh Basin (ca. 467 Ma) provides a unique opportunity to investigate these processes and their potential relationship to the Great Ordovician Biodiversification Event (GOBE; ca. 480–445 Ma), which is the largest sustained marine radiation in Earth history (Servais et al., 2009, 2010; Edwards, 2019). A distinguishing characteristic of this ironstone is the occurrence of large coated Fe grains. The size and variable mineralogy of coated grains provide an unparalleled window into water column and seafloor processes that facilitated ironstone precipitation.

The oceanography and depositional environments of phosphatic ironstone in the Welsh Basin are herein constrained by documenting their sedimentology and sequence stratigraphy. The diagenesis and paragenesis of lithofacies are interpreted in this framework to place these ironstones in their paleoenvironmental context, providing the basis for understanding how changes in the benthic cycling of Fe and P produced coated Fe grains. Collectively, these data assist with understanding the relationship between changes in seawater chemistry, Ordovician tectonic events, and the GOBE.

The Welsh Basin records major tectonic events in the early Paleozoic evolution of Avalonia (Figure 2). Sedimentation began in the Cambrian along a passive margin. By the Floian, the Welsh Basin had evolved into a back-arc basin with the onset of subduction along the southern flank of the Iapetus Ocean (Prigmore et al., 1997; Rushton and Howells. 1998; Brenchley et al., 2006; Howells, 2007; Murphy et al., 2008; van Staal et al., 2012). Such subduction caused rifting of Avalonia from Gondwana and the opening of the Rheic Ocean as a back-arc basin. As Avalonia drifted northward, the widening Rheic Ocean experienced increasing faunal endemism (Cocks and Torsvik, 2002; Fortey and Cocks, 2003; Linnemann et al., 2012; Pothier et al., 2015; Henderson et al., 2016). Arc-related volcanism in the Welsh Basin ceased when Baltica and Avalonia collided during the late Sandbian to earliest Katian (van Staal et al., 1998; Torsvik and Rehnström, 2003; Murphy et al., 2004; Henderson et al., 2016).

In the Welsh Basin, Ordovician strata consist of the ironstone-bearing Ogwen Group (Woodcock, 1990), which is bounded by major, basin-wide unconformities. Its basal unconformity records uplift associated with the onset of Iapetan subduction whereas the upper unconformity developed during collision with Baltica (Woodcock, 1990; Rushton and Fortey, 2000; Brenchley et al., 2006; Howells, 2007). The Ogwen Group ranges in thickness from 1 to 2 km and preserves a continuous record of early Floian to middle Katian (Arenig to Caradoc in the United Kingdom) marine sedimentation (Rushton and Howells, 1998; Rushton and Fortey, 2000; Brenchley et al., 2006; Howells, 2007). From base to top, the Ogwen Group consists of the Wig Bach Formation, Llanengan Mudstone Formation, Tygarn Formation, unsubdivided Nant Ffrancon Subgroup, Ty’r Gawen Mudstone Formation, Cwm Eigiau Formation, and Nod Glas Formation (Figure 3; Rushton and Fortey, 2000). Collectively, these units define a fining upward and deepening succession from fan delta and shoreface deposits to deep-water, pyrite-rich, graptolitic mudstone (Howells and Smith, 1997; Rushton and Fortey, 2000; Young et al., 2002). Paleocurrent directions from cross-bedded and rippled sandstones indicate a predominance of east-northeast directed shelf currents on a south-easterly dipping paleoslope (Beckly, 1988; Young et al., 2002; Howells, 2007).

