Carbon and oxygen isotope characteristics of the Clydach Valley Subgroup, Courceyan, South Wales‐Mendip shelf, UK

The Clydach Valley Subgroup (Courceyan) records a dip section through an Early Carboniferous shallow marine, carbonate shelf and consists of three oolitic formations separated by paludal/peritidal units with abundant evidence of subaerial exposure in proximal areas. The lower part correlates with the Kinderhookian–Osagean Boundary Excursion, with allochem data indicating a minimum δ13C value for marine carbonate of +4.5‰, with associated δ18O of −4.6‰. Marine carbonate δ13C and δ18O values of +2‰ and −2‰, respectively, were estimated for the younger part. Allochem isotopic data yield a well‐defined mixing line consistent with stabilisation in meteoric water at varied water: rock ratios, the degree of stabilisation increasing up‐dip and up‐stratigraphy. Two distinct diagenetic styles closely correlate with evidence for the presence (Diagenetic Regime 1) or absence (Diagenetic Regime 2) of subaerial exposure, non‐ferroan and ferroan calcite cement dominating respectively. Five cement zones (Zones 2–6) defined by iron content occur in Diagenetic Regime 1, irrespective of formation, indicating a similar sequence of palaeohydrological changes affected repeated depositional cycles. Zones 3–5 are considered meteoric based on their isotopic composition (δ18O: −5.8 to −11.1‰ and δ13C: −3.7 to −6.2‰) and form distinct clusters dependent on zone, age and location. Pedogenic carbonates and meteoric cements record a long term increase in meteoric δ18O values: −7.8‰ during the Kinderhookian–Osagean Boundary Excursion, −6.9‰ during diagenesis of the upper part of the subgroup and −6.3‰ associated with initial deposition of the overlying Llanelly Formation. This is consistent with global sea water trends, but an element of climate change cannot be ruled out. Increases in cement δ18O values as meteoric systems become established (Zone 3–4) indicate repeated short term variation in rainwater composition probably driven by climate change, but also suggests a link between climate and the depositional cycle.


| INTRODUCTION
The Early Carboniferous has been the subject of much geological research over the last 20 years because it represents a time of transition between greenhouse and icehouse climates, culminating in the Late Palaeozoic Ice Age (LPIA). Concurrently, there has been a revolution in analytical capability. One outcome has been significant advances in carbon and oxygen isotope stratigraphy. In particular, recognition of a global positive anomaly in δ 13 C values, referred to as the Kinderhookian-Osagean Boundary Excursion (KOBE) (Saltzman, 2002) or Tournaisian Isotope Carbon Excursion (TICE; Yao et al., 2015). The cooccurrence of glacial deposits in South America (Caputo et al., 2008;Isbell et al., 2012;Montañez, 2022) and a significant sea-level fall (Haq & Schutter, 2008;Krammer & Matchen, 2008;Liu et al., 2019), suggests the KOBE event coincided with a transitory glacial episode in the mid-Tournaisian. Consequently, the Early Carboniferous was a time of climate instability, with potential consequences for depositional geometry, facies and early diagenesis.
The South Wales-Mendip Shelf is the only place in Britain where a near complete succession of Tournaisian limestones outcrop, but has been a backwater in terms of recent research. The most recent regional sequence stratigraphic model dates back to the 1990s (Burchette et al., 1990) and has never undergone significant updating, although the recent reassignment of the Gilwern Oolite Formation, the uppermost formation of the Clydach Valley Subgroup, from the Chadian to the Courceyan (Hogancamp & Wright, 2020) illustrates that this is overdue. The absence of progress partly reflects the lack of modern, systematic stratigraphic work which means correlation with other regions is poor.
A number of publications dating from the 1980s discussed climate and climate change on the South Wales-Mendip Shelf based on comparisons between calcretes, karst and early diagenetic styles with modern analogues (Hird & Tucker, 1988;Riding & Wright, 1981;Searl, 1988bSearl, , 1989bWright, 1980Wright, , 1982aWright, , 1984Wright, , 1988, illustrating the potential of the South Wales-Mendip Shelf to make a significant contribution to global understanding of climate change during the transition to the LPIA. Climate change can also leave a geochemical imprint on rocks in terms of the δ 18 O composition of meteoric calcite. However, there is little carbonate δ 18 O data from the South Wales-Mendip Shelf and it is heavily biased towards dolomites (Bhatt, 1976;Faulkner, 1989;Hird, 1985;Searl, 1988bSearl, , 1988c. Searl (1988b) and Wright et al. (1997) provide limited data from pedogenic deposits. Only one publication provides geochemical data on allochems and cements (Searl, 1989a).
The purpose of this paper is threefold: • To present previously unpublished vintage δ 13 C and δ 18 O data from pedogenic carbonates, allochems and cements from the Clydach Valley Subgroup in order to expand the available geochemical database for the South Wales area and provide an evaluation of the carbon and oxygen isotopic composition and origin of the fluids involved in Clydach Valley Subgroup diagenesis; • To place these data in a global stratigraphic and palaeoclimatic context; • To highlight areas where the application of modern analytical techniques and concepts could improve understanding of Clydach Valley Subgroup deposition and diagenesis.

| South Wales-Mendip shelf overview
During the Courceyan the South Wales-Mendip Shelf was located at approximately 12°S and was drifting on a northward trajectory (Figure 1). It remained in tropical latitudes at least through to the early Viséan (Scotese, 2014). It lay to the south-east of the Caledonides, bordering the remnant Rheic Ocean: a narrow, SW-NE trending seaway partly cut-off from the Palaeotethys Ocean. The Lower Carboniferous sediments of the South Wales-Mendip Shelf were deposited on what Ramsay (1991) described as the southward dipping hangingwall of a half graben associated with a major fault system located in the Bristol Channel (Figure 2), forming an approximately 60 km wide carbonate dominated wedge that extended about 100 km along strike. The wedge is over 1,000 m thick in the south, thinning to about 150 m in the most northerly outcrops due to a combination of lower subsidence rates and local and regional syn-depositional uplift of the shelf margin. On a local scale, stresses associated with the closure of the Rheic Ocean caused intermittent reactivation of 'Caledonoid' structures. Of these, the Vale of Neath Disturbance (VND), a complex zone of NE-SW trending faults and associated cross faults, is particularly pertinent to the current study (Barclay et al., 1988;George, 1954;Ramsay, 1991).
Final closure of the Rheic Ocean culminated in the Variscan Orogeny which was responsible for the structures controlling the present day Lower Carboniferous outcrop pattern: the South Wales Coalfield Syncline in the east and central parts and the Pembrokeshire Coalfield Syncline in the west, where substantial thrust related shortening has also occurred (Dunne, 1982). The resulting absence of intervening exposures between the northcrop and southcrop areas, other than along the eastcrop, plus the lack of modern, systematic stratigraphic work on the Courceyan succession has hampered correlation between northern and southern areas. This is compounded by the absence of critical parts of the succession and pervasive dolomitisation in eastcrop exposures that makes recognition of lithostratigraphic units challenging. The diachronous character of the main carbonate sandbodies is an added complexity. The nomenclature for the various outcrop areas used in the text is shown in green on Figure 2, together with the locations of the local studies that are referenced (red boxes).

| Clydach Valley Subgroup stratigraphy
The Clydach Valley Subgroup (CVSGp), the subject of this paper, and its westward equivalent, the Abercriban Oolite Subgroup (AOSGp), outcrops at the eastern end of the northcrop (Figure 2). Further west the AOSGp has been removed by erosion at the sub-Holkerian unconformity, although rocks of equivalent age are preserved in the northcrop extension in Pembrokeshire (George et al., 1976;Sullivan, 1965). Figure 3 summarises the lithostratigraphy of the northcrop and southcrop/Gower areas and should be used as a guide to the abbreviated lithostratigraphic unit names used in the text.
Conodonts suggest that the Pwll-y-Cwm Oolite Formation (PCOFm) belongs to the Siphonodella-Pseudopolygnathus multistriatus interzone, the Blaen Onnen Oolite Formation (BOOFm) spans the Siphonodella-P. multistriatus interzone/P. multistriatus F I G U R E 2 Tournaisian/Viséan outcrops of the South Wales-Mendip shelf showing outcrop area names (green text), 'Caledonoid' structures (purple lines and text), Variscan structures (blue text), the study area (solid red box) and the other study areas discussed in the text (dashed red boxes). Contains British Geological Survey materials © UKRI [2022]. Other studies: 1- Searl (1988aSearl ( , 1988bSearl ( , 1989b boundary and the Coed Ffyddlwn Formation (CFFm) falls entirely within the P. multistriatus Biozone (Barclay, 1989;Riley, 1993;Waters et al., 2013). Consequently, the BOOFm correlates, at least in part, with the more distal Brofiscan Oolite Formation (BrOFm) (Waters et al., 2013) (Figure 3). Until recently the Gilwern Oolite Formation (GOFm) was believed to be Chadian, a proximal equivalent of the Gully Oolite Formation (GuOFm). However, a recent study assigned conodonts from the basal GOFm to the Eotaphrus bultyncki Biozone (Courceyan) (Hogancamp & Wright, 2020), whereas the GuOFm falls within the Mestognathus praebeckmanni Biozone (Chadian), meaning there is an age difference between the two formations of between 4 and 6 Myr (Riley, 1993). Conodont biostratigraphy has been widely used in Europe and the USA as the basis for inter-regional correlation. However, there has been limited work tying UK conodont biozones with the conodont schemes used elsewhere, hampering correlation of the UK with other areas. Bábek et al. (2013) used European conodont and foraminifer biozones (MFZ) to date formations in mid-ramp locations on the South Wales Shelf and this has allowed limited correlation with other basins.
Additionally, table 4 of Barrick et al. (2022), which is a comparison between conodont biozones in deep and shallow water platforms of north-central Europe and biozones based on shallow water British conodont species, has further aided inter-regional correlation and confirmed the conclusions based on the Bábek et al. (2013) results. These data have been combined with the global marine carbonate δ 13 C and conodont apatite δ 18 O curves of Buggisch et al. (2008) and brachiopod δ 18 O curve of Grossman and Joachimski (2020) to produce Figure 4. The positions of the P. multistriatus and E. bultyncki biozones are based on graphical overlay of the Buggisch et al. (2008) and Barrick et al. (2022) charts, so should be considered approximate.
The UK P. multistriatus Biozone straddles the S. isosticha/G. typicus boundary, meaning it is coincident with the KOBE peak. Consequently, the PCOFm, Pantydarren Formation (PFm), BOOFm and, probably, the CFFm encompass the KOBE event. This is consistent with Bábek et al.'s (2013) assignment of the BrOFm, and by implication the BOOFm, to the MFZ4 Biozone, because in the Namur-Dinant Basin ( Figure 1) the KOBE event approximately brackets the upper MFZ2 to basal MFZ5 Biozones, with the peak spanning the S. isosticha/G. typicus boundary (Poty, 2016;Saltzman et al., 2004). The Polygnathus species described by Hogancamp and Wright (2020) from the basal GOFm post-date the KOBE event, which is consistent with the approximated position of the E. bultyncki Biozone shown on Figure 4. This is the first time that specific lithostratigraphic units of the South Wales-Mendip Shelf have been recognised as coeval with the KOBE event. Burchette et al. (1990) recognised at least three third-order depositional sequences in the Courceyan to Arundian succession, the upper parts of which are characterised by strongly progradational shoreline carbonate sandbodies and delimited by regionally significant disconformities, often with evidence of subaerial exposure ( Figure 3). Ramsay (1991) used changes in thickness and facies to highlight the importance of syn-depositional tectonism on depositional architecture and interpreted changes in relative sea level as mainly tectonically driven. The sequence stratigraphic understanding of the South Wales-Mendip Shelf has not evolved since the publication of these papers, either regionally or locally, and no detailed sequence stratigraphic interpretation is available for the CVSGp. The new age assignment of the GOFm and the recognition of possible eustatic sea-level changes associated with the KOBE event (Krammer & Matchen, 2008) mean the old models warrant revisiting. This is beyond the scope of the present study, but should not detract from the interpretations presented here, which are based on field and petrographic observations and geochemical analyses.

