1. Introduction
Unravelling the causes and consequences of major rapid environmental changes requires a detailed understanding of the timing of key events, often observed in sedimentary basins on different continents. The timing of events at the Triassic–Jurassic boundary (the Rhaetian–Hettangian boundary) has been increasingly refined using a combination of carbon isotope stratigraphy, geochronology, magnetostratigraphy and biostratigraphy. This has allowed the timing of the eruption of the Central Atlantic Magmatic Province (CAMP) flood basalts and events in marine sediments to be better understood (Clémence et al., Reference Clémence, Bartolini, Gardin, Paris, Beaumont and Page2010; Boomer et al., Reference Boomer, Copestake, Raine, Azmi, Fenton, Page and O’Callaghan2021; Lindström et al., Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017, Reference Lindström, Pedersen, Vosgerau, Hovikoski, Dybkjær and Nielsen2023; Yager et al., Reference Yager, West, Thibodeau, Corsetti, Rigo, Berelson, Bottjer, Greene, Ibarra, Jadoul and Ritterbush2021; Zaffani et al., Reference Zaffani, Jadoul and Rigo2018), suggesting a two-phased extinction (Lindström, Reference Lindström2021). However, some datasets suggest alternative scenarios (Fox et al., Reference Fox, Cui, Whiteside, Olsen, Summons and Grice2020; Beith et al., Reference Beith, Fox, Marshall and Whiteside2023).
In the UK, the Norian–Rhaetian to early Hettangian interval is a transition from red-green playa mudstones in the upper part of the Mercia Mudstone Group (MMG), though the Penarth Group (Benton et al., Reference Benton, Cook and Turner2002; Gallois, 2007; Reference Gallois2009) and into the marine units of the lowest Lias Group (Gp; Fig. 1b). This lithostratigraphical framework is generally consistent throughout England (Benton et al., Reference Benton, Cook and Turner2002), Northern Ireland (Warrington, Reference Warrington, Meadows, Trueblood, Hardman and Cowan1997) and into the southern North Sea (Lott and Warrington, Reference Lott and Warrington1988; Johnson et al., Reference Johnson, Warrington, Stoker, Knox and Cordey1994; Barnasch et al., Reference Barnasch, Geluk, Warrington, Hauschke, Franz and Bachmann2021). Comparable units occur in the central and eastern parts of the southern Permian Basin (Bachmann et al., Reference Bachmann, Geluk, Warrington, Becker-Roman, Beutler, Hagdorn, Hounslow, Nitsch, Röhling, Simon, Szulc, Doornenbal and Stevenson2010; Fig. 1b). Some of these Rhaetian-age units contain macrofossils, but these do not generally provide sufficiently precise biostratigraphic dating. However, Rhaetian palynostratigraphy allows fairly detailed correlations within NW Europe (Bonis et al., Reference Bonis, Ruhl and Kürschner2010; Lindström & Erlström, Reference Lindström and Erlström2006; Kürschner & Herngreen, Reference Kürschner, Herngreen and Lucas2010; Lindström et al., Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017), although magnetostratigraphy and carbon isotope stratigraphy have the potential to provide a finer-scale chronostratigraphy and correlation in the Rhaetian (Gallet et al., Reference Gallet, Krystyn, Marcoux and Besse2007; Kent et al., Reference Kent, Olsen and Muttoni2017; Hounslow & Andrews, Reference Hounslow and Andrews2024). In the UK, Hettangian units are well-dated by ammonoids, except for the lower parts of the Tilmanni Chronozone (Cz) of the basal Hettangian (Page, Reference Page2003, Reference Page, Lord and Davis2010; Weedon et al., Reference Weedon, Jenkyns and Page2018, Reference Weedon, Page and Jenkyns2019), the base of which outside the GSSP at Kuhjoch can only be inferred by correlations using carbon isotopes, miospores and geochemical datasets.
a) Summary map of locations and environmental facies for the interval occupied by the Branscombe Mudstone Fm and its equivalents. The Staithes S-20 borehole is located at Boulby. Base map modified from Geluk (Reference Geluk2005), with facies concepts from McKie (Reference McKie, Martinius, Ravnås, Howell, Steel and Wonham2014); b) Summary lithostratigraphy for England from this work, with that for the southern North Sea region from Cameron et al. (Reference Cameron, Crosby, Balson, Jeffrey, Lott, Bulat and Harrison1992); other correlations and disconformities based on Barnasch (Reference Barnasch2010), Hounslow & Andrews (Reference Hounslow and Andrews2024) and this work. Summary polarity and substage scale from Hounslow & Gallois (Reference Hounslow and Gallois2023). Numbered subdivisions of the Rhaetian based on Krystyn (Reference Krystyn2008). NRB1 and NRB2 are the two proposed options for the position of the Norian–Rhaetian boundary.
The late Rhaetian–early Hettangian succession in southern England and Northern Ireland has been well studied using a variety of methods (Boomer et al., Reference Boomer, Copestake, Raine, Azmi, Fenton, Page and O’Callaghan2021), but comparatively little detailed work has been carried out on successions in northern England, Scotland or the UK North Sea. This omission is addressed here using magnetostratigraphy, biostratigraphy and astrochronology of the succession from the Staithes No. 20 core from the Boulby Mine in NE England (hereafter called S-20; Fig. 1a). The Norian-age eastern England successions were key for detailed initial studies of this interval in the UK (Elliott, Reference Elliott1961; Taylor, Reference Taylor1982), but the paucity of later studies was superseded by work on better outcrop exposures in SW England. The Rhaetian age units in eastern and NE England have remained less studied, since these are largely known from boreholes and temporary exposures (Kent Reference Kent1953; Reference Kent, Sylvester-Bradley and Ford1968). An aim of this study is to clarify the regional relationships of Norian and Rhaetian units in NE and eastern England, which have been correlated using borehole logs across the southern North Sea to Germany (Barnasch, Reference Barnasch2010). In addition, magnetic susceptibility data from the earliest Hettangian part of the S-20 core allows a re-evaluation of conflicting views (Ruhl et al., Reference Ruhl, Deenen, Abels, Bonis, Krijgsman and Kürschner2010; Weedon et al., Reference Weedon, Page and Jenkyns2019) on the cyclostratigraphy and duration of the early Hettangian, specifically for the Tilmanni Cz and the Planorbis Subchronozone (Scz).
1.a. Uncertainty in the definition of the Norian–Rhaetian boundary
Where to place the Norian–Rhaetian boundary is undecided, and two proposals have been suggested (positions here called NRB1 and NRB2; Fig. 1b). These differ significantly in chronostratigraphic position and use different points in the morphological change of the conodont Misikella posthernsteini from its ancestor Misikella hernsteini (Galbrun et al., Reference Galbrun, Boulila, Krystyn, Richoz, Gardin, Bartolini and Maslo2020). Subdivisions of the Rhaetian into informal units 1 to 4 are used here, based on the conodont ranges and zones of Krystyn (Reference Krystyn2008) and Galbrun et al. (Reference Galbrun, Boulila, Krystyn, Richoz, Gardin, Bartolini and Maslo2020). Rhaetian-1 starts from the base of the Rhaetian as defined by Krystyn et al. (Reference Krystyn, Richoz, Gallet, Bouquerel, Kürschner and Spötl2007) and referred to as NRB1 and ranges through the base of the Rhaetian as proposed by Rigo et al. (Reference Rigo, Bertinelli, Concheri, Gattolin, Godfrey, Katz, Maron, Mietto, Muttoni, Sprovieri and Stellin2016), referred to as NRB2; this is the Epigondolella bidentata–M. posthernsteini Zone. Rhaetian-2 is the M. posthernsteini–M. hernsteini Zone, Rhaetian-3, is the M. rhaetica Zone and Rhaetian-4 the M. ultima Zone, to the base of the Hettangian. Other conodont-based subdivisions of the Rhaetian have been proposed (Rigo et al., Reference Rigo, Mazza, Karádi, Nicora and Tanner2018), but that of Krystyn (Reference Krystyn2008) is used because most Rhaetian conodont-dated magnetostratigraphies have used the zonations of Gallet et al. (Reference Gallet, Krystyn, Marcoux and Besse2007) and Krystyn (Reference Krystyn2008).
1.b. Conflicting astrochronological durations of the Hettangian
Floating astrochronologies for the duration of the Hettangian from the Blue Lias Formation (Fm) have given conflicting interpretations. The multi-section studies of Weedon et al. (Reference Weedon, Jenkyns and Page2018, Reference Weedon, Page and Jenkyns2019) suggested a ≥ 2.9 Myr Hettangian duration if using individual sections or ≥4.1 Myr when splicing together multiple sections, accounting for missing/condensed ammonite biohorizons. Tuning of the lowest observed frequency (statistically significant cycle) to the ∼100 kyr short eccentricity (E2–3) astronomical period was used in these studies. Obliquity and precession cycles were also inferred, with the E2–3 cycles containing up to four limestones per cycle. This analysis was based on 2 to 4 cm spaced measurements of surface magnetic susceptibility (Ksurf). The inferred environmental model for the connection between lithology (i.e., carbonate content) and astronomical cycles was via the degree of storminess controlling sea-floor turbulence and limestone formation (Weedon et al., Reference Weedon, Jenkyns and Page2018).
In contrast, Ruhl et al. (Reference Ruhl, Deenen, Abels, Bonis, Krijgsman and Kürschner2010) and Hüsing et al. (Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014) utilised CaCO3 content, total organic carbon (TOC) and δ13Corg (20 to 30 cm spaced sampling) to identify E2–3 cycles, based around visual interpretation of the bundling patterns of limestones and black shale beds in the St Audrie’s Bay (StAB) section. The environmental interpretation was based principally on analogy with Neogene Mediterranean sapropels (Hüsing et al., Reference Hüsing, Hilgen, Abdul Aziz and Krijgsman2007), which were equated to black shales in the succession and were inferred to be precession forced. In this model, sapropels/black shales matched summer insolation maxima (Hüsing et al., Reference Hüsing, Hilgen, Abdul Aziz and Krijgsman2007), with bottom water anoxia corresponding to peaks in freshwater run-off (Bosmans et al., Reference Bosmans, Drijfhout, Tuenter, Hilgen, Lourens and Rohling2015). These authors also observed splitting of the E2–3 band in power spectra, probably because of accumulation rate changes (Hüsing et al., Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014), a feature widely inferred in the Blue Lias Fm (Weedon et al., Reference Weedon, Jenkyns, Coe and Hesselbo1999). The Hüsing et al. (Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014) c. 1.7 kyr duration estimated for the Hettangian is approximately supported by radioisotopic dates from sections in Peru (suggesting 1.93 ± 0.27 Myr for the Hettangian; Geux et al., 2012) and was additionally correlated (using magnetostratigraphy) with the astrochronology from the continental-lacustrine succession from the Hartford Basin (eastern USA). It was also linked to an astrochronology of the Pliensbachian (Xu et al., Reference Xu, Ruhl, Hesselbo, Riding and Jenkyns2017; Ruhl et al., Reference Ruhl, Hesselbo, Hinnov, Jenkyns, Xu, Riding, Storm, Minisini, Ullmann and Leng2016). Although, as argued by Weedon et al. (Reference Weedon, Page and Jenkyns2019), the Peruvian ammonite data suggest that the radioisotopically dated range may not bracket the entire Hettangian due to misidentification of some of the ammonites and potential reworking of the dated zircons.
The astrochronological assessment presented here focuses on the early Hettangian (Tilmanni Cz and Planorbis Scz), which were identified as more problematic in prior studies due to a lack of lithological bundling in the StAB section (Hüsing et al., Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014). Instead, in the present account, a statistical inference approach is used that includes accumulation rate changes, a factor that was not previously considered in detail but is clearly important in limestone-marl rhythmites (Moghadan & Paul, Reference Moghadam and Paul2000; Arzani, Reference Arzani2006; Westphal et al., Reference Westphal, Munnecke, Böhm, Bornholdt, Algeo, Heckel, Maynard, Blakey, Rowe, Pratt and Holmden2008; Paul et al., Reference Paul, Allison and Brett2008; Brett et al., Reference Brett, Allison, Hendy, Allison and Bottjer2011). Unlike many astrochronologic studies, which utilise single-section/core datasets (excepting Weedon et al., Reference Weedon, Page and Jenkyns2019), a new approach is developed which utilises multi-section data to derive a composite solution with duration uncertainty.
2. Geology of the Staithes S-20 core
The Staithes S-20 borehole (location NZ 76034 18000, at the Boulby Mine, Staithes, NE Yorkshire; Fig. 1a) was completed in December 1968 and recovered the succession from the Upper Lias to the top part of the Zechstein (Woods, Reference Woods1973). Depths are given here in both feet and metres since the core is labelled in (imperial) feet. The bedding dip is near zero. The core was sampled for clay mineralogy by Jeans et al. (Reference Jeans, Mitchell, Scherer and Fisher1994) and Jeans (Reference Jeans, Dunay and Hailwood1995) and for δ34S measurements from evaporites by Salisbury et al. (Reference Salisbury, Gröcke, Cheung, Kump, McKie and Ruffell2022, Reference Salisbury, Gröcke and McKie2023). Southworth (Reference Southworth1987) also used the S-20 core as a reference section for the Middle Triassic successions of the East Midlands Shelf (Fig. 1a). Prior to the sampling for magnetostratigraphy and palynology in 1998, there had been little work on the core, which was then largely in good condition.
For the present magnetostratigraphical work, samples were taken from the succession between 357.3 m and 425.6 m (1172 ft and 1396 ft), through 17.6 m of the basal Redcar Mudstone Fm (Powell Reference Powell1986; Atkinson et al., Reference Atkinson, Wignall and Page2020) of the Lias Gp; 16.6 m of the Penarth Gp; and the upper 34.1 m of the MMG (comprising the Blue Anchor Fm at 5.7 m) and the upper 28.4 m of the Branscombe Mudstone Fm (Fig. 2a). A detailed log of the sampled interval is given in the Supplementary Material (SM Fig. S3).
Summary of petromagnetic data for the Staithes S-20 core. a) Summary lithologic log (see SM Fig. S3 for details); b) surface (Ksurf) and c) specimen volumetric magnetic susceptibility; d) natural remanent magnetisation (NRM) intensity; e) lithostratigraphy and ammonite biostratigraphy. Var.= variegated interval, CM=Cotham Member, LM=Langport Member, Cz=chronozone, Pl.=Planorbis Cz. Hatching in column (e) indicates uncertainty in the position of the base of the Hettangian.