Phosphatic granular ironstone occurs as discrete decimeter-thick horizons in the Wig Bach, Tygarn and Ty’r Gawen Mudstone formations. These horizons are penecontemporaneous with other ironstone beds in eastern North America, southwestern Europe and North Africa (Young, 1989; Young et al., 1989; Woodcock, 1990; Young, 1992; Young et al., 2002; Todd et al., 2019), which also accumulated along the margins of the Rheic Ocean. Research herein focuses on the Darrwilian Hen-dy-Capel Ironstone Member of the Tygarn Formation (Figure 3; Trythall et al., 1987; Young et al., 2002) because it provides the clearest window into ironstone depositional processes. Unlike other ironstone horizons in Wales, the Hen-dy-Capel Ironstone Member is well-exposed in old mining adits and quarries.

Three field sites and two drill cores were examined to investigate the lateral and vertical facies variation of the Hen-dy-Capel Ironstone Member (Figure 2). At Betws Garmon eight adits, mined during the 19th century and through World War 1 for steel production (Strahan et al., 1920), allowed lateral correlation of lithofacies along 1 km of depositional strike. Two field sites on St. Tudwal’s Peninsula in small, abandoned open pit mines yielded information on the regional variability of ironstone and associated facies. Drill cores from near the town of Rhiw (Brown and Evans, 1989) provided further insight into the stacking relationships of lithofacies. Stratigraphic sections were described on a bed-by-bed basis to develop a lithofacies nomenclature, define important sequence stratigraphic surfaces, and interpret paleoenvironments of deposition through time. The Folk classification (Folk, 1980) was used to define terrigenous clastic facies, whereas a modified Dunham classification scheme (Dunham, 1962) was employed to characterize ironstone because these chemical sedimentary rocks share similar textural attributes to limestone (Young, 1989; Pufahl, 2010). Bioturbation was categorized using the Droser and Bottjer (1986) ichnofabric index from 1 (no activity) to 6 (complete homogenization by bioturbation).

Seventy-four polished thin sections were examined using a Nikon Optiphot-POL transmitted and reflected light microscope. Modal compositions of minerals and fossils were estimated using a visual percentage chart (Terry and Chilingar, 1955) with an abundance index of rare (<10%), uncommon (11–40%), common (41–70%), and abundant (>70%). Minerals were identified in thin section by their petrographic characteristics and through profile-fitting of X-ray powder diffractograms. Plagioclase compositions were determined under transmitted light using the Michel-Lévy method (Sørensen, 2013).

Twelve whole rock powders were analyzed on a Malvern-Panalytical Empyrean Series 2 X-ray diffractometer with a programmable PIXcel3D area detector across scattering angles from 4° to 80° using a copper X-ray target source. Although a Cu source causes fluorescence of Fe-rich minerals, programming the peak height discrimination setting on the detector to 54% improved peak-to-background ratios on diffractograms for mineral identification in ironstone lithofacies. Mineralogy was interpreted from X-ray powder diffractograms using Malvern-Panalytical’s HighScore Plus software package.

Important paragenetic relationships were further analyzed using a JEOL JSM6400 scanning electron microscope (SEM) equipped with a Genesis Energy Dispersive X-ray Analyzer at the Microscopy and Microanalysis Facility at the University of New Brunswick—Fredericton. Backscattered electron images and energy dispersive X-ray spectra were generated using an accelerating voltage of 15 kV and a working distance of 27 mm. Carbonate-rich samples were also investigated using a Nikon Eclipse E400-POL microscope equipped with a Reliotron III cathodoluminescence system to understand paragenetic relationships not visible under transmitted and reflected light microscopy and SEM imaging.