PETROGRAPHIC ANALYSIS
The study area comprises a NW-SE trending string of quarries ( Figure 5). The south-eastern limit coincides with the complete dolomitisation of the GOFm, while the north-western limit is defined by the Blaen Onneu Fault (BOF), a segment of the VND. Seventeen vertical sections were described and sampled, which are referred to by locality number in the text and are shown in Figure 5. The sheer nature of the quarry faces hampered access and it was sometimes necessary to piece together vertical profiles from a number of closely spaced sites.

| Petrography
Thin sections (2.4 × 4.7 cm) were prepared to a thickness of 60 μm, stained with Alizarin red-S and potassium ferricyanide (Dickson, 1966) and covered with acetate film.
These were examined under transmitted light to investigate depositional texture and diagenetic history. A subset of thin sections was selected for cathodoluminescence (CL) analysis. The preserved acetate peels and photomicrographs provided a record of the distribution of ferroan and non-ferroan carbonate phases.

| Cathodoluminescence
Thin sections were polished to a mirror finish, taking care not to reduce the thickness below 30 μm. The CL microscopy was carried out using a custom-built cold-cathode instrument with an accelerating potential of 17-26 kV, gun current of 450-600 mA, a focussed beam diameter of 1-10 mm and an air chamber pressure of 0.01-0.05 Torr. The CL chamber was mounted on a petrographic microscope with a camera attachment. Images were captured on black and white film. The CL images were compared with the recorded distribution of ferroan and non-ferroan carbonate phases and candidate samples for carbon and F I G U R E 4 Correlation of selected lithostratigraphic units and sequence tops recognised on the South Wales shelf to the marine δ 13 C curve (δ 13 C CARB , black), highlighting the position of the KOBE event. Also shown are the marine δ 18 O curves derived from conodonts (δ 18 O APATITE , red) and brachiopods (δ 18 O BRACH , blue). Increasing δ 18 O implies a colder climate and vice versa. Modified from Buggisch et al. (2008) and including data from Burchette et al. (1990), Bábek et al. (2013), Waters et al. (2014), Poty (2016), Grossman and Joachimski (2020) and Barrick et al. (2022). oxygen stable isotope analysis (described below) were identified.

| Depositional history
The quarry exposures of the CVSGp in the study area are oriented approximately perpendicular to the VND ( Figure 5) and capture a dip section through the proximal part of a carbonate ramp system (Wright, 1986). Figure 6 illustrates the disposition of the formations comprising the CVSGp and has been extended to the south-east of the study area as far as the Pen-Ffordd-Goch Borehole ( Figure 5) using data from Barclay (1989), in order to capture the most complete preserved section of the GOFm before it is affected by sub-Namurian erosion.
Traditionally, the CVSGp has been described in terms of three packages, the lower two (PCOFm/PFm and BOOFm/ CFFm) consisting of oolitic carbonate sandbodies capped by peritidal limestones and dolomites with evidence of repeated subaerial exposure and the upper one (GOFm) consisting of an oolitic carbonate sandbody capped by a karst that is overstepped by the Llanelly Formation (LFm) (Adams et al., 2004;Wright, 1981Wright, , 1988. In the absence of a modern sequence stratigraphic model for the CVSGp, this approach is continued here. The five broad depositional settings shown on Figure 6 are recognised based on field and petrographic observations, supplemented by published data (George, 1954(George, , 1956Barclay & Jones, 1978;Searl, 1988aSearl, , 1988bSearl, , 1989bBarclay, 1989).
The lowest formation of the CVSGp, the Sychnant Dolomite Formation (SDFm), is poorly exposed and not considered here. Its upper surface has been interpreted as a subaerial exposure surface in the Clydach Gorge area (Barclay, 1989) and similarly at Baltic Quarry F I G U R E 5 Distribution of Courceyan to Holkerian strata at the eastern end of the northcrop showing the study area, numbered localities referred to in the text, area described by Searl (1988aSearl ( , 1988bSearl ( , 1989b and key geological observations. Contains British Geological Survey materials © UKRI [2022]. VNF-Vale of Neath disturbance; BOF-Blaen Onneu fault. (Figure 5), where Ramsay (1991) reported an associated palaeosol that could also be recognised as far west as Pembrokeshire.

| Pwll-y-Cwm and Blaen Onnen oolite formations
The PCOFm and BOOFm consist of massive oolitic grainstones ( Figure 6) with varying proportions of peloidal and bioclastic grains, the latter being particularly common towards the bases. Brachiopods, bivalves and echinoderm plates comprise the majority of bioclasts. In the Clydach Gorge area freshly broken surfaces are medium grey, with a 'rotten eggs' odour. North-west of Locality 8 the colour transitions to pale grey and the pungent odour is lost. This is accompanied by a change in diagenetic character (see below).
The PCOFm is 6-7 m thick in the study area, thinning slightly north-westward across the BOF. Over the same interval the overlying BOOFm initially increases in thickness north-westward from 5 to 9 m, before thinning to 6 m immediately south-east of the BOF. Thickness rapidly increases to 14 m across the fault, indicating syn-depositional movement with downthrow to the north-west. This is consistent with rapid thickness changes across faults oriented perpendicular to the VND reported by Searl (1988a). She also recorded multiple intra-formational calcrete horizons at Cwar yr Hende ( Figure 5), where intense cross-faulting developed in order to release stress at an offset in the VND, meaning it was the focus for complex syn-depositional movement. The lack of intra-formational calcrete in the study area may in part be explained by the fact that the study outcrops run approximately parallel to cross-faults, plus they are located in an area that experienced less intense deformation.
The tops of the PCOFm and BOOFm are mainly exposed in quarry faces where they occur as planar surfaces with no evidence of karst formation. Evidence of subaerial exposure is based on the occurrence of calcrete at the base of the overlying formations and is limited to northwest of Locality 10.
The south-eastern portion of the section is extensively dolomitised, but the oolitic formations can be traced as massive units flanked by the thinly bedded dolomites of the PFm and CFFm (Barclay, 1989;George, 1956). Between the Clydach Gorge and Pen-Ffordd-Goch Borehole the only grains Barclay (1989) recorded were crinoid debris. It is unclear if the lack of ooids represents a facies change or has a diagenetic explanation. Hird (1985) noted that in the BrOFm echinoderm plates were relatively resistant to dolomitisation compared to ooids, meaning crinoid debris was possibly preferentially preserved.

Ffyddlwn formations
In the south-east of the study area the PFm and CFFm form distally thickening wedges of thinly bedded, finely crystalline dolomite, with occasional microbial lamination, interbedded with thin claystones, all interpreted as peritidal deposits ( Figure 6). The CFFm also contains occasional coarser beds containing brachiopod, bivalve and crinoid debris and the topmost few metres remain undolomitised, consisting of claystones (up to 60 cm), micrites, peloidal facies, occasional oncolitic horizons and common detrital quartz. Barclay (1989) recorded mudcracks near the top and interpreted these deposits as lagoonal and/or peritidal. There is no evidence of prolonged subaerial exposure.
The formations thin and change in facies northwestward between Localities 8 and 11. North-west from Locality 10 an increasing proportion of the PFm consists of a wedge of ooid-dominated grainstones which can be traced at least as far as Baltic Quarry ( Figure 5), confirming that facies belts ran parallel to the VND. Above and below the wedge stratiform dolomites alternate with micritic and columnar calcite concretions embedded in claystones that locally include thin coals and evaporite pseudomorphs. Searl (1988bSearl ( , 1989b proposed that the dolomites formed in a brackish, paludal environment, while the concretions were of pedogenic origin. Relict bioclastic grains in the dolomites indicate replacement of marinederived material, possibly washed in by storms, suggesting deposition along a low-lying shoreline. Between Localities 11 and 13 (limit of accessible exposure) the base of the CFFm is marked by a 4-5 cm thick calcrete crust that coats the top of the BOOFm. This is overlain by nodular calcrete followed by fine grained peloidal limestones, micrites, intermittent coarse quartz sand beds (grains up to 2 cm diameter) and thin micrite crusts, similar to those at the base. The micrite crusts are often associated with columnar calcite similar to that found in the PFm. Beds are typically separated by 1-2 cm thick claystone layers. The facies are indicative of alternating subaerial exposure and peritidal deposition.

| Gilwern Oolite Formation
The base of the GOFm is a distinctive 2-3 m thick shallow marine unit (Figure 6), the 'Coral Bed' of George (1954), consisting of coarse grained bioclastic and peloidal grainstones containing abundant crinoid debris, brachiopods, disarticulated bivalves, gastropods and occasional solitary corals. Insoluble residues reveal a diverse conodont fauna (Hogancamp & Wright, 2020) and fish remains. Possible hardground surfaces occur at Localities 1, 4 and 9. 'Coral Bed' remnants were described by George (1954) from Blaen Onneu Quarry (now lost due to quarrying activity), indicating that it extended across the VND.
The majority of the GOFm is a pale grey, massive oolitic grainstone ( Figure 6). Accessory grains include brachiopods, bivalves, echinoderm plates, peloids and intraclasts in varying amounts. Sedimentary structures are poorly preserved, but an intricate syn-depositional marine lithified surface occurs at Localities 7-9 (Figures 7 and 8A), capping a unit displaying upward increases in ooid grain-size, sorting, purity and fabric preservation. Pure oolitic grainstone with good fabric preservation, similar to that seen immediately below the intra-GOFm surface described above, comprise the top 1-2 m of the GOFm in the Clydach Gorge area. The GOFm remains undolomitised further to the south-east than the underlying oolitic formations.
Prior to deposition of the overlying LFm (Arundian), the CVSGp/AOSGp was uplifted, tilted towards the southeast and partially eroded, with complete to near complete removal of the GOFm north-west of Locality 13 ( Figure 6). Thickness progressively increases south-eastward, reaching about 21 m at the Pen-Ffordd-Goch borehole and Wytchwood Quarries (Barclay, 1989) (Figure 5), but nowhere is the original thickness preserved. Between the Clydach Gorge and Blaen Onneu Quarry (10 km) there is close to 18 m relief on the unconformity, based on LFm overstep (Adams et al., 2004;Wright, 1981Wright, , 1988. In local terms, this has been linked to reactivation of the VND, but should be viewed in the context of more widespread uplift of the basin margin, the magnitude of which is uncertain and the exact timing poorly resolved, although it probably occurred no later than the end of Sequence 1 deposition ( Figure 3). A contemporaneous unconformity occurs at the top of the equivalent succession (Pendine Oolite) on the northcrop extension in Pembrokeshire at Pendine (George et al., 1976;Spalton, 1982; Figure 2).

| Karst
A palaeokarst is preserved on the erosional surface at the top of the CVSGp, which is recognised as far south as the Pen-Ffordd-Goch Borehole (Barclay, 1989;Barclay & Jones, 1978) (Figure 5). Its presence immediately west of the VND is confirmed at Blaen Onneu Quarry, where it is developed in the BOOFm, but further west of the VND it is either not exposure or eroded. A palaeokarst occurs in the same stratigraphic position in Pembrokeshire (Spalton, 1982;Sullivan, 1965), supporting the concept of regional uplift and exposure. The karst appears as a rubbly horizon up to 6 m thick consisting of interconnected, centimetre-sized pipes and fissures filled with green clay, with rare, larger dissolution pipes. For a detailed description the reader is referred to Wright (1980Wright ( , 1982aWright ( , 1982bWright ( , 1988Wright ( , 1990, Spalton (1982), Dickson and Wright (1993) and Adams et al. (2004). Wright (1980Wright ( , 1982aWright ( , 1988 likened the karst to kavernosen karren, a tropical karst type developed beneath a soil cover in areas that receive high (about 1,500 mm) annual rainfall (Ireland, 1979), and suggested that it could have formed in as little as a few thousand years. Evidence of the original soil and vegetation cover associated with the karst is missing. This is possibly due to erosion of the karst prior to deposition of the overlying LFm and is consistent with local topography of up to 5 m on the unconformity and the presence of karst-derived cobbles in the Clydach Halt Member (CHMbr), the basal unit of the LFm (Wright, 1982b), indicating reworking of CVSGp derived material. Descriptions of modern kavernosen karren are vague and give no indication of the thickness of recent examples (Gams, 1973;Jennings, 1971Jennings, , 1987, meaning there is no basis on which to estimate the amount of lost section.