The Branscombe Mudstone Fm is dominated by pale red to dark-red mudstones, occasionally laminated, with anhydrite common as nodules, sometimes with chicken-wire texture or veins from 425.8 to 410.0 m (1397 ft to 1345 ft; Fig. 2a, SM Figs. S1g, S3d,e,f). Above 410.0 m (1345 ft), anhydrite occurs more typically as beds 0.07–0.6 m in thickness, with subordinate nodular anhydrite. The upper boundary of the Branscombe Mudstone Fm is at 1303.1 ft (397.18 m), at the top of the last major (∼0.3 m thick) red mudstone, which overlies an anhydritic bed (SM Figs. S1f, S3d). Between 1303.1 and 1311 ft (397.18 m–399.6 m), red mudstones are interbedded with subordinate green-grey blocky mudstones in a ‘variegated interval’. Such units are typical of the uppermost part of the Branscombe Mudstone Fm in sections in SW England (Mayall, Reference Mayall1981; Gallois, Reference Gallois2001, Hounslow et al., Reference Hounslow, Posen and Warrington2004; Howard et al., Reference Howard, Warrington, Ambrose and Rees2008). However, in S-20 the ‘variegated interval’ is only 2.4 m thick in contrast to the ∼20 m present in sections at StAB and Haven Cliff at Seaton (Fig. 1a). The sampled part of the Branscombe Mudstone Fm can be equated to the bulk of the Keuper Anhydrite Member in offshore terminology (Fig. 1b), with the equivalent base of this unit at about 434.34 m (1425 ft) in the S-20 core. A low in δ34S identified by Salisbury et al. (Reference Salisbury, Gröcke, Cheung, Kump, McKie and Ruffell2022, fig. S2) occurs at ∼417.6 m (∼1370 ft).
The Blue Anchor Fm is dominated by grey to dark grey mudstones, with its top marked by 0.3 m of pyritic mudstone overlain by a bed with sandstone flasers, inferred to be at the base of the Westbury Fm (Fig. 2a, SM Fig. S1e). The Blue Anchor Fm in S-20 does not have lithological divisions corresponding to the Rydon and overlying Williton members as in sections such as StAB (Figs. 1a, b) in west Somerset (Mayall, Reference Mayall1981).
The base of the Westbury Formation (WFm) is uneven, with fragments of Blue Anchor Fm lithology in darker mudstones (SM Fig. S1e). The WFm is dominated by dark grey to black shaley mudstone, with a sandstone-dominated unit at 388.43–386.33 m (1274.4–1267.5 ft). Below 384.96 m (1263 ft) are common cm-scale disrupted beds of grey mudstone with sandstone clasts and bioturbated sandstone beds (Fig. 2a, SM Fig. S1c). At 387.20 m (1270.33ft; SM Fig. S1d), a bioturbated and irregular surface with meniscus burrow fills like those seen at the base of the WFm in sections in SW England (Mayall, Reference Mayall1981; Gallois, Reference Gallois2007) may mark a significant hiatus.
The base of the Cotham Member (CMbr) comprises 15 cm of silty sandstone with an irregular bed contact on the underlying mudstone (SM Fig. S1b), possibly marking a disconformity as in sections in SW England (Gallois, Reference Gallois2009). The remainder of this member is dark grey and grey, weakly fissile mudstones, with some reddish-grey mudstones at 381.30 m (1251 ft). The prominent desiccation surface present in the middle of the member in SW England is not present, but darker mudstones are overlain by paler mudstones, a transition which may distinguish lower and upper divisions of the member in the core (SM Fig. S3c). The overlying Langport Mbr (Fig. 2e; SM Fig. S3b) is a well-cemented grey silty-calcareous mudstone, with mm-scale sandstone flasers and lenses. It is divisible into lower and upper units, with the lower unit having more mm-scale sandy layers. The base of the Redcar Mudstone Fm (in the Lias Gp) is marked by a grey, laminated sandy-silty shale with much pyrite, possibly representing a transgressive feature (∼12 cm thick; SM Fig. S1a), but with regular upper and lower contacts. The remainder of this formation consists of black to dark grey shaley mudstone and dark grey to black mudstone, with occasional laminated beds at 371.6–370.1 m (1219–1214 ft). Scattered bivalves and shell-lags progressively increase in abundance up to 364.85 m (1197 ft), above which they are very common (SM Fig. S3a).
3. Methods
3.a. Sampling for magnetostratigraphy and palynology
The S-20 core is 100 mm in diameter and non-slabbed, but at the time of sampling, some parts were fragmented (particularly in the WFm). Re-orientation of the core was attempted by first reassembling the pieces into continuous runs (Hailwood & Ding, Reference Hailwood, Ding, Turner and Turner1995) by rotating and fitting end-pieces together to find a common reference fiducial for as many pieces as possible. This was not possible for the entire core, but 93 core runs were assembled. A few core pieces were found to be inverted, and others were misplaced in the core boxes. From the longer core runs, sample slices were dry cut with a diamond saw and prepared into 2 to 3 cm cubic palaeomagnetic specimens with a mean sample spacing of ∼0.5 m for most of the core. Sampling levels and core runs are marked in SM Figure S3.
The core was sampled for palynology at 29 levels in the upper part of the MMG, the Penarth Gp and the lower part of the Lias Gp, with miospore preparations made at the British Geological Survey (BGS). The spore and pollen parent plant affinity is based on Bonis (Reference Bonis2010), Lindström et al. (Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017) and Gravendyk (Reference Gravendyk2021). The Eco-Plant model classifications of Zhang et al. (Reference Zhang, Lenz, Wang and Hornung2021) were used to assign humidity (EPH) and temperature (EGT) classes to each taxon, using the www.sporopollen.com database (31% of miospore taxa in the core have unclear or unknown Eco-Plant model assignment). For the three major negative carbon isotope excursions (CIE) around the Rhaetian–Hettangian boundary, the names proposed by Lindström et al. (Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017) are used, namely Marshi CIE, Spelae CIE and top Tilmanni CIE.
3.b. Magnetic methods
Prior to any sampling, surface magnetic susceptibility (Ksurf) was measured on the core surface, with a procedure detailed in SM Section 1. The palaeomagnetic measurement procedures follow those used by Hounslow et al. (Reference Hounslow, Posen and Warrington2004), Hüsing et al. (Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014), Hounslow & Gallois (Reference Hounslow and Gallois2023), and Hounslow & Andrews (Reference Hounslow and Andrews2024) on the same formations.
The palaeomagnetic data were analysed in a similar way to Hounslow et al. (Reference Hounslow, Posen and Warrington2004) and Hounslow & Gallois (Reference Hounslow and Gallois2023) using principal component fits with the LINEFIND software (Kent et al., Reference Kent, Briden and Mardia1983). Directional statistics used PalaeomagTools v5.1 (Hounslow, Reference Hounslow2023). The demagnetisation behaviour and nature of characteristic remanence were classified into either line fits (S-class behaviour) or great circle trends (T-class behaviour, fitted to a great circle plane). In each case, the degree of scatter was used for qualitative sub-division into three classes (T1 to T3 and S1 to S3), with T1 and S1 having the least scatter and being most well defined, and T3 and S3 having the most scatter and least well defined (following the procedures in Montgomery et al., Reference Montgomery, Hailwood, Gale and Burnett1998; Hounslow et al., Reference Hounslow, Posen and Warrington2004). A demagnetisation class of X was used for specimens which had either large directional scatter, were insufficiently demagnetised or were inferred to have no characteristic remanent magnetisation (ChRM). The inferred specimen polarity was divided into three quality categories for reverse (R, R? and R??) and normal polarity (N, N? and N??). A polarity category of U was used for specimen data in which the polarity could not be confidently assigned. The same categorisation scheme was used by Hounslow et al. (Reference Hounslow, Posen and Warrington2004), Hounslow & Gallois (Reference Hounslow and Gallois2023) and Hounslow & Andrews (Reference Hounslow and Andrews2024). Magnetic mineralogy has not been investigated in detail, as it is assumed to be like that of the same formations investigated by Hounslow (Reference Hounslow1985), Briden & Daniels (Reference Briden and Daniels1999), Hounslow et al. (Reference Hounslow, Posen and Warrington2004), Hüsing et al. (Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014), Hounslow & Gallois (Reference Hounslow and Gallois2023) and Hounslow & Andrews (Reference Hounslow and Andrews2024), data which are compared in SM Section 4.
3.c. Astrochronological methods and conceptual models of sediment accumulation rates (SAR)
Surface magnetic susceptibility (Ksurf) from the S-20 core was used for the astrochronology, covering the Langport Mbr and Redcar Mudstone Fm (Fig. 2b). For S-20, the Ksurf data were linearly detrended and interpolated using Hermite interpolation with third-order polynomials at the median spacing of 0.051 m (using the pchip function in R; R Core Team, 2013). The astrochronologic analysis principally uses the TimeOpt methodology (Meyers, Reference Meyers2014) but extends the TimeOptTemplate approach as described by Meyers (Reference Meyers2015, Reference Meyers2019), which in this case specifically modelled a more complex assemblage of sediment accumulation rate (SAR) changes. TimeOpt uses a measure of overall fit of r2opt, which is a multiplication of the fits for the amplitude envelope, r2envelope, and the spectral power fit r2spectral. The astrochronological analysis used the astrochron package in R (Meyers, Reference Meyers2014; R Core Team, 2013). A similar analysis was applied to Ksurf datasets from Weedon et al. (Reference Weedon, Jenkyns and Page2018, Reference Weedon, Page and Jenkyns2019), covering the equivalent stratigraphic interval at StAB, Lavernock and Lyme Regis (Figs. 1a, 3b, c, d). These Ksurf data are equally spaced and were not interpolated. For the early Hettangian, the fundamental secular frequencies (g1 to g5, precession rate k) were estimated by linear interpolation of the data in Zhou et al. (Reference Zhou, Wu, Hinnov, Fang, Zhang, Yang and Shi2022) and Meyers & Malinverno (Reference Meyers and Malinverno2018) at 201 Ma (SM Fig. S2). These were converted to periods using the astrochron package, with the used periods and interpolations shown in SM Table S1. Obliquity periods at 201 Ma used those of Berger et al. (Reference Berger, Loutre and Laskar1992).
Surface magnetic susceptibility (Ksurf) records from Langport Mbr and basal Lias Gp. Staithes S-20 from this work and others from Weedon et al. (Reference Weedon, Page and Jenkyns2019). The x-axis is an arbitrary scale with the base of the Lias Gp at zero and those of the Planorbis and Johnstoni subchronozones at 10 and 20, respectively (chronozonal thickness indicated assuming base Lias Group ≅ base Jurassic). Right hand scales are the ammonite biohorizons (orange symbols, and non-underlined bold numbers). Possible hiatus levels (marked as H?) from Weedon et al. (Reference Weedon, Page and Jenkyns2019). Black vertical lines connect biohorizon bases. The inferred short eccentricity cycle (E2–3) is marked within [ ] for St Audrie’s Bay (StAB) from Ruhl et al. (Reference Ruhl, Deenen, Abels, Bonis, Krijgsman and Kürschner2010) and by a red line with a tick for Lavernock (from Weedon et al., Reference Weedon, Page and Jenkyns2019). Grey bands are plausible correlations of the Ksurf changes constrained by position within the chronozones. Original error in exponent of Ksurf corrected for the StAB, Lyme Regis and Lavernock datasets. Bed numbers from Weedon et al. (Reference Weedon, Page and Jenkyns2019). On panel (b), the inferred positions of the base of the Hettangian correlated from the GSSP at Kuhjoch are: J1, J2, J3 (discussed by Jeram et al. Reference Jeram, Simms, Hesselbo and Raine2021), L1, L2 = Lindström et al. (2021) and Lindström et al. (Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017), respectively; C = Clemence et al. (Reference Clémence, Bartolini, Gardin, Paris, Beaumont and Page2010) (using first occurrence of Ps. spelae in the New York Canyon section); K = Korte et al. (Reference Korte, Ruhl, Palfy, Ullmann, Hesselbo, Sial, Gaucher, Ramkumar and Ferreira2019); R= Ruhl et al. (Reference Ruhl, Hesselbo, Al-Suwaidi, Jenkyns, Damborenea, Manceñido, Storm, Mather and Riccardi2020); W=Weedon et al. (Reference Weedon, Page and Jenkyns2019), and correlation (purple dotted line); ws=Whiteside et al. (Reference Whiteside, Olsen, Eglinton, Brookfield and Sambrotto2010). Positions located by bed-by-bed correlation between the slightly differing logs of Weedon et al. (Reference Weedon, Page and Jenkyns2019) and Hesselbo et al. (Reference Hesselbo, Robinson, Surlyk and Piasecki2002). The negative carbon isotope excursions (CIE) on b), c) are the positions of the Spelae ② and top Tilmanni ③ events. White bar in b) is the reverse polarity magnetozone SA5r (UT28r magnetochron). Position of the top Tilmanni δ13Ccarb CIE ③ on the Lavernock section from Korte et al. (Reference Korte, Hesselbo, Jenkyns, Rickaby and Spötl2009; Reference Korte, Ruhl, Palfy, Ullmann, Hesselbo, Sial, Gaucher, Ramkumar and Ferreira2019).
All previous work on cyclostratigraphy from the Blue Lias Fm uses a nominal constant SAR over ranges of ammonite chronozones. In the Redcar Mudstone Fm (and the Blue Lias), as well as longer-term changes in SAR, short-term SAR changes are probably present at the scale of the bedding (Weedon, Reference Weedon1986; Paul et al., Reference Paul, Allison and Brett2008; Brett et al., Reference Brett, Allison, Hendy, Allison and Bottjer2011). We use this starting premise to evaluate probable SAR changes as a framework for a more robust astrochronological assessment, consistent between the three sections and the S-20 core. This analysis is in three stages, involving progressively more complex SAR models.
Stage 1: Baseline SAR changes (SARbase) over a longer height scale than the bedding contrasts were derived from an assessment using evolutive methods of eTimeOpt (Meyers et al. Reference Meyers, Sageman and Hinnov2001; Meyers, Reference Meyers2019; Omar et al., Reference Omar, Da Silva and Yaich2021); the procedure is shown in Figure 4a. Evolutionary methods like eTimeOpt utilise wide data windows and are less sensitive to bedding scale variation, but SAR tracks are sensitive to the width of the data window used, so several height windows were used and a composite track produced (Fig. 4a).
Outline of astrochronologic processing steps used here in developing the sediment accumulation rate (SAR) models for the sections: a) The steps principally using evolutionary TimeOpt and selecting the baseline SAR model (SARbase), b) Steps in selecting the modulated SAR models, which may include hiatus levels (SARβ,H). The blue steps shown are for producing the β- testing, but equally apply to variable hiatus (symbolised as H in purple step)- and β-H testing. {±} indicates β can take positive or negative values for either the TS-SAR or CD-SAR type models. The exit condition from this loop is when any element of SARβ,H is at a minimum >0.
Stage 2: An additional modulation was applied to the better-performing SARbase models to simulate the additional effect of bedding-scale variation of SAR (referred to as SARmod; Fig. 4b). Various fractional contributions of this modulation (as measured by a value β) are added to SARbase.
Stage 3: Based on placement from the ammonite biostratigraphy (Weedon et al., Reference Weedon, Jenkyns and Page2018, Reference Weedon, Page and Jenkyns2019), plausible additional hiatus levels (symbolised as H) were inserted into the SARbase models. This, in effect, tested various degrees of condensation at the suspected hiatus positions, a process indicated in Figure 4b (with β=0) and referred to as hiatus-testing. Using hiatus with the SAR modulation gives a composite SAR model (SARβ,H), which includes baseline changes, hiatuses and bedding-scale SAR modulation. Two likely models of the bedding-scale SAR modulation were evaluated.
