Using U–Pb carbonate dating to constrain the timing of extension and fault reactivation within the Bristol Channel Basin, SW England

The Bristol Channel Basin (BCB) (Fig. 1) initially formed during extension in the Permian–Triassic caused by rifting associated with the break-up of Pangaea (Debenham et al. 2020). This led to the formation of a series of east–west-striking normal faults (Dart et al. 1995; Nemčok et al. 1995; Peacock and Sanderson 1999; Glen et al. 2005). The reactivation of Variscan thrust faults in the underlying pre-Permian basement (Miliorizos et al. 2004) is interpreted to have partly controlled the development of normal faults and may be responsible for the overall east–west structural trend of the entire basin (Dart et al. 1995). The most extensive exposures of Liassic (Early Jurassic) rocks on the south side of the BCB are between Kilve and East Quantoxhead where they are cut by multiple fault populations, which are the focus of this study (Fig. 1). Burial estimates, based on the maturation of organic material in Liassic sedimentary rocks, suggest that maximum burial of c. 1.7 km and temperature of c. 65°C occurred during the Aptian (c. 125–113 Ma), with much of the Jurassic strata having been eroded during uplift from c. 55 Ma to the present day (Cornford 1986).

East–west-striking normal faults in the BCB (Peacock and Sanderson 1991, 1993, 1999; Dart et al. 1995; Nemčok et al. 1995; Kelly et al. 1999; Glen et al. 2005; Peacock et al. 2017) formed during Triassic north–south extension, under relatively low-stress conditions (c. 11 MPa) (Nemcok and Gayer 1996). The timing of deformation is constrained by synsedimentary deformation of the Mercia Mudstone Group (Fig. 1) (Nemčok et al. 1995). The normal faults formed with $σ$₁ subvertical and $σ$₃ plunging gently NNE (Peacock et al. 2017), consistent with the broader interpretation of north–south extension based on kinematic analysis (Nemčok et al. 1995; Kelly et al. 1999; Peacock and Sanderson 1999; Glen et al. 2005).

Strike-slip faults that are conjugate about north–south (Kelly et al. 1998, 1999), thrust faults that crosscut normal faults (Peacock et al. 2017; Rotevatn and Peacock 2018) and the reactivation of pre-existing normal faults (Kelly et al. 1999) document a stage of contraction in the BCB that post-dates initial extension (Dart et al 1995; Nemčok et al. 1995; Kelly et al. 1999; Peacock and Sanderson 1999; Glen et al. 2005; Peacock et al. 2017). During the later stage of contraction, $σ$₁ is interpreted to have plunged gently to the south (Peacock et al. 2017). Reactivation is seen in normal faults with >22 m throw, such as the East Quantoxhead Fault (EQHF) and Blue Ben Fault (Kelly et al. 1999). This period of north–south contraction is interpreted to have occurred in the Cenozoic based on monoclinal folding in the adjacent Wessex Basin (Chadwick 1993). Inversion is also thought to have led to an increase in fluid pressure, which led to increased reactivation (Williams et al. 2005).

Veins are often used as proxies to study palaeo-fluid flow (Nemčok et al. 1995; Passchier and Trouw 2005; Philipp 2012; Spruženiece et al. 2021). The composition, distribution and position relative to other structures reveal the depth, source of fluids and opening mode of veins. Furthermore, dating of mineralization provides a tool for understanding the timing of fluid flow and interplay between fluids and the structural development of an area. Petrographical analysis of veins within the BCB shows evidence for blocky veins formed by epitaxial growth on seed grains in open fractures, which precipitated and sealed in one event (Spruženiece et al. 2021). Geochemical analysis of the calcite veins at Kilve yielded evidence of radiogenic alteration by radiogenic basinal fluids sourced from deeper sediments indicated by higher ⁸⁷Sr/⁸⁶Sr ratios (Bixler et al. 1998; Debenham et al. 2020). Oxygen and carbon isotope data show that the mineralizing fluids were 20–30°C hotter (80–110°C) than the surrounding host rocks (60–80°C) (Philipp 2012).

