1. Palynological processing

The palynological processing was carried out using standard methodologies with 5 g of sample roughly crushed to about a mm in size and then treated with 30 ml of 37 % HCl to remove carbonates. The samples were then decant-washed to neutral to remove any mobilised Ca²⁺ and Mg²⁺ that would subsequently precipitate as an insoluble residue when treated with hydrofluoric acid (HF). The sample was then treated with 30 ml of 60 % HF to dissolve silicate minerals, followed again by decant-washing to neutral. The residue was then sieved over a 15 μm mesh with the coarse fraction washed into a glass beaker, allowed to settle and the water decanted off. To this was added 60 ml of 37 % HCl and the solution was briefly boiled using a Bunsen burner. This solubilised any neoformed fluorides. Initially the organic residue is bound with the neoformed fluoride and denser than the HCl solution. As the temperature gets close to boiling, the neoformed fluorides go into solution and the organic matter floats to the top of the beaker. The contents of the beaker are then transferred into about 100 ml of water to dilute the solution sufficiently to be resieved through a 15 μm nylon mesh without dissolving it. This process is far more effective in removing neoformed fluorides than any heavy liquid separation as the organic matter (post HF treatment) is intricately bound into the neoformed fluorides. Hence, the components of different density are not independent entities. The residue was then transferred onto a small 15 μm sieve and flushed through with MilliQ water and pipetted in a vial. Strew slides were made by evaporating a suspension of the organic matter concentrate in MilliQ water onto 26 × 22 mm coverslips, allowed to dry overnight and then permanently mounted in Elvacite 2044^TM. The Elvacite was made up with analytical grade xylene so as not to have any trace benzene within the mounting medium. No oxidation (in ≥65 % HNO₃) was required as the spores were yellow to pale brown in colour and with negligible pyrite. Amorphous organic matter (AOM) was present in a few samples and removed by a 30 s treatment with a Sonics and Materials ultrasonic probe. This preferentially fragments the AOM that can then be removed as a fine fraction by further sieving at 15 μm. The vialed residue was also top-sieved at 150 μm with the coarse residue mounted to concentrate any megaspores. For selected samples, from Tantallon and Coldstream, the <150 μm residue was also mounted to increase the number of spores present that would otherwise be diluted by the high phytoclast content. The Coldstream samples were dominated by the small spore Anapiculatisporites delicatus, so the <150 μm was sieved at 35 μm with the 35–150 μm fraction mounted in an attempt to increase the diversity of spores recovered. The slides were investigated using an Olympus BHS 313 microscope (230272) and quantitative counts made of over 200 spores on a series of mid coverslip traverses using a ×40 objective. The remainder of the slide was scanned at lower magnification (×20 objective) and any additional species recorded as out of count.

The sub gram fragment of mudstone taken from the partial tetrapod jaw Occidens portlocki (BGS number GSM 28498) posed a significant challenge in palynological processing. It was treated with 37 % HCl, decant-washed and then treated with 60 % HF followed by decant-washing to neutral. It was then sieved at 15 μm. There was a small component of neoformed fluoride which was left in the samples as every cycle of sieving resulted in the loss of a significant proportion of the sample via processes such as surface tension creep. All the resulting thin residue was mounted in Elvacite 2044 on five 18 mm² coverslips. Special care was taken to avoid cross-contamination. At the time, the processing lab was not running any Carboniferous samples; those going through in the same weeks were from the Middle Devonian and Jurassic–Cretaceous boundary. It is also important to be aware that, sometimes, specimens are stabilised and repaired using powdered rock from other formations selected to match the colour of the original matrix.

2. Pederpes finneyae, Auchenreoch Glen

Pederpes finneyae was obtained from a single nodule at the top of the Ballagan Formation section in Auchenreoch Glen at NS 241827, 678323. It was collected in 1971 and initially identified as a fish, so it was not until it was reassigned to the tetrapods (Clack & Finney Reference Clack and Finney2005) that its importance was realised. Hence, there was little attempt immediately following collection to obtain further information or material apart from the field sketch made during its discovery by Aspen (Fig. 3A in Clack & Finney, Reference Clack and Finney2005). The section (Fig. 2, top) is best approached along the Garshake Road to the road end and then by following the path before cutting diagonally across the Auchenreoch Muir to the southern lip of the glen, which can then be descended northwestwards down the somewhat precipitous spur to its confluence with the Murroch Burn. The section is then accessed by climbing the waterfall that occupies the narrow cleft (formerly the Gates of Sodom (e.g., Anon 1862) and subsequently Adam's Gate (Lees Reference Lees1933) but referred to as the Gates of Eden in Clack & Finney (Reference Clack and Finney2005)) made by the Murroch Burn as it cut down through a prominent dyke. The section was vertically logged (Fig. 3) from NS 241822, 678363 to 241827, 678323. These end points have a vertical GPS altimeter height difference of 34 m that tallies well with the 28 m of directly measured thickness but contrasts with the estimated 44 m of section in Clack & Finney (Reference Clack and Finney2005).

Figure 2 Location photographs of the Ballagan Formation exposure at Auchenreoch Glen, Dunbartonshire. (1) The cliff of the main exposure with the two prominent cemenstones marked. It was between these two dolostones that the nodule containing the tetrapod Pederpes finneyae was found. (2) Close-up of the two dolostones; note the red palaeosol in the upper part. (3) The arcuate crack sets (c) that are confined to the interval between the two prominent dolostones and together with the bleached colour confirm the interval as a vertisol. These exposures are difficult to image because of the precipitate nature of the exposure.

Figure 3 Logged Ballagan Formation section from Auchenreoch Glen showing location of palynological samples and position of Pederpes.