3.c.1 Modulated SAR variation models
As demonstrated by Meyers et al. (Reference Meyers, Sageman and Hinnov2001), Meyers & Sagerman (Reference Meyers and Sageman2004) and Meyers (Reference Meyers2019), bedding-scale changes in SAR can generate spectral leakage and the generation of additional harmonics, which can confound the identification of astronomical frequencies using spectral methods (discussed at length in Hilgen et al., Reference Hilgen, Hinnov, Abdul Aziz, Abels, Batenburg, Bosmans, de Boer, Hüsing, Kuiper, Lourens, Rivera, Smith, Bailey, Burgess and Fraser2015).
Modern hemipelagic carbonate–clastic rhythmites (such as those in which the siliciclastics are exclusively aeolian-derived) have their short-term accumulation rate changes controlled by clastic influx (e.g., Clemens & Prell, Reference Clemens and Prell1991), since carbonate production tends to be controlled by water temperature and nutrients, which can have short-term changes suppressed by the sluggishness of the oceans (Strasser, Reference Strasser and Montenari2018). Somewhat comparable clastic-influx type models for the Blue Lias Fm were implied by Ruhl et al. (Reference Ruhl, Deenen, Abels, Bonis, Krijgsman and Kürschner2010) with maximum siliciclastic input, higher TOC, more depleted δ13Corg and larger magnetic susceptibility being observed in shale beds. This was linked to enhanced terrestrial siliciclastic input, using the Mediterranean sapropel model as an analogue (Hüsing et al., Reference Hüsing, Hilgen, Abdul Aziz and Krijgsman2007). The converse model of Weedon et al. (Reference Weedon, Jenkyns and Page2018) had the same net SAR modulation effect, but the SAR minimum was interpreted as coincident with limestone beds due to winnowing of clay, with the carbonate mud being hemipelagic in origin (Weedon, Reference Weedon1986; Arzani, Reference Arzani2006) for the Blue Lias Fm, but perhaps with some shallower water-derived carbonate mud (Sheppard et al., Reference Sheppard, Houghton and Swan2006). The co-varying changes in δ13Corg and TOC indicate that, as in the geochemical tests suggested by Westphal et al. (Reference Westphal, Munnecke, Böhm, Bornholdt, Algeo, Heckel, Maynard, Blakey, Rowe, Pratt and Holmden2008), the lithological cyclicity is principally primary in origin, although this basic pattern is clearly modified by diagenesis (Weedon, Reference Weedon1986, Weedon et al. Reference Weedon, Jenkyns and Page2018; Moghadam & Paul, Reference Moghadam and Paul2000; Arzani, Reference Arzani2006; Bottrell & Raiswell, Reference Bottrell and Raiswell1989). These types of bedding-scale SAR changes are here referred to as the terrestrial-sapropel SAR model (or TS-SAR model).
The early cementation and low compaction of limestone beds is a widely agreed key observation from the Blue Lias Fm (Paul et al., Reference Paul, Allison and Brett2008; Weedon et al., Reference Weedon, Jenkyns and Page2018). In comparison with shales and marls/shales with a greater compaction, this would increase relative SAR in the limestones (Westphal et al., Reference Westphal, Munnecke, Böhm, Bornholdt, Algeo, Heckel, Maynard, Blakey, Rowe, Pratt and Holmden2008). Irrespective of any original SAR changes, if compactional differences between limestones and marls/shales were sufficiently large, it could be that limestones have effectively larger SAR than the other lithologies. Bottrell & Raiswell (Reference Bottrell and Raiswell1989) have proposed a similar possibility with the marls/shales representing the lowest SAR via a linked geochemical model for limestone formation. This is here called the compaction-diagenesis model of SAR changes (or CD-SAR model).
Since Ksurf is strongly related to carbonate content (Weedon et al., Reference Weedon, Jenkyns and Page2018) and hence is a proxy for %CaCO3, it is therefore used here as a proxy for SAR mod. A positive relationship between Ksurf and SAR mod corresponds to the TS-SAR model (high Ksurf = high SAR), and a negative relationship between Ksurf and SAR mod corresponds to the CD-SAR model (high Ksurf = low SAR). For each section, numerically, the SAR mod used is the zero mean and 1σ transformed value of the Ksurf. This transformation enables comparable inter-section modulated values of SAR mod and allows it to be simply added to the more slowly changing SAR base.
Formally, the composite modulated SAR (SAR β) vector for each section (i.e., at each sampling height or depth) is based on SAR mod and SAR base, such that:
(SAR β,H, when hiatus is added)
Varying degrees of SAR modulation are applied by differing β from 0 to the minimum possible SARβ such that min(SARβ) >0. For the TS-SAR model, β is positive, and for the CD-SAR model, it is negative. This procedure is referred to as β-testing. When β = 0, no SAR modulation is used, and only the baseline SAR (SARbase) is applied. Any hiatus (H) can be inserted into SAR β, giving an SAR model with both modulation and hiatus (i.e., SAR β,H, Fig. 4b). When β = 0 and hiatuses are added, this procedure is referred to as hiatus-testing, and β-H testing when both SAR modulation and hiatus are used simultaneously. In each case, the final SAR β,H vector was converted to total duration and scaled to the height scale of the data (in meters; Fig. 4b). TimeOpt was then used to find the maximum value of r2opt for various β or H (or simultaneously β and H) at the optimum astronomical match, from which the optimum SAR vector is obtained, SAR opt. Scaling of the derived duration to section/core height allows the SAR opt value derived from TimeOpt to be compared with the various values of β and/or H. When both H and β are investigated simultaneously, these produce maps of r2opt with H and β (H−β maps), with the H−β position of peaks in r2opt refined by homing in on the r2opt highs identified at lower H-β resolutions. Examples of hiatus-testing are shown in SM Figs. S13 to 15, and for β - H testing in SM Figures S28 to S30. This approach is an extension of the timeOptTemplate method in the astrochron package, which simply stretches or shrinks an SAR template linearly to find the maximum r2opt. The R-scripts used are contained with the datasets (Hounslow, Reference Hounslow2025).
These evaluation processes do not lead to unique solutions for individual sections but to several possible solutions for a range of SAR scenarios. To find optimum solutions, further constraints are applied such that 1: any plausible fits should yield the larger r2opt, 2: the predicted durations of the Tilmanni Cz and Planorbis Scz should be consistent between the four datasets, and 3: the average SAR in each section should be consistent with the SAR estimated from the duration of the Tilmanni and Planorbis chronozones derived from external radioisotopic dates. External SAR constraints are detailed below. These three types of constraints therefore test the fit of the astronomical data, provide an intersection internal consistency check and an external duration constraint.
4. Results
4.a. Palynology of S-20
The palynology of the latest Triassic to earliest Jurassic succession in S-20 (studied by GW) is the most northerly such record from these successions in eastern England (see the regional review of palynology and taxonomic comments in SM Section 2). Miospore recovery from the highest beds of the Branscombe Mudstone Fm and Blue Anchor Fm in S-20 was poor in comparison with coeval sections in Somerset, but richer assemblages appear at the base of the WFm in S-20 and increase in variety upwards (Fig. 5a, SM Fig. S4). Overall, the miospore assemblages from S-20 are like those from StAB (Hounslow et al., Reference Hounslow, Posen and Warrington2004), and four assemblage zones (SAB1 to 4) recognised in that succession by Bonis et al. (Reference Bonis, Ruhl and Kürschner2010) can be applied in S-20 (Fig. 5a). The assemblages in these four zones have been interpreted as comprising palynofloras representing pre-extinction, extinction, recovery and post-extinction phases, respectively (Lindström, Reference Lindström2016; Lindström et al. Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017).
Summary palynomorph dataset from the Staithes S-20 core (full data in SM Fig. S4): a) selected palynomorphs from S-20 with miospore assemblage zones (SAB1 to 4) adapted from the St Audrie’s Bay data of Bonis et al. (Reference Bonis, Ruhl and Kürschner2010), and rarity intervals (MR1, MR2) inferred from criteria in Lindström (Reference Lindström2021). b) relative numbers of pollen and spore taxa, and the probable parent plant groups, c) the average values for Eco-plant model EGT and EPH proxies of the pollen and spore taxa. See SM section 2 for discussion of taxonomic issues.
4.a.1. SAB1 assemblage zone (pre-extinction phase)
At StAB (Bonis et al., Reference Bonis, Ruhl and Kürschner2010), the SAB1 Assemblage Zone (Az) extends from the base of the Williton Mbr to the top of the WFm. SAB1 associations are dominated by ‘Classopollis’ (herein Gliscopollis) meyeriana and other circumpolles, Ovalipollis pseudoalatus, Ricciisporites tuberculatus and Rhaetipollis germanicus; O. pseudoalatus is commoner in the lower part, and small numbers of Granuloperculatipollis rudis and Quadraeculina anellaeformis occur. The last occurrence (LO) of Enzonalasporites vigens is ∼2 m above the base of SAB1. Less diverse associations dominated by G. rudis and Classpollis sp., and with small numbers of Leptolepidites argenteaeformis. O. pseudoalatus, Vesicaspora fuscus and R. germanicus, were recorded from the Rydon Mbr, underlying the Williton Mbr at StAB (Warrington & Whittaker Reference Warrington and Whittaker1984; Warrington in Hounslow et al. Reference Hounslow, Posen and Warrington2004).
In S-20 associations from the WFm (Figs. 5a, SM S4) are like those in SAB1 at StAB, but with high numbers of R. tuberculatus, G. meyeriana and other circumpolles; O. pseudoalatus, Geopollis zwolinskae, R. germanicus and Deltoidospora spp. with each typically comprises at least 5% of these associations; G. zwolinskae was only recorded in SAB1 at S-20. Small numbers of Zebrasporites laevigatus, Vesicaspora fuscus, Limbosporites lundbladiae, Microreticulatisporites fuscus and Cingulizonates rhaeticus are present, and Lunatisporites rhaeticus, Acanthotriletes varius, G. rudis, Convolutispora microrugulata and Semiretisporites gothae appear progressively upwards in the section; G. rudis was not noted above this Az.
4.a.2. SAB2 assemblage zone (extinction phase)
At StAB (Bonis et al., Reference Bonis, Ruhl and Kürschner2010), the SAB2 Az extends from the base of the CMbr into the upper part of that member, below the hiatus/desiccation surface. The base of this Az is marked by an increase in the numbers of a wide range of spores and in the abundance of Vitreisporites spp. and Tsugaepollenites pseudomassulae. The LOs of Ovalipollis pseudoalatus, Rhaetipollis germanicus and Lunatisporites rhaeticus are at the top of this Az. From closely spaced samples, Bonis et al. (Reference Bonis, Ruhl and Kürschner2010) recognised two peaks in spore abundance in SAB2. Components of the lower peak include Porcellispora longdonensis, ‘Heliosporites’ (herein Kraeuselisporites) reissingeri, Deltoidospora spp., Concavisporites spp., Carnisporites anteriscus and Todisporites spp.; the main components of the upper peak are Calamospora tener, Deltoidspora spp. and the bryophyte spore P. longdonensis, the acme of which is in the upper peak.
In S-20, the miospore association from sample MPA 45633 is interpreted as the lowest in the SAB2 Az. This sample (Figs. 5a, SM S4) is likewise dominated by circumpolles and R. tuberculatus but fewer O. pseudoalatus and R. germanicus are present. Other features in higher samples in SAB2 are the absence of G. rudis, increases in the numbers of Deltoidospora spp., Convolutispora microrugulata and Perinopollenites elatoides, and the incoming of small numbers of Calamospora tener and definite specimens of Zebrasporites interscriptus and Kyrtomisporis spp., and species of other spore genera including Carnisporites and Densosporites. The mass rarity interval MR1 of Lindström (Reference Lindström2021) is probably present in MPA 45631, with a rarity of Classopollis sp., the LO of R. germanicus and marked decline in R. tuberculatus (Fig. 5a). The occurrence of MR1 in the SAB2 Az is also a feature at StAB (Lindström, Reference Lindström2021).
4.a.3. SAB3 assemblage zone (recovery phase)
At StAB, the succeeding SAB3 Az extends from the upper part of the CMbr to the top of the Langport Mbr. In this Az, there are wide variations in the relative abundances of circumpolles and a wide range of spore taxa present (Bonis et al., Reference Bonis, Ruhl and Kürschner2010). Two peaks in spore abundance were recognised with the lower peak dominated by Acanthotriletes varius, Concavisporites spp., Conbaculatisporites spp. Deltoidospora spp., Kraeuselisporites reissingeri and Trachysporites fuscus. The upper peak consists mainly of Polypodiisporites polymicroforatus, Calamospora tener, Porcellispora longdonensis, Deltoidospora spp. and Todisporites spp. Notable features of this Az are the absence of Tsugaepollenites pseudomassulae and the very high abundance of K. reissingeri in the upper part of the lower spore peak, ∼1.0 m above the base of the Az. The Spelae CIE also occurs in the lower part of SAB3 at StAB.
Locating a comparable position for the SAB2 /SAB3 boundary in S-20 is difficult because of the smaller number of sample levels from S-20. Differences also probably arise from the hiatus/desiccation surface at StAB between the lower and upper CMbr (Gallois, Reference Gallois2009) and/or because of a fault at 1243.5 ft (378.66 m) in S-20, which may have cut out part of the CMbr. In S-20 (Figs. 5a, SM S4), a change from associations with similar proportions of pollen and spores to one dominated by circumpolles and with fewer spores is probably the clearest way to define this boundary and is here interpreted as between samples MPA 45631 and 45630. The LO of R. germanicus is in MPA 45631, but those of Ovalipollis pseudoalatus and Lunatisporites rhaeticus, which also occur at the Az boundary at StSB, occur higher in the S-20 core. Of the taxa which are relatively common in SAB2 at S-20 but are scarce or absent in SAB3 at StAB, lower numbers of R. tuberculatus occur above MPA 45631, but those of C. tener increase. A greater number of taxa (SM Fig. S4) are recorded above MPA 45631 than at the base of SAB3 at StAB. Of these, Perinosporites thuringiacus and Cornutisporites rugulatus appear in MPA 45630, and Stereisporites perforatus, Neochomotriletes triangularis and C. seebergensis? appear slightly higher. Taxa recorded both below and above the SAB2/3 boundary in S-20, but not from StAB by Bonis et al. (Reference Bonis, Ruhl and Kürschner2010), include Limbosporites lundbladiae, Cingulizontes rhaeticus, Chasmatosporites magnolioides, Convolutispora microrugulata, Semiretisporis gothae, Contignisporites problematicus, Triancoraesporites ancorae, Annulispora folliculosa and species of the genera Zebrasporites, Kyrtomisporis and Stereisporites; but several of these taxa were recorded in the StAB section by Warrington (in Hounslow et al., Reference Hounslow, Posen and Warrington2004, fig. 5).