| Diagenetic history
Four diagenetic regimes have been identified in the study area ( Figure 7): • Diagenetic Regime 1 includes the majority of the oolitic formations. It is characterised by pale grey limestone dominated by non-ferroan calcite cement. Stratigraphic intervals displaying this diagenetic trait are predominantly located beneath subaerial exposure surfaces. Evidence of grain compaction and early diagenetic pyrite is minor, indicating pre-burial cementation and dominantly oxic early pore waters. • Diagenetic Regime 2 occurs where the BOOFm and PCOFm grade distally into medium grey limestones dominated by ferroan calcite cement. Early diagenetic pyrite indicates anoxic syn-depositional pore waters, while common grain compaction indicates limited early cementation. • Diagenetic Regime 3 includes the proximal parts of the PFm and CFFm. The association of pedogenic and paludal deposits (Searl, 1988b(Searl, , 1989b suggests a position close to sea level at the transition from marine to terrestrial environments. The alternation of pedogenic, paludal and marine sediments indicates a complex pore water history. • Diagenetic Regime 4 occurs in the distal part of the CVSGp, where it is extensively dolomitised. This represents the south-eastern limit of the study area. Hird (1985) considered that dolomitisation of peritidal deposits from the southern end of the eastcrop, younger versions of the distal parts of the PFm and CFFm, was penecontemporaneous with deposition. He concluded that pervasive stratal dolomites, including replacement of oolitic formations such as the BrOFm, had a more complex dolomitisation history, starting in the shallow subsurface and continuing during burial with further dolomitisation and/or recrystallisation events. Some dolomitisation is associated with faults.
The focus of this paper is the calcite cements in Diagenetic Regimes 1 and 2 and it is these rocks that are considered in more detail below. The majority of data from Diagenetic Regime 3 is derived from its pedogenic deposits, although some of the cement zones recognised in Diagenetic Regime 1 also occur there. No consideration is given here to Diagenetic Area 4.

| Diagenetic Regime 1
Five calcite cement zones have been recognised in Diagenetic Regime 1 as revealed by staining and CL. So as to keep the nomenclature consistent with that used in Raven (1983) the sequence described below starts with Zone 2, Zone 1 cements being restricted to Diagenetic Regime 2. 4.6.2 | Zone 2 Zone 2 consists of inclusion-rich syntaxial overgrowths on echinoid plates and occurs in all CVSGp formations. It is often delimited by well-defined crystal terminations. Instances of internal sediment immediately overlying Zone 2 occur in most formations. Cathodoluminescence reveals a patchy distribution of irregular, moderately and brightly luminescing areas superimposed on a non-luminescent background ( Figure 8B). The CL character is similar to that of rare fibrous (marine) cements ( Figure 8C).

| Zone 3
Inclusion-poor, faintly ferroan calcite with a dull luminescence commonly overlies Zone 2 cements ( Figure 8B,D) and these Zone 3 cements may, in some cases, exhibit some dull-brighter zonation ( Figure 8E). However, ferroan staining is rarely observed in the GOFm above the 'Coral Bed; although a zone with dull CL is present predating Zone 4. This could be a consequence of the insensitivity of the staining response and/or the small crystal size.

| Zone 4
Inclusion-poor, non-ferroan calcite is the dominant cement phase in all occurrences of Diagenetic Regime 1. Cathodoluminescence reveals an alternation of nonluminescent and brightly luminescent calcite, the latter as micron-wide zones ('bright CL hairlines') ( Figure 8D-F). Diagenetic Regime 1 has been divided into four subregimes (A-D) based on the complexity of the CL zonation pattern ( Figure 7). Diagenetic Regimes 1A and 1B are restricted to the PCOFm and BOOFm, while Regimes 1C and 1D occur in the GOFm.
In Diagenetic Regime 1A the abundance of 'bright CL hairlines' is highly variable within a single crystal, defining subzones where they are alternately sparse or abundant. Fretted crystal faces and crystal faces on which calcite has re-nucleated suggest episodes of dissolution and non-precipitation respectively ( Figure 8E). These observations indicate that Zone 4 encapsulates a number of discrete precipitation events. Diagenetic Regime 1A is the lateral equivalent of the AOSGp where Searl (1988a) also noted complex, laterally variable CL sub-zonation. This CL complexity was attributed to multiple, vertically and laterally restricted precipitation events, evidenced by laterally restricted intra-BOOFm calcrete horizons, multiple influxes of internal sediment on top of meteoric cements and locally restricted geochemical variability. The majority of evidence for this came from Cwar yr Hendre, which was previously highlighted as particularly susceptible to differential subsidence. It cannot be assumed that conclusions based on this locality are applicable elsewhere, especially when conspicuous intra-formational subaerial exposure surfaces are absent, which is the case for the study area described here. In the study area the only formation in which internal sediment has been observed overlying non-ferroan blocky calcite cement is the PFm. An additional factor not considered by Searl (1988a) that could contribute to CL complexity is changes in fluid flow pathways as diagenesis progressively modifies the permeability network. This could have impacted the location and rate at which cements precipitated and could explain why adjacent pores filled at different rates and why the relative proportion of successive zones can differ on a centimetre scale (Raven, 1983). Growth rate will impact the visual appearance of CL zonation, as illustrated by echinoderm syntaxial overgrowths: 'bright CL hairlines' are narrower and more tightly clustered on slower growing faces and less so on faster growing ones.
The CL zonation patterns are less complex in Diagenetic Regime 1B, which is positioned at the distal limit of subaerial exposure ( Figure 7). Typically, a dominantly nonluminescent subzone overlies Zone 3, followed by a subzone with common 'bright CL hairlines' ( Figure 8D). Zone 4 cements also become thinner and less abundant as Diagenetic Regime 1B transitions into Diagenetic Regime 2.
In Diagenetic Regimes 1C and 1D (GOFm) much of the variability in cement CL characteristics is possibly related to textural differences between the basal bioclastrich 'Coral Bed', the overlying impure oolitic grainstones and the pure oolitic grainstones located beneath the intra-GOFm cemented surface and in the top 2 m of the GOFm in the Clydach Gorge area. In the 'Coral Bed', relatively large Zone 4 crystals within moulds display obtuse rhombohedral terminations and rare 'bright CL hairlines' in their outer parts ( Figure 9A). In the impure oolitic grainstone Zone 4 cement grew as smaller lozenge-shaped crystals of dominantly non-luminescent calcite, also with obtuse rhombohedral terminations ( Figure 9B), and the majority of pores remained unfilled after Zone 4 precipitation. In pure oolitic grainstones Zone 4 cement is also dominantly non-luminescent, but equidimensional crystals fill the majority of the pore space leaving little space for Zone 5 and 6 cements.
Diagenetic Regime 1D abuts the top of the CVSGp and the Zone 4 cements are truncated at the unconformity ( Figure 9C; figure 15.8 of Wright, 1988). Calcrete belonging to the CHMbr locally penetrates along cement crystal boundaries ( Figure 9D). This indicates that Zone 4 cement in Diagenetic Regime 1D pre-dates the erosional surface at the top of the CVSGp. 4.6.5 | Zone 5 In Diagenetic Regimes 1A and B (BOOFm and PCOFm) a dull to moderately bright luminescing, faintly ferroan zone commonly occurs overlying Zone 4 ( Figure 8D,F). However, in Diagenetic Regime 1C (GOFm) Zone 5 occurs as alternations of faintly ferroan, dull luminescing calcite and Zone 4-like non-ferroan calcite ( Figure 9A), the number of alternations increasing in a distal direction. This is best observed in crystals growing into mollusc moulds in the 'Coral Bed' between Localities 1 and 9. A similar pattern occurs in cements in the overlying impure oolitic grainstones, although only recognisable using CL ( Figure 9B) where, like Zone 3, the lack of obvious ferroan stain may reflect the small crystal size and relatively low iron content of the calcite. The alternation of ferroan with Zone 4-like non-ferroan calcite suggests a closer genetic relationship with Zone 4 than Zone 6, at least in the GOFm. In Diagenetic Regime 1D (top GOFm), Zone 5 consists of a single moderately brightly luminescing layer of non-ferroan calcite.

| Zone 6
This zone, found in all CVSGp formations and diagenetic regimes, consists of strongly ferroan calcite that frequently displays sector zonation (Raven & Dickson, 1989). It has a dull luminescence that often has a patchy appearance reflecting the difference in iron content between crystal sectors. It fills residual intergranular and biomouldic pores, late fractures and oomouldic and channel porosity created by a post-Zone 5 dissolution event ( Figure 8E,F). This late dissolution event locally caused the disintegration of the rock fabric as manifested by (1) the collapse of oomouldic fabrics, (2) the accumulation of shards of cement with scalloped margins on the floor of dissolution channels, the scallops recording the former presence of ooids, and (3) the occurrence of 'internal intraclasts' (volumes of cemented grains that have become detached from the surrounding rock and fallen into dissolution voids) (plates 3/1 to 3/3 and figure 3/1 of Raven, 1983). Searl (1988a) described a similar ooid dissolution event in the AOSGp, followed by the precipitation of strongly ferroan calcite cement. The similarity in appearance and occurrence of Zone 6 in all diagenetic regimes suggests that it records a discrete late diagenetic event.

| Diagenetic Regime 2
The boundary between Diagenetic Regimes 1B and 2 is transitional ( Figure 7). Some of the cement zones observed F I G U R E 9 Transmitted light (TL) and cathodoluminescence (CL) photomicrographs. Images A and B from Diagenetic Regime 1C; images C and D from Diagenetic Regime 1D; images E and F from Diagenetic Regime 2. (A) Typical cement zonation found in crystals filling bivalve moulds in the GOFm 'coral bed'. The boundary between Zones 4 and 5 is marked by an obtuse termination. Zone 5 consists of alternating mildly ferroan calcite with dull CL and non-luminescent non-ferroan calcite, each with its own crystallographic form. CL, GOFm locality 5, sample 18325. (B) Typical Zone 4 cement above the GOFm 'coral bed' consists of isolated lozenge-shaped crystal growing into intergranular pores between ooids (O). Zone 5 has the same form as in the underlying 'coral bed'. Residual porosity is filled by Zone 6. CL, GOFm locality 4, sample 18306. (C) Zone 4 cement truncated at the top CVSGp unconformity (arrow) and overlain by calcrete (CC) consisting of brightly luminescing rhombic crystals. A rim of Zone 3 surrounds the ooids (O). CL, GOFm locality 4, sample 17843. (D) Oolitic grainstone from immediately below the top CVSGp unconformity cemented by Zone 4 calcite. Two generations of calcrete (CC1 and CC2) infiltrate the rock. A small area is filled by Zone 6 cement. CL, GOFm locality 9, sample 17865. (E) Syntaxial overgrowth on echinoderm plate consisting of Zones 1 and 2b cement. Zone 4 cement is absent and remaining porosity is filled by Zone 6. CL, PCOFm locality 5, sample 18313. (F) Early pyrite (Py) replacement of echinoderm plates in association with much grain compaction. TL, PCOFm locality 7, sample 17830. All scale bars represent 100 μm.
in Diagenetic Regime 2 are unique to that regime, but others are shared with Diagenetic Regime 1. Where the latter is the case the same zone number is used in both regimes. Zones 1-4 are best developed as syntaxial overgrowths on echinoderms. Elsewhere they form only a thin rim around ooids and other grains. This left substantial residual porosity in Regime 2 that was filled by Zone 6 cement in rocks that resisted compaction. 4.7.1 | Zone 1 Zone 1 consists of inclusion-rich, strongly ferroan calcite growing as syntaxial overgrowths on echinoderm plates. It is generally non-luminescent, but often contains irregular patches of variably luminescing (dull to bright) calcite suggesting it has undergone alteration ( Figure 9E). In areas that have undergone compaction these overgrowths are indented into adjacent grains. These syntaxial overgrowths can be correlated laterally with needle-like calcite coating ooids and brachiopods, suggesting a marine origin for both cement fabrics. The iron-rich character is indicative of reducing conditions either during precipitation or alteration and compliments evidence of early anoxia indicated by the presence early diagenetic pyrite ( Figure 9F).