Prior to sampling, faults were mapped in the field and fault types were determined using slickenfibre kinematics, fault dip, bed matching and other indicators such as en echelon veins. Faults with clear kinematics (dip-slip or strike-slip) and evidence of synkinematic calcite slickenfibres were selected for sampling. Twenty-one faults were sampled across the foreshore from East Quantoxhead to Kilve beach (Tables 1 and 2; Fig. 2). Once faults were characterized, slickenfibres from the fault core were sampled and then made into thin (30 µm) and thick (80 µm) polished sections for laboratory analysis.

Fault data underwent kinematic analysis in Stereonet 11 and FaultKin8 (Marrett and Allmendinger 1990; Allmendinger 2012). Principal stress axes were determined using fault orientation data with slickenfibre pitch recorded in the field. Kinematic results have been combined with U–Pb carbonate ages where possible to determine the absolute temporal evolution of the stress field.

Thin sections were used for microtextural analysis and to determine grain-size distribution and recrystallization textures in veins before U–Pb analysis was undertaken on the corresponding thick sections. Thin sections were analysed using plane- and cross-polarized light (PPL and XPL). Undulose extinction and sub-grains were used to infer, respectively, crystal plastic and recovery. Dynamic recrystallization comprised bulging (BLG), and subgrain rotation (SGR) recrystallization types (Passchier and Trouw 2005). Calcite twin type was used to determine palaeotemperature using the technique of Passchier and Trouw (2005).

Minerals within the thick sections were identified using a JEOL 7001F FE-SEM at Plymouth Electron Microscopy Centre, University of Plymouth, with an Oxford Instruments X-Max 50 mm² energy-dispersive spectroscopy detector and AZtec v6.0 software. Operating conditions were accelerating voltage 20 keV, probe current c. 6 nA and working distance 10 mm. This analysis was carried out using internal standards and with a deadtime of 35–50%.

Dating of synkinematic calcite slickenfibres allows for ages to be attributed to the timing of movement along a fault. In situ U–Pb carbonate geochronology was conducted at the University of Portsmouth by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). Laser ablation was conducted using an ASI RESOlution© 193 nm ArF excimer laser coupled to a high-sensitivity Jena Analytic PlasmaQuant Elite© ICP-MS (2021–22 analyses) or Agilent 8900 Triple Quadrupole ICP-MS instrument (2023 analyses). Calcite vein samples were analysed in situ in polished 80 µm thick polished sections. Calcite was analysed using 80 µm spot size, laser fluence of c. 2.5–3 J cm^–2, and a repetition rate of 8 Hz. NIST glasses, NIST SRM612 (NIST612; 38 ppm U and 39 ppm Pb; Jochum et al. 2011, for Jena ICP-MS analyses) and NIST SRM614 (NIST614; 0.8 ppb U and 2.3 ppm Pb, Jochum et al. 2001, for Agilent 8900 analyses) and WC-1 carbonate (254.4 ± 6.4 Ma; Roberts et al. 2017) were used as primary reference materials. Mudtank Zircon (732 ± 5 Ma; Black and Gulson 1978; Jackson et al. 2004) and Duff Brown Limestone (64 ± 2 Ma; Hill et al. 2016) were used as secondary reference materials to verify long-term reproducibility. Analysis of Duff Brown during the analytical period yielded a ²⁰⁶Pb/²³⁸U intercept age of 64.35 ± 0.6 Ma (0.54% reproducibility). Analysis of Mudtank Zircon during the analytical period yielded a ²⁰⁶Pb/²³⁸U intercept age of c. 734.7 ± 2.6 Ma (0.37% reproducibility).

U–Pb data were reduced using Iolite©3 software (Paton et al. 2011). Samples generally have low to very low U content (<1 ppm), low Th/U ratios and have been regressed and plotted using both Tera–Wasserburg (TW) and 86TW plots (see Parrish et al. 2018), using IsoplotR (Vermeesch 2018). Tera–Wasserburg plots were calculated with discordia (model-1) age calculations (as displayed in the Supplementary material), and 86TW plots as isochrons with a type 1 2D regression and using model 1 (maximum likelihood) (Vermeesch 2018). Both 86TW and U–Pb plots have used 2 SE (abs) for input and output errors (see Supplementary material for more detail). In all cases, the 86TW plot has less scatter and lower uncertainties, whereas many of the samples have excess scatter using the traditional TW plot, and the 86TW ages and uncertainties are used as the final quoted ages. Final quoted 2$σ$ uncertainties are propagated in quadrature and include fully propagated analytical uncertainties of 1% (see Supplementary material for full details). Chronostratigraphical ages are based on the International Chronostratigraphic Chart v2023/09 (Cohen et al. 2013, updated).