4.a.4. SAB4 assemblage zone (post extinction)
In the StAB section (Bonis et al., Reference Bonis, Ruhl and Kürschner2010), the SAB4 Az extends from the top of the Lilstock Fm upwards for ∼14 m into the Hettangian of the Blue Lias Fm. Miospore associations in this Az are dominated by G. meyeriana, with K. reissingeri a minor but prominent component at several levels. Carnisporites spp. and R. tuberculatus only occur in the lower ∼4 m of the Az and Cerebropollenites thiergartii first appears ∼4 m above the base of SAB4. Also recorded were scattered occurrences of Porcellispora longdonensis, Quadraeculina annellaeformis, Tsugaepollenites pseudomassulae, Chasmatosporites spp., Vesicaspora fuscus and other bisaccates, and Deltoidospora spp. and other trilete spores. At StAB, the main change between SAB3 and SAB4 is from a relatively varied association to a comparatively impoverished one.
In S-20, the miospore association from sample MPA 45626 is interpreted as the lowest in the SAB4 Az. At this level in S-20, there is a change from a relatively varied association to a comparatively impoverished one (Figs. 5a, SM S4). In the SAB4, Az G. meyeriana is almost the only circumpolles present and dominates an association that includes small numbers of Deltoidospora spp., Alisporites sp., Acanthotriletes varius and Pinuspollenites pinoides, and a few specimens of Cingulizonates rhaeticus, R. tuberculatus and Vitreisporites pallidus. The mass rarity interval MR2 of Lindström (Reference Lindström2021) occurs in the Langport Mbr in S-20 and is marked by the LO of Limbosporites lundbladiae, the absence of Semiretisporites gothae, Perinopollenites elatoides, Lunatisporites rhaeticus and a virtual loss of Convolutispora microrugulata, and the recovery of Gliscopollis meyeriana. Lindström (Reference Lindström2021) placed MR2 in the upper part of the CMbr at StAB, marked by rarity in Lunatisporites rhaeticus, Perinopollenites elatoides, Polypodiisporites polymicroforatus and Ricciisporites tuberculatus. This difference perhaps relates to diachronous variation in the expression of the MR2 event, or the more widely spaced sampling intervals in the Lilstock Fm of S-20 inadequately display MR2.
4.a.5. Dinoflagellate cysts, acritarchs, prasinophyte algae
At StAB, the dinoflagellate cysts Rhaetogonyaulax rhaetica and Dapcodinium priscum appear in the Williton Mbr at the top of the Blue Anchor Fm and dominate aquatic palynomorph associations from the WFm and CMbr, with abundance peaks of R. rhaetica alternating with those of D. priscum (Bonis et al., Reference Bonis, Ruhl and Kürschner2010). These alternations may reflect changes from more fully marine environments, with R. rhaetica, to more marginal ones, with D. priscum (Poulsen, Reference Poulsen1996, p. 45). Courtinat & Piriou (Reference Courtinat and Piriou2002) interpreted D. priscum as a euryhaline form that occupied a range of ecological settings in low to high energy levels in nearshore and restricted marine environments and R. rhaetica as indicative of more open, low-energy marine conditions with greater water depth. Other dinoflagellate cysts were present in very small numbers in these associations (Bonis et al., Reference Bonis, Ruhl and Kürschner2010) and include Heibergella asymmetrica in the Williton Mbr and the middle of the WFm, Beaumontella langii at similar levels, Cleistosphaeridium mojsisovicsii at the base of the WFm and Suessia swabiana throughout that formation and in the CMbr. Bonis et al. (Reference Bonis, Ruhl and Kürschner2010) did not record R. rhaetica above the CMbr, but it occurs in very low numbers in the higher part of the Lilstock Fm and the lowest ∼10 m of the Lias (Warrington in Hounslow et al., Reference Hounslow, Posen and Warrington2004). In the upper part of the Lilstock Fm, dinoflagellate-dominated associations are replaced by ones dominated by acritarchs, predominantly Micrhystridium spp., and prasinophytes, mainly leiospheres; below that level, Micrhystridium occurs infrequently in the Williton Mbr and WFm, but leiospheres are common in the former and the lower half of the latter. Samples MPA45619 and MPA45617 have a higher abundance of acritarchs (SM Fig. S4) and probably relate to the bloom of prasinophytes and acritarchs similarly located in the upper part of the Tilmanni Cz at StAB (Van de Schootbrugge et al., Reference Van de Schootbrugge, Tremolada, Rosenthal, Bailey, Feist-Burkhardt, Brinkhuis, Pross, Kent and Falkowski2007).
The record of aquatic palynomorphs in S-20 (Figs. 5a, SM S4) is broadly comparable with that from StAB, although without leiospheres. The dinoflagellate R. rhaetica is commonest in the lower half of the WFm and the lower CMbr and occurs in very small numbers in the higher part of the Lilstock Fm but was not recorded higher; Dapcodinium priscum was recorded from the middle of the WFm to the lower part of the Planorbis Cz in the Redcar Mudstone Fm, with peaks in the middle of the CMbr and around the boundary between the Lilstock and Redcar Mudstone formations.
4.a.6. Environmental assessment
Mass occurrences of the dinoflagellate cyst R. rhaetica are probably indicators of maximum flooding surfaces (Lindström & Erlström, Reference Lindström and Erlström2006; Gravendyck et al., Reference Gravendyck, Schobben, Bachelier and Kürschner2020). An earlier acme event is the ‘Lunnomidinium interval’ (Lindström & Erlström, Reference Lindström and Erlström2006), in which Lunnomidinium scaniense is associated with a few Beaumontella caminuspina and Suessia swabiana in the lower parts of the Contorta Beds (Lindström & Erlström, Reference Lindström and Erlström2006). This event may be present in the middle of the WFm in S-20, where a possible B. caminuspina occurs in sample MPA 45639 (SM Fig. S4). A more widely recognised flooding event (MFS7, Rh2) occurs around the SAB1–SAB2 boundary, prior to the Spelae CIE (Lindström et al., Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017; Barth et al., Reference Barth, Franz, Heunisch, Ernst, Zimmermann and Wolfgramm2018). This is within the acme of Polypodiisporites polymicroforatus corresponding to the common occurrences of C. microrugulata above the last common occurrence of R. rhaetica in sample MPA45632 in S-20 (Fig. 5a).
The abundance in the diversity of spores seen in S-20 (Fig. 5a, SM S4) is a widely observed feature of the extinction and immediate recovery interval at many other locations (Orbell, Reference Orbell1973; Van de Schootbrugge et al., Reference Van de Schootbrugge, Quan, Lindström, Püttmann, Heunisch, Pross, Fiebig, Petschick, Röhling, Richoz and Rosenthal2009; Bos et al., Reference Bos, Lindström, van Konijnenburg-van Cittert, Hilgen, Hollaar, Aalpoel, van der Weijst, Sanei, Rudra, Sluijs and Van de Schootbrugge2023), and a similar response is apparent in the probable parent plants, with increases in mosses, liverworts, horsetails, ferns and club mosses, and proportional reductions of gymnosperms and conifers (Fig. 5b). Using the Eco-plant model data of Zhang et al. (Reference Zhang, Lenz, Wang and Hornung2021), the change from SAB1 to SAB3 shows an increase in palaeoclimatic humidity (EPH proxy) which begins in the upper part of SAB1 and is largely achieved by the middle of SAB2 (Fig. 5c). From the analysis of miospore data from StAB, Bonis & Kürschner (Reference Bonis and Kürschner2012) inferred an increase in humidity starting near the base of the CMbr with stabilisation by the start of the Lias Gp; a similar trend is shown by the EPH proxy for S-20.
From the upper part of the SAB2 Az, a progressive decline in temperature is apparent from the EGT proxy (Fig. 5c). Eurythermic parent plants (those tolerant to a wide range of temperatures) show a major proportional increase at a rate like that of the change in the EPH proxy. Euryphytes (tolerant to a wide range in humidity) show an erratic proportional increase over the same interval (Fig. 5c). These changes are interpreted as related to the habitat disturbance and ecosystem stress associated with the initial phase of the end-Triassic extinction.
4.b. Ammonites in Staithes S-20 and the base of the Jurassic
The base of the Hettangian Stage is formally defined at the ratified GSSP at Kuhjoch in the Karwendel Mountains (Austria) based on the first occurrence of the ammonite Psiloceras spelae Guex tirolicum Hillebrandt & Krystn (Hillebrandt & Krystyn Reference Hillebrandt and Krystyn2009, Hillebrandt et al. Reference Hillebrandt, Krystyn, Kürschner, Bonis, Ruhl, Richoz, Schobben, Urlichs, Bown, Kment and McRoberts2013) – a species closely related to Ps. tilmanni Lange from South America – hence the use of a Tilmanni Cz at the base of the Jurassic System in both South America and Europe (Page, Reference Page, Lord and Davis2010; Weedon et al., Reference Weedon, Jenkyns and Page2018, Reference Weedon, Page and Jenkyns2019, Boomer et al., Reference Boomer, Copestake, Raine, Azmi, Fenton, Page and O’Callaghan2021; Kment, Reference Kment2021; Hesselbo et al. Reference Hesselbo, Al-Suwaidi and Baker2023; SM Section 3). In the absence of any records of Ps. spelae elsewhere in Europe, alternative means of correlating the base of the Jurassic System (TJB) from the GSSP are necessary, principally using carbon isotope curves (Hillebrandt et al., Reference Hillebrandt, Krystyn, Kürschner, Bonis, Ruhl, Richoz, Schobben, Urlichs, Bown, Kment and McRoberts2013; Korte et al. Reference Korte, Ruhl, Palfy, Ullmann, Hesselbo, Sial, Gaucher, Ramkumar and Ferreira2019; Ruhl et al. Reference Ruhl, Hesselbo, Al-Suwaidi, Jenkyns, Damborenea, Manceñido, Storm, Mather and Riccardi2020), or Hg chemostratigraphy (Yager et al. Reference Yager, West, Thibodeau, Corsetti, Rigo, Berelson, Bottjer, Greene, Ibarra, Jadoul and Ritterbush2021), or using a combination of organic carbon isotopes and palynological changes (Lindström et al., Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017; Boomer et al., Reference Boomer, Copestake, Raine, Azmi, Fenton, Page and O’Callaghan2021). Using these correlation methods, a number of proposed positions for the base of the Hettangian have been suggested at StAB (black arrows, Fig. 3b). The positions in the Langport Mbr principally relate to using palynological data in addition to isotopic changes, whereas the TJB inferred in the basal Lias Gp principally used δ13Corg isotope data. Weedon et al. (Reference Weedon, Page and Jenkyns2019) also transferred the inferred position at StAB to other UK sections using astronomical cycles (dotted purple line, Figs. 3b, c, d). At Lavernock, the base of the Jurassic is probably at or near the base of the ‘Watchet Beds’ of the Langport Mbr based on carbon isotopes and magnetic polarity data (Korte et al., 2009; Reference Korte, Ruhl, Palfy, Ullmann, Hesselbo, Sial, Gaucher, Ramkumar and Ferreira2019; Hounslow & Andrews, Reference Hounslow and Andrews2024), close to an underlying ‘new’ species of the ammonite Neophyllites (Hodges, Reference Hodges2021), a species which could also be of latest Rhaetian age (Page in Hesselbo et al. Reference Hesselbo, Al-Suwaidi and Baker2023, p.18). The standard ammonite zonal framework follows Page (Reference Page, Lord and Davis2010), including the latest available sequence of high-resolution biohorizons (coded as Hn), as used by Weedon et al. (Reference Weedon, Jenkyns and Page2018, Reference Weedon, Page and Jenkyns2019), and summarised in SM Section 3.
The lowest ammonites present in the Staithes borehole correspond to Psiloceras erugatum (Phillips Reference Phillips1829), which occurs between 1192.17 ft and 1192.25 ft (363.37–363.4 m; Fig. 6c), indicating the upper part of the Tilmanni Cz (Hn2), included within calcareous nodules. Ps. erugatum (Figs. 6a, b) is the lowest confirmed Jurassic ammonite recorded across the British and Irish islands and, as well as the biohorizon, can be taken to mark the base of an Erugatum Horizon (i.e., sub-subchronozone; Page, Reference Page2017). The record in S-20 is particularly important, as it is the first time that the species has been recorded in situ close to its type locality in nearby Robin Hoods Bay, where the species has been collected since the early 19th century from loose calcareous concretions on the beach (Howarth Reference Howarth1962, 2022; Fig. 6d). The borehole record confirms that the species is present in situ in the Cleveland Basin in typical preservation and places the beach concretions at a defined level in the Redcar Mudstone Fm. The species is characterised by a nucleus with small bead-like nodes, hence linking it closely to Ps. grp tilmanni, followed by middle and outer whorls which range from plicate to smooth (Bloos & Page Reference Bloos, Page, Hall and Smith2000; Fig. 6d). The holotype of the species (NHM 37981) has been refigured several times, including by Howarth (Reference Howarth1962, pl.14, figs. 2a, b).
Earliest Jurassic ammonites from the northern Cleveland Basin (in following Pnnn= BGS photo assess number): Psiloceras erugatum (Phillips): A, B- GSM BKK 3156 (P1057099, P1057097), C- GSM 3157 (Staithes S-20 at 1192.2 ft, 363.38 m, P1057104), D- NHM37881, “Robin Hoods Bay, Yorkshire”, ex W. Bean coll.1859 (detail of typical concretion recovered ex-situ) (note node-like tubercles on nuclei and variable expression of ribbing on middle and outer whorls). E- cf. Neophyllites sp (GSM 3154, P1057092), Staithes S-20 at 1181.17 ft, 360.02 m). F- Neophyllites sp. cf. antecedens Lange (GSM 3151, P1057084), Staithes S-20 at 1179.00 ft, 359.35 – note relatively steep umbilical wall when compared to typical Psiloceras spp. (scale bar with 1 cm intervals for A-C, E, F; field of view for D. 70x120 mm). A-C, E, F by S. Harris BGS, D by KNP. British Geological Survey materials © UKRI 2024; containing public sector information licensed under the Open Government Licence v3.0.
At 1192.04 ft (363.33 m), a larger (68 mm) essentially smooth Psiloceras with a typical suture suggests a continuation of the erugatum biohorizon (Hn2). At 1189.08 ft (362.43 m), a small evolute cf. Neophyllites (Fig. 6e) indicates the base of the Planorbis Chronozone and Subchronozone and probably Hn3 (with Ne. imitans Lange) with evolute, smooth forms from 1186.04 ft to 1188.0 ft (361.50–362.10 m), suggesting Ne. antecedens Lange of Hn4 (Fig. 6f). Relatively, involute Ps. cf. planorbis indicates the base of planorbis α Biohorizon (Hn5) with large (d = 68 mm) and more evolute forms from 1183.83 ft [360.12 m], indicating the planorbis β Biohorizon (Hn6). As in West Somerset (Bloos & Page Reference Bloos, Page, Hall and Smith2000), rare Neophyllites persists into the lower Planorbis Scz in S-20, at least as far as Hn5, where two specimens with characteristic spiral grooves are present at 1185.79 ft (361.43 m). Ps. plicatulum at 1182.5 ft (360.43 m) indicates the base of the plicatulum biohorizon (Hn7).