The total thickness of Ballagan Formation in the Dumbarton area is estimated at 140 m (Paterson et al. Reference Paterson, Hall and Stephenson1990). Its base is transitional from the underlying sandstone of the Kinnesswood Formation and it passes vertically upwards into the Clyde Sandstone Formation. As such it represents only about 20 % of the >500 m thickness proven in the Tweed Basin (Millward et al. Reference Millward, Davies, Brand, Browne, Bennett, Kearsey, Sherwin and Marshall2018). The logged section is entirely typical of the Ballagan Formation, comprising grey-coloured siltstones and mudstones interbedded with interesting dolostones that are generally 10 to 30 cm thick and which can range from continuous beds to lines of nodules. The section includes (Fig. 3, between 10–15 m) a maroon-coloured palaeosol doublet separated by four conspicuous nodule beds and gypsiferous ribs. The 2.05 m thick bed at 26.7 m (Fig. 2, middle) that included the nodule with Pederpes is between two conspicuously thicker continuous cementstone beds. It is mottled buff grey to red in colour and restricted within it are numerous sub-vertical inclined cracks characteristic of a vertisol palaeosol (Fig. 2, bottom). The uppermost part of the palaeosol is reddened. Vertisols are distinctive palaeosols (Kovda Reference Kovda2020) that represent sustained aridity under a relatively inactive climate system with a low relative P/E ratio. The palaeoclimatic indicators through the logged section are all indicative of aridity apart from at 21.4 m where there is a 0.5 m thick interval of black mudstone representing an interval of wetter conditions.

The Auchenreoch Glen section contains some 33 cementstone layers in 28 m of total section giving a count of 11 dolostones per 10 m of section. This is the effective average number (Bennett et al. Reference Bennett, Kearsey, Davies, Leng, Millward, Smithson, Brand, Browne, Carpenter, Marshall, Dulson and Curry2021) for the West Mains Farm borehole and the Burnmouth section. It shows that the section at Auchenreoch, on the northwestern side of the Midland Valley, is not obviously condensed (Millward et al. Reference Millward, Davies, Brand, Browne, Bennett, Kearsey, Sherwin and Marshall2018) and represents a fraction of rather than the entire thickness of the Ballagan Formation.

Forty-one samples were collected (Fig. 3) and these were all palynologically processed. Only three samples were entirely barren, being devoid of organic matter. Of the remaining samples, 27 contained only degraded phytoclast debris and spores. It is unclear whether palynomorphs were lost from these samples through contemporaneous oxidation in the arid climates of the Ballagan Formation or from recent weathering. It is presumed to be a combination of both. The remaining 11 samples contained well-preserved spores suitable for identification, although none were from or immediately above or below the Pederpes level. This sporadic palynological recovery from what was a carefully selected suite of samples by an experienced collector emphasises that successful palynological recovery in the Ballagan Formation requires significant numbers of spot samples and that these must all be processed for palynomorphs.

2.1 Auchenreoch Glen palynological assemblage

The recovered palynological assemblages (Supplementary Information available at https://doi.org/10.1017/S1755691024000100) were entirely typical of the Ballagan Formation in being dominated by simple smooth and apiculate spores, including Schopfites claviger and Retusotriletes incohatus. More complex abundant spores include Colatisporites decorus, Auroraspora macra and Anaplanisporites baccatus. There are also a number of distinctive spores, generally present in most samples but out of count (i.e., much less than 1 %). These include Knoxisporites (K. literatus and K. triangulatus), Vallatisporites spp., Spelaeotriletes spp., Velamisporites polyptycha, Convolutispora (C. major and C. permixta) and Remysporites magnificus.

Table 1 Spores from the Occipens portlocki jaw. Taxonomic citations not in references can be found in Playford & Melo (Reference Playford and Melo2012) or Owens et al. (Reference Owens, McLean and Bodman2004, Reference Owens, McLean, Simpson, Shell and Robinson2010)

Megaspores are sporadically present in the standard palynological preparations as the isolated spines of Setosisporites. There are also isolated single and paired megaspores of Didyomosporites scottii. The single specimens can be identified by the presence of an adhering much smaller abortive spore (Chaloner Reference Chaloner1958; Fig. 4, 31) which includes its own small trilete mark.

Figure 4 Illustration of important stratigraphic spores from the Ballagan Formation tetrapod localities. All figured spores are in the palynology collections of the British Geological Survey, Keyworth. Microscope co-ordinates refer to Olympus BHS-313 No. 230272 in the School of Ocean and Earth Science, University of Southampton. England Finder co-ordinates (e.g., R20-3) are also provided. Scale bars are all 10 μm. 1. Auroraspora macra Auch-31.2 117.7, 18.3 (J17-1). 2. Colatisporites decorus Auch-31.2 131.9, 11.7 (Q31-2). 3. Plicatispora scolecophora Auch-31.2 122.0, 8.2 (T21-3). 4. Schopfites claviger Auch-31.2 133.0, 6.9 (V32-2). 5. Pustulatisporites gibberosus Auch-23 111.7, 11.1 (Q10-4). 6. Knoxisporites triangulatus Auch-31.2 109.3, 5.8 (W8-1). 7. Knoxisporites concentricus oblique compression Auch-28 120.9, 6.0 (W20-1). 8. Knoxisporites literatus Auch-31 132.4, 11.3 (Q32-3). 9. Remysporites magnificus Auch-31 131.4, 11.9 (P31-3). 10–11. Prolycospora claytonii (10 is distal focus showing verrucate sculpture, 11 is equatorial focus showing minute spines on sculptural elements) Auch-31 134.0, 18.3 (J33-2). 12. Retusotriletes incohatus Auch-31.2 112.8, 18.3 (J12-1). 13. Velamisporites polyptycha Auch-30 127.8, 21.2 (F27-1). 14. Baculatisporites fusticulus Auch-14 135.3, 13.0 (O35-3). 15. Raistrickia spathulata Auch-29 120.2, 17.1 (K19-4). 16. Raistrickia clavata Auch-28 127.9, 6.1 (V27-4). 17. Raistrickia superba Auch-32 127.1, 15.7 (L26-4). 18. Convolutispora tuberosa Auch-23 110.1, 7.0 (V9-1). 19. Convolutispora caliginosa Auch-31 122.5, 12.8 (P22-1). 20. Convolutispora major Auch31.2 110.5, 7.7 (U9-2). 21. Convolutispora vermiformis Auch-28 134.8, 9.3 (S34-4). 22. Crassispora aculeata Auch-14 126.0, 21.1 (F25-2). 23. Cyrtospora cristifer Auch-29 115.6, 9.0 (S14-4). 24. Radiizonates mirabilis Auch-31.2 110.7, 16.7 (K9-4). 25. Grandispora echinata Auch-31 139.3, 7.2 (U39-3). 26. Spelaeotriletes crustatus Auch-29 109.8, 9.8 (S8-2). 27. Spelaeotriletes microspinosus Auch-28 124.8, 22.0 (E24-1). 28. Indotriradites explanatus Auch-14 118.8, 16.0 (L18-3). 29. Chomotriletes sp. Auch-31 137.0, 12.7 (P37-1). 30. Botryococcus sp. Auch-27 119.3, 15.2 (M18-4). 31–32. Didymosporites scottii (31 is a separated single megaspore with the abortive spore preserved (arrowed) next to the monolete mark. There is a barely discernible trilete mark within the abortive spore) Auch-30 113.4, 10.1 (R12-4); (32 is the megaspore pair still preserved within its outer wall layer) Auch-29 137.7, 18.8 (H37-4). These are the spores of the fern Stauropteris burntislandica (Chaloner Reference Chaloner1958). 33. Detached spine of Setosisporites pannosa with its characteristic buttressed spine base. Coldstream-2 125.4, 10.6 (R25-1). 34. Detached spine of Setosisporites. The simple spine base places it within S. pseudoreticulata. Auch-32 110.9, 10.2 (R10-3). Both S. pannosa and S. pseudoreticulata are the megaspores of the creeping lycopod Oxroadia. 35. Anaplanisporites baccatus, the microspore of Oxroadia Auch-31.2 122.6, 5.0 (X22-1).