East–west-striking normal faults are extensively exposed in the foreshore, extending hundreds of metres (Figs 2 and 3), with slips ranging from <1 to >20 m. Normal faults terminate in isolated tips, but are more commonly linked by relay zones, which may then breach to form a larger through-going fault as also noted by Peacock and Sanderson (1991, 1993). Some larger normal faults, such as the Kilve Pill Fault, East Quantoxhead Fault and Blue Ben Fault (Fig. 2), show evidence of inversion, localized thrusting and reverse fibres, and folding of the hanging wall can be observed (Figs 2 and 4). Kelly et al. (1999) also observed inversion features in some reactivated normal faults in the region. Normal faults dip both north and south, with dips ranging from c. 40 to 80°. Faults have been analysed as bulk populations, without division based on location in the field. Kinematic analysis of normal faults generated robust estimates for kinematic tensors, producing a vertical $σ$₁ and a horizontal $σ$₃, which trends broadly north–south (Fig. 3).

All studied veins within normal faults are composed of calcite (Fig. 5). The East Quantoxhead Fault (EQHF) also contains localized celestine. Normal fault calcite crystal sizes range from 0.2 to 5.0 mm. All normal faults show evidence for recrystallization and sub-grain development (Fig. 5) with sub-grain diameter ranging from 0.2 to 0.5 mm. Crystal plasticity is inferred from undulose extinction and dynamic recrystallization by sub-grain rotation (SGR), present in all samples, as well as bulging (BLG), which is present only in selected samples (Table 1). Both type 1 and 2 calcite twins are present in normal faults.

NE–SW-striking sinistral strike-slip faults extend >400 m (Fig. 2), and typically have up to 10 m of lateral offset. They can be seen to cross-cut normal faults (Fig. 4). North–south-striking dextral strike-slip faults dip at c. 60–90° and extend for tens of metres. Dextral faults are less pervasive than sinistral ones and are not observed to offset normal faults; instead they are located between overlapping normal faults, as reported by Rotevatn and Peacock (2018). Sinistral and dextral strike-slip faults mutually cross-cut, suggesting that they are a conjugate system. Thrust faults are the least common structure; they are generally observed in mudstone lithologies proximal to reactivated normal faults (Fig. 2). East–west-striking thrust faults have small throws of between 5 and 30 cm and gentle dips ranging from c. 15 to 40°. Kinematic analysis of contractional features suggests a subhorizontal $σ$₁ that trends broadly north–south or NNE–SSW based on thrust and strike-slip data, with strike-slip faults producing a subvertical $σ$₂ and thrust faults producing a subvertical $σ$₃.

Veins associated with sinistral strike-slip faults have crystal diameters that range from 0.5 to 2.5 mm and have sub-grain diameters ranging from 0.2 to 0.5 mm, with two of four samples showing evidence for sub-grain development. Dextral strike-slip fault vein crystal diameters range from 0.05 to 2.5 mm with sub-grain diameters ranging from 0.1 to 0.3 mm, with three of five showing evidence of sub-grain development. Thrust fault crystal diameters range from 0.1 to 5 mm with sub-grain diameters ranging from 0.1 to 0.5 mm. Thrust faults contain en echelon tension gashes, shear vein jogs and crenulations.

All fault types exhibit evidence for dynamic recrystallization (Table 1), but it is less evident in strike-slip and thrust faults than in normal faults. Sub-grain development has occurred in all types of faults, and diameters typically range from 0.1 to 0.5 mm. Normal faults show SGR in all samples, which is widely distributed throughout the samples, ±BLG in selected samples. This is not the case for contractional structures, where evidence for dynamic recrystallization is less distributed throughout individual samples. Seven out of 14 strike-slip or thrusting samples show some evidence for SGR and BLG with ±SGR and BLG present in the remaining samples. All samples show both type 1 and 2 calcite twins (Passchier and Trouw 2005; Fossen 2016).