No further age diagnostic specimens have been noted below the first record of Caloceras of the overlying Johnstoni Scz at 1177.83 ft (359.00 m). The latter, however, does not have the typical blunt ribbing of basal Johnstoni Scz Ca. aries of Hn10, as its ribs are relatively sharp and hence could represent a slightly younger species, for instance, of Hn11b–11d.
The nearest comparable record of the earliest part of the Hettangian is from the Felixkirk borehole (Ivimey-Cook & Powell, Reference Ivimey-Cook and Powell1991; Fig. 1a), which has an ammonite record like S-20. Ivimey-Cook & Powell (Reference Ivimey-Cook and Powell1991) did not distinguish a Langport Mb at Felixkirk, although in a more detailed assessment Beith et al. (Reference Beith, Fox, Marshall and Whiteside2023) considered the lowest 1.7 m of the Calcareous Shales Mbr as the Langport equivalent, with the Spelae CIE in the lower half of this member. As Ps. erugatum and Neophyllites were not distinguished at Felixkirk, the lowest recorded ammonites (at 11.89 m above the base of the Calcareous Shales Mbr, using the revised base of Beith et al. Reference Beith, Fox, Marshall and Whiteside2023) could either be end Tilmanni Cz or lower Planorbis Scz. Ps. plicatulum at around 4.9 m higher (+16.8m) would at least confirm the upper part of the latter (i.e., Hn7–9), with the highest recorded Psiloceras at +17.6 m and Caloceras spp. from +21.7m to around +22.4m, indicating the Johnstoni Scz. Similarity to Felixkirk is also in the thickness of the Calcareous Shales Mbr below the first occurrence of ammonites, i.e., 11.9 m and 12.45 m above the base of the Lias Group in the Felixkirk borehole and S-20, respectively. However, the Planorbis Scz is thinner at S-20 than Felixkirk (3.5 m compared to 5.7 m). If the succession in S-20 is comparable to that at Felixkirk, the Spelae CIE should be around the lower part of the Langport Mbr in the core (Beith et al., Reference Beith, Fox, Marshall and Whiteside2023), allowing a reasonably precise correlation to StAB (Fig. 3a,b).
4.c. Magnetic and palaeomagnetic results
The variability in Ksurf in the Branscombe Mudstone Fm is largely related to anhydrite content (Fig. 2b). In the overlying formations, variation is likely inversely related to carbonate content, with a variable content of paramagnetic minerals probably in the clay or silt fraction. This is similar to conclusions inferred from other studies of these units (Hounslow, Reference Hounslow1985; Deconinck et al., Reference Deconinck, Hesselbo, Debuisser, Averbuch, Baudin and Bessa2003; Hounslow et al., Reference Hounslow, Posen and Warrington2004; Hüsing et al., Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014; Weedon et al., Reference Weedon, Jenkyns and Page2018), discussed at length in SM section 4. The remanence-carrying mineralogy is similar to these formations investigated in SW Britain (see SM Section 4; Fig. S5).
4.c.1. Magnetisation components
The magnetisations comprise three components. Firstly, a low stability component (LTC), which was typically removed by 100–150oC, occasionally persisting up to 200oC or 300oC; 88% of specimens contained this component. This was often shallow to intermediate in inclination, with 88% showing downward-directed and 12% upward-directed components (SM Fig. S6a). The origin of this component is unclear, but it may represent a short-term viscous component, one acquired during core storage since 1968, or perhaps a combination of this and higher stability components. The large directional scatter of the LTC precludes anything useful being inferred from it (SM Fig. S6b).
Secondly, an intermediate stability component (MTC; 74% of specimens), which largely has a steep inclination and is down-directed (mean inclination +63o, α95 = 2.2o, k = 17.8; mean using the likelihood function of Enkin & Watson, Reference Enkin and Watson1996), although 6.6% of these specimens had steep up-directed components (SM Fig. S6c; example demagnetisation diagrams in SM Fig. S8). The MTC typically displays a marked directional break with the LTC on Zijderveld plots (see SM Fig. S8). The MTC is often a major part of the natural remanent magnetisation intensity. The stability range of the MTC commonly started from 150oC or 100oC (occasionally ranging to 350oC, SM Fig. S7a). The upper stability range was most commonly up to 400 to 450oC during thermal demagnetisation, although this was quite variable, from 200 to 650oC in some specimens (SM Fig. S7a). For specimens using a combined demagnetisation scheme, the upper stability range was typically 10–30 mT but was up to 80 mT in rare some. In 4.4% of specimens, the MTC dominated, and no ChRM was detected. This component is interpreted as a magnetisation acquired during the Brunhes Chron. A similar Brunhes age overprint has been identified in coeval units in southern Britain (Briden & Daniels, Reference Briden and Daniels1999; Hounslow et al., Reference Hounslow, Posen and Warrington2004; Hüsing et al., Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014; Hounslow & Gallois, Reference Hounslow and Gallois2023).
Thirdly, the highest stability characteristic remanence (ChRM) was detected in 89.2% of demagnetised specimens. Of these, 81% had S-class behaviour and 19% had T-class behaviour (Fig. 7b). The ChRM has both positive and negative inclinations (Fig. 8c,d). Converted to positive, the mean inclination of all data from S-class specimens is 47.7° (α95 = 2.4°, k = 15.8, n = 135; method of Enkin & Watson, Reference Enkin and Watson1996). The ChRM was predominantly isolated by thermal demagnetisation (SM Fig. S7b,c) in the Branscombe Mudstone Fm and by AF demagnetisation in the Penarth and Lias groups. The Blue Anchor Fm specimens possess a mix of isolation methods (Fig. 7c). A few ChRM’s were isolated by overlapping thermal and AF demagnetisation ranges. Using only thermal demagnetisation, the starting range of this component was variable but largely between 300oC and 600oC, with a few specimens outside this (SM Fig. S7b). The S-class line fits were predominantly through the origin. For those in which the ChRM was isolated by AF demagnetisation (Fig. 7c), the start AF ranges were 10–70 mT, with the end ranges largely through the origin and others mostly ending in the range 50 to 80 mT (SM Fig. S7b). For some specimens the demagnetisation noise and thermal alteration largely precluded origin fits. The ChRM is interpreted as a Late Triassic–Early Jurassic magnetisation.
Summary magnetostratigraphic data for the Staithes S-20 core. a) Simplified sedimentary log (SM Fig. S3 for details, key in Fig. 2). b) Demagnetisation behaviour classification of specimen data. c) Characteristic remanent magnetisation (ChRM) isolation method during demagnetisation (T=thermal, A= alternating field, C= combined). d) Specimen polarity classification. e) Specimen ChRM inclination. f) Specimen virtual geomagnetic pole latitude (VGPR) with respect to the mean poles for the Branscombe Mudstone Formation and the Penarth and Lias groups (core re-oriented using joint mean-run rotation angle). g) Section polarity, lithostratigraphy (Lith.) and biostratigraphy. MZ=labels of magnetozone couplets (BM = Boulby Mine, the location of the core). LM= Langport Member, CM= Cotham Member, var.=variegated unit. Hatching in column (g) represents the interval of probable base Hettangian.
Component directional data when re-oriented: a) Specimen ChRM directions re-oriented by the mean MT component (inferred Brunhes-age) in each run; b) MT-component (MTC) re-oriented by the mean-run ChRM declination; c) and d) specimen ChRM directions for formational groupings, re-oriented by the combined MTC and ChRM declinations for each run. Fisher mean directions shown for b), c) and d), and the reversal test for c) and d), with classification, observed (γobs) and critical (γcrit) angle of divergence.
For the 6.6% of specimens with a steep up-directed MTC component, this was assumed to be the result of inverted core runs. These occurred in three runs (run codes ST6, ST7 and ST36; SM Fig. 3f, d). Within ST7, the inverted segments were adjacent in the middle of the run, indicating core-segment re-assembly was imperfect for this run. For runs ST6 and ST36, the whole runs were inverted. The magnetisation directions from these inverted segments and runs were rotated by 180o about the horizontal plane.
To reorient the runs, the average declination of the MTC within each run was used to determine a rotation angle to bring the average declination to 0°. Applying this rotation angle to reorient the ChRM directions gives an estimate of the directional distribution of the ChRM directions (Fig. 8a). This reorientation approximately produces directions like the expected Rhaetian–Hettangian directions, supporting the inferences made about the origins of the MTC and ChRM directions. However, ten of the 49 core runs sampled could not be reoriented using the MTC.
Similarly, if the mean declinations of the ChRM in the runs are used to reorient the MTC, it gives a directional distribution (Fig. 8b) similar to that in other Brunhes-age overprints observed in these formations (Hüsing et al., Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014; Hounslow & Gallois, Reference Hounslow and Gallois2023; Hounslow & Andrews, Reference Hounslow and Andrews2024). The ChRM mean declinations expected for S-20 are 037° and 019° for the upper Mercia Mudstone Fm and Penarth/Lias groups, respectively (using palaeopole data from Hounslow et al., Reference Hounslow, Posen and Warrington2004, Hüsing et al., Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014 and Hounslow & Gallois, Reference Hounslow and Gallois2023). The resulting Fisher mean inclination of the reoriented MTC is 68.7° (α95 = 4.4°), which is close to the expected Brunhes-age inclination of 70.4° (Fig. 8b).
4.c.2. Defining the magnetostratigraphy:
In order to utilise the directional information from the T-class specimen ChRM’s and determine an estimate of VGP latitude (VGPR) for each specimen, the core runs were reoriented by using both the MTC and ChRM declinations (i.e., averaging the rotation angles of both sets in each run). These ‘jointly reoriented’ ChRM data are shown in Figure 8c and d. Whilst this procedure introduces dependence on externally derived palaeopole directions, it mostly corrects the additional declination dispersion evident in the ChRM directions (if using only MTC component reorientation; see Fig. 8a) and captures the specimen declination dispersion within and between the core runs. With this procedure, three of the 49 studied runs could not be reoriented. The VGPR from the jointly reoriented core runs for both the S-class and T-class ChRM sets are shown in Figure 7f. An estimate of the formational mean directions (Fig. 8c,d) are comparable to outcrop-based studies of these units.
The magnetostratigraphy for the core can be inferred from the specimen-level interpretations (both S-class and T-class data; Fig. 7d), the ChRM inclinations from the S-class data (Fig. 7e), and lastly, the VGPR, which uses both the T-class and S-class data. The VGPR lacks data for five sampling levels in three core runs, which could not be reoriented (Fig. 7f), but are shown with S-class specimen inclinations instead in Fig. 7e.
The resulting polarity displays four major magnetozone couplets (BM1n/BM1r to BM4n), with five of the magnetozones having submagnetozones (Fig. 7g). Submagnetozones BM1r.1n, BM3r.2n, BM3r.3n and BM3r.4r are defined by sampling from at least two adjacent depths, with one or more specimens from each depth. Tentative submagnetozones, BM1n.2r and BM3r.1n, with ¾ bar width, are defined by two specimens from a single sampling depth. Tentative submagnetozones, BM1n.1r, BM3n.1r, BM3n.2r and BM4n.1r, with a half or ¼ bar width, are defined by a single specimen at a single sampling depth.
Reverse polarity specimens defined by VGPR are more strongly dependent on the T-class type demagnetisations than the normal polarity specimens (Fig. 7f). This issue is also apparent in the coeval section at StAB and in the overlying Hettangian (Hounslow et al., Reference Hounslow, Posen and Warrington2004; Hüsing et al., Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014), as well as in the youngest parts of the MMG in the Seaton sections (Hounslow & Gallois, Reference Hounslow and Gallois2023). As in the prior studies, this is attributed to the difficulty in fully removing the much stronger Brunhes-age overprint magnetisation, which partially remains overlapped with the ChRM in the T-class specimens.
4.d. Astrochronology for the Langport Member and early Hettangian
4.d.1. SAR constraints
The dataset of Guex et al. (Reference Guex, Schoene, Bartolini, Spangenberg, Schaltegger, O’Dogherty, Taylor, Bucher and Atudorei2012) from Peru provides an approximate duration for the Tilmanni plus Planorbis chronozones. An approximately linear SAR is suitable for their data, which extends from the latest Rhaetian to a level which can be correlated to the upper part of the Angulata Cz of the Hettangian (SM Fig. S9). However, correlation of the Peruvian ammonite assemblages to the European ammonite biochrons has some uncertainty, and two possible scenarios are used, which, in combination with the uncertainty in the radioisotopic dates, suggest that the duration of the Tilmanni plus Planorbis chronozones is between 0.725 and 1.175 Myr (SM Fig. S9). In combination with the thicknesses of these chronozones in the UK sections, the likely range in average SAR for sections at StAB, Lavernock, Lyme Regis and S-20 are 1.15–1.87 cm/kyr, 1.24–2.01 cm/kyr, 0.56–0.91 cm/kyr and 2.13–3.45 cm/kyr, respectively. These SAR constraints allow narrow windows of where to expect spectral peaks in cycles/m in the evolutive harmonic spectra. These are marked on Figure 9a for the S-20 dataset (and comparable plots in SM Figs. S10–S12 for the other sections). These simply demonstrate plausible astronomical periods in the data but have not been used beyond this illustration. To allow easier inter-section comparison of the predicted SAR from the durations derived by the astrochronology, the average SAR over the Tilmanni Cz and Planorbis Scz is normalised by the upper SAR constraint for each section, giving a variable nSART+PS. This ranges from 1.0 at the longer duration constraint to 0.62 for the shorter duration constraint, irrespective of the section concerned.
a) Evolutionary harmonic analysis (EHA) map for the Staithes S-20 Ksurf data, with possible astronomical periods indicated (based on the SAR constraints). E=eccentricity cycles, O=obliquity, and P=precession (targets listed in SM Table S1); b) the S20 Ksurf data with depth in metres; c) SAR tracks based on the evolutionary TimeOpt method (Meyers, Reference Meyers2019) for four width windows (5.5 m to 7.0 m; coloured lines); full lines are primary tracks and dashed lines secondary tracks. The SAR constraints (dotted vertical black lines) and inferred composite baseline SAR tracks (in thick grey line, St1, St2, St2x) are shown. Comparable plots for Lavernock, St Audrie’s Bay and Lyme Regis are in SM Figs. S10 to S12.