Whole rock geochemical analysis of 60 samples provides major, minor, and trace element data to further constrain ironstone depositional processes. Analyses were performed at AGAT Laboratories, Mississauga, using a combination of Li borate fusion with an X-ray fluorescence finish and Na peroxide fusion with an inductively coupled plasma optical emission spectrometry (ICP-OES)/inductively coupled plasma mass spectrometry (ICP-MS) finish. Total organic C (TOC) was analyzed using LECO combustion infrared spectroscopy. Replicate analyses indicate a reproducibility for major elemental concentrations using Li borate fusion and an XRF finish of ± 2%. REE + Y and other trace element concentrations determined using a Na peroxide fusion and an ICP-OES/ICP-MS finish have a reproducibility of ±10%. The reproducibility of TOC concentrations using LECO combustion is ±1%. Organic-matter-rich lithofacies contain >0.5 wt% TOC (Trabucho-Alexandre, 2015). The geostandards SY-4 (diorite gneiss), 692 (iron ore), GBM998-10 (multi-metal nickel ore), and Till-2 (till) were used as matrix matching standards for various facies. Whole-rock geochemical data were plotted using GCDkit 5.0 in R 3.4.3 (Janoušek et al., 2016).

Rare earth elements and yttrium (REE + Y) concentrations are normalized with respect to the Post Archean Australian Shale standard (PAAS; McLennan, 1989) to understand redox controlled processes of Fe and P precipitation. Ce and Eu anomalies are calculated as shale normalized (subscript SN) Ce_SN/Ce_SN∗ and Eu_SN/Eu_SN∗, respectively (Bau et al., 2014), where:

Ce_SN∗ = 0.5La_SN + 0.5Pr_SN

Geochemical data are interpreted in paragenetic and sequence stratigraphic context to elucidate the source of Fe to the Welsh Basin and to create an oceanographic model for ironstone accumulation.

Three terrigenous and four ironstone lithofacies are recognized in the Ogwen Group. Facies 1, 2, and 3 are siliciclastic and Facies 4, 5, 6, and 7 are authigenic and related to ironstone deposition. Lithofacies associations suggest deposition on an unrestricted, storm-dominated shelf.

A parasequence is a relatively conformable succession of genetically related beds bounded at the bottom and top by flooding surfaces (Van Wagoner et al., 1988). Parasequences are the primary components of systems tracts, which define a stratigraphic sequence recording deposition through a complete sea level cycle (Catuneanu et al., 2011; Couneanu, 2019). Thus, parasequences record smaller scale fluctuations in relative sea level superimposed on longer wavelengths of sea level oscillation.

Outcrop exposures and lithofacies stacking patterns indicate that two laterally correlative parasequences are preserved in the study area (Figure 7). At Betws Garmon, vertical and lateral facies trends are consistent between mine adits along 0.8 km of depositional strike. With minor variations in facies thickness, these trends are persistent between Betws Garmon, St. Tudwal’s Peninsula and Rhiw.

Parasequence 1 is an aggradational ironstone succession that is at least 8 m thick. This parasequence is interpreted to record a gradual change from distal through middle shelf sedimentation to deposition near fair weather wave base. Basal laminated, organic-rich mudstone (F4) accumulated below storm wave base and grades stratigraphically upward into a bioturbated mudstone (F5) that was the locus of intense authigenesis. These mudstones are rich in chamosite and are, in turn, overlain by an intraclastic, Fe coated grain packstone with locally developed HCS (F6), which records shallowing to just above storm wave base. The top of Parasequence 1 is a trough cross stratified grainstone (F7) reflecting continued aggradation and increased reworking of coated grains near fair weather wave base. Such intense reworking, winnowing, and authigenesis in this chemical sedimentary succession is consistent with stratigraphic condensation in a late transgressive systems tract, suggesting its upper contact is the maximum flooding surface (MFS; Catuneanu et al., 2011; Föllmi, 2016; Catuneanu, 2019).

Parasequence 2 rests sharply on Parasequence 1 and is interpreted to record progradation of clastic middle shelf deposits during early highstand conditions. Unlike Parasequence 1, Parasequence 2 lacks an authigenic overprint and is purely clastic in nature. Parasequence 2 coarsens upward from micaceous shale (F2) and interbedded siltstone (F1), reflecting accumulation below storm wave base, to an HCS sublitharenite (F3) recording deposition between storm and fair weather wave bases. The top of Parasequence 2 is the modern erosion surface.