BLG indicates deformation temperatures up to 250°C and SGR suggests temperatures >250°C (Passchier and Trouw 2005). These observations suggest that normal faults experienced higher deformation temperatures than contractional features.

In total, 21 fault cores have been dated using U–Pb geochronology on calcite slickenfibres and veins. U concentrations range from 0.03 to 2 ppm with >99% of analyses >1 ppm. These are among the lowest U concentrations successfully dated for published samples to date. Twenty to 100 U–Pb calcite analyses were performed on each sample (Fig. 6). Samples are relatively radiogenic and have a significant spread in ²³⁸U/²⁰⁶Pb space (see Supplementary material Table S1). Scatter is relatively minor with MSWDs of 1–3 for all samples using the 86TW plot and regression.

Seven normal faults were dated and yield a range of ages from c. 150.0 ± 2.8 Ma (Kimmeridgian, sample JC-16-21, MSWD = 0.5, n = 50) to c. 120.2 ± 2.0 Ma (Aptian, sample JC-14-21, MSWD = 1.1, n = 45).

Four samples from sinistral strike-slip fault segments range from 41.0 ± 5.7 Ma (Bartonian, sample JC-1a-21, MSWD = 0.95, n = 40) to 34.5 ± 2.5 Ma (Priabonian, sample JC-8-21, MSWD = 0.9, n = 40). Four thrust fault segments range from 41.9 ± 1.3 Ma (Bartonian, sample JC-4-21, MSWD = 0.5, n = 57) to 36.6 ± 2.6 Ma (Priabonian, sample JC-6B-22, MSWD = 1.00, n = 60). Six dextral strike-slip fault segments range from 36.0 ± 1.6 Ma (Priabonian, sample JC-6a-21, MSWD = 1.0, n = 39) to 21.8 Ma ± 0.6 (Aquitanian, sample JC-15-21, MSWD = 1.3, n = 58).

A composite vein (sample JC-3-21) from the EQHF has four domains (A–D) that have been dated individually to constrain its temporal structural evolution (Fig. 7). U–Pb ages range from c. 147 to c. 34 Ma (late Jurassic–Oligocene). Domain A, located at the periphery of the vein, closest to the wall rock yields the oldest age of 147 ± 10 Ma (Tithonian, MSWD = 1.1, n = 39). Domains B–C have younger ages, ranging from 136.5 ± 4.3 Ma (Hauterivian, MSWD 3.1 n = 20) to 116.5 ± 4.9 Ma (Aptian, MSWD = 3.0, n = 20). There are several domains nearer the fault core in which the calcite has a much finer grain size. Celestine occurs as fibrous veins towards the fault core and appears to be spatially associated with the recrystallization of earlier calcite veins. Unfortunately, no age determinations were possible for these recrystallized domains as they did not contain sufficient U. However, the thinnest and finest domain (D) in the fault core was successfully dated and yielded a significantly younger age of 33.5 ± 3.4 Ma (Rupelian, MSWD = 10, n = 64).

Age data show two distinct clusters: normal fault ages range from 150 ± 4 to 120 ± 2.0 Ma whereas strike-slip and thrust fault ages range from 41.9 ± 1.3 to 21.8 ± 0.6 Ma (Fig. 8). The EQHF yields ages that span both these age clusters, suggesting that it has been active for an extended period compared with other individual faults. U–Pb geochronology of the EQHF, combined with microstructural analysis of selected veins, documents the timing of reactivation for the first time (Fig. 7). Vein D yields an age of c. 35 Ma, consistent with the range of overlapping ages obtained from strike-slip faults and thrusts in the area (within error).

Both the relative and absolute timing constraints demonstrate that normal faults were the earliest structures to develop within the Bristol Channel Basin (BCB) (Figs 2, 4 and 8). The range of ages obtained from normal faults indicates a protracted period of regional extension during the Late Jurassic and Early Cretaceous (c. 154–118 Ma; Fig. 8). Along the southern margin of the BCB, normal faults typically strike east–west (Figs 2 and 3), dipping both to the north and south. The largest faults, for example, the EQHF, Blue Ben Fault and the Kilve Pill Fault (Figs 1 and 4), dip to the south and are broadly synthetic to the Central Bristol Channel Fault Zone (BCFZ), which is interpreted to be a reactivated Variscan thrust (e.g. Miliorizos 2004). Kinematic analysis of normal faults suggests a broadly NNE–SSW direction of extension (Fig. 3). Locally, where $σ$₁ axes have more moderate plunges, and normal faults have relatively low angles of dip (<40°), a later tilting of the normal faults, potentially driven by slip on a larger basin bounding fault, is suggested. Peacock et al. (2017) suggested that $σ$₃ within the BCB plunged gently NNE during the Mesozoic, similar to the interpretations presented here.