Apart from spores, the chlorophytean alga Botryococcus was present, and particularly abundant at the base of the black mudstone at 22 m (sample Auch-31), with this interval also containing rare specimens of Chomotriletes that is now known to be an euglenoid (van der Schootbrugge et al. Reference van der Schootbrugge, Koutsodendris, Taylor, Weston, Wellman and Strother2024). No scolecodonts (the jaws of marine and marginal marine annelid worms) were found.

This assemblage from the Auchenreoch section offers the same challenge as found in all Ballagan Formation palynological assemblage in that it is apparently monotonous and with few evident inceptions. In addition, the section is in the northwestern part of the Midland Valley distant from all the detailed palynological information collected during the TW:eed Project. The lower part of the Ballagan Formation at Burnmouth contains the bases of the VI and HD spore zones. As the Auchenreoch Glen assemblages do not include these short-ranging earliest Tournaisian spores, the section is not from this lowest part of the Ballagan Formation.

In the upper part of the Ballagan Formation there are inceptions of Curriculomonoletes orbis, with Monilospora mutabilis becoming persistent. None of these occur in the Auchenreoch section, further indicating a lower, but not lowest, position within the Ballagan Formation. The presence of Didymosporites scottii at Auchenreoch is also significant. This megaspore is generally absent in the lower part of the Ballagan Formation but its presence in abundance above 250 m in the West Mains Farm borehole again indicates a position in the lower part of the Ballagan Formation. In sample Auch-31 there are two specimens of Prolycospora claytonii (Fig. 4, 10–11), which is the in situ spore of the seed fern Genomosperma kidstonii (Reeves et al. Reference Reeves, Marshall, Bennett, Davies, Kearsey and Millward2023). This is the characteristic spore within the middle part of the Ballagan Formation to the general exclusion of Anaplanisporites baccatus and the megaspore Setosisporites spp. The Auchenreoch Glen section thus has the assemblage of the lower A. baccatus acme but the inception of the spore characteristic of the overlying Prolycospora claytonii assemblage. So, it is placed (Fig. 5) above the inception of Didymosporites scottii and most likely close to the inception of Prolycospora claytonii.

Figure 5 Tetrapod localities calibrated against a composite depth scale for the Ballagan Formation based on the Norham West Mains Farm borehole and Burnmouth outcrop section. The main palynological events are based on Reeves Reference Reeves2019 and Marshall et al. Reference Marshall, Reeves, Bennett, Davies, Kearsey, Millward, Smithson and Browne2019. Note that the red bars showing the relative ages are error bars on the age not the age range of the tetrapods. An estimated geochronological timescale is provided scaled against the 12 Myr duration (Aretz et al. Reference Aretz, Herbig, Wang, Gradstein, Ogg, Schmitz and Ogg2020) for the Tournaisian Stage.

The presence of a vertisol at the Pederpes level in the Auchenreoch section is also potentially significant. In the Tweed Basin these only occur (Kearsey et al. Reference Kearsey, Bennett, Millward, Davies, Gowing, Kemp, Leng, Marshall and Browne2016) in the upper 300 m of the West Mains Farm Borehole. Their inception is higher in the Burnmouth coast section, but with fewer palaeosols recognisable in coastal outcrop this is interpreted as a recovery failure. So, the presence of a vertisol at Auchenreoch Glen provides additional support for its stratigraphic position. However, the distribution of Ballagan Formation vertisols in the west of the Midland Valley is poorly known. Carpenter et al. (Reference Carpenter, Falcon-Lang, Benton and Henderson2014) interpreted some palaeosols in the thin Isle of Bute succession as vertisols. Descriptions of siltstones in archive boreholes north of Glasgow and in Renfrewshire are generally grey, rather than red, implying the absence of vertisols. However, the Knocknairshill Borehole NS37SW10 [23056 67438] does contain red/red-brown siltstones (Monaghan et al. Reference Monaghan, Millward, Kearsey, Browne, Leslie, Smith and Strachan2024, Fig. 10.14) in the uppermost part of the Ballagan Formation beneath the Clyde Sandstone Formation, which may include vertisols, although it is difficult to be certain.

The absence of scolecodonts indicates no marine overwash events (Kearsey et al. Reference Kearsey, Bennett, Millward, Davies, Gowing, Kemp, Leng, Marshall and Browne2016; Bennett et al. Reference Bennett, Howard, Davies, Kearsey, Millward, Brand, Browne, Reeves and Marshall2017) and hence a terrestrial environment distant from any marine influence. This contrasts with the Tweed Basin sections, but is in accord with the Ballagan palaeogeography of Millward et al. Reference Millward, Davies, Brand, Browne, Bennett, Kearsey, Sherwin and Marshall2018. The abundance of Botryococcus in the dark-coloured mudstone at 21 m indicates the presence of still, oxygenated fresh water within the system (Batten & Grenfell, Reference Batten, Grenfell, Jansonius and McGregor1996), presumably on an initial flooding surface at the base of the mudstone.