| Zone 2b
This zone consists of non-ferroan, inclusion-poor calcite with dull luminescence ( Figure 9E). It has been denoted Zone 2b to distinguish it from Diagenetic Regime 1 inclusion-rich Zone 2 cement. Its lack of inclusions and CL character suggests it has more affinity to Zone 3 than to Zone 2 cements in Diagenetic Regime 1.

| Zone 3
The subsequent Zone 3 cements consist of mildly ferroan calcite that exhibits dull luminescence. This cement was rarely identified and then only on the proximal side of Diagenetic Regime 2. That in turn tentatively suggests that it is equivalent to Zone 3 cement of the adjacent Diagenetic Regime 1B.

| Zone 4
Zone 4 cement, similar to that encountered in Diagenetic Regime 1, is largely absent from Diagenetic Regime 2, but occasionally occurs as a thin rim on top of the pre-existing cements (Zones 1-3). It is best observed on echinoderm syntaxial overgrowths where it has a non-luminescent inner subzone and an outer subzone of alternating brightly luminescing and non-luminescing calcite, similar to that found in Diagenetic Regime 1B. It is absent from areas that have undergone grain compaction. 4.7.5 | Zone 6 Zone 6 cement in Diagenetic Regime 2 has the same characteristics as Zone 6 cement in Diagenetic Regime 1: it is strongly ferroan, dully luminescent and often shows sector zoning. However, it is more abundant in Regime 2 as it is widely present filling intergranular pore space ( Figure 9E). It post-dates grain compaction.

| Sampling
Serial (up to 20) thin sections were prepared from selected samples by mounting rock slices on glass slides (7.5 × 5 cm) using Lakeside 70, grinding these to approximately 100 μm and staining them with Alizarin red-S and potassium ferricyanide (Dickson, 1966). Material was scraped from these under a transmitted light microscope using a steel scalpel blade. Dickson and Coleman (1980) showed that isotope results were unaffected by the presence of small amounts of Lakeside 70 and stain in sample powders. This extraction technique was used for the majority of cement samples and some allochem samples. Pedogenic calcite and some cement and allochem samples were scraped directly from polished rock slabs, with the slabs incrementally ground down as material was removed. Where possible at least 5 mg of material was collected. Bulk rock samples were not analysed.

| Cements (n = 69)
The philosophy dictating sample selection was to have a standardised approach that allowed the isotopic composition of sequential cement zones to be confidently tracked through time and space. Consequently, priority was given to samples in which as near complete as possible sequence of cement zones was present and crystals were sufficiently large to allow separation of zones with minimal cross-contamination. Suitable samples were limited to shell-rich intervals where large (for the CVSGp) crystals grew into biomouldic, intragranular and/or shelter pores, the 'Coral Bed' being the best source of material. Even in the most favourable cases individual cement zone samples comprise material extracted from multiple crystals and pores. It was not possible to find samples from the BOOFm or PCOFm with a full series of zones that could be sampled, either because certain zones were too thin or were not recognised in stained thin section. The majority of pores with sufficiently large crystals were filled during Zone 4 precipitation, so pores meeting the sequential sampling requirements were unusual and it cannot be guaranteed they record a complete or unbiased history of Zone 4 precipitation. Given that the material was at the limit of what was possible to separate manually, the possibility of cross-contamination between zones has to be borne in mind when interpreting the results.
In the upper part of the GOFm shell-rich intervals were absent, certain zones could only be recognised with CL and inter-ooid pores dominated. Here it was only possible to separate non-ferroan (Zone 4 possibly contaminated from Zones 3 and 5) and later ferroan (Zone 6) calcite. Additionally, ferroan calcite cements from vugs within pedogenic concretions and fracture-fills, assumed to be Zone 6 cements due to their strong ferroan staining, were analysed.

| Allochems (n = 35)
Sampling of allochems was focussed towards complimenting cement zone analysis. Sample selection was not tailored to address isotopic variations beneath exposure surfaces and, consequently, the resulting dataset is not optimum for this purpose.
• Ooids (n = 11)/Ferroan ooids (n = 2): Only ooids with little or no visible nucleus were sampled to minimise contamination from biogenic material forming the nuclei. In some samples staining revealed the ooids to be slightly ferroan, referred to as 'ferroan ooids', suggesting alteration. The ooids from the CVSGp are texturally similar to those in the BrOFm and GuOFm, which Hird and Tucker (1988) and Searl (1989a) concluded originated as high Mg-calcite. • Brachiopod (n = 9)/Brachiopods (n = 12): Two types of sample were collected. The preferred material was from single brachiopod valves that could provide sufficient material for analysis in their own right. Such valves were typically more than 1 mm thick and constitute 'brachiopod' samples. This was rarely possible, in which case material was collected from a number of individual valves. These constitute 'brachiopods' samples. All of the shells sampled had a fibrous structure. Thin section size and the mounting medium used meant it was not possible to undertake CL screening of brachiopod material to check for alteration prior to sampling (Lakeside 70 melts when heated). However, CL examination of thin sections prepared from the same rock samples as used to extract brachiopod material showed that shells which appeared unaltered in transmitted light often displayed a fibrous structure outlined by brightly luminescing calcite in CL, indicating that they had experienced alteration. Dickson and Kenter (2014) describe a similar CL alteration fabric in brachiopods. Brachiopods that showed replacement by ferroan calcite (i.e. were visibly altered) were avoided. • Crinoid ossicles (n = 1): Ossicles several millimetres in size were sampled from rock slabs. No separation of skeleton from stereome fill was possible.

| Pedogenic calcite (n = 54)
Samples were scraped from rock slabs. Five sample types are recognised: • Calcrete coating the CVSGp unconformity (n = 5): This consists of micritic calcite made up of micron-size rhombic crystals with CL similar to that described by Wright and Peeters (1989) (Figure 9C,D). • Calcrete crusts from the CFFm (n = 3): These have a similar CL fabric to the calcrete coating the CVSGp unconformity but have recrystallised to a tightly interlocking network of much larger crystals. • Calcrete concretions from the PFm and CFFm (n = 16): These have a similar CL fabric to the calcrete coating the CVSGp unconformity, but have recrystallised in a similar fashion to the CFFm calcrete crusts. This is referred to as 'recrystallised calcrete'. Samples were collected from the 'recrystallised calcrete' (n = 7) and associated vug and fracture filling calcite (n = 9). • Columnar calcite from the PFm and CFFm (n = 20): This occurs as sheets and spherulitic masses, interpreted as pedogenic by Searl (1989b), that often alternate with recrystallised calcrete. Samples were collected from the columnar calcite crystals (n = 16) and associated vug and fracture filling calcite (n = 4). • Botryoidal calcite (n = 10): This consists of hemispherical radiating fibrous calcite concretions, with central septarian cement-filled fractures, arranged in 'bunch of grapes'-like aggregates (Plate 5/4 in Raven, 1983). A pedogenic origin is presumed due to association with calcrete concretions and columnar calcite. Samples were collected from the fibrous calcite (n = 4) and septarian calcite cement fill (n = 6).

| Analytical procedures
Carbon and oxygen stable isotope analyses were undertaken over the period 1978 to 1980 at the Institute of Geological Sciences, Greys Inn Road, London, under the supervision of Dr. Max Coleman. No pre-treatment for removal of organic material was undertaken since organic carbon content was considered very low. Samples were reacted with 100% phosphoric acid, prepared by adding P 2 O 5 to AnalaR orthophosphoric acid, in evacuated glass reaction vessels which were kept in a water bath at 25°C until reaction was complete (less than 24 h for calcite). Laboratory standards calibrated to Vienna-Peedee belemnite standard (PDB) were prepared in a similar fashion and included in every batch of samples analysed. Following extraction the evolved CO 2 was transferred to a mass spectrometer for analysis. Raw data were automatically corrected for mass and instrumental effects (Craig, 1957;Deines, 1970). Analytical uncertainty for carbon and oxygen was ±0.02‰ and ±0.03‰ respectively. All isotope data are quoted against the PDB standard.
In order to evaluate uncertainty regarding imperfect isolation of cement and allochem components 13 duplicate samples were extracted (Table 1). Eight duplicates represent replicate extracts from the same sample, that is, centimetres apart, and five duplicates represent extracts from two different samples decimetres apart. The mean absolute difference in replicate samples was 0.5‰ (range 0.13-1.56‰) for carbon and 0.6‰ (range 0.01-1.12‰) for oxygen. Absolute differences were slightly larger in the two samples that were decimetres apart, compared to replicates from the same sample, although relative trends were the same.

| RESULTS
The results of δ 13 C and δ 18 O analyses are presented here in graphical form. Means, ranges and standard deviations for the pedogenic carbonate, allochem and cement datasets are presented in Tables 2 through 4 respectively. Standard deviations were only calculated for data sets with at least five samples. Only means and ranges are quoted below. A full set of analytical results is available online as Supplemental Material. Figure 10 presents δ 13 C and δ 18 O results from pedogenic samples displayed by formation and type and compared to Zone 4 cements (pink polygon), the dominant meteoric cement zone, and the estimated isotopic composition of contemporary seawater (see below). Recrystallised/ columnar calcite concretions from the PFm and CFFm provide the largest dataset (n = 11 and 12 respectively) and show a narrow range of δ 18 O values (−8.5 to −7.4‰, mean − 7.8‰) that corresponds to the lower end of the Zone 4 cement range. Recrystallised calcite concretions (n = 7) also display a narrow range of δ 13 C values (−5.8 to −4.2‰, mean −5.0‰), unlike columnar calcites (n = 16) which are more variable, but have a similar mean value (−6.5 to −2.4‰, mean −4.7‰). Data from PFm botryoidal fans (n = 4) plot in a tight cluster with slightly lower δ 18 O (−8.8 to −7.8‰, mean − 8.3‰) than recrystallised calcite concretions and columnar calcite. The δ 13 C values (−4.0 to −3.2‰, mean − 3.6‰) are higher than recrystallised calcite concretions, but towards the upper end of the columnar calcite range. Calcite vug fills in the centre of botryoids have values indistinguishable from the fibrous margins (Table 2). Calcrete crusts from the CFFm (n = 3) also occur in a tight cluster with similar δ 13 C values (−4.1 to −3.6‰, mean − 3.9‰) as the botryoidal fans, but higher δ 18 O values (−7.7 to −6.9‰, mean − 7.3‰). The PFm botryoidal fans and CFFm calcrete crusts plot outside of the Zone 4 cement polygon by virtue of their δ 13 C values, but their δ 18 O values overlap Zone 4 cement and are consistent with the other PFm and CFFm pedogenic carbonates.