Extension in the BCB is likely to have been contemporaneous with extension experienced in nearby basins (Figs 8 and 9). The Mizen Basin in the South Celtic Sea is thought to have undergone NW–SE extension from the Berriasian to Hauterivian (c. 145–133 Ma) and a later period of north–south extension from the Aptian to Cenomanian (c. 125–94 Ma) (Rodríguez-Salgado et al. 2020). This later stage of extension is not recorded in the BCB, suggesting that strain may have partitioned away from the BCB by the Late Cretaceous, or later faults are no longer preserved. Age data demonstrate that extension was active in the BCB for a c. 30 Myr period, and although kinematic analysis of the fault population (Fig. 3) produces a reliable overall kinematic estimate, there is scatter between faults with up to 45° difference between fault trends. It is therefore possible that these faults developed during slightly different stress regimes over time, potentially driven by a gradual evolution of regional stress, or local stress perturbations as individual structures developed.

Strike-slip and thrust faults were active during a period of regional contraction from the early Eocene to early Miocene (c. 47–21 Ma). Kinematic analysis of these faults shows that $σ$₁ remained subhorizontal and broadly north–south throughout contraction (Figs 3 and 8) but that the vertical stress axis flipped between $σ$₂ and $σ$₃. Our data do not document a clear progression in stress axis orientation during the transition from regional extension to contraction, as proposed by Peacock et al. (2017). Rather, our data indicate that strike-slip and thrust faults formed coevally, within the uncertainties of the dataset, between c. 47 and 32 Ma (Fig. 8). Such variability may have resulted from a variety of causes including (1) proximity to major faults, such as the East Quantoxhead Fault (e.g. Fossen 2016), (2) position within relay and transfer zones associated with earlier normal faults (e.g. Rotevatn and Peacock 2018) and (3) slightly different levels of bulk shortening and compartmentalization of strain during contraction (e.g. Butler et al. 2006). The BCB system was modelled to have resulted from low maximum shear stress of 10–15 MPa by Stephenson et al. (2020). High fluid pressure in the basin would have reduced the effective stress required for failure during inversion (e.g. Williams et al. 2005; Holford et al. 2008), t**hus reducing the magnitude of stress perturbation required for localized changes in structural style from strike-slip faulting to thrust faulting during contraction.

The contractional episode lasted for a c. 30 Myr period and there is significant scatter between faults within the dataset. It is therefore possible that the principal shortening direction did not remain consistent over this period. These data highlight the care needed when conducting kinematic analysis on bulk fault populations as two seemingly related faults used to define one palaeo-kinematic setting may have in fact formed at significantly different times in the geological past.

Veins associated with extensional structures show evidence for higher temperature deformation when compared with those associated with contraction structures (Table 1). These observations are in accordance with the burial curve of Cornford (1986) (Fig. 8). The Blue Lias Group, the focus of our study, was at a suggested depth of 1.5–1.7 km and temperature of c. 60–70°C during extension. The exhumation path of the burial curve places the same rocks at a depth of 200–400 m and temperature of c. 20°C during later contraction. Passchier and Trouw (2005) suggested that temperatures of up to 250°C and above are required to produce SGR, BLG and type 2 twins in calcite. We propose therefore that the protracted nature of deformation, as documented by new age data, coupled with a generally low-strain environment, allowed for the development of these microstructures at lower temperatures (e.g. Kennedy and White 2001; Lacombe 2022).

Multiple vein generations developed in the core of the EQHF are evidence for protracted, episodic extension of this structure from the late Jurassic to the later stages of the early Cretaceous (Figs 7 and 8). The structure therefore developed throughout the entire (c. 30 Myr) period of extension as defined by the range of ages recorded from individual (unreactivated) normal faults in the area (Figs 7 and 8). This caused the EQHF to become one of the largest faults in the area, making it more prone to reactivation during contraction.