In the ‘woodland hypothesis’ (Retallack Reference Retallack2011, Reference Retallack2024), tetrapods are hypothesised to have evolved and diversified in parallel with the evolution of the first forests. It was during the Devonian that these forests stabilised the fluvial system, changing it from flashy and ephemeral to a permanent system with established meandering streams and oxbow lakes. Retallack hypothesised that it was within these early forests that tetrapods were able to occupy a diverse range of stable habitats without seasonal climate extremes. Retallack (Reference Retallack2011, Reference Retallack2014) placed Pederpes in the very latest Tournaisian (i.e., above Romer's Gap) and environmentally coded the locality as a brackish lagoon. Clearly this compilation needs to be revised with Pederpes coming from the base of the middle part of the Tournaisian and preserved within a vertisol; i.e., within an environment that was characterised by sustained aridity. It is unclear how this latest Tournaisian age determination in Retallack (Reference Retallack2011) was achieved. It might simply be that it was placed at the top of Romer's Gap on the assumption that there were no tetrapods present within the gap.

3. Willie's Hole, Whiteadder Water, Tweed Basin

This is the locality on the Whiteadder Water (NT 878547) that has yielded a prolific tetrapod fauna including Koilops, Mesanerpeton and Perittodus (Clack et al. Reference Clack, Bennett, Carpenter, Davies, Fraser, Kearsey, Marshall, Millward, Otoo, Reeves, Ross, Ruta, Smithson, Smithson and Walsh2016). The locality represents an isolated exposure in the river and cannot be linked to any contiguous sequence of Ballagan Formation. The palynological assemblage comes from two samples collected from the tetrapod level. They both contain an assemblage dominated by Anaplanisporites baccatus together with a much lower proportion of Colatisporites decorus, Auroraspora macra, with Schopfites claviger being rare. Convolutispora calignosa is present together with Punctatisporites spp. and Plicatispora scolecophora. Megaspores are represented by both Setosisporites pannosus and S. pseudoreticulata. There are rare specimens of the euglenoid Chomotriletes.

This assemblage is dominated by the spores and megaspores of the lycopod plant Oxroadia (Bateman Reference Bateman1992; Stevens et al. Reference Stevens, Hilton, Rees, Rothwell and Bateman2010). In the absence of Prolycospora, Didymosporites and Curriculomonoletes, it indicates a position in the lower part of the Ballagan Formation some 100–200 m above the base of the formation. It is not possible to be more precise given the more limited assemblage.

4. Burnmouth

These data are based on the 107 productive palynological samples collected from the Burnmouth section and reported in Marshall et al. (Reference Marshall, Reeves, Bennett, Davies, Kearsey, Millward, Smithson and Browne2019), a contribution that focused on the position of the Devonian–Carboniferous boundary. The location of these tetrapods is known with reference to the logged stratigraphic section (Kearsey et al. Reference Kearsey, Bennett, Millward, Davies, Gowing, Kemp, Leng, Marshall and Browne2016) that, when combined with the West Mains Farm borehole, makes a composite standard against which the isolated tetrapod localities can be compared.

Two groups of tetrapods are known from (39584 66104), the coastal sections at Burnmouth. These are as yet unnamed tetrapods from 35 m above the top of the Kinnesswood Formation (Smithson et al. Reference Smithson, Wood, Marshall and Clack2012), close to the outer wall of the harbour. This is within the VI spore zone of earliest Tournaisian age.

The second group of tetrapods (Aytonerpeton, Diploradus and Ossirarus) is from the coastal cliffs at Ross end (9627 66054) and from 340–374 m above the base of the formation. This is within the upper interval of the Ballagan Formation rich in Anaplanisporites baccatus.

5. Coldstream, Northumberland

An as yet unnamed tetrapod (Smithson et al. Reference Smithson, Wood, Marshall and Clack2012) was recovered from the east bank of the River Tweed, some 350 m downstream of the Coldstream Bridge (NT 852403). Its location on the east bank means the locality is just in Northumberland (England) rather than the Scottish Borders. As reported by Smithson, this is a short section (15–20 m), only visible when the water level in the river level is low. Two spot samples were provided by Tim Smithson, both originating from blocks that contain vertebrates and ostracods. Both samples were rich in AOM and with abundant framboidal pyrite. Following AOM removal by ultrasonic probe, one sample (Coldsteam-1) was dominated by pale-coloured phytoclasts including cuticular sheets and with spores. The other sample (Coldstream-2) was more typical with thicker, more opaque phytoclasts and spores (Supplementary Information).

In addition to simple and apiculate spores, both samples were dominated by Anaplanisporites baccatus, with Coldstream-2 containing many smaller immature specimens, frequently in tetrads and clusters. Apart from the dominance of Anaplanisporites baccatus the diversity is low, really only containing Colatisporites decorus, Punctatisporites spp., Plicatisporites scolecophora and Convolutispora major. Spores that are generally typical of the Ballagan Formation, such as Retusotriletes incohatus and Schopfites claviger, are rare. There is only a single specimen of Spelaeotriletes sp. The Coldstream-1 sample includes rare Botryococcus.

Given the abundance of Anaplanisporites baccatus it could be considered that the assemblage represents a distal sorted assemblage. However, megaspores of A. baccatus are not uncommon, although often fragmentary. Both Setosisporites pannosa and S. reticula are present. From this, we can infer that the spores were deposited close to vegetation dominated by Oxroadia with the AOM and framboidal pyrite indicating a stratified water column with green sulphur bacteria promoting the formation of pyrite framboids. The >150 μm fraction contains megaspores and significant large fragments of plants, again evidence for locally sourced spores. The abundance of ostracods also supports the interpretation of the palaeoenvironment as a stratified freshwater lake with oxygenated upper waters.

Given the paucity of the palynological assemblage, the age determination is problematic. Currently, it can only be placed within either of the acme of Anaplanisporites baccatus. None of the spores with inceptions in the upper part of the Ballagan Formation are present (e.g., Monilospora mutabilis), but given the low diversity it is impossible to determine if their absence results from stratigraphical position or impoverished assemblage.