| Pedogenic calcite components
The CHMbr calcrete (n = 5) that coats the top of the CVSGp is distinct from the older pedogenic deposits, with greater δ 18 O values (−6.7 to −5.9‰, mean − 6.3‰), although δ 13 C values (−5.4 to −4.6‰, mean −5.0‰) are similar to PFm and CFFm recrystallised calcrete concretions. The stable isotope composition of these calcretes also overlaps the high δ 18 O side of the Zone 4 cement field, with δ 13 C values falling in the middle of the Zone 4 range. Previously published data from CHMbr calcrete (Wright et al., 1997) has a similar δ 13 C range to this study, but a wider range in δ 18 O values (Figure 10), possibly reflecting a greater vertical and lateral distribution of samples. Figure 11 presents δ 13 C and δ 18 O results from allochem samples, displayed by formation and grain type. About half (n = 18) of the data overlap in δ 18 O-δ 13 C space with the isotopic composition of pedogenic calcites. These allochems have δ 18 O values of −7.5 to −6.1‰ and δ 13 C values of −4.8 to −2.6‰ and represent all five allochem sample types and come from all three oolitic formations. Three PCOFm data points have unusually high δ 18 O values (−9.7 to −8.5‰) suggesting a different recrystallisation history (these samples are identified by '?' on Figure 13) and are excluded from the following analysis. The remaining data points (n = 14, PCOFm, PFm, BOOFm and CFFm ooids and brachiopods) have more positive δ 13 C values than recorded in any pedogenic sample (−1.7 to +4.5‰), with associated δ 18 O values of −7.2 to −4.6‰. The PCOFm data lie along a linear mixing (regression) line (solid green line of Figure 11) to which the smaller BOOFm and PFm datasets appear to conform. The PCOFm allochems are consistently less isotopically depleted than allochems from overlying units, with the GOFm allochems being the most consistently isotopically depleted. The BOOFm allochems exhibit isotopic values that tend to range between those two extremes, suggesting an upward trend in allochem isotopic alteration. The small number of PFm and CFFm allochems analysed broadly conform to this trend. There is also a tendency for PCOFm and BOOFm allochems to be more isotopically altered towards the north-west (increasing locality number) where they display similar δ 18 O to GOFm allochems. Within the PCOFm there is a tendency for ooids to be more altered than brachiopods, but in the more isotopically depleted GOFm allochems no distinction can be drawn between allochem types.

| Cement zones
A δ 18 O versus δ 13 C plot ( Figure 12A) displays the isotopic results of cement analyses and shows that individual zones form distinct clusters. The area of isotopic space associated with Zones 3-5 is enlarged in Figure 12B.

OF CaCO 3 PRECIPTIATED IN EQUILIBRIUM WITH SEA WATER
The carbon and oxygen isotopic composition of sea water has fluctuated over time (Cramer & Jarvis, 2020;Grossman & Joachimski, 2020). Consequently, in any study of the diagenetic evolution of marine limestones it is vital to establish the original isotopic composition of depositional CaCO 3 . Despite temporal variations of sea water δ 13 C and δ 18 O values being relatively well constrained in the Carboniferous (Figure 4), the amplitude of fluctuations can vary significantly between different locations dependent on a variety of oceanographic, geographic, climatic and biologic factors (Katz et al., 2007;Patterson & Walter, 1994). Consequently, global average curves cannot reliably provide the δ 13 C and δ 18 O values of carbonate precipitated in equilibrium with sea water for a specific location. Other factors affecting the estimation of the original isotopic composition of allochems are (1) the fact that biostratigraphic zones can span a range of sea water isotope values, especially at times of rapidly changing isotopic composition and (2) difficulties correlating British conodont biozones with European and USA schemes, as discussed above.

| Carbon
The KOBE event (Figure 4) is the most prominent δ 13 C excursion in the Carboniferous (Saltzman, 2002) with δ 13 C values peaking at up to +7‰, although there is significant regional and local variation. After the peak, δ 13 C T A B L E 2 Population statistics for δ 13 C and δ 18 O compositions of pedogenic carbonates. Data from vugs filled with one 6-like cement included on values declined to between +1 and +2.3‰ and continued at this level into the early Viséan (figure 11.8 of Cramer & Jarvis, 2020). The nearest reference section to the UK is from the Namur-Dinant Basin (Figure 1), where the KOBE peak reached a maximum of about +5‰ (Saltzman et al., 2004). The PCOFm allochem data define a mixing line anchored at the marine end by a sample with a δ 13 C value of +4.5‰ (Figure 11). Data from the PFm and BOOFm conform to this line. These three formations were deposited during the KOBE event (Figure 4), suggesting that KOBE marine δ 13 C values for the South Wales-Mendip Shelf were similar to those of the Namur-Dinant Basin. On this basis, the minimum δ 13 C value for marine carbonate for the lower part of the CVSGp is set at +4.5‰. Lower δ 13 C values are expected during post-KOBE GOFm deposition (Figure 4), the estimation of which must rely on global data because only heavily altered allochems with isotopic values similar to pedogenic calcite were analysed from the GOFm. A value of +2‰ is a reasonable estimate (Buggisch et al., 2008;Cramer & Jarvis, 2020).

| Oxygen
The δ 18 O values of marine carbonate increased from −6‰ at the base Carboniferous to −1.5‰ at the base Viséan ( Figure 4 and figure 10.8 of Grossman & Joachimski, 2020), although these values represents global averages and probably do not capture small scale fluctuations. At the KOBE peak, Grossman and Joachimski (2020) suggested a δ 18 O value of −2.5‰, although there is uncertainty associated with this. The two PCOFm brachiopods with the most positive δ 13 C values have δ 18 O of −4.6 and −5.1‰ (Figure 11), which is lower than the value estimate above. This could reflect the PCOFm slightly pre-dating the KOBE maximum and/or some degree of meteoric alteration. For the lower part of the CVSGp it is assumed the original carbonate δ 18 O value was no lower than that of the least altered allochem (−4.6‰). Since no data are available to constrain a value for the later GOFm, a value of −2‰ has been estimated based on Grossman and Joachimski's (2020) curve and the position of the GOFm on Figure 4. This is greater than the value expected for T A B L E 3 Population statistics for δ 13 C and δ 18 O compositions of allochems. Standard deviations (±1σ) are only provided for datasets with at least five members. the KOBE peak but lower than the base Viséan value of −1.5‰. Using the derivations above, the δ 13 C and δ 18 O values of GOFm marine carbonate were +2‰ and −2‰ respectively. A hypothetical mixing line between those values and meteoric calcite ('Coral Bed' Zone 4 cements, see below) can be constructed for the GOFm (red dotted line on Figure 11). The fact that this line helps to explain the position of CFFm (Locality 6) and GOFm (Locality 9) allochem data that plot away from the main GOFm allochem data cloud gives some credence to the estimated composition of GOFm marine carbonate.

COMPOSITION OF CaCO 3 PRECIPTIATED IN EQUILIBRIUM WITH CVSGP METEORIC WATERS
The presence of exposure surfaces on top of the PCOFm, BOOFm and GOFm with extensive pre-compaction calcite cementation in Diagenetic Regime 1 below each of these surfaces suggests that meteoric diagenesis associated with each exposure surface (cementation, mineralogical alteration of allochems and pedogenesis at the exposure surfaces) would have been a common phenomenon. Indeed, T A B L E 4 Population statistics for δ 13 C and δ 18 O compositions of calcite cement zones. Standard deviations (±1σ) are only provided for datasets with at least five members. 'Zone 6 concretion' refers to strongly ferroan calcite vug fills in pedogenic concretions. 'Zone 6 fracture' refers to strongly ferroan cement filling late fractures. the depletion of 18 O and 13 C in the pedogenic carbonates and cement Zones 3-5 relative to calcite in equilibrium with syn-depositional sea water (Figures 10, 11 and 13) is strong evidence for meteoric alteration. The fine CL zonation in these cements, particularly Zone 4, is similar to that of meteoric cements described by Meyers (1978) and Walkden and Berry (1984).

| Oxygen
Interpreting the nuances of meteoric alteration in the CVSGp, especially identifying differential responses between the various units and cement zones, is facilitated by first establishing the general isotopic composition characteristic of meteoric alteration in these rocks. In meteoric systems, the total volume of oxygen in the pore waters far exceeds oxygen derived from alteration of the host carbonates, thus pure meteoric calcites tend to be water buffered with respect to oxygen and have a relatively narrow range in δ 18 O values (Allan & Matthews, 1977Lohmann, 1988). To that end, one measure of the oxygen isotopic composition of CVSGp calcite precipitated in equilibrium with meteoric water is the pedogenic carbonates. The PFm/ CFFm recrystallised/columnar calcite concretions display a narrow (1.1‰) range of δ 18 O values, suggesting a mean value for carbonate precipitated in equilibrium with meteoric water of −7.8‰ (n = 23), providing there was no evaporative fractionation when they formed ( Figure 10 and Table 2). This estimate does not differ significantly from the averages determined from PFm botryoidal calcite (−8.4‰, n = 4), although CFFm and CHMbr micrite crusts are slightly less negative (−7.3‰ and −6.3‰ respectively), which could reflect evaporation of meteoric water as those surface crusts formed, although there is a significant age difference between the two which could play a role. Taken as a whole, pedogenic carbonates (recrystallised/columnar calcite concretions, botryoidal calcite and calcrete crusts) from the CVSGp suggest an average value of −7.8‰ (n = 30) for pedogenic carbonate precipitated in equilibrium with meteoric water. Zones 3, 4 and 5 (meteoric) cement in the three occurrences of Diagenetic Regime 1 (PCOFm, BOOFm and GOFm) provide an additional estimate of carbonate precipitated in equilibrium with meteoric water. Collectively, those cements exhibit a range of −5.8 to −11.1‰ and a mean δ 18 O value of −7.4‰ (n = 25); an average not too dissimilar from the −7.8‰ average for pedogenic calcite.
These averages represent 18 O-depletions in meteoric calcites of 3.2‰ (pedogenic deposits) and 2.8‰ (cement) from the estimated KOBE-age marine value of −4.6‰. Presently, the δ 18 O value of rainwater in the tropics is typically no more than 4‰ depleted with respect to Standard Mean Ocean Water (Yurtsever, 1975). Hence, the observed 18 O-depletion in meteoric calcite is broadly consistent with the expected range for the latitude. F I G U R E 1 0 A δ 13 C and δ 18 O cross plot for CVSGp pedogenic carbonates (individual symbols), shown in relation to CVSGp meteoric Zone 4 cements (pink polygon), data from the LFm published in Wright et al. (1997) (dotted blue polygon) and the expected isotopic composition of carbonate precipitated in equilibrium with syn-depositional sea water (grey box). Numbers refer to sample localities shown in Figure 5. Thick vertical grey lines correspond to the mean δ 18 O values of PFm/CFFm recrystallised/columnar concretions and CHMbr calcrete.

| Carbon
The δ 13 C values of meteoric calcites typically reflect the addition of organic carbon in the form of soil-gas CO 2 , which is then buffered in part by the carbon added to the pore waters as marine carbonates dissolve and alter (Allan & Matthews, 1977Lohmann, 1988). Pedogenic carbonates will also have variable δ 13 C values if formed within 30 cm of the land surface due to the downward diffusion of atmospheric CO 2 at low soil CO 2 respiration rates (Quade et al., 1989). As a result, there is no typical δ 13 C value characteristic of a meteoric calcite. The amount of soil-zone carbon (reflected in the soil-zone pCO 2 ) and the extent of marine carbonate alteration will produce a range of δ 13 C values. Decay/oxidation of organic matter infiltrated through the vadose zone and across the water table can also be a local source of 12 C (Whitaker & Smart, 2007;Wood, 1985).
Quaternary meteoric calcites can have δ 13 C values as low as −12‰, with the total shift (Δ 13 C) from marine carbonates being as great as 14‰ (Allan & Matthews, 1982). That large Δ 13 C is in part reflective of the presence of C4 plants that use a photosynthetic pathway that fractionates against 13 C more so than C3 plants. The C4 plants post-date the Carboniferous, so the maximum Δ 13 C between CVSGp marine carbonate and meteoric calcite should be less. In fact, the maximum observed value of Δ 13 C is 11‰ and occurs between PCOFm pedogenic calcites (minimum −6.5‰, Figure 10) and the estimated KOBE marine carbonate of about +4.5‰. Higher in the stratigraphic section (GOFm), the maximum Δ 13 C is only about 8‰ and occurs between Zone 4 meteoric cements (δ 13 C of −6.2‰) close to the top of the GOFm and the estimated δ 13 C value of marine carbonate during GOFm deposition F I G U R E 1 1 A δ 13 C and δ 18 O cross plot for CVSGp allochems (individual symbols), shown in relation to CVSGp Zone 4 cements (pink polygon) and pedogenic carbonate polygons from Figure 10, referenced against the expected isotope composition of carbonate precipitated in equilibrium with syn-depositional sea water (grey box). Analyses from Zone 1 and 2 cements are also shown for comparison (see text for discussion). A regression line calculated for PCOFm allochems (green line) represents a mixing line indicative of progressive stabilisation of marine carbonate in meteoric water. The dashed green extension represents the expected path PCOFm allochems would follow to complete stabilisation with meteoric water typical of PFm/CFFm pedogenic carbonate formation. See text for derivation of the hypothetical marinemeteoric mixing line (dashed red) for the GOFm. Numbers refer to sample localities shown in Figure 5, brackets indicate Diagenetic Regime 2. Thick vertical grey lines correspond to the mean δ 18 O values of PFm/CFFm recrystallised/columnar concretions, Diagenetic Regime 1C Zone 4 cement and CHMbr calcrete.
(+2‰). Both estimates seem reasonable for pre-Miocene exposures and suggest that −6‰ is probably the most 12 C-depleted meteoric signal of the CVSGp. Less negative values in the pedogenic and meteoric calcites would reflect greater rock-buffering of the aqueous carbon system and lesser values in (all) allochems might also reflect the processes of allochem alteration as discussed below.