Larger normal faults (>20 m offset) are prone to reactivation, as discussed by Kelly et al. (1999). U–Pb geochronology has pinpointed precisely when reactivation occurred. The Blue Ben Fault reactivated at 39 ± 4 Ma via a hanging-wall thrust shortcut (Fig. 2), whereas the fault core of the EQHF underwent direct reactivation and slipped at 34 ± 3 Ma (Fig. 7). All normal faults within the study area, except the EQHF, show evidence for only single slip episodes or they are confined to movement during the extensional phase (stage 1). Reactivation did not seem to occur at the beginning of the contractional phase. It took the EQHF c. 15 Myr to reactivate, suggesting that either an extended period of strain accumulation was required to trigger reactivation or the initial contractional stress field inhibited reactivation and favoured the development of strike-slip faulting.

Both contraction and extension were protracted events where strain accumulated in environments with relatively low stress magnitudes (Nemcok and Gayer 1996; Stephenson et al. 2020). A high fluid pressure was probably required to reduce both the effective and differential stress acting across faults and formations, allowing fault reactivation (Turner and Williams 2004) and the propagation of new structures.

The newly constrained period of inversion presented here for the BCB broadly overlaps with widespread basin inversion, which peaked at c. 34 Ma across northern Europe (Figs 8 and 9) (e.g. Parrish et al. 2018; Blaise et al. 2022; Monchal et al. 2023). The cause of this Oligocene and younger basin inversion across Europe remains a matter of debate, with faulting attributed to either late-stage Pyrenean or Alpine deformation. However, the wealth of recent in situ carbonate dating studies, where >299 samples have been dated across the Pyrenees, Jura and sedimentary basins across Europe (Fig. 9 and references therein), allows for basin inversion drivers to be put into a direct temporal framework for the first time.

Pyrenean fold and thrust belt development occurred from c. 60 to 20 Ma, where peak deformation, as recorded by carbonate and ⁴⁰Ar/³⁹Ar fault gouge ages, occurred at c. 45 Ma (Cruset et al. 2020; Hoareau et al. 2021; Parizot et al. 2021; Muñoz-López et al. 2022a, b; Bilau et al. 2023; Haines and van der Pluijm 2023). Late c. 30–16 Ma faulting in the Pyrenees is attributed to the exhumation of the Pyrenean hinterland (Cruset et al. 2020), whereas the main phase of Alpine foreland deformation occurred c. 25–5 Ma (Pfiffner 2002; Sinclair et al. 2005). Carbonate U–Pb dating has revealed decoupling of the Alpine Molasse Basin formed faults in the Jura from c. 14 to 4.5 Ma, with a dominant peak c. 10 Ma (Looser et al. 2021; Smeraglia et al. 2021; Madritsch et al. 2024).

The BCB is within the foreland of both the Alpine and Pyrenean orogenic belts, and is also adjacent to the Mid-Atlantic Ridge (Fig. 10). The new age data presented here show that BCB inversion occurred from c. 50 to 20 Ma, with a peak at c. 36 Ma. This exactly fits with the pattern of far-field basin inversion across Europe (Fig. 9) including c. 34 Ma in the Wessex Basin (Parrish et al. 2018), c. 48–43 Ma in the Paris Basin (Blaise et al. 2022) and during the Eocene in the Irish basins (Monchal et al. 2023). These events correlate most closely with post-main phase Pyrenean deformation (Figs 8 and 9) and, when seen as a whole, the compiled European carbonate U–Pb dates demonstrate the significance of Pyrenean deformation for basin inversion across Europe (Fig. 9). We therefore propose that the initial inversion of the BCB was probably driven by the Pyrenean Orogeny. The final stages of BCB inversion from the Oligocene onwards may have been driven by the Alpine Orogeny; however, much younger-aged deformation in the Jura suggests that late-stage exhumation from the Pyrenees may have been a more important driver. Far-field stress modelling conducted by Stephenson et al. (2020) showed that significant stress may have also been transmitted to the BCB during the period of inversion from the opening of the Mid-Atlantic Ridge, and it is likely that this event exacerbated the inversion of the BCB. The stress transmitted from these events was unlikely to have been sufficient to cause the basin inversion alone (Williams et al. 2005; Holford et al. 2009; Stephenson et al. 2020).