6. Gin Head, Tantallon and Oxroad Bay, Tantallognathus woodi

Tantallognathus woodi (Chen et al. Reference Chen, Alavi, Brazeau, Blom, Millward and Ahlberg2018) is from the historic Whitecross Collection, initially housed in the Burgh Museum in North Berwick and then transferred to the now National Museums Scotland. The original sampling locality and bed were identified by Stan Wood as being from Gin Head, Tantallon. The locality was revisited but the limestone lithology of the tetrapod bed that is within a volcaniclastic sequences was unsuitable for the preservation of palynomorphs. However, this same limestone bed is almost certainly present within the immediately adjacent sequence at Oxroad Bay and also represented in the uppermost 27 m of the Ballagan Formation of the Spilmersford and East Linton Boreholes. The uppermost Ballagan Formation level in these boreholes is from the Pu spore zone that is generally ascribed a Viséan and younger age. Immediately above the Gin Head locality is the Garleton Hills Volcanic Formation that has a recalculated (Monaghan & Parrish Reference Monaghan and Parrish2006; Monaghan et al. Reference Monaghan, Browne and Barfod2014) U–Pb zircon age of 343 ± 1.0 Ma and a sanidine ⁴⁰Ar/³⁹Ar age of 339.2 ± 3.9 Ma, i.e., from within the Viséan which has its base currently placed at 346.7 ± 0.4 Ma (Cohen et al. Reference Cohen, Finney, Gibbard and Fan2013).

In an attempt to refine and corroborate these correlations, ten samples were collected (by D. Millward) at Oxroad Bay and from the short sections as logged by Bateman & Scott (Reference Bateman and Scott1990). Oxroad Bay is a highly protected fossil locality with anatomically preserved plants so, of necessity, excavation from the foreshore below the level of surface alteration was not possible. The Oxroad Bay locality has previously been palynologically dated (Scott et al. Reference Scott, Galtier and Clayton1984) as CM zone of Tournaisian age. The palynological samples produced varied results, with a number of samples being dominated by degraded plant debris, probably a consequence of surficial alteration. However, several of the samples contained good spores, so representative assemblages could be counted from exposures B (Tantallon 2), D (Tantallon 5B) and close to C (Tantallon 3B) using the short foreshore log notation of Bateman & Scott (Reference Bateman and Scott1990).

The palynological assemblage is a little different from that found in the lower parts of the Ballagan Formation in that the diversity is lower and simple retusoid spores from the VI recovery vegetation are relatively less abundant. In addition, more complex spores such as Monilospora that have inceptions within the Ballagan Formation are absent. Anaplanisporites delicatus is present but not dominant, which is surprising given the relative abundance of Oxroadia within the locality. The tree fern pollen Prolycospora claytonii is common. Larger spores such as Baculatisporites fusticulus are abundant, perhaps reflecting an ecological signal at a plant locality.

One spore more abundant at Oxroad Bay in comparison to older Ballagan Formation samples is Radiizonates mirabilis. This distinctive spore (Fig. 6, 2) was first described from the latest Tournaisian and earliest Viséan of Clare Island, County Mayo, Ireland (Phillips & Clayton Reference Phillips and Clayton1980). Subsequently, it has been recorded in the somewhat older PC spore zone in addition to other CM records (Higgs et al. Reference Higgs, McPhilemy, Keegan and Clayton1988).

Figure 6 Illustration of important stratigraphic spores from Oxroad Bay (1–5) and the Occidens portlocki jaw (6–22). All figured spores are in the palynology collections of the British Geological Survey, Keyworth. All the Occidens portlocki slides are from BGS specimen number GSM 28498 followed by the palynological slide number, e.g., GSM 28498.2. Microscope co-ordinates refer to Olympus BHS-313 No. 230272 in the School of Ocean and Earth Science, University of Southampton. England Finder co-ordinates (e.g., R20-3) are also provided. Scale bars are all 10 μm. 1. Spelaeotriletes pretiosus Tantallon 5B 122.0, 17.0 (K21). 2. Radiizonates mirabilis Tantallon 5B 127.7, 4.8 (X27-1). 3. Lycospora noctuina Tantallon 3B 127.6, 3.9 (Y27-1). 4. Lycospora pusilla Tantallon 2 131.4, 17.2 (K31-1). 5. Lycospora pusilla Tantallon 2 125.1, 12.9 (O24-4). 6. Typical Occidens portlocki jaw spore preservation as three-dimensional and with walls perforated by mineral growth. 7. Retusotriletes incohatus GSM 28498.4 135.4, 9.5 (S35-1). 8. Plicatispora scolecophora GSM 28498.1 137.2, 9.2 (S37-3). 9. Punctatisporites irrasus GSM 28498.1 136.0, 10.0 (R36-3). 10. Cyclogranisporites commodus GSM 28498.1 134.4, 13.8 (N34-3). 11. Knoxisporites literatus GSM 28498.3 136.2, 9.5 (S36-1). 12. Corbulispora cancellata GSM 28498.3 133.7, 19.7 (G33-4). 13. Grumosisporites sp. GSM 28498.1 138.6, 13.1 (O38-4). 14. Cingulizonates bialatus GSM 28498.3 137.7, 11.0 (Q37-4). 15. Tripartites vetustus GSM 28498.3 140.8, 13.6 (O40-2). 16. Triquitrites marginatus GSM 28498.1 129.0, 16.8 (K28-4). 17. Raistrickia corynoges GSM 28498.3 141.0, 14.6 (N41-1). 18. Lycospora pusilla GSM 28498.3 137.6, 14.0 (N37-4). 19. Lycospora noctuina GSM 28498.3 135.1, 14.7 (N24-2). 20. Botryococcus sp. GSM 28498.3 137.6, 12.9 (O37-4). 21. Colatisporites decorus GSM 28498.1 131.2, 13.0 (N31-3). 22. Remysporites magnificus GSM 28498.1 136.0, 13.5 (O35-2).