F I G U R E 1 2 (A)
A δ 13 C and δ 18 O cross plot for CVSGp cement samples highlighted by zone (colour) and formation (symbol). The various cement zones form distinct clusters, as do cements from diagenetic areas 1 and 2. The pink polygon encompasses the spread of Zone 4 cements. (B) Close-up of the distribution of data from Diagenetic Regime 1 Zones 3-5, but highlighted by zone (symbol) and formation (colour), and with polygons outlining the data fields of recrystallises/columnar calcite concretions and calcrete coating the top CVSGp unconformity. Numbers refer to sample localities shown in Figure 5. Thick vertical grey lines correspond to the mean δ 18 O values of PFm/CFFm recrystallised/columnar concretions, Diagenetic Regime 1C Zone 4 cement and CHMbr calcrete.

ALLOCHEM δ 13 C AND δ 18 O ISOTOPIC VARIATIONS: ORIGIN AND RELATIVE TIMING IMPLICATIONS
In the overview of cement isotopic results presented above it was highlighted that individual cement zones differ in their isotopic composition ( Figure 12). Consequently, when viewed in terms of the isotopic evolution of individual samples it should come as no surprise that both δ 13 C and δ 18 O values show largely consistent (but different) evolution trends, irrespective of a sample's lithostratigraphic position (Figure 13). The most complete isotopic record is provided by Diagenetic Regime 1 samples. The analyses from Diagenetic Regime 2 differ in that only Zones 1, 2b and 6 could be sampled, the other zones either being too small to separate or not present. Cements in Diagenetic Regime 2 are less 13 Cdepleted than in Regime 1, as noted above, but again their δ 13 C and δ 18 O compositions show consistent (but different) evolution trends. This section considers each cement zone in detail.

| Zone 2 and Allochems
The occasional presence of internal sediment overlying the inclusion-rich syntaxial overgrowths that constitute cement Zone 2 indicates this cement zone was F I G U R E 1 3 The δ 13 C and δ 18 O data from analyses of sequential cement zones and associated allochems, referenced against the expected isotope composition of carbonate precipitated in equilibrium with syn-depositional sea water (grey boxes). All samples are from Diagenetic Regime 1 except 17829 and 17830 which are from Diagenetic Regime 2. Lines link adjacent zone to highlight the evolution in isotopic composition through the successive cement zones (dashed lines denote the Diagenetic Regime 2 samples). Where data for the adjacent zone is missing the link is broken. In the legend box the number in brackets after the sample number indicates locality and the exact stratigraphic position of the samples is shown in Figure 7. '?' Indicates PCOFm allochems not included in regression analysis shown on Figure 11. syn-sedimentary in origin and precipitated from marine pore waters. The patchy and irregular CL characteristics of this cement phase suggests those marine cements subsequently recrystallised to a stable calcite, with the preservation of euhedral crystal forms indicating the cement was probably originally high Mg-calcite. Consequently, Zone 2 cements do not represent a single precipitation event and are probably diachronous within individual formations. Hird and Tucker (1988) described similar overgrowths on echinoderm plates in the BrOFm and deduced they formed soon after deposition, but were non-committal about the diagenetic fluids from which they precipitated.
The extensive alteration indicated by the patchy CL pattern is supported by highly variable δ 13 C and δ 18 O values that are non-marine, the former mostly decreasing upward through the formations, the latter the converse (Figures 12A and 13). The CL patches have either no, dull or moderate CL responses ( Figure 8B), but with the current data it is not possible to determine if they vary in isotopic composition. Some of the dull CL areas are mildly ferroan and possibly represent Zone 3 calcite infilling etched Zone 2 crystal. Despite this uncertainty, GOFm Zone 2 cements (Diagenetic Regime 1C) cluster together and have similar δ 13 C and δ 18 O values to associated allochems; both have meteoric δ 18 O and δ 13 C values similar to pedogenic and meteoric calcites (Figure 11). This suggests Zone 2 GOFm cements and allochems both recrystallised and completely equilibrated to the same meteoric fluid at high intracrystal and intragranular water: rock ratios respectively.
In contrast, PCOFm and BOOFm Zone 2 cements (Diagenetic Regimes 1A and 1B) also have meteoric δ 18 O values, but have highly variable δ 13 C isotopic compositions ( Figure 11) and show no well-defined relationship with associated allochems. This suggests that PCOFm and BOOFm Zone 2 cements also recrystallised in meteoric waters, but with differing intracrystal water: rock ratios (i.e., semi-closed neomorphic systems) that mostly resulted in complete isotopic equilibration with respect to the δ 18 O value of meteoric water, but often with incomplete isotopic equilibration of the δ 13 C signal (Lohmann, 1988). Such a phenomenon is well-documented and results from the fact that recrystallisation is an intragranular or intracrystal process decoupled from bulk pore water chemistry (Budd & Hiatt, 1993;Machel, 1990;Pingitorre Jr., 1982). The possibility that some of the patchy CL pattern could be due to cement infilling etched areas within Zone 2 cement could also contribute to variability.
The fact that PCOFm and BOOFm allochems lie on a mixing line in Figure 11 indicates that they also altered at varied intragranular meteoric water: rock ratios. The further up the mixing line that allochems are located the lower the intragranular water: rock ratios. Allochems with δ 18 O and δ 13 C isotopic compositions similar to meteoric calcites and pedogenic carbonates altered at high water: rock ratios, like GOFm allochems and GOFm Zone 2 cements. The dotted extension to the PCOFm mixing line in Figure 11 indicates the expected trajectory that Diagenetic Regime 1A and 1B allochems would follow towards complete stabilisation in meteoric water with an isotopic composition similar to that responsible for PFm/CFFm pedogenic deposits. All of the allochems displaying near complete stabilisation plot to the right of this line, close to a value associated with GOFm meteoric cement, including allochems from the PCOFm and BOOFm.

| Zone 3
The generally faintly ferroan nature and dull luminescence of Zone 3 marks an episode of precipitation in all three oolitic formations after formation of inclusion-rich syntaxial overgrowths. The CL and staining indicate iron in the crystals, thus a source of iron and persistent suboxic to anoxic conditions in the pore waters. Zone 3 cannot represent a single precipitation event within the CVSGp because, if so, that would mean meteoric cementation of the PCOFm and BOOFm was delayed until after GOFm deposition. Different timings in the various units is also supported by the different δ 18 O values for Zone 3 cements in the GOFm (mean − 8.4 ‰) and PCOFm (mean − 10.7 ‰), representing a significant decrease in values compared to Zone 2, especially in the PCOFm (Figure 13). Both of the PCOFm samples and one of the GOFm samples have lower δ 18 O values than the lowest PFm/CFFm pedogenic carbonate value, indicating unusual levels of 18 O depletion ( Figure 12B). It was not possible to separate Zone 3 cements from BOOFm samples for isotopic analysis.
Mean δ 13 C values for Zone 3 cements in the GOFm (−4.8‰) and PCOFm (−5.7‰) also represent a significant decrease compared to Zone 2 ( Figure 13). These values fall within the range displayed by pedogenic carbonates and Zone 4 cement, consistent with a meteoric origin for Zone 3 cements.
Processes that could lead to the unusually low δ 18 O values include: • Pore fluid with an elevated temperature: Such pore fluids would probably be derived from the basin and unlikely to have the requisite carbon isotope signature. Expulsion would also have to occur repeatedly at the same stage in the cementation history of each oolitic formation and been synchronised with meteoric cementation events, making this origin highly unlikely. • Addition of meteoric water with lower δ 18 O values than implicated in the formation of other cements and pedogenic carbonates recorded in the CVSGp: The simplest mechanism for this interpretation is climate variations affecting the δ 18 O composition of rainfall. That is, at the onset of exposure of both the PCOFm and GOFm rainfall was initially more 18 O-depleted than when the majority of meteoric cements (Zone 4) and pedogenic products formed. Comparing sequential extracts of Zone 3 and Zone 4 cements, mean δ 18 O values of the former are more negative than the latter by 2.4‰ and 1.5‰, for the PCOFm and GOFm respectively (Figure 13). Based on this small dataset climate driven change in the isotopic composition of rainfall could be a viable explanation for the observed difference in the δ 18 O values of Zone 3 and 4 cements.
It is concluded that Zone 3 cements were most likely precipitated during early exposure of the oolitic units and from meteoric waters that were more 18 O-depleted than the meteoric waters that formed the subsequent Zone 4 cements. The mechanism driving the repeated shift to less 18 O-depleted rainfall after the onset of exposure is considered further below. The δ 13 C values indicate that the early meteoric cements of Zone 3 were associated with ample soil-zone CO 2 . The faintly ferroan nature of the cements may simply reflect suboxic to slightly anoxic conditions that allow the reduction and incorporation of iron from oxyhydroxides within the original sediments.