The link between ironstone deposition and maximum flooding is observed in other Paleozoic ironstones (Young, 1992; Todd et al., 2019; Pufahl et al., 2020). The MFS marks the deepest water facies in a sequence and the change from retrogradational to progradational parasequence stacking (Catuneanu et al., 2011; Catuneanu, 2019). Because the MFS represents the time at which the accommodation is greatest, siliciclastics are trapped in nearshore environments to favour stratigraphic condensation and authigenesis on the distal shelf (Glenn et al., 1994; Taylor and Macquaker, 2000; Pufahl and Groat, 2017). Such low rates of clastic sedimentation stabilize the zone of authigenesis beneath the seafloor (cf. Föllmi, 2016) and permit waves to rework precipitating authigenic muds and coated Fe grains into granular deposits (Todd et al., 2019; Pufahl et al., 2020; Matheson and Frank, 2021).

Authigenic lithofacies (F5, F6, F7) composing Parasequence 1 contains Fe₂O_{3 total}, SiO₂, and Al₂O₃ concentrations that total >80 wt% (Table 1), reflecting the abundance of goethite, hematite, chamosite, metamorphic magnetite, and chert. TOC concentrations are highest at the base of Parasequence 1 (F4) with a maximum value of 4.6 wt% (Table 1). P₂O₅ concentrations are greatest in the middle of Parasequence 1 (F2) with values that range between 0.6 and 8.9 wt% (Table 1). TiO₂ and K₂O concentrations are low and <1 wt% through the thickness of this parasequence (Table 1). In siliciclastic facies (F1, F2, F3) of Parasequence 2, TiO₂ and K₂O concentrations are higher than in Parasequence 1 with maximum values of 1.02 wt% and wt. 3.65%, respectively (Table 1).

Total REE + Y concentrations in all lithofacies vary between 178 and 560 ppm Table 2). Authigenic facies (F5, F6, F7) of Parasequence 1 are enriched in middle rare earth elements (MREE) and heavy rare earth elements (HREE). Ce/Ce* and Eu/Eu* values in both parasequences range from 0.801 to 1.25 and 0.606 to 1.13, respectively. True negative Ce anomalies characterize the base of Parasequence 1 (Figure 8; Bau and Dulski, 1994; Bau et al., 1996; Planavsky et al., 2010). Ce/Ce* and Pr/Pr* values in bioturbated chamositic mudstone at the base (F5) range between 0.85 and 0.93 and between 1.06 and 1.13, respectively. Negative Eu anomalies occur in terrigenous clastic (F1, F2) lithofacies and samples of authigenic facies (F5, F6) with a significant detrital component. In samples of authigenic facies (F5, F6) devoid of siliciclastics and containing ≥1 wt% of synsedimentary CFA, minor positive Eu anomalies are common (Figure 9).

Average concentrations of the redox elements V, Cr, and Zn in authigenic facies are 393 ppm, 163 ppm, and 156 ppm, respectively, which are significantly higher than shale composite values of 130 ppm, 125 ppm, and 100 pm (Table 3; Morgan et al., 1978; Gromet et al., 1984; Condie, 1993). Other Eh sensitive elements such as Mo, Ni, U are near shale composite values, whereas Pb, Co, and Cu are lower (Figure 10; Morgan et al., 1978; Gromet et al., 1984; Condie 1993).

Lithofacies associations in the Hen-dy-Capel Ironstone Member are interpreted to record ironstone accumulation along the northern margin of the Rheic Ocean on a storm-dominated shelf with active coastal upwelling (Figure 12). Organic-rich mudstone, phosphatic nodules, and biogenic silica (F4, F5, F6, F7) are diagnostic of upwelling environments and a productive surface ocean (Jahnke et al., 1983; Behl and Garrison, 1994; Glenn et al., 1994; Sharp, 2007; Föllmi, 1996; Pufahl and Groat, 2017). Modern upwelling systems such as the Humboldt and Benguela current regions along the coasts of Peru and southern Africa, respectively, are characterized by this triad of sedimentary deposits (Suess and Von Huene, 1990; Wefer et al., 2002). Modelling of surface currents in the Rheic Ocean suggest upwelling in this narrow seaway was driven by east-blowing winds and Ekman transport along the southern margin of Avalonia, which formed the northern side of a cyclonic gyre (Figure 11; Pohl et al., 2016; Todd et al., 2019; Pufahl et al., 2020).