A significant but rare component is Lycospora, as both L. pusilla and L. noctuina define the base of the Pu spore zone (Fig 6, 3–5). These spores are not well preserved and the cingulum appears somewhat as a fold rather than a denser ring structure. In this sense they are comparable to the cf. L. pusilla specimens illustrated by Stephenson et al. (Reference Stephenson, Williams, Monaghan, Arkley, Smith, Dean, Brown and Leng2004) from the base of the Pu zone. There is continuing debate (Higgs et al. Reference Higgs, McPhilemy, Keegan and Clayton1988; Stephenson et al. Reference Stephenson, Williams, Monaghan, Arkley, Smith, Dean, Brown and Leng2004) over whether the definition of the Pu zone should be based on the inception of Lycospora pusilla (the original definition of Neves & Ioannides (Reference Neves and Ioannides1974) from the Spilmersford Borehole), or when it became abundant (Owens et al. Reference Owens, Gueinn and Cameron1977b). This abundance of Lycospora spp. is a distinctive feature within many Viséan to Westphalian spore assemblages. It is also evident that this change to a spore assemblage dominated by Lycospora pusilla is meaningful in marking the transition to a flora dominated by large lepidodendroid trees and significantly more humid conditions. This change occurs widely across Euramerica including East Greenland (Vigran et al. Reference Vigran, Stemmerik and Piasecki1999) and Spitsbergen (Lopes et al. Reference Lopes, Mangerud and Clayton2019). In the Tweed Basin, the onset of more humid conditions is marked by a disconformity above the Ballagan Formation with the arrival of the large-scale fluvial system of the Fell Sandstone Formation.

The first appearance of Lycospora pusilla has also been used to palynologically identify the base of the Viséan Stage. However, it is clear that the inception of Lycospora pusilla is inconsistent even within Euramerica and is not a reliable indicator (Lopes et al. Reference Lopes, Mangerud and Clayton2019) for the Tournaisian–Viséan boundary; instead, its inception indicates a position within the Viséan Stage. Hence, both the geochronological age of the Garleton Hills Volcanic Formation and the palynological assemblages for Oxroad Bay indicate that Tantallognathus is Viséan in age and most likely early Viséan.

7. Occidens portlocki from County Londonderry, Northern Ireland

The small available sample came from the partial left jaw of a tetrapod specimen illustrated in Portlock (Reference Portlock1843) that was collected in Ireland during the memoir survey of County Londonderry and adjacent areas. The jaw was preserved in the collections of the BGS, now in Keyworth, Nottinghamshire, as GSM 28498. It was identified as Holoptychius and scheduled for description by Agassiz (Reference Agassiz1844–45) but was never described or figured. It was subsequently identified as a tetrapod by Clack & Ahlberg (Reference Clack, Ahlberg, Arratia, Wilson and Cloutier2004), who described and formally named it (Occidens portlocki), and noted its resemblance to the whatcheeriids and Crassigyrinus. A significant issue is that the single specimen has no detailed location or geological horizon apart from a label identifying it as Archichthys portlocki Ag with a location of Maghera, Derry. It was supplied for palynological analysis without any context. An initial palynological report (Marshall, pers. comm. 1999) was submitted to Jenny Clack with comments on a likely age from palynological elements that included then disjunct Viséan and Namurian forms. This age did not match to the conception of the probable local geology as it was apparently younger than all of the Carboniferous present within the area. So, further palynological analyses were undertaken by Stephenson (in Clack & Ahlberg Reference Clack, Ahlberg, Arratia, Wilson and Cloutier2004) on other specimens in the same drawer as the tetrapod. This gave a CM palynological assemblage, with Auroraspora macra and Schopfites claviger, i.e., characteristic of the Ballagan Formation and of latest Tournaisian age. Within the area studied by Portlock (Reference Portlock1843) there is a Ballagan Formation equivalent present, which is the Altagoan Formation from Draperstown. A study (Owens et al. Reference Owens, Gueinn and Cameron1977a) shows that the Altagoan Formation is also palynologically comparable to the Ballagan Formation, but with very rare specimens of Lycospora pusilla indicating, at least for the sampled level, a position in the early Viséan. This age assignment to the Tournaisian CM spore zone meant that Occidens portlocki became only the second tetrapod to be recognised from within Romer's Gap and in strata that were effectively contiguous to the Midland Valley of Scotland.

Figure 7 Stratigraphy of the Londonderry area and Northwest Carboniferous Basin of Ireland showing likely Brigantian (Viséan) stratigraphic level for the Occipens portlocki tetrapod jaw (*). Compiled from Mitchell & Somerville (Reference Mitchell, Somerville, Waters, Somerville, Jones, Cleal, Collinson, Waters, Besly, Dean, Stephenson, Davies, Freshney, Jackson, Mitchell, Powell, Barclay, Browne, Leveridge, Long and McLean2011), Mitchell & Owens (Reference Mitchell and Owens1990), Whitaker & Butterworth (Reference Whitaker and Butterworth1978), Monaghan et al. Reference Monaghan, Millward, Kearsey, Browne, Leslie, Smith and Strachan2024 and Waters et al. (Reference Waters, Somerville, Jones, Cleal, Collinson, Waters, Besly, Dean, Stephenson, Davies, Freshney, Jackson, Mitchell, Powell, Barclay, Browne, Leveridge, Long and McLean2011).

As the palynological report on the assemblage from the matrix of Occidens portlocki submitted to Jenny Clack was only preliminary, the opportunity is taken here to clarify its age in the context of our much better knowledge of Romer's Gap tetrapods. The palynological assemblage from the matrix of the jaw of Occidens was very thin but over 150 spores were identified and counted. These are listed in Table 1. In dealing with any potentially problematic assemblage it is important to try to find some common elements plus any preservational and thermal maturity characteristics that enable a consistent assemblage to be identified. One unusual feature of many of the spores is that they are three-dimensional (Fig 6, 6), i.e., not compressed as normal and with a common distinctive mineral damage to the exines. This is entirely consistent through preservation within a structure such as a bone, where the matrix would not undergo the normal amount of depth compaction. The most common spore is a large thick-walled species of Punctatisporites (P. irrasus). Diagnostic species include Lycospora (L. noctuina and L. pusilla), Tripartites vetustus and Triquitrites marginatus, Corbulispora cancellata, Grumosisporites and Cingulizonates bialatus. The age (Neves et al. Reference Neves, Gueinn, Clayton, Ioannides, Neville and Kruszewska1973; Owens et al. Reference Owens, McLean and Bodman2004) is clearly Viséan or younger from the presence of Lycospora pusilla. Tripartites vetustus is present, which is the zonal index for the VF spore zone but has a total range into the mid Namurian. This younger age for Occidens is supported by Triquitrites marginatus, which has an inception within the mid Viséan TC spore zone. In the now superseded Namurian spore zonation of Owens et al. (Reference Owens, Neves, Gueinn, Mishell, Sabry and Williams1977b), a spore inception within the NC zone that overlies the VF zone is Grumosisporites rufus. There are two specimens of Grumosisporites sp. within the Occidens spore assemblage that hints the age might be slightly younger than the VF spore zone. So, the assemblage from the matrix of Occidens portlocki is clearly significantly younger than the Tournaisian age preferred by Clack & Ahlberg (Reference Clack, Ahlberg, Arratia, Wilson and Cloutier2004).