1. Structural observations and microtextural analysis have been combined with U–Pb dating to determine absolute timing and kinematic evolution of structures within the Bristol Channel Basin for the first time.

2. Normal faults displaying consistent normal dip-slip slickenfibres have associated veins that yield ages in the range of c. 154–118 Ma and provide absolute age constraints for basin development. These structures exhibit an east–west trend suggesting that there was a protracted c. 30 Myr period of north–south extension from the Late Jurassic to Early Cretaceous. These ages can be extrapolated to better constrain the timing of extension of other rift basins in southern Britain, previously mostly dated through relative constraints. These basins include the Mizen Basin in the South Celtic Sea and the Wessex, Weald and Channel basins.

3. Prominent dextral and sinistral strike-slip faults and localized thrust faults yield ages in the range of c. 47–21 Ma. These faults document a period of protracted north–south contraction acting on the basin from the Early Eocene to Early Miocene. Strike-slip and thrust faults formed coevally, suggesting that $σ$₂ and $σ$₃ remained similar in magnitude throughout the Eocene–Oligocene, and that minor changes in local confining forces and fluid pressure were able to flip these principal stress axes.

4. Structures with clear evidence of reactivation (typically >22 m throw) show evidence of multiple fluid infiltration events associated with both normal and reverse slip episodes. In the case of the regionally significant East Quantoxhead Fault, at least three discrete events are recognized. Each texturally distinct vein yields isotopically robust ages ranging from c. 150 to 34 Ma.

5. Cenozoic tectonic inversion of the Bristol Channel Basin occurred during the Eocene–Miocene, contemporaneous with other basins across Europe including the Wessex Basin, Paris Basin, Ireland and the Mizen Basin. Basin inversion was driven by two main tectonic events that caused far-field stresses across Europe, predominantly far-field stresses associated with Pyrenean deformation during the Eocene, with Oligocene–Miocene drivers from Alpine deformation and the opening of the Mid-Atlantic Ridge.

We would like to thank reviewers D. Chew and R. Shail and editor P. Meere for their feedback on the initial submission of the paper. Their comments and advice have thoroughly improved the quality and overall messaging of the paper. We would like to thank D. Peacock for his expert guidance in the field, as well as extensive help with early drafts of the paper. His wealth of knowledge and detailed publications on the region form the scientific foundations on which this project has been built. We would like to thank G. Long, G. Chapman and J. Dunlop for technical assistance at the University of Portsmouth. We would also like to thank W. Andrews at the University of Plymouth, and P. Connolly for valuable advice in structuring the paper in the early stages as well as assistance in the field.

JC: conceptualization (supporting), data curation (lead), formal analysis (lead), methodology (equal), writing – original draft (lead), writing – review & editing (lead); MA: conceptualization (lead), data curation (supporting), formal analysis (supporting), funding acquisition (lead), investigation (equal), methodology (equal), project administration (supporting), resources (supporting), software (supporting), supervision (equal), validation (equal), visualization (equal), writing – original draft (supporting), writing – review & editing (supporting); CM: conceptualization (supporting), data curation (equal), formal analysis (equal), funding acquisition (supporting), investigation (equal), methodology (lead), project administration (supporting), resources (equal), software (equal), supervision (equal), validation (equal), visualization (equal), writing – original draft (supporting), writing – review & editing (supporting); GDP: data curation (supporting), writing – original draft (supporting), writing – review & editing (supporting); RP: formal analysis (supporting), methodology (supporting); DS: formal analysis (supporting)

This paper contains work conducted during a PhD study undertaken as part of the Centre for Doctoral Training (CDT) in Geoscience and the Low Carbon Energy Transition. It is sponsored by the University of Plymouth (50%) and NeoEnergy Upstream (50%), whose support is gratefully acknowledged. The U–Pb analyses were undertaken in the Laser Ablation Mass Spectrometry Geochronology laboratory at the University of Portsmouth, and the underpinning financial and technical support is gratefully acknowledged.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

All data generated or analysed during this study are included in this published article and its supplementary information files.