| Zone 4
The sequential sample data typically show a modest increase in δ 13 C values from Zone 3 to Zone 4, but a much larger increase in δ 18 O values ( Figure 13). As noted above in the discussion of the isotopic signal of meteoric calcites, the meteoric origin of Zone 4 cements is supported by: • Petrographic and CL characteristics typical of cements interpreted by other authors as meteoric. • The consistent and pervasive occurrence of Zone 4 cement beneath subaerial exposure surfaces, suggesting the bulk of Zone 4 cementation occurred during subaerial exposure immediately following deposition. • Carbon and oxygen isotopic compositions that overlap pedogenic carbonates. • δ 13 C values that reflect addition of organic carbon to meteoric waters in the form of soil-gas CO 2 (Allan & Matthews, 1977. In the case of the PCOFm and BOOFm (Diagenetic Regimes 1A and 1B) the isotopic composition of meteoric water supplied during exposure was captured in PFm/ CFFm pedogenic deposits (−7.8‰). The PCOFm and BOOFm Zone 4 cements from Localities 10, 11 and 14 have a mean δ 18 O value (−8.0‰) very similar to this pedogenic value, suggesting a close genetic and temporal relationship between the cements and immediately overlying pedogenic deposits. However, the PCOFm and BOOFm Zone 4 cement samples from Locality 17 have a mean δ 18 O value (−6.6‰) that is significantly less negative than the pedogenic value ( Figure 12B). Additionally, PCOFm and BOOFm allochems that have reached equilibrium with respect to meteoric water δ 18 O values, notably those from the north-western part of the study area including Locality 17, have a mean δ 18 O value (−6.9‰) similar to that of Locality 17 Zone 4 cement.
Although the number of samples is small, these differences suggest that pedogenesis and Zone 4 cementation may not have been coeval at the north-western end of the study area. The fact that this is true for both the PCOFm and BOOFm makes an explanation in terms of short term climate change problematic since the cause must be site specific. Locality 17 is located close to the BOF where the top CVSGp unconformity cuts down to close to the top of the BOOFm ( Figure 6). Additionally, the δ 18 O values of PCOFm and BOOFm Zone 4 cements at Locality 17 are similar to those of the GOFm (see below). This raises the possibility that at the north-western end of the study areas the PCOFm and BOOFm were affected by meteoric waters introduced during the erosional event that delimits the top of the CVSGp. In this scenario residual porosity experienced a second phase of Zone 4 precipitation. The sampling protocol was biased towards pores with Zone 6 cement and so there could be preferential sampling of the later stages of Zone 4 cementation. To test this hypothesis an understanding of the temporal variations in δ 18 O values through Zone 4 cements is required, which cannot be addressed by manual separation of cement samples.
The δ 18 O values of GOFm Zone 4 cements (Diagenetic Regimes 1C and 1D) (mean − 6.7‰) are less negative than the PFm/CFFm pedogenic carbonates and Zone 4 cements from Localities 10, 11 and 14. Again, climate change resulting in slightly different average rainfall δ 18 O values in the younger GOFm is indicated. The GOFm Zone 4 data fall into two distinct clusters ( Figure 12A,B): (1) Diagenetic Regime 1C, mainly 'Coral Bed' data, has a narrow range of δ 18 O values with a mean of −6.9‰ and (2) Diagenetic Regime 1D, associated with the top of the GOFm, but pre-dating the top CVSGp erosion surface, is distinguished by having lower δ 13 C values than those from Diagenetic Regime 1C and also displays a narrow range in δ 18 O values with a mean of −6.6‰. There are two possible scenarios to explain these data. Scenario 1 is that the Zone 4 cements in Diagenetic Regimes 1C and 1D represent the same cementation event and that the greater δ 18 O value encountered at the top of the GOFm resulted from evaporative fractionation. This interpretation is consistent with the lower δ 13 C values which suggest proximity of a vadose zone. In this case it is preferred to use Diagenetic Regimes 1C data (−6.9‰) as an estimate of the prevailing meteoric water oxygen isotopic composition. Scenario 2 is that Diagenetic Regime 1C Zone 4 cement formed in a meteoric lens that pre-dated the latter part of GOFm deposition and that Diagenetic Regime 1D Zone 4 cement formed later. In this case the data suggest a small (0.3‰) positive shift in meteoric water δ 18 O values over time.
9.1.4 | Zone 5 Only four isotopic analyses are available to help explain the dull to moderately bright luminescing and faintly ferroan Zone 5 cements. The Zone 5 cement analysed from the PCOFm (Diagenetic Regime 1B) was precipitated on top of Zone 4 cement and has δ 18 O and δ 13 C values (−10.0‰ and −5.4‰ respectively) suggesting a return to more 18 O-depleted meteoric water similar to that responsible for Zone 3 precipitation ( Figures 12B and 13). The highly negative δ 18 O signature cannot be explained in terms of contamination by the adjacent Zone 6 cement because this has δ 18 O and δ 13 C values of −7.5‰ and −2.0 ‰ respectively.
As was the case for Zone 3, the three GOFm Zone 5 cement samples (Diagenetic Regime 1C) are not as 18 Odepleted as PCOFm Zone 5 (mean − 6.0 ‰). This is 1.0‰ greater than Zone 4 cements from the same samples. Overall this suggests that the GOFm Zone 5 cements reflect slight 18 O enrichment in the composition of meteoric water and represent a continuation of the trend recorded between Zone 3 and Zone 4 in the same samples ( Figure 13). Zone 5 cements in the GOFm differ from those in the PCOFm in that they consist of alternating ferroan and non-ferroan (Zone 4-like) calcite and the current data represent an average of these two subzones. It is not possible to resolve with the current dataset whether this reflects precipitation from pore fluids with differing isotopic compositions and oxidation states or merely fluctuations in the oxidation state of a pore fluid with a constant isotopic composition. The fine CL zonation in Zone 4 cements suggests that the meteoric systems responsible for their precipitation were not uniformly oxic and the more pronounced fluctuations in GOFm Zone 5 cements could be a more extreme manifestation of this variability. In this case the increase in alternations between ferroan and non-ferroan subzones in a distal direction could reflect a system increasingly close to the oxic-anoxic tipping point. The three Zone 5 samples also show a trend to decreasing δ 13 C values in a distal direction. More data are required to determine whether this is anything more than coincidence. 9.1.5 | Zone 6 The more negative δ 18 O values of Zone 6 compared to Zone 5 can be attributed to increased temperature. This, together with the highly ferroan nature of Zone 6 cement and the fact that it was preceded by an episode of aggressive ooid dissolution, suggests the expulsion of hot basinal fluids into the CVSGp, probably associated with Variscan compression. Dickson and Kenter (2014) described similar ooid dissolution in Tengiz Field associated with ingress of basinal fluids. On reaching Diagenetic Regime 1 the basinal fluids were forced to migrate through microporosity present within ooids because the majority of the macroporosity was cemented. This caused extensive ooid dissolution along preferred migration routes, assimilating variably stabilised ooid carbonate with its generally negative δ 13 C values as it did so. Low permeability would have slowed flow rates allowing time for the fluids to cool as they progressed, hence the decrease in Zone 6 δ 18 O values in a distal to proximal direction and upward through the formations ( Figures 12A and 13). Fluid entering the CVSGp via fractures could bypass the ooid microporosity system, thus experiencing less cooling and precipitating calcite with lower δ 18 O values and a higher and narrower range of δ 13 C values. 9.2 | Diagenetic Regime 2 9.2.1 | Zone 1 and Allochems Zone 1 inclusion-rich syntaxial overgrowths are interpreted as distal equivalents of inclusion-rich syntaxial overgrowths (Zone 2) in Diagenetic Regime 1, i.e. they are of marine origin. Their strongly ferroan character indicates the establishment of anoxic conditions soon after deposition which is consistent with widespread early diagenetic pyrite ( Figure 9F). The δ 18 O values (−7.7‰ and −7.5‰) of Zone 1 PCOFm samples is compatible with recrystallisation (mineralogical stabilisation) in meteoric waters of similar isotopic composition to those associated with the PFm/CFFm pedogenic carbonates. Their δ 13 C values are positive (+0.1‰ and +0.6‰) indicating preservation of a strong original marine component during recrystallisation and, hence, little 12 C in the meteoric fluids during stabilisation. This is compatible with the distal location of Diagenetic Regime 2. By the time meteoric water had fluxed down-dip from the up-dip exposure surfaces, the soil-zone carbon had been buffered by water: rock interactions, but oxygen had not due to the larger amount of meteoric oxygen in the pore water.
The two Diagenetic Regime 2 PCOFm allochem samples not included in the regression analysis shown on Figure 11 (labelled 7) also stand out by having unusually high δ 13 C values compared to δ 18 O. These samples were extracted from the same rock samples as the Zone 1 cement, suggesting that they too have stabilised in the same pore water environment responsible for the Zone 1 cements, although their δ 18 O values are more aligned with Diagenetic Regime 1 Zone 3 cement than the PFm/CFFm pedogenic carbonates.
The above explanation for the isotopic composition of Zone1 cements and associated allochems is not consistent with all of the Diagenetic Regime 2 allochem data. Five of the PCOFm and BOOFm allochems from Diagenetic Regime 2 plot on the same mixing line as allochems from Diagenetic Regime 1 (Figure 11), two of these data points coming from the same rock samples as the Zone 1 cement data. These inconsistencies possibly point to a spatially variable pore water history. It has already been noted that Zone 4 cement is only patchily present in Diagenetic Regime 2. This conundrum cannot be resolved with the current dataset, but the nature of the outcrops means that it should be possible to map the diagenetic transition from Diagenetic Regime 1B into Diagenetic Regime 2 in detail.

| Zone 2b
Non-ferroan Zone 2b displays moderately bright, relatively uniform CL that is probably due to the presence of Mn 2+ in excess of Fe 2+ , with the Mn 2+ incorporation into the calcite lattice indicating sub-oxic conditions. The change to non-ferroan calcite could either mark the exhaustion of the Fe 2+ supply or a slight increase in oxygenation level relative to Zone 1. The isotopic composition of Zone 2b is broadly similar to that of Zone 1 suggesting precipitation in a meteoric water buffered system, but far down the flow gradient such that soil-zone carbon had been buffered by water: rock interaction. There is close similarity between the isotopic composition of Zone 2b cement from sample 17830 and some associated ooids ( Figure 11).

| Zone 4
Zone 4 is only patchily present and where it occurs forms a zone too thin to sample. Its petrographic and CL characteristics, however, suggest it is similar to Zone 4 cement in Diagenetic Regime 1. It most likely represents the distal limits of the meteoric systems developed during exposure of the PCOFm and BOOFm in more proximal locations (Diagenetic Regimes 1A and 1B).

| Zone 6
This zone is also similar to that found in Diagenetic Regime 1, where it was linked to the expulsion of hot basinal fluids (see above). However, in Diagenetic Regime 2 Zone 6 cement has positive δ 13 C values that indicate a more rockbuffered carbon system in the pore waters than in Regime 1 (Figures 12A and 13). Allochems in Diagenetic Regime 2 also still retain much of their original marine carbon signature. The paucity of any negative δ 13 C components (allochems - Figure 11) in the rocks means that when the basinal fluids influxed any water: rock interaction would not have lowered the fluids' δ 13 C values. Further, much intergranular porosity survived early cementation and compaction, meaning that basinal fluids could easily flow through Diagenetic Area 2, with minimal interaction with rock components. Consequently, there is no visible evidence of ooid dissolution.

| Outcrop and petrographic evidence
There have been a number of publications on the Lower Carboniferous limestones of South Wales which suggest humid and arid climates dominated at different times based on comparisons between calcretes, karst and early diagenetic styles with modern analogues (Hird & Tucker, 1988;Riding & Wright, 1981;Searl, 1988bSearl, , 1989aWright, 1980Wright, , 1982aWright, , 1982bWright, , 1984Wright, , 1988. In general, these suggest more humid climates prevailed during the Courceyan than the Chadian and Arundian, consistent with an overall cooling of the climate in the run-up to the LPIA. A variety of pedogenic carbonate deposits occur at multiple levels in the PFm and CFFm in the study area (Figure 7; Searl, 1988bSearl, , 1989b and elsewhere within the BOOFm (Searl, 1988a). Calcretes are usually associated with semi-arid settings and specifically are typical of seasonal climates, especially those with winter rainfall and long, dry summers (James & Jones, 2016). Searl (1988aSearl ( , 1988b variably described the climate conditions under which the PFm was deposited as (1) hot with a strongly seasonal rainfall and (2) seasonally arid. She concluded that micritic calcrete probably formed during dry seasons and pedogenic columnar calcite during wet seasons. The presence of extensive meteoric cement in the PCOFm and BOOFm below subaerial exposure surfaces supports a well-developed wet season. Hird and Tucker (1988) actually described the climate as 'humid' based on the extent of meteoric cementation and low grain compaction in the BrOFm (distal equivalent of BOOFm). These palaeoclimate assessments are consistent with the location of the South Wales-Mendip Shelf within the tropics during the Early Carboniferous (Figure 1), a zone typified by warm temperatures and seasonal weather patterns. It is noteworthy that the PFm and CFFm pedogenic deposits coincide with the KOBE event (Figure 4). If this was indeed associated with a transitory glacial episode (Caputo et al., 2008;Haq & Schutter, 2008;Isbell et al., 2012;Krammer & Matchen, 2008;Liu et al., 2019;Montañez, 2022), then cooling was probably accompanied by a decrease in humidity which would favour calcrete formation over karst development.
The only climate indicator in GOFm outcrops is the karst capping the CVSGp. This is indicative of increased rainfall compared to the KOBE era pedogenic deposits, although this could represent a short-lived climate event (Wright, 1980(Wright, , 1982a(Wright, , 1988. The exact temporal relationship between karst formation and GOFm cementation is unresolved, but the present unconformity cuts GOFm Zone 4 cement, hence the presence of the karst provides little evidence of climate conditions during GOFm deposition and meteoric diagenesis. The most that can be said is that well-developed meteoric cements in the GOFm indicate an abundant supply of meteoric water, at least on a par with that experienced by the underlying oolitic formations.
To the south of the study area ( Figure 2) Hird and Tucker (1988) identified a return to a semi-arid climate during GuOFm (Chadian) meteoric diagenesis. Minor meteoric cementation and high grain compaction below the capping subaerial exposure surface indicate a less well-developed wet season than during the early diagenesis of the CVSGp oolitic formations. Similar conditions persisted into the Arundian as evidenced by the presence of calcrete at the base of the Caswell Bay Mudstone Formation (CBMFm) and its up-dip equivalent the LFm (Wright, 1980(Wright, , 1981(Wright, , 1982b(Wright, , 1988.