4.d.2. Stage 1, evaluation of baseline SAR models
The evolutionary TimeOpt method (eTimeOpt) gives a variety of plausible SAR tracks which could yield an average SAR within the range of the SAR constraints. These tracks were based on the r2opt maps since these gave clearer tracks than the r2envelope or r2spectral eTimeOpt maps. The inferred possible SAR tracks for each section were based on visually estimating a mean path through the primary tracks (i.e., the stronger tracks). An example of this, using the S-20 dataset, is shown in Figure 9c, which generates two probable tracks (St1 and St2). An additional lower SAR track (St2x) was added from the secondary tracks since the Planorbis Scz is rather more condensed (or has more hiatuses) compared to the other sections and that in the Felixkirk borehole (Figs. 1a, 3a). The ratios of the thickness of the Planorbis Scz/Tilmanni CZ are 0.78, 0.73 and 0.73 at StAB, Lavernock and Lyme Regis, respectively, with the values at Felixkirk and S-20 being 0.48 and 0.28. SAR tracks were similarly constructed for the other sections (SM Figs. S10c, S11c, S12c), yielding (stronger) primary-based tracks labelled as SA1 and SA2 at StAB, La1 and La2 at Lavernock, and Ly1a, Ly1b, Ly2a and Ly2b at Lyme Regis. At StAB, an additional larger SAR track (SA2x) was defined on secondary tracks, which also yields a mean SAR within the SAR constraints by joining the SA2 track in the mid part of the Planorbis Scz (SM Fig. S11c). This larger SAR may account for the expansion of Hn5 in the lower part of the Planorbis Scz (Fig. 3b). The primary SAR tracks for Lavernock suggest a possible lower SAR in the upper part of the Tilmanni Cz in a mudstone-rich interval (between the ‘Dual Bed’ and bed 30; Waters & Lawrence, Reference Waters and Lawrence1987) as in SM Figure S10c with larger Ksurf (Fig. 3c).
Using TimeOpt, the largest r2opt were for the baseline SAR models St2, SA2x, La2 and Ly1b (Fig. 10a). At Lavernock, the proxy base of the Tilmanni Cz (at the base of the Watchet Beds) was projected downwards using the average SAR in the overlying part of the chronozone (extension bars on La1 and La2 models in Fig. 10). In S-20, the best-performing St2 model (r2opt =0.246) does not correct for the shorter Planorbis Scz (Fig. 3a), with all the S-20 baseline models (St1, St2, St2x) yielding briefer Planorbis Scz durations compared to the other sections (Fig. 10b).
Evaluation of the baseline SAR models. SA= St Audrie’s Bay (StAB), Ly=Lyme Regis, La=Lavernock and St= Staithes S-20 models, respectively. Those with an appended + indicate the hiatus (or missing interval) inserted (in kyr) into the baseline SAR model (only for S-20 models). a) shows the r2opt and the normalised SAR across the Tilmanni plus Planorbis chronozones interval (nSART+PS), with the upper SAR constraint for each section being the normaliser (giving nSART+PS =1). b) Shows the durations of the Tilmanni Cz and Planorbis Scz. σT,P= standard deviation of the durations of the Tilmanni Cz and Planorbis Scz for the SAR sets indicated inside the marked regions (solid blue line for set-1). The numbers in a) inside […] are the mean nSART+PS ±1σ and mean r2opt for the SAR sets inside the solid blue/black dotted lines (set 1 models marked in blue). The SAR model with the maximum r2opt in each section has a grey background. Examples of hiatus- testing data shown in SM Figs. S13 to 15.
A comparison of the Ksurf correlations to S-20 suggests there may be a missing part around the Hn6 to Hn8 interval (Fig. 3a). If the ratio of the thicknesses of the Planorbis Scz/Tilmanni Cz at Felixkirk is applied to S-20, it suggests that ∼2.5 m may be missing at S-20, representing ∼42% of the Planorbis Scz. This is inferred to be related to a level in the core with near-horizontal slickensides at 1183.38 ft (360.68 m), which probably represents a small fault, providing a position for the missing interval and ‘hiatus’. Therefore, for this initial evaluation, the missing interval was estimated using hiatus testing applied to the S-20 SAR baseline models (see SM Figs. S13 to S15 for hiatus testing examples). These yielded nine additional models with larger r2opt (Fig. 10a) for the baseline models St1, St2 and St2x, with only St2+134 having a lower r2opt (by 0.028) than the St2 baseline model (the ‘+ number’ is the hiatus in kyr inserted into baseline models prior to TimeOpt optimisation; SM Fig. S15). It is important to recognise that these initial baseline hiatus durations will not correspond to the hiatus durations inferred after optimisation using TimeOpt (note the longer ‘hiatuses’ than Planorbis Scz durations (on the y-axis) for models St2+333 and St2x+434; Fig. 10b). The percentage of the Planorbis Scz duration represented by the ‘hiatus’ in the optimised age model is shown in brackets after the ‘model code+hiatus’ in Figure 10b (only for those with values >40%). Many of these hiatus models still yield briefer Planorbis Scz durations but similar Tilmanni Cz durations, compared to other section baseline models. The exceptions are St2x+434, St2x+391, St2+333 and St1+265 (Fig. 10b), which yield Planorbis Scz durations like the other sections. These also yield nSART+PS > 1.0, apart from St2+2 and St2+134 (Fig. 10a).
Selecting those with the larger r2opt and minimal dispersal in zonal durations and nSART+PS, a plausible solution using all four section datasets is using SAR models SA1, La2, Ly2a and St1+265 (those inside the area marked with a blue solid line in Fig. 10a, b). This set has duration dispersion for the Tilmanni Cz and Planorbis Scz (σT,PS) of 46 and 69 kyr, respectively, (top of Fig. 10b), and has a σ for nSART+PS of 0.11 (top of Fig. 10a). This set (Set-1) of SAR models contains one with the largest section-based r2opt (i.e., La2) and a combined mean r2opt of 0.186 (top of Fig. 10a). Ideally, a good solution should maximise the overall r2opt (i.e., the astronomical target fits), contain more of the solutions with the largest section-based r2opt and provide greater intersection consistency in nSART+PS and biochron durations. A second group of SAR models with a larger mean r2opt of 0.204 are La2, Ly2a, SA2x and St2x+391 (Set-2), which give σT,PS = 75, 60 kyr, and σ for nSART+PS of 0.12 (inside the area marked with dashed black lines in Fig. 10a, b). This highlights that there may not be a single outstanding solution but one or more similar alternatives. Many of the S-20 SAR models with larger r2opt suggest the hiatus may occupy about 70% of the Planorbis Scz (models St2x+434, St2x+391, St2+333, St1+265; Fig. 10b), which is larger than the ∼42% expected (compared to Felixkirk). Such large percentages could also relate to the uncertainty on the Caloceras species, which may be Johnstoni Scz, Hn11–11b, rather than Hn10, pointing to a greater loss of the upper part of the Planorbis Scz (alternative on Fig. 3a).
4.d.3. Stage 2, modulation of the baseline SAR models
The better-performing models from stage-1 were used for β-testing, searching for maximum r2opt with respect to β-values selecting peaks in r2opt values greater than or similar to the stage 1 models. High values of mod(β) (when SARβ,H approached the limit possible) generally yield unrealistic age models, since the levels with very low SARβ,H introduce many hiatus-like intervals into the SAR section models, giving a region of r2opt instability at large |β|. Therefore, for the β-H search region, a lower |β| outside this regional of instability was used (see examples in SM Figs. S28–S30). In addition, for the S-20 models containing the faulted-missing interval, this was treated to β-H testing as described in Sections 3.c.1 and 4.d.4.
The resulting models are labelled with the baseline SAR model plus ‘=n’, where n is the rank of the r2opt value, with ‘=1’ indicating the largest r2opt value (‘=2’ next largest, etc, and ‘=0’ if only one r2opt peak) for that baseline SAR model (CD-SAR type models in Fig. 11; TS-SAR type models in SM Fig. S16). Most of these models have larger r2opt (Fig. 11a) than baseline models from stage-1, a model-specific change symbolised as Δr2opt. Of these, the median Δr2opt for the CD-SAR models are 0.050 and 0.031 for the TS-SAR type models, with the maximum Δr2opt of 0.201 for the S-20 CD-SAR model St2x=1 (Fig. 11a). On a section-specific basis the maximum Δr2opt are shown by the CD-SAR models for S-20 (of 0.201; St2x=1) and Lavernock (of 0.140, La1 = 1; Figs. 11a, 12a) and by the TS-SAR models for StAB (of 0.067, SA2=1) and Lyme Regis (of 0.179, Ly1a = 0, SM Fig. S17). Hence, based on TimeOpt, the CD-SAR type models seem slightly better overall, but not universally for all sections.
β- testing of the better-performing baseline models from Figure 10 without hiatus (except for the Staithes S-20 dataset), for the CD-SAR type models. Plot of r2opt with respect to β shown in Fig. 12a. Labelling details as in Figure 10. SAR model set 1 and 2 statistics in Table 1. The equivalent TS-SAR models are shown in SM Figure S16.
a) Plot of r2opt with respect to β for the CD-SAR β-testing models in Fig. 11. Labelling details as in Figure 10. b) Power spectra of the SAR models included in SAR model sets 1 and 2 from Fig. 11, with linear (black) and logarithmic scaling of spectra (in grey), and eccentricity, obliquity and precession frequencies (red vertical dashed-lines) as in SM Table S1. Blue line is the bandpass filter for evaluation of the precession amplitude envelope.
For the CD-SAR type models (with negative β; Figs. 11, 12a), there are two sets of models (Set-1 and Set-2) which have low dispersions in nSART+PS and biochron durations (Table 1; Fig. 11). These share the Ly1a = 1 model but have differing models for the S-20, Lavernock and StAB models. Set 1 maximises mean r2opt and Δr2opt (of 0.291, 0.121; Table 1), with minimum dispersion in Planorbis Scz durations. Set 2 has lower dispersion in Tilmanni Cz durations (Table 1). Both sets show larger mean r2opt than the sets for the baseline models (Table 1). The models in these sets have a wide range of β values (Table 1; Fig. 12a). The power spectra show more prominent matches to the eccentricity and obliquity bands, although model SA2x = 1 is a fair match in the precession bands (Fig. 12b).
Data for the sets of baseline and combined β-testing duration models are indicated in Figs. 10, 11, 12 and SM Figure S16. In column 1, Hopt is the hiatus inferred by the TimeOpt optimised age models shown inside round brackets, as well as the that inserted in the baseline age model (shown as H), with both in kyrs. Column 6: the statistical significance values (P(AR1’)) of the model using a Monte Carlo simulation of an AR1 process (Meyers, Reference Meyers2019), with values listed in the same model order as in column 1 (1000 simulations with 100 sedimentation rates). The AR1 process is modelled with ‘raw’ ρ values for S-20, StAB, Lavernock and Lyme of 0.7325, 0.6189, 0.7481 and 0.7639, respectively. Δ r2opt = the mean improvement in r2opt over the baseline models in Figure 10
For the TS-SAR type models, the section-specific models with the largest r2opt are somewhat scattered in the chronozone duration graph (SM Fig. S16d), and the best choice is a single set with minimum dispersion in nSART+PS and chronozone duration (Table 1; SM Fig. S16b,d); this includes the SAR model at Lyme Regis with the largest r2opt. Apart from the Ly1a=0 model, others in this set have low +β values (Table 1; SM Fig. S16a).
From the β-testing, the CD-SAR models seem to have better overall performance for the following reasons.
1) More of the SAR models from each section have the larger r2opt and cluster in the zonal duration plots better, rather than being more dispersed as in the TS-SAR models (Table 1).
2) The set-1 in the CD-SAR models includes two of the models with the largest r2opt for each section, rather than one in other sets (Fig. S11a; SM Fig. S16b). This set also has the largest mean Δr2opt of 0.121 (Table 1).
3) There is a general positive relationship between Δr2opt and increased SAR modulation amplitude measured by mod(β) for the CD-SAR models when all the section datasets are considered together (SM Fig. S17a).
4) The mostly low β- values for the models comprising the one set for the TS-SAR models indicate that, except generally for the Lyme Regis models, the SAR modulations expressed by +β have limited impact on improving the astronomical fits (SM Fig. S17b).
5) Although none of the CD-SAR sets fully regularise with time the correlation of the ammonite biohorizons, they generally perform better in this respect than the best TS-SAR model set (SM Figs. S21 to S23).
6) Monte Carlo tests of the models against an AR1 process indicate overall that CD-SAR models in set-1 have a closer approach or exceed the 95% confidence level (P(AR1’) <0.05) in more cases (Table 1). For the TS-SAR models only, Ly1a=0 has lower P(AR1’) compared to the corresponding Lyme Regis model for CD-SAR set-1 (Table 1).
However, most of the Lyme Regis TS-SAR models give larger Δr2opt values than those using the CD-SAR type models at this site (SM Fig. 17b), accounting for the smaller P(AR1’). The opposite is the case for the Lavernock models, which show much improved Δr2opt for the CD-SAR type models (SM Fig. 17a). If the β-values usefully express the degree of SAR modulation at the bedding scale, the expectation would be that StAB and S-20 should behave similarly (i.e., similar scale of β) since these contain fewer limestones than the Lavernock and Lyme Regis sections, datasets which also might be expected to have similar behaviour. However, this simple concept could be flawed, and the difference between the responses in the Lyme Regis and Lavernock datasets to β-testing might reflect more fundamental differences in the bedding-scale SAR modulation between those sections. This difference in behaviour might reflect relative contributions and timing of diagenetic carbonate formation or the differing land-proximal to distal shelf positions of these sections. This could yield a differing style of SAR modulations, which could be compatible with evidence for condensation/tractional erosion in some limestone beds at Lyme Regis (Weedon et al., Reference Weedon, Jenkyns and Page2018; Paul et al., Reference Paul, Allison and Brett2008). The inclusion of the Ly1a baseline model (Ly1a is ranked top in each case) in the sets for both the CD-SAR and TS-SAR models suggests that this SAR model is the best irrespective of the type of applied modulation (i.e., Ly1a = 1, Ly1a = 0; Table 1).
4.d.4. Stage 3, Hiatus and β−testing of the baseline SAR models
Detection of hiatuses in the Lias Gp has been based on identifying missing biostratigraphic intervals, sedimentological evidence of hiatus, or steps in Shaw plots based on correlation of biohorizons (Weedon et al., Reference Weedon, Jenkyns and Page2018, Reference Weedon, Page and Jenkyns2019). Whilst these approaches have merit, two of them are built on assumptions of strictly isochronous biohorizons, uniformity of ammonite preservation and near-linear SAR expectations embedded into their use. Ideally, cases of combined sedimentological and biostratigraphic evidence of hiatus provide the stronger conclusive cases, but these are lacking in the interval examined here. Perhaps the best evidence for hiatus is the absence or thinness of Hn3 (Fig. 3) at the base of the Planorbis Scz (Weedon et al., Reference Weedon, Jenkyns and Page2018). Biohorizons Hn6 and Hn5 are also relatively condensed at Lyme Regis and Lavernock, a feature used by Weedon et al. (Reference Weedon, Page and Jenkyns2019) to infer hiatuses. Although this feature may correspond approximately with a maximum flooding surface in the central European Basin, which is provisionally dated at about Hn7 to Hn8 (Barth et al., Reference Barth, Franz, Heunisch, Ernst, Zimmermann and Wolfgramm2018). In the present work, a more conservative view is used, in that hiatuses may exist at the base of the Planorbis Scz at StAB and Lyme Regis (Fig. 3b, d). In addition, it is also impractical to insert two hiatuses and vary β simultaneously.