Within the area formally mapped and monographed by Portlock (Reference Portlock1843) there is the thin Meenymore Formation (Fig. 7) that has been attributed to the VF spore zone (Mitchell Reference Mitchell and Mitchell2004; Mitchell & Somerville Reference Mitchell, Somerville, Waters, Somerville, Jones, Cleal, Collinson, Waters, Besly, Dean, Stephenson, Davies, Freshney, Jackson, Mitchell, Powell, Barclay, Browne, Leveridge, Long and McLean2011). The Meenymore Formation is a laterally extensive tidal flat of evaporitic facies (West et al. Reference West, Brandon and Smith1968) that is only 20 m thick in the Londonderry area, where it rests unconformably on the Asbian Desertmartin Limestone Formation. It is also the youngest preserved Carboniferous sequence in the area. In northwest Ireland (Loch Allen Basin) there has been a systematic attempt (Cózar et al. Reference Cózar, Somerville and Mitchell2006) to identify the Asbian–Brigantian regional substage boundary by combining data from ammonoids, foraminifera, conodonts and spores. These data confirm the value of the ammonoids and foraminifera for correlation to the established international zonation and that the Meenymore Formation has its base effectively coincident with the Asbian–Brigantian boundary. The correlations of Cózar et al. (Reference Cózar, Somerville and Mitchell2006) also emphasised that the NM to VF spore zone boundary lay within the Brigantian stage, the previous correlations to the regional stages being made in sequences with largely terrestrial sediments from the Midland Valley of Scotland. There is no palynological data from the local Meenymore Formation, but the VF spore zone has also been identified in the Ballycastle area of Antrim (Whitaker & Butterworth Reference Whitaker and Butterworth1978) and within the conglomeratic successions of the Fintona Block (Mitchell & Owens Reference Mitchell and Owens1990). Limited data on Tournaisian to Namurian spore assemblages have been reported (Higgs Reference Higgs1984) from the Northwest Carboniferous (Loch Allen) Basin, where the Meenymore Formation is placed in the underlying NM spore zone with Tripartites vetustus not appearing until the Bellavally Formation. The formations in the Northwest Carboniferous Basin studied by Higgs (Reference Higgs1984) are all much thicker (over 500 m of Brigantian Leitrim Group) than in the Londonderry area and there is direct evidence from the ammonoid zones for the attenuation of the Meenymore Formation onto depositional highs (e.g., the Lackagh Hills, Fig. 12 in Brandon & Hodson Reference Brandon and Hodson1984). So, it is possible that only the upper part of a lithologically distinctive transgressive unit (‘Meenymore’ Formation) was deposited on the basement high in the Londonderry area in contrast to the Northwest Carboniferous Basin.

This information enables us to revisit Portlock (Reference Portlock1843) in an attempt to identify the likely locality of the specimen. It is also important to understand how the 1843 Londonderry memoir was researched and compiled (Herries Davies Reference Herries Davies1983). Portlock had a team of surveyors, some geologically trained and some not, and this staff sometimes numbered as many as 35. But by 1840, the slow progress of the truly monographic work coupled with government retrenchment saw the staff cut to four, including Portlock, who were transferred to Dublin. The intention had been to produce a series of similar memoirs for all of Ireland but the project was abandoned. Portlock was rushed to complete the memoir, which he did in January 1843, and was posted back to his regiment which was then in Corfu. In the Londonderry memoir, the fossils were collectively illustrated and described by the talented geologist and artist George Du Noyer. Examination of Plate XIII reveals that the specimen illustrated as Fig. 13a & b is both sides of the jaw of Occidens, to the extent that the illustration of Clack & Ahlberg (Reference Clack, Ahlberg, Arratia, Wilson and Cloutier2004) shows it to be labelled XIIIB. However, it should be noted that an additional front piece to the jaw is present in the original illustration but missing from Clack & Ahlberg (Reference Clack, Ahlberg, Arratia, Wilson and Cloutier2004). Hopefully, it is still present within the BGS collection and can be reunited with the other two pieces.

Analysis of Portlock (Reference Portlock1843) reveals numerous records of Holoptychius, largely scales and some teeth. Holopytychius is a fish that is now known to be restricted to the Devonian, becoming extinct at or immediately below the Devonian–Carboniferous boundary (Sallan & Coates Reference Sallan and Coates2010). These different, largely Carboniferous rhizodont teeth are placed in Archichthys portlocki by Jeffery (Reference Jeffery2006), with the exception of Occidens portlocki. Portlock (Reference Portlock1843) lists a significant number of Holoptychius localities both in the systematic palaeontology, the stratigraphic compilation and the descriptions of the field sections. Although not always clear, these are often the same localities referred to in different ways. They are also not always accurate; for example, the Holywood locality is listed as being in County Antrim rather than County Down and is now a well-known site (Griffith & Wilson Reference Griffith and Wilson1982) for Holoptychius. In addition, the locality information for at least one fossil in Portlock (Reference Portlock1843, p. 464, Megalichthys hibbertii) was regarded as doubtful. Most of the localities are records of scales with the only teeth reported from the Moyola River and Falls of Bannagh. The latter were from mudstones (probably the Drumchoricke Siltstone Formation) of late Arundian age (Waters et al. Reference Waters, Somerville, Jones, Cleal, Collinson, Waters, Besly, Dean, Stephenson, Davies, Freshney, Jackson, Mitchell, Powell, Barclay, Browne, Leveridge, Long and McLean2011). The Moyola River locality is also referred to in Portlock (Reference Portlock1843) as being Maghera and the richest fish site with Gyracanthus as well as Holoptychius teeth and scales. The locality is mapped (Geological Survey of Northern Ireland 2006) as Mormeal Member of the Altagoan Formation and of latest Tournaisian to Chadian age.