| Temporal trends in meteoric water δ 18 O values during CVSGp diagenesis
Glacial ice is strongly 18 O-depleted, meaning an increase in ice volume (glaciation) results in 18 O-enrichment of sea water (Grossman & Joachimski, 2020). This is reflected in an increase in the δ 18 O values of marine authigenic minerals that is reinforced by associated lower temperatures. Consequently, average global increases of about 2‰ in the δ 18 O values of brachiopod calcite and conodont apatite over the time period represented by the PCOFm to basal CHMbr interval (about 9 Myr) is a response to overall global climate cooling and associated increased ice volume (Figure 4). Additionally, conodont data indicate high frequency oscillations of 1-2 Myr duration superimposed on the long term trend in sea water δ 18 O values (Figure 4, Buggisch et al., 2008).
It is reasonable to expect meteoric calcite to mirror changes in sea water δ 18 O values, although it would be simplistic to assume a linear relationship between the two. The following section consolidates the isotopic observations discussed above to provide a summary of the temporal changes in meteoric water δ 18 O values recorded in the CVSGp.

| Meteoric cements
Volumetrically, Zone 4 accounts for the majority of meteoric cement. The δ 18 O values of Diagenetic Regime 1A and 1B (PCOFm and BOOFm) Zone 4 cements from Localities 10, 11 and 14 (mean − 8.0‰) is similar to PFm/ CFFm pedogenic carbonates (mean − 7.8‰), reinforcing the conclusion that the PFm/CFFm pedogenic carbonates provide a reliable estimate of meteoric water δ 18 O values during the KOBE part of the stratigraphy. The more positive value of Diagenetic Regime 1A Zone 4 cements from Locality 17 (mean − 6.6‰) is difficult to explain in terms of climate change, as discussed above, and requires further investigation.
Two scenarios have been presented to explain the δ 18 O values of Zone 4 cements in Diagenetic Regimes 1C and 1D (GOFm). In one scenario meteoric water was determined to have a mean δ 18 O value of −6.9‰, in the other it reduced from −6.9‰ during cementation of the lower part of the GOFm to −6.6‰ during cementation of the upper part of the GOFm. This is an increase compared to the KOBE interval and a trend that is continued in Zone 5 cements in Diagenetic Regimes 1C (mean − 6.0‰).
The Zone 4 cement data are not alone in indicating a positive shift in meteoric cement δ 18 O values from the PCOFm to the GOFm. Mean δ 18 O values of Zone 3 cement data increase from −10.7‰ in the PCOFm to −8.4‰ in the GOFm (Figure 13), a change of 2.3‰. This is larger than the shift between PCOFm and GOFm Zone 4 cements in the same samples (1.0‰), but given the limited dataset more weight should be given to the consistent direction of the change rather than its absolute amount.
The combination of pedogenic and Zone 4 meteoric cement data reveal a long term trend of increasing meteoric water 18 O enrichment over time, the direction and scale of which is similar to that suggested by global δ 18 O carbonate and apatite curves (Figure 4).
There is no evidence for a decrease in δ 18 O values at the end of the KOBE event to support warming and deglaciation, although the extensive meteoric cementation of the GOFm confirms ample rainfall during wet seasons. Similarly, the data fails to provide evidence of an increase in humidity during formation of the top CVSGp karst, although it is possible this did not leave an isotopic record. The only evidence that could indicate a warmer, wetter post-KOBE climate is the higher degree of stabilisation of GOFm allochems compared to the PCOFm and BOOFm. However, the increased water: rock ratios implied could be due to a longer period of exposure to meteoric water rather than an increase in meteoric water flux.

| Short timescale
An increase in δ 18 O values from Zone 3 to Zone 4 cements within individual samples from both the PCOFm and GOFm (mean changes of 2.4 and 1.5‰ respectively) has already been discussed ( Figures 12B and 13), with climate driven changes in rainwater δ 18 O values suggested as the probable cause. The GOFm 'Coral Bed' data also show a consistent increase in δ 18 O values from cement Zone 4 (mean − 6.9‰) to Zone 5 (mean − 6.0‰). The single PCOFm Zone 5 cement sample shows a trend in the opposite direction to that seen in the GOFm, to a Zone 5 value similar to Zone 3 cement. These observations all suggest fluctuations in isotopic composition of meteoric water on a relatively short timescale superimposed on the long-term trend.
Five oscillations in conodont apatite δ 18 O values were recognised by Buggisch et al. (2008) between the KOBE event and the top of Sequence 2 (Figure 4), the first apparently coinciding with the PCOFm and BOOFm (peak KOBE), the second with the CFFm (end KOBE), the third with the GOFm (post-KOBE), the fourth with the interval between the GOFm and GuOFm and the fifth with the GuOFm. If this correspondence between CVSGp formations and sea water δ 18 O values is correct it suggests climate control on both deposition and diagenesis. This suggestion is similar to that of Poty (2016), who proposed that the depositional architecture of the Tournaisian-Viséan of the Namur-Dinant Basin was controlled by orbitally forced variations in climate, the implication being that associated temperature fluctuations controlled ice volume and, thereby, sea level. In Poty's (2016) model highstands correlate with warmer climates and lowstands with colder climates. Consequently, subaerial exposure would be associated with a reduction in temperature and an associated increase in sea water δ 18 O, with a concomitant increase in the δ 18 O of rainfall. This is indeed the trend recorded between Zone 3 and Zone 4 cements at the onset of meteoric diagenesis in the PCOFm and GOFm.
Such a climate driven model of deposition and diagenesis conflicts with the current interpretation that deposition on the South Wales-Mendip Shelf was largely tectonically controlled (Ramsay, 1991). Testing this model will require improved biostratigraphic control of CVSGp formations, better integration of UK biostratigraphic schemes with those from other regions and a rigorous evaluation of the depositional environment of the CVSGp oolitic formations along the lines of that undertaken by Searl (1988a) for the AOSGp. These inputs are all required for the development of a modern sequence stratigraphic model for the CVSGp. There is also the potential to acquire more δ 18 O data from pedogenic deposits and meteoric cements in order to improve precision regarding the temporal changes in meteoric water δ 18 O values. For instance, no isotopic data exists from intra-BOOFm calcretes in the AOSGp. Additionally, the application of laser ablation microprobe techniques, developed since this study was undertaken, would allow collection of (1) data from cement zones that proved too thin to sample manually, (2) multiple, sequential intra-Zone 4 data and (3) a larger, more statistically robust dataset.

| CONCLUSIONS
This paper provides a detailed analysis of the spatially and stratigraphically complex diagenesis and palaeohydrology of a dip section through an Early Carboniferous stacked, shallow marine carbonate succession. The integration of δ 13 C and δ 18 O data from cements, allochems and pedogenic carbonates has allowed a robust evaluation of the original isotopic composition of marine carbonate components and the diagenetic fluids that affected CVSGp diagenesis. The main conclusions of this analysis are: Stratigraphy • Integration of published biostratigraphic data indicates that the PCOFm, PFm, BOOFm and CFFm were deposited during the KOBE event. This conclusion is consistent with brachiopod data that indicate marine carbonate δ 13 C values of at least +4.5‰.

Diagenesis
• Two diagenetic regimes are recognised in the CVSGp oolitic sandbodies where they remain undolomitised. Diagenetic Regime 1, where evidence of subaerial exposure is widespread, accounts for the vast majority of the study area. Non-ferroan calcite cement (Zone 4), with a CL character and isotopic signature typical of a meteoric origin, fills the majority of pore space. Diagenetic Regime 2 is restricted to distal parts of the PCOFm and BOOFm where evidence of subaerial exposure is absent. It is characterised by common early diagenetic pyrite and low levels of early cementation that resulted in common grain compaction. Calcite cement is largely ferroan and mostly of burial origin. Non-ferroan calcite with meteoric characteristics is widely absent and where present is a minor component. • In Diagenetic Area 1 the same series of five cement zones (Zone 2 to 6) occurs in all three oolitic formations, reflecting a similar sequence of palaeohydrological changes affecting repeated depositional cycles. Zone 2 is considered marine, Zones 3 to 5 meteoric and Zone 6 of burial origin.

C and O isotopes trends
• In Diagenetic Regime 1 PCOFm allochem δ 13 C and δ 18 O values yield a well-defined mixing line consistent with stabilisation in meteoric water at varied water: rock ratios. The majority of PFm and BOOFm allochems conform to this mixing line. The degree of stabilisation increases up-stratigraphy and in a distal to proximal direction. Only the GOFm displays near complete stabilisation with respect to both carbon and oxygen. • The higher δ 13 C values observed in cements and some allochems in Diagenetic Regime 2 compared to Diagenetic Regime 1 can be explained in terms of a down-dip increase in buffering of soil-zone carbon by water: rock interactions. • Aggressive ooid dissolution followed by the precipitation of strongly ferroan calcite (Zone 6) is attributed to expulsion of basinal fluids into the CVSGp. The up-dip and up-stratigraphy increase in Zone 6 δ 18 O values is indicative of cooling, while the associated decrease in δ 13 C values is attributed to increasing assimilation of 12 C-enriched ooid calcite.

Climate indicators
• Pedogenic deposits and meteoric cements indicate a long term evolution in meteoric water δ 18 O values that is consistent with the established global sea water trend in terms of both direction and magnitude, but could also be climatically influenced. • Short term variations in meteoric water composition are indicated by differences in δ 13 C and δ 18 O values between successive meteoric cement zones in individual samples. The direction of these changes is similar both within and between formations. In particular, an increase in δ 18 O values between Zones 3 and 4 is an indication of short term fluctuations in rainwater composition due to changing sea water δ 18 O values and/or local climate. The fact that such changes occur across formations suggests a link between meteoric water composition and position in the diagenetic cycle, the latter being controlled by relative sea-level changes. • It has not been possible to confirm either warming at the end of the KOBE peak or the occurrence of a particularly humid episode associated with karst formation at the top CVSGp unconformity. Although, the higher degree of stabilisation evident in GOFm allochems compared to underlying formations indicates higher water: rock ratios, this could be due to a longer period of exposure rather than a wetter climate.
The recognition that deposition of the PCOFm, PFm, BOOFm and CFFm was coeval with the KOBE event opens up new possibilities for stratigraphic correlation on the South Wales-Mendip Shelf, especially in areas not subject to meteoric diagenesis. However, this conclusion is based on biostratigraphic data and a full δ 13 C curve for the CVSGp on the South wales-Mendip Shelf is yet to be established. Locations and stratigraphic units characterised by Diagenetic Regime 2 are candidates to undertake such a study because of the high degree of preservation of marine carbon in the allochems.
been improved as a result of constructive criticisms from the reviewers. In particular the author wished to thank Professor D.A. Budd who went beyond the call of duty to bring scientific rigour, clarity and focus to the text.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are openly available at the British Library e-thesis online service (EThOS) at https://ethos.bl.uk/Order Detai ls.do?uin=uk. bl.ethos.347713.