The better-performing baseline SAR models were tested simultaneously for variation in hiatus (H) and β. The hiatus search window ranged from 0 kyr to upper values identified from hiatus testing, which gave realistic models (i.e., shown in SM Fig. S20). Inserting hiatuses at the base of the Planorbis Scz (for StAB and Lyme Regis datasets) using the better-performing SAR models tends to give comparable or slightly lower dispersion in estimates of nSART+PS and biochron durations to those from β-testing alone (Tables 1, 2).
Data for the sets of combined β-H duration models indicated in Figures 13, 14 and SM Figs. S18, S19. In column 6, Hopt= the hiatus inferred by the TimeOpt optimised age models (in the same model order as in column 1), rather than that inserted in the baseline age model (shown as H, kyrs in column 1). Other columns as in Table 1
The resulting CD-SAR models give three possible groups: set-1 to 3 (Fig. 13). Set-3, marked with a dotted red line in Figure 13, gives low and dispersed nSART+PS, the largest Tilmanni Cz duration (Table 2), and the smallest mean r2opt and is not considered further. The set-1 and set-2 statistics for these CD-SAR models are rather like the comparable models from β-testing, since they include the same models or models with similar β and H (Tables 1, 2). This is except for the included Lyme Regis models. Although set-1 has the largest mean r2opt of 0.321 of the CD-SAR models, the 73 kyr hiatus using the Ly1a = 1 model is unrealistically large at 31% of the Planorbis Scz duration (SM Fig. S21). Accepting the hypothesis that the TS-SAR type models may be more appropriate for the Lyme Regis dataset would give an alternative final set of models, which include the CD-SAR models at the other locations (bottom panel set-final in Table 2 and grouping plots in SM Fig. S19). These give a marginally larger mean r2opt, a smaller hiatus and a lower P(AR1’) for the included Lyme Regis model (Table 2). Hence, our final optimum estimates for the durations of the Tilmanni Cz and Planorbis Scz give 262 ± 53 kyrs and 238 ± 27 kyr, respectively (means: r2opt = 0.325 and nSART+PS = 1.02±0.04; Table 2). Whilst some refinement could be achieved on the Tilmanni Cz duration if the TJB position were more precisely known in each section, the 1σ uncertainty of ±53 kyrs is comparable to the uncertainty on the TJB position at StAB (compare Fig. 3b, SM Fig. S22).
This optimum estimate is just above the upper limit of the expected SAR based on data in Section 4.d.1. For comparison at StAB, using the duration of the Tilmanni Chronozone+ Planorbis Scz would give nSART+PS of 0.49, 2.20 and 1.86 using the timescales in Weedon et al. (Reference Weedon, Page and Jenkyns2019), Husing et al. (Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014) and Ruhl et al. (Reference Ruhl, Deenen, Abels, Bonis, Krijgsman and Kürschner2010), respectively. At Lavernock and Lyme Regis, the equivalent nSART+PS would be 0.42 and 0.52, respectively, using the estimates of Weedon et al. (Reference Weedon, Page and Jenkyns2019). Our final duration estimates therefore fall between those of the earlier studies and are ∼50% of those suggested by Weedon et al. (Reference Weedon, Page and Jenkyns2019) and longer by x2.2 and x1.8 than those suggested by Hüsing et al. (Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014) and Ruhl et al. (Reference Ruhl, Deenen, Abels, Bonis, Krijgsman and Kürschner2010).
β-H- testing of the better-performing baseline models from Figure 10 with hiatus inserted at the base of the Planorbis Scz for Lyme Regis and StAB models. Plot of r2opt with respect to β shown in Fig. 14a. Notation details as in Figures 10 and 11. SAR model set-1, set-2 and set-3 statistics in Table 2.
4.d.5. An anchored astrochronology for the early Hettangian
To anchor this astrochronology, the radioisotopic dates from the Levanto section in Peru are correlated to the StAB section using the correlation relationships between the organic carbon isotope records proposed by Ruhl et al. (Reference Ruhl, Hesselbo, Al-Suwaidi, Jenkyns, Damborenea, Manceñido, Storm, Mather and Riccardi2020, fig. 3). This places the base of the Tilmanni Cz close to the base of magnetozone SA5r (Figs. 3b, 15). These relationships were enhanced by utilising an estimate of the stratigraphic uncertainty in this correlation in the stratigraphic metre scale at StAB. The projected positions and uncertainties were estimated by projecting the Peruvian dates onto the duration scale for the SA2x =1 CD-SAR model (model in set-final), which was projected downwards into the CMbr and the WFm (Fig. 15). The age of the base of the Lias Group is estimated by fitting a regression line between the duration scale and the radiometric dates using uncertainty in both axes (York’s method; Read, 1989; Excel script by P. Kromer) shown as a solid blue line in Figure 15. The three dates spanning 201 to 201.6 Ma were used since these are reasonably well-constrained correlations, and the dates fall on a linear trend. This gives a basal Lias Gp age of 201.394 Ma, which was used to anchor the astrochronology. If there was a perfect match between the astrochronological durations and the correlated radioisotopic dates, lines parallel to the black fixed-duration line in Figure 15 would be expected. Clearly this is not case for all these dates, either because the SAR changes in parts of the section, the correlations of the radioisotopic dates are incorrect, or the dates themselves are biased. Date LM4 117/118 is clearly inconsistent for one or more of these reasons. The regression fit has a lower gradient than the fixed-duration line, possibly because of a lower SAR in the upper part of the StAB section shown. Date LM4 76/77 may also be part of the regression trend if projected downwards. An alternative possibility is that dates LM4 76/77 and LM4 58/59 may reflect an SAR close to the fixed-duration line when projected down from the top of the WFm (dotted line in Fig. 15). If this were the case, there would be a c. 100 kyr hiatus at around the base of the CMbr. Significantly, the coeval nature of magnetozones E23r and SA5n.3r is demonstrated with these correlations and chronometric estimates, with magnetozone bases falling between the regression-fit and the fixed-duration fit at ∼201.60 Ma.
a) Plot of r2opt with respect to β for the CD-SAR β-H- testing models in Fig. 13. Labelling details as in Figure 10. b) Power spectra of the SAR models included in sets-1 and set-2 from Fig. 13, with linear (black) and logarithmic scaling of spectra (in grey), and eccentricity, obliquity and precession frequencies (red vertical dash-lines) as in SM Table S1. Blue line is the bandpass filter for evaluation of the precession amplitude envelope.
Radioisotopic dates from Peruvian sections (Guex et al., Reference Guex, Schoene, Bartolini, Spangenberg, Schaltegger, O’Dogherty, Taylor, Bucher and Atudorei2012; Wotzlaw et al., Reference Wotzlaw, Guex, Bartolini, Gallet, Krystyn, McRoberts, Taylor, Schoene and Schaltegger2014) correlated with the St Audrie’s Bay (StAB) section using organic carbon isotope excursions as in Ruhl et al. (Reference Ruhl, Hesselbo, Al-Suwaidi, Jenkyns, Damborenea, Manceñido, Storm, Mather and Riccardi2020). Each of the dates (circles) has horizontal and vertical error bars representing the 2σ age uncertainty, and the estimated stratigraphic uncertainty (in the duration scale of the SA2x=1 age model), respectively. Reverse magnetozones and negative carbon isotope excursions at StAB as in Figure 16. Ages of Newark Basin (N), Argana Basin (A) and Fundy Basin (F) CAMP pulses from Blackburn et al. (Reference Blackburn, Olsen, Bowring, McLean, Kent, Puffer, McHone, Rasbury and Et-Touhami2013). The x-axis age scale shows Newark ‘E’ magnetozones from Kent et al. (Reference Kent, Olsen and Muttoni2017). The blue regression line is a fit to the dates LM4-86, LM4-90 and LM4 100/101, using the method of Reed (Reference Reed1989). Black fixed-duration line is a fit to the same three dates, using a weighted regression fit, but with a fixed slope to match the durations in both x and y scales. TJB= Triassic–Jurassic boundary age from Wotzlaw et al. (Reference Wotzlaw, Guex, Bartolini, Gallet, Krystyn, McRoberts, Taylor, Schoene and Schaltegger2014) projected onto the SA2x=1 age model.
Using the regression fit (and CD-SAR model SA2x =1 for scaling) gives ages of 201.609 Ma and 201.494 Ma for the bases of the CMbr and Langport Mbr respectively at StAB. Following the same procedure, ages for the Marshi, Spelae and top Tilmanni CIE’s at StAB are 201.936 Ma, 201.499 Ma and 201.113 Ma, respectively. The uncertainty of these ages is on a similar scale to those on the chronozone durations, with a minimum of c. 50 kyr (not including uncertainty from radioisotopic dates).
5. Discussion
5.a. The Norian–Rhaetian boundary interval
The following relationships for the S-20 dataset can be inferred by comparison with the same successions previously studied at StAB, Seaton, and Lavernock (Fig. 16).
Comparison of the Staithes S-20 core magnetostratigraphy with other equivalent sections through the Sevatian and early Rhaetian. Other section data: St Audrie’s Bay (StAB) (Hounslow et al., Reference Hounslow, Posen and Warrington2004; Hüsing et al., Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014), Seaton, Upper Chinle Fm composite and GPTS-B from Hounslow & Gallois (2022), Lavernock (Hounslow & Andrews, Reference Hounslow and Andrews2024), Newark Supergroup (Kent et al., Reference Kent, Olsen and Muttoni2017). Sampling levels marked as ticks on the Seaton, StAB and Lavernock columns, shown as a green bar when densely sampled. Astrochronologic age (in blue) anchored to the Orange Mountain Basalt (Kent et al., Reference Kent, Olsen and Muttoni2017). Hatching in the S-20 column represents uncertainty regarding the position of the base Hettangian in the core. CBz= conodont biozone, SS=Substage, Ala.3= Alaunian 3. Magnetozone (MZ) names from data sources. Additional abbreviations in key to Fig. 17.
Comparison of the Staithes S-20 core magnetostratigraphy from the Penarth and Lias Groups with other key sections of Rhaetian-2 to early Hettangian age. Other section data: St Audrie’s Bay (StAB) (Hounslow et al., Reference Hounslow, Posen and Warrington2004, Hesselbo et al., Reference Hesselbo, Robinson, Surlyk and Piasecki2002; Hüsing et al., Reference Hüsing, Beniest, van der Boon, Abels, Deenen, Ruhl and Krijgsman2014), Lombardian Basin (Muttoni et al., Reference Muttoni, Kent, Jadoul, Olsen, Rigo, Galli and Nicora2010; Zaffani et al., Reference Zaffani, Jadoul and Rigo2018), Oyuklu (Gallet et al., Reference Gallet, Krystyn, Marcoux and Besse2007), Argana Basin (Deenan et al., Reference Deenen, Langereis, Krijgsman, Hachimi, Chellai, van Hinsbergen, Buiter, Torsvik, Gaina and Webb2011) Newark Basin (Kent et al., Reference Kent, Olsen and Witte1995, Reference Kent, Olsen and Muttoni2017). GPTS-B (Hounslow & Gallois, Reference Hounslow and Gallois2023). Hatching in the S-20 and GPTS-B columns represents uncertainty regarding the position of the Triassic–Jurassic boundary, uncertainty in the StAB column that is shown by the various proposed positions for the boundary, which are labelled: L, R, H and J3, from Lindström et al. (Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017), Ruhl et al (Reference Ruhl, Hesselbo, Al-Suwaidi, Jenkyns, Damborenea, Manceñido, Storm, Mather and Riccardi2020), Hillebrandt et al. (Reference Hillebrandt, Krystyn, Kürschner, Bonis, Ruhl, Richoz, Schobben, Urlichs, Bown, Kment and McRoberts2013) and Jeram et al. (Reference Jeram, Simms, Hesselbo and Raine2021), respectively (also in Fig. 3b).
In S-20, the Blue Anchor Fm is of normal polarity, but in the StAB, Seaton and Lavernock sections it is dominantly of reverse polarity (left columns in Fig. 16). Only the Williton Mbr at StAB is of normal polarity. This indicates that the Blue Anchor Fm in S-20 is not coeval with the Rydon Mbr in that formation in SW England. The most likely possibility is that the Blue Anchor Fm in S-20 is only coeval with the Williton Mbr of the Bristol Channel Basin. Therefore, the bulk of the Rhaetian-1 interval (magnetochrons UT22r–UT24r) can be inferred to be missing from the S-20 core (Fig. 16).
In S-20, the BM1n– BM3n polarity interval in the Branscombe Mudstone Fm is a reasonable match with those from StAB (SA2n–SA3n), Lavernock (LP2n–LP3n), and the Newark Supergroup (E15n to E17n) in the eastern USA (Fig. 16). The same interval is present in the Upper Chinle Fm farther west in the USA and in the marine biochronology-based geomagnetic polarity timescale (GPTS-B; Fig. 16). Correlations of Sevatian age sections between SW England and those in North America have been suggested by Hounslow et al. (Reference Hounslow, Posen and Warrington2004), Kent et al. (Reference Kent, Olsen and Muttoni2017), Hounslow & Gallois (Reference Hounslow and Gallois2023) and Hounslow & Andrews (Reference Hounslow and Andrews2024). Within BM1r, the submagnetozone BM1r.1n is probably coeval with E15r.1n in the Newark Supergroup, CC6r.1n in the Upper Chinle Formation and UT21r.1n in GPTS-B (Fig. 16).
Correlatives of the reverse submagnetozones in BM1n in S-20 have also been detected in the coeval CC6n in the Chinle Formation, PM10n at Pizzo Mondello (Kent et al., Reference Kent, Olsen and Muttoni2017) and the F+ magnetozone at Kavur Tepe (see Hounslow & Muttoni, Reference Hounslow, Muttoni and Lucas2010 for details). These reverse submagnetozones are expressed within UT21n in the biochronology-constrained GPTS-B, within the earliest Sevatian (Fig. 16).
In S-20, the UT22r–UT24r hiatus is placed at the top of the variegated interval, probably at the anhydrite-rich bed at 394.91 m (1304.2 ft), with the overlying 0.3 m showing transitional character into the basal Blue Anchor Fm (SM Figs, S1f, S3d). The uppermost sample in the Branscombe Mudstone Fm is of reverse polarity and is tentatively considered equivalent to part of magnetozone SA4r/UT24r (Fig. 16). The Norian–Rhaetian boundary interval (NRB1 to NRB2) is placed within this substantial hiatus, which is associated (Fig. 1b) with the combined D5.1 and D5.2 disconformities in the Germanic Keuper (Barnasch, Reference Barnasch2010).