The REE + Y composition of ironstone lithofacies (F4, F5, F6, F7) corroborates sedimentologic data suggesting upwelling delivered a sustained supply of anoxic bottom water enriched in hydrothermal Fe to an otherwise oxygenated shelf. Near the upwelling front, where primary production and deposition of sedimentary organic matter were highest (Beckly, 1988 Woodcock, 1990; Brenchley et al., 2006; Howells, 2007), intense bacterial sulfate reduction in the sediment is interpreted to have produced a pyritiferous black shale (F4; Figure 12). Inboard, diminished primary production, advection of ferruginous waters away from the upwelling front, and the cyclic, benthic redox recycling of Fe (Raiswell, 2011) into shallower oxygenated environments are interpreted to have stimulated widespread authigenesis and ironstone accumulation (F5, F6, F7; Petránek, 1991; Todd et al., 2019; Pufahl et al., 2020; Matheson and Pufahl, 2021). The presence of HCS constrains the deposition of ironstone to water depths no deeper than ca. 50 m (Figure 12; Dumas and Arnott, 2006).

Such an interpretation is consistent with other upwelling-related ironstones that accumulated along the margins of the Rheic and Iapetus oceans (Petránek, 1991; Todd et al., 2019; Pufahl et al., 2020; Matheson and Pufahl, 2021). What is different about ironstone from the Welsh Basin is the ubiquity of large, coated Fe grains. An upwelled source of Fe contrasts the more traditional view of ironstone deposition that relies on a greenhouse climate and increased chemical weathering for the delivery of continentally derived Fe to the shelf (Van Houten and Bhattacharyya, 1982; Young, 1989, 1992; Van Houten, 2000; Yilmaz et al., 2015).

Ventilation of the deep ocean began during the Neoproterozoic Oxygenation Event (800–500 Ma; Och and Shields-Zhou, 2012; Scott & Lyons, 2012; Shields, 2017), but mounting evidence suggests that a persistently oxygenated deep ocean did not develop until the Permian (Elrick et al., 2011; Kah et al., 2016; Edwards et al., 2017; Lu et al., 2018; Edwards, 2019; Todd et al., 2019; Pufahl et al., 2020; Matheson and Pufahl, 2021). Modelling demonstrates that oxygen concentrations in the ocean-atmosphere system were probably lowest in the Middle Ordovician (Edwards et al., 2017). Occurrences of upwelling-related granular ironstone provide critical insights into this history because they capture nuances of ocean redox structure at the interface between shallow- and deep-water masses. Thus, these occurrences link shelf and abyssal processes to further illuminate the relationship between ocean circulation, seawater chemistry, nutrient availability, and biological evolution.

The ubiquity of upwelling-related ironstone in the Paleozoic suggests that bottom-water anoxia and input of hydrothermal Fe were common in many restricted seaways (Pufahl et al., 2020; Matheson and Pufahl, 2021). Pronounced episodes of ironstone accumulation in the Ordovician and Devonian (Van Houten and Arthur, 1989; Young, 1992) likely correspond to periods of plate reorganization and increased hydrothermal activity during mantle plume events (Murphy et al., 2009; Yang et al., 2012; Algeo et al., 2016). During these tectonic episodes, the input of volcanogenic CO₂ is thought to have acidified the global ocean (Veron, 2008; Kiessling and Simpson, 2011; Torsvik and Cocks, 2017; McKenzie and Jiang, 2019) and greenhouse warming decreased the solubility of O₂ in seawater (Veron, 2008; McKenzie and Jiang, 2019).