Clearly this Brigantian VF spore zone age from the palynological assemblage is at variance with the ages of the Holoptychius teeth reported in Portlock (Reference Portlock1843). But nowhere in the text is there specific mention of a locality for the jaw. As noted, the cost and time overruns meant that the County Londonderry memoir project was wound down rapidly and moved to Dublin in 1840. The transferred collections (Herries Davies Reference Herries Davies1995) included some 4,824 catalogued fossils plus an additional 70 boxes of unnamed specimens. After a brief interval on other duties, Du Noyer was again assigned to the memoir project (December 1840), producing further illustrations of fossils (Hegarty Reference Hegarty2018) until February 1842 and thereafter working casually. One clue that the specimen of Occidens might, in fact, be from Ballycastle (Antrim) is from Baily (Reference Baily1871), who noted that the Portlock Collection on display in the Royal College of Sciences, Stephen's Green, Dublin included a small number of fish specimens collected from Ballycastle, most probably from black mudstones above the Old Salt Pans (formerly the Bath Lodge) Colliery. These black mudstones (Fig. 20 in Wilson & Robbie Reference Wilson and Robbie1966) are immediately above the Bath Lodge Coal and the youngest part of the succession in the Ballycastle Coalfield. This part of the succession is placed in the earliest Namurian (Wilson & Robbie Reference Wilson and Robbie1966) based on marine band correlation with the Midland Valley of Scotland and Macrihanish in the Kintyre. Palynological data from the Ballycastle Coalfield (Whitaker & Butterworth Reference Whitaker and Butterworth1978; Butterworth in Wilson & Robbie Reference Wilson and Robbie1966, p. 81) includes Rotaspora fracta and Tripartites vetustus, which have inceptions in the VF spore zone or younger. Butterworth also recorded Verrucosisporites morulatus that Owens et al. (Reference Owens, McLean and Bodman2004) used to define the Vm spore subzone to which it is restricted and is of Pendleian age, i.e., immediately post-Brigantian.

So, accepting the palynological assemblage from the Occidens jaw, the specimen is not from the Tournaisian but instead from the latest Viséan or younger. If it is from above the Bath Lodge Coal in the Ballycastle Coalfield then the age is early Pendleian. This locality provides a match with the palynological assemblage and there is a documented record of fossil fish from Ballycastle being within the Portlock Collection.

8. Blue Beach, Nova Scotia, Canada

The most comparable section to the Ballagan Formation that is of Tournaisian age and includes tetrapods both as skeletal material (Anderson et al. Reference Anderson, Smithson, Mansky, Meyer and Clack2015) and footprints (Mansky & Lucas Reference Mansky, Lucas, Lucas, DiMichele, Barrick, Schneider and Spielmann2013; Lucas et al. Reference Lucas, Stimson, King, Calder, Mansky, Hebert, Hunt, Lucas, Schneider, Wang and Nikolaeva2022) is Blue Beach in Nova Scotia, Canada. Here, the largely disarticulated tetrapods are in the Blue Beach Member of the Horton Bluff Formation. The section at Blue Beach has been studied extensively, with detailed sedimentology (Martel & Gibling Reference Martel and Gibling1996) and palaeontological collections housed in the Blue Beach Museum (Mansky & Lucas Reference Mansky, Lucas, Lucas, DiMichele, Barrick, Schneider and Spielmann2013; bluebeachfossilmuseum.com). The Horton Bluff Formation has been studied palynologically (Utting et al. Reference Utting, Keppie and Giles1989; Utting & Giles Reference Utting and Giles2004) but largely from sections other than Blue Beach. These palynological studies (Utting et al. Reference Utting, Keppie and Giles1989) show that the Blue Beach and Hurd Creek Members are all within the Spelaeotriletes cabotii spore zone and are of mid to late Tournaisian age, with the earliest Tournaisian represented (above a hiatus) by the largely sandstone lithologies of the Curry Brook and Harding Brook Members. These do not contain tetrapods. In addition, the Blue Beach section only includes the upper half of the exposed total thickness of the Blue Beach Member. So, although frequently portrayed as including the earliest Romer's Gap tetrapods, the Blue Beach section includes only part of the Tournaisian. Blue Beach also differs in its sedimentary environment, which is lacustrine mudstones and sandstones with a marine incursion (Mansky & Lucas Reference Mansky, Lucas, Lucas, DiMichele, Barrick, Schneider and Spielmann2013). These are ever wet environments and so somewhat different from the Ballagan Formation with its matrix of seasonally dry river, lagoon and marine incursions (Bennett et al. Reference Bennett, Kearsey, Davies, Leng, Millward, Smithson, Brand, Browne, Carpenter, Marshall, Dulson and Curry2021). In addition, the palynological assemblages from Blue Beach are different in not being dominated by simple spores and Schopfites claviger but instead by Spelaeotriletes and Vallatisporites. Blue Beach also includes a number of horizons with in situ lycopod stumps (Martel & Gibling Reference Martel and Gibling1996; Rygel et al. Reference Rygel, Calder, Gibling, Gingras, Melrose, Greb and DiMichele2006) representing a forest. Similar in situ lycopod roots are present in the Ballagan Formation at Burnmouth but are rather rare (Bennett et al. Reference Bennett, Kearsey, Davies, Leng, Millward, Smithson, Brand, Browne, Carpenter, Marshall, Dulson and Curry2021). So, Blue Beach and the Ballagan Formation are only really comparable in that both contain Tournaisian tetrapods; apart from that, they are quite dissimilar in age span, palaeoenvironment and vegetation. However, these differences are valuable in that different ecosystems are being sampled and will collectively give us a much better understanding of Tournaisian tetrapods.