5.b. Rhaetian 2 to 4 and the Penarth Group
The magnetic polarity timescale through the Rhaetian-2 to Rhaetian-4 is currently not entirely resolved, although this interval is largely dominated by normal polarity (Fig. 17). Broadly, there are two options for Rhaetian-3. The first is shown as GPTS-B in Figure 16, with two major reverse magnetochrons, UT25r and UT26r, in Rhaetian-2 and Rhaetian-3, respectively. This option largely derives from the conodont-dated Oyuklu section in Turkey (Gallet et al., Reference Gallet, Krystyn, Marcoux and Besse2007), a more poorly dated Argana Basin (Morocco) section (Deenen et al., Reference Deenen, Langereis, Krijgsman, Hachimi, Chellai, van Hinsbergen, Buiter, Torsvik, Gaina and Webb2011), and the Newark Basin, eastern USA, datasets (Kent et al., Reference Kent, Olsen and Witte1995; Kent et al., Reference Kent, Olsen and Muttoni2017). The base of the Oyuklu section is probably truncated by two thrusts, which may have removed a substantial part of Rhaetian-1 (below the shown part in Fig. 17). The reference pattern for UT25r–UT26r is based on the Newark Basin succession, which is well constrained near the TJB (see Hounslow & Gallois, Reference Hounslow and Gallois2023).
The alternative option for the UT25r – UT26r interval is mostly derived from sections in the Lombardian Basin (northern Italy) and is shown next to the GPTS-B (Fig. 17). The Lombardian sections have limited biostratigraphic age control, other than in the latest Rhaetian-4 and from conodonts in the ZU1 interval of the Zu Limestone Fm, below the parts of the section with a magnetostratigraphy; but they do have a detailed δ13Corg record (Fig. 17). These sections display considerably more reverse polarity in UT26.
Reconciling these options requires either that substantial parts are duplicated and/or missing from ZU3a–ZU3b in the Brumano section in Lombardy or that parts are missing from UT26n in both the Newark Basin and Oyuklu sections. A fault in E22n in the Martinsville core in the Newark Basin (Kent et al., Reference Kent, Olsen and Witte1995; Olsen et al., Reference Olsen, Kent, Cornet, Witte and Schlische1996) could account for part of this missing interval (Fig. 17). Some 0.7 Myr missing would bring the predicted Rhaetian durations derived by Galbrun et al. (Reference Galbrun, Boulila, Krystyn, Richoz, Gardin, Bartolini and Maslo2020) and Kent et al. (Reference Kent, Olsen and Muttoni2017) closer together.
Mercury contents from multiple locations associate CAMP volcanism with the end-Triassic extinction and the Spelae CIE, which is in the upper part of the CMbr to basal Langport Mbr (Yager et al., Reference Yager, West, Thibodeau, Corsetti, Rigo, Berelson, Bottjer, Greene, Ibarra, Jadoul and Ritterbush2021). This relationship is confirmed by the correlated radioisotopic dates (Fig. 15). These all suggest that E23r in the Newark Basin and coeval submagnetozones in the Argana and Fundy basins (Deenan et al., Reference Deenen, Langereis, Krijgsman, Hachimi, Chellai, van Hinsbergen, Buiter, Torsvik, Gaina and Webb2011), immediately prior to the onset of CAMP, are equivalent to SA5n.3r at StAB (Fig. 15), BM3r.4r in the basal CMbr in S-20, and UT27r in the GPTS-B (Fig. 17). A single-sample reverse magnetozone is also present in the uppermost part of the Zu Limestone Fm in the Lombardian Basin, which may represent the equivalent of UT27r ≡E23r (≡ symbolises coeval magnetozones or magnetochrons).
There is a dominance of reverse polarity in the WFm in S-20, but conversely of normal polarity in the WFm at StAB and Lavernock (Fig. 16). The simplest explanation is that differing amounts are missing because of hiatuses at the base and top of the WFm, which are inferred to be the D6 and D7 disconformities of the Germanic Keuper (Figs. 1b, 16). Possible explanations are that magnetozone BM3r at S-20 is equivalent to UT25r or that BM3r is coeval with the alternative, more complex version of UT26 (shown next to GPTS-B in Fig. 17). In either case the ‘upper part’ of the WFm is absent at the hiatus at the base of the CMbr in S-20, with rather less missing for the alternative correlation to UT26. Conversely, a ‘lower part’ of the WFm must be missing at StAB, with a more limited part missing at its junction with the CMbr (Gallois, Reference Gallois2009), as is implied by the correlated chronometric dates (Fig. 15). In this scenario the Marshi CIE at StAB (Lindström et al., Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017) may be equivalent to the E4 CIE in the Italcementi section (Lombardy) since this is slightly above the equivalent of UT26r in both sections (Fig. 17). The projected age mismatch between the reverse magnetozones SA5n.1r≡E22r in the Newark Supergroup (Fig. 15) is probably due to the imprecise correlation of δ13Corg records between the Peruvian sections and StAB (as in Ruhl et al., Reference Ruhl, Hesselbo, Al-Suwaidi, Jenkyns, Damborenea, Manceñido, Storm, Mather and Riccardi2020, Fig. 3), which locates date LM4 58/59 too low in the correlation to the StAB section (Fig. 15). Rather more is missing from the basal WFm below LP3r.6r≡UT26r at Lavernock (Fig. 16).
The reverse submagnetozone in the base of the CMbr (BM3r.4r ≡SA5n.3r; Fig. 17) is some 7 to 44 cm thick in S-20 and ∼5 to 20 cm thick at StAB (ranges from sample spacing). Projecting down the SAR in these sections suggests durations for those submagnetozones of 2–7 kyr and 2–8 kyr, respectively; these are shorter, but otherwise like the c. 11 kyr is suggested for the coeval E23r in the Newark Supergroup and elsewhere in eastern North America (Blackburn et al., Reference Blackburn, Olsen, Bowring, McLean, Kent, Puffer, McHone, Rasbury and Et-Touhami2013). At StAB the overlying SA5r submagnetozone is some 28–59 cm thick, corresponding to a duration of 12–25 kyr. SA5r is some 28 kyr younger than the base of the Lias Gp and is at a level that was inferred to be near the base Hettangian by Hillebrandt et al. (Reference Hillebrandt, Krystyn, Kürschner, Bonis, Ruhl, Richoz, Schobben, Urlichs, Bown, Kment and McRoberts2013), Lindström et al. (Reference Lindström, Van de Schootbrugge, Hansen, Pedersen, Alsen, Thibault, Dybkjær, Bjerrum and Nielsen2017), and Ruhl et al. (Reference Ruhl, Hesselbo, Al-Suwaidi, Jenkyns, Damborenea, Manceñido, Storm, Mather and Riccardi2020). Using the age anchor for the astrochronology indicates that the base and top of SA5r are 201.32 and 201.28 Ma in age, respectively. Currently, an equivalent of SA5r≡ LP4n.1r (Fig. 16) has not been detected in North America, where it should lie in the middle of the CAMP successions (Fig. 15). Reversed polarity units were reported by Knight et al. (Reference Knight, Nomade, Renne, Marzoli, Bertrand and Youbi2004) in the middle of the CAMP succession in Morocco, but these were refuted by Font et al. (Reference Font, Youbi, Fernandes, El Hachimi, Kratinova and Hamim2011). Reverse polarity units occur at around 201.5 to 201.3 Ma in the CAMP in Brazil (Moreira et al., Reference Moreira, Ernesto, De Min, Marzoli, Machado, Vasconcellos and Bellieni2023), and brief reversals occur in possibly coeval parts of the Montcornet core in the Paris Basin (Yang et al., Reference Yang, Moreau, Bucher, Dommergues and Trouiller1996). Rather rare CAMP dykes with reverse polarity are also known (Smith, Reference Smith1987; Palencia Ortas et al., Reference Palencia Ortas, Osete, Vegas and Silva2006). It therefore appears that the equivalent of SA5r≡ LP4n.1r≡ UT28r is still to be clearly recognized in other basins following the initial CAMP eruptions.
5.c. End-Triassic extinction scenarios and synchronicity of CAMP
With the improved chronological and magnetostratigraphic correlation to the initial CAMP basalts provided by this work, the lower part of the Cotham Mbr is clearly coeval with the start of CAMP, some 10 to 20 kyr after the top of E23r≡ SA5n.3r≡ BM3r.4r≡ UT27r (Blackburn et al., Reference Blackburn, Olsen, Bowring, McLean, Kent, Puffer, McHone, Rasbury and Et-Touhami2013). This associates the floral changes in the SAB2 Az in the UK (Fig. 5a) with the start of the crisis interval (Lindström, Reference Lindström2021). Since there is a substantial hiatus between the WFm and the CMbr in S-20 (Fig. 17), the abrupt changes in eco-plant proxies EGT and EPH across this boundary also reflect the time missing at this hiatus.
Scenarios for flood basalt-promoted extinctions are broadly twofold. Firstly, extreme greenhouse conditions caused by large volumes of volcanic CO2 and associated intrusion-heating of carbon-rich sediments, and secondly, temporary icehouse conditions prompted by large volumes of volcanic-associated SO2, atmospheric poisoning, associated cooling, and glacioeustatic-forced regression (Self et al., Reference Self, Widdowson, Thordarson and Jay2006; Schaller et al., Reference Schaller, Wright and Kent2011; Guex et al., Reference Guex, Pilet, Müntener, Bartolini, Spangenberg, Schoene, Sell and Schaltegger2016; Algeo & Shen, Reference Algeo and Shen2024). Relative temperature estimates from the EGT proxy in S-20 suggest a decline from an interval around or just below the probable Spelae CIE, a rise again near the base of the Lilstock Fm and a further decline into the Redcar Mudstone Fm (Fig. 5c). This broadly corresponds with temperature changes at a similar time inferred at Hochalplgraben (Austria) by Bonis & Kürschner (Reference Bonis and Kürschner2012) using spore and pollen data. If this cooling was accompanied by ∼16oC temperature seasonality, as inferred by Petryshyn et al. (Reference Petryshyn, Greene, Farnsworth, Lunt, Kelley, Gammariello, Ibarra, Bottjer, Tripati and Corsetti2020), coincident with the Spelae CIE, this could explain the large increase in eurythermic taxa over the SAB2/SAB3 zonal interval (Fig. 5c). Alternative plausible enhancement mechanisms leading to cooling are SO2 generation by heating sediments to 300–500oC caused by CAMP sill and dyke injections (Kaiho et al., Reference Kaiho, Tanaka, Richoz, Jones, Saito, Kameyama, Ikeda, Takahashi, Aftabuzzaman and Fujibayashi2022) or via sulphur liberated from initial lithospheric melting (Guex et al., Reference Guex, Pilet, Müntener, Bartolini, Spangenberg, Schoene, Sell and Schaltegger2016). The SO2 forced cooling scenario seems to offer a better explanation of our dataset, although the changes observed could reflect regional palaeoclimate, rather than the expected global and/or seawater temperatures, which largely implicate a temperature increase due to CAMP CO2 inputs (Korte et al., Reference Korte, Hesselbo, Jenkyns, Rickaby and Spötl2009; Algeo & Shen, Reference Algeo and Shen2024).
In contrast, the analysis of the StAB section by Bonis & Kürschner (Reference Bonis and Kürschner2012) suggests an overall increased temperature from the Spelae CIE, but with a temporary temperature decline in the upper Langport Mbr. This is consistent with a possible ∼8oC temperature increase into the Lias Gp based on δ18O data from Lavernock (Korte et al., Reference Korte, Hesselbo, Jenkyns, Rickaby and Spötl2009) and clumped isotope data from the Cotham Marble (at the Spelae CIE level), which implies no cooling but ∼16oC temperature seasonality (Petryshyn et al., Reference Petryshyn, Greene, Farnsworth, Lunt, Kelley, Gammariello, Ibarra, Bottjer, Tripati and Corsetti2020). However, it is not clear that the ordination coordinates derived from the miospore species composition by Bonis & Kürschner (Reference Bonis and Kürschner2012) reflect similar environmental responses at both StAB and Hochalplgraben, since the major variance is inferred to have been in temperature at the former and in humidity at the latter. This suggests that the ordination scores may be dominated by local effects at both locations. Probably a more consistent inter-section approach should be applied to miospore data to extract regional responses and reduce local controls (Bhatta et al., Reference Bhatta, Mottl, Felde, Flantua, Birks, Cao, Chen, Grytnes, Seddon and Birks2023).
6. Conclusions
The Staithes S-20 magnetostratigraphy indicates that a substantial hiatus exists between the Branscombe Mudstone Fm and the base of the Blue Anchor Fm and that this includes the Norian–Rhaetian boundary interval. This hiatus probably corresponds with the combined D5.1 and D5.2 disconformities of the Germanic Keuper. The Blue Anchor Fm and WFm in S-20 are not synchronous with the equivalent formations in SW England. In S-20, the Blue Anchor Fm is probably equivalent to the youngest (Williton) member of the Blue Anchor Fm in SW England. The WFm in S-20 is older (either late Rhaetian-2 or early Rhaetian-3) than that seen in SW England, which is coeval with late Rhaetian-3 and most of Rhaetian-4. A reverse polarity magnetozone in the basal part of the CMbr in S-20 is coeval with reverse magnetozone SA5n.3r detected at the same stratigraphic level at StAB. Changes in miospore taxa in S-20 are like those at StAB and elsewhere in the UK and include the peak in spore abundance that is typical of the Lilstock Fm and its lateral equivalents. Eco-plant model assessment indicates an increase in humidity in the Lilstock Fm, with increases in eurythermic and euryphyte miospore taxa connected with the peak in spore abundance, followed by cooling into the basal Hettangian.
A joint astrochronology for the earliest Hettangian chronozones from S-20 and the StAB, Lavernock and Lyme Regis sections is anchored to Peruvian radioisotope dates correlated into the StAB section using organic carbon isotope datasets. The astrochronology is anchored on the basal Lias Group at 201.394 Ma. The Tilmanni Cz and Planorbis Scz durations are determined as 262±53 kyrs and 238±27 kyr, respectively, values intermediate between previous estimates.
The anchored astrochronology demonstrates that the reverse magnetozone SA5n.3r (and LBM3r.4r at S-20), in the base of the CMbr, is coeval with magnetozone E23r in the Newark Supergroup, and that the Spelae CIE at StAB and the palynological turnovers are therefore associated with the initial phases of the CAMP. The miospore compositional changes and inferences from the eco-plant model evaluation are consistent with a cooling and enhanced seasonality, conditions prompted by large volumes of CAMP-associated SO2 and atmospheric poisoning. Alternatively, the miospore changes may reflect a regional response superimposed on a more global warming trend.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0016756825100162
Acknowledgements
Authors MWH and GW were in part supported by Norsk Hydro, Saga Petroleum and Deminex. Dennis Kent is thanked for the use of the palaeomagnetic facilities at Lamont-Doherty Observatory. Staff in the core store at BGS are thanked for facilitating access to the core. Simon Harris (BGS) provided imaging of the core and ammonite specimens. Boulby mine Chief Geologist provided permission for sampling. The reviewers, Sophie Lindström and an anonymous one provided constructive comments.
Data availability
All the data and R scripts used here are either hosted on figshare (Hounslow, Reference Hounslow2025) or contained in the Supplementary Material.
Competing interests
The authors declare none.
Open access