Ocean acidification is deleterious to calcareous organisms, causing brittle shells and partial dissolution (Kiessling and Simpson, 2011; Hönisch et al., 2012). In contrast, acidic seawater has a neutral to positive effect on the development of siliceous organisms (Webby, 2002; Kidder and Tomescu, 2016). In upwelling-related ironstone, the absence of a diverse assemblage of carbonate-producing organisms and abundance of sponges with siliceous spicules is consistent with this relationship. In the Welsh Basin, sponges, phosphatic brachiopods, and trilobites (Trythall et al., 1987) apparently flourished in waters too acidic to support a healthy carbonate community. The absence of carbonate skeletons in nutrient-rich environments may also reflect a dysoxic seafloor and dissolution associated with the microbial respiration of accumulating organic matter (James et al., 2005). CFA nodules and cortical layers composed of coated Fe grains record this bacterial process, which also liberates P from degrading organic matter (Jarvis et al., 1994; Pufahl and Groat, 2017). Sponges thrive under such low Eh and Ph conditions and therefore can dominate productive middle and distal shelf environments (Mills et al., 2014; Leys and Kahn, 2018; Matheson and Frank, 2020; Chen et al., 2021).

Under a greenhouse climate and diminished capacity for warmer seawater to retain oxygen, anoxic conditions were probably readily established in the narrow, restricted seaways of the Paleozoic. Such anoxia was a pre-requisite for the concentration of hydrothermal Fe and deposition of ironstone in these volcanically active basins. In addition to Fe, upwelling of these ferruginous waters would have delivered a sustained supply of deleterious trace elements to the shelf (Cd, Cu, As, Zn, Co, Ni, Se, Cr, Ba, Ge, As, Pd, Te, and REEs; Wilde et al., 1990), some of which are elevated in the Hen-dy-Capel Ironstone Member. In conjunction with the impingement of anoxic seawater on the shelf, the availability of these toxic trace elements is interpreted to have triggered regional extinctions that punctuate major biological diversification events such as the GOBE (Vandenbroucke et al., 2015; Pufahl et al., 2020; Matheson and Pufahl, 2021). The accumulation of upwelling-related ironstone in periods of post-extinction recovery suggests that over time ironstone accumulation was also an important sink that lowered the metal toxicity of seawater (Pufahl et al., 2020). The cyclic changes in pore water redox potential required to sequester these trace elements through repeated episodes of scavenging and absorption at the seafloor are preserved in the cortical mineralogy of coated Fe grains. This “kill and cure” mechanism (Pufahl et al., 2020) is interpreted to have contributed to pulses of graptolite diversification in the Welsh Basin (Kokelaar et al., 1985; Brenchley et al., 2006; Howells, 2007) by providing upwelling derived nutrients (Young et al., 2002; Pohl et al., 2016) and lowering metal toxicity (Matheson and Pufahl, 2021). This negative feedback process is thought to have contributed to other post-extinction faunal recoveries that assisted with increasing marine biodiversity through the GOBE (Pufahl et al., 2020). A similar feedback may have also been important in the Devonian when biological turnover was also linked to environmental volatility and periods of shelf anoxia (Bond and Wignall, 2008; Brett et al., 2020). Application of this upwelling model to even younger ironstones and contemporaneous trace element enriched black shales may also provide important new constraints regarding the redox structure and oxygenation history of the Mesozoic oceans.

1) Ironstone of the Hen-dy-Capel Ironstone Member of the Tygarn Formation, Ogwen Group, constitutes a single shallowing upward parasequence (Parasequence 1) recording stratigraphic condensation during maximum transgression. Overlying terrigenous clastic facies form a second parasequence (Parasequence 2) interpreted to record the onset of progradation during highstand conditions.